Electrolyte design combining fluoro- with cyano-substitution solvents for anode-free Li metal batteries

Significance

A cyano-substitution ether, ethylene glycol bis(propionitrile) ether (AN2-DME), was proposed, in which the strong electron-withdrawing cyano-group restrains the loss of lone-pair electrons of ether oxygen and enhances the oxidative stability. Paired with FEC and TTE, the 1 M LiPF₆ in AN2-DME/FEC/TTE electrolyte facilitates the formation of N-containing and LiF-rich solid–electrolyte interphase as well as stable and uniform cathode–electrolyte interphase. A stable cycle of Li metal anode and 4.3 V NCM811 cathode is achieved. Anode-free pouch cells deliver a high energy of 397.5 Wh kg^-1 with 76% capacity maintenance after 100 cycles. This work will inspire battery communities to introspect the very hot fluoro-substitution strategy and act as a primer to ignite the enormous enthusiasm on the cyano-substitution solvents for high-energy Li metal batteries.

Abstract

Fluoro-substitution solvents have achieved great success in electrolyte engineering for high-energy lithium metal batteries, which, however, is beset by low solvating power, thermal and chemical instability, and possible battery swelling. Instead, we herein introduce cyanogen as the electron-withdrawing group to enhance the oxidative stability of ether solvents, in which cyanogen and ether oxygen form the chelating structure with Li⁺ not notably undermining the solvating power. Cyano-group strongly bonds with transition metals (TMs) of NCM811 cathode to attenuate the catalytic reactivity of TMs toward bulk electrolytes. Besides, a stable and uniform cathode–electrolyte interphase (CEI) inhibits the violent oxidation decomposition of electrolytes and guarantees the structural integrity of the NCM811 cathode. Also, a N-containing and LiF-rich solid–electrolyte interphase (SEI) in our electrolyte facilitates fast Li⁺ migration and dense Li deposition. Accordingly, our electrolyte enables a stable cycle of Li metal anode with Coulombic efficiency of 98.4% within 100 cycles. 81.8% capacity of 4.3 V NCM811 cathode remains after 200 cycles. Anode-free pouch cells with a capacity of 125 mAh maintain 76% capacity after 100 cycles, corresponding to an energy density of 397.5 Wh kg^-1.

Li-ion batteries (LIBs) have dominated the market of electric vehicles (EVs), consumer electronics, drones, and grid-scale energy storage. The energy density is prominently restrained by graphite anode (theoretical specific capacity of 372 mAh g⁻¹) and traditional lithium transition metal (TM) oxide cathodes (LiCoO₂, LiMn₂O₄, LiFePO₄, etc.) (1), struggling to satisfy the demand for 400 Wh kg⁻¹ (2). The combination of Li metal anode (LMA) and Ni-rich NCM (LiNi_xM_1−xO₂, M = Mn, Co, and x > 0.6) cathode is considered as a practical alternative to pursue 400 Wh kg⁻¹ at the cell level. LMA with low potential (−3.04 V vs. SHE) and high specific capacity (3,860 mAh g⁻¹) is the ultimate anode choice for high-energy battery systems (3–5), which, however, is afflicted by the notorious issues of low coulombic efficiency (CE) and dendritic Li growth (6, 7). Meanwhile, NCM cathodes suffer from irreversible phase transition (8, 9), gas generation (10), TM dissolution (11), stress-corrosion cracking of the NCM secondary particles (12, 13), and unceasing growth of cathode–electrolyte interphases (CEIs) (9). The issues of LMA and NCM cathode are closely associated with their interface chemistry (14, 15), which renders the operando interface modification by electrolyte engineering imperative for adopting LMA and Ni-rich NCM cathode.

Liquid electrolyte engineering is regarded as a cost-effective and pragmatic alternative to address the root cause, i.e., uncontrollable parasitic reactions between LMA and electrolytes (16). By tailoring electrolyte components, the solid electrolyte interphase (SEI) chemistry and Li morphology can be manipulated to improve the cycling stability and tackle safety issues of LMA. Among the commonly used solvents in LIB electrolytes, ether solvents are considered more compatible with LMA because of their high lowest unoccupied molecular orbital (LUMO) (17–19). Besides, the reduction of ether solvents will construct Li₂O-rich SEI that will boost the Li lateral diffusion along the SEI interface because of its high interfacial energy and suppress the dendritic Li growth and penetration into SEI due to the high Young’s modulus (3, 20). Great progress has been made in ether electrolytes achieving high CEs of >99.5% for Li deposition/stripping (6, 21, 22). Nevertheless, ether solvents hardly withstand 4 V-class cathode chemistry (23), because lone-pair electrons of ether oxygen readily lose at the cathode interface when charged to high voltages (24), hindering their applications in high-voltage Li metal batteries (LMBs).

