A comprehensive overview of the electrochemical mechanisms in emerging alkali metal–carbon dioxide batteries

Abstract

Alkali metal–carbon dioxide (Li/Na/K-CO₂) batteries are emerging electrochemical energy storage technologies in the context of the energy crisis and the urgent demand for carbon neutrality. Alkali metal–CO₂ batteries offer a new strategy for CO₂ fixation and utilization, and thus has been receiving considerable attention in recent years. Considerable progress has been achieved since alkali metal–CO₂ batteries were invented, especially in terms of development of new electrode materials, and yet, research is lacking on the underlying mechanisms of the systems. This is the first typical review focusing on the electrochemical mechanisms of metal–CO₂ batteries that summarizes the current understanding of and provides insights into the thermodynamic reaction pathways, the kinetic characteristics, and the crucial factors determining the reaction mechanisms in alkali metal–CO₂ batteries. The review starts with the fundamental concepts of alkali metal–CO₂ batteries, followed by a comprehensive discussion of the working mechanisms on cathodes and anodes. Moreover, the operation mechanisms of state-of-the-art electrolytes, including liquid and (quasi-)solid-state electrolytes, are also described. Finally, we identify the unsolved problems in current alkali metal–CO₂ batteries and propose potential topics for future research.

INTRODUCTION

With economic development of the modern society, the demand for fossil energies increased rapidly, which has led to severe pollution and emission problems.^1,2 The continuous consumption of fossil energy generates enormous emission of CO₂ greenhouse gases, which has a significant impact on the environment.^3,4 To mitigate the resulting climate change and greenhouse effect, ambitious legislations have been established by many advanced economies worldwide, targeting net-zero greenhouse gas emission before the 2050 s.⁵ Under these circumstances, the development of efficient and clean battery systems has become one of the top priorities of current research. Among many battery systems, the lithium–ion battery (LIB) is the most widely used one.⁶ However, the application of LIBs in certain scenarios was constrained by the relatively low energy density. To solve this problem, lithium–air batteries with ultrahigh energy density have been developed.⁷ However, lithium–air batteries are more like lithium–oxygen batteries because they have to work in a pure oxygen environment rather than in ambient air.^8,9 In 2011, Takechi and co-workers¹⁰ found that adding a moderate amount of CO₂ to the Li–O₂ battery atmosphere (pure O₂) could greatly increase the discharge capacity of the battery (up to almost 3 times), leading to the creation of the original Li–O₂/CO₂ battery. Following this line of thinking, the first Li–CO₂ battery that worked in a pure CO₂ atmosphere was developed in 2013, which, in the next several years, rapidly became a promising strategy for CO₂ fixation and utilization.¹¹ As a unique trait of electrochemical energy storage systems, the operation of a Li–CO₂ battery enables the capture and reduction of CO₂ accumulation in the atmosphere.¹² More importantly, the special charge storage principle endows the Li–CO₂ battery with an ultrahigh theoretical energy density of 1876 Wh kg⁻¹,¹³ which is much higher than that of the traditional Li-ion batteries (LIBs) utilizing graphite/Si–C anodes and ceramic cathodes (200–350 Wh kg⁻¹).^14–16 Other potential application scenarios such as use as power sources in Mars exploration (96% of CO₂ in the atmosphere) have also attracted researchers' attention in developing a Li–CO₂ battery as a prospective energy storage system in the future.^8,17,18

In recent years, for the development of lithium-based batteries, cheaper substitutive elements other than lithium have also been considered due to the scarcity of lithium sources on Earth. This trend also applied for Li–CO₂ batteries, which derives into new systems including Na–CO₂,^19,20 K–CO₂,^21,22 Zn–CO₂,^23,24 and Al–CO₂ batteries.^25,26 Basically, all these systems utilize metal-based materials (Li, Na, Zn, Al, etc.) in the anode side and are thus referred to as metal–CO₂ batteries hereafter. This review focuses on the more widely investigated alkali metal (Li, Na, K)–CO₂ batteries.

Despite the aforementioned advantages, some limitations need to be overcome for practical applications. Most of the full-cell-level issues for alkali metal–CO₂ batteries such as the large discharge/charge polarization (low energy efficiency) and severe capacity fading are closely related to the solid-phase products by the end of discharge. Most solid-state products obtained upon discharge are thermodynamically stable and thus their decomposition upon charging is very difficult, resulting in high charge potential, extremely sluggish charge kinetics, and side reactions involving the electrolyte.^27,28 Also, the incomplete consumption of discharge products leads to the accumulation of an inert coating on the cathode, which is a major reason for the capacity decay and quick failure of the batteries.²⁹

To address these issues, most efforts in current research focus on developing highly active electrocatalysts to accelerate the CO₂ reduction reaction (CO₂RR) and evolution reaction (CO₂ER). Various material categories, for example, carbons,^{17,18,22,30–32} substrate-supported metal single atoms/atomic clusters,^33–36 and transition-metal oxides/sulfides/phosphates/carbides,^29,37–43 and so forth, were explored as bifunctional CO₂RR/ER catalysts in alkali metal–CO₂ batteries. However, the reaction kinetics of the CO₂ cathode, the charge/discharge polarization, and the cycling stability of alkali metal–CO₂ batteries still do not meet practical requirements. In fact, according to reports over the past several years, there is an apparent bottleneck in improving the electrochemical performance of metal–CO₂ batteries by simply conducting different trials on CO₂RR/ER catalysts. There is a severe lack of in-depth studies on the fundamental electrochemical mechanisms of the alkali metal–CO₂ batteries; therefore, limited guidance exists for the design and optimization of CO₂ catalysts, metal anodes, and electrolytes. Different from the studies of electrocatalyst characterization and electrochemical measurements, in-depth studies of electrochemical mechanisms are more complex. Due to the large variety of catalyst materials and discrepancies in experimental conditions, the different experimental phenomena observed were influenced by many uncontrollable factors, which lead to controversy on the understanding and conclusions about the working mechanism in alkali metal–CO₂ batteries. Many studies have included characterization data and discussions on reaction intermediates/products and reaction kinetics for the CO₂ cathode, the metal anode, and electrode/electrolyte interphases.^35,43,44 Therefore, a comprehensive summary of previous studies is highly desirable, based on which in-depth comparisons and analyses of the electrochemical mechanisms could be conducted in alkali metal–CO₂ batteries.

In this review, we provide a comprehensive overview of the electrochemical mechanisms of alkali metal–CO₂ batteries. Different from most previous reviews that mainly included material preparation and electrochemical performances, this review focuses on the mechanisms involved in the electrochemical systems. Electrochemical reactions, chemical reactions, mass transportation, charge transfer, and so forth that occur on CO₂ cathodes, metal anodes, electrolytes, and electrode/electrolyte interfaces are fully covered. Moreover, this review also identifies some unknown aspects of the electrochemical mechanism in alkali metal–CO₂ batteries that need further in-depth studies.

THE CONFIGURATION OF ALKALI METAL–CO₂ BATTERIES

Alkali metal–CO₂ batteries (Li–CO₂ batteries, Na–CO₂ batteries, and K–CO₂ batteries) have attracted considerable attention because of their superhigh theoretical energy densities and the capability of directly converting CO₂ gas.⁴⁵ Considerable efforts have been made to develop advanced alkali metal–CO₂ batteries during the last decade. One clear evidence is the total papers published per year focused on “Li–CO₂ battery”, “Na–CO₂ battery,” and “K–CO₂ battery” (Figure 1A). After the first scientific report of Li–CO₂ batteries by Archer's team in 2013,¹¹ reports on Na–CO₂ batteries and K–CO₂ batteries also appeared in 2016 and 2018, respectively.^19,21 Similar to the Li-ion batteries in terms of the general configuration, an alkali metal–CO₂ battery is also composed of an anode, an ion conductive separator, an organic electrolyte, and a cathode. Figure 1B shows a schematic of the Li/Na/K–CO₂ batteries. The anodes are routinely alkali metal foils (Li, Na, or K metal foils) that provide sufficient alkali metal ions (Li⁺, Na⁺, or K⁺ ions) for the battery system. Essentially different from traditional lithium-ion batteries, which employ solid-state Li⁺ intercalation cathodes,⁴⁷ the reactants in the cathode side of the alkali metal–CO₂ batteries are CO₂ gas and the bifunctional CO₂ reduction/evolution reaction (CO₂RR/ER) catalyst materials. The electrolyte basically consists of an organic solvent and dissolved alkali metal salts. Generally, CO₂ is soluble in organic ester/ether solvents, which is very favorable for electrocatalytic reactions. Also, high CO₂ solubility in the electrolyte solvent is preferred. The ion-conductive and electron-insulate separators separate the anode and the cathode to avoid internal short circuit.

Figure 1. (A) Number of publications related to Li/Na/K–CO2 batteries indexed in the Web of Science database. (B) Schematic of the Li/Na/K–CO2 batteries configurations. Reproduced with permission: Copyright 2016, Elsevier.46

In typical alkali metal–CO₂ batteries, alkali metal anodes lose electrons and become the corresponding alkali metal ions during discharging; meanwhile, the alkali metal ions react with CO₂ to produce discharge products on the cathode side. While charging, the formed discharge products decompose into alkali metal ions and CO₂ on the cathode side, and the alkali metal ions accept electrons and are reduced into alkali metals on the anode side. Based on the electron transferred and the molar mass of alkali metals and CO₂, alkali metal–CO₂ batteries can deliver superior theoretical energy density (1876 Wh kg⁻¹ for Li–CO₂ batteries,¹³ 1130 Wh kg⁻¹ for Na–CO₂ batteries,²⁷ and 923 Wh kg⁻¹ for K–CO₂ batteries), which are much higher than that of commercial LIB (graphite–LiFePO₄, graphite–LiCoO₂, graphite–NMC, Si/C–NMC, etc.), as shown in Figure 2. In addition to the higher energy densities, the advantages of an environmentally friendly nature and unique applicability in a CO₂ environment have promoted the development of alkali metal–CO₂ batteries.

Figure 2. Comparison of the energy densities for Li/Na/K–CO2 batteries versus lithium-ion batteries with various anode/cathode couples (G: graphite, LFP: LiFePO4, LCO: LiCoO2, 333/811: LiNi0.3/0.8Mn0.3/0.1Co0.3/0.1O2, NCA: LiNi1−x−yCoxAlyO2), Li/Na/K–S batteries and Li/Na/K–O2 batteries.

Despite the attractive advantages, the alkali metal–CO₂ batteries significantly suffer from poor CO₂ reduction kinetics and difficult electrochemical decomposition of the discharge product, which lead to the limited battery life, low energy efficiency, and poor rate performance. These issues have markedly hindered the practical application of Li/Na/K–CO₂ batteries. Therefore, a catalyst material is indispensable in the cathode side of the alkali metal–CO₂ batteries. Figure 3 shows the classification of the previously reported CO₂RR/ER catalysts for alkali metal–CO₂ batteries. Various types of catalyst materials including carbon materials, transition-metal nanomaterials, transition-metal compounds, single metal atom materials, and noble metal/alloy nanomaterials were used to enhance the CO₂RR and CO₂ER kinetics and improve the electrochemical performance of alkali metal–CO₂ batteries. Carbon materials are widely used as bifunctional CO₂RR/ER catalysts in alkali metal–CO₂ batteries due to the advantages of the large surface area, high electrical conductivity, good chemical stability, and adjustable active sites.^48,49 Apart from the common Ketjen Black (KB)⁵⁰ and Super P¹¹, nano carbon materials such as carbon nanotube (CNT),⁵¹ graphene,^52,53 graphdiyne,³² nanoporous carbons,^54,55 and so forth with better CO₂RR/ER catalytic activities have been intensively studied. Gao's group used different carbon materials with various pore structures as cathodes of Li–CO₂ batteries to determine the catalytic performance in terms of pore shape, pore size, and surface area.⁵⁶ It has been proven that the pore shape plays a more important role in the Li₂CO₃ deposition among these three factors. The catalytic activities of carbon materials with two-/three-dimensional (2D/3D) interlayer mesopores are superior to the carbons with ink-bottle mesopores or 1D mesopores. In addition, the large pore size favors the resistance of pore blocking by the discharge product and the large specific surface area provides a large number of catalytic active sites for the cathodic reactions. Moreover, the CO₂RR/ER catalytic performance of carbon materials can be improved using the heteroatom doping (e.g., N,^30,57 B,⁸ S,^17,58 P,⁵⁹ O⁶⁰) strategy by modulating the intrinsic electronic structure of the active sites and the electrical conductivity of the graphitic matrices. For instance, the existence of pyridinic nitrogen promotes electron mobility and the decomposition of the discharge products. Wang and co-workers reported that carbon nanotube networks with nitrogen doping promote rapid Li⁺ and CO₂ diffusion to active catalytic sites.⁶¹ Dai's team used N,S-doped carbon nanotubes in Li–CO₂ batteries.¹⁷ S atoms with a spin density different from that of C atoms served as the active sites of the charge process. N atoms promoted charge transfer from adjacent conjugated C atoms, thus accelerating CO₂ reduction and Li₂CO₃ deposition during discharge. The quasi-solid-state flexible Li–CO₂ batteries with a N,S-doped CNT cathode showed a discharge capacity of 23,560 mAh g⁻¹ and could operate stably for 110 days. However, the lack of long-term stability and the low efficiency of alkali metal–CO₂ batteries with carbon material cathodes are still key problems that need to be solved. The poor intrinsic catalytic activity of carbon materials and the structural degradation during cycling lead to the accumulation of discharge products with poor conductivity on the surface blocking active sites. It should also be pointed out that the unstable carbon materials could participate in the reaction during the charge process, which greatly interferes with the cycle stability of the batteries.⁶² Therefore, carbon materials with high conductivity and versatile structures were often used as the matrix in previous studies, and other species were introduced as active sites to promote CO₂RR/ER.

