Redox Mediators for Li₂CO₃ Decomposition

Abstract

Lithium–air batteries (LABs) possess the highest energy density among all energy storage systems, and have drawn widespread interest in academia and industry. However, many arduous challenges are still to be conquered, one of them is Li₂CO₃, which is a ubiquitous product in LABs. It is inevitably produced but difficult to decompose; therefore, Li₂CO₃ is perceived as the “Achilles’ heel of LABs”. Among various approaches to addressing the Li₂CO₃ issue, developing Li₂CO₃-decomposing redox mediators (RMs) is one of the most convenient and versatile, because they can be electrochemically oxidized at the gas cathode surface, then they diffuse to the solid-state products and chemically oxidize them, recovering the RMs to a pristine state and avoiding solid-state catalysts’ contact instability with Li₂CO₃. Furthermore, because of their function mechanism, they can double as catalysts for Li₂O₂/LiOH decomposition, which are needed in LABs/LOBs anyway regardless of Li₂CO₃ incorporation due to the sluggish kinetics of oxygen reduction/evolution reactions. This review summarizes the progress in Li₂CO₃-decomposing RMs, including halides, metal–chelate complexes, and metal-free organic compounds. The insights into and discrepancies in the mechanisms of Li₂CO₃ decomposition and corresponding catalysis processes are also discussed.

1. Introduction

The depletion of fossil fuels and CO₂-induced climate change are the two major energy concerns for the sustainability of human society. To pursue an economy with near-zero CO₂ emissions, fossil fuels should be largely replaced with renewable power sources, such as solar, wind, and tidal energy. To regulate their intermittent and unsteady output, energy storage systems with long cyclability, low cost, and high safety are required. Furthermore, the rapid development of portable electronic devices and electric vehicles imposes urgent demand for batteries with higher energy density, aside from the above-mentioned properties. In the last two decades, lithium-ion batteries (LIBs) have replaced nickel–metal hydride batteries and become the most popular energy storage device in consumer electronics and electric vehicles due to their higher energy density; however, the energy density of LIBs can hardly exceed 500 Wh kg⁻¹ with commonly used lithium intercalation electrode materials. To further improve energy density, lithium metal anode (LMA) and conversion-type cathodes need to be adopted to construct Li–sulfur and Li–air batteries (LABs), etc., which have drawn widespread interest. Especially, LABs possess an ultrahigh theoretical energy density of 11,429 Wh kg⁻¹, excluding the O₂ mass, which is the highest among all energy storage technologies, because they can potentially use ambient oxygen without pre-storage. Some studies have reported LAB pouch cell prototypes with energy densities up to 1214 Wh kg⁻¹ [1,2,3,4]. In pure oxygen atmosphere (making Li-O₂ batteries, or abbreviated as LOBs) and dry electrolytes, most studies [5] reported Li₂O₂ as the main discharge product, as shown in Equation (1), though other discharge products such as Li₂O [6], stabilized LiO₂ [7,8], and Li₃O₄ [9,10] have also been observed. With increasing water content in the electrolyte, the Li₂O₂ would gradually change morphology and convert to LiOH [11,12].

Overall cathode reaction: 2Li⁺ + O₂ + 4e⁻ → Li₂O₂  E⁰ = 2.96 V vs. Li/Li⁺

(1)

Unfortunately, the practicality of LABs/LOBs is plagued by their currently poor cyclic stability, rate capability, and environmental adaptability, which have three origins, namely the high reactivity at both electrodes, the instable cathode interfaces brought by solid-state and insulative discharge/parasitic products, and the contamination of atmospheric CO₂ and H₂O, as shown in Figure 1. At the anode side, the high chemical activity of lithium metal, combined with its huge volumetric variation, causes instability of the solid–electrode interface (SEI), leading to the mutual consumption of metallic lithium and electrolyte, and the thickening of the SEI [13]. At the cathode side, during discharge, the solid-state discharge products (Li₂O₂/LiOH in pure O₂, or Li₂CO₃/LiOH in ambient air) gradually clog the gas diffusion channel and passivate the surface of the air cathode material, causing the termination of discharge. During charge, the unstable solid–solid contact between the air cathode and discharge products significantly raises the charge overpotential. In addition, the high overpotential induces more oxidative decomposition of cell components, such as solvents and air cathode substrates, leading to the formation of parasitic products (Li₂CO₃, lithium carboxylates, polyethers, polyesters, etc.).

Compared to Li₂O₂, LiOH, and other solid-state discharge/parasitic products, Li₂CO₃ has much higher oxidative decomposition potential and overpotential (shown in

Table 1. Common solid-state products in LABs/LOBs and their decomposition potential via possible routes [22].

Discharge Product Species	Possible Decomposition Reaction	ΔGr0 (kJ∙mol−1)	E0 (VLi)	Practical Decomposition Potential in Carbon Electrode (VLi)	Practical Decomposition Potential in Carbon Electrode W/Catalyst (VLi)
Li2O2	Li2O2 → 2Li+ + O2 + 2e−	−15.67	2.96	~4.1 [37]	~3.9 (Pt) [37]
Li2O	2Li2O → 4Li+ + O2 + 4e−	−50.84	2.91	~5 [37]	~4.0 (Pt) [37]
Li2CO3	2Li2CO3 → 4Li+ + 2CO2 + O2 + 4e−	302.16	3.82	~4.8 [37]	~4.2 (Pt) [37]
LiOH	4LiOH → 4Li+ + 2H2O(l) + O2 + 4e−	118.56	3.35	~5 [37]	~3.8 (Pt) [37] ~3.0 (LiI) [38]
HCOOLi (a)	2HCOOLi → 2Li+ + H2 + 2CO2 + 2e−	−112.81	2.46	N/A	N/A
CH3COOLi (a)	2CH3COOLi → 2Li+ + 3H2O(l) + 3C + CO + 2e−	−157.91	2.22	N/A	N/A
R(OCOOLi)2	R(OCOOLi)2 + 2H2O → R(OH)2 + 2CO2 + 2OH• + 2Li+ + 2e−	N/A	N/A	~3.9 [39]	~3.9 (Fe3O4) [39]
	R(OCOOLi)2 + xO2 → yCO2 + zH2O + 2Li+ + 2e−

^(a) Estimated values from the thermodynamic data of HCOONa, CH₃COONa, and Na⁺.

