Carbon capture and utilization (CCU) has been recognized as one of the most promising technologies for mitigating climate change seen in decades because of its capacity for large-scale CO2 reduction. CCU technology is the only group of technologies that can achieve a net-zero emission target by removing direct and balanced emissions1. According to the global roadmap, CCU has the potential to reduce carbon emissions by over 7 gigatons by 2030, and the corresponding market size could reach 800 billion USD2. However, the industrial application of CO2 conversion technology is still challenging due to its low technological maturity, expensive production cost, and high energy consumption3. One way of early settlement of CCU technology is the sequential conversion of CO2 via synthesis gas (syngas) generation. Sequential production via syngas has the great advantage of producing various chemicals by adjusting the H2-to-CO ratio and employing subsequent reaction processes4. However, conventional CO2-derived syngas production methods are highly endothermic, which hinders commercialization and consequently leads to question about their ability to reduce CO2.

Electrochemical CO2 reduction (eCO2R) has attracted much attention as an alternative route for sustainable production of CO or syngas5,6,7,8. eCO2R provides a pathway for achieving net-zero emission during syngas production by utilizing renewable energy. Nevertheless, eCO2R processes are not economically attractive because CO2 conversion through eCO2R generally requires expensive CO2 capture and product conditioning processes for purified raw material and product supply. Recently, novel efforts have been reported to overcome the cost-inefficiency of eCO2R technology by developing new systems excluding pre-/post-conditioning processes, such as utilization of low-concentration CO2 (e.g., flue gas)9,10,11 and direct conversion of CO2 captured in an amine solution12,13. In particular, direct eCO2R in an amine solution is a promising strategy because it could eliminate energy-intensive thermal amine regeneration and consume negligible energy for pressurization. Theoretically, the net-zero CO2 conversion process can be built based on direct conversion of CO2 captured in an amine when the necessary electricity is supplied from a renewable source.

Several attempts have been made to perform eCO2R in commercial CO2-capturing absorbents (e.g., monoethanolamine, diethanolamine, 2-amino-2-methyl-1-propanol, and their mixtures), and their limitations have been discovered12,13. These primary and secondary alkanolamine solvents capture CO2 as a carbamate, which inherently contains a strong C-N bond that makes CO2 conversion difficult14. Chen et al. made an early attempt to demonstrate the possibility of eCO2R in a 30 wt.% monoethanolamine solution12. This process exhibited a moderate CO Faradaic efficiency (FE) of 38.2%, but the authors also found that the active carbon source for the eCO2R was free CO2 in the solution, not captured CO2 as carbamate. Lee et al. conducted eCO2R with captured CO2 in a monoethanolamine solution and first observed that the steric properties of a cation in the Helmholtz layer caused the limited performance of eCO2R13. Addition of electrolytes involving alkali metal cations could constitute a method for improving the eCO2R performance. However, this method may be economically unfavorable due to the expensive supply of electrolytes. Li et al. shed light on the direct-capture CO2 conversion system by suggesting bicarbonate as a valid option for converting captured CO2; this process showed a 37% FE for CO at −100 mA cm−2 without the addition of supporting electrolyte15, and the performance has further increased to 95% FE for CO at −100 mA cm−2 under 4 atm pressure condition, in a recent follow-up study16. In this system, KOH is considered to be a currently available CO2 absorbent. Although KOH is a promising solvent for CO2 capture and conversion, the salt formation and corrosion issues caused by the extremely alkaline condition are considered as major challenges for the commercialization17,18,19,20. Therefore, a specific method for sustainably supplying bicarbonates from flue gas is required. Most importantly, process design and systematic analyses such as a techno-economic analysis (TEA) or life cycle assessment (LCA) have not been applied in spite of the technological potential of achieving net-zero CO2 conversion and producing high-pressure syngas.

Herein, we present an economically feasible and environmentally benign methodology for CO2 reduction called a reaction swing absorption (RSA). RSA is a new syngas production method that directly utilizes CO2 in flue gas without CO2 capture or product separation processes. Throughout this work, we demonstrate experimentally the capability of CO2 absorption and direct eCO2R of RSA. A conceptual design for RSA is also provided, which shows that RSA can be operated without thermal energy consumption; thus, net-zero CO2 emission is possible depending on the energy supply. TEA and LCA are also provided to highlight future research guidelines required for sustainable chemical production via RSA.