Fluoro-substitution is a routine approach to improve the oxidative stability of ether solvents (25–28), because of its strong electron-withdrawing capability to restrain the loss of lone-pair electrons of ether oxygen. Additionally, fluoro-substitution lowers LUMO of ether solvents resulting in their facile reduction to construct a LiF-rich SEI that is widely recognized beneficial for LMA (27). Therefore, in recent years, fluoro-ether solvents have ignited tremendous interest and achieved great success (26–29). Yet, some issues of fluoro-substitution cannot be ignored. The strong electron-withdrawing capability of fluorine will delocalize the electron clouds of fluoro-ether molecules, considerably cutting down the polarity and solvating power (30–32). The lowering solvating power of ether oxygen plus that −F itself can hardly coordinate with Li⁺ leads to the low dissociation of lithium salts (30). Worse still, the thermal and chemical instability of fluoro-ether solvents readily causes the degradation of electrolytes during battery operation, which will generate hydrogen fluoride attacking oxide cathodes and current collectors as well as destroying the SEI of the anode (33, 34). In addition, the lowering LUMO of fluoro-ether solvents leads to their accelerated decomposition on the anode side, resulting in the electrolyte dry-up, battery swelling, and ultimately battery failure (35, 36). Furthermore, the high cost and potential environmental hazards of fluoro-ether solvents will restrain their application as the main solvents of electrolytes (34, 37, 38).

Instead of fluoro-substitution, cyanogen, known as pseudohalogen (39), can also withdraw the lone-pair electrons of ether oxygen to enhance the oxidative stability of ether solvents (40). Compared with fluorine, the electron-withdrawing capability of cyanogen is gentler (39), which will not considerably cut down the solvating power of ether solvents. Moreover, cyano-group and ether oxygen can form a chelating structure with Li⁺ (41, 42), contributing to the dissociation of lithium salts. The reduction of the cyano-group will generate N-containing SEI in the LMA (Fig. 1A) (42), which facilitates the fast Li⁺ migration and dense Li deposition (17, 20, 43). Besides, cyano-ether solvents have decent thermal and chemical stability in favor of long-term cycling stability especially under abuse conditions (44).

Fig. 1.

Challenges and solutions of electrolyte for LMBs with high-Ni cathode. (A) Our electrolyte favors the formation of dense SEI with high Young’s modulus and fast Li⁺ conduction, contributing to a highly reversible LMA. Besides, our electrolyte can eliminate the trace H₂O and HF in electrolyte and restrain the catalytic decomposition of FEC by PF₅. Uniform CEI is formed to suppress the dissolution of Ni⁴⁺, electrolyte penetration, and HF attack. (B) With traditional electrolytes, loose SEI, Li dendrite, and severe parasitic reaction undermine the LMA. Hydrolysis of LiPF₆, decomposition of FEC, and non-uniform CEI lead to severe HF attack, TM dissolution, and particle cracking of NCM811 cathode.

For high-nickel LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) cathode, cyano-ether solvents might contribute to the formation of a stable and uniform CEI (41, 45), which inhibits the violent oxidation decomposition of electrolyte at high voltage and guarantees the structural integrity of NCM811 cathode (Fig. 1A). Besides, the cyano-group is preferentially absorbed on the surface of NCM811 cathode to form a microstructure of R-CN-TM due to the hybridization of N2p orbital of cyanogen with TM-3d orbital of TM^3+/4+ during the charging process (46, 47). The contribution of lone-pair electrons on N2p orbital of cyanogen will effectively reduce the real valence state of TM^3+/4+ for the surface NCM811 crystalline, thereby attenuating the catalytic reactivity of TM ions toward bulk electrolyte. Also, the cyano-group can react with trace water of electrolyte to lower water content (48), which relieves the process of LiPF₆ decomposing into HF. Additionally, due to the catalysis of Lewis acid (PF₅), fluoro-substitution solvent is easily decomposed with the release of hydrogen fluoride (HF) (Fig. 1B) (33, 49), which will damage the CEI layer and cause the TM dissolution of NCM811 cathode. Cyano-group exhibits a strong bonding ability with PF₆^-, restraining the catalytic decomposition of fluoro-substitution solvent and formation of HF (50).

