Bifunctional ligand-bridged electrolyte enables high-capacity and high-energy 600 Wh kg⁻¹ lithium-metal battery pouch cells

Abstract

Lithium metal batteries offer a pathway toward next-generation energy storage due to their high specific energy. However, their implementation remains challenging, particularly under lean electrolyte conditions, where unstable interfaces and inefficient ion transport severely limit cycling stability. Here, we show a bifunctional ligand-bridged electrolyte by integrating ethyl difluoroacetate into a lithium nitrate/triethyl phosphate matrix. Incorporating ethyl difluoroacetate as a dynamic bridging agent drives the formation of interconnected ionic aggregates through synergistic H^δ+ − O^δ− hydrogen bonding with anions and coordination interactions with Li⁺, thereby accelerating interfacial kinetics and suppressing parasitic reactions. The electrolyte further enables weakly coordinating solvents to engage in the solvation sheath via non-classical hydrogen bonding, facilitating the formation of a robust, inorganic-rich solid electrolyte interphase. This design enables the realization of high-capacity (14 Ah) Li‖Ni95 pouch cells, which achieve a specific energy of 606.8 Wh kg⁻¹ while maintaining 92.9% capacity retention after 75 cycles at rates of 0.1 C/0.2 C. These findings demonstrate a viable electrolyte strategy for enabling high-performance lithium metal batteries under practical conditions.

Lithium metal batteries face critical challenges such as low ion transport efficiency under lean electrolyte conditions. Here, authors study a bifunctional ligand-bridged electrolyte containing ethyl difluoroacetate that can form interconnected ionic aggregates through hydrogen bonding and lithium-ion coordination interactions, suppressing parasitic reactions and enhancing interfacial charge transfer, enabling stable cycling of a 600 Wh kg-1 lithium metal 14 Ah pouch cell.

Introduction

The global demand for high-energy-density energy storage solutions is driving the development of lithium metal batteries (LMBs) as one of the most promising next-generation technologies, owing to their high theoretical specific energy exceeding 500 Wh kg⁻¹, significantly surpassing that of current commercial lithium-ion batteries^1–5. However, realizing this potential is contingent upon overcoming fundamental limitations inherent to existing electrolyte systems. Under lean electrolyte conditions (E/C ratio < 3 g Ah⁻¹)³, which are essential for achieving high specific energy, conventional electrolytes struggle to maintain stable cycling performance. Moreover, there exists an intrinsic incompatibility between the chemical stability requirements of high-voltage positive electrodes and lithium metal negative electrodes^6–8. Specifically, carbonate-based electrolytes exhibit good oxidative stability at high voltages but exacerbate dendritic lithium growth^9,10, whereas ether-based electrolytes offer better compatibility with lithium metal yet cannot withstand high-voltage operation^11,12. These challenges ultimately converge at the failure of the solid electrolyte interphase (SEI). Whether due to excessive electrolyte consumption that impairs interfacial self-repair, or persistent side reactions that destabilize interfacial chemistry, the result is a significant compromise in practical battery performance. Therefore, the development of an alternative electrolyte system capable of simultaneously stabilizing both high-voltage positive electrodes and lithium metal negative electrodes under lean electrolyte conditions represents a critical step toward the commercial realization of high-energy-density lithium metal batteries.

To address this critical challenge, localized high-concentration electrolytes (LHCEs) have demonstrated transformative potential due to their unique solvation structures^13–17. With a wide electrochemical stability window (>4.5 V vs Li⁺/Li)^18,19, LHCEs can simultaneously stabilize high-voltage positive electrodes and lithium metal negative electrodes. However, a key limitation of conventional LHCEs lies in the introduction of highly polar diluents to achieve localized high concentration, which leads to overly compact Li⁺ solvation shells. This dense solvation structure hinders interfacial kinetics and induces continuous side reactions at the electrode/electrolyte interface. Extending the concept of LHCEs, Jiao and co-workers engineered a compact ion pair aggregate electrolyte via solvent design, which enhances anion coordination to accelerate interfacial charge transfer, promote stable SEI formation, and enable long-term cycling in pouch cells with energy densities exceeding 500 Wh kg⁻¹³. Nevertheless, the anion-dominated SEI formation mechanism continues to consume lithium salts during cycling, causing the collapse of concentration gradients and ultimately compromising the intrinsic stability of the high-concentration electrolyte. In a different approach, Fan et al. redefined solvation design principles by leveraging intermolecular interactions to activate non-coordinating solvents, constructing a dynamic Li⁺-solvent coordination network²⁰. This enabled non-coordinating solvents to directly participate in SEI formation, achieving a high full-cell Coulombic efficiency (CE) of 99.6% while preserving salt concentration. However, the system still faces inherent limitations: a large proportion of free, non-coordinating solvents creates “kinetically isolated zones” between ion clusters, where Li⁺ transport must repeatedly overcome coordination energy barriers. Current electrolyte design paradigms primarily focus on optimizing isolated solvation clusters, while overlooking the regulatory potential of inter-cluster interactions. Therefore, shifting the design perspective from individual solvation motifs to multi-cluster cooperative networks and establishing cross-scale solvation structures is essential for overcoming interfacial kinetic bottlenecks and addressing instability issues in high-energy lithium metal batteries.

