Electrochimica Acta 462 (2023) 142733 Contents lists available at ScienceDirect Electrochimica Acta journal homepage: www.journals.elsevier.com/electrochimica-acta On the stability of electrochemical CO2 reduction reaction to formate at indium electrodes at biocompatible conditions Paniz Izadi , Aykut Kas , Philip Haus , Falk Harnisch * Department of Environmental Microbiology, Helmholtz-Centre for Environmental Research - UFZ, Leipzig, Germany A R T I C L E I N F O A B S T R A C T Keywords: In a circular economy, carbon dioxide (CO2) has to serve as feedstock that can be utilized by electrochemical CO2 Electrobiotechnology reduction reaction (eCO2RR). Using eCO2RR to C1-compounds such as formate (HCOO− ) allows producing feed Electrobiorefinery for microbial syntheses that generates value-added compounds. However, eCO2RR at biocompatible conditions is Electrocatalysis currently limited to short-term operation facing a gradual performance deterioration. Here, we evaluate the Carbon dioxide reduction possible parameters affecting the stability of performance in terms of formate production rate (rHCOO− ) and coulombic efficiency (CE) of eCO2RR at indium during 72 h batch mode operation. Formate accumulated over time affected catholyte conductivity, but statistical analysis showed this did not have an immediate influence on the performance. However, both are key factors altering the actual cathode potential over time that in turn is leading to changes in rHCOO− (maximum deviation of ± 0.03 mmol h− 1 cm− 2 at the stable performance at each condition) and CE (maximum deviation ± 40% at stable performance at each condition). These effects were more significant after reaching certain formate concentration and catholyte conductivity (ca. 70 mM and 21 mS cm− 1, respectively). These results highlight the potential obstacles needed to be considered and tackled in order to achieve stable rHCOO− and CE over a long-term eCO2RR operation. This study discusses how to overcome these obstacles from different operational perspectives. 1. Introduction were detected [5]. Methane (CH4) was reported as a product of eCO2RR using binary colloidal Cu and silver (Ag) nanoparticles [6]. For process Due to the growing population and mankind’s lifestyle on earth the development and engineering, the selective production of one com­ atmospheric levels of carbon dioxide (CO2) are rising. It is increasing pound is often required rather than a high diversity of products. Indium societal consensus that there is an urgent need to decrease the CO2 (In) is an electrocatalyst with high selectivity for formate with a notable emission and develop technologies utilising CO2 as a carbon feedstock. coulombic efficiency (CE) of even higher than 90%, being used for Utilisation of CO2 and its conversion to chemicals and fuels pave the way eCO2RR on different supports (e.g. carbon black) [7]. For instance, towards circular economy [1]. Offering the possibility to be operated at In2O3 nanoparticles were synthesized on carbon black for eCO2RR [8]. mild reaction conditions (e. g. ambient temperature and atmospheric The authors reported the CE of 87% for formate production at the pressure), the electrochemical CO2 reduction reaction (eCO2RR) is cathode potential of -0.9 V vs. RHE in 0.5 M KHCO3 electrolyte solution highly promising for converting CO2 to chemical products, in particular [8]. Previously, we investigated the eCO2RR to formate using electro­ C1-compounds such as formate [2]. Due to the large amount of energy deposited In on carbon electrodes as electrocatalyst at biocompatible required for dissociation of the C=O bond (795 kJ mol− 1) that is needed conditions, especially neutral pH as well as ambient temperature and for activation of CO2 molecules, electrocatalysts are required for the pressure, in a lab scale (50 mL) [9] as well as scaled-up (1 L) [10] eCO2RR [3]. By far, numerous materials such as tin (Sn), tin oxide electrobioreactors. Over 1 h of eCO2RR at the applied potential (Eapplied ) (SnO2) and especially copper (Cu) have been proposed as electro­ of -2.2 V vs. Ag/AgCl sat. KCl, respective formate production rates catalysts for eCO2RR [4]. Using Cu for eCO2RR for instance, a variety of (rHCOO− ) of 0.061 and 0.038 mmol h− 1 cm− 2, and CE of 94% and 74% products ranging from formate (HCOO− ) and carbon monoxide (CO) to were achieved in the lab scale and scaled-up reactors [9,10]. the compounds needing a C-C-bond formation such as ethylene (C2H4) Microbial based technologies offer another platform for conversion * Corresponding author. E-mail address: falk.harnisch@ufz.de (F. Harnisch). https://doi.org/10.1016/j.electacta.2023.142733 Received 30 January 2023; Received in revised form 26 May 2023; Accepted 11 June 2023 Available online 12 June 2023 0013-4686/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). P. Izadi et al. Electrochimica Acta 462 (2023) 142733 of CO2 to value-added chemicals. Syngas fermentation is among the All chemicals were of at least analytical grade. All solutions were pre­ most prominent platforms, in which microorganisms either reduce CO2 pared with de-ionized water (Merck Millipore, USA). All experiments as a carbon source using hydrogen (H2) as an energy source or exploit were performed in triplicate (except the conductivity experiment, which CO [11]. Generally, more diverse and especially more energy rich was performed in quadruplicate), and the results are reported as average products can be achieved from CO2 through microbial pathways than by with standard deviation. Time point t=0 mentioned throughout the eCO2RR (e.g. typically C2- to C8-compounds [12,13] compared to C1- to manuscript refers to the data recorded when starting the eCO2RR C2-compounds [2], respectively). However, the rate of microbial pro­ experiments. cesses is usually lower than of eCO2RR. For instance, a maximum mi­ crobial production rate of 0.093 mol L− 1 d− 1 caproate (hexanoate) 2.2. Experimental setup under optimal conditions through addition of desired substrates in open-culture biotechnological reactors is reported [14]. In addition, The experimental setup was identical to the previously used one [9]. providing the constant energy source for microbial CO2 reduction is still In brief, a two-chamber electrochemical cell with a 300 mL glass reactor a challenge due to the low solubility of H2 in aqueous solutions (1.6 mg serving as a working electrode (WE, cathode) compartment was used L− 1, 293 K and 1 atm), leading to its low and often limited availability (Fig. 1). The reactor was equipped with the 3D printed polylactic acid for microbial production [15,16]. Considering the advantages of high (PLA) lid. The lid contained several ports that allowed placing the rates and selectivity of eCO2RR, linking it to microbial production seems electrodes fixed throughout the experiments as well as sampling [9,22]. highly promising, as more valuable products than C1-compounds but The counter electrode (CE, anode) was inserted in a glass custom-made with higher production rate can be achieved [17]. As formate can serve tube which was interfaced to the reactor using a cation exchange as both carbon and energy sources for microbial metabolisms, it can be membrane (CEM, fumasep FKE, FuMA-Tech