Assessing the Electrochemical CO2 Reduction Reaction Performance Requires More Than Reporting Coulombic Efficiency

Reporting coulombic efficiency ( CE$\text{CE}$ ) is the common way to assess the performance of electrochemical carbon dioxide (CO2) reduction reaction (eCO2RR) in literature, whereas its carbon conversion efficiency ( CCE$\text{CCE}$ ) is frequently neglected. Herein, the importance of reporting both efficiencies when evaluating the eCO2RR is discussed, using Sn‐based gas diffusion electrodes (GDEs) as model electrodes. It is shown that CCE$\text{CCE}$ can vary remarkably at a constant CE$\text{CE}$ with minor operational changes. Over 120 min experiments with operational conditions being representative of numerous previous studies, the CCE$\text{CCE}$ is increased from ≈20% to 41% (being only 9% below the theoretical maximum). This was achieved by simply adjusting the inlet CO2 flow rate from ≈35 to 16 mL min−1, while CE$\text{CE}$ was identical at both CO2 flow rates (≈85%, 7%, and 4% for production of formate/formic acid, CO, and H2, respectively at both conditions). Thus, it is advocated that reporting of both efficiencies, for electrons and carbon, is required for meaningfully assessing the performance of an eCO2RR system.


Introduction
Electrochemical conversion of carbon dioxide (CO 2 ) to valueadded fuels and chemicals has gained broad attention as a plausible route to effectively store intermittent renewable electric energy as well as capture CO 2 .In the past decades, research on the electrochemical CO 2 reduction reaction (eCO 2 RR) in aqueous solutions has focused overwhelmingly on improving the current or electron efficiency that is the coulombic efficiency (CE) sometimes also denominated as Faradaic efficiency (FE).CE was enhanced, for instance by 1) tuning the catalyst composition and morphology, [1] 2) incorporating the electrocatalyst in gas diffusion electrodes (GDE) to overcome mass transfer limits, [2] and 3) optimizing the pH and composition of the electrolyte solution and cell design (e.g., flow cell instead of H-cell). [3]Various electrocatalysts, showing promising CE, have been discovered for eCO 2 RR to C 1 and C 2 products, such as formate/formic acid, carbon monoxide (CO), methane (CH 4 ), or ethylene (C 2 H 4 ). [4]imultaneously, eCO 2 RR is still associated with low technology readiness levels (TRL) between 2 and 3. Factors limiting the economic feasibility of the eCO 2 RR include limited energy efficiency, low carbon conversion efficiency (CCE) and overall short operational time (low stability).The electric energy consumption of eCO 2 RR is mainly related to 1) thermodynamics of the reaction, i.e., its formal potential, 2) overpotentials of the half-cell reaction, and 3) i Â R drop across the membrane and within the electrolyte solutions in the electrochemical cell.Thus, depending on, for instance, the electrolyte type and concentration, pH and membrane, the half-cell potential during eCO 2 RR to formate (HCOO À ) and thus cell potential varies (Table 1).Using alkaline electrolytes for eCO 2 RR to HCOO À appears to be optimal for achieving a low cell potential.However, the high alkalinity in the catholyte will inevitably lead to formation of K 2 CO 3 and eventually KHCO 3 , causing a waste of both, alkali hydroxide such as KOH and CO 2 .
