Reconciling electrolyte donicity and polarity for lithium carbon fluoride batteries

Abstract

Among the existing electrochemical energy storage technologies, lithium carbon fluoride (Li°||CF_x) batteries have captured substantial attention owing to their surprisingly high energy density and low self-discharge rate. The features of nonaqueous electrolytes play an essential role in determining the electrochemical reactions of the CF_x cathode, subsequently affecting the electrochemical performances of Li°||CF_x batteries. Herein, differing from previous stereotypical perceptions, the fascinatingly entangled parameters of nonaqueous electrolytes, comprising permittivity, donicity, and polarity, are comprehensively investigated and reconciled by adopting the solution mixtures of 1,2-dimethoxyethane (DME) and propylene carbonate (PC). The results demonstrate that the higher donicity and moderate polarity of nonaqueous electrolytes (e.g., DME-rich electrolytes) favor the heterolytic dissociation of carbon–fluorine bonds, resulting in more complete electrochemical conversions of the CF_x cathode. This work provides new insights into the electrochemical reaction paths of the CF_x cathode and encourages the electrolyte design for high-energy batteries with other conversion-type electrode materials (e.g., sulfur and oxygen).

---

Broader context

For certain application scenarios (e.g., aerospace), the rechargeability of batteries becomes less critical, while their energy density and other performance matrices stand out. In this domain, lithium metal carbon fluoride (Li°||CF_x) batteries have attracted considerable attention owing to their superior theoretical energy density (i.e., 2180 W h kg⁻¹), compared with other primary battery technologies. However, the electrochemical performances of the CF_x electrode are still hampered by their inherent properties (e.g., lower electronic conductivity) and sophisticated redox reaction processes. It has been widely acknowledged that the characteristics of nonaqueous electrolytes have a great impact on the discharge process and conversion mechanism of the CF_x cathode. In the present work, from the perspective of electrolyte solvent chemistry, the involved electrolytes parameters (i.e., permittivity, donicity, and polarity) are comprehensively investigated and reconciled. Importantly, the concept of solvent polarity has been introduced into the design of nonaqueous electrolytes for the first time for the conversion-type CF_x cathode, which provides an extraordinary thinking tool for improving the performance of high-energy batteries (e.g., lithium–sulfur, lithium–oxygen, sodium and other related primary and rechargeable batteries).

Introduction

Continuous consumption of non-renewable energy resources (e.g., coal, petroleum, natural gas) has stimulated a quick integration of battery-related energy storage systems in today's energy network since batteries are able to release (or store) electric energy via chemical reactions with high energy efficiency.^1–3 During the past decades, lithium-ion batteries (LIBs) have become a prevailing battery technology for electric vehicles and grid storage owing to their decent energy density (>400 W h kg⁻¹) and cyclability (>1500 cycles).^4,5 However, for some specific application scenarios (e.g., aerospace and medical device) that require an extremely high energy density (>500 W h kg⁻¹) surpassing the theoretical limits of state-of-the-art LIBs, complementary battery technologies are desired. Among these, lithium metal (Li°) carbon fluoride (Li°||CF_x) batteries have attracted considerable attention owing to their moderate operational potential (about 2.8 V vs. Li/Li⁺), high specific capacity of lithium metal (i.e., 3860 mA h per gram metallic lithium)^6,7 and carbon fluoride electrodes (i.e., 865 mA h per gram of CF),^8–10 and low self-discharge rate, among other characteristics.^9,11–13 The gravimetric energy density of Li°||CF_x batteries can reach high values of 2180 W h kg⁻¹ at the material level (i.e., considering only the active materials involved in the redox reaction) and 600 W h kg⁻¹ at the cell level (i.e., partially considering the inactive materials for cell assembly), remarkably outcompeting other kinds of battery technologies [e.g., 2180 (Li°||CF_x) vs. 1470 (Li°||SOCl₂) vs. 1005 (Li°||MnO₂) W h kg⁻¹ at the material level].^11,14,15

The utilization of carbon fluoride materials with a higher degree of fluorination (e.g., CF_x, x > 0.85) is beneficial for attaining high energy density. However, their lower electronic conductivities [approximately 10⁻⁷ (CF_0.80) vs. 10⁻¹² (CF_1.0) S cm⁻¹ at room temperature]^16,17 at initial discharging stages cause sluggish reaction kinetics and severe internal cell polarization, thus leading to insufficient power capability of high-energy Li°||CF_x batteries.^10,18–22 Differing from intercalation-type cathode materials, which are widely used in LIBs, carbon fluoride materials undergo conversion-type reactions during discharge processes, generating graphite intercalation compound (GIC) intermediates, amorphous/crystalline carbon, and lithium fluoride (LiF) at the interface between the electrolyte and cathode compartment.^23–27

It has been demonstrated that the formation and accumulation of LiF particles on the surface of the CF_x electrode has an essential impact on the electrochemical performances (e.g., attainable energy density) of Li°||CF_x batteries, and such processes are highly associated with the characteristics of the nonaqueous electrolyte solution employed therein.^28–34 For example, the operational voltages of the Li°||CF_x batteries could be improved by decreasing the concentrations of lithium salts (i.e., intensifying the interactions between Li⁺ cations and electrolyte solvent).^33,34 The addition of 1,3-dimethyl-2-imidazolidinone (DMI) and N,N-dimethylpropionamide (DMP) with high electron-donicity [characterized by Gutmann donor number; e.g., DN = 29 (DMI),³⁵ DN = 27 (DMP)³⁴ kcal mol⁻¹] facilitates the strong solvation of Li⁺ ions and weakens the carbon–fluorine (C–F) bond, thus elevating the discharge voltage plateau (ca. 2.8 V for DMI-added electrolytes) and energy density.³³

In an early study by Armand,³⁶ the redox reaction of CF_x materials was suggested to follow a bimolecular nucleophilic substitution (S_N2) reaction, in which the electrons act as nucleophiles and the breakup of the C–F bonds and the formation of LiF particles occur simultaneously. As extensively discussed in classic organic chemistry (e.g., nucleophilic substitution reactions of alkyl halides), the nature of the reaction media has a significant impact on the kinetic factors (e.g., activation energy, reaction rate) of the bimolecular reactions.³⁷ In the same vein, it is reasonable to anticipate that the discharge processes of the CF_x electrode are likely to be tightly correlated with the features of the electrolyte components. In a recent work, we reported the involvement of salt anions in the electrochemical breakdown of C–F bonds in CF_x materials during discharge processes, in which salt anions with a moderate donicity could reduce the energy barrier of the S_N2-type conversion reaction and thereby enhance the discharge capacities. Specifically, the salt anions with moderate donicity could favorably interact with slightly positively-charged carbon atoms, thus promoting the heterolytic dissociation of C–F bonds, and minimize the de-solvation energy for Li⁺-ions while receiving negatively-charged fluoride atoms (i.e., push–pull concept).³⁸ This work clearly reinforces the hypothesis that a S_N2 reaction could be adopted to describe the discharge processes of the CF_x electrode.

Considering the importance of the inherent features of nonaqueous electrolytes in determining the kinetics of S_N2 reaction, it is thus urgent to understand the impact of key electrolyte parameters (i.e., donicity and polarity) on the discharge behavior of the CF_x electrode. Herein, a series of 1,2-dimethoxyethane (DME)/propylene carbonate (PC)-mixed nonaqueous electrolytes with different donicities and polarities are designed. Their performances in the Li°||CF_x cells are comprehensively evaluated, aiming to elucidate the unique impact of the electrolyte donicity and polarity on the discharge processes of conversion-type electrode materials. The electrolyte configurations are conceived on the basis of the considerations below (Fig. 1): (1) the selected solvent PC possesses an extremely high dipole moment (ca. 4.94 Debye³⁹), allowing a facile regulation of the electrolyte polarity with wide transition energy values [i.e., E_T(30) = 38.2 (DME), 46.0 (PC) kcal mol⁻¹]⁴⁰ by varying its fractions in the resulting electrolytes, (2) the co-solvent DME is usually used as a benchmark solvent for Li°||CF_x batteries and other kinds of lithium batteries (e.g., Li°||S), in light of its low viscosity (i.e., η = 0.47 cP)⁴¹ and good chemical stability against metallic lithium,^42,43 and (3) the conducting salt lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) has an anion of low donicity (i.e., Gutmann DN = 5.4 kcal mol⁻¹),⁴⁴ which could avoid the potential influence of anion donicity and magnify the critical role of the electrolyte polarity on the discharge processes of the CF_x electrode.