Figure 3. Classification of the previously reported bifunctional CO2RR/ER catalysts in alkali metal–CO2 batteries

Noble metal/alloy nanomaterials show superior CO₂RR/ER catalytic activities and high selectivity due to the unique d-band structures. Various noble metal elements, like Ru,^{20,35,63–67} Ir,^68–70 Au,⁷¹ and Rh,⁶⁶ were found to have excellent intrinsic CO₂RR/ER electrocatalytic capabilities. Ding and co-workers assembled Li–CO₂ batteries with Ru atomic clusters (Ru_AC) and single-atom Ru–N₄ (Ru_SA) composite sites on a carbon nanobox substrate (Ru_AC+SA@NCB) as the electrochemical catalyst.³⁵ The battery can operate stably for more than 60 cycles (~230 h) at voltages ranging from 2.5 to 4.5 V. At current densities of 100, 1000, and 2000 mA g⁻¹, the overpotentials of the Li–CO₂ batteries using the Ru_AC+SA@NCB cathode were only 1.05, 1.65, and 1.86 V, respectively. Theoretical calculations showed that the superb electrochemical performance of Ru_AC+SA@NCB is due to the synergy effect between the Ru_SA and the Ru_AC. The Ru_AC modified the electronic structure of the Ru_SA, thereby optimizing the adsorption of key intermediates and lowering the energy barrier of the rate-determining step during the CO₂RR and CO₂ER process. Moreover, modifying the electronic structure of noble metals via alloying engineering,^66,67,72 ligand engineering,⁵⁹ or metal–support interaction⁶⁴ can also improve the CO₂RR/ER kinetics and change the reaction mechanisms of alkali metal–CO₂ batteries. Guo's group assembled Li–CO₂ batteries with ultrathin triangular RuRh alloy nanosheets as an exceptionally active catalyst.⁶⁶ According to density functional theory (DFT) calculations, after the introduction of Rh, Rh 4d-orbitals and Ru 4d-orbitals were hybridized, which improved the electron transfer ability of the Ru surface and balanced the CO₂ binding strength near the Ru site. The Li–CO₂ battery with a RuRh alloy nanosheet catalyst can stably discharge and charge up to 180 cycles at 1000 mAh g⁻¹ without capacity degradation. During the cycle, the potential gap between the terminal discharge and charge potentials is always maintained at less than 1.35 V. Fan's team developed a facile one-pot solvothermal method to synthesize a series of ultrathin 2D Ru–Co/Ni/Cu nanosheets as catalysts for Li–CO₂ batteries.⁷³ The abundant in-plane Ru–Co alloy active sites, on the one hand, promoted electron interaction with discharge products to enhance the kinetics of CO₂ER, and on the other hand, enhanced adsorption toward Li and CO₂ to boost the reaction rate of CO₂RR.

Considering the high costs of the noble metals, fabrication of non-noble transition-metal/compound nanomaterials as CO₂RR/ER catalysts has become a hot topic in the field of alkali metal–CO₂ batteries. Transition-metal-based catalysts are rich in abundance and low in price. By proper structural adjustment, catalytic activity equivalent to that of noble metal-based catalysts can be achieved. It is noteworthy that more different routes of CO₂ redox have been discovered with the extensive exploration of various transition compounds.^41,45,74 Transition-metal oxides are one of the CO₂RR/ER catalysts widely used in alkali metal–CO₂ batteries. Zhang's group fabricated a self-supporting cathode of 2D Co-doped CeO₂ nanosheets on a graphene aerogel for Li–CO₂ batteries.⁷⁵ According to DFT calculations, the introduction of Co atoms changed the electron cloud density of Ce and O atoms and promoted the reversible formation of Li₂CO₃. The assembled Li–CO₂ batteries achieved a discharge capacity of 7860 mAh g⁻¹ at 100 mA g⁻¹ and excellent cycling stability (>100 cycles). In addition, transition metal sulfides,⁷⁶ carbides,⁴¹ and nitrides⁷⁷ have also been used in alkali metal–CO₂ batteries. However, the main reason for the deactivation of transition-metal-based catalysts is the aggregation or shedding caused by the unstable structure of metal or compound nanoparticles during the discharge and charge process, which, on the contrary, reduces the catalytic activity toward CO₂RR/ER.

Due to the theoretical 100% atom utilization and excellent catalytic properties in many electrocatalytic systems, single metal atom materials are also used as CO₂RR/ER catalysts in alkali metal–CO₂ batteries.^33,78 Because of the low-coordination status, the quantum size effect, and strong metal–support interaction, single-atom catalysts show exceptional catalytic activity/selectivity toward CO₂RR/ER. More importantly, the electronic structure of the active sites in single metal atom catalysts can be largely adjusted by modulating the coordination environments of the central metal atom, and as a result, the catalytic activity and selectivity can be effectively tailored. These features make single metal atom materials a highly promising catalysts for the emerging alkali metal–CO₂ batteries. Cheng's team synthesized N-doped graphene-supported single metal atom catalyst SAMe@NG/PCF (Me = Mn, Fe, Co, Ni, and Cu).⁷⁹ The relatively weak electronegativity and the high d-band center of Cr confer high CO₂ adsorption strength, resulting in excellent cycling stabilization of 350 cycles at 100 μA cm⁻². Gu and co-workers used in-situ environmental TEM to monitor the changes in Pt single-atom-modified N-doped carbon nanotubes (Pt@NCNT) as cathode catalysts for Na–CO₂ batteries during charge and discharge.⁸⁰ Na₂CO₃ formed on the surface of Pt@NCNT in the shape of balls during the discharge process, while they decomposed into Na⁺ and CO₂ in the charge process. SACs agglomerate easily and couple to form large clusters during synthesis or electrochemical reactions due to the large free energy of individual metal atoms, which leads to catalyst degradation. Therefore, stability and loading are major challenges for SACs. Overall, the application of SACs in alkali metal–CO₂ batteries is in the infancy stage and further research is required. The current research of developing CO₂RR/ER catalysts for alkali metal–CO₂ batteries can also be extended to other categories, such as metal–organic framework (MOF)-based materials,⁸¹ covalent organic framework (COF)-based materials,⁸² MXene-based materials,^40,83 and so forth.⁴⁴ Due to the adjustable pore channels for CO₂ capture, MOF- and COF-based materials have enormous potential in CO₂ enrichment and electrocatalytic reduction. However, their synthesis procedures are generally cumbersome and thereby difficult to use for large-scale production.⁸⁴

To date, alkali metal–CO₂ batteries only work in high CO₂ concentration environments, which raise special requirements toward the cell structure. The setup should be such that sufficient CO₂ gas is sent to the catalyst materials in the cathodes to maintain the operation of the batteries. Figure 4 summarizes several types of setups for testing alkali metal–CO₂ batteries in the literature. For the coin cells routinely utilized for LIB tests in laboratories, the cathode shells are open with holes for CO₂ diffusion. Besides, the coin cells are normally placed in poly tetra fluoroethylene (PTFE) and polyvinyl chloride (PVC) tape sealed CO₂ bottles (Figure 4A,B).^29,35 Some more complex setups with adjustable CO₂ pressures were also used. For example, Figure 4C shows a Li–CO₂ pouch battery and the corresponding CO₂ container.⁸⁵ Both the air vent and the electrical wire have good airtightness. A Swagelok cell is another frequently used setup for testing alkali metal–CO₂ batteries. Figure 4D,E shows two types of Swagelok cells in the literature.^59,86 The external gas circuits can provide controllable CO₂ gas flow throughout the alkali metal batteries. Moreover, Swagelok cells can also be connected to many extended equipment, such as an in situ differential electrochemical mass spectrometer (DEMS) for more advanced gas phase product characterization, as shown in Figure 4E.

Figure 4. (A) Digital photograph of the assembled Li–CO2 coin cells and the home-made setup for electrochemical measurements. Reproduced with permission: Copyright 2022, Wiley-VCH.35 (B) Photograph of the CO2 bottle for the coin cell test. Reproduced with permission: Copyright 2020, Wiley-VCH.29 (C) CO2 chamber for Li–CO2 pouch cell tests. Reproduced with permission: Copyright 2017, Wiley-VCH.85 (D) Li–CO2 Swagelok cell and the schematic of the configuration. Reproduced with permission: Copyright 2018, Elsevier.86 (E) Photograph of the customized Swagelok cell and the corresponding schematic illustration of the DEMS analysis system. Reproduced with permission: Copyright 2018, Wiley-VCH.59

ELECTROCHEMICAL MECHANISM OF CO₂ CATHODES

Basically, the thermodynamic reaction pathway is the foundation of the electrochemically active electrodes in batteries. Therefore, we first focus on the crucial experimental data revealing the different reaction pathways of the CO₂ cathodes during charge and discharge processes. As the most commonly investigated systems, Li–CO₂ batteries are discussed first and other lithium (sodium/potassium) metal–CO₂ batteries are covered in a separate section.

Discharge mechanism of CO₂ cathodes in Li–CO₂ batteries Li₂CO₃ and carbon as the discharge products

Studies on the reaction pathway of CO₂ cathodes during the discharge process have been conducted since the Li–CO₂ battery system was first proposed. Like other gas reagents involving electrodes like the cathodes in fuel cells and metal–O₂ batteries, the reactions occurring in the CO₂ cathodes also involve gas–liquid–solid tri-phase interfacial electrochemical charge-transfer processes. In 2013, Archer's group, for the first time, reported the primary Li–CO₂ battery operating at high temperatures.¹¹ To identify the discharge route of the Li–CO₂ battery, the researchers made the hypothesis that the discharge process is identical to the chemical reaction between Li metal and gas CO₂, which is shown as the following equation: [Image Omitted. See PDF]

According to the Nernst equation $$E=-{\rm{\Delta }}G/{zF}$$, where ΔG, z, and F, respectively, represent the Gibbs free energy change, the number of electrons transferred per mole of product, and the Faraday constant, the theoretical discharge potential can be calculated. The dashed line in Figure 5A shows the theoretical discharge potential as a function of temperature. However, the experimental potential at a high temperature surpassed the theoretical values, which indicated the inaccuracy of the reaction equation due to the violation of the Tafel theory.

Figure 5. (A) Comparison of the theoretical equilibrium potential with the actual discharge potential. The theoretical equilibrium potential 1 was calculated based on the reaction 2Li + 2CO2 → Li2CO3 + CO. The theoretical equilibrium potential 2 was calculated based on the reaction 4Li + 3CO2 → 2Li2CO3 + C. (B) Ex situ XRD results of the cathodes after discharge at 100°C. Peaks marked by symbol * are the characteristic peaks of Li2CO3 and peaks marked by symbol # are from the aluminum substrate. Reproduced with permission: Copyright 2013, Royal Society of Chemistry.11 (C) EELS of the porous gold electrode in the first discharge state in the Li–CO2 battery. Reproduced with permission: Copyright 2014, Royal Society of Chemistry.50 (D) CO2 moles detected during the first discharging process after the cell was discharged for 1600 mAh g−1. (E) CO2 moles detected before and after the charge process in the first cycle. (F) Potential-dependent mechanistic analysis of CO32− ion formation on a MoS2/IL cocatalyst that can lead to an increase in the Li2CO3/C discharge product. Reproduced with permission: Copyright 2019, Wiley-VCH.45

To determine the real discharge product, DEMS was applied to analyze the change of the gas phase in the battery during discharge. The absence of CO in the gas phase product further indicated the false assumption of Equation (1). Therefore, another reaction pathway was proposed as shown in the following equation: [Image Omitted. See PDF]

Similar to Equation (1), lithium carbonate (Li₂CO₃) is also the main discharge product in Equation (2). The difference is that no gas phase product appears in the equation. The reduction product of CO₂ is the solid carbon phase. The theoretical potential calculated based on Equation (2) is shown as the dotted line in Figure 5A. All the experimental potential data are below the line, which satisfied the Tafel theory. Moreover, the product of lithium carbonate at the end of discharge was verified by ex situ X-ray power diffraction (XRD, Figure 5B) and spectra Fourier transform infrared (FTIR) results. However, the formation of carbon has not been confirmed due to the lack of an effective characterization technique because the electrodes routinely contain the carbon phase in the pristine state.

Many other groups proposed the same reaction pathway as Archer in Equation (2), but using different catalysts.^45,50,52,57 For example, Li's group assembled a rechargeable Li–CO₂ battery with a KB cathode and a lithium triflate (LiCF₃SO₃)–TEGDME electrolyte.⁵⁰ The XRD analysis on the discharged KB cathode also identified Li₂CO₃ as the main discharge product. However, since the cathode is based on carbonaceous material, it is difficult to distinguish the carbon product generated upon discharge. To rule out the interference of the carbon phase in the pristine cathode, the authors also tested a KB-free cathode using a porous gold catalyst. Both Raman spectroscopy and electron energy loss spectroscopy (EELS, Figure 5C) showed the appearance of amorphous carbon at the end of discharge, which strongly verified the products in Equation (2). Moreover, the carbon formation was also confirmed in the KB-free platinum-based cathode. Nonetheless, the validity of Equation (2) has still not been confirmed because the reaction pathway may not be the same for Au/Pt-based catalysts as that for KB or other carbon-based catalysts.

Another effective strategy to determine the reaction pathway and the discharge product is to determine the e⁻/CO₂ ratio in the reaction, which is the number of electron transfers for one CO₂ molecule consumed. According to this criterion, Equations (1) and (2) are two- and four-electron reaction processes, respectively. The total transferred electron numbers can be recorded by the electrochemical workstation, and the number of CO₂ molecules involved in the reaction can be determined by DEMS. For instance, Salehi–Khojin's group determined the change of CO₂ in the atmosphere in which the Li–CO₂ battery worked by in situ DEMS.⁴⁵ According to the DEMS data (Figure 5D,E), the e⁻/CO₂ ratios for the discharge and charge processes were measured to be 4.05 and 4.07, respectively. The results indicated that a four-electron reaction occurred during discharge, which is a strong evidence of Equation (2). Moreover, the authors used an exclusive methodology to determine the most possible reaction pathway from the thermodynamics perspective based on quantum chemical wave-function-based calculations and density functional theory (DFT) studies. In addition to Equations (1) and (2), two other possible pathways were also included in the calculation, which are described as the the following equations: [Image Omitted. See PDF] [Image Omitted. See PDF]

First, the reaction in Equation (4) was eliminated because it is an endothermic process. According to the calculation, the reactions in Equations (1) and (3) are thermodynamically favorable. However, neither Li₂O nor CO was experimentally detected in the discharge products, which ruled out the feasibility of Equations (1) and (3). Only the reaction in Equation (2) is a four-electron, thermodynamically favorable reaction. Also, the predicted discharge products in the equation are in agreement with the experimental observations.

