Single-atom catalyzed formation of inorganic-rich SEI/CEI for durable anode-free solid-state lithium metal batteries

Abstract

Anode-free solid-state lithium metal batteries are promising for energy storage owing to their maximum energy density, safety, and cost-effectiveness. However, their practical application remains hindered by fragile electrode–electrolyte interfaces (EEI) and the resulting rapid active-species depletion. Here, a single-atom catalysis strategy is proposed for the in situ construction of inorganic-rich EEI by utilizing a single-Ni-atom-anchored covalent organic framework (COF) as a catalyst, which is incorporated into a polymer electrolyte. The B–O–Ni bridge on COF-5 accelerates the electron transfer to the TFSI⁻ anion and improves the decomposition kinetics of lithium salt, thereby building an inorganic-rich solid electrolyte interphase (SEI) to enable smooth Li deposition and remarkable interfacial stability. Additionally, the boron-based COF-5-generated B, F-rich cathode electrolyte interphase (CEI) inhibits the dissolution of transition metal ions and ensures the structural integrity of NCM cathodes upon cycling. Consequently, the NCM622||Li solid-state cell demonstrates an exceptional capacity retention of 92.0% over 1500 cycles at 1.0C while achieving a remarkable capacity of 172.4 mA h g⁻¹ after 200 cycles at 0.2C, under a cutoff voltage of 4.7 V. Moreover, the anode-free NCM622||Cu solid-state pouch cell maintains stable cycling for over 200 cycles in the presence of carbonate electrolytes. This study extends single-atom catalysis into a platform to regulate lithium-salt decomposition for achieving prolonged anode-free solid-state batteries.

---

Broader context

Anode-free solid-state lithium metal batteries (AFSSLMBs), with desired high energy density, high safety, and cost-effectiveness, are considered to have significant potential for next-generation energy storage applications. Unfortunately, the fragile electrode–electrolyte interfaces and the rapid capacity decline have severely hindered the practical implementation of AFSSLMBs. Although some strategies have been developed to address these issues, most are limited by the sluggish lithium-salt decomposition and extensive Li loss upon SEI/CEI layer formation. Here, a single-atom interface catalysis strategy is proposed to promote the decomposition kinetics of lithium salt and the construction of inorganic-rich SEI/CEI layers, which addresses these challenges and demonstrates considerable industrial value. This atomic-level regulation strategy yields a remarkable ion-storage capability with an extraordinary capacity retention of 92.0% over 1500 cycles when coupled with a 4.3 V NCM622 cathode. Moreover, the anode-free NCM622||Cu pouch cell (405 Wh kg⁻¹) delivers a steady cycle for over 200 cycles in the presence of carbonate electrolytes. This study provides a promising avenue for the effective regulation of lithium-salt decomposition and the construction of inorganic-rich SEI/CEI layers, potentially contributing to the practical development of high-performance AFSSLMBs.

1. Introduction

Solid-state lithium metal batteries (SSLMBs) featuring non-flammable solid-state electrolytes (SSEs) and high-specific-capacity Li metal anodes (3860 mA h g⁻¹) provide a promising avenue to boost the energy density and safety of secondary batteries.^1,2 Generally, the production of SSLMBs using excess lithium greatly obstructs the further improvement of energy density and poses safety risks. Anode-free solid-state lithium metal batteries (AFSSLMBs) with all the lithium originating from a lithiated cathode and no pre-stored lithium in the anode will maximize the energy density (≥500 Wh kg⁻¹ and 1500 Wh L⁻¹), safety, and cost-effectiveness of SSLMBs.^3,4 However, the irreversible Li loss caused by the fragile EEI layers and subsequent “dead Li” formation under repeated cycling shortens the lifespan of AFSSLMBs, limiting them to fewer than 100 cycles.⁵

Establishing an inorganic-rich SEI with excellent mechanical strength and ionic conductivity on a Cu current collector is essential for regulating the lithium plating/stripping behavior.⁶ Approaches for building artificial SEIs,⁷ modifying electrolyte components,^8,9 and introducing functional additives^10,11 have been explored to inhibit the undesirable electrolyte decomposition and smoothen Li plating/stripping on the Cu current collector, aiming to enhance the Coulombic efficiency (CE) and prolong the lifespan of AFSSLMBs. For instance, constructing an Ag–C composite SEI or Mg/W bilayer SEI with superior lithiophilicity has proven efficacious in facilitating uniform Li deposition and inducing the dense cohesion of Li deposits with electrolytes.^12,13 The decomposition of the fluorinated 1,4-dimethoxybutane electrolyte generates an inorganic-rich SEI, which can promote the ion transport and inhibit interfacial side reactions.¹⁴ Nevertheless, the uncontrollable reaction processes and extensive Li loss upon SEI formation during the first lap lead to a decreased initial CE and capacity degradation of AFSSLMBs.^4,15 In addition, building an inorganic-rich and oxidation-resistant CEI layer to pair with high-nickel cathodes and inhibit parasitic side-reactions under a high voltage is crucial but rarely related to AFSSLMBs. Therefore, exploring effective strategies to construct an inorganic-rich SEI/CEI layer while reducing irreversible Li loss is urgently desired for the successful implementation of AFSSLMBs.

Regulating the structures and components of lithium salts (anions) is considered a promising approach that consumes a limited quantity of lithium to build robust EEI layers.¹⁶ For example, an asymmetric lithium salt of LiFEA with pseudo-crown ether-like binding geometry is designed to generate a LiF/Li₂O-rich SEI for fast-cycling lithium metal batteries in carbonate electrolytes.¹⁷ A LiF-rich SEI was formed in anode-free lithium pouch cells using a dual-salt LiDFOB/LiBF₄ electrolyte to promote the uniform lithium deposition.¹⁸ Unfortunately, the strong interaction between Li⁺ and the organic segments of the electrolyte hinders anions from penetrating the inner layer of the solvation sheath, leading to a sluggish decomposition of anions at the electrode–electrolyte interfaces and a limited amount of generated inorganic species (such as LiF, Li₂O, Li₃N, and Li₂S).^19,20 Single-atom catalysts, with highly uniform active centers, maximized atom utilization, and tunable coordination environments, are of significant interest to facilitate a specific reaction process.²¹ Typically, the lithium polysulfides (LiPS) redox kinetics in Li–S batteries are significantly facilitated by single-atom catalysts to improve the sulfur utilization efficiency and cycling lifetime.²²

In this study, inspired by the remarkable electrocatalytic activity, a novel single-atom interface catalysis strategy is proposed to enhance the decomposition kinetics of lithium salt, thereby constructing an inorganic-rich EEI toward high-voltage AFSSLMBs. We design a novel composite solid-state electrolyte that incorporates a single-Ni-atom-coordinated boron-based COF (COF-5@Ni). As depicted in Fig. 1(a), the single Ni sites with a positive charge effectively anchor the TFSI⁻ anions via electrostatic interaction, giving rise to more “free” Li⁺. More importantly, the single Ni sites on the tailored B–O–Ni bridge rapidly accelerate the charge transfer to the adsorbed TFSI⁻ anions and promote the multistep decomposition of TFSI⁻ anions, thereby building an inorganic-rich SEI to guarantee the fast Li⁺ transport and smooth Li deposition. The boron-group on COF-5 participates in the formation of a B, F-rich CEI, inhibiting the dissolution of transition metal (TM) ions and ensuring the structural integrity of NCM cathodes upon cycling. Additionally, the confinement of polymer chains within well-ordered COF pores establishes interconnected ion migration pathways to enhance rapid Li⁺ migration and regulate Li⁺ flux. As a result, the as-prepared SSE demonstrates a reversible Li⁺ dissolution/deposition and enhanced high-voltage durability. When coupled with a LiNi_0.6Co_0.2Mn_0.2O₂ (NCM622) cathode, a reversible capacity of 126.0 mA h g⁻¹ with an exceptional capacity retention of 92.0% over 1500 cycles at 1C is yielded at 30 °C. The anode-free pouch cell (405 Wh kg⁻¹) delivers a steady cycle with a remarkable capacity retention of 65.4% over 200 cycles, positioning it among the top-performing AFSSLMBs. The single-atom interface catalysis strategy enables the atomic-level precise control of interfacial chemistry in AFSSLMBs, effectively preventing the uncontrolled side reactions and excessive Li loss during the SEI/CEI film formation. Additionally, it allows for the favorable participation of commercial lithium salts (anions) in the generation of inorganic-rich SEI/CEI film.

