Tailoring Weakly Coordinating Electrolytes via Orbital‐Overlap‐Enhanced Dipole–dipole Interactions for Low‐Temperature Lithium‐Ion Batteries

ABSTRACT

Lithium‐ion batteries (LIBs) suffer rapid capacity fade at low temperatures. Weakly coordinating electrolytes via adding low‐polarity or non‐coordinating co‐solvents (such as fluorinated ethers) have shown promise in rapid desolvation, yet these electrolytes often exhibit low ionic conductivity at low temperature, limiting the application of high‐energy density LIBs. Here, we design a weakly coordinating electrolyte by incorporating the non‐coordinating co‐solvent (pentafluoroethyl)trimethylsilane (PFTMS) into the coordinating solvent diethyl carbonate (DEC) via orbital‐overlap‐enhanced dipole–dipole interactions. The slight Si─O orbital overlap drives strong dipole–dipole interactions between PFTMS and DEC. This interaction lowers the negative electrostatic potential at the carbonyl oxygen of DEC, thereby weakening Li⁺‐DEC coordination. Thus, by leveraging enhanced dipole–dipole interactions, this strategy realizes a weak Li⁺‐solvent coordination through a lower content of PFTMS (10 vol%) compared to traditional fluorinated solvents. Besides, the designed electrolyte delivers sufficient ionic conductivity of 1.56 mS cm⁻¹ at −40°C. Accordingly, the graphite || LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) coin cell shows reversible capacity of 156.5mAh g⁻¹ at −40°C. Notably, 4.7 Ah graphite || NCM811 pouch cell also demonstrates 219.8 Wh kg⁻¹ at −20°C. This work advances the design of traditional weakly coordinating electrolytes via an orbital overlap strategy, which paves the way for application in extreme environments.

Sluggish Li⁺ desolvation generally contributes to rapid capacity fade of lithium‐ion batteries at low temperatures. This work introduces a weakly coordinating electrolyte via orbital‐overlap‐enhanced dipole–dipole interactions, which effectively weakens the Li⁺–DEC binding energy and achieves fast Li⁺ desolvation. Remarkably, a 4.7 Ah graphite || LiNi_0.8Co_0.1Mn_0.1O₂ pouch cell with this designed electrolyte delivers 219.8 Wh kg⁻¹ at −20°C.

1. Introduction

Lithium‐ion batteries (LIBs) consisting of a layered metal oxide cathode, a graphite anode, and ethylene carbonate (EC)/diethyl carbonate (DEC)‐based electrolyte have been regarded as essential energy storage technology for a wider range of applications (such as portable devices and electric vehicles) [1, 2, 3, 4]. However, as the temperature drops below 0°C, electrolytes usually suffer from a significant drop in ionic conductivity and sluggish Li⁺ desolvation [5, 6]. Meanwhile, Li⁺ transport through the solid electrolyte interphase (SEI) and Li⁺ diffusion in the electrodes are also severely hindered [7, 8]. All these unfavorable factors will result in drastic capacity and energy loss in LIBs under low‐temperature conditions. Rational electrolyte design can enable low‐temperature LIB performance by increasing ionic conductivity, accelerating Li⁺ desolvation, and controlling the composition and structure of the SEI and the cathode electrolyte interphase (CEI) [9, 10, 11].

Recently, tuning Li⁺‐solvent interactions enables weakly coordinating electrolytes, which have emerged as a promising strategy to enhance low‐temperature performance of LIBs. Notably, reducing the polarity of solvents via fluorinating [12, 13, 14] and methylating solvents [15, 16] exhibits weak interactions between Li⁺ and solvents, thus achieving fast Li⁺ desolvation and contributing to weakly coordinating electrolytes. For example, the electrolyte employs a mono‐fluorinated linear ether of bis(2‐fluoroethoxy) methane (BFME), the ─O─ groups of BFME exhibit weak affinity to Li⁺ owing to the strong electron‐withdrawing ability of F. Therefore, the electrolyte can realize an anion‐participated solvation structure and decrease the Li⁺ desolvation energy. Finally, 2.4 Ah graphite || LiFePO₄ pouch cells show 1.6 Ah at −20C, which is 66.7% of the room‐temperature capacity [17]. However, they have low ionic conductivities due to poor lithium‐salt dissociation, contributing to poor Li⁺ transport kinetics and low energy densities at low temperatures.

In addition to the decreased polarity of solvent, employing intermolecular interactions (e.g., hydrogen bonds and dipole–dipole interactions) between non‐coordinating co‐solvents (such as fluorinated alkanes, fluorinated ethers, and fluorinated aromatic hydrocarbons, etc.) and coordinating solvents can also realize weakly coordinating electrolytes. For instance, the introduction of fluorinated ether bis(2,2,2‐trifluoroethyl) ether (75% in vol.) regulates the interactions with coordinating solvents to form localized high‐concentration electrolytes, reducing the proportion of free solvents, and promoting an anion‐enriched solvation structure. The graphite || Li cell with designed electrolyte yields 90 mAh g⁻¹ at −20°C [18]. Similarly, adding ethyl heptafluorobutyrate (40% in vol.) to EC can exert considerable dipole–dipole interactions, which can impair the Li⁺‒EC coordination, facilitating the desolvation of Li⁺. Thus, graphite || LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) pouch cells achieve 0.9 Ah at −10°C, retaining 69% of the room‐temperature capacity [19]. Despite these advantages, adding a large amount of non‐coordinating co‐solvents contributes to low ionic conductivity of the electrolyte (below 0.5 mS cm⁻¹ at −20°C), which limits the realization of high capacity and long cycle life at low temperatures.