Second, to probe the detailed pathway by which the Li₂CO₃ and C discharge products were obtained, the potential dependences of the discharge reaction sub-steps were investigated by theoretical calculations. As shown in Figure 5F, all the calculated potentials are referenced to the Li/Li⁺ electrode. The adsorption of both the initial and the second CO₂ molecules is thermodynamically favorable, as evidenced by the decrease in energy by 1.58 and 0.60 eV, respectively. The adsorption of the two CO₂ molecules resulted in the formation of a co-adsorbed carbonate (CO₃*) species and carbon monoxide (CO*) species. Afterwards, the adsorbed CO₃* reacts with Li⁺ to form Li₂CO₃, which is a spontaneous process below 2.31 V, in accordance with the experimental results. However, how the process of amorphous carbon formation occurred still remains unclear, which is indispensable for the stoichiometric balance of the overall reaction. The DFT calculations indicate that the third CO₂ molecule might react with the edges containing adsorbed CO* to form CO₃* and carbon (C*), although this step is unfavorable in thermodynamics (energy increase by 0.94 eV). The investigators speculated that the presence of defect sites on the catalyst or carbon may facilitate the conversion of CO* into CO₃* and C*. Many other studies proved the validity of Equation (2) by analyzing the discharge products.^43,87,88 However, it is worthwhile to mention that because in most cases, as it was difficult to completely remove the carbon phase from the cathode catalysts, the verification of carbon formation in the discharge remains an essential challenge.

The investigation of the reaction mechanism for Li–CO₂ batteries is not just limited to the determination of end products. It is highly instructive to investigate how the catalysts are involved in the CO₂ conversion during discharging and charging. To achieve this aim, more comprehensive characterizations of the reaction intermediates are desired. As a representative example, Manthiram's team used molybdenum disulfide nanosheets (MoS₂-NS) as the CO₂RR/ER bifunctional catalyst in Li–CO₂ batteries.⁸⁹ In terms of the electrochemical performance, the Li–CO₂ battery delivered a discharge capacity of 846 μAh cm⁻² and an energy efficiency of 90.1% in the first cycle. The battery was able to cycle 50 times at a current density of 0.05 mA cm⁻² and a fixed capacity of 0.25 mAh cm⁻². In the study, characterization techniques including XRD, FTIR-attenuated total reflectance (FTIR-ATR), Raman, and X-ray photoelectron spectroscopy (XPS) were utilized to investigate the detailed catalytic mechanism of the MoS₂ catalyst. As shown in Figure 6, during discharge, CO₂ is reduced to oxalate (C₂O₄²⁻), which subsequently reacts with MoS₂ to form the intermediate complex of C₂O₄²⁻–Mo⁶⁺S₂⁴⁻ (Equation 5). This compound dissociates into carbon and CO₃²⁻-Mo⁶⁺S₂⁴⁻ via a two-step process, as elaborated in Equations (6) and (7). Thereafter, CO₃²⁻–Mo⁶⁺S₂⁴⁻ reacts with Li⁺ to form Li₂CO₃ and MoS₂ (Equation 8). [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

Figure 6. (A) XRD, (B) FTIR-ATR, and (C) Raman spectra of MoS2-NS@MWNT/SWNT gas diffusion cathodes from pristine, discharged, and charged Li–CO2 cells. (D) Graphical depiction of the catalytic mechanism of Mo-terminated edge sites of MoS2-NS toward CO2RR/ER. (E) Mo 3d XPS spectra of MoS2-NS@MWNT/SWNT electrodes collected from pristine, discharged, and charged Li–CO2 batteries. Reproduced with permission: Copyright 2019, American Chemical Society.89

As we can see, in the discharge process, there are complex electron transfers between the reagents and the MoS₂ catalyst. The change in Mo valence was revealed by Mo 3d XPS. To highlight the conversion of carbon species, the MoS₂ catalysts are removed from Equations (5)–(8) and Equations (9)–(12) were obtained. [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

Equation (9) is a one-electron reaction in which CO₂ is reduced to C₂O₄²⁻. Subsequently, the unstable C₂O₄²⁻ decomposes to CO₃²⁻ and carbon via a two-step chemical reaction (Equation 10 and 11). Finally, CO₃²⁻ combines with Li⁺ to generate a Li₂CO₃ deposit on the catalyst surface.

Dai's group described more detailed discharging reaction pathways with intermediates in a Li–CO₂ battery, as shown in Figure 7A.⁷⁸ Three possible pathways were proposed. First, because the *COLi intermediate is thermodynamically unstable, the feasibility of pathways II and III was excluded. Pathway I is a widely accepted discharge reaction path reported in previous studies.^33,35 Figure 7B,C demonstrates the sub-steps of pathway I and the corresponding intermediates, responsively. During discharge, two CO₂ molecules successively adsorb on the active site and form C₂O₄*. Subsequently, Li⁺ reacts with C₂O₄* to form LiC₂O₄* and CO*, followed by the detachment of one Li₂CO₃ molecule. Afterwards, the third CO₂ molecule adsorbs and produces C₂O₃*. Then, two Li⁺ molecules react with C₂O₃* in sequence, generating LiC₂O₃* and C*, respectively. Finally, the second Li₂CO₃ molecule and C detach from the catalyst.

Figure 7. (A) Possible reaction pathways and the corresponding intermediates on catalysts in Li–CO2 batteries during the discharge process. The symbol * refers to the active sites of the catalysts. Reproduced with permission: Copyright 2020, Wiley-VCH.78 (B) Simulation path and possible reaction intermediates of a Li–CO2 battery during discharge. (C) Simulation of reactants and reaction intermediates on the “SA Ru–Co3O4” active sites during discharge. Reproduced with permission: Copyright 2021, Wiley-VCH.33 (D) Possible reaction pathways for the formation of Li2CO3 and amorphous carbon. (E) Calculated possible energy profiles for the nucleation of Li2CO3 and C on the basal plane of five samples at a theoretical equilibrium potential. The insets show the top views of adsorption systems on the 3N3SV–ReS2 plane. Reproduced with permission: Copyright 2022, American Chemical Society.76

Cheng's team also proposed the possible pathways for the formation of Li₂CO₃ and C during discharge, as shown in Figure 7D.⁷⁶ The entire discharge process can be divided into two stages: Li₂CO₃ nucleation with a *CO intermediate (the first stage) and a C atom (the second stage). There are five paths for the first stage and only one path for the second stage. Nonetheless, the reaction steps between *CO and CO₂ in the second stage are complex and still unclear. Thus, in this work, the authors used a simplified reaction step to study the possible reaction energy profiles, as shown in Figure 7E. The authors also emphasized that it is difficult to analyze the actual reaction kinetics of the second stage following this simplified reaction path. More importantly, it should be noted that although many catalysts may share the same reaction pathway, the rate-determining step for different catalysts is not necessarily the same.

Li₂CO₃ and CO as the discharge products

In Equation (2), CO₂ is reduced to carbon (C⁰) in discharge, which is a four-electron reaction. From the perspective of reaction kinetics, the reduction of CO₂ to CO is expected be more facile due to the two-electron transfer. Therefore, it is reasonable to make the hypothesis that Li₂CO₃ and CO are the discharge products in a Li–CO₂ battery. By far, there have been many studies to validate this process. For instance, Wang's group assembled a Li–CO₂ battery with a 3D porous fractal Zn (PF–Zn) catalyst, which was completely carbon free.⁹⁰ High-resolution TEM (HRTEM) of the discharge product revealed small crystals with lattice spacings of 0.26 and 0.28 nm on the cathode surface, which were, respectively, ascribed to (−112) and (002) planes of Li₂CO₃. In addition, XPS and Raman spectrum also confirmed the presence of Li₂CO₃. Interestingly, carbon cannot be detected by any characterization technique in the experiments. The gas composition in the enclosed space where the batteries operated was analyzed by gas chromatography (GC). To conduct a quantitative analysis, the Faradaic efficiency (FE) of the products was calculated based on the total amount of charge transfer. The GC and calculation results are shown in Figure 8A,B, respectively. Gases of CO₂, H₂, and CO were detected by GC. CO₂ was the supply gas and H₂ was derived from the reaction between Li metal and residual water in the system. The authors claimed that CO could be one of the discharge products. As shown in Figure 8A, Li–CO₂ batteries produced CO with 3.6% FE at 0.01 mA, which increased to 67% FE at 0.1 mA. Figure 8B demonstrates the FE change of CO as a function of time at 0.1 mA. A continuous increase in FE with CO generation was observed over a time span of 10 h. These analyses, to some extent, proved that Li₂CO₃ and CO are the discharge products in Li–CO₂ batteries, and there is no formation of carbon. This process can be described in Equation (13). [Image Omitted. See PDF]

Figure 8. (A) Max FE of CO at several currents during 10 h discharge. (B) FE of CO during long-term discharge at 0.1 mA. Reproduced with permission: Copyright 2018, Royal Society of Chemistry.90 Galvanostatic discharge profiles of the Li–CO2 cell with a configuration of (sputtered gold cathode)/(CO2-saturated 0.5 M LiClO4–DMSO electrolyte)/lithium foil. The capacity is limited at (C) 10 and (D) 20 mAh. Capacity-dependent in situ Raman spectra recorded during the corresponding discharge processes above (E) 10 mAh and (F) 20 mAh cut-off capacities, respectively. Reproduced with permission: Copyright 2017, Elsevier.12

Li₂O and carbon as the discharge products

The production of Li₂CO₃ can be validated as long as the Li–CO₂ batteries follow the discharge pathways shown in Equations (1) and (13). In this case, the CO₂ involved in the discharge is not fully reduced because the valence of carbon in Li₂CO₃ is still +4. As the battery was discharged to a lower cut-off voltage, no C⁴⁺ can remain and the final products are Li₂O and carbon. For instance, Zhou's group discharged a Li–CO₂ battery to 20 μAh (~1.75 V).¹² As shown in Figure 8C,D, two plateaus can be observed at ~2.5 V (red trace) and 1.8 V (blue trace) in the discharge profile. The red traces indicate the formation of Li₂CO₃ and carbon (Equation 2), which is proved by Raman data in Figure 8E. For the product obtained at a lower cut-off voltage, a new peak appeared at 520 cm⁻¹ in the Raman spectrum (Figure 8F), which is ascribed to the Li₂O phase. Combining the characterization results, the potential of this new four-electron reaction pathway was determined to be 1.89 V versus Li/Li⁺ by theoretical calculation. This mechanism was proposed as the following equation: [Image Omitted. See PDF]

Li₂C₂O₄ and carbon as the discharge products

According to the aforementioned research, C⁰, C²⁺, and C⁴⁺ species have been discovered in the reduction products in Li–CO₂ batteries. It is expected that C³⁺ species, as an intermediate state of C²⁺ and C⁴⁺ species, can also appear in the discharge process if proper catalysts or discharge procedures are used. For instance, a reaction mechanism with Li₂C₂O₄ as the discharge product was proposed by Chen's group.⁴¹ In their Li–CO₂ cells, Mo₂C/CNT was utilized as the cathode catalyst. High energy efficiency of 77% and good stability up to 40 cycles were obtained for the batteries. XRD was used to investigate the discharge products on a Mo₂C/CNT cathode. Interestingly, negligible change was found between the pristine and discharge patterns, as shown in Figure 9A. This observation was apparently not in agreement with the discharge mechanisms described in sections 3.1.1, 3.1.2, and 3.1.3, because no crystalline phase (neither Li₂CO₃ nor Li₂O) formed. Therefore, the reduction product should be in an amorphous phase. Using Raman and XPS analyses, the discharge product was finally determined to be Li₂C₂O₄. The corresponding reaction mechanism was proposed as the following equation: [Image Omitted. See PDF]

Figure 9. (A) XRD patterns of Mo2C/CNT electrodes at different stages (pristine Mo2C/CNT: black line; discharged Mo2C/CNT: red line), with the inset showing an enlargement of the indicated range. Reproduced with permission: Copyright 2017, Wiley-VCH.41 (B) XRD patterns of CC@Mo2C NP electrodes after the first discharge process. (C) High-resolution XPS spectrum of C 1s for the CC@Mo2C NP electrode at a discharged state. Photographs depicting two bottles of solutions containing different discharge products of Mo2C and CNT, respectively, after (D) dissolving the discharge product in oxygen-free distilled water, (E) adding a few drops of a saturated CaCl2 solution, and (F) following excessive addition of 1 mol H3PO4. Reproduced with permission: Copyright 2019, Wiley-VCH.74

As another example, Wang's group also observed the discharge product of Li₂C₂O₄ instead of Li₂CO₃ in the Li–CO₂ battery with a Mo₂C-based catalyst cathode.⁷⁴ The XRD pattern demonstrated that no Li₂CO₃ was formed after discharge, as shown in Figure 9B. Instead, Li₂C₂O₄ was found to be the discharge product. Generally, C₂O₄²⁻ is unstable and easily disproportionates into CO₃²⁻ and carbon, as demonstrated in Equations (10) and (11). The Mo₂C catalyst cathode used here played a special role. The C₂O₄²⁻ anion can connect to the Mo₂C catalyst via a Mo–O bond, facilitating the electron transfer from Mo²⁺ and Mo³⁺ to O atoms in C₂O₄²⁻. It was believed that the delocalized electrons can stabilize the C₂O₄²⁻ anion and lead to the formation of amorphous Li₂C₂O₄, which is described in Equation (16). An ingenious precipitation method was used to confirm the existence of C₂O₄²⁻ in the discharge products. As shown in Figure 9D–F, white precipitation appeared after CaCl₂ and H₃PO₄ were added in the Mo₂C bottle. By contrast, no change occurred in the CNT bottle. The existence of stable C₂O₄²⁻ in the discharge product was confirmed because H₂C₂O₄ is the only stable acid that contains C, H, and O and has stronger acidity than H₃PO₄. Besides, the formation of Li₂C₂O₄ was also confirmed by XPS (Figure 9C). Overall, the formation of Li₂C₂O₄ follows the pathways in Equations (9) and (16) during the discharge process. [Image Omitted. See PDF]

Overall, a new one-electron discharge mechanism of a Li–CO₂ battery can be described as the following equation: [Image Omitted. See PDF]

Based on the above discussions, it can be concluded that there are multiple types of discharge mechanisms for Li–CO₂ batteries in which different reduction products are involved. In most cases, Li–CO₂ batteries follow the reaction pathway of Equation (2) with Li₂CO₃ and carbon as the discharge products. Under certain conditions, reaction pathways of Equations (13), (14), and (17) are also feasible, which generate the discharge products of CO, Li₂O, and Li₂C₂O₄, respectively. The CO₂RR/ER catalysts used played an essential role of determining the reaction products and the pathways.

