The development of cutting-edge technologies to fulfill the demanding energy density and safety standards of lithium-ion batteries (LIBs) is essential for the rapidly expanding electric vehicles and grid storage sectors [1,2]. Recently, Li-metal batteries (LMBs) have garnered interest due to the high specific capacity (3680 mAh/g) and low redox potential (-3.04 V vs. SHE) offered by the Li anode [3,4]. Regrettably, the formation of Li dendrites during the charging process has resulted in safety hazards like fires and explosions [5,6]. These concerns regarding safety have significantly restricted the progress of lithium metal batteries (LMBs).

However, the development of electrolytes has long lagged behind that of electrodes, and there is considerable attention on the research of electrolytes compatible with both positive and negative electrodes, among which carbonate and ether-based electrolytes are the most common [7-9]. In addition, ether solvents are considered to be more compatible with LMA among the frequently used solvents in LIBs electrolytes due to their high lowest unoccupied molecular orbital (LUMO) [10-12]. However, ethers have relatively low flash points, and gas generation takes place in ether-based electrolytes at a high voltage of approximately ≈4.6 V even in a LHCE [13,14], making them unsuitable for operation at very high voltages. Hence, carbonate electrolytes (i.e., EC-based electrolyte), renowned for their robust oxidative stability and high flash points, have been extensively employed in commercial LIBs, making them the most cost-effective and appropriate choice for integration into high-voltage LMBs [15].

Ethylene carbonate (EC) occupies a prominent position in carbonate electrolytes, as the increase in its dielectric constant promotes the effective dissociation of lithium salts (such as LiPF₆), while its reduction can also facilitate the formation of the required SEI layer on the electrode surface, thereby enhancing stability [16,17]. However, the strong interaction between Li⁺ and EC increases the desolvation energy barrier, leading to uncontrolled lithium growth due to poor quality of the interface layer and low CEs [18]. Meanwhile, EC preferentially contacts the electrode and undergoes decomposition, resulting in the formation of a weak SEI abundant in lithium carbonate (Li₂CO₃), lithium ethylene dicarbonate (LEDC), and lithium ethylene monocarbonate (LEMC) [19-21]. Therefore, when designing EC-based electrolytes, it is essential to concurrently induce a protective solid-electrolyte interface by modulating the solvation structure. Given that the solvation structure of Li⁺ is influenced by the competition between different solvents and anions for Li⁺ coordination, it is feasible to reshape the solvation structure of Li⁺ by modifying its coordination configuration.

In this study, we introduced bis(2,2,2-trifluoroethyl) carbonate (TFEC), a carbonate with weak solvation capability. The strong electron-withdrawing effect of the fluorine atom in TFEC (-CF₃ group) reduces the electron cloud density around the oxygen atom of the carbonyl (carbonate) group, weakening the interaction strength between Li and EC. This leads to more participation of PF₆^- anions and TFEC in the solvation structure of Li⁺, promoting the formation of SEI rich in -CF₃ and LiF. At the same time, it enhances the stability and uniformity of CEI. Therefore, the Li||LiMn_0.8Fe_0.2PO₄ full cell exhibits a stable cycling performance showing limit capacity decay with a capacity retention of 62.5% after 800 cycles, which is far higher than those of the base electrolytes (29.8%). In addition, under the TFEC electrolyte system, Li||Li symmetric cells exhibit excellent long-term cycling stability, sustaining 800 h at a current density of 0.5 mA/cm² and a capacity of 0.5 mAh/cm².

The base electrolyte consists of 1.0 mol/L LiPF₆ dissolved in the mixed solvents of EC, ethyl methyl carbonate (EMC), and dimethyl carbonate (DMC) with a quality ration of 1:1:1, denoted as the EED electrolyte. The TFEC were added to the above base electrolytes, denoted as the EEDT electrolyte. The physical properties of EC, DMC, EMC, and TFEC are shown in Table S1 (Supporting information). The ionic conductivity of the EEDT electrolyte (10.90 mS/cm) shows a slight deviation from that of the EED electrolyte (11.48 mS/cm), as shown in Fig. S1 (Supporting information).