GmbH, Germany). The considered as a suitable substrate in order to produce value-added cathode was a graphite rod of high purity (99.997 %, Ø 0.635 × 6 cm, compounds [18]. The first proof was reported by Li et al [19]. In their C-00-RD-000121, Goodfellow, UK) with a net geometric surface area study, formate produced from CO2 at In electrodes was used by genet­ (inserted in the solution) of 2.5 cm2 connected to the rotating electrode ically engineered Ralstonia eutropha (currently known as Cupriavidus holder (CTV 101T, Tacussel, France). CEs were platinized titanium (Ti) necator) for production of higher alcohols. Production of the polymer plates (0.05 × 1 × 7 cm, platinode, Umicore Galvanotechnik GmbH, polyhydroxyrate (PHB) by the microorganism of Cupriavidus necator Germany) and reference electrodes (RE) were Ag/AgCl in saturated KCl using formate gained through eCO2RR is another example for produc­ (SE 11, +0.197V vs. SHE, Xylem Analytics Germany Sales GmbH & Co. tion of valuable compounds from CO2 through integrating electro­ KG Sensortechnik Meinsberg, Germany), respectively. To perform the chemical and microbial syntheses [20]. In addition to pure cultures, experiments under potentiostatic mode, a potentiostat (SP-50, BioLogic microbial mixed cultures were also used and acetate and methane were Science Instruments, France) was used, while a DC-power supply microbially produced from CO2 using formate from the eCO2RR at (Keithley 2230-30-1, Keithley Instruments, USA) was used for galva­ copper electrodes [21]. nostatic operation. Although Eapplied at the cathode electrode was For future process engineering and steering, deciphering the impact controlled by the potentiostat, the actual electrode potential (EWE ) was of environmental as well as operational parameters on the eCO2RR at still measured against RE. For measuring the internal resistance, elec­ biocompatible conditions is required. Thereby, one main concern is trochemical impedance spectroscopy (EIS) was performed under open performance stability over time and its consistency across studies. For In circuit potential (OCP) using frequencies between 100 kHz and 0.1 Hz cathodes, the optimal CE and rHCOO− achieved during the first hour with an amplitude of 20 mV. seemed not to remain stable over time [22]. As operation over days up to Before and during the eCO2RR a constant CO2 flow of 1.2 vvm was weeks or even months is required, performance decline limits the provided using a rotameter (Influx Measurement, United Kingdom) to applicability of eCO2RR. A few reasons are considered to cause the the cathode compartment passed through a wash bottle containing performance decline. One is related to H2 evolution reaction (HER) [23] water to humidify the gas and prevent evaporation. Both catholyte and that is not only competing with eCO2RR and thus lowering CE and anolyte were 50 mL, 0.05 M NaHCO3 (pH 6.5). The catholyte was rHCOO− , but may also lead to the leaching of electrocatalysts, e.g. by constantly stirred using a magnetic stir bar at 1,000 rpm. Prior to use, all spalling off [24]. In addition, various operating conditions can affect the glassware and reactor components were UV sterilized in biosafety cab­ eCO2RR as well as HER and other side-reactions over the long-term inets (Maxisafe-2030i, Thermo Scientific, Germany). Anolyte was operation. One of these is the electrolytic conductivity of the solution continuously refreshed with a peristaltic pump (ISMATEC Reglo Analog (κS ). Due to its correlation with the internal resistance including the i × MS-4/8, Cole-Parmer GmbH, Germany) to the anode compartment with R-drop, the activity and selectivity of eCO2RR may vary with κS and at flow rate of 1.2 mL min− 1. All eCO2RR experiments were conducted at different electrolyte composition [25]. However, it is still unknown to 30◦ C in an incubator hood (TH 30, Edmund Bühler GmbH - 72411 what extend these as well as other operational parameters do affect the Bodelshausen Germany). performance of eCO2RR over a longer-term operation than few hours. In our previous study on eCO2RR to formate using In deposited 2.3. Pre-treatment and In deposition on the WE graphite rod as a cathode, we showed that the highest CE was achieved at the Eapplied of -2.2 V vs. Ag/AgCl sat. KCl with 90% [9]. As these results Pre-treatment and In deposition was performed as described previ­ were derived from 1 h operation, here we investigated the longer-term ously according to the optimal conditions [22] that were later slightly operation of 72 h using the identical setup. We deciphered the effects modified [9]. After mechanical and electrochemical pre-treatment of the of different cathode potentials (ranging from -2.2 to -1.6 V vs. Ag/AgCl graphite electrode, In was electrochemically deposited. During the sat. KCl), presence of already accumulated formate, and catholyte con­ electrodeposition step, electrode holder rotated at 10 rpm to assure a ductivity on long-term operation of eCO2RR to formate as well as elec­ homogenous coverage of the graphite rod with In. Electrodeposition was trode corrosion. performed at -0.9 V with the defined charge of 1.7 C. The solution for electrodeposition step was 50 mL of 0.1 M InCl3 (anhydrous, 99.999%, 2. Experimental Chempur, Germany) in 1 M acetate buffer (pH 4.5) in a single chamber reactor and was sparged with N2 at 1.6 vvm flow rate adjusted with a 2.1. General statements rotameter (Influx Measurement, UK). After electrodeposition, 2.5 cm2 net surface area of the electrode was covered with maximum 2,200 All potentials in this study are reported vs. Ag/AgCl in saturated KCl theoretical atomic layers of In [9]. The pre-treatment and electrodepo­ (0.197 V vs. standard hydrogen electrode (SHE)) if not stated otherwise. sition steps were carried out independently before each single eCO2RR 2 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 Fig. 1. Schematic of the reactor used in electrochemical CO2 reduction experiments in this study (details see text). experiment. Before using the In-coated graphite electrodes for the polarography, samples were randomly selected and measured addi­ eCO2RR experiments, they were rinsed with deionized water. tionally using ELEMENTTMXR sector field ICP-MS (Thermo Fisher Sci­ entific, Germany) and compared with the results from the polarography 2.4. Chemical analyses (Table S1). Due to the resemblance of the data, all the samples were measured using polarography. Throughout the eCO2RR experiments, 400 μL catholyte samples were collected through a sampling port regularly (every 2 hours during the daytime). Samples were analyzed using a pH electrode (InLab Micro Pro, 2.5. Data processing and statistical analyses Mettler Toledo, Switzerland) and a conductivity probe (InLab 731-ISM, Mettler Toledo, Switzerland) connected to the pH/conductivity bench­ CE was calculated for the eCO2RR to formate from the total amount top meter (S470, Mettler Toledo, Switzerland). To detect the products, of charge (Qtotal ) and theoretical amount of charge (QHCOO− ) required for liquid phase analysis was conducted with an HPLC (Shimadzu Scientific the amount of formate (nHCOO− ) measured with HPLC: Instruments, USA) equipped with a refractive index detector (RID) (RID- QHCOO− nHCOO− × z × F 20A, Shimadzu Scientific Instruments, Japan) Hi-Plex H column (300 CE (%) = = ∫t × 100 (1) Qtotal Idt mm × 7.7 mm ID, 8 µm pore size, Agilent Technologies, Germany) with t=0 a pre-column (Carbo-H 4 mm × 3 mm ID, Security Guard, Phenomenex). with z=2 being the number of electrons transferred per CO2 molecule Isocratic elution was performed with 0.005 M H2SO4 at 0.5 mL min− 1 at and F being the Faraday’s constant (96,485 C mol− 1 C). All the CE values 50◦ C for 30 min− 1. Formate was the only product from eCO2RR detected reported within the manuscript were calculated at a certain time interval in the catholyte. Calibration of formate (1.14 mM to 44.4 mM, five-point between the given and last sampling points. calibration with triplicate standards for each point, R2= 0.99) was car­ The rate of formate production rHCOO− was calculated based on the ried out with external standards. amount of formate produced (nHCOO− ) between sampling points (dt) and To measure In leached in the catholyte, 5 mL sample was collected net surface area of the In-coated graphite electrode (A= 2.5 cm2): from the solution every 24 hours, and was replaced by the 5 mL fresh ( ) solution. In case of possible change in the formate concentration rHCOO− mmol = nHCOO− (2) (cHCOO− ) after replacing the medium, it was measured after each sam­ 2 cm h A × dt pling for In measurement. The concentration of In was measured using The formal potentials for eCO2RR to formate (ECO2 /HCOO− 0# ) and polarography (797 VA Computrace, Metrohm, Switzerland) equipped HER (E0# 2H+/H2 ) were calculated (Eqs. (4) and (7)) using the standard with platinum wire as CE and a Ag/AgCl in saturated KCl as RE by differential pulse (DP) mode at a hanging mercury drop electrode electrode potentials for these reactions (E0CO2 /HCOO− =-0.22V vs. SHE and (HMDE) [26]. Calibration was conducted with external In standards E02H+/H2 = 0 V vs. SHE, respectively calculated using the Gibbs free energy (0.906 nM to 0.906 mM, three-point calibration, R2= 0.99) prepared in of individual species at pH=0 extracted from [27]), molar gas constant triple distilled water. To assure the credibility of the data from R=8.314 J mol− 1 K− 1 and temperature T=303.15 K. Equilibrium 3 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 constant of the reaction calculated using cHCOO− measured by HPLC, pH ( ) of the samples and saturated concentration of CO2 and H2 gasses (cCO2 0# E2H+/H 0 = E2H+/H − RT 0 lnQ = E2H+/H − RT ln cH2 (7) and cH2 , respectively) under the experimental conditions (Table S9). 2 2 zF 2 zF (cH+ )2 Overpotentials (ηCO2 /HCOO− and η2H+/H2 ) were calculated (Eqs. (5) and (8)) from the actual EWE at the sampling times measured with either a 0# η2H+/H2 = EWE − E2H+/H 2 (8) digital multimeter (RS PRO RS-14, RS Components GmbH, Germany) or from the continuous data recorded by potentiostat/galvanostat and the In order to evaluate the effect of different parameters (i.e. indepen­ dent variables) on the performance of the eCO2RR, statistical analyses electrode potentials (E0# 0# CO2 /HCOO− and E2H+/H2 ) calculated through exper­ were performed. These selected independent variables were EWE , cHCOO− imental conditions and concentrations of reactants. and κS and their effects on the system performance were studied (Table CO2 + 2e− + H + →HCOO− (3) S2). To monitor the effect of cathode potential, the results of the trip­ licate experiments at 4 different Eapplied (-1.6, -1.8, -2.0 and -2.2 V) were ( ) RT RT cHCOO− considered. The effect of cHCOO− was monitored using the results from the E0# 0 CO2 /HCOO− = ECO2 /HCOO− − lnQ = E0CO2 /HCOO− − ln zF zF cH+ × cCO2 separate quadruplicate experiments, in which 70 mM formate was (4) manually added to the catholyte at t=0. To assess the effect of κS on the eCO2RR performance in separate triplicate experiments, 17 g L− 1 so­ ηCO2 /HCOO− = EWE − E0# (5) dium sulphate (Na2SO4) was added to the catholyte at t=0, leading a κS of 23.0 ± 0.1 mS cm− 1 being identical to the catholyte κS achieved after CO2 /HCOO− 2H + + 2e− →H2 (6) 24 h under Eapplied of -2.2 V. Due to the difference observed between the values of Eapplied and EWE over time (see results), EWE was used in all the statistical analyses. All the statistical analyses were performed using R [28]. When the Fig. 2. Formate production rate (rHCOO− ) and coulombic efficiency (CE) during electrochemical CO2 reduction to formate at 4 different applied potentials (Eapplied ) of A) -2.2 V, B) -2.0 V, C) -1.8 V, and D) -1.6 V. All experiments were conducted for 72 h, in 0.05 M NaHCO3 (pH 6.5) as electrolyte solutions and in triplicate (n=3). Reported values are mean values and the error bars represent the confidence interval (α = 0.05). 4 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 multiple dependent variables were considered, multivariate analysis of between the EWE and cHCOO− (p-value = 1.741 × 10− 2), CE (p-value = variance (MANOVA) test was used. MANOVA tests (and analysis of 2.487 × 10− 2), rHCOO− (p-value = 1.068 × 10− 4), and pH of the catholyte variance (ANOVA) tests, when required) were performed using “stats” (p-value = 2.397 × 10− 8) which can be also observed in the boxplots by package in R [29]. To visualize the data, the box-plots were plotted using differences in medians (Fig. 3). The effect of EWE on eCO2RR to formate “ggplot2” package in R [30]. is also in agreement with previous studies, although in these studies the EWE was not measured but being assumed identical to Eapplied and 3. Results and Discussions considered stable over time [9]. The two parameters impacting the EWE and hence the eCO2RR performance were further evaluated. 3.1. 72 h-operation of eCO2RR to formate at different potentials 3.3. The effect of cHCOO− in the catholyte on the eCO2RR For 1 h eCO2RR, an Eapplied of -2.2 V was shown as an optimum for production of formate with more than 90% of CE at In electrodes [9]. One of the two variables showing a probable effect on the fluctuation Therefore, it was selected for a 72 h-operation. Formate production of rHCOO− and CE was reaching a cHCOO− of ca. 70 mM. Therefore, began immediately and rHCOO− = 0.061 ± 0.002 mmol cm− 2 h− 1 and CE eCO2RR at the Eapplied of -2.2 V was performed with this formate con­ = 82.3 ± 4.8 % were almost stable over the first few hours up to 24 h centration being already present at the beginning of the experiment. (Fig. 2), in line with our previous study at the Eapplied of -2.2 V [9]. Visually, the now obtained rHCOO− and CE seemed similar to the values Although similar rHCOO− and CE were calculated for operation from 24 h after 24 h when no formate was provided initially (Fig. 2A) including to 72 h, strong fluctuations were observed. This was assumed to be observed fluctuations (Fig. 4A). However, MANOVA and further ANOVA related to HER competing with eCO2RR, although H2 in the headspace tests showed no significant correlation between the accumulated was not measured during the experiment due the constant CO2 flushing. formate and rHCOO− (p-value = 0.6618) or CE (p-value = 0.8494), which HER could contribute to 1) the gradual increase in pH of the catholyte can also be seen in the box-plots (Fig. 4B). On the other hand, cHCOO− over time (from ca. 6.4 at t=0 to 7.5 at t=72 h) due