Here we report an enhanced dipole–dipole interaction strategy via orbital overlap that pairs a non‐coordinating co‐solvent (fluorinated silanes) with a strongly coordinating solvent to create an electrolyte with weak Li⁺ coordination and sufficient ionic conductivity. This design enables high‐performance LIB operation at low temperatures. The electron‐withdrawing ─CF₂CF₃ group in PFTMS increases the charge carried by Si, generating a dipole–dipole interaction between Si and O. This interaction leads to the overlapping of their electron clouds. Furthermore, the 3d orbital of Si is empty, which facilitates a weak orbital overlap between Si and O. It further drives the redistribution of intramolecular electron density. Accordingly, the dipole–dipole interaction between Si and O via orbital overlap can be significantly enhanced, which decreases the negative electrostatic potential of the carbonyl oxygen (Figure 1a). Differing from the typical intermolecular interactions between non‐coordinating co‐solvent and coordinating solvent such as atypical hydrogen bonds (H^δ+ and O^δ−, H^δ+ and F^δ−) [20, 21], the electrolyte formulated with a small amount of PFTMS (10 vol%) enhanced dipole coupling via orbital overlap, which effectively weakens the Li⁺–coordinating solvent interactions (Figure 1b,c). The designed electrolyte facilitates rapid Li⁺ desolvation and maintains high ionic conductivity (1.56 mS cm⁻¹ at −40°C). Finally, the graphite || NCM811 cell with the designed electrolyte exhibits an initial high capacity of 221.8 mAh g⁻¹ at room temperature and retains 88.8% capacity after 1000 cycles under a 4.5 V charging cut‐off. When tested at −20°C and −40°C, the cells can retain 83% and 69% of room temperature capacity. In addition, a 4.7 Ah graphite || NCM811 pouch cell can deliver 208.8 Wh kg⁻¹ after 100 cycles at −20°C.

FIGURE 1.

(a) Schematic illustration of enhanced dipole–dipole interactions in an electrolyte via orbital overlap. (b) Schematic illustration of typical intermolecular interactions of an electrolyte with a large amount of non‐coordinating co‐solvent. (c) Schematic illustration of enhanced dipole–dipole interactions in an electrolyte via orbital overlap with a small amount of non‐coordinating fluorinated silane co‐solvent.

2. Results and Discussion

2.1. Enhancing Dipole–dipole Interactions via Orbital Overlap Between Non‐Coordinating Co‐Solvents and Coordinating Solvents

Six non‐coordinating co‐solvents, including 1,1,1,3,3‐pentafluorobutane (PFB), trimethylsilane (TMS), tetramethylsilane (TTMS), difluoromethyltrimethylsilane (DFTMS), (trifluoromethyl)trimethylsilane (TFTMS), and PFTMS, were selected to investigate the influence of orbital overlap on dipole–dipole interaction (Figure 2a and Figure S1). Specifically, for the non‐silicon solvent (PFB), density functional theory (DFT) calculations show that the charge carried by C atom is only +0.912e (the charge ‘e’ denotes the elementary charge) (Figure 2a), rendering its electrostatic interaction with nucleophilic sites (like C═O) intrinsically weak. In contrast, substituting C with Si increases the charge carried by Si. In particular, the higher charge carried by Si, the more likely Si is a favorable electrophilic site, thereby enabling Si to engage in dipole–dipole interactions with a nucleophilic site. When the H in TMS is replaced by ─CH₃ to form TTMS, the charge carried by Si increases from +0.972 to +1.121e (Figure 2a). The increase in the positive charge on Si is attributed to the higher electronegativity of the C in the substituted ─CH₃ group. The C exerts a stronger electron‐withdrawing inductive effect on Si, pulling more electron density away from Si. Subsequently, we also replaced the ─CH₃ group in TTMS with the electron‐withdrawing ─CHF₂, ─CF₃, and ─CF₂CF₃ groups, producing DFTMS, TFTMS, and PFTMS, respectively. The charge carried by Si gradually increases from DFTM to PFTMS (Figure 2a). The increased charge carried by Si enhances dipole–dipole interactions between Si and nucleophilic sites (like C═O), thereby facilitating the overlapping of their electron clouds. In addition, the empty 3d orbitals of Si have the potential to accept electrons. Thus, this electron cloud overlap facilitates a slight orbital overlap between Si and the p orbitals of the carbonyl oxygen (C═O) in coordinating solvents. Such orbital overlap promotes the redistribution of intramolecular electron density, leading to local charge polarization and an altered dipole moment [22, 23].

FIGURE 2.

(a) The charge carried by Si in different non‐coordinating co‐solvents. (b) E(2) values for the orbital interactions in the DEC‐PFB, DEC‐TMS, DEC‐TTMS, DEC‐DFTMS, DEC‐TFTMS, and DEC‐PFTMS (yellow represents regions with increased charge, while light blue represents regions with decreased charge). (c) Orbital overlap between Si (PFTMS) and carbonyl oxygen (DEC). The atoms are represented by spheres in the following colors: red for O, white for H, tan for C, cyan for F, and brown for Si. (d) The influence of E(2) and charge carried by Si on the dipole moment of DEC. (e) ²⁹Si chemical shift of different non‐coordinating co‐solvents upon mixing with DEC. (f) ¹⁷O chemical shift of DEC upon mixing with different non‐coordinating co‐solvents.

Then, DEC was selected as a coordinating solvent to detect the influence of orbital overlap between non‐coordinating co‐solvents and DEC. Besides, the commonly used fluorinated alkane non‐coordinating co‐solvent (like PFB) with no empty orbitals was chosen as a control. Natural bond orbital (NBO) analysis, as a key method for characterizing the properties of molecular orbitals, was employed to directly detect the orbital interaction between Si (C) in the non‐coordinating co‐solvent and the carbonyl oxygen in DEC. Especially, the second‐order perturbation energy, denoted as E(2), derived from NBO analysis, serves as a quantitative metric to characterize the strength of orbital interactions. The calculation result indicates that there are no orbital interactions between DEC and PFB (Figure 2b). In contrast, it is observed that E(2) values for the orbital interactions in the DEC‐TMS, DEC‐TTMS, DEC‐DFTMS, DEC‐TFTMS, and DEC‐PFTMS are 0.11, 0.15, 0.17, 0.21, and 0.24 kcal mol⁻¹ (Figure 2b). This confirms that such orbital overlap between Si and O arises from the presence of empty 3d orbitals in Si. In addition, the trend in E(2) value stems from the varying inductive effects of the substituents. Increasing electron‐withdrawing strength from ─CH₃ to ─CF₂CF₃ elevates the charge carried by Si, which facilitates orbital overlap between DEC and PFTMS. The orbital overlap integral between Si (PFTMS) and O (DEC) was directly calculated to be 0.017 via Multiwfn (Figure 2c).

