Review—CO2 Separation and Transport via Electrochemical Methods

This review focuses on research advancements in electrochemical methods of CO2 separation as part of the broader field of CO2 capture. Such methods are a potentially effective way of separating CO2 from dilute gas mixtures (e.g., flue gas, air) such that it can be sequestered or recycled for other purposes. Electrodialysis using a liquid electrolyte capture solution is the most thoroughly explored electrochemical approach for CO2 capture. The purpose of this review is to provide a broad overview of developments in the field, highlighting and harmonizing relevant figures of merit such as specific energy consumption and faradaic efficiency. In addition, the use of alkaline membranes is separately surveyed as a promising means of electrochemical CO2 separation, as their CO2 transport phenomena are well understood within the context of alkaline fuel cells or electrochemical CO2 reduction. Recent materials advancements enable the use and modification of these membranes to promote electromigration of (bi)-carbonate ions, the result being CO2 concentration on the anode side of an electrochemical cell. © 2020 The Author(s). Published on behalf of The Electrochemical Society by IOP Publishing Limited. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/ 1945-7111/abbbb9]

Persistent anthropogenic greenhouse gas emissions, particularly emissions of CO 2 , necessitate the development of novel mitigation solutions. Global anthropogenic fossil CO 2 emissions exceeded a record estimated 37 Gt in 2017. 1 To limit excessive CO 2 emissions and to minimize impacts on the climate, a range of technologies including a transition from fossil fuel-based to renewable energy, improvement of vehicle and building energetic efficiencies, and CO 2 capture and sequestration must be implemented together. 2 CO 2 can be captured from point sources (e.g., coal-fired power plants) or atmospheric air, the latter theoretically allowing for negative CO 2 emissions. 3 Although various demonstration projects have come online in recent years, the only CO 2 capturing technique that has been utilized at an industrial scale is amine scrubbing. 4 This is a post-combustion process that uses amine-based solvents (e.g., monoethanolamine) that absorb the carbon dioxide contained in the flue gas. The CO 2 is then removed from the CO 2 -containing solvent by means of a regeneration process, driven by a temperature or pressure swing. The solvent is recycled and the captured CO 2 is treated for transport and subsequent storage. The main drawbacks of this technology are related to the corrosive nature of the solvent and to the high cost (60-107 USD/ton of CO 2 captured). [5][6][7] In this context, considerable research efforts are needed to develop more efficient and, at the same time, less costly CO 2 capture technologies. Alternative to amine scrubbing are the capture of CO 2 using solid sorbents such as alkaline earth metal oxides, 8 layered double oxides, 9 carbon, 10 or metal organic frameworks 11 but also electrochemical methods. Using solid sorbent-like metal oxides (e.g., CaO) or carbon present the advantage of using cheap abundant materials, less corrosive and potentially more selective for CO 2 capture. Electrochemical gas separation was successfully used for H 2 separation from gas mixture containing CH 4 , 12 reformate gas, 13 and N 2 . 14 The method consists of selectively oxidizing (or reducing) a gas at one electrode to an ion which is further transported through a membrane or liquid electrolyte to the other electrode where is reduced (or oxidized) back to the gas. The advantages of this technology is that the separation occurs at low temperature and pressure and that, in principle, the energy requirements are low. 15 Moreover, electrochemical methods for CO 2 capture can be performed at atmospheric concentration levels. 16 While electrochemical CO 2 separation methods have been investigated for almost half a century, recent years have brought a surge of interest and encouraging developments. This review is divided into two primary parts; the first focuses on electrochemical processes for CO 2 separation. Although this review is agnostic in the source or purpose of captured CO 2 , a comparison of direct-air capture (DAC) and point-source capture is provided. The second part deals with CO 2 transport in anion exchange membranes (AEM) and proton exchange membranes (PEM) as an ancillary process in fuel cell or co-electrolyzer operation.

Electrochemical CO 2 Separation
Electrochemical CO 2 separation is herein defined as any process by which CO 2 is selectively removed from a gas mixture using electrochemical reactions as the driving mechanism. This is most commonly performed using a liquid electrolyte as the medium for absorbing and releasing CO 2 in a two-step process, as shown in Fig. 1a, with ion-conducting membranes used to regenerate the electrolyte. [17][18][19][20][21][22][23][24][25] Alternatively, the capture and transport of CO 2 can occur in a static environment (e.g., a polymer electrolyte membrane electrochemical cell) in which the CO 2 is delivered as a gas to the feed side and evolved from the permeate side, 15,16,[26][27][28] as shown in Fig. 1b. Depending on the configuration of gases, the cell operates either in a fuel cell mode, which generates electricity, or a driven mode, which consumes electricity.
A summary of notable experimental studies from the past 20 years is presented in Table I. The key performance indicators, faradaic efficiency and molar specific energy consumption, are given for the selected current density value, although they are not necessarily the only values reported for a given study. Faradic efficiency describes the efficiency with which electrons release CO 2 molecules, and is typically defined as the total number of CO 2 molecules transferred over the total number of electrons transferred. In selecting values for the table, higher values of current density were favored, as current density is a critical factor in system scalability. Increasing current density tends to have a negative effect on specific energy consumption and a mixed effect on faradaic efficiency. Table I shows that a wide range of current densities have been reported, varying between 0.5-139 mA cm −2 , and that the corresponding performance indicators have varied widely as well. As the energy consumption calculation generally only includes the electrical input to the cell or stack, it can be assumed that the values given are conservative for a scaled up system. In future electrochemical CO 2 separation studies, efforts should be made to align the reporting of performance metrics, with both faradaic efficiency and specific energy consumption clearly stated or shown at a relevant testing conditions. Variations in the faradaic efficiency and specific energy consumption values shown in Table I primarily arise from differences in the techniques and cell configurations applied. For all techniques shown in the table, faradaic efficiency losses can come from unwanted electrode reactions or unwanted transport of ions. For electrodialysis, this can occur through electrolysis of water at the electrodes 19,21 or transport of protons through the CEMs or hydroxide through the AEMs. 19 Other loss mechanisms that have been proposed are leakage of protons through the AEM, 17,18 loss of BPM perm-selectivity, 17 and unwanted water electrolysis in AEMs. 17 For gas-fed membrane electrochemical cells, the transport of hydroxide instead of (bi)-carbonate ions leads to faradaic losses. Low efficiencies for low concentrations of CO 2 in the feed gas exemplify this issue. 16 Cells operating at higher voltages (>1.23 V) also have the potential to electrolyze water and evolve O 2 and H 2 . 15,27,30 Specific energy consumption, which incorporates voltage across the cell or stack, is negatively affected by activation, ohmic, or mass transport overpotentials. These can arise from slow reaction kinetics at the electrodes, membrane/electrolyte materials with poor conductivities, poor interfacial contact, or insufficient flows of reactants and/or products to or from reaction interfaces. Thus, design aspects such as membrane spacing or electrolyte concentration in electrodialysis can have significant effects. 17,19,21 There is no universal solution to optimize both faradaic efficiency and specific energy consumption of an electrochemical CO 2 separation cell or stack, and the design of such a system requires careful consideration of the process inputs and constraints.