It is instructive to present a summary of the discharge mechanism mentioned in Section 3.1. First, the equations only demonstrate the thermodynamical end products during the discharge of a Li–CO₂ battery. The detailed reaction pathways like the reaction sub-steps and the reaction intermediates are not included in the equation. In fact, the CO₂ reduction reaction electrocatalysis is a highly complicated process comprising various intermediates,^{12,41,74,89,90} while in Li–CO₂ batteries, the CO₂RR pathway with the participation of Li ions is expected to be essentially different from that for CO₂RR electrocatalysis. Second, the reaction in Li–CO₂ batteries involves a complex gas–liquid–solid multiphase catalytic process at the interface, which includes gas diffusion, mass transportation, solid nucleation/growth, charge transfer, and so forth. These factors essentially affect the kinetic behaviors, which are also a vital aspect of the discharge mechanism. Third, a comprehensive survey of the previous studies indicates that the characteristics of the CO₂RR/ER catalysts essentially influence the discharge reaction pathway. Nonetheless, the intrinsic mechanism of the distinct electrocatalysis selectivity in the Li–CO₂ battery needs to be identified by further in-depth investigations.

Charge mechanism of CO₂ cathodes in Li–CO₂ batteries

Currently, the research of alkali metal–CO₂ batteries is mostly focused on secondary batteries. In terms of a rechargeable energy storage system, the Coulombic and energy efficiencies are directly determined by the capacity and potential of the charging process. Similar to the discharge mechanisms, our discussion of the charge mechanisms also mainly focuses on the reaction pathways. It is noteworthy that the starting point of the charging pathway is determined by the end products of the former discharge, which varies depending on the discharge mechanism.

Decomposition of Li₂CO₃

As discussed in Section 3.1, in most cases, Li₂CO₃ is the discharge product in Li–CO₂ batteries. Therefore, the charge process should involve the efficient decomposition of Li₂CO₃ for highly reversible batteries.

Since the carbon in both Li₂CO₃ (reactant) and CO₂ (product) have +4 valence, the redox site active in the charge process should not be ascribed to the carbon in Li₂CO₃. Three different reaction pathways for the charge mechanism of Li–CO₂ batteries were proposed by Zhou et al.  As demonstrated in Table 1,⁹¹ Path I is the decomposition path of Li₂CO₃, which generates CO₂ and O₂, with O₂⁻/O₂ being the redox couple. According to the thermodynamic Gibbs energy change, the corresponding reaction potential is calculated to be 3.82 V. In Path II, carbon species that could be either from the discharge product or from the cathode material participates in the process of Li₂CO₃ decomposition. The carbon is oxidized into CO₂ via a four-electron process. The equilibrium potential of the reaction is calculated to be 2.8 V. Path III is a more complex process with multiple steps. The decomposition of Li₂CO₃ generates Li⁺, CO₂, O₂, and superoxide radicals (O₂^·−) in the first step. The superoxide radical is highly unstable, which can transform into oxygen gas by losing one electron per atom. Also, the highly reactive superoxide radicals may attack the electrolyte with the aid of O₂ to generate some byproducts during the battery charging. To confirm the feasibilities of the three paths, the authors prepared prefilled Li₂CO₃^{– 12}C and Li₂CO₃^{– 13}C electrodes in the cells. Because all carbon atoms in Li₂CO₃ are ¹²C, the ¹³C isotope can be utilized to monitor the origin of the carbon atoms in the charge products. As shown in Figure 10A–D, in situ gas chromatography-mass spectrometry (GC-MS) was used to detect the gas composition in charging. First, no O₂ gas was detected in charging by GC-MS, which disproves the paths decribed in Paths I and III. Second, as shown in Figure 10B,C, the amounts of ¹²CO₂ and ¹³CO₂ formed during charging were nearly identical. This phenomenon  suggested that Path II is also incorrect, because the amount of ¹²CO₂ should be twice that of ¹³CO₂ if the reaction strictly follows Path II.

Table 1 Possible reactions for the decomposition of Li₂CO₃

Path.	Possible reactions	Erev (V) versus Li
I	$${\mathrm{Li}}_{2}{\mathrm{CO}}_{3}\to {\mathrm{CO}}_{2}+{1\text{/}2{\rm{O}}}_{2}+{2\mathrm{Li}}^{+}+{2{\rm{e}}}^{-}$$	3.82
II	$${\mathrm{Li}}_{2}{\mathrm{CO}}_{3}+1\text{/}2{\rm{C}}\to {2\mathrm{Li}}^{+}+{3\text{/}2\mathrm{CO}}_{2}+{2{\rm{e}}}^{-}$$	2.80
III	$${\mathrm{Li}}_{2}{\mathrm{CO}}_{3}\to {\mathrm{CO}}_{2}+{1\text{/}2{\rm{O}}}_{2}+{2\mathrm{Li}}^{+}+{1\text{/}2{\rm{O}}}_{2}^{\text{\unicode{x000B7}}-}+{3\text{/}2{\rm{e}}}^{-}{{\rm{O}}}_{2}^{\text{\unicode{x000B7}}-}\to {{\rm{O}}}_{2}+{{\rm{e}}}^{-}{{\rm{O}}}_{2}^{\text{\unicode{x000B7}}-}+{{\rm{O}}}_{2}+\mathrm{electrolyte}\to \mathrm{uncertain}\unicode{x02007}\mathrm{product}$$	Unknown value

Note: The corresponding Gibbs free energy and reversible potential are also listed.⁹¹

Figure 10. (A) Charge profile of the electrode prefilled by Li2CO3 with 12C or 13C carbon as conductive additives at 36 mA g Li2CO3−1 ${\rm{g}}\unicode{x0200A}{\mathrm{Li}}_{2}{{\mathrm{CO}}_{3}}^{-1}$ under a helium atmosphere with Li metal as the anode. (B) Gas evolution for CO2 of (B) Fragment-44, (C) Fragment-45, and (D) Fragment-15 in the two cells during the charging process. Reproduced with permission: Copyright 2016, Royal Society of Chemistry.91 (E) HPLC analysis and (F) 1H NMR spectra of the electrolyte after polarizing carbon black/Li2CO3/PTFE composite electrodes at the indicated potential to reach a capacity of 0.064 mAh. Reproduced with permission: Copyright 2018, Wiley-VCH.92

Furthermore, FTIR demonstrated the decomposition of the electrolyte solvent as evidenced by the two peaks arising from the C═O bond after charging, which confirmed that the electrolyte was attacked by O₂^·− and O₂ during charging (Path III).

Moreover, it was discovered that the reaction pathway of charging for Li–CO₂ batteries largely depended on the applied current density.¹² At a relatively low current of 0.5 A g⁻¹, only CO₂ was detected during charging, and Equation (18) was proposed accordingly. As the current density increased to 2 A g⁻¹, both CO₂ and O₂ gas were detected by GC-MS. The charge-to-mass ratio is 2 for 2e⁻/CO₂ and 4 for 4e⁻/O₂. Therefore, the reaction pathway during charging at high rates could be described by Equation (19). [Image Omitted. See PDF] [Image Omitted. See PDF]

Nonetheless, there are also different experimental observations on this point reported by other groups. For instance, Freunberger's group found evidence of singlet oxygen (¹O₂) formation along with the decomposition of Li₂CO₃.⁹² As shown in Figure 10E,F, the conversion of 9,10-dimethylanthracene (DMA) into endoperoxide (DMA-O₂) at a charge voltage range of 3.8–4.2 V was confirmed by high-performance liquid chromatography (HPLC) and ¹H nuclear magnetic resonance (NMR) analysis. Since DMA can only react with ¹O₂ and there were no other oxygen species (such as O₂^·−, O₂, or CO₂) that generated DMA-O₂, the experimental results gave solid evidence for the charging pathway in Equation (20) with the involvement of ¹O₂. [Image Omitted. See PDF]

Decomposition of Li₂CO₃ and carbon composite

Because the aforementioned O₂^·− and ¹O₂ may cause electrolyte degradation, restraining the production of O₂^·− and ¹O₂ in the Li–CO₂ battery is imperative. To avoid the formation of O₂^·− and ¹O₂, new reaction pathways for charging other than Equations (18) and (20) should be formulated. The decomposition of Li₂CO₃ accompanied by carbon reduction is a feasible method for an O₂^·−/¹O₂ free charging mechanism. Zhou's group assembled a rechargeable Li–CO₂ battery with Ru nanoparticles as the cathode catalyst, and they proposed the charging reaction pathway described in Equation (21).²⁸ The authors claimed that Ru nanoparticles can catalyze the reaction between Li₂CO₃ and carbon to form Li⁺ and CO₂ gas (Figure 11A). To prove this hypothesis, in situ surface-enhanced Raman scattering was used to monitor the Au electrode and the Au–Ru electrode in discharging and recharging. As shown in Figure 11B,C, in both electrodes, Li₂CO₃ and carbon formed during discharging. During charging, the Au–Ru electrode showed the decomposition of Li₂CO₃ and carbon, whereas the Au electrode showed the decomposition only of Li₂CO₃. This control experiment provides strong evidence that the Ru-based catalyst is able to promote the reaction between Li₂CO₃ and carbon. [Image Omitted. See PDF]

Figure 11. (A) Schematic of the charging mechanism of the Li–CO2 battery with and without the Ru catalyst. In situ SERS characterization of the electrodes (B) without and (C) with the Ru catalyst during discharge and recharge. Reproduced with permission: Copyright 2017, Royal Society of Chemistry.28 (D) High-resolution XPS spectrum of C 1s for the CC@Mo2C NPs electrode at a charged state. (E) Typical SEM images of CC@Mo2C NPs after the charge process at a current density of 20 µA cm−2, with a curtailing capacity of 100 µAh cm−2. Reproduced with permission: Copyright 2019, Wiley-VCH.74 SERS, surface-enhanced Raman scattering.

Some other materials, such as BN-hG⁸ and porous NiO nanofibers,⁹³ were reported to show similar catalytic capabilities in the simultaneous decomposition of Li₂CO₃ and carbon in the charging process. This O₂^·−/¹O₂ free charging mechanism is beneficial for the cyclability of the Li–CO₂ battery due to the successful avoidance of electrolyte decomposition.

Decomposition of amorphous Li₂C₂O₄

As discussed in 3.1.4, amorphous Li₂C₂O₄ could be the discharge product in Li–CO₂ batteries following Equation (17). Therefore, in the subsequent charge process, the Li–CO₂ battery would undergo the Li₂C₂O₄ decomposition process. Chen's group confirmed the decomposition of Li₂C₂O₄ during the charge process in a CC@Mo₂C NPs cathode-based Li-CO₂ battery via XPS and scanning electron microscopy (SEM) characterization.⁷⁴ Compared with the discharged state (Figure 9C), the Li₂C₂O₄ peak intensity in the high-resolution XPS spectrum of C 1s for the CC@Mo₂C NPs electrode at the charged state was significantly reduced, indicating the decomposition of Li₂C₂O₄ (Figure 11D), which is also evidenced by the difference in SEM images after discharge and charge (Figure 11E). The corresponding reaction can be described as Equation (22).⁴¹ Studies have shown that the charging voltage corresponding to the Li₂C₂O₄ decomposition is ca. 3.5 V versus Li/Li⁺,^41,74 which is distinctly lower than that for Equations (20) and (21).

The decomposition products were identified to be Li⁺ and CO₂ gas. No evidence that O₂^·− or ¹O₂ was involved in the charging reaction was found. [Image Omitted. See PDF]

To summarize, for rechargeable Li–CO₂ batteries, the reaction pathways of charging essentially determine the Coulombic and energy efficiencies of the energy storage system. The specific charge mechanism largely depends on the types of CO₂ER electrocatalysts. One basic principle for designing cathode catalysts is enabling the charging process to follow a more kinetically favorable pathway with a low energy barrier in the rate-determining step, which leads to lower charging potential, higher energy efficiency, and better cyclability.