To achieve an efficient eCO 2 RR, CO 2 needs to be collected and concentrated as the reactant.Most relevant is direct air capture, which utilizes alkali hydroxide system and thermal swing to release concentrated CO 2 being a process with typical cycles up to 900 °C.Hence, it requires significant energy input [5] and thus operational expenditures for providing CO 2 for eCO 2 RR.Over the past decades, researchers generally have supplied excess CO 2 to the eCO 2 RR system under study, aiming to obtain the maximum CE: However, most of the supplied CO 2 is not electrochemically converted using this approach, and leaves the electrochemical cell unreacted.So far, in the great majority of studies, CE was used as main, often sole criterion to assess the efficiency of the eCO 2 RR, whereas CCE was paid little to DOI: 10.1002/aesr.202400031 Reporting coulombic efficiency (CE) is the common way to assess the performance of electrochemical carbon dioxide (CO 2 ) reduction reaction (eCO 2 RR) in literature, whereas its carbon conversion efficiency (CCE) is frequently neglected.Herein, the importance of reporting both efficiencies when evaluating the eCO 2 RR is discussed, using Sn-based gas diffusion electrodes (GDEs) as model electrodes.It is shown that CCE can vary remarkably at a constant CE with minor operational changes.Over 120 min experiments with operational conditions being representative of numerous previous studies, the CCE is increased from ≈20% to 41% (being only 9% below the theoretical maximum).This was achieved by simply adjusting the inlet CO 2 flow rate from ≈35 to 16 mL min À1 , while CE was identical at both CO 2 flow rates (≈85%, 7%, and 4% for production of formate/formic acid, CO, and H 2 , respectively at both conditions).Thus, it is advocated that reporting of both efficiencies, for electrons and carbon, is required for meaningfully assessing the performance of an eCO 2 RR system.no attention to.A small number of studies, for instance Ma et al. [6] discussed the importance of the carbon balance in eCO 2 RR using GDE reactors, and both CE and CCE values were considered for assessing eCO 2 RR performance.So far, the HCOO À selective catalysts such as Sn or indium (In) showed high CE (>80%) for eCO 2 RR, [7] while the value of their CCE was rather low.For instance, previous studies reported CCE of only ≈30% and ≈36% using copper (Cu) [6] and tin sulfide (SnS) [8] based electrodes, respectively.These low values of CCE could presumably be one of the reasons that eCO 2 RR has not yet reached a higher TRL or even industrial applications.Recently, efforts have been made to improve the CCE by employing a highly acidic catholyte. [9]Nevertheless, strong acidic conditions may not favor the stability of the catalysts such as tin. [10]ere, we show that CCE needs to be considered to the same extent as CE for reporting the efficiency of any eCO 2 RR systems.Therefore, the overall components of an electrochemical set-up (such as electrolyte salt concentration and membrane type) as well as the operational conditions are crucial to the CCE.
Under alkaline or neutral conditions: Under acidic conditions: Considering only HCOO À as a sole carbon product of eCO 2 RR, the carbon balance for two different types of membrane (cation exchange membrane (CEM), anion exchange membrane (AEM)) and acidic or alkaline anolyte solution are different.This is due to the different ions that are transported between anolyte and catholyte for assuring charge balancing (Figure 1).
Based on the cathode reaction in the KHCO 3 catholyte (Equation (1)), HCOO À formation via eCO 2 RR requires one molecule of CO 2 , and every HCOO À produced via eCO 2 RR will produce one OH À anion, which will react with one molecule of CO 2 available in the cathode compartment, forming a HCO 3 À anion in the catholyte (Equation (2)).Thus, every two electrons transferred consume two molecules of CO 2 (one through Table 1.List of contributors to the cell voltage for an eCO 2 RR system operating with alkaline, neutral, and acidic solutions, while using cation exchange membrane (CEM) or anion exchange membrane (AEM).The contributors listed here include reactions thermodynamics of cathode and anode (E °cathode , E °anode ), overpotentials (η cathode , η anode ), and i Â R drops across the electrolyte solutions and membrane.The cathodic potential (E °cathode þ η cathode ) included here is the least negative potential reported (À1.3 V vs. SHE [14] ) for eCO 2 RR at 100 mA cm À2 .The anode overpotential is taken from the overpotential for oxygen evolution reaction (OER) with commercial RuO 2 /IrO 2 (acidic anolyte) and Ni(Fe)OOH (alkaline anolyte) at 100 mA cm À2 .The i Â R drops are calculated based on the conductivity of membranes and a 5 mm thick layer of solution (conductivity used here: 1 M KOH: 201.3 mS cm À1 , 1 M KHCO 3 : 96.0 mS cm À1 , 0.5 M H 2 SO 4 : 223.0 mS cm À1 ). [15]The i Â R drop across the membrane are calculated based on the area specific conductivity of the membrane soaked in the corresponding electrolyte solution and the thickness of the membrane after swelling (see also Table S3 and S4, Supporting Information).η anode 0.350 [16] 0.390 [17] 0.350 0.390 0.350 [16] 0.390 [17] i Â R drop Membrane 0.213 [18] 0.057 [19] 0.025 0.057 0.213 0.357 [20] 0 Equation (1) and one through Equation ( 2)), or in other words by generation of every molecule of HCOO À via eCO 2 RR, two molecules of CO 2 are consumed.This is most often not considered in literature.When a CEM is used in an electrochemical cell, two cations (K + or H + , depending on the pH of the anolyte), are transferred from anolyte to catholyte to maintain charge neutrality.Thus, the formed HCO 3 À either yields a KHCO 3 that may precipitate due to its low solubility, or forms CO 2 after acidification of HCO 3 À by the proton permeated from anolyte through the CEM (Figure 1a,b).This CO 2 may escape the catholyte via the tail gas, without taking part in the eCO 2 RR.In all these cases, a theoretical maximum CCE of only 50% can be reached.When using an AEM, the theoretical maximum CCE is even lower.The HCO 3 À formed during eCO 2 RR (Equation (2)) is transferred from catholyte to anolyte and will react either with OH À to form CO 3 2À , or with H + to release the CO 2 to the headspace (Figure 1c,d).In acidic anolytes two HCO 3 À will react with two H + generated from the anodic reaction.Therefore, only one out of three molecules of CO 2 that are needed for the reaction is electrochemically converted in eCO 2 RR, leading to a maximum theoretical CCE of ≈33%.These basic considerations already show that CCE can vary in different setups and operational conditions and, hence, it needs to be included, together with CE when reporting the efficiency for eCO 2 RR.