To clarify how such weak orbital overlap between the Si of non‐coordinating fluorinated silanes and the O of DEC modulates the dipole moment of DEC, we subsequently calculated the corresponding dipole moment changes of DEC after the weak orbital overlap. The dipole moment of DEC was 0.72 D, which decreased to 0.65, 0.61, 0.48, 0.39, and 0.21 D after coupling with TMS, TTMS, DFTMS, TFTMS, and PFTMS (Figure S2). Thus, the increase in the charge carried by Si and orbital overlap between Si and O jointly regulate the dipole moment of DEC, inducing a pronounced reduction in its dipole moment (Figure 2d). According to the Born solvation model [24], the reduction in dipole moment facilitates a decline in solvation energy, which weakens the coordination ability between solvents and Li⁺. Thus, PFTMS induced the most pronounced reduction in the dipole moment of DEC, resulting in the weakest coordination interaction between DEC and Li⁺ among all silanes.

Further, to explicitly clarify the dipole–dipole interactions following weak orbital overlap, DFT calculations revealed that the interaction of Si^δ+ and O^δ− in the DEC/PFTMS mixture yields an energy of −18.85 kcal mol⁻¹ between DEC and PFTMS (Figure S3a). This is higher than the interaction between DEC and TFTMS (−16.61 kcal mol⁻¹, Figure S3b), DEC and DFTMS (−15.62 kcal mol⁻¹, Figure S3c), DEC and TTMS (−13.55 kcal mol⁻¹, Figure S3d), as well as the interaction between DEC and TMS (−11.13 kcal mol⁻¹, Figure S3e). In contrast, the interaction of C^δ+ (PFB) and O^δ− (DEC) is −7.35 kcal mol⁻¹ (Figure S3f), which is obviously weaker than the interaction between Si^δ+ and O^δ−. This indicates that the intermolecular interaction via orbital overlap is enhanced compared to no orbital interactions. Notably, the intermolecular interaction between PFTMS and DEC is the most pronounced under the orbital overlap effect. Thus, orbital overlap drives a stronger intermolecular interaction between the non‐coordinating and coordinating solvents, thereby weakening the Li⁺–solvent coordination.

Also, the lowest unoccupied molecular orbital (LUMO) energy level of fluorinated silanes was calculated by DFT. As shown in Figure S4, the LUMO energies of TMS, TTMS, DFTMS, TFTMS, and PFTMS are 1.09, 0.82, 0.71, 0.63, and −0.14 eV. Especially, the lower energy levels of DFTMS, TFTMS, and PFTMS are prone to undergoing reductive decomposition. Compared to the silanes (TMS and TTMS), fluorinated silanes (DFTMS, TFTMS, and PFTMS) are inclined to decompose into a LiF‐rich SEI, which is favorable to the cycling performance of the battery at low temperatures [25, 26]. Thus, we selected DFTMS, TFTMS, and PFTMS as the non‐coordinating co‐solvents.

Subsequently, nuclear magnetic resonance (NMR) was used to measure the interaction between non‐coordinating co‐solvent and DEC. When DEC was blended with non‐coordinating co‐solvent, PFTMS exhibits the most significant upfield shift in the ²⁹Si NMR spectra (δ = −1.21 ppm) (Figure 2e; Figure S5a) compared with TFTMS (δ = −0.43 ppm) (Figure S5b), and DFTMS (δ = −0.07 ppm) (Figure S5c). Correspondingly, as shown in Figure 2f and Figure S6, the chemical shift of carbonyl oxygen (¹⁷O NMR spectra) and carbonyl carbon (¹³C NMR spectra) in DEC also show a more obvious downfield shift after adding PFTMS, compared to that when DEC was mixed with TFTMS and DFTMS (Figure 2f; Figure S7). This demonstrates that the coupling of PFTMS to DEC via the orbital overlap remarkably decreases the electron cloud density around the carbonyl oxygen in DEC. The weakened Li⁺–DEC coordination is further corroborated by the significantly reduced electron density at the DEC carbonyl oxygen in ESP calculations (Figure S8). Except for the Si─O interaction via orbital overlap, both NMR spectroscopy and DFT calculations reveal strong intermolecular interactions between H^δ+ _(DEC) and F^δ− _{(fluorinated silanes)} (Figures S9 and S10). Notably, PFTMS exhibits the strongest H···F affinity, which further weakens Li⁺–DEC binding energy.