Electrodialytical regeneration.-In a two-step CO 2 capture and electrodialytical regeneration system, an electrolyte solution absorbs CO 2 at near-ambient temperatures and pressures before it is pumped to an electrodialysis unit where the CO 2 is removed and the solution is regenerated. Stacks of bipolar membranes (BPM), cation exchange membranes (CEM), and/or AEMs are arranged to create alternating acidic and alkaline compartments through a pH swing, allowing for release of CO 2 and regeneration of the solution, respectively. Gaseous CO 2 is formed by decreasing the pH of the carbonate solution. This is in contrast to the energy-intensive thermal regeneration system of traditional CO 2 capture processes. 31 Various architectures of electrodialytical stacks found in literature are shown in Fig. 2. BPMs can be incorporated to split water into hydroxide ions and protons, but they are not necessary for an electrodialytical separation process (as shown in Fig. 2d

[ ]
In this way, CO 2 is typically released from the CO 2 recovery compartment or downstream gas-liquid separator as a pure gas, meaning concentration factors of 2500 are obtainable for DAC and subsequent electrodialytical regeneration. While an electrochemically-driven regeneration process has theoretically low energy requirements, there is a significant gap between the experimental values reported in literature and the ideal energy requirement for recovery of CO 2 from alkaline carbonate solution, which has been attributed mostly to reactions in the electrodes and ohmic losses in the cell or stack. 19 Most electrodialysis studies use commercially available membrane materials from suppliers like DuPont, 29,32 Ameridia (Eurodia), 17,18,20 Astom, 21,29,33 Tokuyama, 19 or Asahi Glass. 19,21,33 Liquid electrolyte pumping allows for transport of the ions to the electrodes at a relatively high concentration, in turn allowing for higher current densities to be reached and at higher faradaic efficiencies as compared to direct gas phase separation of a dilute CO 2 mixture (e.g., air). This comes at expense of introducing a parasitic energy loss to the system in the form of pump work, additional energy costs associated with the capture step, and technical challenges associated with using a caustic, volatile solution. 34 Inclusion of these costs is largely ignored in electrodialysis studies, arguing that the energy and equipment cost contributions for the absorption step are negligible by comparison. 21 This, however, depends strongly on the concentration of CO 2 being separated and the assumptions of component efficiency. Operational energy estimates for an alkali solution air contactor have ranged from 13 kJ mol −1 CO 2 for fan and pump energy in a pilot plant 35 up to approximately 100 kJ mol −1 CO 2 depending on column design. 36 At the prototype scale, values of 190-390 kJ mol −1 have been obtained for a spray-based air contactor. 34 Past electrodialytical regeneration studies have primarily focused on potassium 17,18,24,32,33,37 or sodium [19][20][21] (bi)-carbonate as the CO 2 capture medium, implying that the CO 2 capture step takes place using potassium hydroxide or sodium hydroxide. One study proposing the use of or amino acid salts 25 as an electrolyte is a variation of this approach. Although sodium and potassium hydroxide are chemically similar, previous experimental work has determined that potassium hydroxide captures more CO 2 and has a higher absorption efficiency than sodium hydroxide in both flue gas 38 and air 39 contactors.
Potassium (bi)-carbonate.-A few early studies in the 1990s proposed electrodialytical CO 2 capture from the atmosphere for the application of renewable fuel production. 32   performed experiments using a potassium hydroxide solution in a packed column, in which sulphuric acid was used to recover the CO 2 , and the solution was subsequently treated using BPMs and CEMs (see Fig. 2a). Total energy demand for such a process of removing CO 2 from air was estimated to be 427 kJ mol −1 CO 2 . Stucki et al. 32 developed a similar method of atmospheric CO 2 capture using a potassium hydroxide solution in a microporous membrane and subsequent Nafion-based electrodialysis membrane stack. The system also generated O 2 and H 2 through the electrolysis of water, with H 2 exiting in the CO 2 stream. Notable findings included that average cell voltage of the system for a constant current density of 100 mA cm −2 had a linear dependence on potassium bicarbonate anolyte concentration for concentrations above 0.2 mol l −1 , and that efficiencies for the regeneration step generally favored more open electrode structures due to enhanced mass transfer. Eisaman et al. performed BPM electrodialysis studies to investigate the ability of aqueous (bi)-carbonate solutions to separate CO 2 at near-atmospheric 17 and elevated pressures. 18 The system employed AEMs to transport (bi)-carbonate from the basic chamber to the acidic chamber and BPMs to split water and provide protons (see Fig. 2b). Various mixtures of potassium (bi)-carbonate and potassium hydroxide solution were evaluated in terms of CO 2 gas generation rate, efficiency, voltage, and energy consumption. 17 Results indicated an energy consumption as low as 100 kJ mol −1 CO 2 and 200 kJ mol −1 CO 2 to regenerate captured CO 2 from aqueous bicarbonate and carbonate solutions, respectively. Current densities in the study were limited to 100 mA cm −2 due to difficulties reaching steady state at higher values. Authors performed a follow-up study with a BPM electrodialysis system working at pressures of 10 atm. 18 As a result, they were able to achieve higher current densities of 139 mA cm −2 and reductions in energy consumption of 29%. They also determined that the improvement in performance outweighs the additional energy costs from pressurization.
