Highly stable quasi-solid-state initially anode-free lithium metal batteries enabled by dynamic integrated interface engineering Authors: Yunyi Chen, Xitang Qian, Yuxiang Lyu, Yican Qiu, Siyu Zhou, Xinyi Lan, Siqi Lu, Minhua Shao Published: 2026 ABSTRACT AbstractInitial-anode-free lithium metal batteries offer the potential for high energy density and simplified manufacturing, positioning as a transformative platform for sustainable energy storage. However, their implementation is hindered by the low reversibility of lithium plating/stripping. Herein, we propose a dynamic integrated interface engineering strategy that incorporates an Ag-based multifunctional initiator into the electrolyte. This additive consists of Ag+ paired with anions capable of generating Lewis acids, enabling electrode-electrolyte interfacial coupling through in situ polymerization, while simultaneously enhancing the lithiophilicity of current collectors via Ag+-driven in situ deposition. The controlled release of the additive regulates the solvation structure, leading to the solid electrolyte interphase with an inverted organic-rich inner/inorganic-rich outer architecture, which represents a deliberate departure from conventional models. The anion anchoring effect reinforces the self-healing capability of the interphase, ensuring the construction of fast Li-ion transport pathways and a durable protective barrier, realizing highly stable operation of initially anode-free quasi-solid-state batteries. Specifically, the Cu | |LiFePO4 cell exhibits 90% capacity retention after 200 cycles at 0.1 C, and a 2 Ah-level pouch cell delivers a specific energy of 467 Wh kg−1 based on total cell weight. This work provides an effective unified approach to address multiple interfacial challenges in initially anode-free lithium batteries.Subject terms: Batteries, Energy, Batteries Introduction Lithium metal has long been regarded as a highly promising negative electrode material owing to its high theoretical specific capacity (3860 mAh g−1) and low reduction potential (−3.04 V vs. standard hydrogen electrode, SHE)1–3. Utilizing a fully lithiated positive electrode and a bare copper current collector to construct an initially anode-free full cell can maximize energy density and reduce manufacturing cost. However, persistent issues such as lithium dendrite growth, active material pulverization, and dead lithium accumulation collectively exacerbate interfacial resistance, compromise the integrity of electrolyte-electrode contact, and raise safety concerns due to the risk of internal short circuits4–6. These detrimental phenomena are particularly pronounced in initially anode-free configurations, posing formidable technical barriers to the development of practical initially anode-free lithium metal batteries (AFLMBs)7–11. It has been established that the lithiophilicity of the current collector, the stability of the solid electrolyte interphase (SEI) layer, and the compatibility of the electrolyte-electrode interface are critical factors for initially anode-free solid-state lithium metal batteries12–14. Conventional Cu current collectors exhibit inherent lithiophobicity, while their non-uniform nucleation sites readily induce lithium dendrite formation15. Additionally, naturally formed SEI layers demonstrate chemical heterogeneity and structural instability, leading to spatially selective lithium-ion diffusion that exacerbates irregular lithium deposition/stripping16. Meanwhile, within solid-state lithium batteries, the physical voids, poor contacts, or micro-scale pores that inherently exist at electrolyte-electrode interface frequently evolve into ion-blocking interphases that severely impede lithium transport17,18. To address these challenges, various strategies have been developed, including novel electrolyte formulations, artificial SEI layers, and modified current collector substrates. For instance, lithiophilic metal coatings can be applied onto the negative electrode surface via magnetron sputtering or roll-pressing techniques to optimize lithium deposition behavior19–21. Among these metals, silver has attracted considerable interest because of its lithiophilicity22–24. Lee et al. achieved a high cycling coulombic efficiency (CE) of 99.8% by coating the copper with an Ag-C layer. However, this approach requires additional pressure to suppress lithium dendrite growth, and the extra coating step complicates battery manufacturing23. Similarly, magnetron sputtering of an Ag layer onto the current collector to form a Li-Ag alloy enhances interfacial lithiophilicity, but the substantial silver usage increases production costs25. Therefore, achieving convenient modification of current collectors using silver particles remains a challenging task. Alternatively, engineered interphases can be constructed on electrodes through artificial SEI strategies to suppress dendrite growth by controlling stress distribution26–29. Optimizing the