Electrochemical mechanisms in Na/K–CO₂ batteries

In the last decade, with the increasing concern about lithium scarcity on Earth, cheaper and more abundant alkali metals (sodium, potassium) attract tremendous attention as alternatives for lithium, leading to the bloom of beyond-lithium (sodium/potassium) battery systems. Specific to the Li–CO₂ batteries, new systems including Na/K–CO₂ batteries have been explored. Basically, the fundamental working procedures of Na/K–CO₂ batteries are essentially the same as those of the Li–CO₂ batteries. Nonetheless, the detailed reaction pathways in discharging/charging could be very different among these systems. Due to the difference in charge carriers, the discharge/charge products, the Gibbs free energy change (ΔG) of discharge/charge reactions, and the redox potentials for Na/K–CO₂ batteries are completely different from those of Li–CO₂ batteries. The detailed discharge/charge mechanisms are significantly affected by the characteristics of the Na and K ions, including size, mass, Lewis pH, and diffusivity. The first rechargeable room-temperature Na-CO₂ batteries were reported by Chen's group in 2016 (Figure 12A).¹⁹ The cathode catalyst was prepared by coating Ni mesh with a tetraethylene glycol dimethyl ether-treated multiwall carbon nanotube (t-MWCNT). The cell showed an ultrahigh discharge capacity of 60,000 mAh g⁻¹ at a current density of 1 A g⁻¹ and a cyclability of 200 cycles with charge voltages below 3.7 V. The authors also investigated the discharge/charge mechanisms using a series of characterization techniques. In situ Raman, XRD, and XPS revealed the formation and decomposition of Na₂CO₃ during discharge and charge, respectively, as shown in Figure 12B–D. The formation of carbon during discharge was confirmed by EELS characterization on a carbon-free Ag nanowire cathode (Figure 12E). Besides, the regeneration of CO₂ during charge was confirmed using a portable CO₂ analyzer (Figure 12F). Based on these data, the discharge/charging pathway in a Na–CO₂ battery was proposed as the reaction in the following equation: [Image Omitted. See PDF]

Figure 12. (A) Structure of Na–CO2 batteries with a metal Na foil anode, an ether-based electrolyte, and a t-MWCNT cathode. (B) In situ Raman spectra and corresponding discharge/charge profiles with 11 selective voltage points (inset). (C) XRD patterns of Na–CO2 batteries at different states. (D) XPS and (E) EELS spectra of a Ag wire cathode at different states. (F) Real time CO2-evolution test. Reproduced with permission: Copyright 2016, Wiley-VCH.19

Moreover, a controlled experiment was conducted to compare the charging behaviors of a prefilled Na₂CO₃/carbon electrode and a Na₂CO₃ electrode. The Na₂CO₃/carbon electrode showed a charging voltage 0.5 V lower than that of the Na₂CO₃ electrode, indicating the favorable effect of the carbon phase on the Na₂CO₃ decomposition.

Compared with the intensively studied Li–CO₂ and Na–CO₂ batteries, studies of K–CO₂ batteries are few and in the infancy stage. To determine the discharge/charge mechanisms in a K–CO₂ battery, Huang's group successfully developed a K–CO₂ nanobattery and observed the in situ discharge/charge processes using an aberration-corrected environmental transmission electron microscope (AC-ETEM), as shown in Figure 13A.²¹ During discharge, K reacted with CO₂ to generate K₂CO₃ hollow balls and CO nanobubbles. The process could be described by Equation (24). During charging, K₂CO₃ decomposed into K and CO₂ with the consumption of carbon in the electrode (Equation 25). [Image Omitted. See PDF] [Image Omitted. See PDF]

Figure 13. (A) Structure and phase identifications for the discharged and partially charged products by electron diffraction patterns. Reproduced with permission: Copyright 2018, Elsevier.21 (B) Ex situ XRD patterns of the Toray paper and N-CNT/RGO after discharging or charging. (C) Raman spectra of the N-CNT/RGO cathode after discharging or charging. (D) Diagrams of a K–CO2 battery with N-CNT/RGO for CO2 electrocatalysis. Reproduced with permission: Copyright 2019, Wiley-VCH.22

Dai's group reported development of a K–CO₂ battery using a composite of nitrogen-doped carbon nanotubes and nitrogen-doped reduced graphene oxide (N-CNT/RGO) as the cathode catalyst.²² Due to the 3D network structure of the composite and the abundant exposure of active sites, this K–CO₂ battery shows a cyclability of 250 cycles under a fixed capacity of 300 mAh g⁻¹. As shown in Figure 13B,C, XRD and Raman characterizations revealed the formation of P12₁/c1-type K₂CO₃ at the discharged state, in line with the K₂CO₃ formation energies calculated using DFT. The K₂CO₃ signals in both XRD and Raman spectra disappeared in the charged state, indicating the reversible decomposition of K₂CO₃. Unfortunately, the study has not confirmed whether carbon is involved in the reactions due to the pure carbon cathode catalyst. Moreover, the DFT calculations and experimental results revealed the decomposition of P12₁/c1-type K₂CO₃, indicating the superior reversibility of this K–CO₂ battery. The reaction mechanism can be described by Equation (26) and Figure 13D. [Image Omitted. See PDF]

Overall, the electrochemical reaction pathways of Li/Na/K–CO₂ batteries with carbonate as the final discharge product are consistent. Nonetheless, the decomposition energy barriers of these carbonates are quite different. The difference in the decomposition potential of Li/Na/K carbonates leads to the different thermodynamic charge potentials, which ultimately results in the different working potentials of the alkali metal–CO₂ batteries. In addition, there are limited options of reaction pathways in Na/K–CO₂ batteries compared with the various reaction pathways in Li–CO₂ batteries (no reaction pathways in Na/K–CO₂ batteries with oxalate as the final discharge product have been discovered so far). Moreover, the sluggish kinetics of the CO₂RR/ER in Na/K–CO₂ batteries is also a reason for their lower energy efficiency and inferior cycling stability compared with Li–CO₂ batteries, which place a higher demand on the electrocatalytic capability of the bifunctional cathode catalysts.

REACTION MECHANISMS OF THE ANODES IN ALKALI METAL–CO₂ BATTERIES Alternatives to lithium

Since the cathodes of the CO₂ batteries contain only the CO₂RR/ER catalysts, which cannot provide alkali metal ions, the anodes must be alkali metal-rich materials. As the alkali metals (Li/Na/K) have the lowest potential and the highest specific capacity,⁹⁴ the CO₂ batteries mostly use the alkali metal foils as the anodes. It should be emphasized that apart from the cathodes typically using CO₂ catalysts, the alkali metal anodes also have an influence on the performance of the devices.

The low relative atomic mass of lithium contributes to the high gravimetric/volumetric energy density of the Li–CO₂ battery.^95,96 However, low abundance, uneven worldwide distribution, and high cost are always concerns for lithium-based batteries.⁹⁷ One promising option is to substitute lithium with other alkali metals, such as the more abundantly available sodium and potassium. As shown in Table 2, the availability of lithium in the Earth's crust and ocean is only 0.0017% and 0.000018%, respectively, much less than that of sodium and potassium.^105,106 Because of the chemical similarity and much higher abundance in the Earth's crust and ocean, sodium and potassium are good alternatives to lithium. Sodium and potassium ions have larger ionic radiuses but smaller Stokes' radiuses in some frequently used ester-based electrolyte (e.g., propylene carbonate) solvents than lithium ions, which is beneficial for enhancing the ion mobility and conductivity.¹⁰⁷ As shown in Table 2, sodium and potassium ions show better conductivities in several ether and ester-based electrolytes compared to lithium ions, which would facilitate fast diffusion kinetics. Compared with Li–CO₂ batteries, Na/K–CO₂ batteries have lower equilibrium potential and energy density (Table 2). On the other hand, even for K–CO₂ batteries (923 Wh kg⁻¹), the energy density still exceeds that of commercial LIBs.

Table 2 Properties of lithium, sodium, and potassium

Alkali metal element	Lithium	Sodium	Potassium
Relative atomic mass	6.94	22.98	39.09
Density (g cm−3)	0.535	0.968	0.856
Melting point (°C)	179	98	64
Abundance in the earth's crust	0.0017%	2.3%	1.5%
Abundance in the earth's ocean	0.000018%	1.1%	0.042%
Distribution	70% in South America	Uniform	Uniform
Ionic radius (Å)95	0.76	1.02	1.38
Stokes' radius in propylene carbonate (Å)98	4.8	4.6	3.6
A+ conductivity in ether electrolyte (ACF3SO3 in TEGDME, mS cm−1)	0.9 (1 M)96	3.9 (1 M)99	Not Rep.
A+ conductivity in ester electrolyte (APF6 in EC/DMC, mS cm−1)100	9.8 (1 M)	10 (1 M)	11.5 (1 M)
A+ conductivity in ester electrolyte (APF6 in PC, mS cm−1)101	4.38 (0.8 M)	6.38 (0.8 M)	6.55 (0.8 M)
Gravimetric capacity (mAh g−1)	3860	1166	687
Volumetric capacity (mAh cm−3)	2062	1131	589
E0 versus SHE (V)	−3.04102	−2.71103	−2.931104
Equilibrium potential versus A/A+ of a A–CO2 battery system (V)	2.8091	2.3565	2.4822
Theoretical energy density of an A–CO2 battery system (Wh kg−1)	187613	113027	923

Alkali (Li/Na/K) metal anodes

For alkali metal (Li/Na/K)–CO₂ batteries, the metal anodes operate in a highly concentrated CO₂ atmosphere, being distinctly different from conventional Li/Na/K batteries in which the metal anodes are only in contact with the organic electrolytes. Therefore, the CO₂ dissolved in the organic electrolyte has significant effects on the behaviors of alkali metal anodes, especially the metal plating/striping behaviors, the side reactions, and the solid electrolyte interface (SEI) layers on the system. It is highly instructive to re-examine the reaction mechanisms of anodes in alkali metal–CO₂ batteries.

First, the fundamental reactions for Li/Na/K metal anodes are identical to those in common Li/Na/K metal batteries, which undergo metal plating and striping processes during charge and discharge of the full cell.^19,85 The reactions are described in the following equation: [Image Omitted. See PDF]

The influence of CO₂ on the lithium metal anode was intensively studied in previous reports. Zhou's group found that CO₂ is beneficial for the formation of a unique SEI layer that is different from the routine SEI formed by the decomposition of organic electrolytes and salts. This SEI exerts an effective protective effect on the lithium metal anode.¹⁰⁸ A proof-of-concept Li–Cu cell was constructed to investigate the behaviors of Li metal in an environment with/without CO₂. The linear sweep voltammetry (LSV) curve in Figure 14A shows that an additional cathodic peak appeared at 1.0 V, which could be ascribed to the CO₂ reduction. Afterwards, the better stability of the Li–Cu cell with CO₂ than that without CO₂ was confirmed by cyclic voltammetry (CV) (Figure 14B,C). Raman and XPS were utilized to investigate the SEI layers formed on the lithium metal surface. As shown in Figure 14D, Raman spectra revealed that Li₂O and LiOH were the main components of the SEI layer in the battery without CO₂, whereas with CO₂, the SEI layer mainly contained dense Li₂CO₃. The XPS results further proved the existence of Li₂CO₃ (Figure 14F). Based on the comprehensive characterizations, the chemical components of the SEI layers formed in different atmospheres are shown in Figure 14E.

Figure 14. (A) Schematic demonstrating the electrolyte reduction on the Cu electrode and the Li plating process in a CO2 environment. CV curves of the Li–Cu batteries upon cycling (B) without CO2 and (C) with CO2. (D) Raman spectra of the electrodes corresponding to (B,C). (E) Structures of the different SEI layers. (F) C 1s, F 1s, and Li 1s XPS spectra of the SEI layers after different etching times. Reproduced with permission: Copyright 2020, Elsevier.108 (G) Schematic diagrams and digital images of the deposition of the graphene on a Li foil in a tetrahydrofuran solution with different reaction times. Reproduced with permission: Copyright 2020, Wiley-VCH.78

Some other studies also reported the constructive role of CO₂ in protecting the Li metal anode from corrosion. For example, Zhang's group reported the positive effect of CO₂ by comparing Li–O₂ batteries and Li–O₂/CO₂ batteries.¹⁰⁹ In a Li–O₂ battery (i.e. an oxygen environment for the lithium metal anode), LiOH is the main product of the side reaction occurring on the lithium metal anode, which has a porous structure, and cannot prevent the further corrosion of lithium metal. On the contrary, in a Li–O₂/CO₂ battery, a dense Li₂CO₃ layer formed on the lithium metal surface, which effectively eliminated the corrosion. As a result, 25.77% Li loss occurred in the Li–O₂ battery after only 50 cycles, and reversible Li plating could last for only 194 cycles. In sharp contrast, in the Li–O₂/CO₂ battery, the Li metal anode showed a very low percentage of loss of only 6.23% after 500 cycles, and the stable Li plating could last for over 8000 cycles. The formation mechanism of the Li₂CO₃ layer was also studied, and the relevant reactions are shown in Equations (28)–(30). The H₂O is derived from the electrolyte decomposition during cycling. Moreover, the CO₂ is able to capture O₂⁻ to inhibit the side reactions that occur in the electrolyte and on the cathode. [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

Apart from the native protection layers generated by the reaction between lithium metal and CO₂, artificial protective coatings on alkali metal anodes are highly beneficial for prolonging battery cycling. For instance, Dai's group established a protective layer by soaking the Li metal anode in a graphene oxide tetrahydrofuran solution for 1 h (Figure 14G). The corrosion of the lithium metal was significantly alleviated by the graphene layer on the metal.⁷⁸ Due to the use of a single Fe atom-decorated N,S-codoped holey graphene (Fe-ISA/N,S-HG) cathode catalyst, this Li–CO₂ battery worked for 210 cycles with a fixed capacity of 1000 mAh g⁻¹ at a current density of 1000 mA g⁻¹.

In the case of sodium and potassium metal anodes, because of their much higher chemical activity than lithium, more severe dendrites, and side reactions issues in the Na/K plating/striping, there is an urgent need to develop new reaction mechanisms for the anodes that can provide for more stable cycling performance and higher Coulombic efficiency.