[22]), further increasing the practical charge polarization, exacerbating parasitic reactions, and thus forming a vicious cycle resulting in rapid cell failure. When operated in ambient air, the situation is further worsened by atmospheric H₂O and CO₂. Despite its low content (~412 ppm) in ambient air, CO₂ has ~50 times higher solubility than O₂ in organic electrolytes [5], thereby fundamentally changing cell chemistry. Many investigations [23,24,25,26,27,28,29] have found that the dominant discharge product shifts from Li₂O₂ in pure O₂ and N₂ + O₂ mixtures to Li₂CO₃ in ambient air; some other papers [30,31,32,33,34] showed that the primary discharge product was LiOH, while there are also studies [35,36] reporting Li₂O₂ staying as the main discharge product, unaffected by CO₂ and H₂O. However, from a thermodynamic perspective, Li₂O₂ and LiOH would eventually convert to Li₂CO₃ after being in contact with CO₂ for sufficient time.

The routes of Li₂CO₃ formation have been well summarized previously by Peng and coworkers, as shown in Figure 2A [20]. Generally, in CO₂-free atmospheres, Li₂CO₃ mainly comes from the degradation of the electrolyte (possible reaction routes are depicted in Figure 2B [40,41]) and carbon substrate of the air cathode [42,43]. In the presence of both CO₂ and O₂ (making an LAB or Li-CO₂/O₂ battery [22,44,45], depending on their contents), CO₂ reacts either with O₂⁻/LiO₂ during the oxygen reduction reaction (ORR) or with Li₂O₂ after the ORR, depending on the dielectric constant and donor number (DN) of the electrolyte solvent [46,47]. Therefore, it can be concluded that Li₂CO₃ is a ubiquitous discharge/parasitic product in LABs/LOBs.

A special case is pure CO₂ atmospheres without O₂ presence (making a Li-CO₂ battery), where the discharge product shifts from Li₂CO₃ to a mixture of Li₂CO₃ and amorphous carbon [48,49], as described in Equation (2).

Cathode reaction in CO₂: 4Li⁺ + 3CO₂ + 4e⁻ → 2Li₂CO₃ + C  E⁰ = 2.80 V vs. Li/Li⁺

(2)

The charge process of the Li-CO₂ battery is the reverse reaction of Equation (2). However, this does not apply to LABs or Li-CO₂/O₂ batteries. In these systems, once Li₂CO₃ becomes the discharge product as a result of CO₂ incorporation during/after ORR, the recharge process is not the reverse reaction of Equation (1), but the decomposition of Li₂CO₃, which has several possible pathways with different sets of products, as shown in Figure 2C and Equations (3)–(5).

Pathway 1: 2Li₂CO₃ − 4e⁻ → 4Li⁺ + 2CO₂ + ³O₂   E⁰ = 3.82 V vs. Li/Li⁺

(3)

Pathway 2: 2Li₂CO₃ − 4e⁻ → 4Li⁺ + 2CO₂ + ¹O₂   E⁰ = 4.07 V vs. Li/Li⁺

(4)

Pathway 3: 2Li₂CO₃ − 3e⁻ → 4Li⁺ + 2CO₂ + O₂⁻

(5)

Apparently, the formation and decomposition of Li₂CO₃ (via the pathways in Figure 2 and Equations (3)–(5)) greatly increase the theoretical charge potential without significantly affecting discharge potential (mainly determined by Equation (1)), thereby diminishing the theoretical round-trip efficiency to 77.5% (2.96 V/3.82 V), which is inferior to that of the co-decomposition reaction of 2Li₂CO₃ + C (100%, Equation (2)). However, the incorporation of even a slight amount (2%) of O₂ would shift Equation (2) to CO₂-incorporated ORR (producing peroxycarbonates and peroxydicarbonates, and finally Li₂CO₃, as shown in Figure 2A, instead of 2Li₂CO₃ + C in Equation (2)) [50,51], which is encountered in most cases with ambient air. The virtual inevitability of Li₂CO₃ production, along with the difficulty to decompose it, make it “the Achilles’ heel of LABs [20,52]”.

To minimize Li₂CO₃ production, many researchers in the field have turned to pure O₂ atmospheres, converting LABs to LOBs. However, the addition of internally stored O₂ and a reservoir (as shown in Figure 3A) would greatly diminish LABs’ core advantage: their ultrahigh energy density. To unleash the full potential of LABs in energy density, the utilization of ambient air is strongly desired. Therefore, tackling the Li₂CO₃ challenge is critical for LABs’ development. It is also highly beneficial for the development of related battery systems, such as Li-CO₂ and Li-CO₂/O₂ batteries (which can find applications in space and underwater missions, especially ones on Mars), and may inspire new remedies to address the carbonation issue in high-nickel cathode materials. The potential approaches to address the Li₂CO₃ challenge are listed below.

(1)

Using H₂O and/or CO₂ absorbents/adsorbents (usually with -OH and/or -NH₂ groups, etc.) [55,56,57] to entrap them before entering the cell. This approach is only feasible for large-scale LAB modules due to the apparently increased heft (including the regeneration apparatus for prolonged operation), as shown in Figure 3B.

(2)

Using oxygen selective membranes (OSMs) to exclusively allow O₂ permeation. Since O₂ has a larger kinetic diameter (0.346 nm, based on Knudsen diffusion values) than H₂O (0.289 nm) and CO₂ (0.330 nm) [5,58], CO₂ and H₂O cannot be filtered out by sheer pore size adjustment, but should be repelled by polarity difference. Recent investigations demonstrated the successful blocking of H₂O by various polymer membranes, such as polyester (PET) [2,3], high-density polyethylene (HDPE) [2], low-density polyethylene (LDPE) [24], polytetrafluoroethylene (PTFE) [59,60,61], perfluoropolyether (PFPE) [54,62], polysiloxanes [63], silicone oil [30,64,65], etc. However, the development of CO₂-repelling (rather than adsorbing/absorbing, which has a limited lifetime) and O₂-permeable membranes is still challenging, and there has been no such literature to the authors’ knowledge. Fortunately, the good news is that Li₂CO₃ formation can also be greatly suppressed by these H₂O-repelling membranes [24,54,62,66], as illustrated in Figure 3C.