electrolyte formulation can modulate the interfacial environment to extend the effective cycle life of AFLMBs30. Nevertheless, these individual approaches often fail to comprehensively cover all the critical aspects of lithium deposition/stripping process. And they also fall short of resolving the persistent depletion of the lithium inventory during long-term cycling. Hence, there is an urgent need to develop a simple and effective integrated method that can accomplish the Cu foil modification, regulate electrolyte performance, and construct an SEI structure in one step, while remaining stable during long-term operation. In this work, we propose a dynamic integrated interface engineering strategy involving the incorporation of a minimal Ag-based multifunctional initiator (AMI) into the electrolyte, and leveraging its controlled-release function to simultaneously achieve Cu current collector modification and in situ electrolyte polymerization after battery assembly. More importantly, it can simultaneously achieve SEI reconstruction, thereby realizing high efficiency and long cycle life of AFLMBs in a unified process (Fig. 1). AMI is composed of silver cations with anions capable of generating Lewis acids (hexafluorophosphate or difluoro(oxalato)borate ions, Supplementary Fig. 1). Firstly, using silver hexafluorophosphate as the research object, by adjusting the plating potential, the Ag+ provided by AMI can be reduced to silver particles and achieve morphology control. This enhanced the lithium affinity of the Cu foil. Additionally, the in situ polymerization induced by anions of AMI facilitates electrolyte-electrode connections, ensuring a robust charge transfer for Li+. More importantly, through the synergistic interactions of interfacial metal modification and AMI anions-induced ring-opening polymerization of solvents (1,3-Dioxolane, DOL), the adsorption capacity of the current collector for anions and the solvation sheath in electrolytes can be both altered, enabling in situ formation of a bilayer-inverted SEI architecture with an organic-rich inner layer and inorganic-rich outer layer (OI-IO) structure on the electrode surface. Unlike conventional SEIs with mosaic organic-rich architectures, the distinctive feature of the OI-IO SEI can integrate fast Li-ion transport with strong mechanical properties. Critically, the Ag substrate anchors anion-derived groups, which maintains a high inorganic content in the outer layer during cycling and thereby constructs a durable protective barrier. Furthermore, the aforementioned integration process is confined to the initial formation cycle. These processes do not recur during subsequent continuous plating/stripping cycling, and thus, they do not introduce ongoing inefficiencies that would degrade cycling performance. Consequently, the initially anode-free Cu | |LiFePO4 (LFP) full cell demonstrates 90% capacity retention over 200 cycles at 0.1 C, while high-loading positive electrode (>12 mg cm−2) under anode-less conditions (N/P = 0.8) maintains stable operation for 300 cycles. The assembled Cu | |LiNi0.8Co0.1Mn0.1O2 (NCM811) pouch cell achieves a capacity of 2 Ah-level and a specific energy of 467 Wh kg−1 based on total cell weight. Multifunctional controlled-release Ag-based multifunctional initiator for regulating quasi-solid-state initially anode-free batteries. A schematic diagram of the triple-synergy sustained-release Ag-based initiators to in situ modify current collector, optimize SEI, and in situ polymerization to enhance interfacial adhesion to achieve high-performance quasi-solid-state AFLMBs. OI-IO SEI: Organic-rich inner and inorganic-rich outer solid electrolyte interphase layer; OIP: ordinary in situ polymerization; LMBs lithium metal batteries. (H: white, Li: green, C: brown/gray, N: pink, O: red, F: cyan, S: yellow). Results Electrochemical deposition induced by AMI In this study, we achieved real-time control over the morphology of silver particles on Cu foil after cell assembly by adjusting the concentration of AMI and the deposition potential (Fig. 2a and Supplementary Fig. 2). Potentiostatic mode was employed for the electrochemical reduction of silver ions. Given the standard reduction potential of silver at +0.80 V vs. SHE, the deposition rate of silver could be modulated by applying potentials lower than this value to the Cu foil. Based on linear sweep voltammetry (LSV) profiles (Supplementary Fig. 3), specific potentials (–0.5, –1.0, –2.0, and –3.0 V vs. SHE) were selected within the range between the reduction potentials of copper and lithium. Deposition rates accelerate progressively with increasingly negative potentials. At relatively high AMI concentrations (2 wt.% and 1 wt.% in electrolyte) and the electrochemical potential is low (–0.5 V, vs. SHE), the silver deposition process is governed by interfacial reactions, leading to the formation of a bulk structure (BS) through particle growth (the type I). When the deposition potential is set to –1.0 or –2.0 V with an AMI concentration around 0.5 wt.