Fig. 1 (a) Schematic of the catalytic mechanism of COF-5@Ni in LiTFSI decomposition. (b) PDOS of the COF-5@Ni and COF-5 systems. Deformation charge density of the (c) COF-5@Ni and (d) COF-5 systems (yellow and blue clouds represent the electron concentration and dissipation area, respectively). Snapshots of the AIMD-simulated LiTFSI decomposition for the (e) COF-5@Ni and (f) COF-5 systems.

2. Results and discussion

2.1 The catalytic mechanism of COF-5@Ni in lithium-salt decomposition

The crystalline framework of COF-5 was synthesized by linking 1,4-benzene diboronic acid (BDBA) and 2,3,6,7,10,11-hexahydroxytriphenylene (HHTP) monomers (Fig. S1).²³ A single Ni atom was anchored onto the pore walls of COF-5 via coordination with an oxygen atom. First, density functional theory (DFT) and ab initio molecular dynamics (AIMD) calculations were performed to explore the catalytic mechanism of COF-5@Ni in lithium-salt decomposition. The partial density of state (PDOS) calculations revealed that the 3d orbital of the Ni atom on the B–O–Ni bridge exhibited robust hybridization with the 2p orbitals of the O, F, and N atoms on a LiTFSI molecule near the Fermi level (E_F) after introducing the Ni sites into COF-5, demonstrating a strong interaction and large charge transfer between COF-5@Ni and LiTFSI (Fig. 1(b)).²⁴ Deformation charge density analysis was conducted to further elucidate the enhanced charge transfer (Fig. 1(c) and (d)). Compared with COF-5, more electrons were transferred from COF-5@Ni to LiTFSI because of the robust orbital hybridization between the B–O–Ni bridge and the LiTFSI molecule, which facilitated the reduction reaction kinetics of the TFSI⁻ anions.²⁵ The single Ni sites acted as a bridge to promote the electron transfer from COF-5 to LiTFSI. AIMD was employed to track the decomposition behavior of LiTFSI under the COF-5@Ni and COF-5 systems. Bader charge analysis indicated that −1.14 |e| transferred from the Ni site to the TFSI⁻ anion when LiTFSI was adsorbed on COF-5@Ni (Fig. 1(e)). After 50 fs, the charge of −0.95 |e| was added to the TFSI⁻ anion, triggering N–S bond breakage. Subsequently, SO₂CF₃⁻ decomposed into CF₃⁻ and SO₂⁻ fragments because of the acquisition of −1.05 |e| at 200 fs. This CF₃⁻ group was further decomposed into CF₂⁻ and F⁻ with the cleavage of the C–F bond at 400 fs, eventually combined with Li to form LiF. The aforementioned reactions can be summarized as follows.

N(SO₂CF₃)₂⁻ + e⁻ → NSO₂CF₃⁻ + SO₂CF₃⁻, (1)

SO₂CF₃⁻ + e⁻ → SO₂⁻ + CF₃⁻, (2)

CF₃⁻ + e⁻ → CF₂⁻ + F⁻, (3)

F⁻ + Li⁺ → LiF. (4)

This LiTFSI decomposition mechanism aligned well with the simulation results.²⁶ In sharp contrast, pure COF-5 exhibited much smaller charge transfers to the TFSI⁻ anion than COF-5@Ni, resulting in a difficult decomposition of LiTFSI with a significantly long duration (Fig. 1(f)).^26–28

2.2 Synthesis and characterization of COF-5@Ni SSE

As illustrated in Fig. 2(a) and Fig. S2, the COF-5@Ni SSE was prepared by the in situ polymerization of vinyl ethylene carbonate (VEC) and poly-(ethylene glycol) diacrylate (PEGDA) monomers (named p(VG)) on the COF-5@Ni-coated PP separator. As a control, the COF-5 SSE was also prepared by the copolymerization of VEC and PEGDA on the pure-COF-5-coated PP separator. First, the structure and chemical coordination of COF-5@Ni were investigated in detail. The powder X-ray diffraction (XRD) pattern shows four diffraction peaks at 3.4, 5.8, 6.7, and 9.0, corresponding to the (100), (110), (200), and (210) planes of COF-5, respectively (Fig. 2(b) and Fig. S3).²³ The similar XRD patterns of COF-5@Ni and COF-5 implied that the crystalline structure was well preserved after the single Ni atom coordination. The Fourier transform infrared (FT-IR) spectra showed the formation of the expected boron-based ring groups in COF-5 and COF-5@Ni (Fig. S4). Additionally, the obvious shift from 1342.2 cm⁻¹ to 1348.1 cm⁻¹ for the B–O band indicated the successful coordination of the Ni atom with O on COF-5. The Brunauer–Emmett–Teller (BET) surface areas of COF-5 and COF-5@Ni were calculated to be 949.6 m² g⁻¹ and 359.8 m² g⁻¹, with pore sizes of 2.3 nm and 2.1 nm, respectively (Fig. S5 and Table S1). The decreased specific surface area and pore size confirmed the successful implantation of single Ni atoms into the channels of COF-5.^29,30 Additionally, a subtle framework collapse inevitably occurred during the synthesis process in a humid air environment. The scanning electron microscopy (SEM) images showed that the morphology remained unchanged after introducing the Ni atoms (Fig. S6). The transmission electron microscopy (TEM) image of COF-5@Ni displayed the regular fringes corresponding to the pore size of 2.1 nm (Fig. S7). Furthermore, the aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) image showed many bright dots, highlighting the atomically dispersed single Ni atoms in the COF-5 framework (Fig. 2(c)).³¹ The energy-dispersive X-ray (EDX) mapping images depicted the uniform distribution of C, B, O, and Ni in the COF-5@Ni (Fig. 2(d) and Fig. S8).³²

Fig. 2 (a) Synthesis of the COF-5@Ni electrolyte. (b) XRD spectra of COF-5@Ni and COF-5. (c) AC-HAADF-STEM image and (d) EDX element mappings of COF-5@Ni. (e) Ni K-edge XANES spectra and (f) Ni K-edge EXAFS spectra of COF-5@Ni, Ni foil, and NiO. (g) EXAFS fitting curve of COF-5@Ni in R-space (inset is the B–O–Ni coordination structure model in COF-5@Ni). WT for the κ³-weighted Ni K-edge EXAFS signals of (h) COF-5@Ni, (i) Ni foil, and (j) NiO. (k) XPS Ni 2p spectrum of COF-5@Ni.