To further confirm the enhanced dipole–dipole interaction via orbital overlap in weakening the Li⁺–DEC binding energy, other factors (such as fluorination, polarity change, and H─F interaction) that might affect Li⁺–DEC coordination have been investigated. To detect the effect of fluorination on the Li⁺–DEC binding energy, the F atoms in PFTMS were replaced with H atoms to yield ethyltrimethylsilane (ETMS). DFT revealed that the interaction energy of DEC with ETMS (−14.16 kcal mol⁻¹) is weaker than that with PFTMS (−18.85 kcal mol⁻¹) (Figure S11), leading to a slightly stronger Li⁺–DEC binding energy (Figure S12). Similarly, the Si in PFTMS was replaced with a C atom to obtain a fluoroalkane. The C^δ+···O^δ− and H^δ+···F^δ− interaction energies are only −7.94 and −10.35 kcal mol⁻¹, respectively (Figure S11), resulting in a significantly stronger Li⁺–DEC binding energy of −43.1 kcal mol⁻¹ (Figure S12). Therefore, it can be deduced that although fluorination exerts a slight effect on reducing the Li⁺–DEC binding energy, the presence of orbital overlap serves as the main factor that weakens Li⁺–DEC binding energy. Then, the influence of polarity was also detected. The electrostatic potentials (ESP) of different fluorinated silanes were mapped. As described in the recent study, the maximum ESP (ESP_max) can serve as an effective descriptor to quantify the polarity difference between different molecules [27]. DFT was performed for DFTMS, TFTMS, and PFTMS to study the polarity change. Notably, the ESP_max values of DFTMS (0.751 eV), TFTMS (0.778 eV), and PFTMS (0.799 eV) show insignificant differences, implying similar polarity (Figure S13). Therefore, the impact of polarity changes of these co‐solvents on the weakened Li⁺–DEC coordination is negligible. The interaction energies of Si─O and H─F between PFTMS and DEC were also calculated (Figure S14). The Si─O interaction (−18.85 kcal mol⁻¹) is significantly stronger than the H─F interaction (−11.49 kcal mol⁻¹). Thus, the weakened Li⁺−DEC coordination is mainly attributed to the Si─O interaction via orbital overlap.

Besides, the universality of the enhanced dipole–dipole interaction via orbital overlap in other fluorinated silanes was also verified. The interaction energies between the carbonyl oxygen of DEC and different fluorinated silanes were evaluated using DFT (Figure S15a). The calculated interaction energies of DEC with short‐chain trimethylfluorosilane (TMF) and long‐chain trimethyl(nonafluorobutyl)silane (TMSNF) are −12.67 kcal mol⁻¹ (Figure S15b) and −16.96 kcal mol⁻¹ (Figure S15c), respectively. Besides, the second‐order perturbation energy (E(2)) was also calculated to evaluate the degree of Si and O orbital overlap. The E(2) values for the DEC‐TMF and DEC‐TMSNF are 0.156 and 0.227 kcal mol⁻¹, respectively (Figure S15d,e). Thus, it confirms that the dipole–dipole interaction via orbital overlap applies to other fluorinated silanes. Compared with the TMF and TMSNF, PFTMS exhibits a stronger interaction energy (−18.85 kcal mol⁻¹) and a higher E(2) value (0.24 kcal mol⁻¹). It indicates that the enhanced dipole–dipole interaction via orbital overlap is more pronounced in the PFTMS. This can be primarily attributed to the higher charge carried by Si in PFTMS.

To investigate the correlation among the Si atomic charge density, orbital overlap, and enhanced dipole–dipole interaction, we first examined how the degree of fluorination determines the charge carried by the Si. It shows that as the F atom increases from 1 to 5 (from TMF to PFTMS), the charge carried by Si increases non‐linearly (Figure S16a). However, when the F atom exceeds five (such as TMSNF), the charge carried by Si decreases. This is attributed to the long molecular chain and subsequent steric hindrance, which counteract the initial electron‐withdrawing effect and ultimately reduce the positive charge carried by the Si. Then, the relationship between the charge carried by Si and E(2) was evaluated (Figure S16b). An increase in the charge carried by Si leads to a sharp rise in E(2), eventually reaching the maximum orbital interaction between Si and O (near 0.24 kcal mol⁻¹). Furthermore, the near‐linear relationship between E(2) and the Si─O interaction energy suggests that as the degree of orbital overlap increases, the interaction energy becomes significantly stronger (Figure S16c). Therefore, it confirms that the orbital overlap results in enhanced dipole–dipole interactions between the Si and the O.

2.2. Electrolyte Solvation Structure

To directly characterize the solvating ability of DEC after it couples with different non‐coordinating co‐solvents, the donor number (DN) of DEC was measured by NMR. Specifically, the DN value can be regarded as a reflection of the electron‐donating ability of a solvent and further used to quantify the coordination capability between the solvent and Li⁺ [28, 29]. The DN of different mixing solvents is significantly decreased as DEC couples with the non‐coordinating co‐solvent (Figure 3a), indicating that the coordination between DEC and Li⁺ is weakened. In particular, the coordination between Li⁺ and DEC is weakened most significantly after Si─O coupling between PFTMS and DEC.

FIGURE 3.

(a) The DN of DEC and the corresponding DN of DEC upon mixing with PFTMS, TFTMS, or DFTMS. The full name of EA is ethyl acetate. (b) ⁷Li NMR spectra of different electrolytes. (c,d) The Raman spectra of different electrolytes. (e) The radical distribution functions and coordination numbers were calculated from MD simulations of the LDP electrolyte. (f) Li⁺ desolvation energies determined by a graphite || graphite cell. (g) The ionic conductivity of different electrolytes. (h) The LSV of different electrolytes.