%, the morphology of the silver particles is influenced by the competition between interfacial reactions and diffusion processes, resulting in anisotropic structures (AS) such as dendrites and flower-like formations (the type II). In contrast, at high negative potential with a low interfacial AMI concentration (0.5 wt.%, –3V; 0.1 wt.%, –2 V; 0.1 wt.%, –3 V), the deposited silver on the collector surface primarily consists of small nanoparticles (SN) (the type III). Under conditions of limited diffusion (0.1 wt.%, –3 V), uniform silver nanoparticles are formed (Supplementary Fig. 4). The successful electrochemical preparation of Ag particles on the Cu foil was confirmed by Energy-dispersive spectroscopy (EDS) mapping and X-ray diffraction (XRD) characterization (Supplementary Figs. 5 and 6). X-ray photoelectron spectroscopy (XPS) depth profiling also reveals that the Ag signal is detectable at different depths on the current collector surface (Supplementary Fig. 7). In situ electrochemical regulation of silver particles and optimization of lithium deposition morphologies. a The SEM images of silver particles with different morphologies on the current collector surface as a function of the potential and AMI concentration (30 °C). Lithium symmetrical cells performance as induced by (b) bulk silver particles, c anisotropic silver particles, and d small nano silver particles at 1 mA cm–2 and 1 mAh cm–2. SEM and FIB-SEM images of lithium deposition morphologies after 30 cycles and deposited to 0.3 mAh cm–2 at 1 mA cm–2 induced by (e and f) AMI and (g and h) OIP electrolytes. i Calculated adsorption energies of a Li atom with Ag and Cu substrates. j Morphological evolution of electrodeposited Li characterized by real-time optical microscopy on Li surface, and silver-modified Li surface at 2 mA cm–2 and 2 mAh cm–2 (30 °C). k Morphological evolution of electrodeposited Li characterized by real-time optical microscopy on Cu collector surface, and silver-modified current collector surface at 5 mA cm–2 and 10 mAh cm–2 under 30 °C (The red dotted line represents the thickness of lithium growth). Phase field simulation results of lithium deposition induced by (l) uniform nucleation with AMI, and (m) heterogeneous nucleation with OIP. The effects of different silver particle morphologies on lithium deposition behavior and electrochemical performance were further investigated. Following the initial lithium deposition and repeated charge-discharge cycling, SN silver particles can induce more uniform lithium plating, transforming the lithium deposition morphology from loose structures like dendrites to dense bulk forms (Supplementary Fig. 8 and Supplementary Note 1). This improvement is primarily attributed to the ability of SN silver particles to provide more lithium nucleation sites, allowing lithium ions to be reduced more uniformly on the current collector surface. Energy dispersive spectroscopy (EDS) results in Supplementary Fig. 9 show that SN can induce a uniform and dense lithium deposition structure from the initial stage of lithium plating (SOC > 20%). In addition, by assembling Cu@Li | |Li@Cu cells, it is found that within 300 h, the overpotential of cells with the BS and AS silver particles exceeds 0.1 V and 0.2 V, respectively (Fig. 2b, c, Supplementary Fig. 10). In contrast, the SN silver particles enable longer stable plating/stripping, maintaining the overpotential within 25 mV (Fig. 2d). Furthermore, the Li | |Cu cells prepared with SN particles also achieve a higher average CE of 99.9% at a current density of 1 mA cm–2 for lithium plating/stripping, outperforming the 98.7% and 98.4% obtained with BS and AS particles, respectively (Supplementary Fig. 11). After establishing the optimal in situ electroplating conditions for AMI, scanning electron microscopy (SEM) and focused ion beam scanning electron microscopy (FIB-SEM) were employed to conduct a comparative analysis of lithium deposition morphologies mediated by AMI in situ polymerization electrolyte (1.2 M LiTFSI-pDOL/5 wt.%FEC-0.1 wt.% AMI) versus ordinary in situ polymerization (OIP, Using lithium hexafluorophosphate initiators, LiPF6) electrolyte (1.2 M LiTFSI-pDOL/5 wt.%FEC-0.1 wt.% LiPF6) and baseline electrolyte (1.2 M LiTFSI-DOL/5 wt.