In this work, we proposed a cyano-substitution ether, ethylene glycol bis(propionitrile) ether (AN2-DME), in which the strong electron-withdrawing cyano-group restrains the loss of lone-pair electrons of ether oxygen and enhances the oxidative stability to >5.0 V. Additionally, both ether oxygen and cyano-group of AN2-DME take part in the coordination with Li⁺ in favor of the dissociation of lithium salts. Paired with FEC and TTE, the 1 M LiPF₆ in AN2-DME/FEC/TTE electrolyte facilitates the formation of N-containing and LiF-rich SEI as well as stable and uniform CEI, contributing to the dense Li deposition and the structural integrity of NCM811 cathode, thereby supporting the stable long-term cycling of LMA and 4.3 V NCM811 cathode.

Results and Discussion

Design Logic of AN2-DME Molecular Structure.

The electrostatic potential (ESP) distribution of four solvent molecules (DME, AN, AN2-DME, and F6DEE) was evaluated using density functional theory (DFT) calculations (Fig. 2A). For DME and AN, the negative charges are concentrated on ether O atoms and cyanogen N atoms, respectively. While the negative charges of AN2-DME are observed on both O and N atoms, indicative of the Li⁺-coordination ability of O and N atoms, which is consistent with the DFT-optimized coordination structures of Li⁺-AN2-DME (Fig. 2D) and consolidated by donor number (DN) of AN2-DME solvent (SI Appendix, Fig. S1 and Table S1). By contrast at the same iso-potential scale, the negative charges on O atoms of F6DEE are considerably attenuated due to the high electron-withdrawing capability of -F group, suggesting that F6DEE hardly coordinates with Li⁺. In the DFT-optimized coordination configurations between Li⁺ and each type of solvent molecules (Fig. 2 B–D), the Li⁺ ions exhibit tripod and linear coordination geometry in Li⁺-DME (Fig. 2B) and Li⁺-AN complex (Fig. 2C), respectively, with oxygen and nitrogen atoms interacted with Li⁺ ions. While in the coordination structure of Li⁺-AN2-DME, the Li⁺ ions display a tetrapod coordination geometry with both oxygen and nitrogen atoms corresponding to the distance of 2.12 and 2.14 Å, respectively (Fig. 2D). Such interaction between Li⁺ and oxygen and nitrogen atoms can be rationalized by the fact that the ether oxygen and cyanogen nitrogen are locally polar and negatively charged in the ESP (Fig. 2A).

Fig. 2.

Design concept of the AN2-DME solvent. (A) ESP distribution of four solvents: DME, AN, AN2-DME, and F6DEE. The ESP distribution is depicted in the range of −1.70 to 1.20 eV. The blue and red regions represent negative and positive charges, respectively. (B–D) Coordination structures and binding energies between one Li⁺ ion and one solvent molecule calculated using DFT: Li⁺-DME (B), Li⁺-AN (C), and Li⁺-AN2-DME (D). (E) HOMO and LUMO of four solvents: DME, AN, AN2-DME, and F6DEE.

The highest occupied molecular orbital (HOMO) and LUMO energy levels were calculated using DFT to decipher the stability window of DME, AN, AN2-DME, and F6DEE (Fig. 2E). Among these four solvents, DME has the highest values of HOMO and LUMO energies in contrast with the lowest values of HOMO and LUMO energies of AN. Accordingly, DME has the lowest propensity to accept electrons being reduced and meanwhile the highest propensity to be oxidatively decomposed at the 4 V-class cathode interface. After the substitution by electron-withdrawing –CN and –F groups, both HOMO and LUMO energies are lowered, indicating that the resistance to be oxidized is enhanced in AN2-DME and F6DEE at the expense of the preference to be reduced. It is well known that the lone pair electrons of ether oxygen are the culprit of the poor oxidative stability of DME (24). In AN2-DME, the electron-withdrawing cyano-group lowers the propensity to lose the electrons, which can be experimentally verified using linear sweep voltammetry (LSV) measurements (SI Appendix, Fig. S2) and potentiostatic intermittent titration technique (PITT) on Li|Al cells (SI Appendix, Fig. S3). Unlike the low oxidation voltage of DME-based electrolyte (~4.3 V), the AN2-DME-based electrolyte shows considerable high-voltage tolerance by giving oxidation voltage at ~6.0 V. Meanwhile, the electron-rich ether oxygen will donate electrons to cyanogen, which will elevate the LUMO of AN2-DME compared to AN and lower the propensity of cyano-group to be reduced.