Here, we report a multi-cluster coordination approach through the design of a bifunctional ligand-bridged electrolyte (BLBE) system via molecular interfacial engineering. In contrast to traditional LHCEs characterized by isolated ion clusters, our system introduces ethyl difluoroacetate (EDFA) as a molecular bridge to construct an interconnected solvation network across multiple ion aggregates (AGGs). Specifically, the high-donor-number solvent triethyl phosphate (TEP) solvates lithium nitrate (LiNO₃) to form contact ion pairs (CIPs) and AGGs^1,2,21, while the EDFA ligand with its bifunctional structure anchors ${\mathrm{NO}}_3^{-}$ anions through C–H···O hydrogen bonding (via the −CHF₂ group) and simultaneously coordinates to adjacent Li⁺ ions via its carbonyl oxygen. This molecular bridging reconstructs the solvation structure at multiple length scales, resulting in two major advantages. First, the interconnected solvation network significantly enhances ion transport kinetics by eliminating the coordination energy traps commonly found between isolated clusters in LHCEs, reducing the Li⁺ migration barrier. Second, the hydrogen-bond stabilization of ${\mathrm{NO}}_3^{-}$ anions suppresses oxidative decomposition at high voltage, extending the oxidative stability of the electrolyte to 4.5 V vs Li⁺/Li. In addition, component-level synergy is achieved by coupling EDFA with fluoroethylene carbonate (FEC), which preferentially decomposes on the lithium surface to form LiF nanocrystals. Meanwhile, the EDFA-regulated solvation environment promotes the formation of a dense, inorganic-rich SEI, effectively mitigating lithium dendrite penetration. Benefiting from these synergistic mechanisms, the BLBE system exhibits highly stable electrochemical performance in both pouch-cell and large-format configurations. In particular, it enables extended cycling in Li‖Ni95 pouch cells and enables durable operation of 14 Ah high-energy cells. Collectively, this study validates the multi-cluster cooperative solvation concept and demonstrates its practical scalability, highlighting a scalable electrolyte design paradigm for next-generation high-energy lithium metal batteries.

Results

Insights into electrolyte solvation structures

In the BLBE system (Fig. 1a), the bifunctional molecule EDFA serves as a key regulator of the solvation environment by bridging adjacent ion clusters. The −CHF₂ group of EDFA forms directional hydrogen bonds with ${\mathrm{NO}}_3^{-}$ anions, while the carbonyl group weakly coordinates with neighboring Li⁺ ions. This dual interaction promotes the aggregation of CIPs and AGGs, resulting in a cross-cluster solvation network that enhances ionic connectivity and reduces coordination energy barriers. Furthermore, the formation of these enlarged AGGs accelerates interfacial reduction kinetics at the lithium surface, promoting the rapid establishment of an inorganic-rich, mechanically robust SEI layer³. To elucidate the nature of these intermolecular interactions, density functional theory (DFT) calculations were performed to assess the free energy change (ΔG) associated with hydrogen bond formation between various solvent and anion pairs (Fig. 1b and Supplementary data 1). Among all combinations tested, only EDFA and ${\mathrm{NO}}_3^{-}$ exhibited a negative ΔG, indicating spontaneous hydrogen bond formation. This behavior is attributed to the strong electron-withdrawing effect of the −CHF₂ group, which enhances the acidity of the hydrogen atom and enables the formation of a stable H^δ+···O^δ− interaction with ${\mathrm{NO}}_3^{-}$. Furthermore, the presence of multiple electron-rich oxygen sites on ${\mathrm{NO}}_3^{-}$ allows it to simultaneously interact with several EDFA molecules, reinforcing its incorporation into the solvation structure. The existence of these hydrogen bonds was corroborated by spectroscopic analysis. In the ¹H NMR spectrum (Fig. 1c), the proton signal of EDFA in the BLBE system exhibited a notable downfield shift compared with pure EDFA, indicative of reduced electron shielding caused by hydrogen bond formation. This interpretation is further supported by FTIR spectroscopy (Fig. 1d), where the vibrational peaks of EDFA showed a red shift, consistent with electron cloud delocalization associated with hydrogen bonding.

Fig. 1. Schematic of the BLBE and analysis of solvation structures.

a The mechanistic schematic diagram of BLBE. b Free energy of H-bond formation between electrolyte species from DFT calculations. Color code: white ball, H; gray ball, C; red ball, O; light blue ball, N; purple ball, Li; cyan ball, F. c ¹H-NMR of various solvents or electrolytes. d FTIR spectra of different electrolytes and solvents.

Raman spectroscopy was employed to analyze the different electrolytes (Fig. 2a and Supplementary Fig. 1). Figure 2a shows the Raman spectra in the range of 1000−1075 cm⁻¹, revealing that in low-concentration electrolytes (LCEs), nitrate ions predominantly exist as CIPs (52.9%) due to their inherently strong solvation ability. With the increase of LiNO₃ concentration, AGGs emerge in the solvation structure of the electrolyte, promoting the decomposition of nitrate ions and the formation of the solid electrolyte interface (SEI). With the addition of FEC, the solvation structure of NTF does not change significantly compared with HCE (3 M LiNO₃ in TEP), and the beneficial solvation structure remains preserved. Notably, the addition of EDFA leads to a significant increase in the number of AGGs. For this phenomenon, the typical solvation structure obtained from molecular dynamics (MD) simulations shows that EDFA molecules form hydrogen bonds with solvated ${\mathrm{NO}}_3^{-}$ (Supplementary Figs. 2−4), while simultaneously coordinating with Li⁺ through carbonyl oxygen, bridging two ion clusters, and forming more abundant AGGs. The more abundant AGGs cause an increase in the electron cloud density around Li⁺, which is reflected in the high-field shift of ⁷Li in NMR (Supplementary Fig. 5).

Fig. 2. Solvation structures of electrolytes and their electrochemical behaviors with electrodes.

a Raman spectrum of different concentrations of electrolytes and solvents. RDF results of NTF (b), and BLBE (c). d Binding energy of per Li⁺ in various electrolytes. e Reduction energy of one Li⁺ in various electrolytes. f HOMO/LUMO of different solvation structures in electrolytes. g Reaction pathway of EDFA. Color code: white ball, H; gray ball, C; red ball, O; light blue ball, N; purple ball, Li; cyan ball, F.