X-ray absorption fine structure spectroscopy (XAFS) was employed to explore the dispersion and coordination structure of Ni in COF-5@Ni. Fig. 2(e) shows the Ni K-edge X-ray absorption near-edge structure (XANES) spectra of COF-5@Ni with Ni foil and NiO as the references. The absorption edge position of COF-5@Ni was slightly higher than that of NiO, indicating that the valence state of Ni in COF-5@Ni was slightly higher than +2. The Fourier-transform of the extended X-ray absorption fine structure (FT-EXAFS) plots is displayed in Fig. 2(f). The COF-5@Ni showed a main peak at ∼1.61 Å, in accordance with the coordination of Ni–O.³³ Moreover, no peak of the Ni–Ni band at ∼2.18 Å was detected in COF-5@Ni, further confirming that the Ni sites in COF-5@Ni were atomically dispersed. The coordination structure near the Ni atom was analysed via fitting the EXAFS spectra (Fig. 2(g) and Fig. S9–S11). The result illustrates that Ni atoms are fixed at the atomic level in the COF-5 and are coordinated with four O atoms with an average Ni–O bond length of 2.06 Å, thereby forming the B–O–Ni bridge (Table S2). The inset of Fig. 2(g) shows the B–O–Ni coordination structure model in COF-5@Ni. Furthermore, the wavelet transform (WT) in the Ni K-edge EXAFS was employed to further investigate the coordination environment of Ni sites in COF-5@Ni. Different from Ni foil and NiO (Fig. 2(i) and (j)), the WT contour plot of COF-5@Ni (Fig. 2(h)) showed that there was only one intensity maximum around 5.8 Å⁻¹ (attributed to Ni–O coordination), showcasing that Ni sites were atomically distributed throughout COF-5@Ni. The coordination environment of the Ni site was further explored by X-ray photoelectron spectroscopy (XPS). As shown in B 1s spectra (Fig. S12), an obvious shift of the B–O peak to higher binding energy can be detected in COF-5@Ni compared with that in COF-5, which was due to the coordination of O atoms to Ni ions, in alignment with the EXAFS analysis. Additionally, the Ni 2p XPS spectrum of COF-5@Ni exhibited two peaks at 856.3 eV and 874.2 eV corresponding to Ni 2p_3/2 and Ni 2p_1/2, respectively (Fig. 2(k)). The binding energies of Ni 2p_3/2 and Ni 2p_1/2 are slightly higher than that of Ni²⁺ 2p_3/2 (856.2 eV) and Ni²⁺ 2p_1/2 (873.8 eV), suggesting that the valence of the Ni atom in COF-5@Ni was near +2.³⁴ The inductively coupled plasma mass spectrometry (ICP-MS) revealed that the loading amount of single Ni atoms was 4.2 wt% in COF-5@Ni. Correspondingly, the density of the active site of the single Ni atoms in COF-5@Ni was calculated to be 4.3 × 10²⁰ atoms g⁻¹ (7.16 × 10⁻⁴ mol g⁻¹) and 1.2 × 10¹⁴ atoms cm⁻².

As shown in Fig. S13a, the COF-5@Ni was coated on the surface of the PP separator uniformly and densely via vacuum filtration. The thickness was ∼1 µm (Fig. S13b). The minimal thickness fluctuation over a large scale of ∼200 µm further revealed the excellent thickness uniformity of the COF-5@Ni-coated PP separator (Fig. S14). Then, VEC and PEGDA (crosslinker) were in situ polymerized on the as-prepared membrane to obtain composite SSE. The photographs in Fig. S15 validate the successful solidification of the precursor into a gel after heating at 70 °C. FT-IR was utilized to analyse the polymerization reactions. The C=C vibrating peaks of VEC (1650.1 cm⁻¹) and PEGDA (1636.1 cm⁻¹) almost disappeared after polymerization (Fig. S16), indicating that the monomers were successfully converted into a cross-linking polymer.³⁵ The thermogravimetric analysis (TGA) analysis revealed that the residual monomer content was ∼50.5% (Fig. S17). Furthermore, the XRD pattern in Fig. S18 shows an amorphous state of p(VG) polymer, which facilitates the segmental mobility and Li⁺ migration in the electrolyte.³⁶

2.3 Analysis of Li⁺ transport kinetics

The FT-IR spectra showed that p(VG) and COF-5 SSEs exhibited the same O=S=O stretching peak (1349.0 cm⁻¹) and –CF₃ asymmetric stretching peak (1177.4 cm⁻¹) of LiTFSI, suggesting that there was no obvious interaction between COF-5 and LiTFSI. In stark contrast, the obvious shift of the –CF₃ and O=S=O stretching peaks in COF-5@Ni SSE implied a robust interaction between LiTFSI and COF-5@Ni (Fig. 3(a)).³⁷ As shown in Fig. 3(b), the peak at 745.0 cm⁻¹ corresponds to undissociated Li-TFSI ion clusters, while the peak at 740.4 cm⁻¹ represents the TFSI⁻ anions that successfully dissociate from LiTFSI. In comparison to COF-5, the COF-5@Ni showed more dissociated TFSI⁻ anions and fewer undissociated Li–TFSI, pointing to a higher dissociation degree of lithium salt because of the electrostatic interaction between COF-5@Ni and TFSI⁻ anions.³⁸ Cyclic voltammetry (CV) test was performed to explore the redox reaction of SSE in the SEI formation. As shown in Fig. 3(c), the decomposition of the electrolyte happens over a broad voltage range from 1.7 to 0.2 V (vs. Li/Li⁺). The TFSI⁻ anions decomposed in the high-voltage region around 1.4 V, while polymer chain reduction occurred in the low-voltage region below 1.0 V. Notably, COF-5@Ni delivered a relatively high peak intensity and wide reduction area in the high-voltage region, alongside a diminished peak intensity in the low-voltage area, revealing that the B–O–Ni bridge significantly accelerated the decomposition kinetics of LiTFSI.^26,28,39 Furthermore, in situ electrochemical Raman spectroscopy was conducted to track the electrolyte decomposition (Fig. 3(d) and Fig. S19). The peak at ∼742 cm⁻¹ corresponded to the TFSI⁻ anion. Compared with the slow decay on COF-5, the TFSI⁻ signal attenuated more quickly on COF-5@Ni as the electrodeposition progressed, demonstrating the rapid decomposition kinetics of TFSI⁻ on COF-5@Ni.^40,41 The enhanced decomposition of the TFSI⁻ anions promotes the construction of inorganic-rich SEI and thus inhibits the undesirable decomposition of the polymer chain.

Fig. 3 (a) FT-IR spectra of the COF-5@Ni, COF-5, and p(VG) electrolytes. (b) Raman spectra of the COF-5@Ni and COF-5 electrolytes. (c) CV curves of the Li||Cu cells with the COF-5@Ni and COF-5 electrolytes at a scan rate of 0.1 mV s⁻¹. (d) In situ Raman spectroscopy during the initial discharge process on the COF-5@Ni and COF-5 electrolytes. (e) Arrhenius plots of the COF-5@Ni and COF-5 electrolytes. (f) Chronoamperometry polarization curve and the impedance spectra before and after the polarization of the Li|COF-5@Ni|Li symmetric cell. (g) Summarized σ_Li⁺ and t_Li⁺, (h) LSV curves, (i) ⁷Li solid NMR, and (j) Tafel plots of the COF-5@Ni and COF-5 electrolytes. (k) CCD tests of the Li||Li symmetric cells with the COF-5@Ni and COF-5 electrolytes.

The electrochemical impedance spectroscopy (EIS) over a temperature range of 30–70 °C was carried out to evaluate the temperature-dependent ionic conductivity. As shown in Fig. S20, the Li⁺ conductivity (σ_Li⁺) of the COF-5@Ni electrolyte is calculated to be 7.4 × 10⁻⁴ S cm⁻¹ at 30 °C, which is 3.2 times that of COF-5 (2.3 × 10⁻⁴ S cm⁻¹). Correspondingly, Fig. 3(e) shows that the activation energy (E_a) of COF-5@Ni is 0.23 eV, which is much lower than that of COF-5 (0.40 eV). Moreover, the Li⁺ transference number (t_Li⁺) was tested at room temperature (RT) to examine the selective ionic conduction of the SSEs (Fig. 3(f) and Fig. S21). COF-5@Ni exhibited an impressive t_Li⁺ of 0.83, which was 2.6 times that of COF-5 (0.32). The enhanced t_Li⁺ resulted from the anchoring effect of TFSI⁻ anions by the single Ni sites on COF-5@Ni. The σ_Li⁺ and t_Li⁺ of COF-5@Ni and COF-5 SSEs are summarized in Fig. 3(g). Linear sweep voltammetry (LSV) measurements revealed that COF-5@Ni provided an expanded electrochemical stability window (ESW) of 5.0 V, owing to the diminished interfacial overpotential and the enhanced electrode–electrolyte interfaces (Fig. 3(h)).^9,42 The local chemical environments of COF-5@Ni interacting with LiTFSI were further assessed via⁷Li solid-state nuclear magnetic resonance (NMR). As depicted in Fig. 3(i), the chemical shift of ⁷Li for COF-5@Ni (δ = −0.02 ppm) is considerably downfield relative to that of pure COF-5 (δ = −0.59 ppm), implying a diminished electron cloud density around the Li atom and looser coordination with the electron-donating nitrogen in TFSI⁻, highlighting that the single Ni sites on COF-5 play a vital role in dissociating lithium salts and promoting rapid Li⁺ transport.⁴³ Tafel curve analysis showed that the exchange current density (j₀) for COF-5@Ni (0.42 mA cm⁻²) was ∼2.3 times that of COF-5 (0.18 mA cm⁻²), suggesting an enhanced Li⁺ interface transport kinetics (Fig. 3(j)).⁴⁴ As shown in Fig. 3(k), Li|COF-5@Ni|Li cell shows a large critical current density (CCD) of 2.2 mA cm⁻², significantly surpassing that of COF-5 (0.6 mA cm⁻²). The improved CCD can be attributed to more effective Li⁺ migration and the enhanced EEI.