To further estimate the rationality of using the solvating power of solvents as a solvent indicator, we discuss the DFT calculations, which are used to study the electronic surface potentia (ESP) distribution and binding energy of solvents (Figs. S2a and b in Supporting information), where the absolute value of the binding energy for TFEC-Li⁺ (-0.087 eV) is lower compared to EC-Li⁺ (-0.207 eV), DMC-Li⁺ (-0.225 eV), and EMC-Li⁺ (-0.205 eV). This indicates that the strength of the ion-dipole interaction between TFEC and Li⁺ is weaker than that of the other solvents studied. This implies that TFEC has a lower solvating power in terms of its ability to effectively solvate or interact with Li⁺ ions. In addition, DFT calculations have revealed that TFEC has the lowest highest occupied molecular orbital (HOMO) compared to the other three solvents (Fig. S2c in Supporting information). This suggests that the free fluorinated carbonate solvent possesses a greater capability to resist oxidation.

Molecular dynamics (MD) simulations were performed to obtain a more profound understanding of the solvation sheath surrounding Li⁺ ions (Figs. 1a and b). The radial distribution functions (RDFs) depicted in Figs. S3a and b (Supporting information), indicate that the intensity of Li-F(PF₆) in the EED electrolyte are weaker compared to those in the EEDT electrolyte, indicating the interaction between Li⁺ and F(PF₆) ions becomes stronger with the addition of TFEC. This is beneficial as it helps reduce the decomposition of electrolyte, leading to the formation of the PF₆^- anion-derived interfacial layer that is enriched in fluorine [22]. Furthermore, the Li-O(TFEC) peak in the RDF profile is observed, suggesting that TFEC is participation in the Li⁺ solvation structure. This result is consistent with the previous DFT calculations observations. Based on the results, the solvation structure of the EEDT electrolyte has been visually represented in Fig. 1d, with that of the conventional electrolyte (EED) shown in Fig. 1e for comparison.To further examine the influence of TFEC on the local environment of Li⁺ in the electrolyte, ¹⁷O nuclear magnetic resonance (NMR) spectroscopy was conducted. Fig. 1f demonstrates an upfield shift, which suggests an increase in electron density around oxygen (O) in solvent upon the addition of TFEC. This is further confirmed by the ⁷Li NMR (Fig. 1c), where with EEDT electrolyte demonstrates a greater downfield shift (i.e., more negative) compared to the EED electrolyte. This can be attributed to the weak solvation ability of TFEC, which reduces the interaction between Li⁺ and the solvent, allowing more anions to squeeze into the primary solvation shell [23-25]. This observation validates the previous DFT and MD simulations results.

To gain a deeper understanding of the solvation structures in EED and EEDT electrolytes, we utilized Raman and ATR-FTIR analysis (Figs. 1g-j). Specifically, the stretching vibration at approximately ~720 and ~894 cm^-1 could be attributed to free EC molecules, while the bands around ~730 and ~905 cm^-1 are associated with solvated EC species [26,27]. Besides, DMC exhibits two primary conformations in the electrolyte: cis-cis (cc) and cis-trans (ct) [28], the bands at ~862 and ~917 cm^-1 belonged to the pure DMC (ct) and DMC (cc) components, the bands at ~875 and ~934 cm^-1 belonged to the solvated DMC (ct) and DMC (cc) components. In addition, the bands at ~938 and ~934 cm^-1 belonged to the free EMC and solvated EMC, respectively [29,30]. It is important to note that the addition of TFEC leads to significant changes in the proportion of EC, DMC, and EMC molecules. Specifically, the presence of free EC, EMC, and DMC decreases or even disappears, giving way to the formation of solvated EC, EMC, and DMC species (Figs. 1g and h). Further analysis reveals the presence of two bands at approximately 740 and 745 cm^-1, suggesting an increased presence of CIPs within the EEDT electrolyte [30]. Moreover, this phenomenon can also be observed in the shift of peak positions in FTIR. Specifically, the bands ~1260 and 840 cm^-1 associated with the coordination of Li⁺ and solvents display a blue shift in the presence of TFEC (Figs. 1i and j, Fig. S3 in Supporting information) [31]. These spectral analyses indicate a weakening of the interaction between Li⁺ and solvents in the solvation structure, while highlighting an enhancement in the interaction with the anion.

