Introduction

All-solid-state Li metal batteries (ASSLBs) using polymer as electrolyte are widely recognized as the most promising system to achieve high energy density and improved security1,2,3,4,5,6. Polyethylene oxide (PEO)-based solid polymer electrolytes show great potential to realize this goal for easy processability and good physical contact7,8,9, but suffer from poor oxidation stability at high voltages10,11,12,13,14. The notorious interfacial adaptability between the state-of-the-art layered Ni-Rich materials and PEO-based electrolytes remains a critical challenge15,16,17,18.

Tremendous efforts have been devoted to solving the interfacial problems of polymer electrolytes at high voltage. The chemical deterioration of PEO is generally acknowledged as the primary factor leading to battery failure19. Protective methods, including inorganic filler20,21, cathode coating22,23,24, and molecular grafting25,26 have been pursued to extend the working voltage windows of solid-state batteries. The 4.5 V-class PEO-based solid-state batteries have been successfully developed26. However, the inherent low oxidative potential of ether oxygen (EO) chains impedes their applications in ultra-high voltage solid-state batteries (for example, 4.8 V-class ASSLBs)27,28,29. It is urgently necessary to explore the failure mechanism within the voltage range of 4.5–4.8 V and create new design of electrolytes.

Recently, the high-concentration salt concept has proved to be effective in improving the high-voltage resistant ability of electrolytes30,31, as the highly reactive EO chains and anions are involved in the more stable coordination structure32. Despite this strategy providing insight into tuning the coordination environment to overcome the poor stability, the ultralow ionic conductivity makes electrolytes unsuitable for Li+ migration during battery charging/discharging33,34. Further, the metal ions-coordinated polymer was innovatively proposed to realize rapid Li+ conduction by chelating matrix molecules and expanding ion channels1 but still challenged by the incompatibility of electrolytes with high-voltage cathodes35. Inspired by the above coordination effect, it is hopeful that the tradeoff between oxidative stability and ionic conductivity can be decoupled by regulating the coordination mode in electrolytes.

Here we demonstrate a Lewis-acid coordinated PEO-based electrolyte to achieve high-energy 4.8 V-class batteries. The strongly electronegative EO chains are captured by introducing Mg2+ and Al3+ with strong electron-withdrawing capability to weaken its solvating ability for high valence Ni at high voltage36. This effect decreased the strong reactivity between the EO group and LiNi0.83Co0.12Mn0.05O2 (Ni83) cathodes, which significantly improved the interface compatibility at 4.8 V37. Besides, the typical coordination structure involved by Mg2+ and Al3+ and more anions shows low energy of the highest occupied molecular orbital (HOMO) and simultaneously forms an inorganic-enriched interphase38. As a consequence, the Mg2+ and Al3+-coordinated electrolytes (PEO-Mg-Al-LiTFSI) integrate high electrochemical oxidation potential (>5 V) with good ionic conductivity (0.23 mS cm–1 at room temperature). With the modified electrolyte, the 4.8 V-class all-solid-state batteries achieve 300 cycles. The pouch cells with the energy density of over 586 Wh kg‒1 exhibit good cycling stability. The industrial-scale manufacture of electrolyte membranes demonstrates the great engineering application potential of this strategy.

Results

Ni-poor layers induced by strong solvating ability of PEO at high-voltage

To design high-voltage stable electrolytes against Ni-Rich layered cathodes, understanding the failure mechanism between PEO-based electrolytes and cathodes is desirable. Especially charging over 4.3 V, which significantly improves the energy density, but needs a deep understanding of interface failure mechanism. Operando synchrotron transmission X-ray microscopy (TXM) was conducted to probe the correlation of PEO-cathode (Ni-rich) interface evolution with the voltage increase in solid-state batteries (Fig. 1a, b)39,40,41,42. To conduct Operando TXM, an in-situ cell was constructed according to the design shown in Figs. S1–2 and installed it in the center of the sample stage.

Fig. 1: Strong interface interaction at high voltage.
Fig. 1: Strong interface interaction at high voltage.
Full size image

a Operando TXM mapping of cathode particles at different PEO and liquid voltages. b The charge curve of PEO batteries and liquid batteries. c The Ni concentration of Ni-Rich cathodes with different voltages in PEO batteries and (d) in liquid batteries. e Adsorption energy of EC and PEO (f) on the half-lithiated LixNiO2 (104), gray, red, and purple balls represent Ni, O, and Li atoms, respectively. g Energy shift at the near edge of Ni extracted from the XAS of each region at different voltages. h Energy shift of the whole particle in PEO and liquid batteries. i Schematic of the formation mechanism of the “Ni-poor” layer.

In the range of 3.5–4.3 V, both PEO-based and liquid batteries show similar shape of charge curve, the discrepancy in voltage stage and capacity can be ascribed to sluggish migration of Li+ in PEO electrolyte43. Besides, as the outward direction along the radius (the point from No.1 to No.5) shown in Figs. S3, S4, the Ni intensity of each point keeps steady with the voltage increasing44. This result demonstrates unchanged Ni content from the surface to the bulk of Ni-rich cathodes in both PEO-based and liquid batteries. Although it has been proven that PEO electrolytes will decompose on the Ni-Rich cathode over 4.2 V45, the Ni content at the interface is not significantly affected by the decomposition during the first cycle.