In addition to the metal plating/striping reactions of alkali metals, anodes involving other reaction mechanisms (e.g., alloying/dealloying) were used in alkali metal–CO₂ batteries, especially Na/K–CO₂ batteries. Chen's group reported a rechargeable K–CO₂ battery with a KSn anode.¹¹⁰ The KSn anode was fabricated by in situ electrochemically potassiating a Sn foil in a K||Sn cell. Compared with K metal, the KSn anode can effectively alleviate dendrite formation and improve the safety and stability of a K–CO₂ battery. Figure 15A shows the configuration of the K–CO₂ battery with a KSn anode and the reaction mechanisms are shown as follows: [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

Figure 15. (A) Schematic illustration of a K–CO2 battery utilizing a KSn alloy anode, an MWCNT-COOH cathode catalyst, and an ester-based electrolyte. (B) XRD pattern and corresponding Rietveld refinement of the KSn anode; inset: photograph of the KSn anode. (C) Crystal structure of the KSn alloy. In situ observation of the electrochemical K deposition/potassiation behavior for (D) a K foil substrate and (E) a Sn foil substrate. (F) XRD patterns of the KSn anode at pristine, fully discharged, and fully charged states. Ex situ SEM images of the KSn anode at (G) fully discharged and (H) fully charged states. Reproduced with permission: Copyright 2021, Wiley-VCH.110

XRD pattern and Rietveld refinement results confirmed the tetragonal structure of the KSn alloy with the I4₁/acd space group, as shown in Figure 15B,C. By comparing the potassiation of the K foil and the Sn foil, the authors observed uneven K deposition and dendrite formation on the K foil, indicative of unstable K deposition. In addition, bubbles were observed, which may have been caused by the side reactions between the K metal and the electrolyte (Figure 15D). By comparison, stable potassiation was found on the Sn foil, which can be attributed to the completely different reaction mechanisms of K and Sn (alloying/dealloying vs. K plating/striping, Figure 15E). As shown in Figure 15F, strong Sn peaks were found after discharge, indicating the dealloying process of KSn. Upon recharge, KSn peaks reappeared, indicating a highly reversible alloying/dealloying process. Furthermore, the morphology of the KSn anode did not change obviously after discharge and recharge (Figure 15G,H), indicating the excellent stability of the KSn alloy anode. Eventually, the K–CO₂ batteries with the KSn anode showed superior cycling stability and rate performance.

To summarize, most of the charge/discharge mechanisms of metal anodes in alkali metal–CO₂ batteries are based on alkali metal plating/stripping. Nonetheless, due to the differences in the growth morphology of dendrites, the Coulombic efficiency and stability of the alkali metal–CO₂ batteries with Na/K anodes are inferior to those with Li anodes. Therefore, it is highly desired to modify the properties of the Na/K anode by using other more intrinsically stable and efficient reactions.

THE STATE-OF-THE-ART ELECTROLYTES IN AN ALKALI METAL–CO₂ BATTERY

To optimize the electrochemical performance of alkali metal–CO₂ batteries, various types of electrolytes have been developed in recent years. Basically, a good electrolyte for alkali metal–CO₂ batteries should have the following characteristics: (1) the solvent should have high CO₂ solubility, which provides sufficient CO₂ supply for the cathode catalysts; (2) the electrolyte should have excellent alkali metal ion conductivities, which endow the battery with low internal resistance and excellent power characteristics; (3) the electrolyte should have a wide voltage window and superior stability at high reducing/oxidizing potentials, which minimize the side reactions and electrolyte degradation during discharge or charge; and (4) the electrolyte should ideally be environmentally friendly.¹⁴ So far, few electrolytes fulfill all these requirements. Therefore, more research on optimizing electrolytes are required to break the bottleneck of electrolytes in the alkali metal–CO₂ batteries.

Liquid electrolytes in alkali metal–CO₂ batteries

Common electrolyte solvents in alkali metal–CO₂ batteries mainly include organic solvents, water, and ionic liquids.^111–114 Ether-, sulfone-, and carbonate-based solvents are the main choices for organic electrolytes in alkali metal–CO₂ batteries. The salts that are most commonly used are lithium/sodium/potassium bis(trifluoromethanesulfonyl)imide (Li/Na/KTFSI) and lithium/sodium/potassium trifluoromethanesulfonate (Li/Na/KCF₃SO₃). As mentioned previously, the reactions in alkali metal–CO₂ batteries involve a gas–liquid–solid tri-phase interfacial electrochemical charge-transfer process. However, because of the high solubility of CO₂ in organic electrolyte solvents, the reaction of alkali metal–CO₂ batteries in certain areas of the cathode/electrolyte interfaces is analogous to a solid–liquid two-phase reaction. Taking Equation (2) as an example, the discharge potential of the Li–CO₂ battery can be expressed by the following Nernst equation.¹¹⁵ [Image Omitted. See PDF]where $$\varphi (+)$$ is the discharge potential of the cathode, $${\varphi }^{\theta }(+)$$ is the standard potential of the cathode, R is the gas constant, T is the temperature, F is the Faraday's constant, and [CO₂] and [Li⁺] are the concentrations of CO₂ and Li⁺ in the electrolyte, respectively. As can be deduced from Equation (31), the concentrations of Li⁺ and CO₂ in the electrolyte are the key factors affecting the discharge voltage of Li–CO₂ batteries.

Gallant's group proposed that different concentrations of LiCF₃SO₃ affect the solubility of CO₂ in tetraethylene glycol dimethyl ether (TEGDME) and the concentration of Li⁺ available to react with CO₂ and/or its discharge intermediates.¹¹⁶ A moderate Li⁺ concentration (0.7–2 M) can ensure that the electrolyte has high ionic conductivity and good CO₂ solubility. Compared with other organic solvents such as dimethyl sulfoxide (DMSO) and propylene carbonate (PC), Li⁺ desolvation in TEGDME has the lowest thermodynamic barrier, indicating that Li⁺ is more “available” in TEGDME for electrochemical reactions.

Precisely because ether-based electrolytes are highly CO₂ soluble and compatible with most categories of electrocatalysts, many studies of alkali metal–CO₂ batteries are based on ether-based solvents (especially TEGDME). For instance, Guo's group fabricated Li–CO₂ batteries with crumpled Ir nanosheets on porous carbon nanofibers (Ir NSs-CNFs) as the cathode and used 1 M LiTFSI in a TEGDME solvent as the electrolyte.⁷⁰ This battery is able to operate for 400 cycles with a fixed capacity of 1000 mAh g⁻¹ at a current density of 500 mA g⁻¹. Besides, at 100 mA g⁻¹, with a capacity limit of 1000 mAh g⁻¹, a low charge voltage below 3.8 V was obtained, indicating the excellent electrochemical performance of this battery. Wang's group used 1 M LiTFSI in TEGDME as the electrolyte of Li–CO₂ batteries as well.¹¹⁷ Equipped with a MnO nanoparticle-decorated graphene-interconnected N-doped 3D carbon framework (MnO@NC-G) cathode, the assembled Li–CO₂ battery yielded an excellent discharge capacity of 25,021 mAh g⁻¹ at a current density of 50 mA g⁻¹. At a high current density of 1 A g⁻¹ and a cut-off capacity of 1000 mAh g⁻¹, the battery showed an outstanding cycle performance of 206 cycles. Besides, the battery achieved a charge capacity of 23,811 mAh g⁻¹, and the Coulombic efficiency was calculated to be 95.2%. Many other studies used 1 M LiTFSI in TEGDME as the electrolyte,^{8,38,59,81,118,119} proving that it is the most suitable liquid electrolyte in LiCO₂ batteries to date.

Some other salts were also used in Li–CO₂ battery electrolytes. For instance, Chen's group used 1 M LiCF₃SO₃ salt dissolved in TEGDME as the electrolyte. A Mo₂C/CNT catalyst was used as the cathode.⁴¹ The battery achieved a good round-trip efficiency of 77% and a stable cyclability of 40 cycles at a current of 20 μA and a limited capacity of 100 μAh. Zhou's group also used LiCF₃SO₃ in TEGDME (mole ratio 1:4) as the electrolyte.²⁸ Combined with the Ru@Super P cathode, the Li–CO₂ battery delivered a high discharge capacity of 8229 mAh g⁻¹ at a current density of 100 mA g⁻¹. Besides, the battery was able to operate for 80 cycles at a current density of 100 mA g⁻¹ and a fixed capacity of 1000 mAh g⁻¹, indicating the good cycling performance.

Compared with the various electrolytes used in Li–CO₂ batteries, fewer types of electrolytes have been utilized in Na–CO₂ batteries so far. 1 M NaClO₄ salt in TEGDME is a widely used electrolyte in Na–CO₂ batteries. In 2016, Chen's group fabricated the first Na–CO₂ battery employing 1 M NaClO₄ in TEGDME as the electrolyte.¹⁹ The electrolyte is highly compatible with the electrodes in the system; hence, a high reversible capacity of 60,000 mAh g⁻¹ was achieved at a current density of 1 A g⁻¹. The battery operated well for 200 cycles with a cut-off capacity of 2000 mAh g⁻¹ at a high current density of 1 A g⁻¹, indicating the negligible degradation of the electrolyte during the prolonged cycling. In 2019, Song's group also designed 1 M NaClO₄ dissolved in a TEGDME electrolyte for a Na–CO₂ battery with a Ru@KB composite cathode catalyst.²⁰ This battery delivered a high discharge capacity of 11,537 mAh g⁻¹ at 100 mA g⁻¹ and a Coulombic efficiency of 94.1%. Moreover, the Na–CO₂ battery was able to operate for more than 130 cycles at a fixed capacity of 1000 mAh g⁻¹ and a current density of 200 mA g⁻¹ with a charge voltage below 4.0 V. This electrolyte is also compatible with other catalysts like Co₂MnO_x nanowire-decorated carbon fibers (CMO@CFs),¹²⁰ MoS₂/SnS₂,¹²¹ and ZnCo₂O₄@CNT.¹²² Basically, there are several advantages of using 1 M NaClO₄ in a TEGDME electrolyte for Na–CO₂ batteries: (1) relatively high ionic conductivity (ca. 0.2 S m⁻¹); (2) low volatility; and (3) intrinsic stability of Na metal.¹⁹ These characteristics are highly beneficial for the cyclability and rate performance of Na–CO₂ batteries. In recent years, the sodium salt options available for use in Na–CO₂ battery electrolyte have rapidly increased. In addition to NaClO₄, NaCF₃SO₃ and NaTFSI dissolved in TEGDME electrolytes were also utilized in Na–CO₂ batteries. Wang's group applied 0.5 M NaCF₃SO₃ in a TEGDME electrolyte in Na–CO₂ batteries with cobalt-modified carbon nanofibers as self-standing cathode materials.¹²³ The Na–CO₂ battery showed good cycling stability, with 128 cycles at a current density of 0.14 mA cm⁻². Liu and co-workers assembled Na–CO₂ batteries using 1 M NaTFSI in a TEGDME electrolyte and a Ru nanoparticle-decorated CNT cathode.⁶⁵ The batteries achieved a high discharge capacity of 20,277 mAh g⁻¹ and worked stably for 100 cycles at a cut-off capacity of 500 mAh g⁻¹ at 100 mA g⁻¹.

To date, reports of K–CO₂ batteries are much fewer than those of Li–CO₂ batteries and Na–CO₂ batteries due to the shortage of high-performance catalysts and the poor understanding of the reaction mechanism. In the limited literature available on K–CO₂ batteries so far, 1.0 M KTFSI dissolved in TEGDME is almost the only electrolyte option. This electrolyte does not seem to be the bottleneck for the cyclability of K–CO₂ battery. For instance, Dai's group used 1.0 M KTFSI in a TEGDME electrolyte and a 3D network of a N-doped carbon nanotube/N-doped reduced graphene oxide composite (N-CNT/RGO) as the cathode catalyst.²² Superior cycling performance (250 cycles) was achieved at a cut-off capacity of 300 mAh g⁻¹, which is compatible with the counterpart values of Li/Na–CO₂ batteries. Gallant's group also demonstrated that in the TEGDME solvent, although the desolvation barrier of K⁺ is lower than that of Li⁺/Na⁺ $$\left({\rm{\Delta }}{G}_{\left({{\rm{K}}}^{+},\mathrm{desolv}\right)}=323.1\unicode{x0200A}\mathrm{kJ}\unicode{x0200A}{\mathrm{mol}}^{-1},\unicode{x02007}{\rm{\Delta }}{G}_{\left({{\rm{Li}}}^{+},\mathrm{desolv}\right)}=385.0\unicode{x0200A}\mathrm{kJ}\unicode{x0200A}{\mathrm{mol}}^{-1},\unicode{x02007}{\rm{\Delta }}{G}_{\left({\mathrm{Na}}^{+},\mathrm{desolv}\right)}=383.1\unicode{x0200A}\mathrm{kJ}\unicode{x0200A}{\mathrm{mol}}^{-1}\right),$$ the activity for CO₂ reduction is limited.¹¹⁶ Therefore, in K–CO₂ batteries using KCF₃SO₃ in a TEGDME electrolyte, the CO₂ reduction activity is dominated by the Lewis acidity of K⁺.

According to Equation (34), the concentration of metal cations at the interface between the electrolyte and the cathode greatly affects the discharge voltage. Meng's team proposed that the solvated Li⁺ size in different electrolyte solvents has an influence on the Li⁺ concentration at the interface. In the case of 1 M LiTFSI as the solute, the size of Li(DMSO)₄⁺ is smaller compared to Li(TEGDME)₂⁺, making the density of Li⁺ in the DMSO electrolyte around CO₂ molecules higher.¹¹⁵ For instance, Wang's group reported that the MOF-derived single iron atom catalyst with 1 M LiTFSI/DMSO electrolyte showed high electrochemical performance in Li–CO₂ batteries.¹²⁴ A high capacity of 13,228 mAh g⁻¹ and good cycling performance of 140 cycles were found in these Li–CO₂ batteries operating at a current density of 200 mA g⁻¹. Wang and co-workers used 1 M LiCF₃SO₃/DMSO as the electrolyte and a graphitized polyamic acid fiber decorated with copper single atoms as the cathode to assemble Li–CO₂ batteries.³⁴ The batteries delivered superior electrochemical performance with a high capacity of 14,084 mAh g⁻¹ and a long life of 133 cycles. Although DMSO is widely used as an electrolyte solvent in Li–CO₂ batteries, it is rarely reported in Na/K–CO₂ batteries.

In addition to the mainstream pure ether solvent-based electrolyte, some other special chemical compositions have been reported in Li–CO₂ batteries to date, for instance, 0.3 M LiNO₃ in 1 M LiTFSI/DMSO,⁶⁰ 1 M LiClO₄ in TEGDME,⁶⁸ and so forth. The introduction of nitrate into the electrolyte was shown to facilitate the formation of a passivation layer on the surface of the metal anode, which retarded the corrosion of the metal caused by the intermediates.¹²⁵ In addition to nitrate, small amounts of other metal cation (Li⁺/Na⁺/K⁺) salt additives were included in the bulk electrolytes; yet, the specific functionalities are still not fully understood.