Taniguchi et al. 33 proposed a similar method of electrodialytic CO 2 capture using CEMs and AEMs to treat a potassium bicarbonate solution (see Fig. 2d), but instead proposed vacuum desorption for the CO 2 recovery step, resulting in theoretically lower energy requirements compared to thermal regeneration. Preliminary thermodynamic calculations were performed showing a potential minimum work of 83 kJ mol −1 CO 2 at a pressure of 6 mbar using commercial membranes. This value, authors claim, could be pushed even lower through the use of membranes with improved ion conductivity, but no bench-scale experimental work was performed.
Wang et al. 24 investigated a membrane electrodialysis and electrolysis setup with a single CEM and AEM to capture atmospheric CO 2 . Potassium hydroxide was electrochemically generated from O 2 and H 2 O reduction to form hydroxide ions at the cathode (see Fig. 3a). CO 2 was then recovered from the solution by reacting the potassium (bi)-carbonate with electrochemically generated sulfuric acid from protons formed at the anode through the oxygen evolution reaction. Using current densities of 2.0 mA cm −2 and cell Sodium (bi)-carbonate.-The use of BPM electrodialysis for CO 2 capture from flue gas was explored by Nagasawa et al. 19 with three membrane configurations of BPM and CEM (see Fig. 2a); BPM and AEM (see Fig. 2b); and BPM, CEM, and AEM (see Fig. 2c). Units were arranged in stacks of ten, and sodium bicarbonate and sodium chloride solutions were used in the feed and recovery compartments, respectively. Under a constant current density of 17 mA cm −2 , faradaic efficiencies of 40%-50% were observed for the two configurations involving CEMs and a value 30% was observed for the BPM-AEM configuration. Leakage of protons through the CEM and hydroxide ions through the AEM were determined to be significant detrimental factors for efficiency. The BPM-CEM configuration had the best performance in terms of energy requirement per CO 2 recovered. Other findings were that specific energy requirement was reduced with more units in a stack and less distance between membranes. At reduced current densities of between 2.4-9.5 mA cm −2 , higher current densities were associated with higher specific energy requirements and higher CO 2 recovery rates. As can be observed in Table I, the study achieved one of the lowest minimum energy requirements of 92 kJ mol −1 CO 2 at a low current density of 2.4 mA cm −2 .
Further development of the BPM electrodialysis approach was presented in a study by Iizuka et al., 21 which examined the effects of numerous factors on specific power consumption and faradaic efficiency under steady-state conditions. This was in the same membrane arrangement as shown in Fig. 2a. In general, power consumption was favored by higher sodium concentrations in the feed solution, lower current densities, and a higher number of cells. Faradaic efficiency was favored by higher sodium concentrations in the feed solution, higher current densities, and higher flow rates. Both power consumption and faradaic efficiency were favored by higher extents of CO 2 absorption and higher extents of CO 2 recovery. A cost analysis determined electricity cost and membrane cost to dominate the economics of the overall process.
Datta et al. 20 developed an electrochemical pH-swing process for CO 2 capture from flue gas, employing resin-wafer electrodeionization and a process liquid of monosodium dihydrogen phosphate and disodium hydrogen phosphate. The system contained alternating CEM and BPM (see Fig. 2a) in a buffer solution to form diluate and concentrate chambers. This system distinguishes itself from other electrodialysis approaches by including ion-exchange resin beads (resin wafers) in the diluate chambers to promote ion transfer and improve pH control. Up to 80% CO 2 was captured with purity exceeding 98% from a 15% CO 2 in N 2 inlet gas. Enhancement of reaction kinetics was a key challenge in this method of CO 2 capture.
A novel membrane electrodialysis and electrolysis system was investigated by Mehmood et al., 29 which examined both a CEM-AEM and a CEM-CEM arrangement to produce sodium hydroxide from aqueous sodium chloride (see Fig. 3). Protons were generated at the anode through the hydrogen oxidation reaction (HOR) and hydroxide was generated at the cathode through the hydrogen evolution reaction (HER). Sodium ions combined with hydroxide in the alkaline compartment to form sodium hydroxide, which was used in turn to capture CO 2 as sodium (bi)-carbonate. Recirculation or treatment of the sodium (bi)-carbonate was not explored in this study. In the second arrangement, the AEM was replaced by a CEM to transport protons to the acidic chamber, which granted improved performance due to the generally higher stability and conductivity of CEMs compared to AEMs. Under optimized conditions, cell voltage was 1.25 V at 50 mA cm −2 , leading to a specific energy consumption of 367 kJ mol −1 CO 2 .
Another variation of the BPM electrodialysis CO 2 capture process was described by Jiang et al. 25 with the side process of amino acid production. The BPM electrodialysis unit consisted of three repeating BPM, AEM, and CEM units (see Fig. 2c). Experiments were performed under constant current density conditions from 20-50 mA cm −2 with inlet gas concentrations between 10%-30% CO 2 in N 2 . Faradaic efficiencies of up to 87% were reached for the highest inlet concentration of CO 2 case. Reported energy consumption was relatively high at 1109-1505 kJ mol −1 CO 2 , but the process simultaneously produced methionine and sodium hydroxide, which are potentially valuable chemical products.