Radial distribution functions (RDFs) of various electrolyte systems were obtained from MD (Fig. 2b, c and Supplementary Figs. 6 and 7). For conventional commercial electrolytes (CE), the coordination degree of ${\mathrm{PF}}_6^{-}$ anions is low, with an average of only 0.2 anions per Li⁺ (Supplementary Fig. 8). This observation suggests that Li⁺ ions are predominantly solvated by solvent molecules, resulting in a solvation structure dominated by solvent-separated ion pairs (SSIPs). During battery discharge, a solvation structure dominated by SSIPs leads to solvent-driven interfacial reactions, rendering solvent molecules more susceptible to reduction. This results in continuous solvent decomposition and excessive electrolyte consumption. In contrast, the solvation structure in NTF exhibits fewer solvent coordination sites, yielding an average Li⁺ solvation environment of Li⁺TEP_1.30FEC_0.80(${\mathrm{NO}}_3^{-}$)_1.57. This composition is characterized by a higher proportion of CIP and AGG. The average solvation structure of BLBE is Li⁺TEP_0.68FEC_0.59EDFA_0.79(${\mathrm{NO}}_3^{-}$)_1.81. Although EDFA alone does not significantly coordinate with Li⁺ in conventional LiNO₃ solutions, it exhibits strong solvation capability in BLBE due to its bifunctional coordination behavior and hydrogen bonding interactions, becoming second only to ${\mathrm{NO}}_3^{-}$ in coordination strength. The behavior is attributed to the relatively high binding energy of partially fluorinated EDFA with Li⁺ (Fig. 2d) and the further enhancement of the solvation capability of EDFA due to atypical hydrogen bonding. Meanwhile, as a high-dielectric-constant additive, FEC partially participates in the solvation structure, working synergistically with EDFA and ${\mathrm{NO}}_3^{-}$ to facilitate the formation of a stable SEI. This cooperative effect enhances CE and enables long-term cycling stability in lithium metal batteries (Supplementary Figs. 9 and 10). This behavior overcomes the limitation of conventional LHCE diluents, which typically do not dissolve lithium salts and therefore cannot participate in the solvation structure.

To clarify the formation mechanism of the SEI, the molecular orbital energy levels of various anions, solvents, and ion pairs were further investigated by DFT (Supplementary Figs. 11 and 12). LiNO₃ exhibits the lowest unoccupied molecular orbital (LUMO) energy among all species, making it the most readily reduced. For the solvents, the LUMO energies follow the order EDFA < FEC < TEP, suggesting that EDFA and FEC preferentially decompose before TEP in the BLBE system. To more accurately characterize the solvation environment, the HOMO/LUMO energy levels of solvated complexes were further calculated. The results reveal that the lowest LUMO remains predominantly localized on BLBE-derived complexes (Fig. 2f), consistent with the reduction sequence inferred from the isolated species, indicating that the solvation environment does not alter the primary reduction pathway. It should be noted that HOMO/LUMO energy levels provide only approximate descriptors of electrolyte redox behavior, as they do not account for decomposition processes such as ring-opening and defluorination²². This decomposition pathway effectively suppresses the unfavorable reduction of TEP at the lithium metal surface²³, thereby enhancing electrolyte stability. To further probe the reduction behavior, the reduction energy (E_red) of the Li⁺–UNIT complexes was computed as the energy difference between the neutral and one-electron-reduced states (Fig. 2e). Among all species, the EDFA−Li⁺ ion pair exhibits the strongest electron-accepting capability, which can be attributed to its ability to form a stable five-membered chelate structure with the lithium ion. Upon electron uptake, the F atom in the −CHF₂ group tends to dissociate from the C atom, thereby forming LiF and subsequently detaching from the complex (Fig. 2g and Supplementary Figs. 13–15), which demonstrates that EDFA can serve as an effective LiF precursor to optimize the SEI composition.

Characterization and analysis of the lithium metal negative electrode interface

The CE of Li‖Cu cells measured using the Aurbach method²⁴ (Fig. 3a and Supplementary Fig. 16) demonstrates that BLBE exhibited a significantly higher CE of 99.31% compared to conventional CE and NTF electrolytes containing only FEC as the film-forming solvent. Furthermore, Li‖Cu coin cells utilizing BLBE achieved a cycle life exceeding 300 cycles with an average CE of 98.85% (Supplementary Fig. 17), clearly indicating the enhanced reversibility of lithium metal deposition/stripping in BLBE. Cyclic voltammetry (CV) tests conducted within the potential range of −0.3 V to −0.6 V (Supplementary Fig. 18) reveal that BLBE possessed the highest response current and the lowest nucleation overpotential (36 and 6 mV lower than those of CE and NTF, respectively), reflecting a marked enhancement in the kinetics of lithium deposition reactions. In Li‖Li symmetric cell tests (Fig. 3b), BLBE exhibits a cycle life of up to 1500 h with a polarization overpotential of only 44 mV. By contrast, NTF failed rapidly owing to excessive polarization caused by high viscosity, while CE failed within 400 h due to poor compatibility with lithium metal. Electrochemical impedance spectroscopy (EIS) tests (Supplementary Fig. 19 and Supplementary Table 1–3) show that the interfacial impedance of BLBE gradually decreased during cycling and remained consistently lower than that of CE and NTF at the comparable cycle stage. These results confirm that BLBE promotes the formation of a robust SEI and effectively suppresses electrolyte degradation induced by lithium dendrites. Rate performance tests (Supplementary Fig. 20) demonstrate that BLBE exhibits the highest rate capability, which can be attributed to its relatively high lithium-ion transference number (0.644, Supplementary Fig. 21 and Supplementary Table 4) and optimized charge transfer kinetics (Supplementary Fig. 22). Specifically, the higher lithium-ion transference number effectively suppressed ion concentration polarization at high current densities, while the enhanced charge transfer kinetics further facilitated rapid electrochemical reactions. To investigate the differences in lithium deposition behavior among various electrolytes, the morphologies of lithium deposition at a capacity of 1 mAh cm⁻² were examined via scanning electron microscopy (SEM). CE results in significant lithium dendrite growth (Fig. 3c), NTF forms a loose and porous deposition layer (Fig. 3e), whereas BLBE produced a dense, dendrite-free lithium deposition layer with a thickness of only 5.16 μm (Fig. 3g and Supplementary Fig. 23). Even after 100 cycles, the lithium deposition morphology in BLBE remains smooth (Supplementary Fig. 24). Transmission electron microscopy (TEM) analysis (Supplementary Fig. 25) further reveals that lithium particles formed in BLBE exhibit a larger diameter (251.3 nm) and more uniform distribution compared with those in NTF (81.8 nm). Cryo-TEM characterization demonstrates that the SEI formed in CE (Fig. 3d) and NTF (Fig. 3f) is thick and uneven, whereas the SEI generated in BLBE is only 11.14 nm thick and uniform (Fig. 3h), enabling effective regulation of uniform lithium ion deposition.