to decrease of proton seemed to significantly affect the EWE (p-value= 2.2 × 10− 16) and pH (p- concentration (Eq. 6) and lead to 2) increase in the indium leaching in value= 2.2 × 10− 16). Therefore, we assume that a gradual increase in the solution (Table S3). The increase in the leaching rate (from ca. 37 at the cHCOO− leads to the change in EWE being different from Eapplied , which 24 h to 780 ng mL− 1 day− 1 at 72 h, or considering the cathode surface influences rHCOO− and CE and this leads to the occasionally observed from ca. 733 at 24 h to 15,606 ng cm− 2 day− 1 at 72 h) could be related to fluctuations. the increasing generation of H2 bubbles at the surface of the electrode, It is worth mentioning that at the initial eCO2RR at Eapplied of -2.2 V leading to spalling off of In particles. The reason for occasional increase without formate addition, the κS was ca. 5 mS cm− 1, whereas addition of in CE to more than 100% was not clear. However, a similar trend was 70 mM formate to the catholyte changed the κS value to ca. 9 mS cm− 1 also observed in separate experiments with the Eapplied of -2.0 and -1.8 V, (Table S5). This showed that the κS of 21 mS cm− 1 after 24 h when 70 in which a CE of ca. 120% was measured after 48 h of operation (Fig. 2). mM formate was accumulated by eCO2RR (Fig. 2A) was mostly related Diffusion of formate through the cation exchange membrane (CEM) due to charge balancing ion transfer through the membrane [31] and not to to the concentration gradient can be excluded, as no formate was formate accumulation. detected in the anode compartment of all reactors. On the contrary, rHCOO− and CE (in average 0.018 mmol h− 1 cm− 2 3.4. The effect of κS of the catholyte on the eCO2RR and 60%, respectively) calculated at the Eapplied of -1.6 V was almost stable over the 72 h-operation. Interestingly, sudden strong fluctuations One of the important parameters in eCO2RR is κS . A conductive so­ in rHCOO− and CE began after 24, 30 and 46 h at the Eapplied of -2.2, -2.0 lution is required to facilitate charge balancing ion transfer at low in­ and -1.8 V, respectively. These time points at different potentials had ternal resistance. Internal resistance is not only important for the entire two facts in common: 1) reaching a similar cHCOO− accumulated in the electrochemical cell and hence for the electric power per produced catholyte (ca. 70 mM), and 2) reaching a similar catholyte conductivity product, but also for the i × R-drop between RE and WE [32]. In the (κS of ca. 21 mS cm− 1 from initially 4.5 mS cm− 1). We speculated that eCO2RR at the Eapplied of -2.2 V, κS was almost 5 mS cm− 1 at t=0, however these two factors as well as change in pH may have caused the occa­ it increased one order of magnitude after 72 h, reaching ca. 48 mS cm− 1. sional inconsistencies and thus studied them further. Similar increase in the κS was also observed in the eCO2RR at Eapplied of -2.0 and -1.8 V. As mentioned above, the occasional fluctuation in the 3.2. Set and actual cathode potential and statistical analysis of pH, κS rHCOO− and CE appeared to start after reaching κS of ca. 21 mS cm− 1. and cHCOO− Therefore, in separate experiments 17 g L− 1 Na2SO4 was added to the bicarbonate solutions for pre-setting a κS of 23.0 ± 0.1 mS cm− 1 and the During eCO2RR using a conventional potentiostat we noticed that the experiments were conducted for 48 h. As the anolyte solution was EWE monitored against RE were different from Eapplied (-2.2, -2.0 and -1.8 refreshed permanently during all experiments, its κS value remained V, Tables S4-7). At the Eapplied of -2.2 V, the EWE began to change few stable at ca. 5 mS cm− 1. Despite the occasional increases in rHCOO− and hours after starting the experiment and reached to almost -1.7 V. Similar CE (Fig. 5A), both performance parameters remained stable during the potential change was recorded at Eapplied of -2.0 and -1.8 V, in which the experiment. Although an increase in the κS was expected to influence the EWE increased and reached to -1.7 V, this was not the case at Eapplied of performance of the eCO2RR, a MANOVA test showed that there was no -1.6 V. In order to check if this was related to the operational mode, an correlation between the κS of the catholyte and rHCOO− (p-value: 0.8909) almost identical experiment was performed but using galvanostatic or CE (p-value: 0.3177). Yet, again the change in κS affected the EWE operation. Here, the current of -15 mA was fixed using a DC power (p-value: 8.617 × 10− 4) and pH (p-value: 2.2 × 10− 16), which may in supply, which corresponded to the current during potentiostatic oper­ turn influence the performance (Fig. 5B). ation with Eapplied of -2.2 V. However, similar to the potentiostatic operation the EWE changed to around -1.7 V over time (Table S8). 3.5. It is not only catholyte, but also cathode surface Due to the difference between the Eapplied and EWE , the latter was used as an independent variable for statistical analysis, thus the effect of EWE We showed that the longer-term operation of eCO2RR is feasible, yet on cHCOO− , CE, rHCOO− , and pH was studied. MANOVA test (p-value= 2.2 the best performance was only achieved over the first few hours. As × 10− 16) showed that there was at least one group different from others. shown above, this performance decrease seems mainly to be related to The summary of the MANOVA test revealed the significant relationship changes in cHCOO− and κS of the catholyte. Therefore, we assumed that 5 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 Fig. 3. Box-plots of formate concentration (cHCOO− ) accumulated in the catholyte, formate production rate (rHCOO− ), coulombic efficiency (CE) and pH at different actual cathode potentials (EWE ) denoted at the x-axis. All data are from 72 h electrochemical CO2 reduction to formate at 4 different applied potentials (Eapplied ) of -2.2 V, -2.0 V, -1.8 V, and -1.6 V each performed in triplicate (n=3). In each box, horizontal black lines denote median values, extending from the 25th to the 75th percentile of each group’s distribution of values. The error bars denote the 5th and 95th percentile values. Black dots denote the observations outside the range of adjacent values. refreshing the catholyte should enhance or at least stabilize the perfor­ catalysts [35]. HER needs to be suppressed in order to achieve an effi­ mance. Thus, the catholyte was completely replaced with a fresh solu­ cient eCO2RR. For instance, it was reported that increasing the mass tion after 72 h of experiment in the reactors operated at the Eapplied of -2.2 transport in the system using the high-speed rotating gold electrode V, and the operation was further carried out for another 28 h and the leads to an increase of eCO2RR rather than HER by adjusting the pH at data monitored after 24 h was compared with the data collected from electrode surface [36]. In addition, some studies report that higher the first 24 h (first 24 h of 72 h-operation). In contrast to our expecta­ cathodic potentials (more negative) at indium can surprisingly be used tions, lower CE (around 50%) and rHCOO− of 0.06 ± 0.02 mmol h− 1 cm− 2 to decrease the HER; even at a higher production rates of eCO2RR [9]. were achieved during the further 24 h while the