Further insights into the solvation behavior of electrolytes in different solutions can be gained by examining the performance of the lithium metal anode. When employing the Aurbach method [32], it is observed that the Li||Cu cell cycled at current density of 0.5 mA/cm² with constant areal capacity of 1 mAh/cm² utilizing the EEDT electrolyte exhibits superior reversibility (Fig. 2a). Moreover, our research illustrates that under similar conditions, our performance of the Li||Cu cell surpasses those of other research teams (Fig. 2b) [32-35]. Meanwhile, the Tafel curves show that the exchange current density of the EEDT electrolyte is relatively higher at 0.158 mA/cm² compared to 0.126 mA/cm² (Fig. 2c). A higher value for exchange current densities signifies greater interfacial charge-transfer kinetics and a reduced overpotential in the electrochemical reaction [35,36]. Moreover, the utilization of the EEDT electrolyte in the symmetric Li||Li cell results in remarkably low overpotentials and demonstrates excellent cycling stability throughout the lithium plating/stripping process when the areal capacity is maintained at 0.5 mAh/cm² under a current density of 0.5 mA/cm², lasting for over 800 h (Fig. 2d). These comprehensive results provide substantial evidence of the significant potential of EEDT electrolyte studied during long-term cycling tests, which are further corroborated by the subsequent evaluation of the Li||LiMn_0.8Fe_0.2PO₄ full cell cycling performance (Fig. 2e).

The typical charge-discharge curves at different cycles for Li|| LiMn_0.8Fe_0.2PO₄ full cells in the EED and EEDT electrolyte are shown in Fig. S4 (Supporting information). Notably, at the 800^th cycle, the EED electrolyte exhibits significantly higher polarizations (Fig. S4a). This phenomenon is clearly depicted in Fig. 2e, which shows the cycle performance between 3.0 V and 4.5 V (vs. Li⁺/Li). After 800 cycles, the full cell utilizing EEDT electrolyte demonstrates a significantly improved specific capacity of 90.6 mAh/g in comparison to the EED electrolyte (44.7 mAh/g), exhibiting much greater stability throughout the cycling process. Even after 800 cycles, when two cells are connected in parallel, the general power LED indicators loaded in them can still emit high brightness (Fig. 2f). Furthermore, electrochemical impedance spectra (EIS) are also included to examine the impact of EEDT electrolyte on the resistance of Li|| LiMn_0.8Fe_0.2PO₄ full cells (Fig. S5 in Supporting information). Fig. S5a reveals an interesting finding that the newly assembled full cells exhibit higher interfacial resistance in the EEDT electrolyte compared to the EED electrolyte. This observation corresponds well with the larger polarization observed in the first charge-discharge curves of the EEDT electrolyte (Fig. S4). Additionally, the impedance in the EED electrolyte surpasses 2500 Ohms, higher than the < 1500 Ohms observed in the EEDT electrolyte after 800 cycles (Fig. S5c). The EIS results, combined with the aforementioned charge and discharge curves and cycling performance, demonstrate that the addition of TFEC effectively alleviates capacity degradation and enhances the movement of lithium ions and electrons.