Fig. 3. Characterization of the lithium metal negative electrode interface.

a Plating/stripping profiles of Li‖Li symmetric cells tested at a current density of 1 mA cm⁻² with a capacity of 1 mAh cm⁻². b Long-term electrochemical stability of Li‖Li symmetric cells under Li plating/stripping at a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻². c, e, g SEM images of the initial lithium plating on Cu foil at 0.5 mA cm⁻² and 1 mAh cm⁻² for CE (c), NTF (e), and BLBE (g) electrolytes. d, f, h Cryo-TEM images of lithium plating on lacey carbon grids after 60 min at a current density of 0.5 mA cm⁻² for CE (d), NTF (f), and BLBE (h) electrolytes. i, j Sputtering component maps and 3D depth profile images of LiF⁻₂(i) and C₂H⁻ (j) obtained by TOF-SIMS for different electrolytes. For TOF-SIMS, a Li‖Cu coin cell was subjected to galvanostatic cycling at a current density of 1 mA cm⁻². The cell was discharged for 3 h and subsequently charged to a cutoff voltage of 1 V at the same current density for five cycles.

X-ray photoelectron spectroscopy (XPS) depth profiling was employed to further investigate the composition and structure of the SEI (Supplementary Figs. 26–30). The SEI formed in BLBE primarily consists of LiF (with a peak proportion of 37.36%) and Li₃N, with their contents remaining consistently high across all depths (Supplementary Fig. 31). Highly conductive nitrogen-containing species such as Li₃N/LiN_xO derived from the decomposition of LiNO₃ enhanced interfacial conductivity. Time-of-flight secondary ion mass spectrometry (TOF-SIMS) analysis (Fig. 3i−j and Supplementary Figs. 32−36) confirms that the SEI thickness in BLBE is approximately 24 nm, consistent with TEM observations. The LiF⁻ signal remained strong throughout the 0−28 s sputtering period, and the 3D depth profile revealed a more uniform distribution of LiF/Li₃N. Notably, almost no P⁻ signal was detected in BLBE (Supplementary Fig. 36)²⁵, suggesting that the unfavorable TEP decomposition on the lithium metal surface was effectively suppressed due to the preferential reductive decomposition of FEC, EDFA, and ${\mathrm{NO}}_3^{-}$ upon their incorporation into the solvation structure. Collectively, these findings confirm that BLBE can construct a stable SEI with both robustness strength and ionic conductivity, providing a critical basis for enhancing the performance of lithium metal batteries.

Characterization and analysis of the NCM811 positive electrode

The electrochemical stability windows of various electrolytes were evaluated by linear sweep voltammetry (LSV). The decomposition voltage exceeding 4.7 V for BLBE indicates its improved stability at high voltages (Supplementary Fig. 37), which is essential for compatibility between the electrolyte and high-voltage positive electrodes. To further investigate electrolyte compatibility with high-voltage positive electrodes, in situ EIS tests were performed on Li‖NCM811 cells during their initial charge-discharge cycle within the voltage range of 3.0−4.4 V (Supplementary Fig. 38). Compared with other electrolytes, BLBE exhibits consistently lower impedance throughout the entire charge-discharge process. This behavior can be attributed to its good wettability toward battery components (Supplementary Fig. 39 and Supplementary Table 5) and the formation of low-impedance SEI and CEI layers at both electrode interfaces during cycling. To more intuitively observe the CEI, the surfaces of NCM811 positive electrodes were examined after long-term cycling in various electrolytes using SEM (Fig. 4a−c). The results demonstrate that CE is incapable of withstanding a high cutoff voltage of 4.4 V, leading to significant decomposition and the formation of thick, uneven organic layers on NCM811 particles. Similarly, the CEI formed in NTF also exhibits an irregular morphology. In contrast, the CEI generated in BLBE is thin and uniform, effectively preserving the structural integrity of NCM811 particles with clearly defined edges. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) analysis (Fig. 4d−f) reveals that the CEI formed in BLBE is only 2.33 nm thick and uniformly coats the NCM811 surface. This uniform coverage significantly reduces interfacial impedance and facilitates stable lithium-ion intercalation/deintercalation. To investigate the specific composition of the CEI, XPS analyses were conducted (Supplementary Figs. 40−43). Spectral analysis reveals that BLBE promotes the formation of a LiF-rich interfacial layer while simultaneously regulating inorganic species derived from the solvation structure. This LiF-rich interface exhibits robust mechanical reinforcement, effectively suppressing surface reconstruction of NCM811 materials during long-term cycling and significantly mitigating structural degradation. To investigate the impact of different electrolytes on the evolution of the internal crystal structure of the NCM811 positive electrode during cycling, in situ X-ray diffraction (XRD) was employed to monitor the first charge-discharge cycle of Li‖NCM811 (Fig. 4g−i). The data reveal that the transition from H1 phase to the H2 phase causes the 003_H2 diffraction peak to shift to lower 2θ angles, indicating lattice expansion along the c-axis within the layered structure of the Ni-rich NCM811 positive electrode material²⁶. During the entire charge-discharge process, the change in 2θ angle for the CE sample exceeded 2°, indicating substantial structural evolution that may lead to damage to the internal structure of the positive electrode. In contrast, the 2θ angle changes for the NTF and BLBE samples were only 1.12° and 1.09°, respectively, suggesting that the CEI formed in NTF and BLBE effectively mitigates structural degradation of the NCM811 positive electrode during cycling²⁷. To further substantiate the stabilizing effect of BLBE on the NCM811 positive electrode, transition metals (TM) dissolution was quantified by using inductively coupled plasma mass spectrometry (ICP-MS) (Supplementary Fig. 44). Compared with CE, both NTF and BLBE significantly suppress TM dissolution in NCM811, consistent with the enhanced structural integrity of NCM811 in these electrolytes. The lowest dissolution level observed in BLBE confirms that the robust CEI formed therein provides effective protection against lattice degradation and metal leaching.