In the range of 4.3–4.8 V, for the liquid batteries, the redox reaction that occurs along all directions follows the phase-transition mechanism of layered cathode particles41,42, as depicted by the Operando chemical mapping (Fig. 1a) and Ni K-edge X-ray absorption spectrums (XAS, Fig. S5a) from the surface to the bulk (No.1-No.5). However, for PEO batteries, there appears an extended voltage stage before reaching 4.5 V in the charging curve, the Ni concentration tested by inductively coupled plasma optical emission spectrometer (ICP-OES, Fig. 1c, d) and the peak intensity of the No.5 point extracted from Ni XAS (Fig. S5b) shows anomalous phenomenon of Ni content decrease on particle surface, indicating the formation of “Ni-poor” layer in this process. In addition, as the XAS extracted from the particle of 4.5 V shown in Fig. S4, the absorption edge shift of the surface (No.5) is less than that of the inner layer (No.4)46, which demonstrates a lower valence state of Ni at the PEO-cathode interface. When the PEO batteries are last charged to 4.8 V, the Ni ratio on the surface decreases drastically and is accompanied by a more lower Ni valence state.

This “Ni-poor” effect at higher potentials (over 4.5 V) indicates that, in addition to the decomposition of PEO electrolytes, the PEO-cathode interface also exhibits a phenomenon of Ni dissolution on the surface of Ni-rich particles during the first charging. This is similar to the so-called “Ni dissolution” in liquid batteries after long cycling47. Besides, the reduced absorption edge shift in the “Ni-poor” layer (Fig. 1e) demonstrates that, at high potential (4.5–4.8 V), high valence Ni is reduced to a low valence state during the oxidation of PEO, although the Ni valence of the whole cathode particle showed an increasing trend (Fig. 1f). The decrease of the Ni ratio can be explained by the strong interaction (−0.76 eV) between PEO and cathodes48, while the adsorption energy of liquid electrolyte (Ethylene carbonate, EC) for high valent Ni is only—0.42 eV, as the adsorption energy calculated in the Fig. 1g, h. Under the strong adsorption of the EO chains, Ni migrates from the surface of the cathode to the electrolyte phase (Fig. 1i), resulting in interface degradation and battery failure. Therefore, high-voltage stable electrolytes against layered cathodes should satisfy both intrinsic high oxidation stability and, most importantly, weak interaction between EO chains and cathodes.

Lewis-acid coordinated strategy to weaken the solvating ability of electrolytes

Here we creatively introduce polyvalent cations as electron-withdrawing ligands for EO chains to control the interfacial interaction49,50,51,52. The first criterion is ionic conductivity, electrolytes that only contain Al3+ have the highest ionic conductivity, followed by Mg2+ (Fig. S6a). High oxidation potential should be the second crucial requirement, electrolytes with Mg2+ alone show the highest oxidation stability window, followed by Al3+ (Fig. S6b). To improve the intrinsic oxidation-stability of electrolytes and weaken the interactions between cathodes and EO chains, Al3+ and Mg2+ (Fig. S6c, d) are simultaneously used to achieve complementarity and achieve their synergistic effects on ionic conductivity and voltage window. Additionally, the optimal molar ratio of Mg/Al: LiTFSI = 0.25 and Mw (2,000,000) of PEO was determined in electrolytes, which have the highest conductivity and improved oxidation potential (Figs. S7, S8). Therefore, the strong Lewis-acid additive of Mg(ClO4)2 and Al(ClO4)3 (EO: Li: Mg: Al = 18: 1: 0.25: 0.25) are employed to improve the intrinsic oxidation-stability of electrolytes and weaken the interactions between high valence Ni and EO chains53,54.

Scanning electron microscope (Fig. S9) shows the homogeneous and dense structure. Karl-Fisher titration results demonstrate that perchlorates and electrolytes (165 ppm) have almost no water content after drying, which has almost no impact on battery performance55 (Figs. S10–12). Thermogravimetric curve shows good thermal stability of the PEO-Mg-Al-LiTFSI electrolyte (Fig. S13).

The Mg2+ and Al3+ lead to a competitive coordination environment in electrolytes, the Li+ migration rates can be reinforced due to the weakened Li-O interactions and encapsulated TFSI‒51,56,57 (Fig. 2a). Raman spectroscopy was employed to gain insights into electrolyte coordination behaviors (Fig. 2b). The band at 744 cm‒1 is assigned to the ion pairs of TFSI‒ and the band at 740 cm‒1 is assigned to the dissociated TFSI‒20,58. The monotonic blue shift of these peaks demonstrates more TFSI‒ occupying the solvation sheath59, which will reduce the migration number of anions. Additionally, the reduced peak intensity at 744 cm‒1 suggests the facilitated dissociation of Li-TFSI ion pairs, indicating more free-Li+ for rapid ion transfer.