Fig. 1 Design concept for elucidating the critical role of electrolyte polarity and donicity on the performances of lithium carbon fluoride batteries. The balls in black, dark green, and pink represent the carbon, fluorine, and lithium atoms, respectively. The ellipsoids in grey and purple colors indicate the solvent molecules.

Results and discussion

Physical properties

For chemical and electrochemical reactions (e.g., conversion reaction of the CF_x electrode), the reaction kinetics and equilibrium positions are tightly associated with the employed solvent, as casually expressed by Albery—“In searching to understand the rate of a reaction in solution, the baby must not be separated from its bath water”.⁴⁵ In the same vein, there exists a wide array of inter- and intramolecular interactions among solvent and solute molecules, which governs the transport properties (i.e., total ionic conductivity, cation transference number) and chemical reactivities (e.g., electrophilicity, nucleophilicity) of the nonaqueous electrolyte solutions.^37,40 In this regard, an explicit understanding of the features of solvents and their interactions with electrolyte salts are of particular importance to design suitable electrolytes for advanced batteries.^{33,34,38,46–48} Currently, several kinds of solvent parameters (e.g., permittivity, polarity, and donicity) have been introduced to evaluate the properties of electrolyte solvents from different perspectives. Herein, a comprehensive investigation on the key solvent parameters has been carried out in this work for the DME/PC mixed electrolytes, and the results are summarized in Fig. 2 and Table S1 (ESI†). The acronyms of D₁₀₀P₀, D₇₀P₃₀, D₅₀P₅₀, D₃₀P₇₀, and D₀P₁₀₀ represent the electrolytes of 1.0 M LiTFSI-DME, 1.0 M LiTFSI-DME/PC (70 : 30, by wt) 1.0 M LiTFSI-DME/PC (50 : 50, by wt), 1.0 M LiTFSI-DME/PC (30 : 70, by wt), and 1.0 M LiTFSI-PC, respectively.

Fig. 2 (a) Schematic diagram of the dipole moment of the solvent PC and DMC. (b) The transition energies in the dye involve the variation in the chemical structure from a more dipolar zwitterionic ground state to a less dipolar transition state. (c) Definition of the donor number for solvents, as suggested by Gutmann.^49,50 The geometry of SbCl₅ is generated and optimized using the open-source molecular visualizer Avogadro and VESTA software.^51,52 (d) UV-vis absorption spectra of the different PC-based solvent systems in the range of 400–900 nm. (e) ²³Na-NMR spectra of the different PC-based systems with NaTFSI. (f) Dependence of the solvent parameters [i.e., E_T(30), dielectric constant, and donor number] on the fraction of PC at 25 °C.

As early as the 1950s, unusually high dielectric constant (ε) values, being comparable to that of water [e.g., ε = 64.9 (PC)⁵³vs. 78.6 (H₂O)⁵⁴ at 25 °C], have been identified for cyclic carbonates, which is anticipated to be critical for their superior dissolving ability towards a wide gallery of metal salts (e.g., LiBr, NaI, ZnCl₂, FeCl₃, HgCl₂, etc.).⁵⁵ The nonaqueous solutions of metal salts in cyclic carbonates are mostly ionically conductive, indicating the sufficient ionization of metal salts in the presence of cyclic carbonate solvents.⁵⁵Fig. 2(a) shows the optimized geometries of the representative cyclic carbonate PC and linear carbonate dimethyl carbonate (DMC). The corresponding density functional theory (DFT) calculations and the Cartesian coordinates are provided in ESI† (Table S2). Following the valence shell electron-pair repulsion (VSEPR) model,⁵⁶ the center carbon in the carbonyl group with an sp² hybridization holds a trigonal planar structure. Ideally, the O–C(=O)–O bond angles should be 120°. However, a slight compression is observed for both PC and DMC molecules (e.g., 108° for PC and 110° for DMC), owing to the steric repulsion of methylene (–CH₂–) and methyl (–CH₃) groups.⁵⁷ Nevertheless, the carbonate center in both PC and DMC possesses comparable geometries, which signifies its negligible impact on the overall dielectric properties of the molecules. For the linear carbonate DMC, the dipole force of the sp³ carbon–oxygen (C–O) ether bond adjacent to the carbonate center is in an opposite direction as compared to that of the carbonyl (C=O) group (Fig. 2(a)), which effectively cancels out the overall dipole force brought by the strong dipole of the carbonyl group. For cyclic carbonate PC, the sp³ hybridization of the C–O ether bond places it in a cis conformation, which could positively contribute to the overall dipole moment of the carbonate molecule (Fig. 2(a)). Thus, the net permanent dipole moment of PC is remarkably stronger than that of DMC, and consequently results in a stronger dielectric permittivity of the former one.⁵³

It is important to note that the solvent parameter of the dielectric constant and dipole moment described above are independent of the identity of the electrolyte salt, considering only the interactions among the solvent molecules. Hence, the concept of the solvent polarity and donor number was suggested in the 1960s for the inclusion of solvent–solute interactions. Following the definition suggested by Reichardt,⁵⁸ the polarity of a solvent is determined by its solvation capability (or solvation power) for reactants and activated complexes, as well as for molecules in their ground and excited states. To quantitatively characterize the polarity of various kinds of solvents, several methodologies have been developed in the past decades.^59,60 These empirical solvent polarity scales are generally obtained through kinetic measurements (e.g., “Y” and “Ω” scales, see Table S3, ESI†) and spectroscopic methods [e.g., “Z” and “E_T(30)” scales, ultraviolet-visible (UV-vis) spectroscopy, see Table S3, ESI†]. Among them, the E_T(30) scale, suggested by Reichardt et al.,^59,60 has been extensively employed in chemistry in light of its practical feasibility in experimental procedures, together with good suitability for a wide array of solvents and solvent mixtures.^40,60 Founded on the interactions between the solvent molecule and a reference solute (i.e., Reichardt's dye), the E_T(30) value is defined as the molar electronic transition energy (historically in kcal mol⁻¹) of the Reichardt dye 30, using eqn (1):^40,60–62

E_T(30) = 28 591/λ_max (1)

where λ_max (nm) is the maximum wavelength of the absorption band in the UV-vis region, corresponding to the structural change from the dipolar zwitterionic ground state to the less dipolar transition state, as shown in Fig. 2(b).

Considering that interactions between the solvent and solutes mainly stem from the electron delocalization between an electron-pair donor (EPD) and an electron-pair acceptor (EPA), the concept of electron donicity was proposed by Gutmann to describe the ability of a given molecule to share their electrons with a reference Lewis acid compound, antimony pentachloride (SbCl₅).^49,50 As presented in Fig. 2(c), a donor solvent (D) could generally interact with the acceptor molecules (e.g., SbCl₅), releasing a certain enthalpy (ΔH, kcal mol⁻¹) that could be detected by calorimetric methods. In this sense, a quantitative value of the Gutmann donor number (DN, in kcal mol⁻¹)⁴⁹ is given by the following eqn (2):

DN = −ΔH (2)

Effectively, most metal ions are electron pair acceptors, and their interactions with donor solvent molecules are critical for the solvation of metal cations in nonaqueous electrolyte solutions. On the basis of the extensive DN values determined by calorimetric methods, one may employ sodium (²³Na) nuclear magnetic resonance (NMR) spectroscopy to indirectly estimate the DN values of a new solvent or solvent mixture without handling the toxic and corrosive reference reagent SbCl₅.^63,64

Based on the understanding of the aforesaid solvent properties, the dielectric constant, E_T(30), and Gutmann DN values of the neat PC and its mixtures with DME are determined in this work, and the corresponding numeric values are collected in Table S1 (ESI†). As seen in Table S1 (ESI†), the values of the dielectric constant increase monotonically by increasing the fraction of PC in the solvent mixture of DME/PC, e.g., 7.5 (D₁₀₀P₀) vs. 31.4 (D₅₀P₅₀) vs. 64.0 (D₀P₁₀₀), indicating that the addition of cyclic carbonate could remarkably improve the overall permittivity of the resulting solvent mixtures.^65,66