By adding appropriate additives to the electrolyte, a solid electrolyte interphase/cathode electrolyte interphase (SEI/CEI) film can be constructed at the interface of the cathode/anode and the electrolyte, which greatly enables maintenance of the stability of the electrode structure. Wang's group fabricated a high-performance K–CO₂ battery using a durable K anode with artificial SEI.¹²⁶ The K anode with a KF-rich SEI film was obtained by charging/discharging in 2.7 M KFSI–TEGDME electrolyte, as shown in Figure 16A. Then, the K–CO₂ battery was assembled using the K anode with an SEI film, a bamboo-like N-doped carbon nanotube cathode, and a KTFSI–TEGDME electrolyte. The SEM image shows a clear film covered on the K sheet and a uniform compact surface morphology, as shown in Figure 16B. Besides, the XRD pattern of SEI-formed K metal shows the obvious existence of diffraction peaks of KF, indicating the successful deposition of the KF SEI film (Figure 16C). Moreover, high-resolution XPS spectra of F 1s for the KFSI salt and the KF-SEI K anode proved that the S–F bond contributed to the formation of the KF-SEI film, as shown in Figure 16D,E. This assembled K–CO₂ battery delivered a discharge capacity of 9436 mAh g⁻¹ and a low overpotential of 0.81 V at a current density of 50 mA g⁻¹. Furthermore, the battery could work for 450 cycles at a high current density of 300 mA g⁻¹ with a cut-off capacity of 500 mAh g⁻¹, which offers an inspiring idea for the development of alkali metal–CO₂ batteries.

Figure 16. (A) Schematic illustration of the preparation procedure of a KF-rich SEI-coated K anode and operation of a K–CO2 battery. (B) SEM morphology of the KF-rich SEI-coated K anode after cycling in a concentrated KFSI/TEGDME electrolyte. (C) XRD patterns of the pristine K and KF-rich SEI-coated K detached from a K||KFSI-TEGDME||carbon paper cell. High-resolution XPS spectra of F 1s for (D) KFSI salt and (E) the SEI K anode (after cycling in 2.7 M KFSI/TEGDME). Reproduced with permission: Copyright 2022, Wiley-VCH.126 (F) Schematic illustration of the CO2RR mechanism in the existence of quinone. Cyclic voltammograms at 100 mV s−1 of quinones (2.5 mM AQ [top], 5 mM PAQ [middle], and 5 mM DBBQ [bottom]) in Ar- (red lines) and CO2− (blue lines) saturated MeCN containing (G) 0.1 M TBAClO4 and (H) 0.1 M LiClO4, respectively. Reproduced with permission: Copyright 2018, American Chemical Society.55

In terms of the working mechanism, the functionalities of the electrolyte are not limited to the ion conduction, anode/cathode insulation, and SEI/CEI modification. The electrolyte can participate in the CO₂RR/ER reactions by releasing electrolyte-soluble catalytic active species. Recently, many studies have proposed that addition of soluble catalysts or redox mediators (RM) in the electrolyte is an effective method to reduce the charge overpotential.¹²⁷ Redox mediators are soluble molecules with reversible redox couples. The existence of independent central metal species and adjacent bonding elements or functional groups provides electrocatalytic activity for these redox mediators in alkali metal–CO₂ batteries.⁶² Zhang and co-workers proposed a possible reaction pathway for Li₂CO₃ decomposition by redox mediators as shown in the following equations¹²⁷: [Image Omitted. See PDF] [Image Omitted. See PDF]

Electrons are transferred from the redox mediator to the cathode at the beginning of the charge process up to its oxidation potential. Subsequently, Li₂CO₃ is decomposed into CO₂ and Li⁺ by the RM^O+, while the RM^O+ is reduced to the original state (RM). Therefore, the charge voltage of the battery is the oxidation potential of the redox mediator, instead of the decomposition potential of Li₂CO₃.

Grimaud and co-workers added quinones as RM to the electrolyte of Li–CO₂ batteries for the first time.⁵⁵ As shown in Figure 16F, the quinone additives will undergo the following reactions in turn during the discharge process. [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

First, quinone will acquire an electron to form a quinone anion radical (Q^·−) and it will continue to acquire electrons to form dianion Q²⁻. Next, in the presence of quinone, intermolecular coupling between CO₂ and dianions Q²⁻ yields Q(CO₂)_p²⁻ to complete the CO₂RR process. In fact, the redox potential of RM greatly affects the overpotential of the CO₂RR/ER process. It was found that the redox potential of quinones may depend on their molecular structure, the solvents supporting cations, and electrolytes. It can be inferred from Figure 16G,H that 9,10-anthraquinone (AQ) and 2,5-ditert-butyl-1,4-benzoquinone (DBBQ) react with CO₂ at the second redox peak. That is, the discharge voltage of the Li–CO₂ battery to which quinone RM is applied is determined by the redox potential of Q⁻/Q²⁻. In addition, Li₂CO₃ only formed in the DBBQ/CH₃CN electrolyte system. Despite the inevitable degradation of DBBQ and other battery components, this study provides a novel idea to develop stable additives to enhance the CO₂ reduction process in Li–CO₂ batteries.

Apart from the organic RMs, various types of inorganic RMs were also introduced into the electrolyte of alkali metal–CO₂ batteries. According to theoretical computations, Br₂ could decompose the discharge products (Li₂CO₃ and C) of Li–CO₂ batteries, expressed as 2Li₂CO₃$$+$$ + C + 2Br₂ = 3CO₂ + 4LiBr, which makes Br species a potential redox mediator for alkali metal–CO₂ batteries. Zhou's group reported that LiBr was used as a redox mediator in the electrolyte of Li–CO₂ batteries.¹²⁸ As shown in Figure 17A, through the two-step reaction of Br⁻ → Br₃⁻ (3.6 V vs. Li/Li⁺) and Br₃⁻ → Br₂ (4.0 V vs. Li/Li⁺), the obtained Br₂ can decompose Li₂CO₃. The charge potential of the Li–CO₂ battery decreases from 4.5 to ~4.1 V (Figure 17B) and shows a discharge capacity as high as 11500 mAh g⁻¹. In addition, XRD patterns (Figure 17C) can also confirm the occurrence of an oxidation–reduction reaction between Br₂ and Li₂CO₃. Once the discharge product Li₂CO₃ comes in contact with Br₂, it will promptly oxidize and produce Br⁻ as the reduction product. Species with active metal centers could also be used redox mediators. For instance, Wang's team introduced Ru(bpy)₃Cl₂ as a redox mediator into the electrolyte.⁴⁴ As shown in Figure 17D, the Ru^II centers can not only interact with dissolved CO₂ to facilitate the electroreduction reaction but also stabilize the discharge intermediate and inhibit its conversion into Li₂CO₃. The charge potential of the Li–CO₂ battery containing a Ru(bpy)₃Cl₂ catalyst was reduced to 3.86 V (Figure 17E), and over 60 cycles were achieved with a cut-off capacity of 1000 mAh g⁻¹ at a current density of 300 mA g⁻¹. This work also compared the effect of discharge depth for discharge products of Li–CO₂ batteries with or without a Ru^II catalyst. According to the FTIR spectrum (Figure 17F), the characteristic peaks at 860 cm⁻¹ (bending vibration of O–C═O), 1430 cm⁻¹ (stretching vibration of O-C═O), and 1000 cm⁻¹ (stretching vibration of C–O) of Li₂CO₃ were observed on the cathode after discharge to 10,000 mAh g⁻¹, while these peaks were absent at a discharge depth of 1000 mAh g⁻¹. This means that at a shallow discharge depth, Li–CO₂ batteries containing the Ru^II catalyst may form the amorphous intermediate discharge product Li₂C₂O₄, which has a smaller charge-transfer resistance, thereby being conducive to reducing the charge overpotential. Additionally, o-phenylenediamine (OPD) has been proven to be a bifunctional liquid catalyst for Li–CO₂ batteries.¹²⁹ Figure 17G shows the charge mechanism of Li–CO₂ batteries with an OPD redox mediator. OPD is oxidized preferentially at the beginning of the charge process (Equation 40). The oxidation state OPD* migrates to the surface of discharge products and decomposes them into Li⁺ and CO₂, while the OPD* returns to its pristine state (Equation 41). [Image Omitted. See PDF] [Image Omitted. See PDF]

Figure 17. (A) Proposed mechanism in the charge process with LiBr as the redox mediator in Li–CO2 batteries. (B) Initial cycle voltage profiles with a cut-off capacity of 500 mAh g−1 at a current density of 200 mA g−1. (C) XRD patterns of Li2CO3 and C without (black line) or with (red) oxidized Br2. Reproduced with permission: Copyright 2017, Wiley-VCH.128 (D) Illustration of the function of the electrolyte soluble Ru(bpy)3Cl2 catalyst in Li–CO2 batteries. (E) Initial discharge/charge curves with a cut-off capacity of 1000 mAh g−1 at 300 mA g−1. (F) FTIR spectra of cathodes of the batteries with a RuII catalyst at different states of charge. Reproduced with permission: Copyright 2021, Wiley-VCH.44 (G) Illustration of the oxidization reaction mechanism of OPD toward discharge products in Li–CO2 batteries. (H) The Gibbs free energy change of CO2RR processes with OPD at different reaction steps. Reproduced with permission: Copyright 2022, Elsevier.129

DFT simulations were carried out to elaborate the mechanisms of the discharge process in Li–CO₂ batteries with an OPD catalyst (Figure 17H). The specific intermediate process reactions are as follows: [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF] [Image Omitted. See PDF]

CO₂ molecules dissolve in the electrolyte first and acquire electrons to form CO₂ radicals (CO₂⁻). Next, CO₂⁻ preferentially combines with OPD to form OPD:CO₂⁻, and then Li⁺ is added to form OPD:LiCO₂, which easily falls off from the OPD and is converted into LiCO₂. Li₂C₂O₄ synthesized using a combination of LiCO₂ tends to undergo a self-disproportionation reaction and reacts with the disproportionation product to yield the final discharge products Li₂CO₃ and carbon. The charge platform of Li–CO₂ batteries with and without an OPD redox mediator at 200 mA g⁻¹ with a cut-off capacity of 1000 mAh g⁻¹ observably decreased from 3.99 to 3.21 V.

It is noteworthy that the possible chemical interaction between the electrolyte and the intermediates in CO₂RR/ER is a critical issue for the development of new electrolyte components, which needs to be studied in depth. Zhou's group pointed out that at a high charge potential (>4.2 V), the intermediates released from the process of Li₂CO₃ decomposition would nucleophilically attack DMSO molecules to generate some by-products (lithium carboxylate, etc.).¹² It has been suggested that TEGDME molecules are unstable during the decomposition of Li₂CO₃.⁹¹ Therefore, the instability of the organic electrolyte solvent during the charge process is the key factor that adversely affects the electrochemical performance of alkali metal–CO₂ batteries.

In recent years, the liquid electrolyte was expanded from organic electrolytes to molten salt and aqueous electrolytes. For instance, Zhou's team reported a low-charge-overpotential Li–CO₂ battery consisting of a binary molten salt electrolyte and a porous carbon cathode.¹³⁰ In the molten state of 140°C, the electrolyte of the LiNO₃/KNO₃ mixture showed high ionic conductivity, high solubility, and a fast diffusion rate of CO₂. In particular, the charge overpotential was less than 1 V with a SuperP cathode. Aqueous electrolytes are also used in alkali metal–CO₂ batteries. Limited by the high reactivity of alkali metal anodes with water, aqueous electrolytes are usually only applied on the cathode side in alkali metal–CO₂ batteries and a “water-in-salt” (WIS) electrolyte strategy is used to create an efficient cationic solvation sheath structure. Kang and co-workers assembled a Na–CO₂ battery with a WIS electrolyte (cathode side)–ether electrolyte (anode side) separated by a NASICON film.¹¹² The electrochemically stable window was expanded to 3.45 V using a 17 M NaClO₄ aqueous electrolyte. Na–CO₂ batteries with this WIS electrolyte and a Ru@carbon cathode showed cycling stability of over 75 cycles without significant alteration.

Solid-state electrolytes in alkali metal–CO₂ batteries

Due to the issues of volatility, leakage, and flammability of liquid electrolytes and the increasing demand for flexible battery devices, the development of solid-state electrolytes for alkali metal–CO₂ batteries has become a hot topic in recent years. A gel polymer electrolyte (GPE) is a commonly studied solid-state electrolyte for alkali metal–CO₂ batteries. Generally, GPEs comprise liquid electrolytes immobilized in solid polymer matrices and are hence generally called quasi-solid-state electrolytes. The most commonly used polymer matrices include polyethylene oxide (PEO), polyacrylonitrile (PAN), poly(vinyl alcohol) (PVA), poly(vinylidiene fluoride-hexafluoropropylene) (PVDF-HFP), and so forth. Due to the quasi-solid nature, GPEs normally have high ionic conductivity (0.1–10 mS cm⁻¹) under ambient conditions because the ion transportation actually occurs in the liquid phase instead of the polymer phase. The liquid phases in GPEs for alkali metal–CO₂ batteries are mostly organic solvents (esters and ethers). Some inorganic additives, including SiO₂, Al₂O₃, and so forth, are usually added in the GPE to modify the ion conductivity, mechanical strength, and so forth. Specifically in terms of alkali metal–CO₂ batteries, the liquid phases in GPEs fall within the scope of the liquid electrolytes mentioned in Section 5.1.

A representative early study of a solid-state electrolyte-based Li–CO₂ battery was carried out by Chen's group.⁸⁵ The gel polymer matrix was poly(methacrylate)/poly(ethylene glycol) with a 3 wt% SiO₂ inorganic additive (PMA/PEG-LiClO₄-3 wt% SiO₂). LiClO₄ was utilized as the lithium salt. On adding 3 wt% amorphous SiO₂ (~10 nm in diameter) to the GPE, the Li-ion conductivity increased from ~6 × 10⁻² to 7.14 × 10⁻² mS cm⁻¹ at 55°C. The CR2032 coin cell-type Li-CO₂ battery with GPE can last for 100 cycles with a cut-off capacity of 1000 mAh g⁻¹. Besides, the GPE electrolyte overcame the problem of leakage of liquid electrolyte-based Li–CO₂ batteries (Figures 18A,B). More importantly, due to the high flexibility of GPE, flexible pouch cells can be easily assembled, yielding a large reversible capacity, high energy density, and long work time upon repeated bending.