Fig. 2: Regulating the coordination structure to weaken the interfacial interaction.
Fig. 2: Regulating the coordination structure to weaken the interfacial interaction.
Full size image

a Schematic of the function of Lewis-acid additives. b Raman spectra of PEO-LiTFSI and PEO-Mg-Al-LiTFSI electrolytes. c Snapshots obtained from MD simulations of the PEO-LiTFSI electrolyte and (d) the corresponding radial distribution functions. e Snapshots of the PEO-Mg-Al-LiTFSI electrolyte and (f) Radial distribution functions centered on the Li atom, (g) Mg atom, and (h) Al atom, white balls represent H atoms, as well as blue-N atoms, light purple-O atoms, yellow-S atoms, light blue-F atoms, pink-Al atoms, and orange-Mg atoms.

To gain further insight into the working mechanism of Lewis acid, molecular dynamics (MD) simulations were performed to reveal the coordination configurations of electrolytes. The molecular legend is depicted in Fig. S4d. As the snapshot of PEO-LiTFSI shown in Fig. 2c, the typical configuration displays that the Li+ form coordination with O atoms (EO chains) and TFSI‒. The radial distribution functions (RDF) (Fig. 2d) show strong peaks of Li-EO chains and Li-TFSI coordination, and their high coordination numbers (CN), meaning the strong constraints on Li+. This result suggests that strong binding of Li-EO and Li-TFSI will cause concentration polarization in batteries60. In comparison, benefitting from the electron-withdrawing effect of additives, the typical coordination configurations of PEO-Mg-Al-LiTFSI show that anions extensively enter the inner solvation sheath (Fig. 2e), which results in a preferential formation of an anion-derived cathode electrolyte interphase (CEI)61. Besides, the decreased peak intensity of Li-EO and Li-TFSI coordination and reduced coordination number indicate a faster Li+ diffusion. The increased distance (from 2.056 Å to 2.075 Å) between Li+ and EO in PEO-Mg-Al-LiTFSI also demonstrates the weakened interaction of the Li-EO bond, which corresponds to the RDF analysis30. Finally, the Mg2+ and Al3+ show strong interaction with EO chains and TFSI− anions, as the RDF of Mg/Al/O/TFSI− calculated in Fig. 2f–h, compared to Li, Mg2+, and Al3+ have greater intensity peaks and coordination numbers when interacting with TFSI– and EO chains, which result in mostly coordinated EO chains and protect them from oxidation32.

Electrolyte performance for high-voltage cathodes

To investigate the oxidation resistance of the coordination structure, the electrostatic potential (ESP) distribution is explored to account for the local interaction at the electrolyte-cathode interface. As seen in Fig. 3a, the Lewis-acid coordinated EO chains show decreased electron density compared with Li-coordinated EO chains (PEO-LiTFSI), which manifests a weak interaction of PEO-Mg-Al-LiTFSI against the Ni-rich cathode at high state of charge62. Note that the enlarged region of negative area on the surface of the structure sheath indicates a preferential decomposition of inorganic anions, which is beneficial for forming high-voltage stable interphases63. In addition, the Lewis-acid coordinated structure shows a low energy level (−9.08 eV) of the highest occupied molecular orbital (HOMO) (Fig. 3b), confirming its good oxidation resistance ability61.

Fig. 3: Electrolyte properties aiming at high voltage cathodes.
Fig. 3: Electrolyte properties aiming at high voltage cathodes.
Full size image

a Electrostatic potential distribution of the typical coordination structure in PEO-LiTFSI and PEO-Mg-Al-LiTFSI electrolytes. b The HOMO energy values of each component. c Linear sweep voltammetry curve of PEO-LiTFSI and PEO-Mg-Al-LiTFSI electrolytes, with the threshold current of 2.5 µA. d The binding energy between Li and the typical coordination structure in PEO-LiTFSI and PEO-Mg-Al-LiTFSI electrolytes. e The ionic conductivities of different electrolytes at different temperatures. f The ion diffusion coefficient of anions and cations in electrolytes. g Average Li-O, Mg-O, and Al-O coordination numbers from MD simulations of PEO-LiTFSI and PEO-Mg-Al-LiTFSI electrolytes. h Chronoamperometry curve for the PEO-Mg-Al-LiTFSI electrolyte with a polarization voltage of 10 mV and electrochemical impedance spectroscopy (EIS) plots before and after the polarization (inset).