Fig. 4. Characterization of the NCM811 positive electrode.

SEM images of NCM811 positive electrode particles after 100 cycles between 3.0 and 4.4 V at 1 C/1 C with a CE, b NTF, and c BLBE. TEM images of NCM811 positive electrode particles after 100 cycles between 3.0 and 4.4 V at 1 C/1 C with d CE, e NTF, and f BLBE. In situ XRD patterns of the first charge-discharge cycle of NCM811 in a lithium metal battery between 3.0 and 4.4 V at 0.1 C/0.1 C with g CE, h NTF, and i BLBE.

Electrochemical performance of LMB

The electrochemical performance of the electrolytes was initially evaluated using Li‖NCM811 coin cells. At a high cutoff voltage range of 3.0 − 4.4 V and a rate of 1 C, BLBE achieved a capacity retention rate of 94.4% (Supplementary Figs. 45 and 46). Supplementary Fig. 47 presents the rate performance of different electrolytes in full cells. Benefiting from rapid interfacial kinetics and the formation of a high-performance SEI, BLBE exhibits the highest capacity across the rate range of 0.1–4 C (1 C = 200 mA g⁻¹), underscoring its high-rate performance. To further evaluate electrolyte performance under lean electrolyte conditions, Li‖NCM811 coin cells were assembled using high-mass-loading NCM811 positive electrodes (13 mg cm⁻²) and tested for long-term cycling at a low electrolyte-to-capacity ratio (E/C = 1.61 g Ah⁻¹). Cells using CE and NTF electrolytes exhibited rapid capacity decay to nearly zero within fewer than 100 cycles, whereas the BLBE system maintained 90.3% of its initial capacity after 150 cycles, demonstrating its potential for practical application (Supplementary Fig. 48). The applicability of BLBE was further validated in pouch cells, Li‖NCM811 pouch cells with a specific energy of 502.5 Wh kg⁻¹ showed robust cycling stability in BLBE, retaining 90.5 % capacity retention after 120 cycles (Fig. 5a, Supplementary Fig. 49, and Supplementary Table 7). In contrast, neither CE nor NTF sustained stable cycling beyond 100 cycles, and even LHCE electrolytes in Li‖NCM811 pouch cells achieved only 77.4% capacity retention after 120 cycles, confirming the clear performance advantage of BLBE.

Fig. 5. Electrochemical performance characterization of LMB.

a Cycle performance of lithium metal pouch cells with a specific energy of 502.5 Wh kg⁻¹. b Cycle performance and charge-discharge curves of lithium metal pouch cells with a specific energy of 606.8 Wh kg⁻¹, operating within a voltage range of 3.0−4.4 V. c Comparison of specific energy and capacity between this work and recently reported high-energy-density pouch cells (Supplementary Table 6).

To further probe its practical viability, BLBE was applied to 14.02 Ah Li‖Ni95 pouch cells (Supplementary Table 8). Under a 4.4 V cutoff voltage, the cell achieved a high specific energy of 606.8 Wh kg⁻¹ while retaining 92.9% of its initial capacity after 75 cycles (Fig. 5b and Supplementary Fig. 50). This specific energy rivals the current state of the art works reporting large-format lithium metal pouch cells (Fig. 5c and Supplementary Table 6)^28–42. Moreover, BLBE enabled stable cycling (>30 cycles) in 13.5 Ah Li‖Ni90 pouch cells at 601.7 Wh kg⁻¹ under an even higher cutoff voltage of 4.5 V (Supplementary Figs. 51 and 52 and Supplementary Table 9). These results demonstrate that BLBE is competitive with both conventional and advanced LHCE systems, highlighting its strong potential for practical deployment in next-generation high-energy lithium metal batteries.