Figure 18. Leakage test of the Li–CO2 coin cells with GPE or liquid electrolyte (A) before and (B) after pressing on a piece of dry paper. Reproduced with permission: Copyright 2017, Wiley-VCH.85 (C,D) Textile integrated with a double fiber-shaped Li–CO2 battery as a power accessory. (E,F) Three green LEDs lit up by the self-powered textile bent in different directions. Fire-proof properties of (G,H) the quasi-solid-state fiber-shaped battery using GPE and (I,J) the traditional fiber-shaped battery using a liquid electrolyte. Reproduced with permission: Copyright 2019, Wiley-VCH.74 (K) Full discharge/charge curves of the Li–CO2 batteries using GPE and Bi-CoPc-GPE between 2.0 and 4.5 V. (L) Voltage curves for the Li–CO2 batteries with GPE and Bi-CoPc-GPE at a fixed capacity of 1000 mAh g−1 at a current density of 200 mA g−1. Reproduced with permission: Copyright 2019, Wiley-VCH.131

Wang's group prepared GPE using N-methyl-2-pyrrolidinone (NMP)/PVDF-HFP and a 2-hydroxy-2-methyl-1-phenyl-1-propanone (HMPP)/trimethylolpropane ethoxylate triacrylate (TMPET) gel polymer matrix with LiTFSI in the TEGDME liquid component.⁷⁴ With the Mo₂C nanoparticle catalyst, the Li–CO₂ battery delivered a low charge potential at 3.4 V and a high energy efficiency of 80%. The GPE endowed the batteries with excellent extendibility toward flexible and wearable electronics, which can operate under various deformation conditions (Figure 18C–F). Moreover, the leakage problem and the flammability issue were significantly suppressed by taking advantage of the solid-state GPE (Figure 18G–J).

The function of GPE is not just limited to conduction of ions. The GPE can also participate in the electrochemical processes of the cathodes if the GPE contains certain a redox mediator. Functionalization of the electrolyte with a redox mediator is a classic strategy to modulate and improve the electrochemical performance of the electrodes; this strategy has been applied in Li–O₂ batteries,^132–134 Zn–air batteries,^135,136 and Li–sulfur batteries.^137,138 In alkali metal–CO₂ batteries, the effective redox mediator should be electrolyte-soluble substances that can participate in the CO₂RR/ER. As a representative example, Guo and co-workers fabricated GPE containing 0.0025 M binuclear cobalt phthalocyanine additive (Bi-CoPc-GPE) for a Li–CO₂ battery.¹³¹ The Bi-CoPc acted as a redox mediator that could promote the decomposition of the discharge product, thereby decreasing the charge polarization. Besides, the ion conductivity was also improved upon addition of Bi-CoPc. Li–CO₂ batteries using a Bi-CoPc-GPE electrolyte showed an outstanding discharge capacity of 27,196 mAh g⁻¹ with a voltage platform of 2.90 V, superior to that of the Bi-CoPc free GPE-based counterpart (Figure 18K). The overpotential of the Bi-CoPc-GPE-based Li–CO₂ battery was only 1.14 V, much lower than that of the GPE-based Li–CO₂ battery (1.90 V, Figure 18L), proving the crucial role of Bi-CoPc as redox mediators in promoting the decomposition of discharge products.

Apart from Li–CO₂ batteries, solid-state electrolytes are also utilized in Na–CO₂ batteries. A pioneering study on a solid-state electrolyte Na–CO₂ battery was conducted by Chen's group in 2017.²⁷ The as-prepared GPE in this work used a PVDF-HFP-4 wt% SiO₂ gel polymer matrix. The liquid component immobilized in the polymer matrix was a NaClO₄-TEGDME solution. Due to the high ionic conductivity, robust toughness, nonflammable matrix, and strong electrolyte-locking ability of the GPE, this Na–CO₂ battery system can work for 400 cycles at a current rate of 500 mA g⁻¹ with a cut-off capacity of 1000 mAh g⁻¹ in a pure CO₂ atmosphere. A pouch-type Na–CO₂ battery using the identical GPE can deliver a high capacity of 1.1 Ah, high energy density of 232 Wh kg⁻¹, and cyclability of 50 cycles. High-performance GPE is also a critical component for flexible solid-state Na–CO₂ batteries. For instance, Chen and co-workers fabricated GPE with a PEO matrix, NaClO₄ sodium salt, and a 3 wt% SiO₂ additive.¹³⁹ The integrated Na–CO₂ batteries with this GPE showed excellent performance stability upon repeated bending, folding, and other deformations, indicating the great potential for wearable energy storage devices based on Na–CO₂ electrochemistry.

In addition to the quasi-solid-state electrolyte, the Na⁺ superionic conductor (NASICON) solid-state electrolyte was also applied in alkali metal–CO₂ batteries. The working potential of these solid-state electrolyte-equipped batteries could be modulated by selecting transition metal/polyanion groups in the NASICON-type structure.^140,141 The structural formula of NASICON can be summarized as A_xMM'(XO₄)₃, where A = Li/Na/K, and so forth, M or M' = Fe/V/Ti/Zr/Sc/Mn/Nb/In, and so forth, and X = S/P/Si/As, and so forth. MO₆ and M'O₆ octahedra share apex corners with three XO₄ tetrahedra, which provides a good transport channel for metal ion diffusion. The NASICON-type solid electrolytes used in Li–CO₂ batteries are mainly Li_1+xAl_2−xGe_x(PO₄)₃ (LAGP). For instance, Xu's group applied the Li_1.5Al_0.5Ge_1.5P₃O₁₂ solid-state electrolyte synthesized using the solid-phase method to assemble Li–CO₂ batteries.¹⁴² This highly conductive solid-state LAGP electrolyte (~3.9 × 10⁻⁴ S cm⁻¹) can suppress the cross-reaction and improve the utilization of CO₂ gas, thereby avoiding the effect of CO₂ dissolved in the electrolyte on the lithium anode. Liu and co-workers reported a solid-state Na–CO₂ battery with an inorganic Na₃Zr₂Si₂PO₁₂ solid-state electrolyte (NZSP) and a succinonitrile interphase.¹⁴³ The high conductivity of the NZSP (0.8 mS cm⁻¹) promoted the room-temperature operation of the battery. This solid-state Na–CO₂ battery showed a high discharge capacity of 28830 mA h g⁻¹ at 100 mA g⁻¹ with a Ru/CNT cathode and good stability of 70 cycles (at 50 mA g⁻¹) without an evident increase of the potential gap and capacity decay. There are no reports on the application of solid-state electrolytes for K–CO₂ batteries. All-solid-state alkali metal–CO₂ batteries can easily avoid the problems of poor stability and easy leakage of the liquid electrolyte, as well as the alkali metal corrosion caused by the dissolved CO₂. Therefore, it is highly desired to study and establish the structure–activity relationship between solid-state electrolytes and alkali metal–CO₂ batteries in future research.

SUMMARY AND PERSPECTIVE

To summarize, use of alkali metal–carbon dioxide batteries (Li/Na/K–CO₂ batteries) provides a new strategy for CO₂ fixation and utilization. Unfortunately, the limited understanding of the electrochemical mechanism to date significantly impedes the development of alkali metal–CO₂ batteries. This field is quite young, and hence, a systematic summary of the previous studies, especially the ones focusing on the reaction mechanism exploration, is highly instructive. This is the first article of this type, that is, a comprehensive review of electrochemical mechanisms in alkali metal–CO₂ batteries. The working mechanisms of CO₂RR/ER catalysts in the cathode side, the metal anodes, and the electrolytes are fully covered in this study.

In the first section, the reaction mechanisms of CO₂ cathode catalysts in alkali metal–CO₂ batteries were introduced. During discharge, several different routes may arise, depending on the specific reaction condition in Li–CO₂ batteries. Correspondingly, different discharge products, that is, Li₂CO₃/carbon, Li₂CO₃/CO, Li₂O/carbon, and Li₂C₂O₄, can form in discharge. For the rechargeable alkali metal–CO₂ battery, in the reverse charging process, the discharge products become the initial reactants for the charge reaction, which would basically decompose into Li⁺ and CO₂. The reaction mechanisms on the cathodes of Na/K–CO₂ batteries were reviewed parallel to that of Li–CO₂ batteries. As for the reaction mechanism on the anode side of alkali metal–CO₂ batteries, it can be briefly summarized that reversible alkali metal plating/striping processes occur in discharge and charge. Apart from the electrodes, the previously used electrolytes including both liquid- and solid-state ones in alkali metal–CO₂ batteries are also comprehensively reviewed, including the chemical components, the intrinsic properties, and design strategies.

The previous studies clearly demonstrate the great potential of alkali metal–CO₂ batteries in the application scenarios of high energy density energy storage, stationary large-scale energy storage, power source for aerospace exploration, and so forth. Nonetheless, the commercialization process of the alkali metal–CO₂ batteries is greatly hindered by the insufficient understanding of the electrochemical mechanism in the systems. Specific to the topic of this review, we propose that significant efforts should be made in the following directions to study the electrochemical mechanisms of alkali metal–CO₂ batteries.

• 1)

  The exploration of the reaction pathways for the CO₂ cathodes should not be limited to the end products of the discharge/charge reactions. Both CO₂RR and CO₂ER are highly complex reactions. The detailed reaction sub-steps and the intermediates involved need to be determined to draw the whole Gibbs free-energy diagram. This is of vital importance for the design of bifunctional catalysts because the adsorption/desorption behaviors of the reaction intermediates on the active sites in catalysts essentially determine nearly all aspects of the battery performance including but not limited to the voltage polarization, energy efficiency, and rate capability. Advanced operando characterization techniques combined with in situ electrochemical battery setups are highly desired to investigate the intermediates participating in the reactions. This may not be an easy task because the intermediates could be electrolyte dissoluble or gas in phase and, in most cases, require a high detection limit for the characterization technique. Therefore, a characterization system with various mutually complementary spectrum and microscope techniques needs to be established to gain an in-depth understanding of the cathode reaction mechanism.

• 2)

  The factors influencing the reaction pathway of CO₂RR/ER in the cathode need further investigation, which is an indispensable aspect of the study of the working mechanism in the future. Previous research demonstrates that different catalysts may have different reaction pathways for CO₂RR/ER. However, the underlying cause of this phenomenon is unclear. The electrolytes that provide the chemical environments for the charge transfer, gas diffusion, mass transportation processes have a major influence on the reaction mechanism like reaction kinetics and reversibility. Moreover, due to the open structure of alkali metal–CO₂ batteries, the factors of the external environment, such as the CO₂ pressure, the moisture level, the external physical field (light, electromagnetics), and so forth, should not be neglected. As discussed in this review, most of the studies demonstrated that despite using different CO₂RR/ER catalysts, the reaction pathways in Na/K–CO₂ batteries only follow the mechanism of Na₂CO₃/K₂CO₃ and carbon products, which is quite different from the varied reaction pathways in a Li–CO₂ battery. Therefore, one important aspect of the study of Na/K–CO₂ is exploration of other alternative reaction pathways that could be more kinetically facile. Moreover, it is highly instructive to explore the potential selectivity of CO₂RR/ER catalysts toward the different reaction pathways.

• 3)

  Apart from the bifunctional catalysts in the cathode side, the metal anode exerts an identical effect on the performances of alkali metal–CO₂ batteries. In fact, the issues present in the alkali metal anodes (e.g., dendrite formation, electrolyte corrosion, large polarization) are the disadvantages present in all metal anode-based battery systems including Li–O₂, Li–S, and Li-inorganic cathodes battery systems. Alkali metal anodes may also face other challenges since the alkali metal–CO₂ batteries work in a special CO₂-rich atmosphere. Basically, the alkali metal anodes undergo a reversible metal plating/striping mechanism and dendrite formation is, to some extent, inevitable. Therefore, it is highly instructive to expand the working mechanisms in the anode side. For instance, the alloying reaction, conversion reaction, and interaction reaction-based anode materials have great potential for use in CO₂ batteries. Nonetheless, it is noteworthy that the anodes should be in the Li/Na/K-containing condition because the CO₂ cathode cannot provide alkali metal ions for the system. Therefore, for the different types of Li/Na/K free anodes, pre-lithiation/sodiation/potassiation procedures are essential. Due to the much larger volume change of Na/K metal plating/striping, the lower resistance to the electrolyte erosion, and greater difficulty in building a robust SEI, it is much more challenging to construct highly stable Na/K metal anodes than a Li metal anode. Therefore, there is a more urgent need to expand the reaction mechanism of the anodes in Na/K–CO₂ batteries by minimizing the energy density sacrifice as compared to the pure Na/K metal anode-based ones.

• 4)

  To date, specialized research on the electrolyte optimization for alkali metal–CO₂ batteries is still quite limited. Most of the previous studies utilized conventional ether-based electrolyte compositions. In fact, there is considerable room for improvement of the electrochemical performance of alkali metal–CO₂ batteries by modulating the liquid electrolyte to gain the following characteristics. First, the solvent should have high solubility toward CO₂. According to the Nernst equation, the potential of the reaction can be determined by the concentration of the reagent. In most cases, a high CO₂ concentration in the electrolyte is favorable. Second, special attention should be paid to the properties of the electrolyte/electrode interfaces, which is strongly affected by the electrolyte degradation. For instance, the SEI on the anode, which is derived from the decomposition of electrolyte components (both salts and solvents), strongly affects the anode performance. Also, since the CO₂RR/ER occurs at the triple-phase interface, the ionic solvation structure and the diffusion layer at the interface are of critical importance. Third, since the alkali metal–CO₂ batteries normally have an open structure, the possible problems that may arise such as electrolyte leakage, flammability, and volatilization should also be taken into consideration for the development of new electrolytes.

ACKNOWLEDGMENTS

This work received financial support from the National Natural Science Foundation of China (52072257) and the National Key Research and Development Program of China (2019YFE0118800).

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.