Physical Properties and Solvation Structure.

The physical properties of electrolytes were studied with spectroscopies (Fig. 3). The Raman spectra of FEC ring breathing mode (51) and PF₆⁻ anion vibration band (52–54) drift downward in the DME-based electrolyte (Fig. 3A), indicating that DME is prone to bond with Li⁺ and promotes the complete dissociation of LiPF₆ to produce more free PF₆⁻ anions than AN2-DME and AN solvents. The ratio of uncoordinated -CN (55) in AN2-DME is much higher than that in AN solvent (Fig. 3B), indicative of the lower donor number of AN2-DME, which is consolidated by Fourier-transform infrared (FT-IR) spectra of ν(C≡N) (Fig. 3C) (56). The FT-IR spectra of C=O stretching mode of FEC (3, 57–60) show that AN2-DME-based electrolyte has the highest ratio of coordinated FEC (Fig. 3D), which can be ascribed to low permitivity of AN2-DME causing the majority of FEC coordinated with Li⁺. Furthermore, the low permitivity of AN2-DME leads to the weak dissociation of LiPF₆ and high ratios of coordinated PF₆⁻ anion (SI Appendix, Fig. S4) (61), as well as high ratios of ether oxygen (C–O–C) coordinated with Li⁺ (SI Appendix, Fig. S5) (62, 63). NMR was adopted to probe the ion-solvent interaction in three electrolytes (Fig. 3 E–I and SI Appendix, Figs. S6–S8). The chemical shift of ³¹P nuclei in PF₆⁻ anions (64–66) exhibits a blue shift in the order of DME > AN > AN2-DME (Fig. 3E), denoting that the affinity of PF₆⁻ with Li⁺ becomes increasingly strong, which is confirmed by the chemical shift of ¹⁹F nuclei in PF₆⁻ (Fig. 3F) (66). The chemical shift of ¹⁹F nuclei in FEC (66) is largest in the AN2-DME-based electrolyte (Fig. 3G), signifying that the affinity between FEC and Li⁺ is strongest whilst the interaction between AN2-DME and Li⁺ is weakest in the electrolyte, which can be consolidated by the chemical shift of ¹H nuclei in FEC (Fig. 3I) (66–68). The chemical shift of ¹⁹F nuclei in TTE decrease in the order of AN2-DME > AN > DME (Fig. 3H and SI Appendix, Fig. S6) (69). Given that TTE hardly coordinates with Li⁺, the interaction between TTE and other solvents, due to the dipole–dipole interaction and/or dispersion force, gradually increases, which is cross-validated by the chemical shift of ¹H nuclei in TTE and FEC (Fig. 3I and SI Appendix, Figs. S7 and S8) (69–71). Accordingly, the solvating power of AN2-DME is lower than that of DME and AN, resulting in the weak dissociation of LiPF₆ in AN2-DME-based electrolyte. Despite that, the AN2-DME-based electrolyte supports a high ionic conductivity of 3.14 mS cm⁻¹ (SI Appendix, Fig. S9), sufficiently guaranteeing a stable cycle for LMB.

Fig. 3.

Physical properties of electrolytes. Raman (A and B) and FT-IR (C and D) spectra of AN2-DME-, DME-, and AN-based electrolyte. The chemical shift of NMR spectra in various electrolytes: ³¹P (E) and ¹⁹F (F) of PF₆^-, ¹⁹F of FEC (G) and TTE (H), and ¹H of TTE and FEC (I).

The electrochemical performance of LMA was investigated using Li/Cu half cells. An average CE of 98.4% is achieved within 100 cycles in the AN2-DME-based electrolyte, while DME- and AN-based electrolytes hardly endure a stable cycle (Fig. 4A). The typical voltage profiles of LMA in AN2-DME-based electrolyte exhibit a low overpotential of ~40 mV with a marginal increase over 100 cycles (Fig. 4B), whereas the overpotentials in other electrolytes unceasingly climb with unstable cycling (SI Appendix, Figs. S10 and S11), probably due to the continuous crack/re-construction of SEI and the soft short circuit caused by “dead lithium” (72, 73).

Fig. 4.