(3)

Shifting the CO₂-involved reaction routes to prevent Li₂CO₃ deposition. For example, in 2018, Zhou’s group [15] reported a Li₂CO₃-free Li–O₂/CO₂ battery with a [Li(DMSO)₃]⁺–[TFSI⁻] contact ion pair (CIP) as the electrolyte. The positive charge of highly solvated Li⁺ cations is efficiently neutralized by the DMSO sheath, preventing its combination with the peroxydicarbonate (C₂O₆²⁻) anion to form solid-state Li₂CO₃; thus, C₂O₆²⁻ becomes a soluble discharge product, rather than intermediate. This is demonstrated by the presence of CO₄²⁻ and C₂O₆²⁻ and the absence of Li₂CO₃ signals in in situ Raman and infrared spectroscopy tests. As a result, the charge voltage plateau was significantly reduced from ~4.2 V to ~3.5 V. In 2021, Wang’s group [67] discovered tris(2,2′-bipyridyl)-dichloro-ruthenium(II) (Ru(bpy)₃Cl₂) as a bi-functional redox mediator (RM) for Li-CO₂ batteries. Through X-ray diffraction (XRD) and Fourier-transform infrared spectroscopy (FTIR) characterizations, they found that the cell with Ru(bpy)₃Cl₂ contained little Li₂CO₃ after a shallow discharge to 1000 mAh g⁻¹, in sharp contrast to that without Ru(bpy)₃Cl₂. They speculated that the shallow discharge product was lithium oxalate, and the Ru^II center could interact with dissolved CO₂ molecules to promote the formation and stabilization of oxalate, inhibiting the transformation from oxalate to carbonate. During recharge, the cell with Ru(bpy)₃Cl₂ exhibited a much lower voltage plateau at 3.7~3.8 V than the blank control (4.2~4.3 V), attributed to the easier decomposition of soluble oxalate than solid-state Li₂CO₃. This approach is highly desirable because the soluble nature of C₂O₆²⁻ and oxalate significantly reduces the charge potential. However, to date there have been only a few pieces of literature reported [15,67,68,69,70,71,72,73,74,75], and more investigations should be made in the future.

(4)

Developing Li₂CO₃-decomposing catalysts. Aside from addressing the CO₂-induced Li₂CO₃ issue, this approach can also handle Li₂CO₃ produced via internal parasitic reactions. Due to the sluggish kinetics of the ORR and oxygen evolution reaction (OER), cathode catalysts are usually used to improve LABs/LOBs’ performance, which constitutes a large portion of research in the LABs/LOBs field [5]. Despite the primary goals of ORR catalysis and Li₂O₂/LiOH decomposition, some of them are also found to be capable of decomposing Li₂CO₃, such as Ru [76] and its compounds (RuO₂ [77], RuP₂ [78]), Au [79], NiO [80,81], Ir/B₄C [82], LiCoO₂ [83], etc. Aside from this, some catalysts are found to be capable of catalyzing the co-decomposition reaction of 2Li₂CO₃ + C (Equation (2)), such as MoS₂ [84], Ti₃C₂T_x MXene [85], etc., while their effects on Li₂CO₃ decomposition need to be further investigated.

However, an issue of solid–solid contact instability emerges when the proximate solid-state products are decomposed and a gap forms between the products and catalyst; this would raise the charge polarization, diminish the decomposition efficiency of solid-state products, and induce more parasitic reactions. To address such issues, soluble redox mediators (RMs) are introduced into LABs/LOBs [86,87,88,89,90], which are also called soluble or mobile catalysts. They can be electrochemically oxidized at the air cathode surface, then they diffuse to the solid-state products and chemically oxidize them, recovering the RMs to a pristine state, as illustrated in Figure 4A. With suitable RMs, the charge overpotential and parasitic reactions can be greatly suppressed [91,92].

Based on the above description, developing Li₂CO₃-decomposing catalysts, especially the soluble ones, is a more convenient and versatile approach to address the Li₂CO₃ challenge. Because of their function mechanism, they can double as catalysts for Li₂O₂/LiOH decomposition, which are needed in LABs/LOBs anyway regardless of Li₂CO₃ incorporation. Therefore, it is helpful to summarize the proceedings in Li₂CO₃-decomposing RMs, which is the scope of this review. The reported Li₂CO₃-decomposing RMs can be generally classified into three categories, namely halides, metal–chelate complexes, and metal-free organic compounds, which are discussed in Section 2, Section 3 and Section 4, respectively.

2. Halides

Halides have long been used as RMs in dye-sensitized solar cells (DSSCs) [96], and have been introduced into LABs/LOBs since 2014 (LiI [97,98]) as RMs, primarily for Li₂O₂ and LiOH decomposition. However, the utilization of LiBr in LABs/LOBs can be traced back to 2009 [99,100], where it was used for improving Li⁺ conduction. In 2016, Aurbach and Sun [101] reported the first application of LiBr as an RM for an LOB, using a Br⁻/Br₃⁻ redox couple (at ~3.5 V vs. Li/Li⁺ in glymes) to facilitate Li_xO_y decomposition. In their work, they suggested limiting raising the charge voltage to avoid the formation of Br₂ due to its strong corrosive/reactive nature. Soon after, Lu and co-worker [92] proposed that, with bromine’s second redox couple Br₃⁻/Br₂ at ~4.0 V, (shown below), LiBr could mediate both Li₂O₂ and Li₂CO₃ decomposition and act as a bi-functional mobile catalyst for Li-O₂ batteries. Through differential electrochemical mass spectroscopy (DEMS) characterization, they observed not only a reduction in charge overpotential, but also an earlier onset of CO₂ evolution (from 4.35 V of the LiBr-free cell to 3.90 V of the LiBr-containing cell). Furthermore, the CO₂ evolution rate almost doubled. They suggested that the surge in CO₂ evolution rate was caused by the LiBr-mediated acceleration of Li₂CO₃ decomposition, as shown in Figure 4B.

3Br⁻ → Br₃⁻ + 2e⁻    E_diglyme = 3.48 V vs. Li/Li⁺

(6)

2Br₃⁻ → 3Br₂ + 2e⁻    E_diglyme = 4.02 V vs. Li/Li⁺

(7)

In 2017, Zhou’s group [94] used LiBr to facilitate a Li-CO₂ battery. With LiBr in the electrolyte, the specific capacity of the Li-CO₂ battery was significantly improved from ~3200 mAh g⁻¹ to 11,500 mAh g⁻¹. In the 500 mAh g⁻¹ fixed-capacity cycling test, the charge voltage of the cell with LiBr was also lowered by ~0.34 V to ~4.0 V, in accord with the potential of the Br₃⁻/Br₂ couple, and the cycling life was improved from 16 to 38 cycles. Furthermore, using ex situ XRD and UV-vis spectra characterization (displayed in Figure 4C), they found direct evidence of Li₂CO₃ removal with LiBr facilitation, and with UV-vis spectroscopy, they determined that Br₂ was reduced to Br₃⁻ while chemically oxidizing Li₂CO₃.