To further illustrate the changes in the electrolyte solvation structure after the coupling of Si and O, the ⁷Li‐NMR was also conducted to reveal the solvation structure of electrolytes. In particular, LiDFOB was selected as the lithium salt due to its lower lowest unoccupied molecular orbital (LUMO) energy compared to conventional anions, facilitating the formation of a LiF‐rich SEI (Figure S17). Then, 1 m LiDFOB in DEC and 1 m LiDFOB in DEC with 10 vol% DFTMS, TFTMS, and PFTMS are referred to as LD, LDD, LDT, and LDP electrolytes (Table S1), respectively. Following concentration optimization, the formulation with 10 vol% PFTMS (LDP electrolyte) was identified as the optimal electrolyte to balance ionic transport and desolvation energy (Figures S18 and S19). As shown in Figure 3b, the ⁷Li peak exhibited an obvious downfield shift from LDD electrolyte to LDP electrolyte, indicating a decreased electron cloud density around Li⁺ in the solvation shell and a weaker shielding effect around Li⁺ in LDP electrolyte. Raman spectroscopy was further used to characterize the solvation structure of electrolytes. In Figure 3c, the Raman peak at around 902 cm⁻¹ originates from the vibration of free DEC and the peak at around 855 cm⁻¹ represents the coordination between Li⁺ and DEC. Notably, the Li⁺‐solvating DEC ratio decreases from 21% (LD electrolyte) to 10.8% (LDP electrolyte) (Figure 3d; Figure S20), indicating the weaker Li⁺‐solvent interaction in LDP electrolyte. This suggests that the weaker Li⁺‐O coordination binding energy results from lowered Li⁺‐DEC interactions, which is due to the decreased solvating capability of the [C═O] group of DEC in the LDP electrolyte. The solvation structures of the LDP and LD electrolytes were also characterized via classical molecular dynamics (MD) simulations to elucidate the underlying influence of PFTMS. For the LD electrolyte, the coordination numbers (CNs) of oxygen in DFOB⁻ and DEC were determined to be 1.17 and 2.58 (Figure S21), respectively. In contrast, upon the incorporation of PFTMS, the CN of DFOB⁻ increased to 1.48 and that of DEC decreased notably to 2.24, while PFTMS showed no participation in the Li⁺ primary solvation shell (Figure 3e; Figure S22). This confirms that PFTMS can effectively alter the solvation structure of the electrolyte.

To further elucidate the effect of Li⁺–DEC interactions on desolvation behavior, electrochemical impedance spectroscopy (EIS) at different temperatures was carried out to unveil the Li⁺ desolvation processes (Figure S23). As shown in Figure 3f, the Li⁺ desolvation energy is 26.3 kJ mol⁻¹ for LDP electrolyte, lower than those for LDT electrolyte (33.2 kJ mol⁻¹), LDD electrolyte (34.1 kJ mol⁻¹), and LD electrolyte (46.5 kJ mol⁻¹). This is because the Si‒O coupling weakens the ion‐dipole interaction between Li⁺ and DEC, thereby enabling rapid desolvation in the LDP electrolyte. Except for Li⁺ desolvation, the ionic conductivity of the electrolyte directly controls Li⁺ mobility in the bulk electrolyte, thereby affecting the low‐temperature performance of the battery. Thus, to detect the impact of enhanced dipole–dipole interactions on the ionic conductivity of electrolytes, we tested the ionic conductivity of four electrolytes at different temperatures. At −40°C, the ionic conductivities were 1.56, 1.58, 2.99, and 3.12 mS cm⁻¹ for the LDP, LDT, LDD, and LD electrolytes in Figure 3g, respectively. Although the addition of fluorinated co‐solvents inevitably leads to a decrease in ionic conductivity (Figure S24), all modified electrolytes maintain sufficient conductivity to ensure rapid bulk ion transport kinetics even at −40°C. Subsequently, to decipher the influence of Si─O coupling on the oxidation stability of electrolytes, the highest occupied molecular orbital (HOMO) energy level of DEC was calculated. Following the coupling of PFTMS, TFTMS, and DFTMS with DEC, the HOMO energies are −8.38, −8.31, −8.29 eV, respectively (Figure S25). The lower HOMO energy implies that materials can more difficultly donate electrons and undergo oxidation [28]. Especially, DEC exhibits the highest oxidation stability after the coupling of Si (from PFTMS) and O. Besides, linear sweep voltammetry (LSV) was utilized to directly evaluate the oxidation stability of electrolytes after the coupling between Si and O. As shown in Figure 3h, LDP electrolyte showed the highest oxidation stability (5.78 V) compared to LDT electrolyte (5.26 V) and LDD electrolyte (5.08 V). It indicates that the LDP electrolyte demonstrates superior compatibility with high‐voltage LIBs.

2.3. Electrochemistry Performance of graphite || NCM811 at Room Temperature

The electrochemical performance of LDP, LDT, and LDD electrolytes was compared by assembling graphite || NCM811 coin cells at room temperature. As shown in Figure 4a, cells cycled at 0.3C to a 4.5 V charge cut‐off with the LDT and LDD electrolytes delivered initial capacities of 188.4 and 154.5 mAh g⁻¹, respectively. By contrast, cells with the LDP electrolyte delivered an initial capacity of 221.8 mAh g⁻¹ and retained 88.8% of their capacity after 1000 cycles (Figure 4a), indicating superior high‐voltage cycling stability (Figure S26). These are significantly superior to those of the LD electrolyte (135.1 mAh g⁻¹) in Figure S27. As shown in Figure S28a, the graphite || NCM811 full cell with the LDP electrolyte also shows a high average Coulombic efficiency (99.93%) at room temperature. In comparison, the average Coulombic efficiency of the graphite || NCM811 full cell with the LDT electrolyte and LDD electrolyte is only 99.35% and 99.04% (Figure S28b,c). These results indicate a better reversibility of cells with LDP electrolyte.

FIGURE 4.

(a) Cycling performance of the graphite || NCM811 full‐cell using different electrolytes at room temperature. In situ DRT data representing different SOCs of a Li || Li ref. || Graphite cell with (b) LDP and (c) LDT electrolytes at room temperature. (d) Rate performance of graphite || NCM811 cells and (e) corresponding charge–discharge curves of LDP electrolyte. (f) The corresponding dQ/dV curves for the graphite || NCM811 full cell with LDP electrolyte. (g) Cycling performance of 1 Ah graphite || NCM811 pouch cell.