Electrochemical performance and characterizations of LMA. (A) Coulombic efficiency of Li deposition/stripping at 0.5 mA cm⁻² for 1 mAh cm⁻². (B) Corresponding voltage profiles in AN2-DME-based electrolyte. Inset is the top view of deposited Li morphology. (Scale bar, 5 μm.) (C) Li metal thickness calculated from FIB-SEM. Inset is the cross-section view of deposited Li in AN2-DME-based electrolyte. (Scale bar, 10 μm.) (D) TOF-SIMS depth profiles of chemical fragments for LMA in AN2-DME-based electrolyte. XPS spectra of F 1s (E), N 1s (F), O 1s (G), and C 1s (H) for deposited Li in AN2-DME-based electrolyte with various sputtering time: 0, 300, 600, 900, and 1,200 s. Nyquist spectra of Li/Li systematical cells after 3 (I) and 50 (J) cycles in three electrolytes and corresponding resistance (K).

The electrochemical behaviors of LMA are primarily determined by its interfacial chemistry. Different from the highly porous and whisker-like Li in AN-based electrolytes (SI Appendix, Fig. S12B), Li deposits in AN2-DME- and DME-based electrolytes are notably close-packed (Inset of Fig. 4B and SI Appendix, Fig. S12A), in which compact Li deposits form columnar structures with the thickness of 21.22 (Inset of Fig. 4C) and 16.78 μm (SI Appendix, Fig. S13A), respectively. The slightly higher thickness of Li deposits formed in AN2-DME-based electrolyte might be due to the lower LUMO of AN2-DME solvent and its higher propensity to be reduced by Li metal than that of DME. Resultantly, the smaller Li deposits formed in AN2-DME-based electrolyte will lead to the higher thickness. In contrast, the deposited Li in AN-based electrolyte is highly permeable with a thickness of 51.34 μm and a clear crack between Li deposits and Cu substrate (SI Appendix, Fig. S13B). The compact Li deposits in AN2-DME- and DME-based electrolytes will diminish the surface exposure to liquid electrolyte and alleviate the side reactions in favor of high CE and stable cycle of LMA. Immersed in three solvents for 7 d, Li disks cause AN solvent turning brown while the color of AN2-DME solvent does not distinctly change (SI Appendix, Fig. S14). Additionally, the morphology of Li disks in AN2-DME solvent maintains well with Li disks apparently corroded by DME and AN solvent (SI Appendix, Fig. S15), indicating that the SEI formed in AN2-DME solvent can well protect Li from side reactions.

The SEI chemistry of LMA was unraveled using time-of-flight secondary ion mass spectrometry (TOF-SIMS) (Fig. 4D and SI Appendix, Fig. S16) and X-ray photoelectron spectroscopy (XPS) (Fig. 4 E–H and SI Appendix, Figs. S17–S20). TOF-SIMS depth profiles of representative fragments exhibit that the intensity of F- and O- dominantly outcompete that of C- referring to the organic species (Fig. 4D). The signals of N-, CH₂N₂-, and CH₃NO- from the reduction of AN2-DME solvents are readily observed. Such a SEI structure in AN2-DME-based electrolyte can be clearly visualized by 3D reconstruction (SI Appendix, Fig. S16). According to XPS results, the organic decomposition products (RO–Li, C–H, and O–C=O) of the AN2-DME-based electrolyte indicated by C 1s (Fig. 4H) and O 1s signals (Fig. 4G) considerably diminish after sputtering, whereas LiF signals remain prominent over sputtering (Fig. 4E) (74–76). Even after 1,200 s sputtering, N1s signals persist (Fig. 4F), revealing the important role of AN2-DME reduction in constructing SEI. No N1s signal can be detected in DME-based electrolyte (SI Appendix, Fig. S18A), hinting that the N-containing SEI in AN2-DME-based electrolyte comes from the reduction of AN2-DME cyanogen. In contrast in AN-based electrolyte, the C 1s signals indexed to the C–H and Li–C can be readily observed after sputtering (SI Appendix, Fig. S20), indicative of the significant reduction of AN solvent due to its low HOMO.