Later, in 2020, Kim and Byon [95] used Raman spectroscopy to characterize the extracted electrolyte of a LiBr-containing Li-O₂/CO₂ cell at various states of charge (SOCs). They observed Br₂…Br₃⁻ complex formation during charge, evidenced by the asymmetric stretching mode of Br₃⁻ at 201 cm⁻¹ in the Raman spectra and the red-shifted UV-vis absorption band at ~474 nm (Figure 4D). They thus proposed that the Br₂…Br₃⁻ complex plays a key role in Li₂CO₃ decomposition catalysis by avoiding the nonpolar Br₂ molecule from precipitating on the cathode before accessing the Li₂CO₃ surface. They also noticed that, from the second cycle forth, the discharge curve of the LiBr-containing Li-O₂/CO₂ cell showed a plateau at ~3.5 V and a following slope at 2.75–3.0 V, ascribed to the reduction of remaining Br₃⁻ and Br₂…Br₃⁻, respectively, whereas the slope was absent in the Li-O₂ cell. In addition, they found that a lower Li⁺ concentration is beneficial for stabilizing Br₂…Br₃⁻. However, due to the reduction of Br₂…Br₃⁻ during discharge, the unbound Br₂ would still precipitate on the cathode in the subsequent cycles, decreasing its activity. Co-solvents and additives could be a remedy to stabilize Br₂ and mitigate its precipitation. It is noteworthy that BrO, Br₂O, HOBr, and OBr⁻ species were not detected by UV-vis spectra during the entire discharge and charge processes; the authors speculated that no undesired chemical reactions were induced by Br⁻ or its oxidized forms. Nevertheless, further investigations should be conducted to ascertain the compatibility of Br₂ with other battery components.

In contrast to Br₂, much less attention has been paid to I₂ for Li₂CO₃ decomposition, despite extensive investigations on Li₂O₂ [97,98] and LiOH [38] decomposition and there being disputed mechanisms [102,103], primarily due to its lower redox potential than Li₂CO₃ decomposition. However, in some reports [31], the deposited Li₂CO₃ was removed during charge with the presence of I₂, possibly due to the involvement of the co-decomposition of 2Li₂CO₃ + C. Furthermore, it has been demonstrated that the cell chemistry and redox potential of I₃⁻/I⁻ are heavily influenced by the solvent and additives [102,104]. Shiga et al. found that [105,106] replacing the solvent (propylene carbonate, PC) with trimethyl phosphate (TMP) can dramatically elevate the redox potential of I₃⁻/I₂ from ~3.5 V (in PC) to 4.0 V (in TMP), and I₂ can thus successfully decompose Li₂CO₃. They also used a LiI–TMP-based electrolyte to investigate its influence on the performance of Li-CO₂ batteries [106]. However, the specific capacity was quite low at ~80 mAh g⁻¹, probably due to the use of blank carbon paper as the cathode, without high-surface-area materials.

The primary advantages of halide RMs are their high solubility and strong stability. However, their corrosive nature limits their practical application, especially with metal casings and current collectors; their volatility, toxicity, and environmental hazards (especially for Br species) are also concerns that limit their application.

3. Metal–Chelate Complexes

With much more variable molecular structures, organic compounds could theoretically offer more tunable redox potentials and electron transfer numbers than their inorganic counterparts. Among them, metal–chelate complexes, especially metallocycles (such as metal phthalocyanines and porphyrins), are given considerable interest because their central metal ions can offer redox center(s) with the desired potential(s), while the chelates can fine-tune redox potential and other properties, such as steric configuration, solubility, diffusivity, adsorbability, etc. Furthermore, these metallocycles can form oligomers and polymers to expand the π-π structure, offer more redox-active sites at the same potential to favor catalysis for multi-electron involved reactions, and enable various molecule conformations to serve different catalysis purposes, such as gas desulfurization [107], SOCl₂ reduction (in Li-SOCl₂ batteries) [108], mercaptan oxidation (for gasoline) [109], ORR [110,111,112], OER [113], etc. These features make them promising candidates as Li₂CO₃-decomposing catalysts.

In 2017, Liu et al. [22] reported studies on mononuclear and binuclear cobalt phthalocyanines (denoted by mono-CoPc and bi-CoPc, respectively, shown in Figure 5A) for their Li₂CO₃-decomposing potentiality. Mono-CoPc exhibited two redox couples at 3.90 V (Co^II/Co^III) and 4.31 V (Pc(−2)/Pc(−1)), whereas bi-CoPc exhibited two groups of redox couples at 3.74~3.82 V (Co^IICo^II/Co^IIICo^II and Co^IIICo^II/Co^IIICo^III) and ~4.10 V (Pc(−2)Pc(−2)/Pc(−1)Pc(−2) and Pc(−1)Pc(−2)/Pc(−1)Pc(−1)). Through potentiostatic charge tests of a glassy carbon (GC) working electrode (to electrochemically oxidize the RMs) and subsequent ex situ XRD characterization of the Li₂CO₃-preloaded carbon paper (which was electrically insulated from the electrodes, as shown in Figure 5B), Li₂CO₃ decomposition was observed only above the potential of the second active redox couple. “Active” means that the redox couple’s potential should be higher than the onset potential of Li₂CO₃ decomposition (3.71 V vs. Li/Li⁺); thus, the active redox couples are Co^IICo^II/Co^IIICo^II and Co^IIICo^II/Co^IIICo^III for bi-CoPc, and Co^II/Co^III and Pc(−2)/Pc(−1) for mono-CoPc. Based on these results, it was proposed that the successful chemical facilitation of Li₂CO₃ decomposition should be realized by two-electron RMs. Bi-CoPc was determined as a more suitable mobile catalyst for Li₂CO₃ decomposition due to its lower second active redox potential (3.82 V) compared to mono-CoPc (4.31 V), which can alleviate electrolyte degradation. Furthermore, using DEMS tests coupled with Li₂¹³CO₃ isotope labelling, a 59-fold higher ¹³CO₂ production amount, 3-fold higher ¹³CO₂ evolution rate, and 4-fold lower ¹²C-gas proportion (from parasitic reactions with cathode and electrolyte, etc.) were recorded. The cyclability of LAB (N₂-O₂, 78:22, v/v) was improved from 31 cycles without an RM, to 54 cycles with mono-CoPc, and further to 88 cycles with bi-CoPc, with a current density of 100 mA g⁻¹ and a fixed capacity of 500 mAh g⁻¹. Under the same operation settings, the cyclability of the Li-CO₂/O₂ (2:1, v/v) battery was also improved from 20 cycles without an RM, to 37 cycles with mono-CoPc, and further to 171 cycles with bi-CoPc.

