The rapid increase in CO₂ emissions caused by the overuse of fossil fuel has resulted in many severe environmental issues [1]. Therefore, great efforts have been made to search for valuable use of CO₂, including the development of electrochemical and photochemical CO₂ reduction technologies [2, 3]. Among them, energy storage devices such as Li metal based CO₂ batteries (4Li⁺ + 3CO₂ + 4e^- ↔ 2Li₂CO₃ + C) have attracted increasing interest, which can operate with a high discharge potential (~2.8 V) and considerable theoretical energy density of 1876 Wh/kg [4-10], through the reduction of CO₂ in the discharge process.

With regard to Li-CO₂ batteries, the activation of CO₂ to form CO₂-related intermediate species is important to the discharge reaction. However, the existence of carbon-oxygen double bond makes the CO₂ molecule very stable and difficult to directly accept electron and reduce to CO₂^•- [10-15]. For example, CO₂ was initially proposed for use in Li-O₂ battery as a "gas assist" additive, in this case, O₂ is the electroactive species, and CO₂ can only reacts with reduced O₂ species by subsequent chemical reactions [16-18]. Recently, by in situ ambient pressure X-ray photoelectron spectroscopy, Wang et al. verified that pure CO₂ reduction is not electrochemically active at room temperature on porous carbon electrode in organic ionic liquid [19]. In order to activate CO₂ and promote CO₂ reduction kinetics, homogeneous liquid phase catalyst has been used. Yin et al. investigated the possibility of using quinones (Q) as liquid catalysts for CO₂ reduction in Li-CO₂ system [20]. Quinones were reduced at the cathode first to form Q^2-, then chemical reaction between Q^2- and CO₂ occurred to form quinone-CO₂ adducts, which were further reduced to discharge product. Slightly different from the activation mechanism of quinones, Khurram et al. employed an alkyl amine, 2-ethoxyethylamine (EEA) to react with CO₂ to form an EEA-CO₂ adduct by the formation of a N-C bond [9]. The EEA-CO₂ adduct is not only highly electroactive in the electrolyte, but that the N-C bond is selectively cleaved upon electron transfer, ultimately facilitating the conversion of CO₂ gas to Li₂CO₃. Except for the effect of catalyst molecules with high e^-/CO₂ affinity, in both of the two cases, Li⁺ is also implicated in the formation process of the active adduct species, and enables crucial shift to discharge product.

Recently, Khurram et al. proposed that coupled e⁻/Li⁺ transfer can activate CO₂ to form the Li⁺-CO₂^- intermediate, avoiding the formation of unstable CO₂^- radical only, and suggested that Li⁺ can also form a contact ion pair (CIP) with CO₂-derived anion, facilitating the subsequent CO₂ reduction steps [21]. However, the amount of available Li⁺ is usually limited because of the high desolvation energy of Li⁺ in organic electrolyte (526.7 kJ/mol for dimethyl sulfoxide, 494.9 kJ/mol for propylene carbonate and 385.0 kJ/mol for tetraethylene glycol dimethyl ether), thus the formation of Li⁺-paired species in organic electrolyte is difficult, which leads to sluggish CO₂RR kinetics [21]. What is more, increasing the concentration of Li salt cannot enhance the availability of Li⁺ due to the formation of solvent-sheathed CIP, in which Li⁺ and anion are completely aggregated to form a fluid network [22]. Considering the factors above, adding an appropriate positive ion (M⁺) with lower desolvation energy than that of Li⁺ to couple an electron and assist the transfer of the electron to CO₂, may induce strong interactions between M⁺ and CO₂ reduction intermediate in electrolyte, and therefore facilitate to achieve better CO₂ reduction kinetics.

In addition to reaction kinetics, cathode passivation caused by insulator discharge products is also a key factor that prevents the battery from achieving high energy density. It has been reported that the film-like discharge products produced by the surface growth pathway can hinder the conduction of electrons during the discharge process, resulting in passivation of the cathode, while the solution phase growth pathway can keep the cathode surface sufficiently exposed, so that the discharge process can continue, thus greatly improving the capacity of the battery. The reason why the solution phase growth pathway can be realized lies in the enhanced solubility and/or reactivity of discharge intermediate, which can be achieved by using soluble catalysts in electrolyte. In this regard, until now, few researches on the strategy to increase the capacity by regulating discharge paths have been explored in the Li-CO₂ battery. With these motivations, in this work, soft Lewis acid cetyl trimethyl ammonium bromide (CTAB, which structure was shown in Fig. S1 in Supporting information) with quaternary ammonium CTA⁺ cation was introduced into a dimethyl sulfoxide (DMSO) based Li-CO₂ battery for the first time. The choice of long alkyl chain is based on the principle that the hydrophobic microenvironment can enhance the diffusion of CO₂ gas [23]. Additionally, the DMSO used in this system has a better ability to dissolve CO₂ [21, 24]. With the CTAB additive, the Li-CO₂ battery can operate at high discharge current from 0.2 mA/cm² to 0.35 mA/cm² and displayed an excellent CO₂RR performance (a real discharge capacity up to 20 mAh/cm² at a current density of 0.2 mA/cm²). A series of evidence have shown that the addition of CTAB greatly improves the dissolving ability of discharge intermediates, facilitating the formation of discharge product in the solution phase, which can alleviate the cathode passivation. Ab initio molecular dynamics (AIMD) simulations showed that the quaternary ammonium N⁺ group can form more stable CIP with e^-/CO₂ in DMSO than Li⁺, thus enhance the ability of activating CO₂ and improve the CO₂ reduction process.