2.4 Analysis of Li plating/stripping behavior

The Li||Li symmetric cells and Li||Cu half-cells were assembled to investigate the Li plating/stripping behaviors. As shown in Fig. 4(a), the Li|COF-5@Ni|Cu half-cell exhibits a stable cycle over 300 cycles with an average CE of ∼98.4% at a current density and capacity of 0.5 mA cm⁻² and 1.0 mA h cm⁻², while the Li|COF-5|Cu cell breaks down after 250 cycles with a much lower average CE of ∼89.6%. Correspondingly, the Li plating curves showed that the nucleation overpotential at the initial cycle was only 15.1 mV for COF-5@Ni (Fig. 4(b)), which was considerably lower than that of COF-5 (30.6 mV), indicating that the B–O–Ni bridge can effectively decrease the nucleation barrier of lithium metal on Cu foil. As a result, COF-5@Ni achieved a high Li plating/stripping CE (such as CE_240th = 99.2%) with a low and stable overpotential in the following Li||Cu half-cell test, significantly outperforming COF-5 (Fig. 4(c)). Furthermore, another Li|COF-5@Ni|Cu half-cell tested under the same conditions showed an average CE of ∼98.3% over 300 cycles (Fig. S22). As shown in Fig. S23, compared with the smooth and seamless surface of Li-deposited Cu foil by COF-5@Ni, the Cu foil deposited by COF-5 displays a coarse surface with numerous Li dendrites, resulting in continuous side reactions and cell failure. Benefiting from the excellent lithium plating/stripping reversibility, the Li|COF-5@Ni|Li symmetric cell exhibited steady cycle over 1500 h with a small polarization voltage of ∼120 mV at 0.2 mA cm⁻² and 0.2 mA h cm⁻² (Fig. 4(d)). In stark contrast, the Li|COF-5|Li cell suffered a sharp increase in polarization voltage until a short circuit occurred after ∼310 h. The EIS results showed that the interfacial resistance of the Li|COF-5@Ni|Li symmetric cell was much less than that of Li|COF-5|Li after cycling (Fig. 4(e)), showcasing improved interfacial compatibility and stability between COF-5@Ni and the Li anode. As mentioned above, the smooth and highly reversible Li plating/stripping behavior was due to the construction of an inorganic-rich SEI film by the catalysis of LiTFSI decomposition, which not only boosted the electrode–electrolyte interface stability but also enhanced the ion transport capability.

Fig. 4 (a) Coulombic efficiency of the Li||Cu half-cells with the COF-5@Ni and COF-5 electrolytes under a current density of 0.5 mA cm⁻² and Li deposition capacity of 1.0 mA h cm⁻². (b) Corresponding initial Li plating profiles. (c) Selected Li plating/stripping curves. (d) Voltage–time plots at 0.2 mA cm⁻²/0.2 mA h cm⁻². (e) EIS curves of the Li||Li symmetric cells with the COF-5@Ni and COF-5 electrolytes after 300 cycles. AFM images of the Li metal anodes in the (f) and (g) Li|COF-5@Ni|Li and (h) and (i) Li|COF-5|Li cells after cycling. Finite element simulations of Li deposition and electric field distribution within the Li||Li symmetric cells with the (j) COF-5@Ni and (k) COF-5 electrolytes. In situ optical images of the Li deposition on the stainless-steel electrode with the (l) COF-5@Ni and (m) COF-5 electrolytes at 2.0 mA cm⁻².

An atomic force microscopy (AFM) test was performed to visually analyze the surface morphology and mechanical properties of the SEI formed on lithium metal. Compared with the coarse and random surface with a large average roughness of 39.7 nm for COF-5 (Fig. 4(h)), the SEI film formed by COF-5@Ni SSE presented a smooth surface with a much lower average roughness of 9.1 nm (Fig. 4(f)), highlighting a uniform and thin SEI film on the lithium anode. In addition, the Young's modulus of SEI was tested to be 6.2 GPa for COF-5@Ni SSE (Fig. 4(g)), which was 2.3 times (2.7 GPa) that of pure COF-5 (Fig. 4(i)), showcasing an enhanced interface mechanical strength due to the formation of inorganic-rich SEI. This Young's modulus value exceeds the threshold value of 6.0 GPa for Li dendrite growth, indicating that the COF-5@Ni-originated SEI can provide the desired mechanical stability during Li plating/stripping, as well as sufficient strength to suppress the Li dendrite growth.^45,46 Numerical simulations were utilized to investigate the deposition behavior of Li⁺ at the lithium anode interface. The electric field distribution simulation with time demonstrated that COF-5@Ni facilitated the uniform distribution of electric field intensity and significantly diminished the tip effect, thereby guaranteeing a uniform nucleation of Li⁺ on both the protrusions and flat regions of the substrate, leading to more homogeneous and rapid Li⁺ deposition (Fig. 4(j)).⁴⁷ The simulations of interfacial Li⁺ concentration (Fig. S24) and current density (Fig. S25) depicted the Li⁺ migration in the electrolyte, further revealing that COF-5@Ni enabled a more dispersed Li⁺ deposition pathway rather than agglomerating at the tip, resulting in uniform Li⁺ deposition. The simulation results clearly correspond with the morphological characterizations. The in situ optical microscopy was employed to dynamically monitor the interfacial evolution during Li deposition. As shown in Fig. 4(l), the surface of the stainless steel (SS) electrode remains stable and smooth with dense Li deposition in Li|COF-5@Ni|SS half-cell throughout Li plating. In stark contrast, the Li|COF-5|SS half-cell displayed severe Li dendrites with loose and rough Li deposition (Fig. 4(m)), which led to a high residue of dead Li and poor CE during the battery cycling.⁴⁴