Bi-CoPc has also been applied in Li-CO₂ batteries. In 2019, Guo and coworkers [117] constructed a Li-CO₂ battery with a pencil-trace cathode and a gel polymer electrolyte (GPE) containing bi-CoPc. The incorporation of bi-CoPc in the GPE resulted in the improvement of discharge capacity from 22,570 mAh g⁻¹ to 27,196 mAh g⁻¹ and Coulombic efficiency (CE) from ~39.9% to ~100% in the first full discharge–charge cycle, under a current density of 100 mA g⁻¹. In addition, the discharge voltage plateau was also raised from ~2.50 V to ~2.90 V. When operated under 1000 mAh g⁻¹ fixed capacity, the cell with bi-CoPc not only exhibited a reduced charge–discharge voltage gap from 1.90 V to 1.14 V, but also improved the cycle stability from 60 to 120 cycles, with a current density of 200 mA g⁻¹. By selected area electron diffraction (SAED) and energy-dispersive X-ray spectroscopy (EDS) characterizations of the discharged Au-decorated nickel foam cathode, they demonstrated that the discharge products were Li₂CO₃ and carbon. According to the formation and disappearance of thin-sheet products in the scanning electronic microscopy (SEM) images after discharge and recharge, respectively, they concluded that bi-CoPc can catalyze not only Li₂CO₃ decomposition, but also the co-decomposition of 2Li₂CO₃ + C. In 2021, they further combined bi-CoPc with an ethylenediamine-based LMA-protecting layer and in situ-formed gel polymer electrolyte, achieving 96 cycles in a 1000 mAh g⁻¹ fixed-capacity test under a high current density of 1000 mA g⁻¹ [118].

Later, in 2024, a comparative study between bi-CoPc and binuclear cobalt–manganese phthalocyanine (bi-CoMnPc) [114] was conducted to investigate their catalytic capability towards Li₂CO₃ decomposition. Based on potentiostatic, galvanostatic charging tests and ex situ XRD characterization, it was again demonstrated that Li₂CO₃ decomposition was only accelerated above the potential of the second active redox couple of the RMs. Since one of the central metal ions in bi-CoMnPc is replaced with Mn (compared to bi-CoPc), its redox potential is reduced from 3.74 V (Co^II/Co^III) to 3.48 V (Mn^II/Mn^III); thus, the second active redox couple changes from Co^IIICo^II/Co^IIICo^III (with bi-CoPc) to Pc(−2)Pc(−2)/Pc(−1)/Pc(−2) (with bi-CoMnPc), elevating the effective potential for Li₂CO₃ decomposition. But on the other hand, the reduced potential of Mn^II/Mn^III favors Li₂O₂/LiOH decomposition because of their lower decomposition potentials (Table 1) [18,119,120]. In the LAB (N₂-O₂, 78:22, v/v) cycling tests and subsequent ex situ FTIR characterizations, the cell with bi-CoMnPc exhibited even less Li₂CO₃ and Li-carboxylates accumulation in the air cathode, and the lifetime was improved from 193 cycles with bi-CoPc to 261 cycles with bi-CoMnPc at a current density of 100 mA g⁻¹_carbon and fixed capacity of 500 mAh g⁻¹. Furthermore, density functional theory (DFT) calculations were also carried out, considering three possible sets of reaction products, namely CO₂ and ³O₂ (via Equation (3)), CO₂ and ¹O₂ (via Equation (4)), and CO₂ and O₂⁻ (via Equation (5)), respectively. As shown in Figure 5C,D, the calculations reveal that, compared to the simultaneous transfer of two electrons (*CO₃²⁻ → *C₂O₆⁴⁻ → C₂O₆²⁻ accompanied by bi-CoPc²⁺ → bi-CoPc), the process of tandem electron transfer (*CO₃²⁻ → CO₃⁻ → C₂O₆³⁻ → C₂O₆²⁻) is not energetically favorable, regardless of whether the two electrons are gained by bi-CoPc²⁺ → bi-CoPc⁺ → bi-CoPc, 2bi-CoPc²⁺ → 2bi-CoPc⁺, or 2bi-CoPc⁺ → 2bi-CoPc. It was also revealed that the pathways to form CO₂ and ³O₂ (Equation (3)) or CO₂ and O₂⁻ (Equation (5)) are kinetically disadvantageous. Therefore, the dominant products of Li₂CO₃ decomposition are CO₂ and ¹O₂, as shown in Figure 5E. This is in accord with previous experimental results, where singlet ¹O₂ was detected by reacting with a 9,10-dimethylanthracene (DMA) probe and forming a DMA-O₂ adduct [121]. The reaction pathways for Li₂CO₃ decomposition on bi-CoMnPc³⁺ are similar to bi-CoPc²⁺, but CO₃²⁻ is preferably adsorbed at the Mn site due to its more positive charge than Co. These results draw a similar conclusion to Ref. [22], that for metal phthalocyanine-type RMs, two-electron ones are far superior to one-electron ones in catalyzing Li₂CO₃ decomposition.

Compared to their monomer and oligomer counterparts, polymerized metal–chelate complexes have diminished solubility in many solvents, but expanded π-π conjugation networks to favor multi-electron reactions. In 2019, Li’s group [115] developed a conjugated cobalt polyphthalocyanine (CoPPc, structure shown in Figure 5A) by a facile microwave method, using CoCl₂ and 1,2,4,5-tetracyanobenzene (TCNB) as the precursors. Although CoPPc is only slightly soluble in water, TEGDME, and common alcohols, they found it has solubility ≥0.1 mg mL⁻¹ in DMSO and NMP. By drop-casting the CoPPc/NMP solution onto a piece of carbon cloth with a loading of ~1 mg cm⁻², they prepared the gas cathode and assembled Li-CO₂ batteries with it. The cyclic voltammogram of the battery exhibited an onset oxidation potential at ~3.8 V and a reduction one at ~2.8 V in a CO₂ atmosphere. The cell exhibited both a high areal discharge capacity of 13.6 mAh cm⁻² and a high CE of 88%. In contrast, the cell with a bare carbon cloth air cathode showed a similar discharge capacity of 12.5 mg cm⁻² but a much poorer CE of 22%. Under a current density of 0.05 mA cm⁻² and fixed-capacity of 1 mAh cm⁻², the cell with CoPPc stably operated for 50 cycles. And the catalytic effect of CoPPc on decomposing Li₂CO₃ was demonstrated by galvanostatic charge tests of Li₂CO₃-preloaded electrodes, in which the charge voltage was lowered by ~0.2 V with the use of CoPPc. Later in 2022, a similar CoPPc with a defective π-π extended structure was used in an LOB by Han’s group [122] to achieve a high discharge voltage (2.7~3.0 V at current densities ranging from 0.01 to 0.5 mA cm⁻²) and reduced charge–discharge voltage gap (0.88 V at 0.5 mA cm⁻²).