Reaction swing absorption

The term RSA refers to processes that selectively separate a target chemical from a gas stream using an absorbent and chemically convert it to desired products. The products are released from the absorbent as the reaction takes place due to solubility differences between reactants and products. For example, the CO2 RSA separates CO2 from flue gas, and subsequent eCO2R in an amine solvent releases high-purity syngas because the product has low solubility in the amine solvent. Figure 1 shows the generalized concept for RSA production of high-pressure syngas (i.e., a CO and H2 gas mixture) from CO2. RSA consists of chemisorption, pressurization, and electrochemical conversion (e-chemical)21 processes. The chemisorption process generates bicarbonate as an absorbent and captures CO2 from the flue gas. The bicarbonate solution is pressurized with a pump and then selectively reduced to CO in a zero-gap membrane electrode assembly (MEA) electrolyzer. The unreacted CO2 remains in the solution, so a gas product separation process is no longer required.

Fig. 1: Schematic diagram comparing the RSA pathway with other CCU pathways.
figure 1

Graphical representation of the reaction swing absorption process. The light blue stream represents the RSA pathway. Compared to the RWGS (dark blue) and gas eCO2R (orange) pathways, the RSA pathway exhibits a simple process configuration, retains the highest portion of electricity, and is potentially capable of providing a viable CCU solution.

Figure 1 also describes two conventional syngas production pathways; the reverse water gas shift reaction (RWGS) and gas phase eCO2R9,13,22. Both pathways exhibit far more complicated process configurations than CO2 RSA because they require CO2 separation, pressurization, and product separation. In conventional processes, thermal solvent regeneration in CO2 separation and high-temperature syngas generation (i.e., RWGS) consume considerable amounts of thermal energy, so CO2 reduction in these systems shows limited capability23. Additionally, considering the low one-path conversion to CO in eCO2R (<50%), an additional process for separating unreacted CO2 and recycling it back to the electrolyzer should be considered to maximize carbon utilization24. Furthermore, even the most advanced eCO2R electrolyzers (e.g., a zero-gap MEA electrolyzer) have experienced systematic problems such as carbonate crossover, which causes system instability and low energy efficiency. The RSA successfully overcomes the abovementioned drawbacks while producing high-pressure and purified syngas. We first describe experimental observations for unit operability of the RSA process and present comprehensive comparisons among CO2 RSA, RWGS, and gas phase eCO2R via TEA analysis and LCA.

Amine selection for selective CO2 absorption as bicarbonate

Alkanolamines are commonly used in industry due to their fast CO2 absorption rates and high capacities, but most of them capture CO2 as carbamate25,26. According to previous studies conducted on eCO2R with CO2 captured in amine solutions, CO2 is not easily released from carbamates due to the strong C-N bonding13,14. Thus, it is important to find a new amine that captures CO2 in a mild form such as bicarbonate, and directly converts amine-captured CO2. After screening various amines, including primary, secondary, and tertiary amines, we found that triethylamine (TREA) is an ideal solvent for both CO2 capture and bicarbonate utilization, which contains aliphatic groups but not hydroxyl groups. We confirmed the presence of bicarbonate as the major form of CO2 captured in a TREA aqueous solution by nuclear magnetic resonance (NMR) analysis (Fig. S1). In the 13C-NMR spectrum of CO2 captured in a 3 M TREA/H2O solution, a single peak derived from bicarbonate (and/or carbonate) ion was observed. Carbamate species were not observed. Considering that the CO2 saturated 3 M TREA/H2O is a neutral solution (pH 7–8), it can be deduced that bicarbonate was selectively generated during the absorption of CO2 by TREA according to the chemical equilibrium determined by pH27. Figure 2a suggests a simplified mechanism for CO2 capture in a 3 M TREA solution. The mechanism shows an equimolar reaction between CO2 and TREA as bicarbonate generation yields TREA protonation. Note that the theoretical absorption capacity of the TREA is twice that of monoethanolamine, although monoethanolamine has higher reactivity.