Alternative approaches using liquid electrolytes.-Novel electrochemical approaches have been proposed based on seawater electrolysis and mineral calcination, [40][41][42] ionic liquids, 26,[43][44][45] or amines paired with a transition metal. 22,23 Rau 40,41 proposed the use of a calcium carbonate solution in tandem with electrolysis of a saline water solution (seawater) to capture atmospheric CO 2 as calcium bicarbonate, similar to that process shown in Fig. 3 without ion-selective membranes. This was accomplished through the following electrochemically-driven overall reaction: with calcium carbonate being split at the anode and calcium ions moving to the cathode via electromigration. Calcium carbonate was deemed an interesting storage medium for CO 2 due to its relative  24 Red text is exclusive to the system in Ref. 24. abundance and low cost. The concept was tested at a laboratory scale, and the electrochemically treated seawater absorbed CO 2 at a rate approximately 3 times that of untreated seawater. However, energy input was orders of magnitude higher than that expected for a commercial electrolyzer. The author noted the additional significant energy penalty of regeneration if concentrated CO 2 is desired as the end product, which would be partially offset by reduced supply and transportation costs if the reactant were recycled. A later study by Rau et al. 42 investigated a similar process using silicates and found an experimental energy expenditure of 426-481 kJ mol −1 CO 2 removed. Reinhardt et al. 43 wrote a perspective focusing on CO 2 capture using electrogenerated nucleophiles. In such a process, carbon in the CO 2 reversibly binds with an electrochemically reduced quinone (Q) within a liquid electrolyte on the cathode and is released on the anode as a pure gas. One benefit of these novel electrolytes is enhanced tunability of the chemistry, which can theoretically reduce the potential differences between the capture and release step. 43 This can be implemented in a system that either utilizes electrolyte pumping or electromigration across a static liquid electrolyte, with the following generic overall reaction: Scovazzo et al. 44 developed an aqueous redox active CO 2 carrier in a proof-of-concept batch-type cell and achieved pumping of CO 2 at <1% levels to pure concentrations at atmospheric pressures with a transference of 0.427 moles CO 2 per electron mole (solution CO 2 release step, 0.5 for the ideal case). Poor stability in O 2 was a critical issue for the system that would preclude operation with ambient air. Stern et al. 22,23 developed an electrochemically-mediated amine regeneration system to capture CO 2 from flue gas. The study used polyamines as the CO 2 sorbents and copper electrodes to produce cupric ions that assist in displacing CO 2 . The relevant reactions were: with A representing a generic amine molecule. Pure CO 2 was used to saturate the sorbent, and the authors achieved a faradaic efficiency of 42% operating at 2.5 mA cm −2 . 23 The authors also reported a specific energy consumption of 100 kJ mol −1 CO 2 for their system. 22,23 Tuning the chemistry to optimize binding constants for CO 2 and reduce the potential difference between the capture and release process is the subject of ongoing work. 43 Legrand et al. 46 proposed a membrane capacitive deionization technique to separate CO 2 , in which ions in a CO 2 -containing deionized water solution are removed via an anion exchange or cation exchange membrane and stored in the electrical double layer of a carbon electrode. The experimental cell operated in batch mode at very low current densities of 0.02-0.06 mA cm −2 and feed concentrations 15%, 30%, and 100% CO 2 in N 2 were tested. Authors reported low energy requirements of approximately 40 kJ mol −1 CO 2 , but also reported low amounts of CO 2 exchange (<9%) and very low capture CO 2 rates of 6•10 −9 mol/s•g carbon . In addition, captured CO 2 was desorbed back to the solution rather than to a concentrated gas stream, making direct comparison with other methods difficult. Authors acknowledged the necessity of an additional membrane contactor for separating liquid and gas phases to generate pure streams of CO 2 .
Direct separation from a gas mixture via an electrochemical cell.-Other existing literature in the field describes electrochemical cells that directly separate gas-phase CO 2 via electromigration of carbon-containing ions across a physical barrier. The electrochemical cell approach offers a distinct advantage over conventional membrane separation processes, 47 namely that it allows for highselectivity concentration from relatively dilute mixtures through active electrochemical pumping of carbon-containing ions and does not require large pressure gradients for operation. In fact, such a method opens possibilities of pumping against concentration gradients (i.e., electrochemical compression), as has already been demonstrated for hydrogen. 48 However, there are still significant challenges in operation when the feed gas is extremely dilute, such as the high energetic costs of delivering large volumes of gas to the cell and limitations in applicable current density.
In the first reported devices for electrochemical CO 2 separation, the electrochemical reactions occur on electrodes separated by an aqueous carbonate solution supported in a thin asbestos matrix. In such a device, water is produced accompanied by the generation of electrical energy (fuel cell-like operation) and the transfer of CO 2 from the cathode to the anode side. The National Aeronautics and Space Administration explored electrochemical methods for separation of CO 2 in the early 1970s to scrub CO 2 from spacecraft cabins. 49,50 These early studies note a number of benefits an electrochemical process has over a cyclic adsorption process, which include continuous operation, low CO 2 partial pressure capabilities with low system weight, concentrated CO 2 without air contamination, premixed H 2 and CO 2 at relevant ratios for downstream processing, elimination of mechanical CO 2 compressors, and low temperature operation with low flows. The electrochemical reactions hypothesized were: [ ] with reactions 9-12 taking place on the cathode side and reactions 13-14 taking place on the anode side. Wynveen et al. 49 designed and fabricated a one-man, self-contained electrochemical system for CO 2 separation and concentration, capable of removing 1.9 kg CO2 /day from air. Humidified CO 2 and O 2 were fed at the cathode and H 2 was fed at the anode, which were separated by a cesium carbonate electrolyte at 21°C-29°C. The authors also tested rubidium carbonate and cerium carbonate electrolytes with less success. The electrocatalyst used was platinum, which exhibited better performance than palladium. The primary figure of merit for system performance was transfer index, or mass of CO 2 removed over mass of O 2 consumed, which was affected by current density and CO 2 partial pressure. In typical operation, 50% or more of the inlet CO 2 stream was transferred to the H 2 stream, and the system showed no signs of deterioration after 260 d of operation. A follow-up study by Winnick et al. 50 furthered the concept with a scale-up of the technology. The system was tested with one, three, and 90 cells, each with an aqueous carbonate solution (either potassium or cesium carbonate) suspended in an asbestos matrix with platinum catalyst at the electrodes. The system operated in the same configuration as described previously. 49 Partial pressures of CO 2 up to 12 mbar and current densities up to 32 mA cm −2 were tested. Important findings were that CO 2 transfer rate increased with increasing inlet CO 2 concentration and increasing current density. At a given current density, there was a peak effective inlet CO 2 concentration beyond which the transfer rate was unaffected. Other results were that performance was unaffected by cell temperature and the scaled-up system performed as predicted based on the single cell system. Chin and Winnick 51 also developed a steady-state numerical model for the aforementioned system and found good agreement between the simulations and experiments.