Fig. 2(d) shows some representative UV-vis spectra measured with the DME/PC mixed solvent mixtures in the presence of Reichardt dye 30, and their corresponding appearances are exhibited in Fig. S1 (ESI†). For the PC-free solvent (i.e., labelled as D₁₀₀P₀), a characteristic absorption peak (λ_max) appears at the wavelength of 754 nm, corresponding to an E_T(30) value of 37.9 kcal mol⁻¹ using eqn (1). This value is comparable to that reported in the literature [e.g., 37.9 (this work) vs. 38.2 (ref. 40) kcal mol⁻¹], indicating the reliability of our experimental procedure, particularly the control of residual water in the solvent.^67,68 With the addition of PC, the absorption band shifts towards a lower wavelength direction (i.e., blue shift), indicating an increase of the corresponding E_T(30) value [i.e., 37.9 (D₁₀₀P₀) to 42.7 (D₇₀P₃₀) kcal mol⁻¹; Fig. 2(d) and Fig. S2, ESI†]. Generally, a stronger stabilizing effect between the solvent and dipolar zwitterionic ground state (Fig. 2(b)) leads to a higher value of the transition energy for the excitation process, and thereby a higher E_T(30) value (i.e., higher solvent polarity).⁴⁰ In addition, as shown in Table S4 (ESI†), a negligible difference of the molar absorption coefficient is observed for the resulting electrolyte solvents, which fulfils the experimental conditions required by the Beer–Lambert law.^69,70

Fig. 2(e) presents the ²³Na NMR spectra measured with the diluted electrolyte of 0.01 M NaTFSI in DME/PC mixtures using a co-axial NMR tube, in which the capillary tube is filled with an aqueous solution of 0.1 M NaCl in deuterated water for calibrating the chemical shift (δ, in ppm) of ²³Na NMR spectra. For the neat DME sample (i.e., D₁₀₀P₀), two isolated peaks are observed in the ²³Na NMR spectra corresponding to the signals of the solvated sodium cations in deuterated water (δ = 0 ppm) and DME (δ = −5.77 ppm), respectively. By increasing the weight fraction of PC in the DME/PC mixtures, the chemical shifts assigned to the solvated sodium cations in nonaqueous environments move to the higher field domains [e.g., δ = −5.77 (D₁₀₀P₀) vs. −6.60 (D₇₀P₃₀) ppm], indicating stronger shielding effects of the sodium cations with the increase of the PC fractions (Fig. 2(e) and Fig. S2, ESI†).

Following the seminal studies by Erlich and Popov,⁷¹ one may correlate the Guttman DN values of the solvent with the chemical shift of the solvated sodium cations in non-aqueous solutions, and the DN values of the DME/PC mixtures could be readily obtained (cf. Experimental section of the ESI†). Clearly, the addition of PC into DME decreases the overall electron-donicity of the solvent mixture [e.g., 20.6 (D₁₀₀P₀) to 18.8 (D₇₀P₃₀) kcal mol⁻¹], and therefore enhances the delocalization of the electron cloud from solvent molecules (i.e., electron donors) to sodium cation (i.e., Lewis's acid)—a higher chemical shift in the presence of higher fractions of PC.

Fig. 2(f) compares the key solvent parameters measured for the DME/PC mixed solvent. As a general trend, both dielectric constant and E_T(30) values of the solvent mixtures increase monotonically with increasing PC weight fraction. Meanwhile, the DN values decreases at higher fractions of PC. For the dielectric constant values, a linear fitting could be achieved with a decent coefficient of determination (R² = 0.9805; Fig. S3, ESI†) for the five kinds of samples containing different contents of PC, as observed similarly in the literature.^65,66 Therefore, using these results in this work, one may calculate the dielectric constant for the DME/PC mixed solvent with a random ratio (Fig. S3, ESI†). This is because the change of the dielectric constant mainly involves the van der Waals interactions of the solvent molecules, and this could be a linear superposition by adding more PC. Differing from the pattern observed with dielectric properties, the dependence of the E_T(30) and DN values on the weight fraction of PC does not follow a linear behavior [R² = 0.8761 for E_T(30) values, R² = 0.9280 for DN values; see Fig. S3, ESI†]. This is because not only the effects of the solvent itself, but also the effects of the solvent and solute should be considered when understanding the polarity and donicity of solvents.

Furthermore, only a small quantity of PC could bring about a remarkable change in the E_T(30) polarity scale, e.g., 37.9 (D₁₀₀P₀) vs. 42.7 (D₇₀P₃₀) kcal mol⁻¹; while a further increase of the PC content leads to smaller changes in the E_T(30) values [e.g., 44.0 (D₅₀P₅₀), 44.7 (D₃₀P₇₀) to 46 (D₀P₁₀₀) kcal mol⁻¹]. Generally, the enhancement of a small quantity of a polar solvent into a less polar solvent could result in a disproportionately large hypochromic band shift in the betaine dye 30, which coincides with an excessively large increase in E_T(30). This can easily be explained by the strong preferential or selective solvation of the dipolar betaine dye by the more polar component of the binary solvent mixtures.⁴⁰ With respect to DN values, apart from a steady decrease of the DN values with increasing weight fraction of PC, a sharper decrease is observed in the PC-rich domain (>50 wt% PC), e.g., 17.3 (D₃₀P₇₀) to 13.4 (D₀P₁₀₀) kcal mol⁻¹. This phenomenon could mainly be ascribed to the sudden transformation of the solvated structure and the coordination ability of the solvent by the low donicity of the solvent mixtures.

In spite of the expected trend in the dielectric properties, the incorporation of a small amount of PC (e.g., D₇₀P₃₀) leads to significant changes in the polarity of the electrolyte solvents, while the utilization of PC-rich solvents (e.g., D₀P₁₀₀) sharply reduce the coordination ability of the solvents. These two key parameters could greatly impact the electrochemical conversion reaction of the CF_x electrodes, since the solvated Li⁺ ions essentially participate in the heterolytic dissociation of C–F bonds during electrochemical processes. High donor number and favorable polarity are required to give useful solubilities. A high dielectric constant is not necessary, but it is useful to keep the dissolved ions apart. Therefore, reconciling the electrolyte polarity and donicity is particularly important to deciphering the exact reaction mechanisms of the CF_x conversion electrodes in solution media, as detailed below.

Electrochemical reaction kinetic and discharge behavior of the CF_x cathode

To underline the impact of the electrolyte donicity and polarity on the performance matrices of the CF_x cathode, the discharge performances (e.g., rate capability and discharge behavior) of the prototype Li°||CF_x cells utilizing various fractions of PC are systematically evaluated. Fig. 3(a) depicts the dependence of discharged capacities vs. current-rates for the Li°||CF_x cells with different electrolytes, and the corresponding discharge profiles are given in Fig. S4 and S5 (ESI†).

Fig. 3 (a) Dependence of the discharged capacities vs. current-rates for the Li°||CF_x cells with different electrolytes. (b) Discharge profiles of the Li°||CF_x cells with five kinds of electrolytes at a relatively low discharge rate of 0.1C. (c) GITT curves measured with the Li°||CF_x cells utilizing various kinds of electrolytes at 0.1C. (d) Apparent D_Li⁺ values at different DoD values of the Li°||CF_x cells with different electrolytes. (e) Dependence of discharge capacity on the dielectric constant of the solvent mixture. (f) Dependence of the E_T(30) and donor number values in the identity of solvents on specific capacity.

As seen in Fig. 3(a), the DME-rich samples show significantly higher values of discharge capacity as compared with the PC-rich ones under the same current-rates [e.g., 649.2 (D₁₀₀P₀) vs. 678.2 (D₇₀P₃₀) vs. 202.3 (D₀P₁₀₀) mA h g⁻¹ at 0.1C], and such behavior becomes more pronounced under the higher discharge currents [e.g., 525.4 (D₁₀₀P₀) vs. 434.1 (D₇₀P₃₀) vs. 108.7 (D₀P₁₀₀) mA h g⁻¹ at 1C]. This clearly suggests that a higher fraction of DME supports a better discharge performance of the CF_x cathode. However, the addition of a small portion of PC in DME endows a slight increase in the discharge capacity at low current-rates [e.g., 0.05C and 0.1C, Table S5, (ESI†), e.g., 648.1 (D₁₀₀P₀) vs. 684.3 (D₇₀P₃₀) vs. 234.7 (D₀P₁₀₀) mA h g⁻¹ at 0.05C]. This implies that an optimal amount of PC, instead of a full DME solution, is preferred for a complete conversion of CF_x into LiF and carbon.