To observe the effect of EEDT electrolyte on Li anodes during cycling process, the morphology of the cycled Li anodes recovered from practical Li||LiMn_0.8Fe_0.2PO₄ cells were characterized using scanning electron microscopy (SEM). The Li anode cycled in the EEDT electrolyte exhibited a dense and compact surface in Fig. S3d (Supporting information), with no visible surface cracks or evidence of Li dendrite growth observed. In stark contrast, when cycled in the EED electrolyte, the cycled Li anode exhibited a nonuniform and porous surface, prominently featuring the presence of Li dendrites (Fig. 3a). X-ray photoelectron spectroscopy (XPS) was utilized to analyze and identify the components of the SEI on cycled Li anodes, investigating the considerable contrast in surface morphology. The XPS analysis involved different etching times and was conducted on samples using EED and EEDT electrolytes (Figs. 3b and e). The outer layer of the SEI (sputtering time: 0 min) formed in both electrolytes comprised of both inorganic substances such as Li₂CO₃ and LiF, as well as organic compounds like ROCO₂Li which are derived from the reduction of the electrolyte [37]. The presence of relatively high intensity of products, such as ROCO₂Li and Li₂CO₃, was observed in the C 1s spectra (Fig. 3b) of EED electrolytes, indicating a significant occurrence of side reactions involving EC-based organic solvents [37]. Differences in electrolyte composition are clearly evident in the F 1s spectra obtained from the Li surface (Figs. 3c and f). The relative percentage of LiF is higher in EEDT electrolytes at 56%, compared to 18% in EED electrolytes (Figs. 3i and j). Furthermore, the atomic composition ratios at the lithium anode interface in the EEDT electrolyte indicate a higher fluorine content compared to the EED electrolyte (Figs. 3g and h), pointing towards the formation of a substantial inorganic SEI on the lithium anode surface. Notably, in the EEDT electrolyte (Figs. 3d and f), the additional peak at 293.0 eV (C 1s spectrum) and 688.6 eV (F 1s spectrum) corresponding to –CF₃ bonding are observed at various depths of the SEI [37,38]. The DFT calculations provide further evidence supporting this observation, showing that TFEC, possessing the lowest LUMO energy level compared to EC, EMC, and DMC (Fig. S2c in Supporting informaiton), selectively decomposes on the Li metal anode surface. Figs. 3g and h illustrate the atomic ratios of four selected elements on the cycled Li anodes, acquired at various depths. During the cycling of Li anodes in EED electrolytes, the inorganic component (represented by Li) within the SEI layer exhibits an initial increase and subsequently a decrease as the sputtering time is extended. Moreover, the uneven distribution of the four elements at the two depths within the SEI creates an inhomogeneous Li⁺ flux and leads to non-uniform lithium deposition. This lack of uniformity ultimately negatively affects the cycling stability of the Li anodes over time (Fig. 3g). On the other hand, in stark contrast to the previous case, the SEI formed in the EEDT electrolyte demonstrates a nearly constant atomic ratio of the four elements at two distinct depths (Fig. 3h). The uniform internal composition of the SEI derived in the EEDT electrolyte imparts unique properties to the SEI layer. In addition, the presence of LiF on the SEI surface enhances the efficient distribution of lithium ions parallel to the interface and promotes uniform deposition of lithium during cycling processes. As a result, side reactions are minimized, leading to improved cycling performance.