Fig. 1a shows a typical configuration model of Li-CO₂ battery with CTAB additive, in which carbon paper (CP) was used as cathode and DMSO containing 1 mol/L LiCF₃SO₃ as the electrolyte. Fig. 1b displays the proposed discharge mechanism of CTA⁺ assisted pathway of CO₂RR. With the existence of CTA⁺, CO₂ is reduced by CTA⁺/e^- pair to form CTA⁺-CO₂^- CIP, and then reacts with Li⁺ to produce Li₂CO₃ and C with the regeneration of CTA⁺. The electrochemical performances of Li-CO₂ batteries with and without the addition of CTAB (20 mmol/L) were investigated. Fig. 1c presents the cyclic voltammograms (CVs) response of batteries at a constant scan rate of 0.1 mV/s. It can be clearly seen that the battery with CTAB exhibits a significantly higher reduction onset potential and a larger peak current density compared to the battery without CTAB, implying faster CO₂ reduction kinetics with the assistance of CTAB [20]. The discharge capacities of the batteries are given in Fig. 1d, which shows that with the addition of CTAB, a significantly increased discharge capacity of more than 20 mAh/cm² was achieved, while the battery without CTAB additive can only displayed a pitiful capacity of about 2 mAh/cm², at a current density of 0.2 mA/cm² with a discharge terminal voltage of 2.0 V. Additionally, even at higher current densities, batteries with CTAB addition still exhibit considerable discharge capacity (Fig. S2 in Supporting information), further confirmed the significantly improved CO₂RR performances with the addition of CTAB. The effect of the concentration of the added CTAB on the discharge capacity was also investigated. As shown in Fig. S3 (Supporting information), the discharge capacity was increased with increasing the concentration of CTAB from 2 mmol/L to 20 mmol/L, while decreased slightly when the concentration was raised to 30 mmol/L. The increase in discharge capacity is due to the more CTA⁺ provided with increasing CTAB concentration [25, 26], and the decrease of discharge capacity may be related to the reduced ion mobility of the cations and anions associated with CO₂RR caused by the high concentration of salt. In order to prove that the capacity of CTAB catalyzed Li-CO₂ battery comes from CO₂ reduction, rather than the side reaction in which CTAB participates, the battery is discharged in an Ar atmosphere and the results show that the capacity is pitiful (Fig. S4 in Supporting information). The results above showed that CTAB can facilitate the reduction reaction of CO₂, and significantly improve the electrochemical performances, including reduced discharge over-potential and increased discharge capacity. The influence of Br^- anion on the CO₂ reduction process can be ruled out by investigating the LiBr (20 mmol/L) added battery, the reason is concerned that the addition of LiBr does not improve the capacity of the battery (Fig. S5 in Supporting information). Br^- anion in CTAB can act as an effective oxidizing intermediator due to the generation of Br₂ which can chemical oxidize the discharge products [27], and the CTAB catalyzed battery exhibited a stable round-trip performance of 47 cycles, illustrating the potential application of CTAB for the rechargeable Li-CO₂ battery (Fig. S6 in Supporting information). The corresponding scanning electron microscope (SEM) images, X-ray diffraction (XRD) and Raman patterns of charged or cycled cathodes were shown in Figs. S7 and S8 (Supporting information), further verified the catalytic action of CTAB in charging process.

To verify the morphology and distribution of the discharge products, deep discharged cathodes with and without CTAB additive were examined by SEM, respectively. Obviously, in the absence of CTAB, the CP surface is almost completely covered with fine particles of discharge products (Fig. 2a), leading to the cathode passivation and poor CO₂RR kinetics of the battery. While vertical growth of large-size discharge products can be observed in the CTAB added battery, displaying a solution-phase growth pattern (Fig. 2b). To further illustrate this, the glass fiber separator in the deep discharged batteries with and without CTAB were also investigated as shown in Fig. S9 (Supporting information). For the battery without CTAB (Fig. S9a), the glass fiber is almost clean, while the glass fiber in the CTAB added battery was covered by foliated products (Fig. S9b), further demonstrated that the discharge products exhibit obvious solution-phase growth behavior, since the glass fiber is insulating. The XRD pattern and Raman shift spectroscopy were applied to investigate the discharge products in the CTAB added battery. As shown in Fig. 2c, the diffraction peaks at 21.2°, 30.5°, 31.6° can be assigned to the (110), (202), (002) planes of Li₂CO₃, respectively [6, 28, 29], and the characteristic Raman spectra peak at 1089 cm^-1 is highly consistent with the standard patterns of Li₂CO₃ (Fig. 2d) [6], indicating the existence of Li₂CO₃ in the discharge products. To further confirm the component of discharge products in CTAB added Li-CO₂ batteries, sufficient HCl aqueous solution (1 mol/L) was used to remove Li₂CO₃. It can be found that the products presented a plate like morphology after being treated with HCl (Fig. S10 in Supporting information), which is characterized as carbon by Raman scattered spectrum (Fig. S11 in Supporting information), suggesting that the discharge products are Li₂CO₃ and C.