Fig. 3(b) further compares the discharge profiles of the Li°||CF_x cells with five kinds of electrolytes at a relatively low discharge rate of 0.1C. The corresponding discharge curves at higher C-rates are provided in Fig. S4 and S5 (ESI†). Here, a low current rate is selected for comparing the discharge behavior in different electrolytes to avoid the severe electrochemical polarization brought by the inherently low electron-conductivity and relatively sluggish diffusivity of CF_x materials.^{17,31,33,34,38,72} As seen in Fig. 3(b), the DME-rich samples (i.e., D₁₀₀P₀ and D₇₀P₃₀) show much higher discharge plateaus [e.g., 2.92 (D₇₀P₃₀) vs. 2.83 (D₃₀P₇₀) V at a discharge capacity of 300 mA h g⁻¹] and significantly lower internal polarizations than the PC-rich ones [i.e., D₃₀P₇₀ and D₀P₁₀₀], and the employment of neat PC-based electrolytes leads to a quick drop in the cell voltage. Clearly, the different capacity values (cf.Fig. 3(a)) are primarily due to the internal electrochemical polarizations among the assembled Li°||CF_x cells. Furthermore, at a relatively low depth of discharge (DoD < 50%), the discharge profiles of the Li°||CF_x cells are nearly superimposed for the DME-rich electrolytes (i.e., D₁₀₀P₀ and D₇₀P₃₀). However, the electrolyte with a small quantity of PC promotes the completion of the conversion reaction of the CF_x materials at the end of the discharge processes.

It has been reported that the ionic conductivity of electrolytes (Fig. S6, ESI†) is not the leading factor for the conversion-type CF_x cathode.^30,38 To uncover the electrode reaction kinetics of the Li°||CF_x cells, the galvanostatic intermittent titration technique (GITT)⁷³ was applied to determine the apparent D_Li+ in the CF_x cathode. The experimental details and calculation of D_Li+ values are provided in the Experimental Section.

Fig. 3(c) shows the GITT curves measured with the Li°||CF_x cells utilizing various kinds of electrolytes. For all of the Li°||CF_x cells, the voltages abruptly increase upon the removal of the external current, and then gradually approach a constant value (cf. zoomed-in profiles in Fig. S7, ESI†), which signifies that the relaxation time (i.e., 150 min) reserved for the GITT tests is sufficient to distinguish the transport kinetics among these electrolytes.^32–34,74 Generally, the DME-rich electrolytes (i.e., D₁₀₀P₀ and D₇₀P₃₀) present very similar GITT curves at the initial stage (<10 h of discharge time), which suggests that the ability for recovering to the equilibrium state is comparable for these three kinds of electrolytes without the disturbance of an applied electric field. However, for the PC-rich electrolytes (i.e., D₃₀P₇₀ and D₀P₁₀₀), the recovery of the equilibrium state becomes obviously hindered in the GITT discharge curves, which implies sluggish kinetics for the diffusion of the ionic species in the PC-rich electrolytes (i.e., D₃₀P₇₀ and D₀P₁₀₀) during the discharge processes of the CF_x cathode.

Following the non-steady state diffusion processes described by Fick's second law, the apparent D_Li⁺ values at different DoD are calculated and gathered in Fig. 3(d). For the DME-rich electrolytes (i.e., D₁₀₀P₀ and D₇₀P₃₀), the corresponding Li°||CF_x cells show comparable D_Li⁺ values throughout the discharge processes (e.g., 3.2 × 10⁻¹¹ cm² s⁻¹, 30% DoD). However, the Li⁺ diffusion process is obviously impeded for the PC-rich electrolytes [e.g., 7.9 × 10⁻¹² (D₃₀P₇₀) vs. 1.6 × 10⁻¹¹ (D₀P₁₀₀) cm² s⁻¹, 30% DoD]. These results are consistent with the discharge behaviors of the Li°||CF_x cells. Clearly, the variations of the D_Li⁺ values depend on the identity of the electrolytes and DoD of the cells. For all five samples, the apparent D_Li⁺ values obtained at the initial state (DoD < 10%) are relatively high, and they decrease with the continuation of the discharge processes. This is related to the trade-off between the Li⁺ diffusion in the bulk of the electrolyte and that in the CF_x cathode, where a fast diffusion of Li⁺ ions in the bulk electrolyte contributes to higher apparent D_Li⁺ values at the initial stage, and the sluggish Li⁺ diffusion in the CF_x cathode outweighs the overall process with the consumption of physically accessible active materials at the very surface of the CF_x particles.^26,75 By continuing the discharge processes, the apparent D_Li⁺ values increase monotonically due to the formation of carbon materials with superior electronic conductivities, thus reducing the internal polarization of the Li°||CF_x cells. Additionally, the D_Li⁺ values spanning from 10⁻¹² to 10⁻⁸ cm² s⁻¹ are obtained for all of the samples, which are close to the typical values reported in the literature for the CF_x-type electrode materials.^{33,34,76–78} It is noteworthy that the D_Li⁺ values in the bulk electrolytes are usually above 10⁻⁶ cm² s⁻¹,⁷⁹ being two to six orders of magnitudes higher than that for the electrode materials. This illustrates that the diffusion process in the cathode materials is the determining factor governing the overall transport of ionic species in the conversion reactions of the CF_x cathode.

The above results clearly prove that the solvent properties of the employed electrolytes are responsible for the distinctive discharge behaviors obtained with the DME/PC mixed electrolytes. Fig. 3(e) compares the dependence of the discharge capacity on the dielectric constant of the solvent mixture. When increasing the fraction of PC, the dielectric constant of the resulting solvent mixture increases monotonically [e.g., 7.5 (D₁₀₀P₀), 18.4 (D₇₀P₃₀) to 64 (D₀P₁₀₀)], while the corresponding discharge capacities show a maximal value at a relatively low PC fraction [e.g., 678.2 (D₇₀P₃₀) vs. 649.2 (D₁₀₀P₀) vs. 202.3 (D₀P₁₀₀) mA h g⁻¹]. This suggests that unveiling the discharge behavior of the Li°||CF_x cells from the stand point of the dielectric property tends to be insufficient, and further elaboration from other perspectives is effectively desired.

Fig. 3(f) compares the dependence of the discharge capacities on the polarity [i.e., E_T(30)] and donicity (i.e., Gutmann DN) of the solvent mixtures. From the perspective of the electrolyte donicity, the DME-rich electrolytes with higher DN values are beneficial for achieving higher discharge capacities with the Li°||CF_x cells. Meanwhile, the PC-rich ones with relatively lower DN values generally impede the discharge processes. This suggests that the electrolyte donicity plays a key role in the conversion reactions of the CF_x cathode, and electrolyte components with relatively high donicity are thus preferred. Effectively, the employment of solvents with higher DN values (e.g., DMI, DMP)^33,34 have been reported to be efficient in improving the discharge performance of the Li°||CF_x cells. However, a straightforward correlation between the electrolyte donicity and discharge capacities seems to be difficult. This is due to the enhanced discharge performance at an optimal fraction of PC in the mixture of DME/PC. Considering that the conversion reaction of the CF_x cathode could be treated as a typical S_N2 reaction and the polarity of the solvent mixture would inevitably influence the discharge processes, the polarity measured with the E_T(30) scale is therefore incorporated to reconcile the above divergence. As seen in Fig. 3(f), with increasing fraction of PC, the E_T(30) values show an opposite trend as compared to the DN values, in which the presence of PC in the DME/PC mixture favors the improvement of the electrolyte polarity. Clearly, the electrolyte with a small fraction of PC shows significantly higher polarity [i.e., E_T(30) = 42.7 kcal mol⁻¹] without much decrease in the electrolyte donicity (DN = 18.8 kcal mol⁻¹). This results in more complete discharge processes (i.e., higher discharge capacity) of the CF_x cathode under the same conditions. Thus, it is reasonable to anticipate that a trade-off between donicity and polarity of the solvent in the electrolyte is responsible for the distinctive discharge performances of the Li°||CF_x cells.

Morphological and compositional analyses

To manifest the synergistic influence of the electrolyte polarity and donicity on the conversion reactions of the CF_x cathode materials, the morphologies and surface compositions of the fresh electrode and recovered CF_x electrodes have been characterized using scanning electron microscope (SEM), energy-dispersive X-ray (EDX), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS).

Fig. 4 shows the SEM images of the CF_x cathode recovered from the DME/PC mixed electrolytes, together with the corresponding EDX mapping images of selected elements. For the as-prepared fresh CF_x cathode, a relatively smooth surface without agglomeration of large particles is observed, indicating a homogeneous distribution of CF_x particles and polymeric binder in the resulting cathode laminate (Fig. S8, ESI†). For the CF_x cathodes recovered after discharge, the outer surface becomes rougher and fluffier, accompanied by the presence of fine crystals for all samples. This is mainly due to the electrochemical conversion of pristine CF_x particles [i.e., CF_x (s) → C (s) + LiF (s)], which results in an expansion of the electrode volume and the sedimentation of new solid phases (i.e., LiF). As shown in Fig. 4, the EDX mapping images show the dispersion of carbon and fluoride elements on the electrode surface after the discharge processes. Meanwhile, the co-existence of oxygen, sulfur, and nitrogen elements is also detected (Tables S6 and Fig. S9, ESI†), which is attributed to the residual electrolyte salt (i.e., LiTFSI).