To elucidate the impact of SEI chemistry on Li deposition/stripping process, electrochemical impedance spectroscopy (EIS) was carried out after 3 and 50 cycles (Fig. 4 I–K). R_s is predominantly associated with electrolyte resistance (77), while R_SEI and R_ct represent the process of Li-ion migration through SEI (78) and the following reduction of a Li-ion to a Li atom by obtaining an electron (17), respectively. Both the R_SEI and R_ct augments over cycling, probably due to the accumulative SEI and electrolyte degradation, respectively (79). The high R_SEI in AN-based electrolyte after 50 cycles indicates low Li⁺ diffusion ability, making it difficult for Li ions to migration through SEI. Consequent Li depletion readily results in the scarcity of Li ions beneath SEI. The insufficient Li ions preferentially tend to gather on the nucleation sites, inducing the highly porous and whisker-like Li deposition in AN-based electrolyte. While the low R_SEI in AN2-DME-based electrolyte, primarily ascribed to the N-containing and LiF-rich SEI chemistry (80), denotes fast Li ion diffusion through high-conductive SEI. The number of Li ions beneath the SEI is considerably enriched even on the edges of Li nucleation sites. Li ion can be uniformly converted to Li atom on Li nucleation sites (81), leading to the close-packed and columnar Li deposition.

Performance and Cathode Chemistry of NCM811.

The electrochemical behaviors of NCM811 cathode were evaluated in various electrolytes within 3.0 to 4.3 V (Fig. 5). A high capacity of 195 mAh g⁻¹ is delivered with the capacity retention of 81.8 % after 200 cycles in AN2-DME-based electrolyte (Fig. 5A). In contrast, only 67.7% and 66.9% capacities maintain in DME and AN-based electrolyte, respectively. The initial CE of AN2-DME-based electrolyte (89.6%) is higher than those of DME- (88.4%) and AN-based electrolyte (68.5%) (Fig. 5B). The CE in AN2-DME-based electrolyte readily surpasses 99% in the second cycle, whereas CEs in DME- and AN-based electrolytes gradually climb to 99% after more than 10 cycles. Worth noting that the average CE in AN2-DME-based electrolyte approaches 99.85% within 200 cycles, remarkably higher than those in DME- (99.5%) and AN-based (99.4%) electrolytes. Besides, AN2-DME-based electrolyte shows more stable voltage profiles (Fig. 5C) than the other two electrolytes (Fig. 5D and SI Appendix, Fig. S21). The galvanostatic intermittent titration technique measurements (GITT) (SI Appendix, Fig. S22) and rate performance of NCM811 cathode (SI Appendix, Fig. S23) in the AN2-DME-based electrolyte exhibit small overpotentials and fast reaction kinetics, indicating the structural integrity of NCM811 cathode and weak side reaction of NCM811 cathode with electrolyte. The NCM811 cathode was further charged to higher voltages than 4.3 V (SI Appendix, Fig. S24), in which the stable cycling and high CE were achieved for 4.4 and 4.5 V, highlighting the compatibility of AN2-DME-based electrolyte with high voltages.

Fig. 5.

Electrochemical performance and cathode chemistry of NCM811. Cycling performance (A) and Coulombic efficiency (CE) (B) of NCM811 cathode at 0.5 C rate (0.2 C for the first three cycles, 1 C = 200 mAh g⁻¹) within 3.0 to 4.3 V. Corresponding voltage profiles of NCM811 cathode for the first 200 cycles in AN2-DME- (C) and DME-based (D) electrolytes. (E) Leakage currents during the 4.3 V constant-voltage floating test of the NMC811 cathodes cycled in various electrolytes for 3 cycles. (F) TM dissolution measured by ICP-MS after 200 cycles. High-resolution TEM, SEM, and FIB-SEM images of NCM811 particles cycled in AN2-DME-(G, H, and I, respectively) and DME-(J, K, and L, respectively) based electrolyte. [Scale bar: 10 nm, 5 μm, and 2 μm for TEM, SEM, and FIB-SEM images, respectively.]

The electrochemical performance of NCM811 cathode is prominently dictated by the cathode interfacial chemistry. Accelerated degradation tests were performed to uncover the side reactions between the cathode and electrolyte by persistently exposing the cathode at 4.3 V (Fig. 5E). The float-test leakage current represents side reaction rates, in which the AN-based electrolyte remains high leakage current of 12.3 μA mg⁻¹ in the 10,000 s hold, probably due to the incompatibility of AN with LMA. In contrast to ~1.24 μA mg⁻¹ in the DME electrolyte, the leakage current in the AN2-DME electrolyte fleetly attenuates to 0.51 μA mg⁻¹, even lower than that in 1M LiPF₆ EC/DMC electrolyte, revealing a pale side reaction and stabilized interphasial chemistry. The cathode degradation is often concomitant with the TMs dissolution (11, 82) which was investigated by inductively coupled plasma mass spectrometry (ICP-MS) measurements. Compared in DME-based electrolyte, the dissolved TMs in AN2-DME electrolyte are notably less after 200 cycles (Fig. 5F), indicating the restrained cathode degradation.