In order to further understand the effect of CTAB on the discharge process in Li-CO₂ battery, the evolution of discharge products in CTAB added battery was tracked at a current density of 0.2 mA/cm². As shown in Fig. 3, it can be seen that there are some aggregated and isolated discharge products formed after being discharged to 1.5 mAh/cm² (Fig. 3a), and the amount of which was increased with the discharge capacity raising from 1.5 mAh/cm² to 12 mAh/cm² (Figs. 3b-d). The characteristic of the formation of discharge products is that as the discharge process proceeds, large sized discharge products were continously aggregated on the cathode surface with a mode of expanding in both width and depth, avoiding the blocking of e^- transfer channels on the cathode-electrolyte interface, which exhibits typical solution-phase growth behavior and is favorable to consecutive discharge process as well as the improvement of discharge capacity.

Based on the results above, the mechanisms of surface and solution-phase growth modes with and without CTAB are proposed and given in Fig. 3e, the formation and migration of M⁺−CO₂^- CIP play the key role in the solution-phase growth of discharge products, in which the interaction of CTA⁺ with CO₂^- is important to understanding the enhanced CO₂RR. In this regard, AIMD was used to investigate the structures and properties of the CIP formed by Li⁺ and CTA⁺ pair with CO₂^- in DMSO solvent, respectively. Representative snapshots of solvent-separated ion pair (SSIP) and CIP solvation models were shown in Figs. 4a-d. Each calculated solvent boxes contain DMSO solvent, three CO₂^- (CO₂ + e^-), and three simplified ammonium CH₃CH₂(CH₃)₃N⁺ (denoted as N⁺) or Li⁺ ions. After 15 ps, it can be seen from Fig. 4b that three CO₂^- ions are almost all around an N⁺ group, and the distance of CO₂^- from N⁺ is 3.21, 3.379 and 4.9 Å, respectively, demonstrating that N⁺ can easily contact with CO₂^-. However, only one CO₂^- is closed to the target Li⁺ (the distance between them is 3.34 Å) (Fig. 4d), the strong solvation interaction between DMSO and Li⁺ makes DMSO form a tight solvation shell around Li⁺, hindering the contact of CO₂^- and Li⁺. In contrast, N⁺ can easily contact with CO₂^- and form more stable N⁺-CO₂^- CIP than that of Li⁺ due to the weak interaction between N⁺ and DMSO. The energy barrier for the formation of the two CIP systems was further investigated as shown in Fig. 4e. Energy snapshots were conducted on four structures with an integration time-step of 1 fs during the AIMD time. It could be found that both the two CIP structures have lower energy over the entire AIMD time than their corresponding SSIP structures, confirming that the CIP structures are thermodynamically preferred in the electrolyte and isolated CO₂^- radicals are less likely to form alone. Obviously, the energy difference of N⁺-CO₂^- CIP (-1.672 eV) is larger than that of Li⁺-CO₂^- CIP (-0.893 eV), which indicates that N⁺ can form stable CIP with CO₂^- easily. Therefore, according to the AIMD results above, it can be known that CTA⁺ can enhance the CO₂ reduction kinetics by forming a more thermodynamically favorable CTA⁺-CO₂^- CIP than Li⁺-CO₂^- CIP. The former can combine with Li⁺ in the electrolyte during its migration process to generate discharge products, enhancing the ability of forming products in the solution phase. The comparison of discharge pathway of Li-CO₂ batteries with and without CTA⁺ was shown in Fig. 4f, obviously, the different energy barrier for the formation of the two CIP systems changed the reaction pathway of CO₂ reduction, and the consequences of the CTA⁺ catalyzed reaction route is the formation of stable CIP transfer, which dominates solution phase product formation and thus improves the discharge capacity of the battery.

In summary, in this work, CTAB was successfully introduced into Li-CO₂ battery system to greatly improve the electrochemical performances, including discharge over-potential and discharge capacity (a real discharge capacity can up to 20 mAh/cm² at a current density of 0.2 mA/cm² by using carbon paper cathode). Experiments coupled with AIMD showed that the CIP formed by CTA⁺ and CO₂^- is more stable than that of Li⁺ and CO₂^-, thus the CO₂ reduction process can be accelerated with the assistance of CTA⁺. In addition, the introduced CTAB promotes the mobility of the discharge intermediates and makes the discharge products grow through the solution phase pathway, greatly eliminating the passivation of the cathode and finally releasing the battery energy. This work can facilitate the development of Li-CO₂ battery and provide a novel understanding of the CO₂ reduction chemistry in organic systems.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work is financially supported by National Science Foundation of China (Nos. 21701145 and 21701146), China Postdoctoral Science Foundation (Nos. 2017M610459 and 2018T110739).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2021.10.089