Fig. 4 SEM and EDX images of the discharged CF_x cathodes recovered from the lithium carbon fluoride batteries with different electrolytes as follows: (a) D₁₀₀P₀, (b) D₇₀P₃₀, (c) D₅₀P₅₀, (d) D₃₀P₇₀, and (e) D₀P₁₀₀.

For the DME-rich samples, the fine cubic crystals, representing one of the discharge products LiF, tend to be more abundant on the cathode surface, according to the SEM and EDX mapping images (Fig. 4). Meanwhile, carbon signals are found to be predominant for the PC-rich samples, which is very similar to the situation presented in the fresh electrode (Fig. 4 and Fig. S10, ESI†). The aforementioned results are in line with the phenomenon that the DME-rich samples are more favorable for the electrochemical conversion of the CF_x electrode as compared to the PC-rich samples. Additionally, the LiF crystals become powdery in the neat PC-based electrolytes, instead of the fine cubic crystals observed in the DME-rich samples. This suggests that the crystallization processes of LiF tend to be more sluggish in the presence of higher fractions of PC (cf. XRD and mechanism). The morphological results provide reliable proof that the electrolyte donicity and polarity greatly affect the electrochemical conversion process of the CF_x electrode. The DME-rich samples with higher donicity and moderate polarity could reduce the energy barrier, thus promoting the electrochemical cleavage of the C–F bond, compared with the PC-rich samples with lower donicity and higher polarity.

Fig. 5(a) shows the XRD patterns acquired from the CF_x cathode after discharge tests. As reported from previous studies,^27,33,38,80 the pristine CF_x cathode displays only the peaks assigned to graphite (100) (2θ = 42°) and fluorinated carbon material (002) (2θ = 13°). For all of the samples, new diffraction peaks [2θ = 38° (111), 45° (200), 65° (220) and 23° (002)] are observed after discharge, associated with the diminution of the line of the initial CF_x materials (i.e., 2θ = 13°), which indicates the successful conversion of CF_x materials into crystalline LiF (Fm3̄m) and graphitic carbon [i.e., 23° (002)].^{27,28,81–83} The detected intensities of the as-formed crystalline LiF are significantly higher for the DME-rich samples as compared to the PC-rich ones [e.g., 310 (D₇₀P₃₀) vs. 163 (D₃₀P₇₀)], which conforms to the trend in the discharge capacity [e.g., 678.2 (D₇₀P₃₀) vs. 578.1 (D₃₀P₇₀) mA h g⁻¹, Table S5, ESI†]. Additionally, the signals assigned to the pristine CF_x materials (i.e., 2θ = 13°) appear to be more intense in the PC-rich samples. These XRD results again suggest that the DME-rich electrolytes favor a more complete conversion reaction of the CF_x electrode during the discharge processes.

Fig. 5 (a) XRD patterns, (b) C 1s and (c) F 1s XPS spectra of the CF_x cathodes recovered from the discharged Li°||CF_x cells employing DME/PC mixed electrolytes.

Fig. 5(b) further compares the XPS spectra of C 1s and F 1s measured with the discharged CF_x cathodes, and the spectra of the fresh CF_x cathode are given in Fig. S11 (ESI†). As seen in Fig. S11 (ESI†), the presence of the C–F bonds in both ionic (i.e., 286.7 eV in C 1s, and 687.5 eV in F 1s) and covalent (i.e., 289.6 eV in C 1s, 688.4 eV in F 1s) features are confirmed, which could be ascribed to the active materials of CF_x particles. Additionally, the signals related to the –CF₂/–CF₃ groups (ca. 689.5 eV in F 1s, 291.3 eV in C 1s) are associated with the presence of the PVdF binder and pristine CF_x materials.^{22,27,80,84,85} After discharge tests, the C 1s spectra show several peaks assigned to the species containing sp² hybridized C=C bonds (i.e., 284.5 eV), C–O bonds (i.e., 286.2 eV), and C=O bonds (i.e., 290.1 eV), which are related to the formation of graphitic carbon as the discharge product and the adsorption/absorption of solvent molecules (Fig. S12, ESI†). In parallel, the F 1s spectra also validate the appearance of LiF as a discharge product in all of the cases (i.e., 684.5 eV).^{27–29,75,86} One may also note that the atomic percentages of sulfur and nitrogen-containing species are generally low (<5%), indicative of a small fraction of LiTFSI as the residual species on the surface of the discharged CF_x cathode [(i.e., being tentatively removed by washing with solvent) (Table S7 and Fig. S13, ESI†)]. It is important to highlight that the atomic percentages of Li-containing species are higher for the DME-rich samples vs. PC-rich ones, which is clear evidence for the more complete conversion of CF_x in the former electrolytes.

Combining the discharge behaviors with the morphological and compositional analyses, it could be concluded that the discharge performances directly reflect the degree of conversion reactions for the CF_x electrode, and the DME-rich electrolytes with higher donicity and moderate polarity solvent allow the Li°||CF_x cells to achieve a higher degree of conversion reactions, as compared with the PC-rich ones.

Solvation structure and discharge mechanism analysis

Since the solvated Li⁺ ions essentially participate in the heterolytic dissociation of the C–F bond during electrochemical processes, the solvation structure and coordination environment of the Li⁺ ions in the bulk electrolyte have been investigated by Raman spectroscopy (Fig. 6 and Fig. S14, S15, ESI†). The stretching vibration of the S–N–S band of sulfonimide anions (i.e., TFSI⁻ anion) appears in the region of 675–775 cm⁻¹ in the Raman spectra, which is very sensitive to the surrounding molecules/ions and can be adopted as a reference region to track the interactions between the lithium cations and salt anions.^87–90 In the same vein, the interactions between the lithium cations and solvent molecules (i.e., DME) can be followed in the region of 775–900 cm⁻¹ in the Raman spectra.^87,91

Fig. 6 Local Raman spectra and solution composition diagrams of the three selected kinds of electrolytes: (a₁) D₁₀₀P₀, (a₂) D₅₀P₅₀, and (a₃) D₀P₁₀₀ electrolyte in the region of 675–775 cm⁻¹; (b₁) D₁₀₀P₀, (b₂) D₅₀P₅₀, and (b₃) D₀P₁₀₀ electrolyte in the region of 775–900 cm⁻¹; composition diagram of the (c₁) D₁₀₀P₀, (c₂) D₅₀P₅₀, and (c₃) D₀P₁₀₀ electrolyte. The ellipsoids in grey and purple colours indicate the solvent molecules.

Fig. 6 compares the Raman spectra extracted from these two selected regions for the DME/PC mixed electrolytes. For the neat DME-based electrolyte (i.e., D₁₀₀P₀), a single peak (i.e., 739 cm⁻¹) is observed in the region of 675–775 cm⁻¹ (Fig. 6(a)), which is ascribed to the stretching vibration of the S–N–S band for the fully dissociated LiTFSI salt.^88,89,92 Meanwhile, the characteristic peaks associated with the free (i.e., 821 and 847 cm⁻¹) and solvated (i.e., 872 cm⁻¹) DME molecules are present in the region of 775–900 cm⁻¹ (Fig. 6(b)).^93,94 This result suggests the absence of a contact-ion pair (i.e., a TFSI⁻ anion interacting with one Li⁺) and aggregate solvates (i.e., a TFSI⁻ anion interacting with two or more Li⁺) in the neat DME-based electrolytes. With the incorporation of PC into the electrolytes, the first region (i.e., 675–775 cm⁻¹) shows additional peaks related to free (i.e., 712 cm⁻¹) and solvated (i.e., 725 cm⁻¹) PC molecules,^95–98 and the intensity of the solvated Li⁺–DME complexes is reduced (Table S8, ESI†). This implies the concurrent participation of PC molecules in solvating Li⁺ cations in the DME/PC mixed electrolytes. For all five kinds of electrolytes, a complete solvation of LiTFSI salt has been achieved, as indicated by the absence of any contact or aggregated ion pairs in the Raman spectra (Fig. 6 and Fig. S15, ESI†). However, for the equal weight mixture of PC and DME (being close to equimolar mixture), the solvated Li⁺–DME complexes are in higher fractions in contrast to the solvated Li⁺–PC compounds (Fig. 6 and Table S8, ESI†), indicating the preferential solvation interactions between the Li⁺ cations and DME solvents.