Fig. 2: The reaction mechanism for CO2 captured in TREA solution and CO2 absorption performance in 3 M TREA solvent.
figure 2

a A representation of the proposed CO2 capture process is shown, where the TREA solution captures CO2 by stabilizing the bicarbonate on an equimolar basis. b A schematic figure of the CO2 absorption column used for the capture experiments. The column is 3 meters in height with a diameter of 3 inches. c CO2 absorption experiment results with varying TREA flow rates at fixed gas flow rate (0.8 m3 h−1) and CO2 concentration in gas flow (3% CO2). d Effect of L/CO2 ratios on CO2 absorption capacity at various values of TREA flow rates (2–5 L h−1), gas flow rates (0.5–0.8 m3 h−1), and CO2 concentrations (3–5%). CO2 absorption capacity shows a linear correlation with L/CO2 ratio, regardless of TREA flow rate, gas flow rate, and CO2 concentration.

The CO2 absorption capacity of TREA solvent was measured with a bench scale absorption column equipped with structure packing (Figs. 2b and S2). We measured the CO2 absorption rates and capacities at liquid feed/gas CO2 feed (L/CO2) ratios ranging from 0.07 to 0.21 with synthesized flue gas containing 3–5% CO2. Figure 2c shows that the higher the L/CO2 ratio at fixed gas flow and CO2 concentration, the greater the CO2 absorption rate but the less CO2 absorption capacity. Figure 2d presents a linear relationship between L/CO2 ratio and CO2 absorption capacity, regardless of liquid flow (2–5 L h−1), gas flow (0.5–0.8 m3 h−1), and CO2 concentration (3–5%). All CO2 absorption experiments with TREA in this study exhibit high performance, in terms of CO2 absorption rate from 84.5% at fast flue gas flow rate (0.8 m3 h−1 with 3% CO2) to 95.1% at slow flow (0.5 m3 h−1 with 3% CO2) shown in Fig. S2.

Desirable system configuration for direct conversion of amine-captured CO2

CO2-captured TREA solution was then supplied to the electrochemical system for CO2-derived syngas generation. The eCO2R system was designed based on the MEA electrolyzer, which is the most suitable system applied commercially (Fig. 3a)28,29. The eCO2R system configuration is similar to that of the catholyte-free MEA electrolyzer except that an amine-captured aqueous CO2 solution is provided to the cathode instead of humidified CO2 gas. Since the targeted product is syngas, Ag nanoparticles (a representative catalyst for CO production)9,30,31,32,33,34 on a carbon supporter (Ag/C) were initially selected as a standard catalyst in developing the system. Ni foam was used as the anode while 1 M KOH was supplied as an anolyte. Between the cathode and anode, a bipolar membrane was positioned to separate the two electrodes and supply proton and hydroxide ions to each electrode (Fig. 3b). Gas products from CO2 conversion were delivered back to a liquid reservoir with the CO2 capture solutions, and then analyzed by gas chromatography (GC). Only gaseous products were detected in this system.

Fig. 3: Syngas production with the eCO2R system.
figure 3

a Schematic of the eCO2R system configuration. b Schematic of chemical reactions with a cathode, anode, and bipolar membrane. c CO FEs for Ag/C measured with various applied current densities in monoethanolamine, diethanolamine, and TREA. d CO FEs for Ag NP, Ag/C, coral-Ag, and coral-Ag/C in 3 M TREA. e SEM images (scale: 500 nm) and contact angles for all Ag electrodes. f CO FEs for Ag/C measured with various membranes, including an anion exchange membrane, cation exchange membrane and bipolar membrane, in 3 M TREA. g Long-term CO production represented as the H2:CO ratio, CO FEs, and cell voltages (blue) for coral-Ag/C at −100 mA cm−2 in 3 M TREA.

To demonstrate the excellent capability of TREA, eCO2R with various CO2 absorbents such as monoethanolamine and diethanolamine were compared to that with TREA in the abovementioned system configuration (Figs. 3c and S3a). CO2 conversion was evaluated with CO FE at applied constant current densities ranging from −20 to −200 mA cm−2 by using chronopotentiometry. CO FEs at all conditions were below 15% when 30 wt.% monoethanolamine and 30 wt.% diethanolamine were used. However, the 3 M TREA outperformed the other amines, demonstrating that it is the best option for a feasible system. Considering that monoethanolamine and diethanolamine transform CO2 into carbamates, bicarbonate formation with TREA is the most favorable for direct conversion of CO2. Moreover, experiments in which the bicarbonate concentration in TREA was decreased over 6 h of eCO2R confirmed that bicarbonate served as the carbon source for CO production (Supplementary Note 3 and Table S1). Initially, there 2.78 M bicarbonate was present in the TREA solution prior to the reaction, and the concentration linearly decreased to 2.47 M as the reaction progressed in 6 h. This result indicated that bicarbonate was utilized as the major CO2 source. During the eCO2R, CO2 was not detected by GC, implying that unreacted CO2 was efficiently captured as bicarbonate in the TREA solution (Fig. S4).