Further studies by Winnick et al. 52,53 developed CO 2 separation from air using a molten carbonate electrochemical cell operating at high temperatures (>500°C). This approach benefits from its insensitivity to relative humidity and fast reaction kinetics. Experiments were run in both a hydrogen mode (i.e., fuel cell) and a driven mode feeding only N 2 at the anode. In the driven mode, the reaction at the anode was oxygen evolution from the oxidation of carbonate or bicarbonate, i.e.: [ ] A more recent study by Spinelli et al. 54 used thermodynamic system modeling to examine the possibility of retrofitting natural gas and coal power plants with molten carbonate fuel cells for CO 2 capture. The study estimated a specific primary energy consumption of approximately 57 kJ LHV /mol CO 2 avoided, which was significantly lower than conventional amine scrubbing methods.
Li and Li 55 investigated the use of an electrochemical membrane cell to remove CO 2 from breathing gas mixtures. Authors employed a 1-mm porous polyamide sheet saturated with potassium bicarbonate between two nickel screens. The feed gas, consisting of 4% CO 2 , 56% O 2 , and 40% N 2 was fed to the cathode side, resulting in mostly CO 2 with some O 2 collected at the anode. Current densities of up to 25 mA cm −2 were tested, at which point the CO 2 removal rate became relatively independent of current density. A follow-up study by Xiao and Li 56 used the same experimental setup to separate a humidified mixture of 4.8% CO 2 , 17% O 2 , and 78.2% N 2 on the cathode side with humidified N 2 on the anode side and modeled its performance. Experiments found the CO 2 removal rate to be independent of feed and carrier gas velocities within the range of 0-1.5 m s −1 . Based on simulation and experimental results, transfer of CO 2 was mainly controlled by resistances in the electrolyte solution (i.e., adsorption of CO 2 at the cathode, migration as carbonates, and evolution at the anode).
As described previously, electrochemically-mediated quinones can be used as redox-active CO 2 carriers in an electrochemical cell configuration. Gurkan et al. 26 characterized potential ionic liquids for their quinone solubility and stability in the presence of CO 2 . This operated through the same principle as that shown in Eq. 5. The high polarity of the examined quinone enabled more effective CO 2 separation by mitigating diffusive back-transport. Watkins et al. 28 investigated an electrochemical separation cell containing a quinone liquid-soaked membrane to separate CO 2 from a simulated flue gas mixture. The quinones, represented as Q, were used to transfer protons in the overall reaction: leading to the net transfer of CO 2 from the cathode to the anode side. Platinum was found to be the most effective catalyst in terms of CO 2 transport, while palladium and ruthenium catalysts showed significant O 2 due to water splitting at higher potentials. Eisaman et al. 16 developed an electrochemical method of CO 2 separation using an ion-conducting membrane. The proof-of-concept study employed either 390-μm-thick cellulose paper saturated with a cesium carbonate solution or a 500-to-600-μm-thick AEM saturated with a potassium carbonate solution. Ideal faradaic efficiencies for pure bicarbonate and pure carbonate transport are 100% and 50%, respectively, and ionic current carried by hydroxide and hydrogen ions negatively affects this parameter. In the experiments, higher inlet CO 2 concentrations led to higher faradaic efficiencies, but lower fractions of CO 2 separated. Using an AEM, air containing 400 ppm CO 2 was separated with an energy consumption of 350 kJ mol −1 and 23% faradaic efficiency. Water management was an issue when operating with fully humidified gases, prompting experiments with a room-temperature ionic liquid electrolyte solution containing cesium bicarbonate, which achieved comparable but slightly lower efficiencies.