From the above-mentioned Raman results, the schematic illustrations of the solvation structure in the DME/PC mixed electrolytes are provided in Fig. 6(c). At a moderate salt concentration, full dissociation of lithium salts is likely to occur in all of the cases, without forming any ion pairs or ion aggregates. For the DME/PC mixed electrolytes, the solvation processes of Li⁺ cations predominantly involve the DME molecules, although PC is also involved in the solvation structures of the Li⁺–solvent complexes. This is primarily due to the higher electron-donicity of the DME molecules as compared to that of PC [i.e., 20.6 (neat DME) vs. 13.4 (neat PC) kcal mol⁻¹], and the chelating effect of DME, which renders this solvent stronger in terms of the Lewis acid–base interactions in the solvation processes.

As revealed in recent reports, the solvation structure of nonaqueous electrolytes has an important impact on the electrochemical performance for both primary batteries (e.g., Li°||CF_x cells) and other related rechargeable batteries. For primary lithium batteries, the incorporation of electrolyte solvents with a small molecular volume and high donor number (e.g., DMSO) contributes to enhanced solvation of Li⁺ ions in the resulting electrolyte. Such solvated Li⁺ ions could intercalate between the fluorinated carbon layers with larger distances, and subsequently improves the discharge specific capacity for the Li°||CF_x cells.^30,99 For rechargeable lithium-based batteries, the concentration of Li⁺ ions at the surface of the Li° electrode gradually decreases during the plating process in the diluted electrolytes, which leads to weakened interactions between the Li⁺ ions and solvent molecules, thus liberating an increasing amount of free solvents that may react with the Li° metal for generating solvent-derived SEI films. With highly salt-concentrated electrolytes, the content of free solvents is greatly suppressed due to their strong coordinating interactions with the excessive amount of Li⁺ ions, and the reduction of salt anions predominates over the solvent molecules, which leads to the formation of anion-derived SEI films.¹⁰⁰ Clearly, the exploration of electrolyte chemistry performed in this work also provides interesting insights into rechargeable lithium metal batteries and other kinds of batteries employing conversion-type electrode materials.^101–104

On the basis of the above experimental results, the special role of the electrolyte donicity and polarity on the discharge reaction of the Li°||CF_x cells is illustrated in Fig. 7. During the conversion reaction of the CF_x cathode, the stabilization of the positively-charged carbon atoms tends to be the rate-determining steps, as generally observed in the S_N2 reactions. In this sense, the DME-rich electrolytes with higher electron-donicity could better promote the coordination and stabilization of positively-charged species (i.e., fluorinated carbon atoms), as compared to that of the PC-rich electrolytes (Fig. 7(a)). This effectively enhances the reaction kinetics for the electrochemical breakdown of C–F bonds in the CF_x cathode (i.e., stronger push effect), although the desolvation processes of Li⁺ cations are less energetically favorable in the DME-rich electrolytes.

Fig. 7 Discharge mechanisms in the electrolytes with different donicities and polarities. (a) DME-rich electrolytes with a higher donicity and moderate polarity, (b) PC-rich electrolytes with a lower donicity and higher polarity. The balls in black, dark green, and pink represent carbon, fluorine, and lithium atoms, respectively. The ellipsoids in grey and purple colors indicate the solvent molecules.

It is important to note that the DME-rich electrolytes with a small portion of PC (i.e., D₇₀P₃₀) shows improved polarity, which gives the electrolyte a stronger capability for attracting F⁻ anions via electrostatic interactions (i.e., Li⁺ cation and F⁻ anion) and van der Waals force (i.e., the H atom of –CH₂– in PC and F⁻ anion). Consequently, the synergy between the electrolyte donicity (i.e., push effect) and polarity (i.e., pull effect) jointly leads to a higher degree of conversion reactions during discharge processes for the CF_x cathode, in comparison with the neat DME-based electrolytes.

However, the nucleation and precipitation of LiF is inhibited at a high fraction of PC. This is due to the strong interactions between PC and the F⁻ anion, as indicated by the higher polarity values of the PC-rich electrolytes and its higher acceptor number (AN_PC = 18.3 kcal mol⁻¹).¹⁰⁵ Therefore, the lower donicity and higher polarity of the PC-rich electrolytes cause sluggish kinetics for the breakdown of the C–F bonds and formation of the LiF crystals, and thus inferior discharge performances of the Li°||CF_x cells (Fig. 7(b)).

Implication for the design of high-energy batteries based on conversion-type electrodes

For the Li°||CF_x batteries, it has been widely acknowledged that the conversion processes of the CF_x electrode are tightly associated with the identity and features of the electrolyte component.^{25,30,33,34,38,106} By regulating the electrolyte compositions and deepening the mechanistic understanding of the conversion reactions, a higher discharge voltage and enhanced energy density could be attained for the Li°||CF_x batteries. In this work, the fascinatingly entangled parameters of nonaqueous electrolytes, enlisting permittivity, donicity, and polarity, are comprehensively investigated and reconciled, and the impact of these electrolyte descriptors on the electrochemical conversions of the CF_x cathode are elucidated. In projection, the mechanistic understanding provided herein could be transferred to other related battery systems. For instance, one may readily extends the present lithium-based CF_x chemistry to post-lithium batteries, including Na°||CF_x, and K°||CF_x batteries, based on the physical and chemical similarities among these alkali metal cations (Fig. 8). Furthermore, various conversion-type materials (Table S9, ESI†) beyond CF_x cathodes have emerged as promising candidates for high-energy primary and/or rechargeable battery systems, and their electrochemical performances show a clear dependence with the inherent features of the adopted electrolytes. The design and optimization of high-energy batteries utilizing these emerging conversion-type electrode materials could also benefit from the electrolyte-dependent CF_x chemistries revealed in the present work, for elucidating the intricate correlations between electrolyte components and the reaction processes of conversion-type electrode materials (Fig. 8). Lastly, the combination of sodium and other metal cations with these emerging conversion-type electrode materials provides an alternative solution to improve the technological sustainability and electrochemical performances of today's battery systems. Again, the insights provided herein are useful for the design and optimization of these battery configurations.

Fig. 8 Projection of the CF_x chemistry revealed in the present work to other kinds of primary and rechargeable batteries, which includes the extension of the CF_x chemistry from lithium to other monovalent or multivalent cation-based batteries (e.g., Na⁺, K⁺; left panel), and the shift from the CF_x chemistry to other conversion-type electrodes (e.g., sulfur or oxygen cathodes; right panel).

Conclusions

In summary, a series of nonaqueous electrolytes with different donicity and polarity are designed, and their performances in the Li°||CF_x cells are comprehensively evaluated, aiming to reconcile the unique impact of the electrolyte donicity and polarity on the discharge processes of conversion-type electrode materials. The concept of solvent polarity has been introduced into the design of nonaqueous electrolytes, for the first time, for the conversion-type electrode, which provides an extraordinary thinking tool for improving the performance of high-energy batteries.

Our results reveal that the electron-donicity and polarity of the electrolyte components are highly associated with the electrochemical reactions of CF_x materials. More specifically, the DME-rich electrolytes with higher donicity and moderate polarity solvent show a better completion of the conversion reaction [e.g., 649.2 (D₁₀₀P₀) vs. 678.2 (D₇₀P₃₀) vs. 202.3 (D₀P₁₀₀) mA h g⁻¹], as compared to the PC-rich ones with lower donicity and higher polarity. The morphological and compositional analyses validate that the DME-rich electrolytes promote the formation of graphitic carbon and crystalline LiF as discharge products, thereby boosting the specific capacity during discharge processes. Based on the aforesaid observations, a new mechanistic understanding of the conversion reactions of the CF_x cathode is proposed, in which a higher donicity of the nonaqueous electrolyte is beneficial for better stabilizing the positively-charged carbon atoms (i.e., stronger push effect). Simultaneously, a moderate polarity is favorable for the detachment of the negatively-charged fluorine atoms (i.e., stronger pull effect).

The present work not only advances the mechanistic understanding on the conversion reactions of the CF_x cathode, but also provides profound insight into the development of high-performing electrolytes for other kinds of high-energy batteries.

Author contributions

Z. Z. and H. Z. supervised the work. X. W. performed the experiments. The manuscript was written through contributions of all authors.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

The authors acknowledge the financial support from the National Natural Science Foundation of China (no. 22279037 and 52203223) and the Fundamental Research Funds for the Central Universities, HUST (2020kfyXJJS095).