κS remaining below 21 mS cm− 1 (Fig. S1). Thus, we hypothesized that the performance decline was as a result of corrosion of the electrocatalyst at the cathode surface 3.6. Thermodynamic considerations during the first 72 h of operation. Catalyst corrosion and the resulting leaching of In to the catholyte is inevitable. Particularly at the cathode Accumulated formate and pH change could cause shifts in formal potentials of eCO2RR, which can be calculated by the Nernst equation. potentials needed for eCO2RR having a high probability of HER leading to gas bubble formation and spalling off of metal particles. At carbon- An experimental measurement of formal potentials by approximating the midpoint of anodic and cathodic peaks using cyclic voltammetry can based electrodes, the HER commonly starts at cathode potentials of -0.8 to -0.9 V [33]. At the surface of the electrode, particularly at the only be performed for reversible electrochemical reactions with low rate kinetics, which is not the case for the eCO2RR [37]. Hence, formal po­ areas that In was probably not homogenously deposited, generated H2 bubbles could cause In leaching in the catholyte. This was observed in tentials were calculated starting from the standard potential and using experimental conditions and saturated concentration of CO2 [38]. Shift the experiments. The corrosion rate was almost similar for Eapplied of -1.8, in the formal potential of eCO2RR at different applied potentials and -2.0 and -2.2 V and increased after 72 h of operation to the average of operational parameters found to be between -49 mV to -64 mV within 760 ng mL− 1 day− 1 or 15,248 ng cm− 2 day− 1 (Table S3). The slight the first 24 h of operation and around -78 mV to -93 mV for 72 h differences in the corrosion values at different potentials (Table S3) were operation except the formate addition experiments resulting in lower more likely related to the higher HER rate at more negative potentials. shift of formal potential (≈-30 mV at t=24 h, ≈-45 mV at t=48 h). Predictably, the corrosion rates were much lower for Eapplied = -1.6 V (in Concentrations of both H+ and HCOO− contributed to the shift in formal average 18 ng mL− 1 day− 1 after 24 h to 44 ng mL− 1 day− 1 after 72 h). potential significantly, for instance formal potential shift of -84 ± 1 mV The competition of HER and eCO2RR has been reported in previous at the Eapplied of -2.2 V was recalculated under an assumed constant studies using indium/indium oxide [34] as well as other types of (initial) pH but at increased formate concentration as only -54 ± 6 mV. 6 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 Fig. 4. A) Formate production rate (rHCOO− ) and coulombic efficiency (CE) during electrochemical CO2 reduction to formate at the applied potential (Eapplied ) of -2.2 V with formate provided at t=0. All experiments were conducted for 48 h, in 0.05 M NaHCO3 (pH 6.5) with formate concentration (cHCOO− ) of 70 mM as catholyte and 0.05 M NaHCO3 (pH 6.5) as anolyte in quadruplicate (n=4). Reported values are mean values and the error bars represent the confidence interval (α = 0.05). B) Box- plots of formate production rate (rHCOO− ), coulombic efficiency (CE), actual cathode potentials (EWE ) and pH at different cHCOO− accumulated in the catholyte denoted at the x-axis. All the data are from 48 h electrochemical CO2 reduction to formate at the Eapplied of -2.2 V with 70 mM formate provided in catholyte at t=0, performed in quadruplicate (n=4). In each box, horizontal black lines denote median values, extending from the 25th to the 75th percentile of each group’s distribution of values. The error bars denote the 5th and 95th percentile values. Black dots denote the observations outside the range of adjacent values. In summary, formal potential shifts were less than -100 mV for all selectivity between eCO2RR and HER is based on mechanistic or kinetic eCO2RR experiments (Table S11). Similar shifts in formal potential were reasons, similar to the findings of the previous study with the same observed for all potentials applied and experimental conditions, there­ reactor setup [9]. fore the changes in CE are less likely to be explained thermodynamically. Formal potentials of eCO2RR and HER also shift similarly during continuous reactor operation further indicating that any shift in 7 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 Fig. 5. A) Formate production rate (rHCOO− ) and coulombic efficiency (CE) achieved during electrochemical CO2 reduction to formate at the applied potential (Eapplied ) of -2.2 V with catholyte conductivity (κS ) adjusted at ca. 23 mS cm− 1 at t=0. All experiments were conducted for 48 h, in 0.05 M NaHCO3 (pH 6.5) with 17 g L− 1 Na2SO4 as catholyte and 0.05 M NaHCO3 (pH 6.5) as anolyte in triplicate (n=3). Reported values are mean values and the error bars represent the confidence interval (α = 0.05). B) Box-plots of formate production rate (rHCOO− ), coulombic efficiency (CE), actual cathodic potentials (EWE ) and pH at different catholyte κS denoted at the x-axis. All the data are from 48 h electrochemical CO2 reduction to formate at the Eapplied of -2.2 V with 17 g L− 1 Na2SO4 provided in catholyte at t=0, performed in triplicate (n=3). In each box, horizontal black lines denote median values, extending from the 25th to the 75th percentile of each group’s distribution of values. The error bars denote the 5th and 95th percentile values. Black dots denote the observations outside the range of adjacent values. 3.7. Implications of reactor design and operation cations (esp. Na+) inevitably passes through the membrane from the anolyte to the catholyte. Yet, under biocompatible conditions not only It was observed that changes in the catholyte especially in κS affected cations but also anions are transferred through CEM from catholyte to the EWE leading to a difference in the set Eapplied . This κS change is as a the anolyte [40]. Besides, in addition to formate production, there could result of charge balancing ion transfer through the membrane between be some shifts between carbonate/bicarbonate equilibrium and con­ the compartments [39]. The CEM placed between the anodic and centrations. Consequently, the concentration of anions and cations in cathodic compartments is assumed to be cation selective, therefore the the catholyte increases during eCO2RR. This was confirmed by not only 8 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 increase in κS , but also decrease in the internal resistance as EIS analysis the eCO2RR was reported to re-activate the catalyst. It was shown that showed (6 to 1 Ω, after 72 h experiment at Eapplied = -2.2 V). Therefore, the pulsed oxidation steps lead to formation of a “fresh” copper oxide the change in catholyte solution led to further changes in EWE , and layer at the surface of the electrode, allowing the formation of eventually the occasional fluctuations observed for the production rate surface-bound intermediates favouring hydrocarbon production by and efficiency (also confirmed statistically). Interestingly, the cation eCO2RR over HER [49]. The repeated cyclic voltammetry scans with transfer through the membrane during eCO2RR was recently discussed anodic reverse potentials seemed to show