Results and Discussions
In this study, efficiencies of eCO 2 RR are evaluated using a setup and operating conditions being representative of previous studies (Table 2).In brief, the eCO 2 RR on a Sn-GDE was studied (details in the Experimental Section in Supporting Information), as Sn is one of the most promising catalysts for eCO 2 RR to formate. [11]For this, an inlet CO 2 flow of 35 AE 1 mL min À1 was applied for 120 min, while the outlet gas flow rate and composition were measured (Figure S1, Supporting Information).This value of the CO 2 flow was selected since it was almost the minimum flow rate used in previous studies (Table 2).Since the reactor was used in flow through mode by closing the outlet of the gas compartment, the overall carbon balance of the system could be determined.The effect of KHCO 3 concentration in the catholyte on the eCO 2 RR efficiencies was examined by using various diluted KHCO 3 solutions, i.e., 0.5, 0.05, and 0.005 M KHCO 3 , since the equilibrium of CO 2 /HCO 3 À shifts at different KHCO 3 concentration, leading to change in the catholyte pH.
Before starting the eCO 2 RR, all catholytes were purged with only CO 2 until reaching a stable pH, i.e., 0.5 M KHCO 3 to ≈7.5, 0.05 M KHCO 3 to ≈6.5 and 0.005 M KHCO 3 to ≈5.7 (Table S1, Supporting Information) and conductivity was adjusted in all experiments to ≈38 mS cm À1 by adding the electrochemically inert electrolyte K 2 SO 4 (Table S1, Supporting Information).Although the pH of the catholytes at the beginning and throughout the experiments differed from each other, the selectivity of eCO 2 RR to HCOO À was not affected.The CE for HCOO À production (CE HCOO À ) (Equation (1) and ( 5)) in all conditions were similar at more than 85% and with a similar HCOO À production rate (r HCOO À ) of ≈32 mM h À1 (Figure 2a).CO from reduction of CO 2 (Equation (3) and ( 6)) and hydrogen (H 2 ) from the hydrogen evolution reaction (HER, Equation ( 4) and ( 7)) were the other products, corresponding to an individual CE of ≈7% and 4%, respectively.It is worth mentioning that leaching of Sn from the GDE to the solution at different catholyte pH from the beginning (Table S1, Supporting Information) and throughout the experiment (Figure S3, Supporting Information) was below the limit of detection.Despite the high CE at all conditions, being in-line with previous studies, the CCE in these experiments was low.At all conditions, the CCE was only ≈20% (Figure 2b-d).This showed that 80% of the carbon supplied to the system in the form of CO 2 was not electrochemically converted.In contrast, the carbon was either partially (≈60%) released via the off-gas to the atmosphere (Figure 2b-d), or fixed chemically as bicarbonate in the solution (e.g., Equation (2)).The almost identical CCE in all conditions with different KHCO 3 concentrations, and hence different catholyte pH, showed that release of CO 2 in form of gas or bicarbonate was inevitable and different pH provided did not affect the CCE.
For enhancing the CCE, the amount of CO 2 supply can be limited.The minimum amount needed can be calculated when considering the (fixed) current and the achieved CE.Thereby, the CO 2 required by eCO 2 RR (Equation (1)) and further inevitable chemical conversion of CO 2 need to be considered.First, when assuming the CE = 100% at a current of 1 A, 7.5 mL min À1 of CO 2 is required at 22 °C for eCO 2 RR to HCOO À .At the same time, the generation of OH À under alkaline conditions (Equation (1)-( 4)), or H + consumption under acidic conditions (Equation ( 5)-( 7)) that cause the chemical conversion of gaseous CO 2 to carbonate or bicarbonate in the liquid phase need to be considered.When now considering a minimum CE HCOO À of 80% (according to the experimental results, Figure 2a) as well as carbonate or bicarbonate generation in the catholyte, a CO 2 flow rate of almost 15 mL min À1 is theoretically minimally required for eCO 2 RR to HCOO À .In practice, the CO 2 flow rate of minimum 16 AE 1 mL min À1 was also sufficient for eCO 2 RR to HCOO À in our setup at a constant CE.In addition, this flow rate value was also confirmed by trying eCO 2 RR at lower CO 2 flow rate than 15 mL min À1 in our setup.When the flow rate of 13 AE 1 mL min À1 was tried, an increase in H 2 evolution was observed with CE (>20%) for HER (e.g., Equation ( 4) and ( 7)), which was also discussed previously. [12]Our calculations were considering a constant CE HCOO À of 80%, which of course, could vary and hence lead to minimal changes in the required CO 2 flow.