To deeply understand the superiority of the LDP electrolyte, the charge transfer resistance (R_CT), and the impedance of SEI and CEI (R_SEI and R_CEI) of the cell were measured by a three‐electrode system (Figure S29). Especially, R_CT is highly related to the Li⁺ solvation and desolvation at the anode and cathode, which have a significant impact on the performance of the cell. To visualize the trends of R_CT and R_SEI during cycling, in situ EIS‐combined distribution of relaxation times (DRT) was performed on the Li || Li ref. || Graphite cell with different states of charge (SOC). The Nyquist plots of the Li || Li ref. || graphite in LDP, LDT, and LDD electrolytes are in Figure S30a–c. Compared to the cell with LDT and LDD electrolytes, the negligible R_CT and R_SEI in LDP electrolyte demonstrate a much faster Li⁺ desolvation process and easier Li⁺ migration through SEI as shown in Figure 4b,c and Figure S30d. Besides, in situ EIS with DRT analysis was also extended to the NCM811 electrode. Consistently, the cells with LDP electrolyte exhibited significantly diminished peak intensities for both R_CEI and R_CT (Figure S31a,d) compared to LDT and LDD electrolytes (Figure S31b,c,e,f). This confirms the superior compatibility of the LDP electrolyte with the NCM811 cathodes, effectively lowering the Li⁺ desolvation and migration barriers at CEI. Then, the electrochemical rate test was conducted to reveal the reaction kinetics performance of the LDP electrolyte. As shown in Figure 4d,e, graphite || NCM811 full cells with the LDP electrolyte deliver the excellent rate performance with a capacity of 232.3, 219.3, 206.8, 192.8, and 180.7 mAh g⁻¹ from 0.2 to 3C. On the contrary, cells with LDT and LDD electrolytes exhibit poor rate performance in Figure S32. This superior rate capability is attributed to the rapid Li⁺ transport across the interphases both on anodes and cathodes, and low Li⁺ desolvation energy. Galvano‐static intermittent titration technique (GITT) was performed to further confirm the Li⁺ kinetics at the cathode interphase. As shown in Figure S33, the cell with LDP electrolyte exhibits a small overpotential (51 mV) compared to those with LDT (84 mV) and LDD electrolyte (106 mV), implying fast kinetic performance of LDP electrolyte.

To further understand the superiority of the LDP electrolyte, the XRD of graphite in graphite || NCM811 after 200 cycles was first tested (Figure S34). According to the XRD patterns, the graphite cycled in LDP electrolyte shows negligible change from the pristine state (Figure S34a), verifying the intact crystal structure after cycling. Conversely, the graphite cycled in LDT and LDD electrolytes exhibits obvious change (Figure S34b,c), indicating severe structural degradation after cycling. In addition, the differential capacity (dQ/dV) curves of graphite || NCM811 were conducted to directly verify the phase transition in graphite and NCM811 after cycling. As shown in Figure 4f, the phase transition from C₆ to Li_xC₆ exhibits high reversibility after 200 cycles in LDP electrolyte compared to that in LDT and LDD electrolyte (Figure S35a,b). Especially, the H2–H3 peak is associated with the phase transition of NCM811. In addition, H2 and H3 represent the two hexagonal phases that coexist during the deep delithiation of NCM811. The transition from H2 to H3 involves a drastic collapse of the interlayer spacing, which is typically triggered when the lithium content is significantly depleted at high cut‐off voltages [21]. The H2–H3 peak remains almost stable in LDP electrolyte, while the H2–H3 peak gradually weakens in LDT electrolyte and disappears in LDD electrolyte. This implies the highly reversible H2/H3 transition of NCM in LDP electrolyte. Thus, the high performance of the cell with the LDP electrolyte can be ascribed to the high reversibility of the phase transition of the graphite anode and NCM811 cathode during subsequent cycling. To further explore the practicality of the LDP electrolyte, we assembled 1Ah pouch cells. As shown in Figure 4g, when operated at room temperature and cycled at a 0.3C rate with a charging cut‐off voltage at 4.5 V, the pouch cell with LDP electrolyte achieves a capacity retention of 81.7% after 300 cycles, demonstrating better stability of LDP electrolyte. We also compared it with recent literature in terms of cut‐off voltage and capacity retention. As shown in Figure S36, the coin cell with LDP electrolyte exhibits excellent capacity retention (95.1% after 400 cycles), and cut‐off voltage (4.5 V).

To better highlight the superiority of the designed electrolyte, we compared it with three other electrolytes based on non‐coordinating co‐solvents of PFB, FB, and TTE. 1 m LiDFOB in DEC with 10 vol% PFB, 10 vol% FB, and 10 vol% TTE are referred to as PFB‐based, FB‐based, and TTE‐based electrolytes, respectively. The cells with PFB‐based, FB‐based, and TTE‐based electrolytes show a rapid capacity fading (Figure S37), while the cell using LDP electrolyte stands out with significant capacity retention (Figure S38). This confirms that the electrolyte designed through the Si─O coupling between non‐coordinating co‐solvents and coordinating solvents possesses distinct superiority. Besides, the physical properties of the co‐solvent are shown in Table S2. The ionic conductivity and Li⁺ transference number of different electrolytes were tested in Table S3. The Li⁺ transference numbers of LDD, LDT, and LDP electrolytes are around 0.7. Despite the LDP electrolyte exhibiting low bulk ionic conductivity (5.42 mS cm⁻¹), the cell with LDP electrolyte still achieves superior performance. Therefore, the physicochemical properties, including viscosity, dielectric constant, ionic conductivity, and Li⁺ transference number, are not key parameters in determining the electrochemical performance. We also prepared electrolytes using TMF and TMSNF, which are denoted as the TMF‐based electrolyte and the TMSNF‐based electrolyte. As shown in Figure S39a, a faster Li⁺ desolvation in LDP electrolyte (26.3 kJ mol⁻¹) can be achieved compared with that in the TMF‐based electrolyte (36.78 kJ mol⁻¹) and that in the TMSNF‐based electrolyte (31.59 kJ mol⁻¹). The graphite || NCM811 cells with the LDP electrolyte delivered a higher initial capacity of 221.8 mAh g⁻¹ and 99.77% capacity retention after 100 cycles. In contrast, cells using the TMF‐based electrolyte and the TMSNF‐based electrolyte exhibited lower initial capacities (183.9 and 177.8 mAh  g⁻¹) and low retentions (33.1% and 95.3%) (Figure S39b).