It is important to note that differences in coordination structures can explain variations in the electrochemical performances of electrolytes. Consequently, the features of the coordination structure have a significant impact on the oxidation potential. Linear sweep voltammetry (LSV) tests were further conducted to evaluate the antioxidation stability of electrolytes. The decomposition current that corresponds to the decomposition potential of the PEO-LiTFSI electrolyte can be used as the threshold for determining the decomposition potential of PEO-Mg-Al-LiTFSI. The PEO-LiTFSI electrolyte exhibits a noticeable current rise over 4.2 V (vs Li+/Li), while the PEO-Mg-Al-LiTFSI electrolyte shows improved antioxidation ability up to 5.0 V (vs Li+/Li) (Fig. 3c). Additionally, benefitting from the Lewis acid-modified competitive coordination for EO chains, the dissociation energy of Li+ from the coordination structure drops from 0.75 eV to as low as 0.58 eV (Fig. 3d), which is considered as key for boosted Li ions transfer64. Consequently, the dried (Table S2, Fig. S14) Lewis-acid coordinated electrolytes exhibit high ion conductivity of 0.23 mS cm–1 at room temperature (Fig. 3e). Indeed, it demonstrates superior ionic conductivity compared to PEO-LiTFSI electrolyte at temperatures below 60 °C.

The ion diffusion coefficient, subsequently, was calculated to explore the transport properties of each ion in electrolytes65. As shown in Fig. 3f, the Mg2+ and Al3+ in PEO-Mg-Al-LiTFSI are not mobile, as indicated by a negligible diffusion coefficient of 1.62 × 10‒10 cm2 s‒1 and 7.05 × 10‒11 cm2 s‒1, respectively. However, the Lewis-acid additives play an indispensable role in allowing fast Li+ transport, the PEO-LiTFSI displays a low Li+ conductivity (4.80 × 10‒2 mS cm–1) and diffusion coefficient (1.21 × 10‒9 cm2 s‒1), by sharp contrast, the PEO-Mg-Al-LiTFSI displays a much higher diffusion coefficient of 5.30 × 10‒8 cm2 s‒1. Additionally, in PEO-Mg-Al-LiTFSI electrolyte, the diffusion coefficient (5.3 × 10‒8 cm2 s‒1) of Li ions is 2.24 times the sum of the diffusion coefficients of TFSI– (1.9 × 10‒8 cm2 s‒1) and ClO4– (4.6 × 10‒9 cm2 s‒1). And the higher coordination number (from RDF) of Mg-O (3.80) and Al-O (3.57) than Li-O (2.32) is also consistent with the diffusion analysis (Fig. 3g), resulting in the high transference number (\({t}_{{{Li}}^{+}}\) = 0.67) in PEO-Mg-Al-LiTFSI (Fig. 3h). These results demonstrate the great practicality potential of PEO-Mg-Al-LiTFSI electrolytes in high loading composite cathodes.

For the compatibility between electrolyte and lithium anode, the Lowest Unoccupied Molecular Orbital (LUMO) calculations (Fig. S15) show that, compared to PEO electrolyte (−1.51 eV), the LUMO energy of PEO-Mg-Al-LiTFSI electrolyte (−2.11 eV) is slightly reduced. This is characterized as an increase in anode interface impedance and cyclic overpotential (Figs. S16–17). The XPS analysis reveals that that in PEO-Mg-Al-LiTFSI batteries (Fig. S18), the anode solid electrolyte interphase (SEI) components may include organic compounds66, carbonates, LiF, MgF2, AlF3, and Al2O367,68. These low conductivity components increase the interfacial impedance but may be beneficial for long-term cycling stability51.

The practicality of Lewis-acid coordinated electrolytes in high-energy batteries

To evaluate the capability of electrolytes that support higher voltages, we assembled all-solid-state Li lithium batteries (Fig. 4a, b), with PEO-Mg-Al-LiTFSI as cathode side electrolytes membrane and ion conductor in the composite cathode (Fig. S19). Note that the EO chains of Lewis-acid coordinated electrolytes show a rather small reaction energy of −0.42 eV on the half-lithiated LixNiO2 (104), while the EO in PEO-LiTFSI is energetically favorable with a reaction energy of −1.48 eV (Fig. 4c, d), indicating good interface stability of Lewis-acid coordinated EO chains against the high voltage cathodes30,69.

Fig. 4: Batteries performance assembled by Lewis-acid coordinated electrolytes.
Fig. 4: Batteries performance assembled by Lewis-acid coordinated electrolytes.
Full size image

a Schematic of the structure of PEO-Mg-Al-LiTFSI batteries. b Comparison chart of properties of electrolytes. c The adsorption energy of PEO-LiTFSI electrolyte on the half-lithiated LixNiO2. d Adsorption energy of PEO-Mg-Al-LiTFSI (EO- LixNiO2) on the half-lithiated LixNiO2 (104). e Cycling performance of PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries. f EIS of PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries before and after cycling. g Galvanostatic charge/discharge voltage profiles of PEO-Mg-Al-LiTFSI. h Galvanostatic charge/discharge voltage profiles of pouch cells. i Cycling performance of PEO-Mg-Al-LiTFSI pouch cells. j Optical photos of PEO-Mg-Al-LiTFSI pouch cells and their application in micro drones.