Other studies by Landon and Kitchin 15 and Pennline et al. 30 employed an AEM in an electrochemical cell to separate mixtures of approximately 50% CO 2 in O 2 (Eqs. [9][10][11][12][15][16]. Only humidified argon was used as a carrier gas on the anode side, resulting in oxygen evolution and CO 2 formation from primarily bicarbonate. Both studies employed nickel and platinum as electrocatalysts, and cell potentials of up to 1.2 V were used in order to stay below the water-splitting potential of 1.23 V. The cell setup of Landon and Kitchin 15 reached current densities up to 6 mA cm −2 with an average ratio of 3.56:1 CO 2 :O 2 measured at the anode. Authors determined that the investigated membrane conductivity and stability were too low and electrocatalysts were not active enough for practical application in coal power plants. In addition, the gas mixture examined was not analogous to an air-combustion flue gas mixture, but the results compared favorably with previous studies of electrochemically separating CO 2 from breathing gas mixtures. 55,56 Rigdon et al. 27 examined the use of AEM to separate mixtures of CO 2 and O 2 in an electrochemical cell (Eqs. [9][10][11][12][15][16]. Platinum was used as the electrocatalyst for both the anode and cathode, and the polymer membrane was approximately 57 μm thick. Humidified N 2 was supplied to the anode side of the cell, and humidified 50% CO 2 in O 2 was supplied to the cathode. Similar to the study by Landon and Kitchin, 15 CO 2 was transported via membrane carbonation and evolved at the anode along with O 2 . Cell potentials of up to 1.5 V were reached at current densities of approximately 2 mA cm −2 with a resulting transference of approximately 0.67 CO 2 molecules per electron, indicating a CO 2 transport across the membrane via a mixture of carbonate and bicarbonate ions. The dominant transport pathway shifted from carbonate to a carbonate/bicarbonate mixture as the cell potential increased. There are important implications about the quality of CO 2 at the outlet of an electrochemical concentration cell system depending on the methodology used. While water can be easily separated from the gas mixture through condensation, gas phase separations are more difficult and costly to address. Inert carrier gases are often used to facilitate gas analysis, but other options would be more practical in a scaled-up system, e.g., recirculation of the anode gas. If carbonate and bicarbonate are neutralized at the anode in an inert atmosphere, the highest ratio of CO 2 to O 2 possible is 4:1 based on the stoichiometry of Eq. 16. Authors have proposed using the resulting gas mixture for oxy-combustion, resulting in a pure CO 2 stream. 30 If the cell is operated in a fuel cell mode (H 2 at the anode), hydrogen oxidation produces a mixture of CO 2 , water, and residual H 2 . Running the cell at low stoichiometries would in theory allow for higher concentrations of CO 2 to be obtained, although this would also approach an upper limit due to mass transport limitations and lead to possible degradation in the cell. In spite of this limitation, mixtures of H 2 and CO 2 could be useful chemical feedstocks, for example, in reverse water gas-shift reactors or methanol synthesis. Use of quinones could theoretically produce a pure stream of CO 2 in the same way that electrodialysis methods do, although O 2 evolution is still possible. 28 Energy consumption of CO 2 pumping with AEMs is still relatively high in the pioneering studies, but may be decreased below the state of the art for absorption through advancements in the catalyst and membrane materials, as well as cell engineering. 27 As with most AEM fuel cell studies, 57 the majority of electrochemical cell studies reviewed in this work use platinum as the electrocatalyst, but cheaper, platinum-group-metal-free catalysts will need to be employed to obtain significant reductions in stack cost. Cheaper membrane materials with high conductivities will also be critical in providing a scalable system, especially if the applicable current densities remain low. If current densities can be increased by an order of magnitude while limiting cell overpotentials and maintaining reasonable faradaic efficiencies, this will lead to an order of magnitude reduction in stack cost for a given production rate of CO 2 . Table II summaries some of the pros and cons for electrochemical methods of CO 2 capture.

Point-Source vs Direct-Air Capture
Point-source (e.g., flue gas) capture of CO 2 is a more thermodynamically advantageous approach compared to DAC. Capture of CO 2 from fossil fuel combustion is primarily performed in one of three configurations with different corresponding CO 2 concentrations: oxy-fuel post-combustion capture (>90% CO 2 ), pre-combustion capture (e.g., partial oxidation, 30%-35% CO 2 ), and air postcombustion capture (5%-25% CO 2 ). Minimum work to separate CO 2 depends logarithmically on the concentration; the minimum work to separate CO 2 from flue gas (15% CO 2 in N 2 ) at 300 K is 7 kJ mol −1 CO 2 . Given practical limitations to process efficiency, 32 kJ mol −1 CO 2 represents a rough lower limit of energetic cost for amine scrubbing, 4 with literature showing energetic inputs to be 4-5 times higher. 31,58 Membrane separation is typically discussed in reference to postcombustion gases, as it generally requires gas mixtures containing 10% or more CO 2 . 47,59,60 Two key parameters that require optimization for such systems are membrane permeability and selectivity, as the pressure gradient through the membrane provides the driving force. A number reviews on non-electrochemical, membrane-based methods of CO 2 capture from combustion gases have been published, focusing on conventional polymeric membranes, 47 facilitated transport membranes, 59 mixed matrix membranes, 61 organosilica membranes, 62 and membrane contactors as compared to membrane gas separators. 60 In terms of an electrochemical separation process, the chief benefit of point source capture is that the relatively high concentration of CO 2 allows for higher current densities to be reached at higher faradaic efficiencies in a direct gas separation cell. For a liquid electrolyte absorption step, less gas would need to be flowed through the absorber to saturate the solution, reducing energy costs. However, common flue gas contaminants (e.g., NOx, SO 2 , soot) would potentially pose additional challenges such as electrocatalyst poisoning or electrolyte degradation.
High energy requirements are the primary detriment of DAC compared to point-source capture. Due to the extremely dilute concentration, at least 2500 tons of air need to be processed in order to capture one ton of CO 2 , so a minimal pressure drop is desired. There are therefore implications for duration of the capture step and methodology to effectively deliver air to the capture medium and sufficiently saturate it with CO 2 . Lackner et al. 63 reviewed the state of the art and challenges in CO 2 capture from ambient air. Some of the critical requirements the authors note for air capture are the thermodynamic limits associated with the free energy of mixing CO 2 in air at 300 K and the mechanical work required to compress CO 2 from 1 to 60 bar isothermally, 22 and 11 kJ mol −1 CO 2 , respectively. The minimum theoretical separation work for the range of CO 2 partial pressures between 0.1 mbar and 1 bar is shown in Fig. 4, with the partial pressure of CO 2 in air indicated. However, there are also significant advantages to air capture, namely, that it allows for geographical decoupling from the emission source. This is of particular relevance to the transportation sector, which is still heavily reliant on carbon-based fuels. DAC can also reduce the need to transport CO 2 over long distances and enable the production of synthetic fuels in a closed carbon cycle. An additional benefit is the significantly reduced scrubbing requirement of air vs flue gas. 64 Electrochemical methods that operate isothermally at low temperatures can approach the theoretical separation work shown in Fig. 4 due to removing the inefficiencies associated with thermal cycling, 23,27 although overpotentials and faradaic losses present barriers to reaching those limits. Other important thermodynamic limits are set by the absorption reaction enthalpy between CO 2 and the aqueous sorbent solution. When an alkaline capture solution is used, e.g., NaOH or KOH, the enthalpy of the exothermic absorption reaction is 109.4 or 95.8 kJ mol −1 CO 2 , respectively. The enthalpy of absorption/desorption of a typical monoethanolamine solution has been estimated to be approximately 80 kJ mol −1 CO 2 . 65 For the AEM electrochemical cell methods described in Refs. 15, 27, 30, the minimum energy consumption as dictated by the standard electrode potentials of the half-reactions 66 is 35.1 and 56.2 kJ mol −1 CO 2 for a cell operating fully in the bicarbonate and carbonate mode, respectively. Although theoretically low energy consumption is obtainable using electrochemical approaches, it is also important to consider how a practical system would operate and the energy associated with mass transport (e.g., pumping of the capture solution, delivery of the feed gas mixture).