Thus, the CO 2 flow rate was further adjusted to constant 16 AE 1 mL min À1 for eCO 2 RR.Since CE, CCE, and r HCOO À were similar in all the catholytes with different KHCO 3 concentration (0.5, 0.05, and 0.005 M) as shown before (Figure 2 -the respective CE of ≈85%, 7%, and 4% for HCOO À , CO and H 2 production, CCE of ≈20%, and r HCOO À of ≈32 mM h À1 ), the most diluted catholyte (0.005 M KHCO 3 ) was used for studying the effect of the decreased CO 2 flow rate.Interestingly, CE of the products and r HCOO À were still similar to those when a higher CO 2 flow rate of 35 AE 1 mL min À1 was used (Figure 3a).However, CCE increased by factor two compared to the higher flow rate (35 AE 1 mL min À1 ), and reached ≈41%, which is already only 9% below the theoretical maximum (Figure 3b).This confirms that using a high flow rate during eCO 2 RR, as performed in many previous studies (Table 2), is not necessary for achieving a high CE, but it is diminishing the CCE.At the same time, it needs to be considered that CO 2 flow rate needs to be adjusted at a value that covers the minimum requirement of CO 2 for eCO 2 RR, as well as the inevitable chemical conversion of CO 2 in the liquid phase.In addition, even when using the CO 2 flow rate of 16 AE 1 mL min À1 , gaseous CO 2 , although in smaller shares than before, was still released to the headspace (≈18% at 16 AE 1 mL min À1 compared to ≈60% at 35 AE 1 mL min À1 during 120 min experiments, Figure 2 and 3).Hence, one may never achieve the theoretical maxima of 50% CCE at the same time with 100% CE, especially not at an acceptable rate.If electrochemical conversion of the CO 2 in industrial flue gas is targeted, CCE is considered as a crucial factor.Also, here, CO 2 dissolving in the solution is governed by thermodynamic principles and hence only a theoretical maximum CCE of 50% can be achieved (Figure 1).At the same time, the release of gaseous CO 2 from the reactors to the atmosphere can be technically circumvented, for instance, by adding a CO 2 recycling line to the gas compartment of the reactor. [13]In essence, we strongly advocate to report the CCE side by side to the CE when assessing the eCO 2 RR, as only one of these two efficiencies tells less than half of the story.

Figure 1 .
Figure1.Schematics of carbon balance for the eCO 2 RR in a KHCO 3 catholyte to formate (HCOO À ) using cation (CEM) or anion (AEM) exchange membranes with combination of acidic (dark gray) and alkaline (light gray) anolyte.a) When CEM and alkaline anolyte (KOH) are used, the ionic current is conducted through the membrane via K + , which leads to the formation of KHCO 3 and HCOOK in the catholyte, b) the application of CEM and acidic anolyte leads to the transport of ionic current via H + through the membrane, which produces CO 2 and HCOOK.If AEM are used, the species transporting the charge balancing ionic current across membrane is HCO 3 À .In this case, the increase of HCOO À concentration in the catholyte is accompanied with the decline of HCO 3 À concentration.Depending on the anolyte used, the HCO 3 À transported from catholyte to the anolyte can either be c) alkalified by OH À to yield CO 3 2À or d) acidified by H + to become CO 2 that is exhausted.

Figure 3 .
Figure3.a) Coulombic efficiency (CE) of the eCO 2 RR at Sn-GDE (10 cm 2 ) reactors and the formate (HCOO À ) production rate (r HCOO À ) using the catholyte of 0.005 M KHCO 3 with the inlet CO 2 flow rate of 35 AE 1 mL min À1 and 16 AE 1 mL min À1 .b) Carbon balance considering inlet and outlet CO 2 , total inorganic carbon (TIC) and products (CO and HCOO À ) at the CO 2 flow rate of 35 AE 1 mL min À1 and 16 AE 1 mL min À1 .All experiments were conducted for 120 min in triplicate (n = 3).

Table 2 .
Experimental conditions and results from previous studies on eCO 2 RR using gas diffusion electrode (GDE) reactors.