Figure 4e shows that the performance of full batteries with PEO-Mg-Al-LiTFSI as cathode side electrolytes (referred to below as PEO-Mg-Al-LiTFSI batteries) outperforms the PEO-LiTFSI batteries. Specifically, the PEO-Mg-Al-LiTFSI batteries exhibit a long cyclic performance over 300 cycles in the voltage range of 2.8–4.8 V. Even with a high loading of 15.1 mg cm‒2 (Fig. S19a), it still delivers a high retention of 80.8% after 100 cycles. On the contrary, the PEO-LiTFSI batteries cannot charge and discharge at high cut-off voltage of 4.8 V. Electrochemical impedance spectroscopy (EIS) investigation of cycled batteries was conducted, as depicted in Fig. 4f, the PEO-Mg-Al-LiTFSI batteries show low interface impedance (Rinter) after 100 cycles while the PEO-LiTFSI batteries suffer from a sharp increase of Rinter after only 3 cycles. These results agree well with the density functional theory (DFT) calculations. The corresponding charge-discharge voltage profiles of the batteries at different cycles, as seen in Fig. 4g, demonstrate low electrochemical polarization in PEO-Mg-Al-LiTFSI batteries. On the contrary, the highly reactive interface severely degrades the capacity of PEO batteries (Fig. S20a)17.

To better represent practical conditions, the PEO-Mg-Al-LiTFSI pouch cells were evaluated and delivered a discharge energy density of 586 Wh kg‒1 (bare cell = 645 Wh kg‒1, Fig. 4h) at 0.1 C. More importantly, high retention of capacity (80.6%) is observed even after 50 cycles (Fig. 4i, 100 cycles with retention of 63.5%). The high energy density of batteries indicates the ability of intelligent equipment to endure ultra-long. Figure 4j shows the potential for uncrewed aerial vehicles expected to operate at much longer mileage. As shown in Figs. S20b–d, S21, we further assembled PEO-Mg-Al-LiTFSI batteries and liquid batteries with Li-Rich Li1.14Ni0.136Co0.136Mn0.542O2 materials as cathode, the batteries show a high retention of 74.7% after 100 cycles. While liquid batteries have a retention rate of 61.8% after 100 cycles, and they are incapable of operating at elevated temperatures. In addition, the battery’s capacity retention using PEO-LiTFSI after 100 cycles, within the voltage range of 2.8–4.2 V, was found to be only 28.1%. Comparatively, the PEO-Mg-Al-LiTFSI batteries were evaluated for their cycle performance within the voltage range of 2.8-4.6 V and 2.8-4.7 V, as shown in Fig. S22. The improved stability of the high voltage batteries supports the fact that the “Lewis-acid coordinated electrolytes” have great potential to promote the energy density of PEO-based batteries significantly.

Furthermore, a roll of PEO-Mg-Al-LiTFSI electrolyte membranes with a length of 50 m, width of 0.3 m was prepared using an industrial-scale continuous slurry-casting production line (Fig. 5a, b)70, by coating the electrolyte slurry on a polyethylene (PE) substrate film. This flexible and transparent electrolyte shows uniformity of thickness and high quality of surface (Fig. 5c, d). The solid cathodes can be manufactured using similar processes by replacing PE and electrolyte slurry with current collector and composite electrode slurry, respectively. In addition, a 20 Ah all-solid-state polymer battery was assembled by the electrolyte membrane obtained from the industrial-scale continuous process (Fig. 5e). The high capacity, high coulomb efficiency, and good cyclic performance of this device (Figs. S23–24) present the great potential of the Lewis-acid coordinated strategy in the engineering application.

Fig. 5: Industrial-scale membranes manufactured by continuous process.
Fig. 5: Industrial-scale membranes manufactured by continuous process.
Full size image

a Schematic of the manufacturing process of membranes. b Optical photos of industrial-scale continuous slurry-casting production line. c, d A roll of electrolyte membranes with a length of 50 m, width of 0.3 m. e Optical photos of 20 Ah all-solid-state polymer battery.

Ultimately, we conducted the electrolyte moisture content (Table S3) under different dry environments with varying standing times, as well as the corresponding electrolyte impedance, the first Coulombic efficiency of the battery, and the impedance of lithium symmetric batteries. The PEO-Mg-Al-LiTFSI electrolyte exhibits stable ion conductivity, oxidation potential (Fig. S25), cathode interface stability (Fig. S26), and lithium metal interface stability (Fig. S27) in a moisture-free environment with standing time increasing, hence confirming its good chemical stability.

Investigation of the high voltage stable interface

Synchrotron TXM was further conducted to investigate the interaction between electrolyte and Ni-Rich cathode. As the 2D chemical phase mapped in Fig. 6a, the Ni83 particle (discharge state) in PEO-LiTFSI shows an enlarged area of local high SOC region after cycling41. In contrast, the cathode particle disassembled from PEO-Mg-Al-LiTFSI shows homogeneous SOC. The X-ray absorption spectrum (Fig. 6b) also presents that a clear shift of 1.5 eV occurs at the absorption edge of nickel in PEO-LiTFSI cathode particles, compared with PEO-Mg-Al-LiTFSI batteries42. This result indicates an improved interface stability benefited by the Lewis-acid coordinated effect, which promotes interfacial Li+ transport and SOC homogeneity. Besides, the statistical intensity of nickel (Fig. 6c) and absorption edge shift (Figs. 6d, S28) shows that the content and valence of Ni element are consistent with the phase-transition mechanism in liquid batteries43, manifesting a locally inhibited interaction between EO chains and cathode.