2.5 Analysis of Li plating/stripping behavior

Solid-state lithium metal cells paired with the NCM622 cathode were assembled to assess the practicality of the COF-5@Ni SSE and measured at 30 °C. As shown in Fig. 5(a), the NCM622|COF-5@Ni|Li cell exhibits an excellent reversible capacity of 126.0 mA h g⁻¹ at 1.0C after 1500 cycles under a cutoff voltage of 4.3 V, with a remarkable capacity retention of 92.0% and a capacity decay of only 0.005% per cycle. Conversely, the capacity declined to 53.6 mA h g⁻¹ after 600 cycles for the NCM622|COF-5|Li cell. Impressively, even at a cutoff voltage of 4.7 V, the NCM622|COF-5@Ni|Li cell achieved a high reversible capacity of 172.4 mA h g⁻¹ at 0.2C after 200 cycles with a capacity retention of 85.8% (Fig. 5(b)), highlighting its great potential for high-specific-energy SSLMBs. The impressive cycling performance at high voltage was attributed to the construction of the B, F-rich CEI, which could effectively inhibit the dissolution of transition metal ions, mitigate the oxidation decomposition of electrolyte, and ensure the structural integrity of the NCM cathodes upon cycling.^48,49 Correspondingly, the capacity–voltage curve of the NCM622|COF-5@Ni|Li cell indicated its reliable cycling without obvious polarization at the cutoff voltage of 4.3 V and 4.7 V (Fig. S26 and S27). Furthermore, NCM622|COF-5@Ni|Li delivered specific capacities of 195.7, 186.0, 166.9, 148.4, and 127.5 mA h g⁻¹ at stepwise current densities of 0.1, 0.2, 0.5, 1.0, and 2.0C, respectively, and the capacity was reversible upon returning to 0.1C (Fig. 5(c)). Conversely, the NCM622|COF-5|Li cell exhibited relatively low capacities at various current densities, with a capacity of merely 57.6 mA h g⁻¹ at 2.0C. The charge/discharge plateaus of NCM622|COF-5@Ni|Li always remain stable with small polarization, owing to the rapid Li⁺ transport (Fig. 5(d)). The EIS profiles showed that the charge transfer resistance of NCM622|COF-5@Ni|Li was notably lower than that of NCM622|COF-5|Li after the rate test (Fig. S28), suggesting the rapid Li⁺ migration and superior interface compatibility of COF-5@Ni.^50–52 With a high loading of up to 21.6 mg cm⁻², the NCM622|COF-5@Ni|Li cell also provided a stable cycle with a reversible capacity of 141.8 mA h g⁻¹ after 50 cycles at 0.2C (Fig. 5(e)). The cycling performance was much higher than that of COF-5 electrolyte, showcasing its excellent potential for practical application. Furthermore, the NCM622|COF-5@Ni|Li cell was tested at −20 °C to probe the cycling performance of COF-5@Ni under extreme conditions. As displayed in Fig. 5(f), the NCM622|COF-5@Ni|Li cell delivers a high reversible capacity of 114.4 mA h g⁻¹ at 0.1C after 100 cycles, with a capacity retention of 80.3%, demonstrating an exceptional Li-storage capability at a low temperature. The capacity–voltage profiles showed that the NCM622|COF-5@Ni|Li cell exhibited a high charge–discharge reversibility with a slight polarization increase at −20 °C (Fig. S29). The EIS at −20 °C was conducted to assess the ionic conductivity of COF-5@Ni at a low temperature. As shown in Fig. S30, COF-5@Ni exhibits an ionic conductivity of 2.1 × 10⁻⁴ S cm⁻¹ at −20 °C, which is near the ionic conductivity of the COF-5 electrolyte at room temperature. The excellent ionic conductivity at −20 °C guaranteed the effective ion transport and stable cycling of the NCM622||Li cell at a low temperature. The cycling stability and capacity retention of the as-prepared COF-5@Ni SSE are among the top when compared with those of previously reported COF-based composite SSEs (Fig. 5(g) and Table S3). As shown in Fig. S31, many bright and isolated dots, which refer to single Ni atoms, are clearly observed after long-term cycling, indicating the firm anchoring of the single Ni atoms on the COF-5 framework. Furthermore, the Ni 2p XPS spectrum of COF-5@Ni after cycling exhibited two peaks located at 856.3 eV and 874.2 eV, corresponding to Ni 2p_3/2 and Ni 2p_1/2, respectively (Fig. S32). The peaks remained unchanged compared with the fresh COF-5@Ni (Fig. 2(k)), indicating that the valence of Ni had not changed and remained near +2.

Fig. 5 (a) Cycling performance of the NCM622||Li cells with the COF-5@Ni and COF-5 electrolytes at 1.0C. (b) Cycling performance of the NCM622|COF-5@Ni|Li cell at 0.2C with a cutoff voltage of 4.7 V. (c) Rate performance of the NCM622||Li cells with the COF-5@Ni and COF-5 electrolytes at current densities from 0.1 to 2C. (d) Corresponding charge–discharge profiles. (e) Cycling performance of the NCM622||Li cells with the COF-5@Ni and COF-5 electrolytes at a high loading. (f) Cycling performance of the NCM622|COF-5@Ni|Li cell at −20 °C. (g) Comparison of the SSE reported in this work with other reported COF-based composite SSEs.

2.6 Interfacial stability and cycling performance of AFSSLMBs

The interfacial evolution and electrode structure were investigated to elucidate the mechanisms of enhanced Li-storage performance. XPS depth profiling was carried out on the lithium metal surface after 200 plating/stripping loops to characterize the composition of SEI. As depicted in Fig. S33b, the C 1s spectra show that the primary components of SEI from COF-5 are the polymer-derived organic species (C–C/C–H at 284.8 eV, C–O at 286.7 eV, C=O at 288.5 eV, and –CF₃ at 292.8 eV).^53–55 Concurrently, a large number of incomplete LiTFSI decomposition-derived inorganic compounds are detected in Fig. 6(b) and Fig. S34b, such as N–S species (399.5 eV in N 1s spectra) and S_xO_yⁿ⁻ (168.3 eV in S 2p spectra). However, only limited amounts of complete LiTFSI decomposition-derived inorganic compounds are observed, including LiF (685.0 eV in F 1s spectra), Li₂O (528.2 eV in O 1s spectra), Li₃N (395.5 eV in N 1s spectra), and Li₂S (160.7 eV in S 2p spectra). This confirmed the poor catalyzing capability of pure COF-5 in promoting the decomposition of LiTFSI into the desired SEI components (LiF, Li₂O, Li₃N, and Li₂S). In sharp contrast, the SEI generated from COF-5@Ni predominantly consisted of complete LiTFSI decomposition-derived species, such as LiF, Li₂O, Li₃N, and Li₂S (Fig. 6(a) and Fig. S33a, S34a), with only a limited amount of incomplete LiTFSI decomposition-derived N–S and S_xO_yⁿ⁻ components. The detailed ratios of the inorganic/organic species in SEI are summarized in Fig. S35. Compared with the SEI formed by the COF-5 electrolyte, the COF-5@Ni-induced SEI exhibited a significantly lower proportion of organic C and a higher content of LiF, Li₂O, Li₃N, and Li₂S species at different etching depths, with internal organic C and LiF content being ∼24.5% and 23.4%, compared with 52.8% and 8.7% for COF-5. The sharp contrast highlights the formation of an inorganic-rich SEI. Furthermore, the intensities of LiF and Li₂O using COF-5@Ni markedly increased with sputtering depth, whereas the C peaks intensities dropped sharply, showcasing a dual-layer SEI featuring an amorphous organic outer layer and an inorganic-rich inner layer. The thin organic outer layer can enhance the mechanical strength, while the inorganic-rich inner layer promotes homogeneous Li⁺ deposition and fast Li⁺ transport, thereby enabling dendrite-free lithium deposition and durable interfacial stability.⁵⁶

Fig. 6 XPS F 1s, N 1s, and S 2p depth profiles of the Li anodes cycled in the (a) NCM622|COF-5@Ni|Li and (b) NCM622|COF-5|Li cells. (c) TOF-SIMS 3D mappings in the formed SEI using the COF-5@Ni (up) and COF-5 (bottom) electrolytes. (d) Corresponding TOF-SIMS depth profiles of various segments. Cryo-TEM images of SEI on the Li anode surface formed by the COF-5@Ni electrolyte at (e) low magnification and (f) high resolution. (g) EDX mappings of the surface of the deposited Li. (h) Cryo-TEM image of SEI on the Li anode surface formed using the COF-5 electrolyte.

Time-of-flight secondary-ion mass spectrometry (TOF-SIMS) was conducted to further analyze the surface components and depth distribution of SEI. The three-dimensional (3D) reconstruction images and two-dimensional (2D) mappings depicted the distribution of the organic and inorganic constituents in SEI, with the LiF₂⁻, LiO⁻, Li₃N⁻, and LiS⁻ fragments representing the inorganic component of LiF, Li₂O, Li₃N and Li₂S, respectively, while the C₂HO⁻ fragment denoted the organic species (Fig. 6(c) and Fig. S36, S37). Clearly, for COF-5, relatively few signals from the inorganic species, such as LiF₂⁻, LiO⁻, Li₃N⁻, and LiS⁻, could be detected in the SEI layer, whereas the signals for organic ones, such as C₂HO⁻, were much stronger with prolonged sputtering time, indicating a thick SEI layer predominantly consisting of organic components. In contrast, the SEI from COF-5@Ni exhibited a considerably high and uniform signal intensity of LiF₂⁻, LiO⁻, Li₃N⁻, and LiS⁻, along with a relatively low signal intensity of C₂HO⁻, demonstrating a dense and thin inorganic-rich SEI layer that could effectively suppress interface side reactions and electrolyte decomposition. The depth profile presented in Fig. 6(d) further confirms the distribution of typical C₂HO⁻, LiF₂⁻, and LiO⁻ fragments, corresponding with the 3D views and XPS analysis.