2.4. Chemistry at Anode and Cathode Interphases

The composition of the electrode electrolyte interphases plays a crucial role in the cycling stability of the battery. The LUMO energy of DEC and LiDFOB was calculated by DFT. The LUMO level of LiDFOB (−3.24 eV) is significantly lower than that of DEC (−0.73 eV) (Figure S40a). It demonstrates that DFOB⁻ decomposes more easily, which may be conducive to the formation of a stable interface. Then, the reduction behavior was evaluated via Cyclic Voltammetry (CV) of the Li || graphite cell. According to previous studies, LiDFOB typically reduces at ∼1.6 V [29, 30]. As shown in Figure S40b, the LDP electrolyte shows an initial reduction peak at 1.55 V, which is notably higher than those of LDT (1.45 V) and LDD (1.40 V). This higher reduction potential suggests the preferential decomposition of the salt over the solvent in LDP electrolyte, effectively mitigating the detrimental decomposition of DEC.

To further identify the interphasial chemistry, X‐ray photoelectron spectroscopy (XPS) with Ar⁺ sputtering depth profiling was employed to investigate the changes in SEI compositions on the graphite anode. As shown in Figure 5a and Figure S41a,d, for the battery using LDP electrolyte, the SEI is primarily composed of inorganic species (mainly LiF), and some organic species (mainly C─C, C═O, and RO‐Li). After 40 s of sputtering with Ar⁺, all the C 1s and O 1s spectra show a relatively diminished intensity, while the content of inorganic substances, especially LiF, increases dramatically and LiF becomes the predominant component in the SEI. Combining the results from the C 1s, O 1s, and F 1s spectra, it is concluded that the inner SEI layer is rich in LiF and the outer SEI layer is rich in organic and LiF. However, for the battery using LDT electrolyte, the SEI contains primarily more organic species (mainly C─C, C═O, and RO‐Li), and less LiF (Figure 5b; Figure S41b,e). As the etching time increases, the content of organic substances remains dominant in the SEI from C 1s and O 1s (Figure S41b,e). Although the SEI contains a certain amount of LiF after 40 s of etching (Figure 5b), the concentration of LiF is still lower than that of the organic species. Thus, it is determined that a bilayer SEI, consisting of a less LiF inner layer and an organics‐rich outer layer, was established. For the LDD electrolyte, the SEI is rich in organic species after 40 s of etching in Figure S41c,f. Especially, the content of LiF is very low in comparison to the organic species in Figure 5c. Therefore, such an SEI contains highly abundant organic in the LDD electrolyte. Especially, the organic components are primarily derived from the decomposition of DEC, and LiF and B‐F are mainly derived from the decomposition of DFOB⁻. Compared to the SEI composition of LDT and LDD electrolytes, it reveals a higher inorganic content of SEI in the LDP electrolyte. This is attributed to more anions in the solvation structure of the LDP electrolyte. Higher inorganic content of SEI in the cell with the LDP electrolyte also ensures the integrity of the graphite structure during cycling, which is consistent with the previously observed graphite XRD results [21].

FIGURE 5.

XPS depth profiles of F 1s on graphite in (a) LDP, (b) LDT, and (c) LDD electrolytes. SEM images of (d) Pristine NCM811 particles, NM811 cycled in (e) LDP, (f) LDT, and (g) LDD electrolytes. XPS depth profiles of C 1s on NCM811 in (h) LDP, (i) LDT, and (j) LDD electrolytes.

Except for the impact of SEI on battery performance, the composition of CEI after cycling is crucial for the stability of the battery at high voltage. Notably, the favorable composition of the CEI effectively preserves the structural integrity of NCM811 particles under high‐voltage. Thus, the morphology of NCM811 cathodes was characterized using scanning electron microscopy (SEM) after 200 cycles in LDP, LDT, and LDD electrolytes. SEM characterization showed that the pristine NCM811 electrode was intact without any cracks (Figure 5d). As shown in Figure 5e, the NCM811 cathode particles cycled in LDP electrolyte exhibit integrality after 200 cycles. The result indicates that NCM811 cathode material exhibits outstanding stability at high voltage in the LDP electrolyte. In contrast, particle cracking is clearly observed for the cycled cathodes in LDT and LDD electrolytes (marked by circles) in Figure 5f,g. These cracks lead to the loss of active materials and an increase in cathode impedance. Furthermore, the structural stability of the NCM811 cathode was also confirmed by XRD measurements. In Figure S42, the NCM811 cycled in LDP electrolyte maintains high structural stability. However, NCM811 cycled in the LDT and LDD electrolytes exhibits obvious changes, originating from the dissolution of Ni, Mn, and Co from NCM811, thus inducing phase changes on the NCM811 surface [31, 32].

XPS analysis with Ar⁺ sputtering (Figure 5h–j; Figure S43) also confirms that LDP electrolyte facilitates the formation of LiF‐rich interphases (SEI and CEI) dominated by LiF, whereas LDT and LDD electrolytes result in organic‐rich layers (C─C, C─O, and RO─Li) derived from DEC decomposition. This favorable enrichment of inorganic species in LDP‐derived interphases effectively stabilizes the NCM811, thereby enhancing the high‐voltage cycling stability of graphite || NCM811 batteries. The F 1s XPS of the cycled electrodes in LD electrolyte was also tested. It reveals the presence of the B─F, confirming that DFOB⁻ is an interphase activity. However, the LiF in both the SEI and CEI was negligible (Figure S44). In contrast, a LiF‐rich SEI/CEI is prominently detected on the electrode surfaces cycled in the LDP electrolyte. This is because the strong Si─O interaction effectively weakens the Li⁺‐DEC coordination. The interaction drives more DFOB⁻ anions into the solvation sheath, creating an anion‐rich solvation structure. Consequently, the LiF‐rich SEI/CEI is derived from the decomposition of LiDFOB in LDP electrolyte. Additionally, B‐containing species are also detected on SEI and CEI, which directly confirms the decomposition of LiDFOB (Figure S45). Thus, it can be concluded that the improved interphase chemistry is mainly attributed to the LiF‐rich SEI/CEI derived from the anion‐rich solvation structure regulated by PFTMS.