Syngas production by the RSA system was further improved by cathode development (Figs. 3d and S3b). In a typical catholyte-free MEA system, CO2 is supplied in the gas phase, and thus, hydrophobicity is required for the catalyst electrode to form a three-phase interface, which acts as an active site35,36. In contrast, this direct amine-captured CO2 conversion system utilizes CO2 from the liquid electrolyte, so a hydrophilic catalytic electrode would be preferred for sufficient interaction of the catalyst layer with CO2. To simultaneously achieve a suitable electronic structure enabling high CO selectivity for the Ag catalyst as well as hydrophilicity for the carbon supporter, we electrochemically modified the Ag/C into a coral like-structure (coral-Ag/C) through electrochemical oxidation and reduction processes (Fig. 3e)32. To identify the effects of the coral-Ag structure and carbon hydrophilicity, additional catalysts, such as (i) Ag nanoparticles (Ag NP) and (ii) coral-structured Ag nanoparticles (coral-Ag), were prepared as a control group, and their characteristics were correlated with their CO production performance (Figs. S5S10). As expected, the coral-Ag exhibited a higher CO FE than the Ag NPs, and this result clarified that the coral structure was favorable for CO production in the amine-captured CO2 conversion system. Randomly connected 3D coral structure with partially oxidized Ag (Ag+) near the surface provided suitable surface binding affinities for reaction intermediates32. More importantly, the coral-Ag/C catalyst achieved an outstanding CO FE of 70% at a current density of −20 mA cm−2 and 30% at a high current density of −200 mA cm−2 (Fig. 3d). Considering the moderate performance improvement from Ag NP to Ag coral and from Ag NP to Ag/C, the improved performance of coral-Ag/C implies that there may be other factors to contribute to the performance in addition to the structural modification of Ag and the introduction of carbon support. Thus, we suggest that the hydrophilicity of the electrode is a novel critical factor determining catalytic performance in the RSA system. As shown from contact angle measurements, coral-Ag/C showed the lowest contact angle among the electrodes (Fig. 3e). Similar contact angles for coral-Ag and Ag NPs implied that the electrochemical oxidation and reduction processes were not influenced by the hydrophilicity of Ag but by that of carbon support, as we expected. This influence might derive from increases in the number of oxygen functional groups present on the carbon supporter during electrochemical processes, as observed via X-ray photoelectron spectroscopy (XPS) analysis (Fig. S11). We further compared the eCO2R performance according to hydrophobicity of electrodes (Fig. S12). As expected, hydrophobic electrodes showed worse performance with the same Ag catalysts.

System sustainability was secured through selection of an appropriate membrane and confirmed by long-term stability. In the RSA system, since 3 M TREA and 1 M KOH solutions are respectively utilized as catholyte and anolyte, both electrolyte separation for recycling and proton supply for releasing CO2 from bicarbonate are important. In this regard, a bipolar membrane is the best option for supplying protons to the cathode and hydroxides to the anode, respectively, while guaranteeing negligible ion transfer between the catholyte and anolyte. In cases with other membranes, such as anion exchange and cation exchange membranes, ion transfer inevitably occurs due to the intrinsic ion-conducting properties, and this ion transfer interferes with the long-term use of the electrolytes. In practice, only bipolar membrane systems have exhibited acceptable CO production performance, while other types of membranes have shown poor CO production performance (Figs. 3f and S3c). It was also confirmed that the bipolar membrane prevented bicarbonate crossover to the anode, since no CO2 gas evolution was detected from the anode (Fig. S4). This optimized system configuration (coral-Ag/C cathode and bipolar membrane) demonstrated stable performance with 35% of CO FE, during a 70 h operation of chronopotentiometry experiment at −100 mA cm−2 (Fig. 3g).