The microstructures of the cycled NCM811 cathode were analyzed by high-resolution TEM (HRTEM) (Fig. 5 G and J). A thin CEI layer of ~2 nm can be observed in the NCM811 cathode cycled in our electrolyte with the well-preserved morphology and no distinct phase transition in the interface (Fig. 5G). By contrast in DME-based electrolyte, the cycled NCM811 cathode is coated by a thick non-uniform layer of >10 nm (Fig. 5J). The well-preserved CEI morphology in AN2-DME-based electrolyte suggests that our electrolyte greatly mitigates the cathode surface degradation over cycling. Besides, the NCM811 secondary particles maintain intact after 200 cycles in our electrolyte (Fig. 5H). No apparent intergranular cracking is observed on the cross-sectioned NCM811 particles (Fig. 5I). In sharp contrast, the microstructural degradation of NCM811 cathode is prominent in DME-based electrolyte with broken NCM811 particles (Fig. 5K) and intergranular cracking (Fig. 5L). The non-uniform CEI might cause heterogeneous lithiation/delithiation of the primary particles, thereby leading to the anisotropic strain changes which will trigger the fragmentation of NCM811 secondary particles and loss of electrical contact between primary particles. Consequently, the high surface area augments the probability of severe parasitic reactions and massive electrolyte consumption.

Performance of Anode-Free Cu|NCM811 Pouch Cells.

To systematically gauge the compatibility of the AN2-DME-based electrolyte with LMA and NCM811 cathode, anode-free Cu/NCM811 pouch cells were assembled with high NCM811 loading of 4.25 mAh cm⁻² and lean electrolyte condition of approximately 3 g (Ah)⁻¹. The two-layer pouch cells with our electrolyte can maintain 76% capacity after 100 cycles (Fig. 6 A and B). However, the pouch cells survive only less than 40 cycles in the DME-based electrolyte with rapid depletion of Li or electrolyte and consequently catastrophic capacity drop (Fig. 6 A and C). Adapting the parameters of practical multilayer pouch cells (SI Appendix, Table S2), the specific energy is estimated to be 397.5 Wh kg⁻¹, which is highly competitive among the reported LMBs (5, 27).

Fig. 6.

Electrochemical performance of anode-free Cu|NCM811 pouch cells. (A) Cycling performance and corresponding voltage profiles using AN2-DME-based electrolyte (B) and DME-based electrolyte (C) at 0.1 C charge and 0.5 C discharge rate within 3.0 to 4.3 V with electrolyte absorbance of 3 g (Ah)⁻¹.

Conclusion

In this work, cyano-substitution was demonstrated to enhance the oxidative stability of ether solvent, in which cyano-group and ether oxygen form the chelating structure with Li⁺ not notably lowering the solvating power of solvents. A stable cycle of LMA is achieved in our AN2-DME-based electrolyte with CE of 98.4% within 100 cycles. Compact and column Li deposits are formed benefiting from the N-containing and LiF-rich SEI facilitating the fast Li⁺ migration. Besides, the AN2-DME-based electrolyte supports a stable long-term cycling of 4.3 V NCM811 cathode with the negligible side reaction. 81.8% capacity remains after 200 cycles with CE of 99.85%, much higher than that in DME-based electrolyte. The stable cycling and high CE in our electrolyte are predominantly due to a uniform CEI sheltering the electrolyte from the violent oxidation decomposition and the strong bond of cyano-group with the TMs attenuating the catalytic reactivity of TM ions toward bulk electrolyte. Resultantly, anode-free pouch cells with the capacity of 125 mAh retain 76% capacity after 100 cycles, corresponding to the energy density of 397.5 Wh kg^-1. Cyano-substitution will provide an instructive methodology instead of fluoro-substitution on precise electrolyte engineering for high-energy lithium metal batteries.

Materials and Methods

Materials.