2.5. Electrochemistry Performance of Graphite || NCM811 at Low Temperatures

To further verify the influence of the superior solvation structure and electrolyte electrode interphase in the designed electrolyte at low temperatures, we subsequently tested the low‐temperature performance of the graphite || NCM811 coin cells and 4.7 Ah pouch cells. As shown in Figure 6a, the coin cells with LDP electrolyte display a high capacity of 184 mAh g⁻¹ at −20°C. In contrast, the capacity of the cells with LDT and LDD electrolytes shows capacities of 172 and 160.8 mAh g⁻¹. Particularly, the cell with the LDP electrolyte exhibits steady cycling performance with higher capacity retention (95.2%) at 0.2C (Figure 6b), while the capacity of the cells with LDT and LDD electrolytes experience fast decay with only 46.1% and 39.4% capacity retention after 150 cycles at 0.2C (Figure 6c; Figure S46). The cells using LD electrolyte exhibited much inferior performance, delivering a lower initial capacity of 158.4 mAh g⁻¹ andae capacity retention of 26.7% after 100 cycles (Figure S47). Additionally, the cells with LDP electrolyte exhibited an average Coulombic efficiency of 99.86% (Figure S48a), which is much higher than those of LDT (98.72%), and LDD (98.72%) (Figure S48b,c). It indicates a better reversibility of cells with LDP electrolyte at −20°C.

FIGURE 6.

(a) Cycling performance of the graphite || NCM811 full‐cell using different electrolytes at −20C and corresponding charge–discharge curves of LDP (b) and LDT (c) electrolytes. (d) DRT data of graphite || NCM811 cell with LDP and LDT electrolytes at −20C. (e,f) Cycling performance of 4.7 Ah graphite || NCM811 pouch cell at −20C. (g) Cycling performance of graphite || NCM811 cells with LDP electrolyte in the range 2.8–4.5 V at different temperatures. (h) Cycling performance of the graphite || NCM811 full‐cell using LDP electrolyte at −40 C. (i) Comparison of LIB coin cells cycling performance between LDP electrolytes and published work at −20C [26, 28, 31, 33, 34, 35, 36, 37, 38]. (j) Comparison of LIB pouch cells cycling performance between LDP electrolytes and published work at −20 C [19, 26, 31, 38, 39, 40, 41, 42].

Subsequently, we utilized DRT in the graphite || NCM811 cell to analyze the R_CT and R_SEI to investigate the reasons for the excellent performance at low temperatures. As shown in Figure 6d, the R_SEI and R_CT of the cell are negligible in LDP electrolyte compared to that of LDT and LDD electrolytes (Figure S49), which helps graphite || NCM811 achieve high cycle stability at −20C. Notably, 4.7 Ah graphite || NCM811 pouch cell with LDP electrolyte delivers a capacity over 4.0 Ah and an·energy density of 219.8 Wh kg⁻¹ at −20C (Figure 6e,f and Table S4). In addition, LDP electrolyte supports high reversibility of graphite || NCM811 cells in a temperature window ranging from −40 to 20C (Figure 6g; Figure S50). The capacity of the cell with LDP electrolyte shows 223.5, 198.3, 184.6, 166.3, and 152.1  mAh  g⁻¹ when cycled at 30, 0, −20, −30, −40C, respectively. Especially, the graphite || NCM811 cells with LDP electrolyte show highly stable and reversible capacity over 150 mAh g⁻¹ at −40C in Figure 6h. Here we summarize the recent work at −20C and evaluated our developed electrolyte in both coin cell (Figure 6i; Table S5) and pouch cell configurations (Figure 6j; Table S6). Remarkably, the cells utilizing our designed electrolyte demonstrate significant superiority, outperforming previously reported works [19, 26, 31, 33, 34, 35, 36, 37, 38, 39, 40, 41, 43, 44]. Moreover, graphite || NCM811 cells with the LDP electrolyte delivered a superior initial capacity of 233.1 mAh g⁻¹ and maintained 84.9% retention after 300 cycles at 50°C (Figures S51). Conversely, cells utilizing LDT, LDD, and LD electrolytes exhibited lower initial capacities (227.4, 183.9, and 177.8 mAh g⁻¹, respectively) and suffered rapid capacity fading. It indicates excellent wide‐temperature adaptability of the LDP electrolyte (from −40 to 50 C).

3. Conclusion

In summary, we propose an electrolyte featuring enhanced dipole–dipole interactions via orbital overlap to induce a weakly coordinating electrolyte. This design successfully addresses the insufficient ionic conductivity of LIBs at low temperatures. The key mechanism lies in the formation of Si−O coupling via a weak orbital overlap, which strengthens high dipole–dipole interactions between PFTMS and DEC. The unique interactions can effectively weaken the interaction of Li⁺‐DEC with 10 vol% PFTMS, contributing to fast Li⁺ desolvation. Additionally, the designed electrolyte achieves high ion conductivity (1.56 mS cm⁻¹ at −40°C), enhances the oxidative stability, and facilitates the formation of anion‐derived interphases on both anodes and cathodes. The performance of graphite || NCM811 cells with the designed electrolyte also realizes a high capacity of 221.8 mAh g⁻¹ at room temperature and retains 88.8% capacity after 1000 cycles. Even when the temperature drops to −20 C and −40°C, the cells can retain 83% and 69% of room temperature capacity. Meanwhile, 4.7 Ah graphite || NCM811 pouch cell can deliver 219.8 Wh kg⁻¹ at −20°C. Ultimately, employing dipole–dipole interactions via orbital overlap between non‐coordinating solvent and coordinating solvent in the electrolyte showcases an encouraging avenue for advancing LIBs at low temperatures.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma73982‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.