In 2020, He’s group [123] reported a ruthenium-based metal–chelate molecule, ruthenocene (Ruc), as an RM in an LOB, using its redox couple Ru^II/Ru^III at 3.59 V to mediate Li₂O₂ decomposition. In the cycling test, the use of Ruc improved the Li-O₂ battery from 23 to 83 cycles under a current density of 0.1 mA cm⁻² and cutoff capacity of 500 mAh g⁻¹. By deconvolution of the ex situ X-ray photoelectron spectroscopy (XPS) peaks of the cycled air cathodes, they assessed the amount of Li₂O₂ and Li₂CO₃ decomposed during the charge of each cycle based on the peak areas. For the cell without Ruc, Li₂CO₃ content increased by 13.8% and 15.7% after the charge in the 5th and 10th cycle, respectively, indicating cathode and/or electrolyte degradation. In contrast, in the cell with Ruc, 59.9% and 57.3% of Li₂CO₃ content was decomposed after the charge in the 10th and 30th cycle, respectively. Thus, they determined that adding Ruc also favors the decomposition of Li₂CO₃, aside from Li₂O₂. However, the redox potential of Ru²⁺/Ru³⁺ at 3.59 V is insufficient to mediate Li₂CO₃ oxidation; thus, the actual reaction may be the co-decomposition of 2Li₂CO₃ + C, more investigations should be conducted to further ascertain its actual role.

In 2022, Liu’s group [124] screened ruthenium acetylacetonate (Ru(acac)₃) as an RM in a Li-CO₂ battery. During the CO₂ evolution reaction (CO₂ER) scan of the rotating disk electrode (RDE) test, they observed a significantly lowered onset potential from 4.21 V to 3.87 V vs. Li/Li⁺ with the addition of Ru(acac)₃, indicating its good catalytic activity towards the co-decomposition of 2Li₂CO₃ + C. In addition, by charging Li₂CO₃-preloaded electrodes and subsequent XRD and SEM characterizations, they also demonstrated the good catalytic effect of Ru(acac)₃ on Li₂CO₃ decomposition. The use of Ru(acac)₃ in the Li-CO₂ cell showed little influence on the full discharge performance, but significantly lowered the recharge plateau by 0.64 V to ~3.8 V and improved CE from 18.1% to 100%, with a moderate terminal voltage of 4.26 V. However, in the cyclability tests, the cell with Ru(acac)₃ showed an even poorer performance (36 cycles) than the blank control (40 cycles) under a current density of 100 mA g⁻¹ and cutoff capacity of 1000 mAh g⁻¹, attributed to the shuttling effect of RM to LMA. By using a Zn-MOF nanoplate modified glass fiber (M-GF) separator to suppress Ru(acac)₃ shuttling, the cycling life was significantly improved to 142 cycles.

In 2022, Qian’s group [125] used a chiral salen-Co(II) complex, (1R,2R)-(-)-N,N-bis(3,5-di-t-butylsalicylidene)-1,2-cyclohexanediaminocobalt(II) (abbreviated as Co(II)), as a multi-functional RM in a Li-O₂ battery. The use of Co(II) improved the discharge capacity of the Li-O₂ cell from 4900 to 22,000 mAh g⁻¹, presumably due to its solvation effect on Li⁺ and thus the induction of solution pathway of the ORR. During charge, they observed a much lower (~3.5 V) voltage plateau than the blank control (~4.1 V), attributed to Co(II)’s first redox couple at 3.4 V. In line with its second redox couple at 4.1 V, they also proposed its effect on Li₂CO₃ decomposition. They conducted galvanostatic charging tests for a Li₂CO₃-preloaded electrode. The cell with Co(II) was successfully charged to the theoretical capacity (1000 mAh g⁻¹) at a moderate voltage plateau of ~4.2 V, while the cell without Co(II) was cut off at 4.7 V and ~400 mAh g⁻¹. XRD, SEM, XPS, and FTIR characterizations of the charged electrode demonstrate the complete removal of Li₂CO₃ in the cell with Co(II), in contrast to the considerable Li₂CO₃ residue without Co(II). The cyclability of the cell was also improved from 50 to 252 cycles under a current density of 200 mA g⁻¹ and cut-off capacity of 500 mAh g⁻¹.

Inter-electrode crosstalk (shuttling) is an important issue affecting LABs/LOBs’ cyclability and anode overpotential, especially during charge [126]; the situation is even more severe when RMs are incorporated. A common countermeasure is placing SSE membranes between the RM-containing electrolyte (catholyte) and the LMA; however, most currently used SSEs, such as lithium aluminum germanium phosphate (LAGP) [127], lithium aluminum titanium phosphate (LATP) [128], lithiated Nafion–polyvinylidene difluoride (PVDF) composites/blends [22,129], etc., suffer from low ionic conductivity at room temperature, as well as high cost, fragility (for LAGP, LATP, and many other ceramic SSEs), and swelling/dissolution in organic solvents (for Nafion–PVDF composites/blends). Possible alternatives include developing grafted RMs [130] and enhancing RMs’ adsorbability on the air cathode. In 2023, Liu and coworkers [116] compared three cobalt porphyrin derivatives, namely cobalt tetramethoxyphenyl porphyrin (CoTMPP), cobalt mesotetraphenylporphyrin (CoTPP), and cobalt octaethyl-porphyrin (CoOEP), as RMs in LABs (N₂-O₂, 78:22, v/v), as shown in Figure 5F. Through cyclic voltammetry (CV) tests and calculations with Randles–Ševčík [131,132] and Brown–Anson [133] equations, it was found that the redox reactions of RMs are controlled by adsorption on the cathode surface rather than diffusion, and the side substituent groups exert critical influences on RM adsorbability, and consequently cell cyclability. With the highest adsorbability and suitable redox couples (the other two, CoTPP and CoOEP, lack a second active redox couple) to facilitate Li₂CO₃ decomposition, CoTMPP enabled 200 cycles of operation of an LAB cell under 100 mA g⁻¹_carbon and 500 mAh g⁻¹_carbon, even without an SSE membrane to suppress RM shuttling and deactivation.