Techno-economic analysis and life cycle assessment of RSA

We carried out a comprehensive TEA and an LCA of RSA using a process model based on the abovementioned experimental results. The result was compared with those of conventional processes using the RWGS and gas phase eCO2R (Fig. 1) to confirm the superiority of the proposed RSA process. A global sensitivity analysis (GSA) was also performed to assign cost contributions and global warming potentials (GWPs) and eventually to highlight the crucial factors for further improvement of RSA37. Current and optimistic scenarios were compared through analysis of several related factors: (i) the electricity generation cost based on different energy sources, (ii) improvements to the bicarbonate electrolyzer, (iii) purchase cost of H2 produced from an on-site water electrolysis system as an alternative option to satisfy the H2-to-CO ratio of 2 for syngas, and (iv) CO2 capture rate of the absorber. H2 in syngas comes from purchase, from the CO2 electrolyzer as a by-product, or from both. There is competition of H2 supply between the bicarbonate electrolyzer and on-site purchase. Therefore, the third factor (purchase cost of H2) is added for the sensitivity analysis. Whereas the operating conditions of RWGS and gas eCO2R processes are well known, those of the RSA should be optimized due to its early development stage. As a result, the optimal cell voltage, which is pertinent to the CO FE and current density, was found to minimize the break-even price of syngas. The details of the process modeling methodology and TEA and LCA results are explained in Supplementary Note 4.

As illustrated in Figs. 4a, b, S18 and S19, the RSA process outperforms the other two CCU processes, in terms of operating expenditure (OPEX) and the break-even price of syngas in the optimistic scenario, regardless of energy sources. In the current scenario, all three processes have a similar OPEX and break-even price, although the RSA shows an enormous capital expenditure (CAPEX) due to the inefficient bicarbonate electrolyzer. However, current density can be increased at lowered voltage with higher CO FE, as the performance of bicarbonate electrolysis is improved in the optimistic scenario. This improvement considerably reduces CAPEX of the RSA and converts the H2 supply from the bicarbonate to water electrolyzer. Considering lower cell voltage in water electrolysis than that in bicarbonate electrolysis, the H2 supply change results in the RSA beating other two processes for CAPEX, OPEX, and break-even price in the optimistic scenario. When renewable energy is used, the optimistic break-even prices for syngas will be dropped to $0.65/kg of syngas for wind and $0.56/kg of syngas for solar. These prices can compete with fossil fuel-based syngas processes (Fig. 4f). Although the wind case has a slightly higher syngas price than that in the solar case, wind is a more promising energy source from an environmental perspective (only one-third of the GWP100 value in comparison to the solar case).

Fig. 4: Results of TEA, LCA, and GSA for the wind case.
figure 4

a Capital expenditures for three processes. b Operating expenditures and break-even prices for syngas from the three processes. c Graphical representation of LCA system boundaries. d LCA results for the three processes. The left and right bars in each scenario represent global warming potential (GWP) and fossil resource scarcity (FRS), respectively. e LCA results for the RSA process with different cases and scenarios. f Cash flow charts for an optimistic scenario with a selling price of $0.8/kg syngas. g 1st order Sobol indices from global sensitivity analysis.

Figure 4c presents the system boundaries for LCAs containing CO2 sources within the boundaries. Two important impacts for an LCA of CCU technology are GWP100 and fossil resource scarcity (FRS). In the case of GWP100, the RSA process has a lower impact than the other processes in all cases except the current energy mix case (Fig. S20). The GWP100 of the RSA can be minimized to 0.27 kg CO2 eq./kg syngas (Fig. 4d) when wind energy is used for electricity generation. However, GWP100 is highly sensitive to the energy source, so this value can be increased up to 5.52 kg CO2 eq./kg syngas in the energy mix case (Fig. 4e). The FRS result shows a similar trend to GWP100, because the RSA is highly energy-intensive process.

A GSA was also conducted to support decision-making for establishing appropriate CCU strategies and policies and to provide the priority of research targets. Unlike local sensitivity analyses, such as a one-factor-at-a-time method, GSA changes all uncertain input variables simultaneously and monitors the variances of dependent variables38. Thus, a GSA generates Sobol indices, which indicate the impacts of uncertain (input) variables on dependent (output) variables. There are six uncertain variables in the RSA for the solar and wind cases: CO2 capture rate, CO FE improvement, current density improvement, unit cost of the electrolyzer, electricity generation improvement, and additional cost (except electricity cost) when H2 in syngas is supplied by an on-site water electrolysis system, instead of H2 production in the bicarbonate electrolyzer as by-product. For the energy mix case, portions of solar and wind energies in electricity generation mix are added to the uncertain variable set.