Sorption-based methods are the only commercialized approach for DAC, mainly due to the large volumes of air that need to be processed. Methods based on aqueous sorbents typically regenerate the sorbent using a causticization/thermal regeneration process. The cost of DAC is the subject of much debate, and has been estimated to be $30-1000 per ton of CO 2 captured. 64 A techno-economic analysis by Fasihi et al. 67 based on estimates from companies commercializing DAC technology gave a value of 243-317 kJ mol −1 CO 2 for combined electrical and thermal energy requirements, although current values are likely to be higher. According to a thermodynamic analysis of DAC performed in 2011, an energetic cost of greater than 400 kJ mol −1 CO 2 was considered potentially counterproductive due to the CO 2 intensity of fossil-based electricity sources in the United States at the time. 68 However, this value will shift upwards as adoption of renewable energy becomes more widespread.

CO 2 Transport in Alkaline Membranes
Ion-conducting membranes are of great interest in electrochemical separation processes, as they allow for selective transport of species through a physical barrier. AEMs are particularly relevant for CO 2 separation because of their active transport of (bi)carbonate. As the name implies, AEMs use negatively charged ions (e.g., hydroxide) as the charge carrier, as opposed to PEMs, which use hydrogen cations (see Fig. 5). In general, AEMs are a more nascent technology than PEMs; commercial implementation of PEMs in automotive fuel cell applications has already begun, albeit with some challenges that still need to be overcome. 69 The past decade has brought a significant level of interest in alkaline polymer membranes, a subgroup of AEMs, for fuel cells or co-electrolysis of CO 2 to form CO as a fuel or chemical precursor. Developments in AEMs for electrochemical applications are discussed in a perspective by Varcoe et al. 70 Alkaline membranes in fuel cells.-Alkaline membranes enable the use of non-noble metal catalysts in fuel cell applications, but face complications when air is used as the oxidant, as there is a significant negative influence due to CO 2 . CO 2 neutralizes hydroxide in the membrane via Eqs. 9-12 and reduces catalytic activity and ion conductivity. However, bicarbonate in the membrane will be converted to carbonate and hydroxide and transported out of anode side as CO 2 during power generation via Eq. 14. Early studies describe the so-called "self-purging mechanism" in alkaline exchange membrane fuel cell systems. [71][72][73][74][75] Under some conditions, almost all of the CO 2 fed from ambient air can be absorbed, transported, and released on the anode side of the alkaline membrane, and a higher degree of self-purging is observed at higher current densities. 72,74,[76][77][78] A modeling study by Krewer et al. 78 found that temperature has an unclear effect on membrane carbonation during AEM fuel cell operation, and that carbonate formation is heavily favored over bicarbonate at current densities of 0-1500 mA cm −2 . The same Table II. Pros and cons of various electrochemical methods.

Electrodialytical Regeneration
Electrochemical Cell (Direct Separation of Gas Mixture) +Decouples capture and release steps, making it more viable for low CO 2 concentrations +Allows for separation of a gas mixture in a single-step process +Allows for higher current densities to be reached, improving scalability +Polymer electrolytes are rapidly improving in terms of conductivity and stability +Can generate streams of pure CO 2 -Large ohmic losses due to membrane stack and electrolyte solution -Low concentrations of CO 2 lead to low faradaic efficiencies -Requires the use of a caustic solution with evaporative losses -Limited current densities in experimental literature -Gas bubbling creates localized high current densities -Difficulties in generating a pure CO 2 stream study determined that at current densities greater than 500 mA cm −2 , the carbonate builds up in the membrane at the anode in an enrichment zone less than 2 μm thick that is critical to transport and release of CO 2 . Other studies found that the ability to flush (bi)carbonate from the anode side efficiently as CO 2 is a key challenge. 74,79 The buildup of carbonate at the anode catalyst layer has also been identified as a source of the high anode overpotential in anion-exchange membrane fuel cells. 75,76,80,81 While the introduction of CO 2 on the cathode side leads to a clear degradation in performance of the cell due to increased ionic and reaction resistance, 82 experiments show that the effect is at least partially reversible. 75 Zheng et al. 83 76,79,85,86 and water management is a key issue. 79,87 As shown in Fig. 5, AEM fuel cells have two reactions involving water compared with one for a PEM fuel cell, and sufficient water must be supplied to the cathode to produce hydroxide. A recent review by Ziv et al. 88 comprehensively details the developments in understanding the effect of CO 2 on AEM fuel cells. Some studies have examined the use of carbonates to transport charge in AEM fuel cells (i.e., carbonate cycle) in contrast to the more common hydroxide. [89][90][91][92] In this case, CO 2 is fed intentionally on the cathode side at relatively high concentrations to maintain the membrane in (bi)-carbonate form. This potentially allows for higher power densities and improved stability in long-term operation. 92 In such a configuration, an increase in the CO 2 at the cathode can actually improve cell performance due to an increase in kinetic current. 91 One challenge with this route is the reduced ion mobility of carbonate and bicarbonate as compared to hydroxide. 80,85 Another area that requires improvement is the selection of catalysts that promote carbonate formation over hydroxide formation. 91 Alkaline membranes for CO 2 reduction.-A number recent of CO 2 co-electrolysis studies also contain relevant information on CO 2 transport in alkaline membranes. The main drawback of using liquid electrolytes instead of polymer electrolyte membranes in electrochemical CO 2 reduction is the poor solubility of CO 2 , which limits applicable current densities and potential scale-up. Polymer electrolyte membranes have lower gas permeability than electrolyteimpregnated porous media, and thus can be made thinner to reduce ohmic losses. 