Discussion

In this work, we report a BLBE system that achieves favorable electrochemical performance in lithium metal pouch cells under lean electrolyte conditions. Central to this design is the introduction of EDFA, a dual-functional molecular bridge that simultaneously forms hydrogen bonds with solvated ${\mathrm{NO}}_3^{-}$ anions and coordinates with Li⁺ cations. This cooperative interaction drives the formation of interconnected ionic AGGs, transforming the solvation structure from isolated clusters into a cross-linked ionic network. Such a tailored architecture not only accelerates interfacial charge transfer by eliminating kinetic bottlenecks but also activates weakly coordinating solvents to participate in electrochemical reactions. As a result, the electrolyte promotes the in situ formation of a robust, dual-phase SEI composed of LiF and Li₃N, effectively suppressing parasitic reactions and dendrite growth. Furthermore, the localized high-concentration effect extends the oxidation stability beyond 4.4 V and effectively mitigates solvent depletion, enabling good compatibility with high-voltage positive electrodes. When deployed in Li‖NCM811 pouch cells (502.5 Wh kg⁻¹), the BLBE system delivers stable cycling with 90.5% capacity retention over 120 cycles. More significantly, we demonstrate its scalability and industrial relevance through the initial application in large-capacity (14 Ah) Li‖Ni95 pouch cells, achieving a specific energy of 606.8 Wh kg⁻¹. These large-format cells sustain 75 cycles at 4.4 V cutoff, retaining 92.9% of their initial capacity. This work establishes an alternative paradigm for electrolyte design—leveraging multi-cluster coordination and molecular-level interfacial engineering—to overcome the long-standing trade-offs between stability, reactivity, and scalability in lithium metal batteries.

Methods

Theoretical calculations

Quantum chemical calculations were performed by the Dmol3 module in the Materials Studio package. The B3LYP functional was employed with customized Grimme DFT-D parameters and the DNP 4.4 basis set to ensure simulation accuracy. All molecules used the ESP charges after ultra-fine optimization^43,44. The k-points were set to the Gamma point (1 × 1 × 1) and the convergence tolerances were set to 1.0 × 10⁻⁵ Ha for energy, 2.0 × 10⁻³ Ha/Å ⁻¹ for maximum force, and 5.0 × 10⁻³ Å for maximum displacement.

The solvation energy was calculated by the following formula:

$$E_{sol}=E_{total}-E_1-E_2$$ 1

where E_total is the energy of the optimized Li⁺–solvent complex, and E₁ and E_2, correspond to the energies of the isolated Li⁺ ion and solvent molecule optimized in their respective monomer geometries. Only the lowest-energy conformers were considered.

The reduction energy (E_red) of the Li⁺–UNIT complex was evaluated to characterize the stability of the reduced species upon electron attachment. The optimized Li⁺–UNIT complex was used as the initial structure, and one electron was added to simulate the reduction process:

$${Li}^{+}-UNIT+e^{-}\to Li-UNIT\left(reduced\right)$$ 2

The reduced complex was then fully reoptimized at the same level of theory. The reduction energy was obtained as the total energy difference between the reduced and neutral complexes:

$$E_{red}=E_{Li-UNIT\left(reduced\right)}-E_{{Li}^{+}-UNIT}$$ 3

A negative E_red indicates that electron attachment stabilizes the complex, implying a higher tendency for reduction. These energies were used to compare the relative reduction stability of different solvent– or anion–Li⁺ coordination structures.

Three solution models were developed to simulate the different electrolytes. The CE box contained a mole ratio of 20:100:80:60 for LiPF₆, EC, DMC, and EMC, respectively. The NTF box contained a mole ratio of 40:80:120 for LiNO₃, TEP, and FEC, respectively. The BLBE box contained a mole ratio of 40:80:120:120 for LiNO₃, TEP, FEC, and EDFA, respectively. All MD simulations were performed using the Forcite module with the COMPASSIII force field, and ion charges were scaled by 0.7 following similar previous studies^45–48. The simulation was performed with a time step of 1 fs. Prior to the production simulations, the solution models were equilibrated in the NPT ensemble. A Berendsen barostat was employed to maintain a constant pressure of 0.1 MPa with a relaxation time of 0.1 ps, and the equilibration was conducted for 20 ps. Simultaneously, a Nosé thermostat was used to maintain the temperature at 298 K. Following equilibration, production runs were carried out in the NVT ensemble for 1000 ps, using the final configuration from the NPT equilibration as the initial structure. The duration of the simulations was adequate to guarantee that the solution system reached equilibrium. Calculation of electrolytes’ conductivity: In our simulations, the ionic conductivity was calculated using the Nernst-Einstein equation based on the diffusion coefficients obtained from the MSD of the ions:

$$\sigma =\frac{n{q}^{2}D}{V{k}_{\mathrm{B}}T}$$ 4

where $\sigma$ is the ionic conductivity, $q$ is the elementary charge, $V$ is the system volume, $k_B$ is the Boltzmann constant, $T$ is the temperature, $n$ is the number of charge carriers, and $D$ is the diffusion coefficient of ion species. The Nernst-Einstein equation does not account for correlated ion motions, which can lead to an overestimation of ionic conductivity. However, the Nernst-Einstein approach provides a reasonable first-order approximation of ionic mobility and is widely used in electrolyte simulation studies for comparative analysis.

Electrolyte preparation

LiNO₃ (metal basis, 99.99%), TEP (AR, ≥99.5%), EFA (99.5%), FEC (>99%), and dimethyl carbonate (DMC, ≥99.9%) were purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd. All solvents were dehydrated and purified using 4 Å molecular sieves before use. All reagents were stored in an argon-filled glove box (O₂ < 0.1 ppm, H₂O < 0.1 ppm). The commercial electrolyte, denoted as CE, consisted of 1 M LiPF₆ dissolved in a mixture of ethylene carbonate (EC), DMC, and ethyl methyl carbonate (EMC) in a 1:1:1 volume ratio, and was obtained from Dodo Chem. Electrolytes containing 1 M and 3 M LiNO₃ in TEP were designated as LCE and HCE, respectively. An electrolyte comprising 1.8 M LiNO₃ in TEP/FEC (2:1, v/v) was referred to as NTF. Another formulation containing 1.1 M LiNO₃ in TEP/FEC/EDFA (2:1:2, v/v/v) was denoted as BLBE.