similar results to pulsed to cause changes in the electrical double layer at the cathode, limiting oxidation steps, increasing the methane production by eCO2RR to up to the active sites required for eCO2RR [41]. Therefore, the specific cations 50 h [48]. in the electrolytes have to be further taken into consideration. Anion Another approach towards stability of electrodes and therefore long- exchange membranes (AEMs) are commonly not a suitable alternative to term eCO2RR operation is using bimetallic catalysts such as Cu-In, by CEMs for eCO2RR, due to the more likely crossover of formate [42]. The decreasing the eCO2RR overpotentials [7]. Depending on lattice cation transfer through the CEM could be prevented by using a bipolar compatibility of the metals, the bimetallic catalysts can be in form of membrane (BPM) [43]. However, the larger cell potential required when alloy or heterostructure. An electronic interaction of the alloy catalysts using BPM (due to the lack of proton transfer through the membrane and and a synergistic interaction at the interfacial regions for the hetero­ the extra driven force required for water splitting reaction of 800 mV) is structured catalysts could enhance the adsorption of ●COOH species a significant disadvantage [31]. Alternatively, membraneless single and weaken the adsorption of ●H species, leading to HER inhabitation chamber reactors could be used, but that would lead to new challenges [7]. Although the activity of the interfacial sites between the Cu and In such as preventing oxidation of formate at the CE. Additionally, increase in the Cu-In catalysts is not completely understood, the suppression of in cHCOO− accumulated in the solution and its effect on the cathode EWE HER could be the reason for stability of Cu-In reported previously. For could not be neglected. This could be potentially tackled by constant instance, the 7 h operation of Cu-In catalyst at -0.6 V vs. RHE in 0.1 M refreshing the catholyte during operation. However, in case of further KHCO3 with a selectivity of 95% towards CO was reported [50]. Oper­ downstream processes to separate formate from the catholyte, the ating eCO2RR for 60 h with stable CE at 90% for CO production at the concentrated formate in the electrolyte solution is more favourable for potentials of -0.6 to -0.8 V vs. RHE was achieved by Luo at al using Cu the separation of formate from the matrix solution [18]. nanowires covered by a thin layer of In [51]. Long-term stability of In this study, we showed that eCO2RR to formate in a two-chamber In2O3/Cu2O and In2O3/Cu3N at -0.6 V vs. RHE in 0.1 M KHCO3 was also electrochemical cell using an In electrode can be operated over more reported. Being CO as the main product again, stable CE of 80% was than only few hours. Nevertheless, changes within the catholyte (e.g. κS achieved during 50 h eCO2RR [52]. Many other cathodic electro­ and cHCOO− ) and cathode (e.g. catalyst leaching) over the long run in­ catalysts were tested in combination with In. One example is palladium fluence the eCO2RR and eventually limit the performance. As some of (Pd)-In/three-dimensional graphene which showed the highest CE of these changes seem inevitable, keeping the performance constant during 85% being stable for 24 h at -1.6 V vs. Ag/AgCl in 0.5 KHCO3 [53]. The the long-term operation is a challenge with Eapplied being the key authors discussed the small particle size and high particle dispersion to parameter. A very negative Eapplied has advantage of higher production be the reason for stable catalytic performance. rates, albeit a higher energy demand per product. More importantly it also leads to higher rate of catalyst leaching that is deteriorating the 3.9. Integration of eCO2RR and microbial synthesis performance. If less negative Eapplied is set, it rather keeps the catalyst structure firmer during the operation, but only at lower production rate. One of the main reasons of targeting formate production from CO2 at Adjusting stability and performance needs a delicate balance, maybe biocompatible conditions is the role of formate as both carbon and en­ also including pulse operation or (periodically) different Eapplied . In ergy sources for microbial metabolisms and further biosynthesis of addition to affecting the cathode, the changes in catholyte bring further value-added compounds [18,54]. According to previous successful challenges, for instance on charge balancing ion transfer through the studies on CO2RR combining the electrochemical and microbial pro­ membrane and changes in κS and formate accumulating over time. These cesses [19,55], the cHCOO− , rHCOO− and CE achieved in our study is suf­ challenges originate from the influence of κS and cHCOO− on the EWE , and ficient for further biosynthesis using formate in case of integrating as shown here the set electrode potential does not remain stable and is in eCO2RR and microbial synthesis. In fact, one of the factors causing turn affecting the production rate and efficiency. Hence, the electrode changes in EWE at the cathodic potentials of -1.8, -2.0 and -2.2 V was potential during the long-term operation of eCO2RR needs to be moni­ accumulation of formate in the catholyte, which could be tackled by tored and re-adjusted regularly. simultaneous consumption of formate by microbial cells after its pro­ duction through eCO2RR. Nevertheless, two potential issues need to be considered and examined when integrating eCO2RR and microbial 3.8. Towards catalyst stability for long-term eCO2RR operations synthesis. First, the catholyte used in our setup was only 0.05 M NaHCO3, which does not provide an optimal condition for microbial This study strongly highlights the importance of catalyst stability growth and activities. If microbial media is used as a catholyte the high over long-term operation of eCO2RR on the example of In cathodes. κS of the catholyte due to the additional chemicals vital for microbial Stability of the electrocatalysts depends on the electrode architecture, growth will lead to more changes in EWE and therefore the fluctuations but especially material. The exception are very few electrocatalysts for in rHCOO− and CE, as observed in this study. Second, the single or eCO2RR such as Ag that showed high stability for conversion of CO2 to bimetallic catalysts may provide toxicity for microbial cells particularly syngas (>1200 h) [44]. In particular, the long-term stability of the over the long run. Therefore, to what extend these two issues are electrodes with only In as the catalyst was not studied before. However, affecting the eCO2RR and further biosynthesis is a crucial question pulsed electrolysis has been shown as an effective method for Cu elec­ which needs to be systematically studied, in order to find optimal con­ trodes [45]. Cu containing electrocatalysts also suffer from only ditions favourable for both electrochemical and microbial processes. short-term stability and decreasing selectivity of eCO2RR, e.g. for The answer to this question will potentially incite electrochemical and methane and ethylene [46]. This was mainly assigned to morphological microbial technologies further towards industrial applications. changes in the electrocatalyst surface [5,47]. One possible causation for