%FEC). After the initial lithium electroplating at 0.5 mA cm–2 to a capacity of 5 mAh cm–2, the lithium deposition morphology with AMI appears dense with few visible voids, whereas the OIP electrolyte exhibited extensive dendritic growth (Supplementary Fig. 12). Further, after the charge/discharge cycling tests in Li | |Cu cell, the lithium electroplating layer (deposited to 0.3 mAh cm–2) with AMI mainly consists of a large sheet structure with minimal voids (Fig. 2e, f). FIB cross-section images reveal an average thickness of approximately 1.7 μm for the deposited lithium layer, corresponding to a volume expansion rate of only 16.8%. In contrast, numerous dendrite structures are displayed in the OIP electrolyte, with an average thickness of about 5.5 μm for the deposited lithium layer (Fig. 2g, h). Such rod-like lithium is prone to becoming dead lithium during cycling, leading to irreversible active lithium loss and posing a major challenge to the long-term operation of AFLMBs31,32. Through statistics of the voids area, it is found that the porosity of the lithium deposition layer can be reduced from the original 42.1–1.6% through the modification of the electrode surface by the AMI initiator (Supplementary Fig. 13). Furthermore, SEM and FIB-SEM images demonstrate that many holes and dendrites also appear in the deposited layer through the baseline electrolyte due to the lack of silver particles (Supplementary Fig. 14). Importantly, the lithium morphology remains loose and porous even in systems using ex situ Ag plating combined with OIP electrolyte (EAO, Supplementary Fig. 15a). Similarly, a dendritic structure still forms if SEI modulation is attempted only after the electrolyte has fully polymerized, thereby bypassing the controlled-release function of AMI (Supplementary Fig. 15b). These results collectively indicate that a dense and highly reversible lithium deposition process is achievable only through the integrated strategy enabled by AMI. To assess the completeness of Li stripping, we performed Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) analysis on the Cu surfaces of fully discharged cells (SOC = 0%). The Li signal on the AMI system surface is weak, whereas the OIP system surface exhibits an intense Li signal, approximately eight times higher than that of the AMI system (Supplementary Fig. 16). This indicates that Li stripping is highly efficient in the AMI, with minimal residual Li. In contrast, the accumulation of dead Li on the current collector surface in the OIP system creates a physical barrier that impeded subsequent Li plating/stripping. This strategy concurrently modifies the current collector, regulates the SEI, and facilitates in situ electrolyte polymerization. Density functional theory (DFT) calculations were employed to determine the adsorption energies between a lithium atom and Ag or Cu slab (Fig. 2i and Supplementary Table 1)22,33. The Ag (111) surface in AMI system demonstrates a larger adsorption energy of –3.09 eV compared to the pristine copper collector (111) at –2.30 eV, revealing that the electroplated Ag layer is much more lithiophilic than the Cu substrate. This lithiophilic Ag coating can reduce electroplating overpotentials, facilitating smooth lithium deposition. In situ optical microscopy further illustrates the distinct lithium deposition behaviors in AMI and OIP electrolytes. A large number of loose structures are observed on the electrode after electroplating at 2 mA cm–2 with a capacity of 2 mAh cm–2, while no obvious accumulation layer is found on the silver-modified electrode surface (Fig. 2j). The benefits of AMI are even more vividly demonstrated during the plating process on the Cu surface at 5 mA cm–2 with a capacity of 10 mAh cm–2 (Fig. 2k). The deposited layer thickness in OIP system exceeded that of the AMI by more than threefold. In contrast to the extensive development of dendritic flower-like morphologies on pristine copper substrates, the modified current collector demonstrates interfacial stability during deposition processes. The lithium deposition process was also modeled using COMSOL Multiphysics to simulate the phase field, concentration field, and potential field under uniform nucleation conditions (Fig. 2l, m, Supplementary Figs. 17–19). The results corresponding to the multi-site uniform growth process of the AMI system show that the lithium deposition morphology is relatively compact, with only a few small branches. The lithium ions concentration and potential distribution are also relatively stable. In contrast, the OIP electrolyte showcases a random few-point growth process, with lithium primarily forming in rod-like structures and showing chaotic lithium ions concentration and potential distributions. All of the above results indicate that the in situ electrochemical regulation process based on AMI is helpful to achieve reversible lithium plating/stripping behavior. Optimization of SEI layer composition and structure XPS depth-profiling confirms this compositional distribution within the SEI. The Li 1S spectra reveal a stable peak near 56 eV in the outer layer of the SEI, indicating the presence of more LiF along with a small number of inorganic components such as Li2O, Li3N, and Li2CO3. However, this peak diminishes abruptly as sputtering proceeds, marking a boundary between the outer and inner SEI regions (Fig. 3b and Supplementary Fig. 21). Commensurately, the F 1 s spectra (Fig. 3c) show that the LiF peak appears and diminishes with sputtering time, being replaced by the C-F bond. Notably, LiF exhibits