Fig. 6: Interface investigation in high voltage batteries.
Fig. 6: Interface investigation in high voltage batteries.
Full size image

a TXM mapping of the whole particle in PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries after 5 cycles. b The Ni XAS of the whole particle in PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries after 5 cycles. c The normalized Ni intensity of each region (No.1–No.5) in PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries. d Energy shift at the near edge of Ni extracted from the XAS of each region (No.1–No.5) at different voltages. e 3D structure of synchrotron X-ray tomography for PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries after cycling, the viewport size of electrolyte is 10 × 10 × 10  µm3, green represents strong absorption of X-rays and blue represents weak absorption. f Quantitative analysis of inorganic area proportion. g The variation of porosity with distance of the Ni83 particle after cycling. h XPS spectra of F 1 s of cycled Ni83 particle in PEO-LiTFSI and PEO-Mg-Al-LiTFSI batteries.

Synchrotron X-ray nano-tomography, subsequently, was used to explore the interphase structure after cycles. As reconstructed in Fig.6e, a large area of X-ray weakly absorbed substance on the PEO-LiTFSI cathode surface can be assigned as organic components, while the surface of the PEO-Mg-Al-LiTFSI cathode particle shows a larger area of strong X-ray absorbent substance that corresponds to inorganic components45. The quantitative analysis (Fig. 6f) also demonstrates an inorganic-dominant interphase of the PEO-Mg-Al-LiTFSI particle, which can inhibit continuous electrolyte oxidation and provide excellent interfacial stability71. These results also agree well with the DFT calculations of ESP and reaction energy. Besides, the higher porosity of PEO-LiTFSI particles (Fig. 6g) after cycles indicates more micro-cracks due to the severe surface degradation, which impedes ion transport in solid-state batteries72. By comparison, the PEO-Mg-Al-LiTFSI particle keeps low porosity, demonstrating structural integrity thanks to the uniform and stable CEI.

Interfacial chemistry was further investigated by conducting X-ray photoelectron spectroscopy (XPS). The cathode powder sample is obtained by dissolving the disassembled cathode sheet in acetonitrile solvent with further filtering and drying. As seen in Fig. 6h, the apparent signals of LiF, MgF2, and AlF3 were observed on the surface of the PEO-Mg-Al-LiTFSI cathode, while PEO-LiTFSI cathode showed weaker signals of LiF66,73. We know that inorganics-abundant (LiF, MgF2, and AlF3) interphases which are dense and robust, can block electron leakage and give rise to a low interface resistance51,74,75. Thus, good interface compatibility can be obtained to improve the high-voltage stability of electrolytes.

Discussion

In summary, the strong interaction between EO chains and Ni-Rich cathode was diagnosed as a critical origin responsible for the failure of high voltage PEO-based batteries. Accordingly, Lewis-acid Mg2+ and Al3+ were introduced as electron-withdrawing ligands to control the solvation ability of EO chains and enhance the oxidation resistance of electrolytes. As a consequence, the Lewis-acid coordinated electrolytes demonstrate high electrochemical oxidation potential (>5 V) with good ionic conductivity (0.51 mS cm–1). Further, the batteries with Lewis-acid coordinated electrolytes deliver good cyclic performance in the range of 2.8–4.8 V. A high energy density pouch cells over 586 Wh kg‒1 (bare cell = 645 Wh kg‒1) achieved 100 cycles with a capacity retention of 63.5%. A roll of electrolyte membranes was prepared using an industrial-scale continuous process also shows the potential of a Lewis-acid coordinated strategy to significantly promote practical processes for high energy density ASSLBs.

Methods

Materials preparation

The PEO-LiTFSI electrolytes were prepared by a solution casting method and hot pressing at 80 °C. 2.763 g Polyethylene oxide (PEO, Mn = 2,000,000, Aladdin, 99.9%) and 1 g lithium bis(trifluoromethane sulfonimide) (LiTFSI, Aladdin, 99%) were added into 100 g anhydrous acetonitrile (ACN, Aladdin, 99%) solutions, the EO:Li ratio is 18:1. After stirring for 12 h, the mixture of electrolyte solution was poured into a polytetrafluoroethylene mold and dried under vacuum at 60 °C for 48 h. Finally, a solid-state electrolyte (PEO-LiTFSI) membrane was obtained through hot pressing process. For PEO-Mg-Al-LiTFSI electrolytes, 5 g Al(ClO4)3 9H2O (Aladdin, 99.9%) and 2.5 mg Al debris were mixed and heated 15 h at 150 °C in the inert atmosphere (Ar) for dehydration, then the anhydrous Al(ClO4)3 was obtained. Next, 0.284 g anhydrous Al(ClO4)3, 0.194 g anhydrous Mg(ClO4)2, 1 g LiTFSI and 2.76 g PEO were added into 100 g anhydrous acetonitrile, stirring for another 12 h, then the obtained PEO-Mg-Al-LiTFSI electrolyte solution was poured into a polytetrafluoroethylene mold and dried under vacuum at 60 °C for 48 h. Finally, PEO-Mg-Al-LiTFSI membrane was obtained through hot pressing process.