References

1. J. M. Tarascon and M. Armand, Nature, 2001, 414, 359–367.
2. M. Armand and J. M. Tarascon, Nature, 2008, 451, 652–657.
3. B. Dunn, H. Kamath and J. M. Tarascon, Science, 2011, 334, 928–935.
4. P. Kurzweil and J. Garche, Lead-Acid Batteries for Future Automobiles, 2017.
5. M. Armand, P. Axmann, D. Bresser, M. Copley, K. Edström, C. Ekberg, D. Guyomard, B. Lestriez, P. Novák, M. Petranikova, W. Porcher, S. Trabesinger, M. Wohlfahrt-Mehrens and H. Zhang, J. Power Sources, 2020, 479, 228708.
6. D. Lin, Y. Liu and Y. Cui, Nat. Nanotechnol., 2017, 12, 194–206.
7. H. Zhang, L. Qiao, H. Kühnle, E. Figgemeier, M. Armand and G. G. Eshetu, Energy Environ. Sci., 2023, 16, 11–52.
8. R. Yazami, A. Hamwi, K. Guérin, Y. Ozawa, M. Dubois, J. Giraudet and F. Masin, Electrochem. Commun., 2007, 9, 1850–1855.
9. S. S. Zhang, D. Foster and J. Read, J. Power Sources, 2009, 188, 601–605.
10. Q. Zhang, S. D’Astorg, P. Xiao, X. Zhang and L. Lu, J. Power Sources, 2010, 195, 2914–2917.
11. N. Sharma, M. Dubois, K. Guerin, V. Pischedda and S. Radescu, Energy Technol., 2021, 9, 2000605.
12. Z. Fang, Y. Peng, X. Zhou, L. Zhu, Y. Wang, X. Dong and Y. Xia, ACS Appl. Energy Mater., 2022, 5, 3966–3978.
13. S. Zhang, L. Kong, Y. Li, C. Peng and W. Feng, Energy Environ. Sci., 2023, 16, 1907–1942.
14. Y. Ma, H. Zhang, B. Wu, M. Wang, X. Li and H. Zhang, Sci. Rep., 2015, 5, 14949.
15. G. Zhong, H. Chen, X. Huang, H. Yue and C. Lu, Front. Chem., 2018, 6, 50.
16. T. Nakajima, R. Hagiwara, K. Moriya and N. Watanabe, J. Electrochem. Soc., 1986, 133, 1761–1766.
17. A. Hamwi, M. Daoud and J. C. Cousseins, Synth. Met., 1988, 26, 89–98.
18. H. Groult, C. M. Julien, A. Bahloul, S. Leclerc, E. Briot and A. Mauger, Electrochem. Commun., 2011, 13, 1074–1076.
19. J. L. Adcock, P. F. Fulvio and S. Dai, J. Mater. Chem. A, 2013, 1, 9327–9331.
20. K. Yoshida, Y. Sugawara, M. Saitoh, K. Matsumoto, R. Hagiwara, Y. Matsuo, A. Kuwabara, Y. Ukyo and Y. Ikuhara, J. Power Sources, 2020, 445, 227320.
21. R. Zhou, Y. Li, Y. Feng, C. Peng and W. Feng, Compos. Commun., 2020, 21, 100396.
22. X.-X. Yang, G.-J. Zhang, B.-S. Bai, Y. Li, Y.-X. Li, Y. Yang, X. Jian and X.-W. Wang, Rare Met., 2021, 40, 1708–1718.
23. M. S. Whittingham, J. Electrochem. Soc., 1975, 122, 526–527.
24. K. Ueno, N. Watanabe and T. Nakajima, J. Fluorine Chem., 1982, 19, 323–332.
25. N. Watanabe, R. Hagiwara, T. Nakajima, H. Touhara and K. Ueno, Electrochim. Acta, 1982, 27, 1615–1619.
26. S. S. Zhang, D. Foster, J. Wolfenstine and J. Read, J. Power Sources, 2009, 187, 233–237.
27. B. Sayahpour, H. Hirsh, S. Bai, N. B. Schorr, T. N. Lambert, M. Mayer, W. Bao, D. Cheng, M. Zhang, K. Leung, K. L. Harrison, W. Li and Y. S. Meng, Adv. Energy Mater., 2021, 12, 2103196.
28. P. Chen, C. Jiang, J. Jiang, J. Zou, Q. Ran, X. Wang, X. Niu and L. Wang, ACS Appl. Mater. Interfaces, 2021, 13, 30576–30582.
29. Z. Ding, C. Yang, J. Zou, S. Chen, K. Qu, X. Ma, J. Zhang, J. Lu, W. Wei, P. Gao and L. Wang, Adv. Mater., 2021, 33, e2006118.
30. J. Jiang, H. Ji, P. Chen, C. Ouyang, X. Niu, H. Li and L. Wang, J. Power Sources, 2022, 527, 231193.
31. J. Ban, X. Jiao, Y. Feng, J. Xue, C. He and J. Song, ACS Appl. Energy Mater., 2021, 4, 3777–3784.
32. Z. Fang, Y. Yang, T. Zheng, N. Wang, C. Wang, X. Dong, Y. Wang and Y. Xia, Energy Storage Mater., 2021, 42, 477–483.
33. A. Fu, Y. Xiao, J. Jian, L. Huang, C. Tang, X. Chen, Y. Zou, J. Wang, Y. Yang and J. Zheng, ACS Appl. Mater. Interfaces, 2021, 13, 57470–57480.
34. Y. Xiao, J. Jian, A. Fu, C. Tang, Y. Zou, X. Chen, J. Wang, Y. Yang and J. Zheng, ACS Sustainable Chem. Eng., 2022, 10, 6217–6229.
35. M. Baek, H. Shin, K. Char and J. W. Choi, Adv. Mater., 2020, 32, e2005022.
36. M. Armand, “Electrode materialsjavascript:void(0); with internal redox activity”: M. Armand, PhD thesis, Grenoble Institute of Technology, France, 1977, 61.
37. A. J. Parker, Chem. Rev., 1969, 69, 1–32.
38. X. Wang, Z. Song, H. Wu, H. Yu, W. Feng, M. Armand, X. Huang, Z. Zhou and H. Zhang, Angew. Chem., Int. Ed., 2022, 61, e202211623.
39. R. J. Sengwa, S. Choudhary and P. Dhatarwal, J. Mol. Liq., 2017, 225, 42–49.
40. C. Reichardt and T. Welton, Solvents and Solvent Effects in Organic Chemistry, John Wiley & Sons, 2011.
41. M. Ue and S. Mori, J. Electrochem. Soc., 1995, 142, 2577–2581.
42. K. Xu, Chem. Rev., 2004, 104, 4303–4417.
43. E. Park, J. Park, K. Lee, Y. Zhao, T. Zhou, G. Park, M.-G. Jeong, M. Choi, D.-J. Yoo, H.-G. Jung, A. Coskun and J. W. Choi, ACS Energy Lett., 2022, 8, 179–188.
44. W. Linert, A. Camard, M. Armand and C. Michot, Coord. Chem. Rev., 2002, 226, 137–141.
45. W. J. Albery, Ann. Rev. Phys. Chem., 1980, 31, 227–263.
46. Y. Wang, Z. Cao, Z. Ma, G. Liu, H. Cheng, Y. Zou, L. Cavallo, Q. Li and J. Ming, ACS Energy Lett., 2023, 8, 1477–1484.
47. P. Xiao, X. Yun, Y. Chen, X. Guo, P. Gao, G. Zhou and C. Zheng, Chem. Soc. Rev., 2023, 52, 5255–5316.
48. Y. Zou, G. Liu, Y. Wang, Q. Li, Z. Ma, D. Yin, Y. Liang, Z. Cao, L. Cavallo, H. Kim, L. Wang, H. N. Alshareef, Y. K. Sun and J. Ming, Adv. Energy Mater., 2023, 13, 2300443.
49. V. Gutmann, Coordination Chemistry in Non-Aqueous Solutions, Springer-Verlag, 1968.
50. V. Gutmann, Coord. Chem. Rev., 1967, 2, 239–256.
51. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
52. M. D. Hanwell, D. E. Curtis, D. C. Lonie, T. Vandermeersch, E. Zurek and G. R. Hutchison, J. Cheminf., 2012, 4, 1–17.
53. R. Payne and I. E. Theodorou, J. Phys. Chem., 1972, 76, 2892–2900.
54. F. H. Drake, G. W. Pierce and M. T. Dow, Phys. Rev., 1930, 35, 613–622.