To further investigate the determinants of the enhanced Li-storage performance, cryo-TEM was employed to examine the structure and composition of SEI at the atomic level. Lithium metal was plated onto a Cu grid with a capacity of 0.2 mA h cm⁻² at a current density of 0.4 mA cm⁻² for the cryo-TEM test. As shown in Fig. 6(e), the generated SEI using COF-5@Ni predominantly consists of inorganic components, with a considerable amount of LiF, Li₂O, Li₃N, and Li₂S species, while organic components (amorphous phase) are scarce. Specifically, the primary crystalline lattices with calibrated interplanar spacings of 2.32 Å, 2.66 Å, 3.90 Å, and 3.30 Å are recorded from the high-resolution images, corresponding to the LiF(111), Li₂O(111), Li₃N(110), and Li₂S(111) planes, respectively (Fig. 6(f)). This configuration facilitates Li⁺ transport through SEI and restrains uncontrolled dendrite growth.⁵⁷ The EDX mapping also shows the uniform distribution of the C, N, F, O, and S elements throughout the SEI (Fig. 6(g) and Fig. S38). In sharp contrast, the SEI from COF-5 contained a relatively large proportion of organic components and few inorganic nanocrystals (Fig. 6(h)). These results are consistent with the theoretical simulation and experimental characterizations, showcasing a significantly enhanced catalytic ability of the B–O–Ni bridge in the decomposition reactions of LiTFSI.

The surface-phase change from a layered (Rm) structure to a rock-salt (Fmm) arrangement is acknowledged to result in the serious structural deterioration of the NCM cathode during cycling. In contrast to the layered structure, the rock-salt phase was not electrochemically active and ionically conductive, resulting in rapid capacity decay and sluggish Li⁺ transport. The rock-salt phase emerged from the reduction of Ni³⁺/Ni⁴⁺ with the electrolyte at a high voltage, subsequently accompanied by the mixed arrangement of Li⁺/Ni²⁺ during the lithiation/delithiation process.⁵⁸ To elucidate the degradation of the NCM622 cathode, TEM was utilized to monitor the lattice-phase change after cycling. Compared with the rough CEI layer with a thickness of ∼11 nm for the COF-5 electrolyte (Fig. 7(b)), a smooth and uniform CEI with a thickness of ∼4.5 nm formed on the NCM622 surface when cycled with the COF-5@Ni electrolyte (Fig. 7(a)), highlighting an efficient suppression of interfacial side reactions. Finally, the bulk and surface of NCM622 with the COF-5@Ni electrolyte preserved their initial layered structure, as confirmed by the fast Fourier transform (FFT) patterns and high-resolution TEM images. However, two separate areas were observed in the NCM622 cycled with the COF-5 electrolyte. In addition to the original layered structure in the interior (region I), a drastic phase transition to the rock-salt phase was observed on the NCM622 surface (region II). The sharp contrast implied that the COF-5@Ni-derived CEI could prevent the formation of the harmful rock-salt phase, thereby sustaining the structural stability to alleviate the capacity decay under high voltage.⁵⁹ As a result, the morphology of the NCM622 particles was integrally preserved after cycling by the COF-5@Ni electrolyte (Fig. S39a), while severe fragmentation occurred for the COF-5 electrolyte (Fig. S39b).

Fig. 7 TEM images and corresponding FFT patterns for the NCM622 particles cycled in the (a) NCM622|COF-5@Ni|Li and (b) NCM622|COF-5|Li cells. XPS F 1s, B 1s, and Ni 2p depth profiles of the cycled NCM622 cathodes in the (c) NCM622|COF-5@Ni|Li and (d) NCM622|COF-5|Li cells. (e) Cycling performance of the anode-free NCM622||Cu cells with the COF-5@Ni and COF-5 electrolytes at 0.5C. (f) Cycling performance of the anode-free NCM622|COF-5@Ni|Cu pouch cell at 0.2C. (g) Corresponding charge–discharge profiles of the NCM622|COF-5@Ni|Cu pouch cell. (h) Comparison of this work with other state-of-the-art reported works on the anode-free lithium metal batteries.

XPS depth profiling was carried out to assess the CEI on the cycled NCM622 cathode after cycling (Fig. 7(c), (d), and Fig. S40). The relative concentrations of organic C, LiF and B-species are summarized in Fig. S41. Compared with that of COF-5, the generated CEI using COF-5@Ni had a much lower concentration of organic C species at different etching depths, showcasing the effective suppression of the electrolyte decomposition. Moreover, the NCM622 cycled with COF-5@Ni displayed a significantly higher intensity of LiF compared with that of COF-5 in the F 1s depth profiling, with an inner content exceeding 27.3%, compared with only 7.9% for COF-5, which contributed to preventing electron tunneling and enhancing CEI stability. Impressively, the concentration of the B-species (B–O and B–F species) in the CEI with COF-5@Ni (2.6% after etching for 150 s) was higher than that with COF-5 (0%), implying that more B-groups in COF-5 participated in the formation of CEI for COF-5@Ni, owing to the activation of COF-5 by the single Ni sites. The B–O species are widely recognized for suppressing the dissolution of transition metals, while the B–F species can capture harmful H⁺ and O⁻ in the electrolytes.^42,60 Additionally, reduced Ni⁰ species were observed in the CEI formed by the COF-5 electrolyte at different etching depths, while no signal was detected in the CEI formed by COF-5@Ni, further proving the prohibited formation of the rock-salt phase. ICP-MS analyses were conducted to quantify the dissolution of TM ions. As illustrated in Fig. S42, the amounts of Ni, Co, and Mn elements that dissolved into the Li anode in the NCM622|COF-5@Ni|Li cell are significantly lower than those in the NCM622|COF-5|Li cell after cycling, suggesting the inhibited TM dissolution and cathode deterioration.

Inspired by the excellent lithium plating/stripping reversibility of the Li||Cu half-cell and the cycling stability of the NCM622||Li full-cells, the anode-free NCM622|COF-5@Ni|Cu full-cells were assembled to assess the potential of the COF-5@Ni SSE in AFSSLMBs. Following two cycles of activation at 0.1C, the NCM622|COF-5@Ni|Cu cell achieved a specific capacity of 164.8 mA h g⁻¹ at 0.5C, and 64.5% of its initial capacity was retained after 200 cycles (Fig. 7(e)). In stark contrast, the NCM622|COF-5|Cu suffered a rapid capacity decay to 28.4 mA h g⁻¹ after 42 cycles with huge fluctuations of CE under the same conditions. Another anode-free NCM622|COF-5@Ni |Cu cell was assembled and tested under the same conditions to demonstrate the reproducibility. As shown in Fig. S43, the cell delivers a specific capacity of 160.8 mA h g⁻¹, with 64.2% of its initial capacity retained after 200 cycles, showcasing the superior reproducibility of the COF-5@Ni electrolyte for the anode-free cells. In view of the superior performance of the anode-free coin cell, the practical feasibility of COF-5@Ni was further validated in a homemade scale-up pouch cell. As shown in Fig. 7(f), the anode-free NCM622|COF-5@Ni|Cu pouch cell with an energy density of 405 Wh kg⁻¹ demonstrates a superior cycling stability over 200 cycles at 0.2C, with a remarkable capacity retention of 65.4% and an average CE of 97.5%, indicative of robust cycling stability and long-term cycling lifespan for successful implementation in AFSSLMBs. The overview of the NCM622|COF-5@Ni|Cu pouch cell is presented in Table S4. Despite a slight increase in voltage polarization over 200 cycles, the capacity–voltage profiles revealed the favorable charge–discharge reversibility of the anode-free NCM622|COF-5@Ni|Cu pouch cell (Fig. 7(g)). Notably, the COF-5@Ni electrolyte demonstrated a remarkable advantage compared with previously reported anode-free lithium metal batteries in terms of reversible capacity, cycle number, and retention, positioning it among the top-performing AFSSLMBs (Fig. 7(h) and Table S5).

To sum up, the results discussed above demonstrate that the incorporation of the single-atom sites into SSEs can effectively promote the multistep decomposition of lithium salt, thereby enabling an inorganic-rich SEI that guarantees the smooth Li deposition and excellent interface stability, ultimately boosting the Li-storage performance in anode-free solid-state batteries.