93 Indeed, faradaic efficiencies of CO of >90% have been reported in literature for alkaline membrane CO 2 -splitting. [94][95][96] Liu et al. 94 performed a CO 2 -electrolysis study using imidazoliumfunctionalized polymer membranes with silver catalyst at the cathode and iridium dioxide catalyst at the anode, and achieved very high CO selectivity (>95%) and current densities up to 600 mA cm −2 . A potassium bicarbonate solution was fed to the anode to assist ion conduction, and water management was identified as a key issue in effective operation. The ratio of CO 2 :O 2 at the anode was close to 2, suggesting carbonate as the charge carrier. Yin et al. 96 reported on a cationic polymer membrane co-electrolyzer with gold catalyst at the cathode and iridium dioxide catalyst at the anode that operated with pure water. Unlike the setup by Liu et al., 94 no alkaline solution was fed during operation, but the duration of the experiment was limited to 100 h as opposed to 4000 h. The cell was capable of maintaining current densities of 500 mA cm −2 and CO 2 crossover was low (<1 μl min −1 ·cm −2 ), suggesting that hydroxide was the main charge carrier.  CO 2 transport through the membrane is an undesirable sideprocess in CO 2 -splitting applications, as it is a parasitic loss for the system and reduces CO 2 utilization efficiency. Pătru et al. 95 evaluated cell designs for gas-phase CO 2 reduction using CEM, AEM, BPM, and a novel BPM-like configuration. Authors also quantified CO 2 production at the anode of the various cell configurations. In alkaline membrane co-electrolysis of CO 2 , carbonate and bicarbonate formation at the cathode results in one-half to one CO 2 molecule being transferred to the anode side per electron. CO 2 is also transported via the hydrogen evolution reaction at the cathode in concert with other side reactions. For the co-electrolyzer, an alkaline membrane was prepared with gold catalyst at the cathode and iridium dioxide and titanium dioxide catalyst at the anode. Similar to Liu et al., 94 the study found that at high current densities, twice the amount of CO 2 as O 2 was produced at the anode, indicating the formation and transport of carbonate through the membrane. At lower current densities, less CO 2 was produced but constituted a larger proportion of the anode outlet gas. Using a BPM with the acidic cation exchange side facing the anode greatly suppressed this CO 2 -pumping effect. One challenge with a BPM is delamination at the interface due to the formation of water and CO 2 , and a novel configuration using an alkaline ionomer within and on top of the cathode catalyst layer and an acidic PEM was able to mitigate this problem while still suppressing unwanted CO 2 transport.

On the Effects of CO 2 in Proton Exchange Membrane Fuel Cell Systems
Although active transport of CO 2 (i.e., electrochemical pumping) does not occur in PEM fuel cells, the topic is briefly discussed here. A review of the common perfluorosulfonic acid (PFSA, e.g., Nafion) membranes by Kusoglu et al. 97 notes the slight variation in measured CO 2 permeability values found in literature, but it is generally the third most permeable molecule after He and H 2 . CO 2 has also been observed to have a pressure-dependent permeability due to its tendency to plasticize fluorocarbon chains. For CO 2 in PFSA membranes, both permeability and selectivity increase with increasing relative humidity.
The effects of CO 2 in PEM fuel cells are well studied due to the potential for CO to poison the catalyst. 98,99 Even if clean (CO 2 -free) air is used as the oxidant, CO 2 can be formed through oxidation of carbon at the electrode, which is a particularly relevant mode of corrosion during transient start-up or shut-down of the cell. 100 A pertinent study by Erbach et al. 101 examined the effect of CO 2 crossover to the anode side through diffusion when ultra-thin (⩽15 μm) PEMs are used. CO 2 permeation was examined over a range of different membrane materials, thicknesses, temperatures, relative humidities, pressures, and flow rates. A hydrocarbon-based membrane introduced significantly less CO 2 crossover compared to PFSA membranes. Higher pressures, higher relative humidities, and higher temperatures were all associated with higher CO 2 permeation levels. With a feed concentration of 1% CO 2 in Ar, up to 135 ppm CO 2 was observed on the permeate side in a single cell configuration using an 8 μm PFSA membrane. Increasing the current density in the full size fuel cell system did not appear to affect CO 2 on the anode side, as expected in the absence of electrochemical CO 2 pumping.

Conclusions
The state of the art in electrochemical CO 2 separation methods is presented in this work, which provides a basis for comparisons and future work. Significant research advances and material improvements during the past several decades have enabled performance enhancements to make electrochemical CO 2 separation commercially interesting. As shown, CO 2 separation is relevant for a wide range of applications, and will only become more critical in the coming years due to the ubiquity of CO 2 emissions. Current industrially-implemented methods of CO 2 separation have significant drawbacks-sorbent regeneration requires large heat inputs and/or pressure swings, and cryogenic distillation is very inefficient. 47 Electrochemical membrane separation could maintain the desirable aspects of membrane separation, such as high modularity, continuous operation, and relatively low energy intensity, while improving aspects such as selectivity and ability to operate with low CO 2 concentrations.
While early studies show interesting possibilities for electrochemical CO 2 capture, there is still significant research and development work required to reduce costs and increase performance. Critical to these cost reduction pathways are the catalyst and membrane contributions. Research efforts should move towards platinum-group-metal-free catalysts and higher current densities. If the applicable current densities remain restricted to the 0-10 mA cm −2 range, very large membrane/electrode areas will be required to separate appreciable amounts of CO 2 , incurring high capital costs. The importance of faradaic efficiency will depend strongly on the price and CO 2 intensity of electricity used to drive the process. For example, a relatively low faradaic efficiency process may become attractive if capital costs are low and off-peak renewable electricity can be utilized. Another topic that needs to be investigated further is system durability across long operation times, and this should be reported on in future studies. The parallel advancements in AEM fuel cell and electrolyzer technology may allow further improvements to the state of the art in electrochemical CO 2 separation.