Preparation of high-voltage composite cathode and anode

LiNi0.83Co0.12Mn0.05O2 (Ni83) cathodes or Li-Rich Li1.14Ni0.136Co0.136Mn0.542O2 materials, electrolyte (PEO-Mg-Al-LiTFSI electrolyte solution), and carbon nanotubes (CNTs) were added to N-Methyl pyrrolidone (NMP, Aladdin, 98%) solvent in a mass ratio of 85:12:3, and the uniform electrode slurry was obtained through a planetary agitator. Then the obtained slurry was coated on the aluminum current collector. Subsequently, the electrode is dried at 65 °C for 24 h under vacuum to eliminate residual solvent. Finally, to prepare a dense composite cathode, press the dried cathode to 1 MPa cm‒2 at 80 °C. The pure lithium metal electrode used for assembling Coin-type batteries and the copper mesh-lithium metal composite foil used for assembling pouch cells were purchased from China Energy Lithium Co., Ltd.

Battery assembly: For PEO-Mg-Al-LiTFSI batteries, the CR2025 coin-type full batteries were assembled at a sealing pressure of 0.70 T in glovebox filled with argon, where the lithium metal foil is used as the anode. The double layer electrolyte is used as the membrane. The PEO-LiTFSI (≤15 µm) is used as anode side membrane, PEO-Mg-Al-LiTFSI is used as cathode side membrane and ion conductor in cathode. different composite cathodes as working cathodes, all the cathode with loading over 15 mg cm–2. For pouch cells, Lithium copper (Copper mesh as current collector) composite foil is used as anode, the total thickness of anode is less than 20 µm, the double layer electrolyte membrane is less than 40 µm, the loading of cathode sets 50 mg cm–2 (double-sided coating), with the area of 32 × 42 mm2, made through hardware knife molds. Then four cathodes, five anodes and corresponding number of membranes were used to assemble an all-solid-state bare cell with 3.93 g. Next, polypropylene (PP) separator is used to wrap the bare cell to prevent short circuits. Aluminum-plastic film is used to encapsulate the battery (4.31 g) at the vacuum condition.

Electrochemical measurements

Galvanostatic charge/discharge measurements were performed on Neware battery test system (BTS-2004, Shenzhen, China) under 60 °C, at atmospheric pressure, and the voltage ranges of batteries with Ni83 as cathode were adjusted to 2.8–4.8 V (1 C = 200 mA g–1), the batteries with Li-Rich as cathode were adjusted to 2.0-4.8 V (1 C = 200 mA g–1). Pouch cells were tested under 1 MPa pressed by splint. The ionic conductivity of electrolyte films and all electrochemical impedance polymer electrolytes were evaluated by the electrochemical working station (Par-Solartron 663). The SS|PEO-LiTFSI|SS and SS|PEO-Mg-Al-LiTFSI|SS batteries were assembled with stainless steel sheet and electrolyte membrane. Ionic conductivity is calculated by the Eq. (1)1:

$$\,{{{\rm{\sigma }}}}=\frac{L}{{RS}}$$
(1)

where L represents thickness the of electrolyte membrane, the R resistance of battery, and S the contact area between steel sheet and electrolyte. The electrochemical window of the SPE was conducted by linear sweep voltammetry (LSV) technique in a battery that sandwiched the electrolyte between steel sheet and lithium metal at a sweep rate of 1 mV s–1, 60 °C. The lithium tLi+ was calculated by Eq. (2):

$${t}_{{Li}+}=\frac{{I}_{s}}{{I}_{0}}\frac{\varDelta V-{I}_{0}{R}_{0}}{\varDelta V-{I}_{s}{R}_{s}}$$
(2)

The initial current I0 and the steady current Is of a Li|SPE|Li cell were measured with a voltage pulse (ΔV) of 10 mV at the mode of DC polarization. Electrochemical impedance spectroscopy was implemented before and after the DC polarization measurement to obtain the initial resistances R0 and the final interfacial Rs. For the pouch cells test, a pressure mold is used for applying a pressure of 1 MPa on the battery. The screw can be adjusted using a torque wrench to ensure even distribution of force on the battery. The testing environment is atmospheric pressure and 60 °C. The activation conditions for all batteries are two cycles of charging and discharging at 0.1 C.

Energy density calculation

According to the definition of battery energy density, it refers to the amount of energy (Wh) that a battery can release per unit mass (kg). Battery energy density (Wh kg–1) = discharge energy (Wh)/battery quality (kg). The energy density of the pouch cell in this study was determined to be 586 Wh kg–1 based on the discharge capacity acquired from the charge and discharge tester (2.53 Wh) and the battery mass measured using an electronic balance (4.32 g).