deactivation of the catalyst could be poisoning by deposition of the 4. Conclusions impurities from the solution on the catalyst surface. Although this was the case in some studies [47], no deposited impurities were found in This study investigated the potential parameters limiting the stable other studies [48]. Adding pulsed oxidation as intermittent steps during eCO2RR to formate using In catalyst over 72 h operation. We showed 9 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 that in order to achieve a stable and efficient eCO2RR, the interrelation products with electrochemical reduction of CO2, Sustainable Energy & Fuels (2021). between few operational parameters needs to be taken into consider­ [3] A. Ruiz-Valencia, D. Benmeziane, N. Pen, E. Petit, V. Bonniol, M. Belleville, ation. In the case studied here being exemplarily for interfacing eCO2RR D. Paolucci, J. Sanchez-Marcano, L. Soussan, CO2 valorization by a new and bioproduction, a cHCOO− increase in the catholyte and change in κS of microbiological process, Catalysis Today 346 (2020) 106–111. the catholyte may not directly influence the rHCOO− and CE. Yet, they [4] J. Qiao, Y. Liu, F. Hong, J. Zhang, A review of catalysts for the electroreduction of carbon dioxide to produce low-carbon fuels, Chemical Society Reviews 43 (2014) affect the EWE , leading to fluctuations in rHCOO− (maximum ± 0.03 mmol 631–675. h− 1 cm− 2 at stable operation at each condition) and CE (max. ± 40% at [5] K.P. Kuhl, E.R. Cave, D.N. Abram, T.F. Jaramillo, New insights into the stable operation at each condition) after reaching the certain value of electrochemical reduction of carbon dioxide on metallic copper surfaces, Energy & Environmental Science 5 (2012) 7050–7059. cHCOO− and κS in the catholyte (ca. 70 mM and 21 mS cm− 1, respec­ [6] L. Wu, K.E. Kolmeijer, Y. Zhang, H. An, S. Arnouts, S. Bals, T. Altantzis, J. tively). Therefore, the concentration should be kept below such a value P. Hofmann, M.C. Figueiredo, E.J. Hensen, Stabilization effects in binary colloidal which can be easily achieved by process steering allowing balancing the Cu and Ag nanoparticle electrodes under electrochemical CO 2 reduction conditions, Nanoscale 13 (2021) 4835–4844. rates of eCO2RR and microbial consumption. In addition, the stability of [7] J. Li, M. Zhu, Y.F. Han, Recent advances in electrochemical CO2 reduction on the In catalyst needs to be improved in order to operate long-term stable indium-based catalysts, ChemCatChem 13 (2021) 514–531. eCO2RR. Confirmed by the increase observed in In corrosion rate at [8] K. Mou, Z. Chen, S. Yao, L. Liu, Enhanced electrochemical reduction of carbon dioxide to formate with in-situ grown indium-based catalysts in an aqueous more negative potentials and change in the pH values, H2 generation at electrolyte, Electrochimica Acta 289 (2018) 65–71. the surface could lead to the damage at the surface of the electrode [9] R. Hegner, L.F. Rosa, F. Harnisch, Electrochemical CO2 reduction to formate at coated by the electrocatalyst. Therefore, the durability of the electrode indium electrodes with high efficiency and selectivity in pH neutral electrolytes, Applied Catalysis B: Environmental 238 (2018) 546–556. can be closely related to the electrode potential. One can consider less [10] R. Hegner, K. Neubert, L.F. Rosa, F. Harnisch, Engineering electrochemical CO2 negative potentials to decrease the catalyst leaching but to increase the reduction to formate under bioprocess-compatible conditions to bioreactor scale, HER. However, that would consequently decrease rHCOO− from eCO2RR. ChemElectroChem 6 (2019) 3731–3735. Alternatively, more C1-compound selective electrodes can be explored [11] X. Sun, H.K. Atiyeh, R.L. Huhnke, R.S. Tanner, Syngas fermentation process development for production of biofuels and chemicals: A review, Bioresource and used. Technology Reports 7 (2019), 100279. [12] L.T. Angenent, H. Richter, W. Buckel, C.M. Spirito, K.J. Steinbusch, C.M. Plugge, D. CRediT authorship contribution statement P. Strik, T.I. Grootscholten, C.J. Buisman, H.V. Hamelers, Chain elongation with reactor microbiomes: open-culture biotechnology to produce biochemicals, Environmental science & technology 50 (2016) 2796–2810. Paniz Izadi: Conceptualization, Methodology, Formal analysis, [13] W. de Araújo Cavalcante, R.C. Leitão, T.A. Gehring, L.T. Angenent, S.T. Santaella, Investigation, Visualization, Writing – original draft. Aykut Kas: Formal Anaerobic fermentation for n-caproic acid production: A review, Process Biochemistry 54 (2017) 106–119. analysis, Investigation, Writing – review & editing. Philip Haus: [14] M. Roghair, T. Hoogstad, D.P. Strik, C.M. Plugge, P.H. Timmers, R.A. Weusthuis, Investigation, Formal analysis. Falk Harnisch: Conceptualization, M.E. Bruins, C.J. Buisman, Controlling ethanol use in chain elongation by CO2 Methodology, Formal analysis, Supervision, Funding acquisition, loading rate, Environmental science & technology 52 (2018) 1496–1505. [15] N. Wan, A. Sathish, L. You, Y.J. Tang, Z. Wen, Deciphering Clostridium metabolism Writing – review & editing. and its responses to bioreactor mass transfer during syngas fermentation, Scientific reports 7 (2017) 1–11. Declaration of Competing Interest [16] Q. Fei, M.T. Guarnieri, L. Tao, L.M. Laurens, N. Dowe, P.T. Pienkos, Bioconversion of natural gas to liquid fuel: opportunities and challenges, Biotechnology Advances 32 (2014) 596–614. The authors declare that they have no known competing financial [17] P. Izadi, E. Yu, Realizing Full Potential of Bioelectrochemical and interests or personal relationships that could have appeared to influence Photoelectrochemical Systems, Joule 4 (2020) 2085–2087. [18] P. Izadi, F. Harnisch, Microbial| electrochemical CO2 reduction: To integrate or not the work reported in this paper. to integrate? Joule 6 (2022) 935–940. [19] H. Li, P.H. Opgenorth, D.G. Wernick, S. Rogers, T.-Y. Wu, W. Higashide, P. Malati, Data availability Y.-X. Huo, K.M. Cho, J.C. Liao, Integrated electromicrobial conversion of CO2 to higher alcohols, Science 335 (2012) 1596, 1596. [20] M. Stöckl, S. Harms, I. Dinges, S. Dimitrova, D. Holtmann, From CO2 to Data will be made available on request. Bioplastic–Coupling the Electrochemical CO2 Reduction with a Microbial Product Generation by Drop-in Electrolysis, ChemSusChem 13 (2020) 4086–4093. [21] K.R. Chatzipanagiotou, L. Jourdin, C.J. Buisman, D.P. Strik, J.H. Bitter, CO2 Conversion by Combining a Copper Electrocatalyst and Wild-type Microorganisms, Acknowledgement ChemCatChem 12 (2020) 3900–3912. [22] C. Gimkiewicz, R. Hegner, M.F. Gutensohn, C. Koch, F. Harnisch, Study of electrochemical reduction of CO2 for future use in secondary microbial The authors acknowledge the support of the VIVALDI project that electrochemical technologies, ChemSusChem 10 (2017) 958–967. has received funding from the European Union’s Horizon 2020 research [23] W. Luo, W. Xie, M. Li, J. Zhang, A. Züttel, 3D hierarchical porous indium catalyst and innovation program under grant agreement 101000441. This for highly efficient electroreduction of CO 2, Journal of Materials Chemistry A 7 research is also financed by the German Federal Ministry of Education (2019) 4505–4515. [24] F. Bienen, A. Löwe, J. Hildebrand, S. Hertle, D. Schonvogel, D. Kopljar, N. Wagner, and Research (BMBF) under the GAMES project (Grant nr: 33RC031E). E. Klemm, K.A. Friedrich, Degradation study on tin-and bismuth-based gas- The responsibility for the content lies with the authors. This work was diffusion electrodes during electrochemical CO2 reduction in highly alkaline supported by the Helmholtz-Association in the frame of the Integration media, Journal of Energy Chemistry 62 (2021) 367–376. [25] W. Ma, S. Xie, X.