3. Conclusions

In summary, we developed an innovative single-atom catalysis strategy to facilitate lithium-salt decomposition and the construction of inorganic-rich electrode–electrolyte interfaces toward prolonged AFSSLMBs. Experiments and theoretical simulations confirm that the TFSI⁻ anions are effectively anchored by single Ni atoms via electrostatic interaction, giving rise to more free-moving Li⁺. Furthermore, the single Ni atoms as a catalyst boost the multistep decomposition of TFSI⁻ anions, thereby affording an inorganic-rich SEI to guarantee rapid Li⁺ transport and smooth Li deposition. Additionally, the boron-based COF-5-generated B, F-rich CEI inhibits the TM dissolution and ensures the structural integrity under high-voltage cycling. As a result, the assembled NCM622||Li cell delivers a remarkable capacity retention of 92.0% over 1500 cycles at 1.0C, and the cell is capable of stable cycling when paired with a 4.7 V NCM622 cathode. The anode-free pouch cell (405 Wh kg⁻¹) also demonstrated a steady cycle using the COF-5@Ni SSE, positioning it among the top-performing AFSSLMBs. Based on the impressive electrochemical performance of the COF-5@Ni electrolyte, the cost of synthesizing COF-5@Ni is estimated to be ∼$6.2 USD g⁻¹, indicating economic feasibility for use in practical AFSSLMBs. In the future, synthesis methods, such as solvent-free flux synthesis and the sonochemical method, are expected to scale COF-5@Ni. Compared to the building of artificial SEI and the introduction of functional additives, our proposed single-atom interface catalysis strategy enables the atomic-level precise control of interfacial chemistry, effectively avoiding the uncontrolled side reactions and excessive Li loss upon SEI/CEI film formation. Furthermore, the single-atom interface catalysis strategy allows for the favorable participation of commercial lithium salts (anions) in the formation of the inorganic-rich SEI/CEI film.

Author contributions

T. S. Zhao supervised this work. X. Xu conducted the synthesis and the experiments. X. Xu carried out the material characterizations and electrochemical measurements. X. Xu, J. Chen, J. Li, Z. Wang, Z. Guo, P. Lin, Y. Wang, J. Sun, and B. Huang performed the data analysis. X. Xu wrote the manuscript. T. S. Zhao, B. Huang, and J. Sun revised the manuscript. All authors discussed the results and checked the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been presented in the article and its supplementary information (SI) and are available from the corresponding author upon reasonable request. Supplementary information is available. See DOI: https://doi.org/10.1039/d5ee05317j.

Acknowledgements

This study was funded by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (project no. R6005-20). We would like to acknowledge Prof. Khalil Amine for providing valuable insights on the analysis of the interface chemistry and electrochemical performance.