Currently, the main drawback of metal–chelate-type RMs is their high prices, even comparable to precious metals (tens to hundreds of USD per gram). However, mass production would dramatically lower the price. For example, bi-CoPc at AR grade costs ~USD 150 per gram, whereas the widely used gas desulfurization catalyst, binuclear cobalt phthalocyanine sulfonate, costs only ~USD 15 per kilogram. The synthesis of bi-CoPc (and its derivatives) would produce considerable mono-CoPc (or its derivatives) impurity, which is costly to separate with chromatography, but the chromatography separation may not be needed in the first place, because mono-CoPc can function as a Li₂O₂/LiOH-decomposing RM in LABs/LOBs. Another issue is their variable solubility; for example, unmodified (also known as “neat”) metal phthalocyanines and porphyrins typically have a low solubility of several mmol L⁻¹ in commonly used organic solvents [22,114], but modifications on side substituent groups can readily address this.

4. Metal-Free Organic Compounds

Similar to metal–chelate-type RMs, the diverse organic structures of metal-free organic compounds provide wide tunability for their redox potential and other physical/chemical properties. In recent years, many metal-free organic RMs have been discovered for catalyzing Li₂CO₃ decomposition catalysis.

In 2019, Manthiram’s group [134] reported the utilization of phenyl disulfide (PDS) in Li-CO₂ batteries to facilitate CO₂ reduction. PDS can spontaneously react with metallic lithium to break the weak S-S bond and form thiophenolate anions (TP⁻, as shown in Figure 6A). Then, in the discharge process, the dissolved CO₂ molecules can attach to TP⁻ to form an intermediate S-phenyl carbonothioate (SPC⁻). Through a series of reduction reactions, SPC⁻ can release O₂⁻ and amorphous C, and then O₂⁻ chemically reacts with CO₂ to form CO₃²⁻, which finally combines with Li⁺ to form Li₂CO₃, as shown in Figure 6A. They proposed that TP⁻ could also assist Li₂CO₃ decomposition by operating the cycle in the reverse direction. This hypothesis is supported by the completed removal of Li₂CO₃ in the cell with PDS after charge, while the Li-CO₂ cell without PDS showed many residual Li₂CO₃ granules. However, as all the cycling tests were conducted in Li-CO₂ cells, whether the actual reaction was Li₂CO₃ decomposition or the co-decomposition of 2Li₂CO₃ + C remains elusive, and should be further investigated in the future.

In 2020, Zhang’s group [136] reported a tri-functional RM, 2,5-di-tert-butyl-1,4-dimethoxybenzene (DBDMB), which can capture highly reactive O₂⁻ during discharge, forming a complex with O₂⁻ and Li⁺ to mediate the solution pathway of the ORR; thus, the parasitic reactions could be alleviated, with a high Li₂O₂ yield of 96.6%. This also leads to alleviated cathode passivation and improved discharge capacity. Moreover, with a redox couple of DBDMB/DBDMB⁺ at ~4.20 V vs. Li/Li⁺, it can decompose the three major solid-state species in the battery, namely Li₂O₂, LiOH, and Li₂CO₃, which is demonstrated by XRD and SEM characterizations of the Li₂O₂/LiOH/Li₂CO₃ preloaded electrode after galvanostatic charge. The LOB with DBDMB completed 243 and 90 cycles with 1000 and 3000 mAh g⁻¹ cutoff capacity, respectively, under a high current density of 1000 mA g⁻¹, significantly higher than the blank cell, which survived only 23 and 3 cycles, respectively.

In 2021, Chen and coworkers [137] used phenoxathiine (PHX) to assist a Li-CO₂ battery. With the redox potential of PHX/PHX⁺ at 4.275 V, they found that PHX⁺ can readily chemically decompose commercial Li₂CO₃ powder to release CO₂. Furthermore, using DMA as a molecular trap for singlet oxygen, they detected the presence of a DMA-O₂ adduct using ¹H NMR, thus demonstrating ¹O₂ as the reaction product of Li₂CO₃ decomposition. They proposed the PHX-facilitated Li₂CO₃ decomposition as follows:

4PHX⁺ + 2Li₂CO₃ → 4Li⁺ + 2CO₂ + ¹O₂

(8)

Aside from Li₂CO₃ decomposition, they also observed a decrease in D-band intensity for carbon in the Raman spectra after charge, indicating the decomposition of amorphous carbon. Using isotope labelling and mass spectroscopy, they observed ¹³CO₂ evolution after the exposure of the Li₂CO₃-¹³C mixture to PHX⁺. These results demonstrate that carbon was also decomposed along with Li₂CO₃, suggesting that PHX can also catalyze the co-decomposition of 2Li₂CO₃ + C.

In 2022, Liu’s group [138] reported o-phenylenediamine (OPD) as a bi-functional RM in a Li-CO₂ battery. The Li-CO₂ cell with OPD exhibited a high discharge capacity of 20,314 mAh g⁻¹ under a current density of 200 mA g⁻¹, significantly higher than that without OPD (6620 mAh g⁻¹). Under current densities of 400 and 600 mA g⁻¹, the cell with OPD still maintained considerable performances, 8398 and 4338 mAh g⁻¹, respectively, much higher than that without OPD (4145 and 2842 mAh g⁻¹, respectively). By natural population analysis (NPA) and SEM characterization of the discharged cathode, they proposed that OPD can form a stable adduct with the CO₂ reduction reaction (CO₂RR) intermediate CO₂⁻• to promote the liquid phase growth of discharge products, thereby alleviating cathode passivation and improving discharge capacity. During charge, on the other hand, with the two redox pairs OPD/OPD⁺ and OPD⁺/OPD²⁺ positioned at 3.22 V and 3.75 V, respectively, they observed a significant decrease in charge voltage from 3.99 V to 3.21 V when OPD was used in the Li-CO₂ cell. Under a current density of 200 mA g⁻¹ and cutoff capacity of 1000 mAh g⁻¹, the cyclability of the Li-CO₂ cell was improved from 56 to 125 cycles with OPD employment. They also substituted OPD with benzene, finding that the discharge and charge performance were similar to the blank control, while the substitution of OPD with p-phenylenediamine (PPD) gave a similar performance to OPD, demonstrating the key role of the -NH₂ group in CO₂RR and CO₂ER mediation.