55. R. Kempa and W. H. Lee, J. Chem. Soc., 1958, 1936–1938.
56. R. J. Gillespie and R. S. Nyholm, Q. Rev., Chem. Soc, 1957, 11, 339–380.
57. M. Oki and H. Nakanishi, Bull. Chem. Soc. Jpn., 1971, 44, 3148–3151.
58. C. Reichardt, Angew. Chem., Int. Ed. Engl., 1965, 4, 29–40.
59. K. Dimroth, C. Reichardt, T. Siepmann and F. Bohlmann, Justus Liebigs Ann. Chem., 1963, 661, 1–37.
60. C. Reichardt, Chem. Rev., 1994, 94, 2319–2358.
61. O. Pytela, Collect. Czech. Chem. Commun., 1988, 53, 1333–1423.
62. I. M. Herfort and H. Schneider, Liebigs Ann. Chem., 1991, 27–31.
63. Y. Marcus, J. Solution Chem., 1984, 13, 599–624.
64. A. R. Harifi-Mood, F. Sharifi and H. R. Bijanzadeh, ChemistrySelect, 2021, 6, 600–608.
65. Y. Matsuda and H. Satake, J. Electrochem. Soc., 1980, 127, 877–879.
66. N. Yao, X. Chen, X. Shen, R. Zhang, Z. H. Fu, X. X. Ma, X. Q. Zhang, B. Q. Li and Q. Zhang, Angew. Chem., Int. Ed., 2021, 60, 21473–21478.
67. V. G. Machado and C. Machado, J. Chem. Educ., 2001, 78, 649–651.
68. M. F. Vitha, J. Chem. Educ., 2001, 78, 370–372.
69. D. F. Swinehart, J. Chem. Educ., 1962, 39, 333–335.
70. W. Mäntele and E. Deniz, Spectrochim. Acta, Part A, 2017, 173, 965–968.
71. R. H. Erlich and A. I. Popov, J. Am. Chem. Soc., 1971, 93, 5620–5623.
72. C. Pang, F. Ding, W. Sun, J. Liu, M. Hao, Y. Wang, X. Liu and Q. Xu, Electrochim. Acta, 2015, 174, 230–237.
73. W. Weppner and R. A. Huggins, J. Electrochem. Soc., 1977, 124, 1569–1578.
74. W. Li, L. Kong, Y. Hu, L. Sun, C. Peng, M. Qin, Z. Zhao, Y. Li and W. Feng, ACS Appl. Energy Mater., 2023, 6, 6132–6140.
75. K. Leung, N. B. Schorr, M. Mayer, T. N. Lambert, Y. S. Meng and K. L. Harrison, Chem. Mater., 2021, 33, 1760–1770.
76. N. Chen, G. Zhang, H. Chen and H. Yue, Coatings, 2023, 13, 812.
77. L. S. Yao, H. P. Zhou, G. T. Chen, S. Zhang, T. T. Feng, Z. Q. Xu, Z. X. Fang, H. M. Zhang, L. Zhang and M. Q. Wu, Diamond Relat. Mater., 2023, 131, 109545.
78. J. Yu, D. Wang, G. Wang, Y. Cui and S. Shi, Adv. Mater., 2023, 35, e2209210.
79. W. Martin, Y. Tian and J. Xiao, J. Electrochem. Soc., 2021, 168, 060513.
80. W. Liu, T. Yan, R. Guo, G. Dang, Y. Li, H. Pei, Q. Yan and J. Xie, J. Materiomics, 2021, 7, 1225–1232.
81. C. Jiang, B. Wang, Z. Wu, J. Qiu, Z. Ding, J. Zou, S. Chen, P. Gao, X. Niu, L. Wang and H. Li, Nano Energy, 2020, 70, 104552.
82. G. Zhong, H. Chen, Y. Cheng, L. Meng, H. Liu, Z. Liu, G. Zheng, Y. Xiang, X. Liu, Q. Li, Q. Zhang, H. Yue, C. Lu and Y. Yang, J. Mater. Chem. A, 2019, 7, 19793–19799.
83. S. Long, F. Chen, F. Ding, Y. Han, X. Liu and Q. Xu, J. Phys. Chem. C, 2019, 123, 28048–28057.
84. S. Xia, L. Luo, Y. Pang, J. Yang, S. Zheng, X. Zhang and Y. Huang, Small, 2021, 17, e2006002.
85. Z. Luo, X. Wang, D. Chen, Q. Chang, S. Xie, Z. Ma, W. Lei, J. Pan, Y. Pan and J. Huang, ACS Appl. Mater. Interfaces, 2021, 13, 18809–18820.
86. V. Krishnamurthy and V. Viswanathan, Chem. Mater., 2022, 34, 5799–5807.
87. L. Suo, F. Zheng, Y.-S. Hu and L. Chen, Chin. Phys. B, 2016, 25, 016101.
88. M. Martinez-Ibañez, N. Boaretto, L. Meabe, X. Wang, H. Zhu, A. Santiago, O. Zugazua, M. Forsyth, M. Armand and H. Zhang, Chem. Mater., 2022, 34, 7493–7502.
89. L. Meabe, S. R. Peña, M. Martinez-Ibañez, Y. Zhang, E. Lobato, H. Manzano, M. Armand, J. Carrasco and H. Zhang, J. Phys. Chem. C, 2020, 124, 17981–17991.
90. L. Qiao, S. Rodriguez Pena, M. Martinez-Ibanez, A. Santiago, I. Aldalur, E. Lobato, E. Sanchez-Diez, Y. Zhang, H. Manzano, H. Zhu, M. Forsyth, M. Armand, J. Carrasco and H. Zhang, J. Am. Chem. Soc., 2022, 144, 9806–9816.
91. P. Huang, X. Liu, Y. Wada, K. Katoh, M. Arai and M. Tamura, Fuel, 2013, 105, 364–367.
92. M. Martinez-Ibañez, E. Sanchez-Diez, L. Qiao, Y. Zhang, X. Judez, A. Santiago, I. Aldalur, J. Carrasco, H. Zhu, M. Forsyth, M. Armand and H. Zhang, Adv. Funct. Mater., 2020, 30, 2000455.
93. N. Xiao, W. D. McCulloch and Y. Wu, J. Am. Chem. Soc., 2017, 139, 9475–9478.
94. P. Zeng, Y. Han, X. Duan, G. Jia, L. Huang and Y. Chen, Mater. Res. Bull., 2017, 95, 61–70.
95. Y.-S. Kim and S.-K. Jeong, J. Spectrosc., 2015, 2015, 323649.
96. Z. Wang, W. Gao, X. Huang, Y. Mo and L. Chen, J. Raman Spectrosc., 2001, 32, 900–905.
97. K. Kondo, M. Sano, A. Hiwara, T. Omi, M. Fujita, A. Kuwae, M. Iida, K. Mogi and H. Yokoyama, J. Phys. Chem. B, 2000, 104, 5040–5044.
98. D. Battisti, G. A. Nazri, B. Klassen and R. Aroca, J. Phys. Chem., 1993, 97, 5826–5830.
99. H. Wang, J. Jiang, P. Chen, Z. Wu, X. Niu, C. Ouyang, J. Liu and L. Wang, J. Energy Chem., 2024, 89, 208–215.
100. J. Wang, J. Luo, H. Wu, X. Yu, X. Wu, Z. Li, H. Luo, H. Zhang, Y. Hong, Y. Zou, S. Cao, Y. Qiao and S. G. Sun, Angew. Chem., Int. Ed., 2024, 63, e202400254.
101. Q. Huang, T. C. Lourenco, L. T. Costa, Y. Zhang, E. J. Maginn and B. Gurkan, J. Phys. Chem. B, 2019, 123, 516–527.
102. J. Holoubek, H. Liu, Z. Wu, Y. Yin, X. Xing, G. Cai, S. Yu, H. Zhou, T. A. Pascal, Z. Chen and P. Liu, Nat. Energy, 2021, 2021, 303–313.
103. T. Hou, K. D. Fong, J. Wang and K. A. Persson, Chem. Sci., 2021, 12, 14740–14751.
104. Y. Zou, Z. Ma, G. Liu, Q. Li, D. Yin, X. Shi, Z. Cao, Z. Tian, H. Kim, Y. Guo, C. Sun, L. Cavallo, L. Wang, H. N. Alshareef, Y. K. Sun and J. Ming, Angew. Chem., Int. Ed., 2023, 62, e202216189.
105. H. Ishida, S. Azuma, N. Yamasaki, H. Kurita, T. Hasegawa, S. Ogo and T. Ueda, Anal. Sci., 2021, 37, 1131–1137.
106. S. S. Zhang, D. Foster and J. Read, J. Power Sources, 2009, 188, 532–537.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ee01793e

---