A comparison between the two cathode SEM images reveals that the cycled LiMn_0.8Fe_0.2PO₄ cathode in the EED electrolyte exhibits severe blockage of pores, whereas in the electrolyte EEDT, the degree of blockage is noticeably reduced (Figs. 4a and d). The findings are consistent with the CEI thicknesses observed through high-resolution transmission electron microscopy (HRTEM) in Figs. 4b and e. Notably, the EEDT electrolyte yields a thinner and more uniform CEI layer, suggesting that the presence of TFEC effectively mitigates the continuous decomposition of electrolyte solvents. Moreover, the composition of the CEI layer formed on the LiMn_0.8Fe_0.2PO₄ further confirms the distinctions arising from the variances in electrolyte properties (Figs. 4c and f-h, Fig. S2c in Supporting information). A severe side reaction of organic solvents is indicated by the relatively high intensity of products, such as ROCO₂Li and Li₂CO₃ [39], observed in the C 1s spectra of EED electrolytes. This indirectly indicates that TFEC can inhibit the decomposition of EC solvent [40,41]. Moreover, the Li_xPO_yF_z peak (685.8 eV for F 1s, 534.2 eV for O 1s and 134.0 eV for P 2p in Fig. 4f and Fig. S6 in Supporting information) [42,43] is significantly weaker for the LiMn_0.8Fe_0.2PO₄ cathode cycled in the EEDT electrolyte compared to that in the EED electrolyte, indicating that less Li_xPO_yF_z is generated in the EEDT electrolyte during cycling. In addition, we observed small peaks at 293.0 eV (C 1s) and 688.6 eV (F 1s), attributed to the -CF₃ group of TFEC, indicating the involvement of TFEC in building the CEI (Figs. 4c and f) [36]. Furthermore, the EEDT electrolyte effectively suppresses the dissolution of Mn and Fe from the LiMn_0.8Fe_0.2PO₄ cathode, as observed in the Fe 2p and Mn 2p spectra (Figs. 4g and h) [44]. The restrained degradation of LiPF₆, which results in the generation of scarce HF, is the primary factor responsible for the leaching of metals from LiMn_0.8Fe_0.2PO₄ cathode. Therefore, in order to provide evidence for the dissolution of the cathodes, we conducted a simulation experiment. We aged the charged LiMn_0.8Fe_0.2PO₄ cathode at room temperature for 20 days in EED and EEDT electrolytes, and compared the dissolution content of Mn and Fe (Fig. 4i). It is evident that the EED electrolyte exhibits significantly higher levels of Mn and Fe dissolution compared to the containing TFEC electrolyte (EEDT), indicating that the presence of TFEC in the electrolyte effectively suppresses the dissolution of Mn and Fe from the LiMn_0.8Fe_0.2PO₄ cathode. The XPS spectra and inductive coupled plasma optical emission spectrometer (ICP) findings (Fig. 4i) are consistent with the observations obtainedfrom SEM, HRTEM, XRD and FTIR (Figs. S7 and S8 in Supporting information).These results indicate that TFEC can suppress the decomposition of solvent and lithium salt in the electrolyte during the charge-discharge cycle of LiMn_0.8Fe_0.2PO₄ cells, leading to a thin CEI formation on the surface of LiMn_0.8Fe_0.2PO₄ cathode and suppress the dissolution of transition metal, thereby enhancing the performance of lithium batteries.

This study demonstrates that TFEC enhances the stability of EC at electrode interface by leveraging the strong electron-withdrawing effect (-CF₃), which reduces the electron cloud density of the carbonyl oxygen and subsequently lowers the solvating power of solvents, leading to the creation of electrode-electrolyte interfaces rich in inorganic compounds (SEI and CEI), thereby enhancing the stability of both the lithium anode and high-voltage cathode. As expected, the Li||LiMn_0.8Fe_0.2PO₄ full cell demonstrates exceptional capacity retention at 62.5% after 800 cycles, significantly surpassing that of the base electrolytes (29.8%). Our work offers a new perspective on the application of EC-based electrolyte systems in high-voltage LMBs.

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.

Guihuang Fang: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Ying Liu: Resources, Project administration, Methodology, Data curation. Yangyang Feng: Funding acquisition, Formal analysis, Data curation. Ying Pan: Formal analysis, Data curation. Hongwei Yang: Software, Resources, Project administration, Methodology. Yongchuan Liu: Project administration, Methodology. Maoxiang Wu: Resources, Investigation, Funding acquisition, Formal analysis.

This work is supported by National Natural Science Foundation of China (Nos. 12275274 and 22209182), the Fujian Provincial STS program supporting project of Chinese Academy of sciences (No. 2022T3001).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110385.