In 2022, Park and coworkers [135] investigated several p-type organic molecules, including methylphenothiazine (MPTZ), N,N,N,N’-tetramethyl-p-phenylenediamine (TMPD), 5,10-dimethylphenazine (DMPZ), and 1,4-ditert-butyl-2,5-dimethoxybenzene (DBB), for their catalytic effects on Li₂CO₃ decomposition. MPTZ and DBB exhibit a single redox peak at 3.71 and 4.18 V, respectively, whereas TMPD and DMPZ both show a first redox peak at ~3.2 V, and a second peak at 3.73 and 3.90 V, respectively. They conducted 10 cycles of repeated galvanostatic charging tests with these RMs and a cutoff voltage of 4.2 V, resulting in cumulative capacities that corresponded to 10~108% of the theoretical capacity of the complete Li₂CO₃ decomposition. Using FTIR and XRD analyses, they observed a significant decrease in the Li₂CO₃ signal in the samples with TMPD, DMPZ, and DBB. Through Li₂¹³CO₃ labelling and DEMS tests, they further determined the efficacy of TMPD, DMPZ, and MPTZ. As shown in Figure 6B, TMPD exhibited the highest activity toward Li₂CO₃ decomposition, with a 38-fold higher ¹³CO₂ production (from Li₂¹³CO₃ decomposition) and a 10-fold smaller ¹²CO₂ proportion (from cathode and electrolyte degradation, etc.) than the blank control. Furthermore, they used the DMA probe to quantitatively detect ¹O₂ production during Li₂¹³CO₃ decomposition, finding that the use of MPTZ, TMPD, and DMPZ can lower the ¹O₂ proportion (to 2 molar equivalents of oxidized Li₂CO₃) from 51% to 16%, 9%, and 7%, respectively. They also observed an unusually low onset potential for ¹³CO₂ evolution below the redox potential of TMPD⁺/TMPD²⁺ (3.73 V), accompanied by H₂ evolution above 3.2 V. Combined with ex situ NMR spectra, they conjectured the H-abstraction from TMPD to induce a side reaction of 2H⁺ + Li₂CO₃ → 2Li⁺ + 2CO₂ + H₂O. Due to the limited stability of TMPD, they suggested DMPZ as an optimal RM with moderate activity (8-fold higher ¹³CO₂ production and 4-fold smaller ¹²CO₂ proportion), good stability, and good ¹O₂ suppressing effects. It is also noted that, contrary to the binuclear metal phthalocyanine-type RMs which necessitate one-step two-electron transfer by a single molecule to oxidize Li₂CO₃, they found that these p-type molecules could catalyze Li₂CO₃ oxidation through tandem electron transfer.

5. Conclusions and Prospects

Along with the development of other components and technologies in LABs/LOBs, many achievements have been made in Li₂CO₃-decomposing RMs, and hope has been kindled to finally cure LABs’ “Achilles heel”. However, there are still many questions left unanswered, and arduous challenges lie ahead. Some examples are listed below.

(1)

To address the ¹O₂ issue in Li₂CO₃ decomposition and enable true reversible cell chemistry, stable and efficient ¹O₂ quenchers are urgently needed. Despite many recent investigations for ¹O₂ quenchers in Li₂O₂ decomposition [139,140,141,142,143], to the authors’ knowledge there has been only one study conducted on Li₂CO₃ decomposition by Fruenberger’s group [121]. In that research, the ³O₂ evolution rate is 2~3 orders of magnitude higher when using a ¹O₂ quencher, 1,4-diazabicyclo[2.2.2]octane (DABCO), than that without a quencher; nevertheless, only ~1% of the produced ¹O₂ was quenched even when using the quencher. The reason for such unsatisfactory quenching efficiency may be attributed to three aspects. First, the potential required to decompose Li₂CO₃ (3.71~3.82 V vs. Li/Li⁺) is beyond the electrochemical window of many quenchers, such as azides and aliphatic amines (with upper limit to 3.5~3.6 V). Second, amine-based quenchers have logarithmical correlations between quenching efficiency and ionization potential (and hence oxidation potential). The quenching efficiency drops by four orders of magnitude when the oxidation potential increases by 1 volt [140,144]. Third, the Lewis basicity of amide- and azide-based quenchers can impair quenching [144], which is even exacerbated in Li₂CO₃ decomposition by adsorbing CO₂. Therefore, finding efficient approaches to address the ¹O₂ issue remains a highly challenging task.

(2)

More efforts should be devoted to ascertaining the mechanisms of Li₂CO₃ decomposition, as well as the corresponding catalysis processes. As discussed above, there have been contradicting experimental results. On the one hand, many Li₂CO₃-decomposing RMs, such as Br₂ (or the Br₂…Br₃⁻ complex) [95], I₂ (in TMP) [105,106], bi-/multi-nuclear metal phthalocyanines [22,114,115], and porphyrins [116], can be categorized into two-electron (or multi-electron) RMs, which is supported by potentiostatic charging tests (with Li₂CO₃-preloaded carbon paper insulated from electrodes, as shown in Figure 5B) and DFT calculations [75,114]. On the other hand, some recently reported one-electron RMs, such as MPTZ [135], TMPD [135], DMPZ [135], DBDMB [136], PHX [137], Ru(aca)₃ [124], and the salen-Co(II) complex [125], can also catalyze Li₂CO₃ decomposition. Other one-electron RMs (Ruc [123] and OPD [119]) and two-electron RMs (SPC⁻ [134] and I₂ in ethereal electrolytes [31]) have demonstrated their capability in promoting the co-decomposition of 2Li₂CO₃ + C, but their efficacy in Li₂CO₃ decomposition should be further investigated.

Here, we provide some suggestions for future investigations on Li₂CO₃-decomposing RMs to gain more insights into the mechanisms of Li₂CO₃ decomposition and the corresponding catalysis processes. First, DEMS combined with isotope labelling is a powerful tool to quantify evolved (or consumed) gases, offering plentiful information to gain insights into the mechanisms, and is highly recommended. Second, machine learning (ML) has been emerging as a useful computational tool to accelerate DFT [145] and molecular dynamics (MD) [146,147] simulations, as well as bridge different models [148]. It can also be utilized to analyze Li₂CO₃ decomposition and corresponding catalysis processes in the near future. Third, potentiostatic charging (to oxidize the RM to a certain redox state) combined with the electrically insulated Li₂CO₃ target is an effective method to avoid the interference of direct electron transfer between the solid-state working electrode and Li₂CO₃, and can give more insights into the mechanisms. Fourth, carbon-free working electrodes (such as nanoporous gold [149], Au-decorated nickel foam [117], TiC [150], nanostructured Co₃O₄ [151], etc.) are also highly recommended to ascertain or avoid the involvement of the co-decomposition of 2Li₂CO₃ + C. Fifth, amorphous carbon and trace amounts of H⁺ [135,152] could be purposefully added into the LAB system and function as additives to shift cell chemistry, improving reaction reversibility and round-trip efficiency. With the development of efficient Li₂CO₃-decomposing RMs and ¹O₂ quenchers, we believe this review will shed light on future works in the field and contribute to enabling LABs that can stably operate in ambient air.