Figure 4g compares the 1st order Sobol indices of the break-even price of syngas and the GWP100 for the RSA process under a given range of uncertain variables from current to optimistic scenarios. When wind energy is used for the RSA, the most sensitive economic factor is the electricity generation price (electricity generation improvement in Fig. 4g). Wind energy is well-known for its very low CO2 emission, so electricity becomes no longer a major environmental issue. Consequently, uncaptured CO2 to be released to air from the absorber is the main contributor to CO2 emission when wind energy is used. This result indicates that the near net-zero CO2 emission is achievable as the CO2 capture rate increases in the chemisorption process. For the energy mix case, no factor dominantly impacts the break-even price. From an environmental perspective, the additional cost except electricity cost to the on-site water electrolysis system is the most significant factor. In other words, the improvement of water electrolysis infrastructure and technology economically attracts this system for H2 production and substantially reduces electricity usage. Consequently, the amount of emitted CO2 in the energy mix case can be largely decreased by this improvement. The detail TEA, LCA, and GSA results for different scenarios and cases can be seen in Supplementary Note 4.


We report RSA as a potential net-zero CO2 emission methodology for syngas production from CO2. By utilizing a bicarbonate-generating amine solvent, TREA, RSA successfully captures CO2 from flue gas and selectively reduces it to syngas via direct electrochemical CO2 reduction. The CO2 absorption capability of TREA was measured with a bench scale absorber and a scalable system configuration for eCO2R, and this further supported the feasibility of RSA in industrial applications. After optimization of variables in the eCO2R system, such as catalyst, electrolyte, and membrane, RSA produced high-purity syngas composed of 30 to 70% CO with a balance of H2, depending on applied current densities ranging from −200 mA cm−2 to −20 mA cm−2. It is worth noting that the impact of trace amounts of impurities, such as NO3-, on the bicarbonate electrolysis system should be investigated before upscaling the process16. A TEA based on the experimental results showed that RSA is the most promising CCU process with the lowest CO2 emissions, and it outperforms current CCU processes.

CCU technology has been studied substantially but commercialization is still limited by many challenges and difficulties originating from the high stability of CO2. If syngas, a product of the endothermic reactions of CO2, can be produced economically, then it is possible that CCU technology can be industrialized because the downstream processes are exothermic (Fig. 5). For example, methanol, ethanol, and dimethyl ether synthesis and the Fischer-Tropsch reaction are highly exothermic, implying that these downstream processes can be operated without excessive energy consumption. Considering the technological maturity of downstream processes utilizing syngas, we believe that RSA can change the paradigm for chemical processes by realizing the eco-friendly and cost-effective production of syngas from CO2. In Fig. 5, energy-intensive syngas production accounts for the highest proportion in terms of cost and energy consumption. Therefore, employing an environmentally benign and economically feasible process of syngas production could accelerate the commercialization of CCU technology. Such acceleration may be achieved through the RSA process proposed in this study because it requires neither an energy-intensive step for amine solvent regeneration nor a separation system of unreacted CO2. According to the modeling results of this study, it is crucial to improve the bicarbonate electrolyzer used in RSA and integrate the system with renewable energy sources in order to make the price highly competitive with that of fossil fuel-based syngas and achieve net-zero CO2 emission.

Fig. 5: Syngas as a versatile intermediate chemical having a competitive price when using the eco-friendly CO2 RSA process.
figure 5

Estimated levelized cost of target chemicals produced from consecutive RSA and downstream processes39 (red), current market prices of chemicals (blue), and heat of reactions of syngas production and downstream reactions40,41 (green). (Market prices from industries42,43,44 and a report45.