-G. Zhang, F. Sun, J. Kang, Z. Jiang, Q. Zhang, D.-Y. Wu, Y. Wang, Platform “Tapping nature’s potential for sustainable production and a Promoting electrocatalytic CO2 reduction to formate via sulfur-boosting water healthy environment” at the UFZ. activation on indium surfaces, Nature communications 10 (2019) 1–10. [26] J. Barek, A.G. Fogg, A. Muck, J. Zima, Polarography and voltammetry at mercury electrodes, Critical reviews in analytical chemistry 31 (2001) 291–309. Supplementary materials [27] W.M. Haynes, D.R. Lide, T.J. Bruno, CRC handbook of chemistry and physics, CRC press (2016). Supplementary material associated with this article can be found, in [28] R.C. Team, R Core Team R. R, A Language and Environment for Statistical Computing R Foundation for Statistical Computing, Vienna, Austria, 2016. the online version, at doi:10.1016/j.electacta.2023.142733. [29] R.C. Team, R: A language and environment for statistical computing [Computer software manual], Vienna, Austria, 2019. References [30] H. Wickham, W. Chang, M.H. Wickham, Package ‘ggplot2’, Create elegant data visualisations using the grammar of graphics, Version 2 (2016) 1–189. [31] R.A. Tufa, D. Chanda, M. Ma, D. Aili, T.B. Demissie, J. Vaes, Q. Li, S. Liu, D. Pant, [1] O.S. Bushuyev, P. De Luna, C.T. Dinh, L. Tao, G. Saur, J. van de Lagemaat, S. Towards highly efficient electrochemical CO2 reduction: Cell designs, membranes O. Kelley, E.H. Sargent, What should we make with CO2 and how can we make it? and electrocatalysts, Applied Energy 277 (2020), 115557. Joule 2 (2018) 825–832. [2] D. Li, H. Zhang, H. Xiang, S. Rasul, J.-M. Fontmorin, P. Izadi, A. Roldan, S.F. R. Taylor, Y. Feng, L. Banerji, A.J. Cowan, E.H. Yu, J. Xuan, How to go beyond C1 10 P. Izadi et al. Electrochimica Acta 462 (2023) 142733 [32] R.A. Rozendal, H.V. Hamelers, R.J. Molenkamp, C.J. Buisman, Performance of [44] T. Haas, R. Krause, R. Weber, M. Demler, G. Schmid, Technical photosynthesis single chamber biocatalyzed electrolysis with different types of ion exchange involving CO2 electrolysis and fermentation, Nature Catalysis 1 (2018) 32–39. membranes, Water research 41 (2007) 1984–1994. [45] Y. Jännsch, J.J. Leung, M. Hämmerle, E. Magori, K. Wiesner-Fleischer, E. Simon, [33] P. Izadi, J.-M. Fontmorin, A. Godain, H.Y. Eileen, I.M. Head, Parameters M. Fleischer, R. Moos, Pulsed potential electrochemical CO2 reduction for influencing the development of highly conductive and efficient biofilm during enhanced stability and catalyst reactivation of copper electrodes, Electrochemistry microbial electrosynthesis: the importance of applied potential and inorganic Communications 121 (2020), 106861. carbon source, npj Biofilms and Microbiomes 6 (2020) 1–15. [46] K. Schouten, Y. Kwon, C. Van Der Ham, Z. Qin, M. Koper, A new mechanism for the [34] W. Yang, Y. Zhao, S. Chen, W. Ren, X. Chen, C. Jia, Z. Su, Y. Wang, C. Zhao, selectivity to C 1 and C 2 species in the electrochemical reduction of carbon dioxide Defective Indium/Indium Oxide Heterostructures for Highly Selective Carbon on copper electrodes, Chemical Science 2 (2011) 1902–1909. Dioxide Electrocatalysis, Inorganic Chemistry 59 (2020) 12437–12444. [47] Y. Hori, H. Konishi, T. Futamura, A. Murata, O. Koga, H. Sakurai, K. Oguma, [35] M.C.O. Monteiro, A. Goyal, P. Moerland, M.T.M. Koper, Understanding Cation Deactivation of copper electrode” in electrochemical reduction of CO2, Trends for Hydrogen Evolution on Platinum and Gold Electrodes in Alkaline Media, Electrochimica acta 50 (2005) 5354–5369. ACS Catalysis 11 (2021) 14328–14335. [48] B. Jermann, J. Augustynski, Long-term activation of the copper cathode in the [36] A. Goyal, G. Marcandalli, V.A. Mints, M.T.M. Koper, Competition between CO2 course of CO2 reduction, Electrochimica acta 39 (1994) 1891–1896. Reduction and Hydrogen Evolution on a Gold Electrode under Well-Defined Mass [49] A. Engelbrecht, C. Uhlig, O. Stark, M. Hämmerle, G. Schmid, E. Magori, K. Wiesner- Transport Conditions, Journal of the American Chemical Society 142 (2020) Fleischer, M. Fleischer, R. Moos, On the electrochemical CO2 reduction at copper 4154–4161. sheet electrodes with enhanced long-term stability by pulsed electrolysis, Journal [37] J. Ho, M.L. Coote, C.J. Cramer, D.G. Truhlar, 6 Theoretical Calculation of of The Electrochemical Society 165 (2018) J3059. Reduction Potentials, (2015). [50] S. Rasul, D.H. Anjum, A. Jedidi, Y. Minenkov, L. Cavallo, K. Takanabe, A highly [38] J.G. Speight, Lange’s handbook of chemistry, McGraw-Hill Education, 2017. selective copper–indium bimetallic electrocatalyst for the electrochemical [39] M. Ramdin, A.R. Morrison, M. De Groen, R. Van Haperen, R. De Kler, L.J. Van Den reduction of aqueous CO2 to CO, Angewandte Chemie 127 (2015) 2174–2178. Broeke, J.M. Trusler, W. De Jong, T.J. Vlugt, High pressure electrochemical [51] W. Luo, W. Xie, R. Mutschler, E. Oveisi, G.L. De Gregorio, R. Buonsanti, A. Züttel, reduction of CO2 to formic acid/formate: a comparison between bipolar Selective and stable electroreduction of CO2 to CO at the copper/indium interface, membranes and cation exchange membranes, Industrial & engineering chemistry Acs Catalysis 8 (2018) 6571–6581. research 58 (2019) 1834–1847. [52] G.O. Larrazábal, T. Shinagawa, A.J. Martín, J. Pérez-Ramírez, Microfabricated [40] F. Harnisch, U. Schröder, Selectivity versus mobility: separation of anode and electrodes unravel the role of interfaces in multicomponent copper-based CO2 cathode in microbial bioelectrochemical systems, ChemSusChem: Chemistry & reduction catalysts, Nature Communications 9 (2018) 1477. Sustainability Energy & Materials 2 (2009) 921–926. [53] G. He, H. Tang, H. Wang, Z. Bian, Highly selective and active Pd-In/three- [41] B. Pan, Y. Wang, Y. Li, Understanding and leveraging the effect of cations in the dimensional graphene with special structure for electroreduction CO2 to formate, electrical double layer for electrochemical CO2 reduction, Chem Catalysis (2022). Electroanalysis 30 (2018) 84–93. [42] H. Strathmann, A. Grabowski, G. Eigenberger, Ion-exchange membranes in the [54] M. Stöckl, N. Claassens, S. Lindner, E. Klemm, D. Holtmann, Coupling chemical process industry, Industrial & Engineering Chemistry Research 52 (2013) electrochemical CO2 reduction to microbial product generation–identification of 10364–10379. the gaps and opportunities, Current Opinion in Biotechnology 74 (2022) 154–163. [43] D.A. Salvatore, D.M. Weekes, J. He, K.E. Dettelbach, Y.C. Li, T.E. Mallouk, C. [55] Y. Tashiro, S. Hirano, M.M. Matson, S. Atsumi, A. Kondo, Electrical-biological P. Berlinguette, Electrolysis of Gaseous CO2 to CO in a Flow Cell with a Bipolar hybrid system for CO2 reduction, Metabolic engineering 47 (2018) 211–218. Membrane, ACS Energy Letters 3 (2017) 149–154. 11