Anhydrous Ethylene Glycol Bis(propionitrile) Ether (AN2-DME, >97%), Ethylene glycol dimethyl ether (DME, 99.9%), acetonitrile (AN, 99.9%), fluoroethylene carbonate (FEC, 99%), and 1,1,2,2-Tetrafluoroethyl 2,2,2-Trifluoroethyl Ether (TTE, 99.0 %) were purchased from TCI, Alfa Aesar, Acros, Sigma-Aldrich, and TCI, respectively. All solvents were dried with >10% (by weight) activated molecular sieves for at least 24 h before use. Lithium hexafluorophosphate (LiPF₆, 98.0%) was purchased from TCI and used as received. LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) powder was obtained from Shenzhen Kejing Star Technology Company.

Electrolyte Preparation.

All the electrolytes were prepared and stored in an argon-filled glovebox with O₂ and H₂O level <0.1 ppm. LiPF₆ was dissolved and stirred in the solvent of AN2-DME, FEC, and TTE with the volume ratio of 2:2:1 to obtain 1 M LiPF₆ AN2-DME: FEC: TTE electrolyte (denoted as AN2-DME-based electrolyte for short). DME- and AN-based electrolytes were obtained by replacing AN2-DME with DME and AN, respectively. Molality [M, moles of salt in liters of solution (mol L⁻¹)] is used to denote the salt concentration in electrolytes.

Material Characterizations.

The morphologies of samples were investigated by scanning electron microscopy (Hitachi S-4800) and transmission electron microscopy (JEM 2100Plus, JEOL Limited Corporation, Japan). The surface chemistry of cycled electrodes was analyzed by XPS (ESCALAB 250 Xi, Thermo Fisher) and TOF-SIMS (ION-TOF). All binding energies of XPS were corrected using the signal of carbon at 284.8 eV as an internal standard. Cross-sectioned Li deposits were prepared using a focused ion beam (IM-40000, Hitachi) and inspected by SEM. Raman spectra for electrolytes were collected with an NRS-5100 spectrometer (JASCO) using a 532 nm diode-pumped solid-state laser between 4,000 and 100 cm⁻¹. FT-IR spectra were measured using a Nicolet iS50 with a diamond-attenuated total reflectance attachment. ¹H and ¹⁹F-NMR spectra were measured on a Varian Mercury 400-MHz NMR spectrometer. The ²³Na NMR measurements were performed on a Bruker AvanceIII 800-MHz NMR spectrometer. ²³Na NMR was carried out in a 10 mM solution of NaTFSI in the solvents. Note that NaTFSI cannot completely dissolve in the AN2-DME. Then, 0.1 M sodium chloride aqueous solution was used as the reference that was placed in a sealed 1-mm melting point capillary. The capillary was placed in a standard 5-mm NMR tube inserted coaxially into the sample tube. All the cell disassembly was carried out in an argon-filled glovebox with O₂ and H₂O level <0.1 ppm and the electrodes were washed in pure DME three times to remove the electrolyte, and then the drying samples were obtained and moved to the machine with an argon-filled sealing tube as a transfer box. In this process, all samples would not be exposed to air.

Electrochemical Measurements.

NCM811 cathodes were fabricated by mixing NCM811, carbon black (super P), and polyvinylidene fluoride (PVDF) at a weight ratio of 96:2:2 with N-methyl-2-pyrrolidone to form a uniform slurry, which was coated onto Al foil using a doctor blade. Cell assembly was carried out in an Ar-filled glovebox with O₂ and H₂O levels below 0.1 ppm. EIS was carried out on an Autolab PGSTAT302N (Metrohm, Switzerland) over a frequency range of 1 to 10 mHz. The ionic conductivities of the electrolytes are obtained by measured EIS using electrochemical workstation (IM6e Zahner). The rate, GITT, and cycling tests for coin cells and pouch cells were carried out on a Land instrument. For Li stripping/deposition CE, cycling was done by depositing 1 mAh cm⁻² of Li onto the Cu electrode at 0.5 mA cm⁻² followed by stripping to 1 V. The average CE was calculated by dividing the total stripping capacity by the total deposition capacity. The Li|NCM811 cells were cycled at C/2 rate between 3 and 4.3 V after the first three activation cycles at C/5 rate. The anode-free Cu|NCM811 pouch cells were cycled at 0.1 C charge and 0.5 C discharge between 3 and 4.3 V.

Theoretical Simulations.

The molecular geometries for the ground states were optimized by DFT at the B3LYP-D3(BJ)/def2-TZVP level, and then the binding energies and electrostatic potential surfaces of molecules were evaluated at the B3LYP-D3(BJ)/ma-TZVPP level (83–85). All DFT calculations were carried out with Gaussian16 program without any constraints.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.