CO2 capture rate measurement

The CO2 capture rate of 3 M TREA solution is measured in a pilot scale absorber. A micro gear pump (GA-V21) and a mass flow controller are used to control the simulated flue gas and absorbent flow rates. The 3–5% of simulated flue gas are prepared using 99.99% CO2, N2, and air. The flue gas and absorbent are count currently contacted in an absorber equipped with 3-inch Sulzer structure packing (DX) in order to maximize the mass transfer between the gas and liquid. We calculated CO2 capture rate by measuring inlet and outlet CO2 flow rates. To achieve the steady state condition of the absorber operation, each experiment lasted at least 45 min. More Detail information is available in Supplementary Note 1.

Catalyst preparation and characterization

For all Ag electrodes, a carbon paper (AvCsrb MGL190) with 200 nm of Ag primary layer, which was deposited by e-beam evaporator (UL VAL Inc.) with 3 A s−1 deposition rate under 10−6–10−7 Torr of vacuum condition, was used as a substrate. The Ag nanoparticle (Ag NP) catalysts (Alfa Aesar, 99.9%) were basically deposited on the prepared substrate by spraying-coating method and modified by mixing carbon or by conducting electrochemical structural tuning. They were named according to their composition and structure such as Ag NP, Ag/C, coral-Ag, and coral-Ag/C. First, Ag NP electrode was prepared by spraying of ink solution which was composed of Ag NP (120 mg), isopropanol (5 ml), and 5 wt.% Nafion perfluorinated resin solution (128 µl) with catalyst loading amount 1 mg cm−2. The only difference for Ag/C electrode was that Ag NP was mixed with Ketjen black 600JD with mass ratio 75% of Ag NP in the catalyst ink solution. Coral-Ag and coral-Ag/C electrodes were respectively prepared by oxidizing of Ag NP and Ag/C in the 0.1 M Ar-saturated KCl (Sigma-Aldrich, >99%) solution at 0.3 V (vs Ag/AgCl (3 M KCl) reference and Pt foil counter electrode for three-electrode system) for 12 h and then reducing them in 0.1 M KHCO3 at −1.2 V (vs Ag/AgCl (3 M KCl)) for 30 min as previously reported method32. The morphology of the prepared Ag catalysts were characterized by scanning electron microscope (SEM) and energy dispersive X-ray spectroscopy (EDX) mapping (Hitachi Regulus 8230 with UHR cold type field emitter gun). The chemical structure of Ag electrodes was analyzed by XPS spectra using Nexsa (Thermo Fisher Scientiffic) with a monochromated Al-Kα (1486.6 eV) source. To understand the effect of the hydrophilicity/phobicity of the prepared Ag electrodes, contact angle was measured by a contact angle meter (DSA 25, Kruss) with 2.0 µl of DI water.

Electrochemical measurement

All electrochemical CO2R were performed in a commercial MEA electrolyzer (complete 5 cm2 CO2 electrolyzer, Dioxide materials). The as-prepared Ag electrode and nickel foam (200 mm length × 300 mm width × 1.6 mm thickness, MTI Korea) were used as the cathode and anode, respectively. The bipolar membrane (Fumasep FBM) was positioned between cathode and anode. The active area was controlled as 1 cm2. 3 M TREA (Sigma-Aldrich, 99%) was used as the CO2 capture solution and 1 M KOH (Sigma-Aldrich, >90%) was employed as anolyte. It is assumed that the TREA solution absorbs only CO2 from the flue gas. Therefore, 3 M TREA solution was saturated with CO2 for 1 h before eCO2R. The catholyte and anolytes were respectively provided from liquid reservoirs to MEA electrolyzer and continuously circulated by using a peristaltic pump. For comparison of CO2 captured solution, the 30 wt.% monoethanolamine or 30 wt.% diethanolamine were supplied instead of 3 M TREA. The eCO2R of all prepared Ag electrodes were examined by chronopotentiometry using a potentiostat (VSP, Biologic with booster 20 A) with current density range from −20, −50, −100, −150, to −200 mA cm−2. The gas products from the eCO2R were periodically quantified by online GC with a thermal conductivity detector and a flame ionization detector (Agilent 7890). MolSieve 5 A (6FT, Agilent) and Hayesep D (11FT, Agilent) packed columns were equipped, and ultrapure Ar (99.9999%) was used as the carrier gas. The products from CO2 conversion were CO and H2, and the total FE was confirmed to be ~100%. The FE of the products (i.e., H2 and CO) was calculated by the ratio of each partial current of the products to the total current.