Battery assembly

Except for lithium metal pouch cells, all Li‖Cu (Cu foil diameter: 12 mm), Li‖Li, and Li‖NCM811 cells were assembled into CR2032-type coin cells. The battery cases, spacers, and springs used for coin cell assembly were made of 304 stainless steel and purchased from Guangdong Canrd New Energy Technology Co., Ltd. A 9 μm double‑smooth copper foil with a purity of ≥99.95% (Guangdong Canrd New Energy Technology Co., Ltd.) was used as the current collector. Lithium metal foil with a thickness of 300 μm, supplied by Zhejiang Funlithium New Energy Technology Co., Ltd., was used for coin cell assembly. Unless otherwise specified, the electrolyte volume for each cell was fixed at 60 μL. All cell assembly processes were carried out in an argon-filled glove box (H₂O < 0.1 ppm and O₂ < 0.1 ppm). For the preparation of the NCM811 positive electrode, NCM811, Super P, and PVDF were mixed in a mass ratio of 8:1:1, with PVDF dissolved in N-methyl-2-pyrrolidone (NMP) at a concentration of 2.5 wt%. The slurry was stirred for 8 h to ensure homogeneity, then uniformly coated onto aluminum foil using a spatula. A 16 μm thick aluminum foil with a purity of ≥99.65% was purchased from Guangdong Canrd New Energy Technology Co., Ltd. The electrodes were dried overnight in a vacuum oven at 110 °C and subsequently punched into circular disks with a diameter of 12 mm. The mass loading was approximately 3 mg cm⁻². An additional NCM811 electrode sheet with a higher mass loading of 13 mg cm⁻² was provided by Hangzhou Haber New Blue Technology Co., Ltd. The electrode consisted of 96 wt% NCM811, 3 wt% PVDF, and 1 wt% Super P. All lithium metal pouch cells were supplied and tested by Zhejiang Funlithium New Energy Technology Co., Ltd. Li-metal pouch cells with specific energies of 606.8 Wh kg⁻¹ and 601.7 Wh kg⁻¹ were assembled in an argon-filled glovebox at 25 ± 2 °C. A predetermined volume of electrolyte was injected through a preset port using a precision syringe, followed by a 4 h rest period to ensure sufficient wetting of the electrodes and separator. Subsequently, three vacuum–pressurization cycles were then applied (vacuum: −0.060 MPa; backfill: −0.005 MPa; hold time: 1 min per cycle). The first sealing was performed under a vacuum of −75 to −85 kPa, after which the cells were rested for 24 h prior to formation. During formation, the cells were placed between parallel clamping plates under a preload pressure of 410 kPa, charged galvanostatically at 1.3 A to 4.4 V and 4.5 V (for the 606.8 and 601.7 Wh kg⁻¹ cells, respectively), rested for 15 min, and then discharged galvanostatically at 1.3 A to 3.0 V, followed by another 24 h rest. The clamping plates were then removed for the second sealing, which was conducted under a vacuum of −0.095 to −0.103 MPa with a 12 s hold time. The residual electrolyte mass was controlled at 13.0 ± 0.5 g and 11.0 ± 0.5 g for the 606.8 and 601.7 Wh kg⁻¹ cells, respectively. For capacity grading, the cells were reclamped at 410 kPa, charged galvanostatically at 2.6 A to 4.4 V, and held at constant voltage until the current decreased to ≤0.65 A, followed by a 15 min rest and subsequent galvanostatic discharge at 2.6 A to 3.0 V. After an additional 15 min rest, cycling tests were conducted under an external pressure of 410 kPa, with 1 C defined as 13 A. Li-metal pouch cells with a specific energy of 502.5 Wh kg⁻¹ underwent the same first-sealing procedure, without further post-treatment. These cells were cycled directly without external pressure, with 1 C defined as 2.4 A. Detailed parameters are provided in Supplementary Tables 7–9.

Electrochemical tests

CV measurements were performed on Li‖Cu coin cells using an electrochemical workstation (CHI 660E) within a voltage range of −0.3 V to 0.6 V at a scan rate of 10 mV s⁻¹. EIS tests were conducted on Li‖Li cells over a frequency range of 0.01 Hz to 1 MHz in potentiostatic mode, with the steady-state potential set to the corresponding open-circuit potential. The acquired data were fitted using Autolab NOVA software. LSV was carried out on Li‖stainless steel cells in the voltage range of 2−6 V with a scan rate of 1 mV s⁻¹. The lithium-ion transference number (t_Li⁺) was determined using the steady-state current method reported in previous literature. A small polarization voltage of 10 mV was applied to the Li‖Li cell with a relaxation time of 1000 s. EIS measurements were performed before and after the current reached a steady state. The transference number (t_Li⁺) was calculated using the following equation:

$$t_{{\mathrm{Li}}^{+}}=\frac{I_{\mathrm{S}}\left(\mathrm{\Delta }V-I_0{R}_0\right)}{I_0\left(\mathrm{\Delta }V-I_{\mathrm{S}}{R}_{\mathrm{S}}\right)}$$ 5

where $I_o$ and $I_s$ represent the initial and steady-state currents after polarization, respectively. $R_o$ and $R_s$ denote the interfacial resistance of the lithium electrode before and after polarization, respectively.$△V$ is the applied polarization voltage^49,50. The CE of the Li‖Cu battery was evaluated using a LAND battery testing system (Wuhan, China). Copper foil served as the working electrode, while lithium metal sheets were utilized as both the reference and counter electrodes. In each cycle, lithium was deposited onto the copper foil at a current density of 1 mA cm⁻² and then stripped to a cutoff potential of 1 V (vs Li⁺/Li). To eliminate interference from the copper substrate, the CE of the Li‖Cu battery was determined using Aurbach’s method⁵¹. Specifically, a pre-lithiation layer (Q_T) with a capacity of 4 mAh cm⁻² was first deposited onto the copper foil at a current density of 0.5 mA cm⁻². This was followed by “n” repeated charge-discharge cycles at a fixed capacity of 0.5 mAh cm⁻² per cycle (Q_C). Finally, the remaining lithium in the reservoir layer was stripped at a current density of 0.5 mA cm⁻² until the cutoff voltage of 1 V was reached (Q_S). The average Coulombic efficiency (ACE) was calculated using the following equation:

$$\mathrm{ACE}=\frac{n{Q}_{\mathrm{C}}+Q_{\mathrm{S}}}{n{Q}_C+Q_{\mathrm{T}}}$$ 6