References

1. Z. Ning, G. Li, D. L. R. Melvin, Y. Chen, J. Bu, D. Spencer-Jolly, J. Liu, B. Hu, X. Gao, J. Perera, C. Gong, S. D. Pu, S. Zhang, B. Liu, G. O. Hartley, A. J. Bodey, R. I. Todd, P. S. Grant, D. E. J. Armstrong, T. J. Marrow, C. W. Monroe and P. G. Bruce, Nature, 2023, 618, 287–293.
2. X. Xu, J. Chen, J. Li, J. Shen, P. Lin, Z. Wang, Z. Guo, J. Sun, B. Huang and T. Zhao, J. Am. Chem. Soc., 2025, 147, 25896–25909.
3. J.-G. Zhang, Nat. Energy, 2019, 4, 637–638.
4. W. Z. Huang, C. Z. Zhao, P. Wu, H. Yuan, W. E. Feng, Z. Y. Liu, Y. Lu, S. Sun, Z. H. Fu, J. K. Hu, S. J. Yang, J. Q. Huang and Q. Zhang, Adv. Energy Mater., 2022, 12, 2201044.
5. T. Zhang, K. Huang, Y. Zheng, D. Luo, W. Yan, J. Zhang and Z. Chen, Adv. Mater., 2025, 37, e05695.
6. Y. Zhou, P. Wang, K. Wang, X. Fang, W. Li, J. Nai, Y. Liu, Y. Wang, S. Zou, H. Yuan, X. Tao and J. Luo, Adv. Funct. Mater., 2025, 35, 2424022.
7. Z. Ouyang, S. Wang, Y. Wang, S. Muqaddas, S. Geng, Z. Yao, X. Zhang, B. Yuan, X. Zhao, Q. Xu, S. Tang, Q. Zhang, J. Li and H. Sun, Adv. Mater., 2024, 36, 2407648.
8. P. Li, H. Zhang, J. Lu and G. Li, Angew. Chem., Int. Ed., 2023, 62, e202216312.
9. M. Mao, X. Ji, Q. Wang, Z. Lin, M. Li, T. Liu, C. Wang, Y. S. Hu, H. Li, X. Huang, L. Chen and L. Suo, Nat. Commun., 2023, 14, 1082.
10. Y. Qiao, H. Yang, Z. Chang, H. Deng, X. Li and H. Zhou, Nat. Energy, 2021, 6, 653–662.
11. Y. Liu, X. Meng, Y. Shi, J. Qiu and Z. Wang, Adv. Mater., 2023, 35, 2305386.
12. Y.-G. Lee, S. Fujiki, C. Jung, N. Suzuki, N. Yashiro, R. Omoda, D.-S. Ko, T. Shiratsuchi, T. Sugimoto, S. Ryu, J. H. Ku, T. Watanabe, Y. Park, Y. Aihara, D. Im and I. T. Han, Nat. Energy, 2020, 5, 299–308.
13. J. Oh, S. H. Choi, H. Kim, J. Y. Kim, G.-J. Lee, K. Y. Bae, T. Lee, N. Lee, Y. Sohn, W. J. Chung and J. W. Choi, Energy Environ. Sci., 2024, 17, 7932–7943.
14. Z. Yu, H. Wang, X. Kong, W. Huang, Y. Tsao, D. G. Mackanic, K. Wang, X. Wang, W. Huang, S. Choudhury, Y. Zheng, C. V. Amanchukwu, S. T. Hung, Y. Ma, E. G. Lomeli, J. Qin, Y. Cui and Z. Bao, Nat. Energy, 2020, 5, 526–533.
15. J. Xiao, Q. Li, Y. Bi, M. Cai, B. Dunn, T. Glossmann, J. Liu, T. Osaka, R. Sugiura, B. Wu, J. Yang, J.-G. Zhang and M. S. Whittingham, Nat. Energy, 2020, 5, 561–568.
16. H. Xiao, X. Li and Y. Fu, Nano-Micro Lett., 2025, 17, 149.
17. Y. Xia, P. Zhou, X. Kong, J. Tian, W. Zhang, S. Yan, W.-H. Hou, H.-Y. Zhou, H. Dong, X. Chen, P. Wang, Z. Xu, L. Wan, B. Wang and K. Liu, Nat. Energy, 2023, 8, 934–945.
18. R. Weber, M. Genovese, A. J. Louli, S. Hames, C. Martin, I. G. Hill and J. R. Dahn, Nat. Energy, 2019, 4, 683–689.
19. T. R. Jow, K. Xu, O. Borodin and M. Ue, Electrolytes for Lithium and Lithium-Ion Batteries, Springer, 2014, vol. 58.
20. C. Jin, A. Xiang, Z. Wang, Q. He, B. Li, X. Zhang, Y. Xiang, P. Zhai and Y. Gong, Adv. Energy Mater., 2024, 15, 2402811.
21. A. Wang, J. Li and T. Zhang, Nat. Rev. Chem., 2018, 2, 65–81.
22. Z. Liang, J. Shen, X. Xu, F. Li, J. Liu, B. Yuan, Y. Yu and M. Zhu, Adv. Mater., 2022, 34, 2200102.
23. A. P. Côté, A. I. Benin, N. W. Ockwig, M. O’Keeffe, A. J. Matzger and O. M. Yaghi, Science, 2005, 310, 1166–1170.
24. C. Dong, C. Ma, C. Zhou, Y. Yu, J. Wang, K. Yu, C. Shen, J. Gu, K. Yan, A. Zheng, M. Gong, X. Xu and L. Mai, Adv. Mater., 2024, 36, 2407070.
25. X. Xu, F. Xu, X. Zhang, C. Qu, J. Zhang, Y. Qiu, R. Zhuang and H. Wang, Nano-Micro Lett., 2022, 14, 91.
26. Y. Liu, X. Tao, Y. Wang, C. Jiang, C. Ma, O. Sheng, G. Lu and X. W. Lou, Science, 2022, 375, 739–745.
27. W. Liang, X. Zhou, B. Zhang, Z. Zhao, X. Song, K. Chen, L. Wang, Z. Ma and J. Liu, Angew. Chem., Int. Ed., 2024, 63, e202320149.
28. L. Yang, H. Zhang, H. Xu, W. Peng, L. Wu, Y. Cheng, Y. Liu, L. Xu and L. Mai, Adv. Energy Mater., 2024, 14, 2401829.
29. L. Wang, J. Li, S. Ji, Y. Xiong and D. Wang, Energy Environ. Sci., 2024, 17, 8482–8528.
30. X. Xu, Y. Qiu, J. Wu, B. Ding, Q. Liu, G. Jiang, Q. Lu, J. Wang, F. Xu and H. Wang, Chin. J. Chem. Eng., 2021, 32, 416–422.
31. M. Zhou, Z. Wang, A. Mei, Z. Yang, W. Chen, S. Ou, S. Wang, K. Chen, P. Reiss, K. Qi, J. Ma and Y. Liu, Nat. Commun., 2023, 14, 2473.
32. X. Xu, X. Li, K. Liu, J. Li, Q. Feng, L. Zhou, F. Cui, X. Liang, Z. Lei, Z. Liu and H. Xu, Cryst. Growth Des., 2018, 18, 2844–2850.
33. F. Raziq, C. Feng, M. Hu, S. Zuo, M. Z. Rahman, Y. Yan, Q. H. Li, J. Gascon and H. Zhang, J. Am. Chem. Soc., 2024, 146, 21008–21016.
34. Y. Li, X. Cao, Q. Chen, R. Pan, J. Zhang, G. Meng, Y. Yang, Y. Li, J. Mao and W. Chen, Small, 2024, 20, 2405367.
35. K. Liu, H. Cheng, Z. Wang, Y. Zhao, Y. Lv, L. Shi, X. Cai, Z. Cheng, H. Zhang and S. Yuan, Adv. Energy Mater., 2024, 14, 2303940.
36. X. Xu, J. Chen, J. Li, Z. Wang, J. Shen, P. Lin, J. Sun, B. Huang and T. Zhao, Adv. Funct. Mater., 2024, 35, 2415298.
37. S. Qi, M. Li, Y. Gao, W. Zhang, S. Liu, J. Zhao and L. Du, Adv. Mater., 2023, 35, 2304951.
38. Y. Cheng, X. Liu, Y. Guo, G. Dong, X. Hu, H. Zhang, X. Xiao, Q. Liu, L. Xu and L. Mai, Adv. Mater., 2023, 35, 2303226.
39. C. Hao, X. Zhang, Z. He, M. Gao, Y. Liu, H. Pan and W. Sun, Angew. Chem., Int. Ed., 2025, 64, e202503691.
40. Z. Chen, J. Pan, W. Huang, K. Shi, Z. Yang, H. Wu, S. Wei, G. Jiang, W. Zou, R. Zhang, X. Li and Q. Liu, ACS Nano, 2025, 19, 13160–13174.
41. X. Song, C. Liu, A. Zhang, L. Ding, T. Zeng, Y. Lu, Y. Ou, W. Hou, P. Zhou, Q. Cao, S. Yan, Z. Liu, X. Peng, H. Zhou, Y. Xia, W. Zhang, H. Liu and K. Liu, Adv. Mater., 2025, 37, 2500478.
42. H. Ren, X. Wang, W. Ding, C. Xu, W. Zhao, H. Ji, H. Yi, Z. Zhan, Y. Song, L. Zhou, Q. Zhao and F. Pan, Adv. Funct. Mater., 2025, 35, 2504165.
43. S. Lv, X. He, Z. Ji, S. Yang, L. Feng, X. Fu, W. Yang and Y. Wang, Adv. Energy Mater., 2023, 13, 2302711.
44. J. Li, J. Chen, X. Xu, J. Shen, Z. Wang, Z. Guo, P. Lin, J. Sun, B. Huang and T. Zhao, Adv. Mater., 2025, 37, 2501006.
45. C. Monroe and J. Newman, J. Electrochem. Soc., 2005, 152, A396–A404.
46. J. Pokharel, A. Cresce, B. Pant, M. Y. Yang, A. Gurung, W. He, A. Baniya, B. S. Lamsal, Z. Yang, S. Gent, X. Xian, Y. Cao, W. A. Goddard, 3rd, K. Xu and Y. Zhou, Nat. Commun., 2024, 15, 3085.
47. J. Li, J. Chen, X. Xu, J. Sun, B. Huang and T. Zhao, Energy Environ. Sci., 2024, 17, 5521–5531.
48. Z. Sun, F. Li, J. Ding, Z. Lin, M. Xu, M. Zhu and J. Liu, ACS Energy Lett., 2023, 8, 2478–2487.
49. Y. Mu, Z. Liao, Y. Chu, Q. Zhang, L. Zou, L. Yang, Y. Feng, H. Ren, M. Han and L. Zeng, Nano-Micro Lett., 2025, 17, 163.
50. X. Xu, F. Xu, C. Qu, G. Jiang, H. Yu, H. Repich, H. Han, F. Cao, L. Li and H. Wang, Adv. Funct. Mater., 2021, 31, 2101059.
51. X. Xu, Y. Xu, F. Xu, G. Jiang, J. Jian, H. Yu, E. Zhang, D. Shchukin, S. Kaskel and H. Wang, J. Mater. Chem. A, 2020, 8, 1636–1645.
52. Y. Qiu, Y. Su, X. Jing, H. Xiong, D. Weng, J. G. Wang, F. Xu and H. Wang, Adv. Funct. Mater., 2025, 35, 2423559.
53. J. Zhan, L. Deng, Y. Liu, M. Hao, Z. Wang, L. T. Dong, Y. Yang, K. Song, D. Qi, J. Wang, S. Wang, H. Liu, W. Zhou and H. Chen, Adv. Mater., 2025, 37, 2413420.
54. O. Kong, G. Jiang, K. Wang, A. Yang, J. Zheng, S. Wu, W. Han, J. Li, L. Li, X. An, J. Wang, C. Sun, L. Li, F. Xu and B. An, Adv. Energy Mater., 2025, 15, 2405593.
55. L. Li, G. Jiang, K. Wang, A. Yang, T. Cao, O. Kong, J. Li, J. Zhu, J. Zheng, C. Sun, L. Li, F. Xu and B. An, Adv. Funct. Mater., 2025, 35, 2505700.
56. M. Li, H. An, Y. Song, Q. Liu, J. Wang, H. Huo, S. Lou and J. Wang, J. Am. Chem. Soc., 2023, 145, 25632–25642.
57. P. Xu, Y. C. Gao, Y. X. Huang, Z. Y. Shuang, W. J. Kong, X. Y. Huang, W. Z. Huang, N. Yao, X. Chen, H. Yuan, C. Z. Zhao, J. Q. Huang and Q. Zhang, Adv. Mater., 2024, 36, 2409489.
58. J. Yang, X. Liang, H.-H. Ryu, C. S. Yoon and Y.-K. Sun, Energy Storage Mater., 2023, 63, 102969.
59. Z. Wang, C. Zhu, J. Liu, X. Hu, Y. Yang, S. Qi, H. Wang, D. Wu, J. Huang, P. He and J. Ma, Adv. Funct. Mater., 2023, 33, 2212150.
60. K. Ma, Y. Cao, S. Zhang, Y. Zhang, S. Fang, X. Han, F. Jin and J. Sun, Nano Lett., 2024, 24, 8826–8833.

---