SEM, Raman, ICP-OES, TG

The morphology of cathode and powders was observed by field-emission scanning electron microscopy (SEM, Phenom Pro), and the smooth cross-section of the cathode was achieved by a cross-section ion polisher (accelerating voltage: 2 ~ 6 kV, beam spot size: 500 μm). Raman spectrum was recorded by Raman spectrometer (Renishaw). Inductively coupled plasma optical emission spectrometer (iCAP 7400) was used to test the element concentration of Ni-Rich cathodes disassembled from batteries at different voltages. The heating rate of the thermogravimetric (TG) curve is 5 °C for Al(ClO4)3 (keep 135 °C for 3 h) and 10 °C for electrolytes, respectively.

XPS

the battery was disassembled in a drying room and the cathode was dissolved again using acetonitrile solvent to remove the PEO-based electrolyte. The cycled cathode powder can be further obtained through filtration and drying for XPS testing. X-ray Photoelectron Spectroscopy (XPS, ESCALAB 250Xi) was employed to investigate the surface elvolution.

Transmission X-ray microscopy (TXM), TXM mapping and corresponding data analysis

The CT image of TXM with synchrotron X-ray was recorded at Shanghai Synchrotron Radiation Facility (SSRF), beamline BL18B. For CT samples, due to the need for rotation of the sample stage, the cathode sheet is precisely cut and positioned inside a capillary glass tube, securely fastened at the midpoint of the sample stage, ensuring that the capillary is perfectly perpendicular to the sample stage. At above 15 keV, for ~180 projections over an angular range of 180° with a field of view of 20 × 20 μm2. The raw data obtained is tomographically aligned and reconstructed76. The reconstructed cathode 3D data was further processed by the software package Avizo (Thermo Fisher Scientific, Waltham, Massachusetts, USA) for 3D visualization and quantized data analysis. The in-operando TXM mapping imaging was performed on the model coin cell during the initial charging (delithiation) process. To study the chemical state evolution, in the glove box, disassemble the coin-type battery. In order to prevent oxidation of the lithium metal and subsequent damage to the sample, it is recommended to introduce acetonitrile solvent between the electrolyte and lithium metal. This will effectively separate the lithium metal anode. Next, the cathode sheet is cut to an appropriate size and installed it on the sample stage, then a full XAS image series is collected at each charging stage (voltage) during the delithiation process. Each XAS image series was measured by scanning Ni absorption K-edge from 8280 to 8500 eV72. The recorded data were processed by an open-source software of TXM-Wizard.

Density functional theory (DFT) calculations

Vienna Ab-initio Simulation Package (VASP) code was used to carry out the density functional theory (DFT) calculations77, and the model was established using Materials Studio software. The exchange-correlation was processed using the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA), and the expansion of the electronic eigenfunctions was described by applying the projector augmented-wave pseudopotential (PAW) with a kinetic energy cut-off of 500 eV78,79. A Γ-centered 1 × 1 × 1 Monkhorst–Pack k-point sampled the Brillouin-zone integration. Until energy and force reached a tolerance of 1 × 10–5 eV and 0.03 eV/Å, respectively, all atomic positions were totally loosened. The long-range interactions were taken into consideration using the dispersion-adjusted DFT-D technique80. During the process of establishing the model, an organic compound with 18 EO chains was used as the PEO substrate, this compound has a 1: 1: 0.25: 0.25 molar ratio to LiTFSI, Al(ClO4)3, and Mg(ClO4)2, and a box of 50 × 50 × 50 Å was constructed. The interatomic potential of components was derived from the Optimized pseudo potential for Liquid Simulations all-atom (OPLS-AA) force field81. For PEO the parameters are from Jorgensen’s original paper, for TFSI– are developed by Minhaeng Cho, and Al3+/Mg2+ are developed by Biswarup Pathak’s job82,83. As for the entire system, we refer to the previous work on electrolyte solution systems84,85,86,87,88, and use VASP’s machine learning force field. During the molecular dynamics (MD) simulation, we employed a timestep of 3000 for the time integration, and the systems were running with a temperature of 300 K in the NVT ensemble for 500 ps. Calculations for the determination of binding energy (Eb) for decomposition reactions on a four-layer slab model of LiNiO2 (104) were performed30. The binding energy (Eb) of a complex formed between two molecules, A and B, can be calculated using the following Eq. (3):

$${E}_{b}={E}_{{complex}}-({E}_{A}+{E}_{B})$$
(3)

Where Ecomplex is the total energy of the molecular complex of A and B. EA and EB are the total energies of isolated molecules A and B, respectively. The electrostatic potential (ESP) and lowest unoccupied molecular orbital (LUMO)-highest occupied molecular orbital (HOMO) energy were conducted with the Gaussian09 package63 at the level of B3LYP functional level with 6–311++(d, p) basis sets89.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.