To assess the stability of different electrolytes against lithium metal, Li‖Li symmetric cells were cycled at a current density of 1 mA cm⁻² with a fixed areal capacity of 1 mAh cm⁻². Rate performance tests were carried out on Li‖Li cells at the same capacity condition (1 mAh cm⁻²), with current densities sequentially set at 0.5, 1, 2, 3, 4, and 5 mA cm⁻². All electrochemical tests on Li‖NCM811 coin cells were performed at a nominal specific capacity of 200 mAh g⁻¹. All battery cycling and rate capability tests were performed in a climatic chamber at 25 °C ± 1 °C. Other electrochemical measurements were carried out under controlled ambient laboratory conditions at 25 °C ± 2 °C. For coin cell electrochemical tests, three parallel samples were evaluated to ensure reproducibility of the experimental results. For cycling tests of lithium metal pouch cells conducted by Zhejiang Funlithium New Energy Technology Co., Ltd., two cells were tested in parallel for each electrolyte.

Characterization of materials

Raman spectroscopy was employed to investigate the solvation structures of electrolyte components. Measurements were conducted using a LabRAM HR Evolution spectrometer, with a 532 nm laser as the excitation source. Nuclear magnetic resonance (NMR) spectroscopy was performed on a Bruker AVANCE III 500 MHz spectrometer. All electrolyte samples were dissolved in deuterated dimethyl sulfoxide (DMSO-d₆) prior to analysis. SEM images were acquired using a Verios G4 UC microscope at an accelerating voltage of 5.0 kV. Cryogenic transmission electron microscopy (Cryo-TEM) was conducted on a Talos 200x microscope at an acceleration voltage of 200 KV to observe the morphology of the SEI. For Cryo-TEM analysis, lithium was deposited onto a copper grid, and residual surface contaminants were removed using DMC inside an argon-filled glove box⁵². The prepared grid was then transferred to the microscope for imaging under cryogenic conditions. XPS depth profiling was carried out on a Therma Scientific K-Alpha instrument with Al Kα radiation (hv = 1486.6 eV) as the excitation source. Depth etching was performed using an argon cluster ion gun at a rate of 0.25 nm s⁻¹, with spectra collected every 60 s. All binding energies were calibrated to the C 1 s peak at 284.8 eV. The entire XPS process was conducted under an inert atmosphere to prevent exposure to moisture and oxygen. TOF-SIMS analysis was performed using a TOF.SIMS5-100 instrument. Depth profiling was conducted using a 1 keV Cs⁺ ion beam at an etching rate of 0.5 nm s⁻¹, while a 30 keV Bi⁺ ion beam was used for surface analysis over a 50 μm × 50 μm area. For TOF-SIMS characterization, a Li‖Cu coin cell was subjected to galvanostatic cycling at a current density of 1 mA cm⁻². The cell was discharged for 3 h and subsequently charged to a cutoff voltage of 1 V at the same current density, completing five cycles. After cycling, the cell was disassembled in an argon-filled glove box. The lithium metal foil was retrieved, rinsed three times with DMC, dried, and used for subsequent characterization. All procedures were carried out under strict moisture- and oxygen-free conditions. HAADF-STEM was conducted on a Spectra 300 microscope operated at 300 kV. The cycled NCM811 powder was ultrasonically cleaned in ethanol for 20 min and mounted onto a microgrid for imaging. In situ XRD measurements were performed using a Bruker D8 Advance diffractometer at a scan rate 0.1° s⁻¹. ICP-MS was conducted on a PE NexION 300X instrument to quantify TM dissolution from the NCM811 positive electrode after cycling. Specifically, a Li‖NCM811 coin cell (cycled at 1 C/1 C within 3.0–4.4 V for 50 cycles) was disassembled in an argon-filled glove box. The lithium metal foil was dissolved in 9 mL of deionized water, followed by the addition of 1 mL of hydrochloric acid to extract any TM deposited on its surface. The resulting solution was analyzed via ICP-MS. All operations were performed under inert conditions to avoid contamination from air or moisture.

Author contributions

Hongwei Yu conceived and designed the experiments, performed the experimental work, analyzed and processed the experimental data, and drafted the manuscript. Xing Xin, Zoran Mandic, and Xu-Feng Zang contributed to the experimental design, participated in data discussion, and provided critical feedback on the manuscript. Jiawen Ge and Yanming Cui fabricated and evaluated all lithium-metal pouch cells. Yunyun Luo reviewed and corrected the manuscript formatting. Tonghui Xu contributed to discussions on theoretical calculation methodologies and provided technical suggestions. Tianle Zheng carried out the theoretical calculations and computational analyses. Yongyao Xia contributed to data interpretation and provided guidance on manuscript preparation. Mingjiong Zhou supervised the project, guided the experimental design, revised and polished the manuscript, and finalized the manuscript for submission. All authors discussed the results and contributed to the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks Wei Tang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. [A peer review file is available].

Data availability

The data generated in this study are provided in the Supplementary Information Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-71850-w.