negligible presence within the OIP-derived SEI architecture (Supplementary Fig. 22). Meanwhile, in the O 1 s, N 1 s and S 2p spectra, the AMI-SEI exhibits a higher concentration of inorganic components compared to the control case (OIP-SEI), with these components predominantly located in the outer layer of the SEI (Fig. 3d–f and Supplementary Fig. 23). Statistical analysis (Supplementary Fig. 24) indicates that inorganic composition on the surface of the AMI-SEI outer layer accounts for as much as ~99%, with LiF constituting ~64%. In the main body of the AMI-SEI outer layer, LiF even reaches ~95.2%. Conversely, the inner layer of the AMI-SEI shows a greater presence of organic components. This inverted structure deviates from the traditional mosaic-like organic-rich SEI structure recorded in previous studies. Moreover, the AMI-SEI demonstrates structural stability, retaining substantial inorganic content (LiF, Li2O, and Li3N) in its outer layer even after 30 palting/stripping cycle tests (Supplementary Fig. 25). In stark contrast, the SEI in the baseline electrolyte is predominantly composed of organic components due to the dominant role of DOL in solvation sheaths (Supplementary Fig. 26). Unlike the dynamic regulation of AMI, XPS analysis reveals that the ex situ silver-modified Cu foil could not replicate this inverted structure during in situ polymerization. The EAO preferentially adsorbed electrolyte anions (TFSI-) from the beginning of SEI formation, forming an inorganic-rich mosaic interfacial phase (Supplementary Fig. 27). Furthermore, even if AMI is added to the electrolyte, if the electrolyte-electrode interface reaction process is not controlled during the in situ polymerization process, the SEI formed after complete polymerization will still consist of inorganic-organic composite components (Supplementary Fig. 28). These comparative observations substantiate the distinctive methodology of leveraging AMI sustained-release kinetics to construct OI-IO SEI. Composition distribution of OI-IO SEI layer. a Schematic of the interface structure regulated by AMI after the formation process and first palting/stripping cycles at 0.1 C and 30 °C (The frame lines represent local magnifications). b–f XPS spectra depth-profiles of the formed SEI of (b) Li 1 s, and (c) F 1 s with AMI (Sputtering time: 10 min). XPS spectra of the formed SEI of (d) O 1 s, (e) N 1 s, and (f) S 2p in AMI and OIP electrolytes. TOF-SIMS 3D distribution of SEIs formed in (g) AMI and (h) OIP electrolytes. Single-color gradient maps of LiF2-, LiO-, and LiS- in (i) AMI and (j) OIP electrolytes. The three-dimensional (3D) architecture of SEI and the spatial distribution of each component were further scrutinized using ToF-SIMS. The outer layer of the AMI-SEI shows a higher concentration of LiF and other inorganic species than that of the OIP-SEI. The substances in the OIP electrolyte are intricately intertwined, exhibiting a marked increase in organic matter (Fig. 3g, h and Supplementary Fig. 29a, b). This phenomenon is particularly pronounced in the superimposed 3D distribution diagrams (Supplementary Fig. 30) and depth profiles (Supplementary Fig. 31). Single-color gradient maps of LiF2-, LiO-, LiS-, and LiCO3- reveal that the distribution of inorganic components on the surface of the AMI-SEI is more homogeneous, which is likely to promote the formation of a more stable and isotropic lithium ions diffusion channel, thereby promoting uniform lithium deposition behavior (Fig. 3i, j and Supplementary Fig. 29c, d). These findings confirm that the incorporation of the AMI facilitates the development of a stable OI-IO LiF-rich SEI on the current collector. To further investigate the regulatory role of AMI on the SEI structure and component distribution, cryo-transmission electron microscopy (Cryo-TEM) and annular dark field scanning transmission electron microscopy (ADF-STEM) were employed to confirm the structural integrity of the OI-IO SEI structure. The SEI formed in the AMI electrolyte exhibits a uniform thickness, forming a stable passivation layer of about 20 nm (Fig. 4a). In contrast, the SEI randomly formed in the control case displays a wavy morphology with irregular thickness variations (~35-80 nm) (Fig. 4b), which severely compromises interfacial lithium-ion diffusion homogeneity. High-resolution Cryo-TEM imaging further reveals distinct contrast differentiation between the inorganic-rich outer layer (darker substrate) and the amorphous organic inner layer of the AMI-SEI, demarcated by a sharp interfacial boundary (Fig. 4c). Further magnification reveals distinct lattice fringes indexed to LiF (111), Li2S (200), and Li2O (111) crystallographic planes (Supplementary Fig. 32a–c). However, the SEI formed in OIP electrolytes predominantly exhibits amorphous characteristics corresponding to organic constituents (Supplementary Fig. 32d, e). Selected area electron diffraction (SAED) patterns collected from various regions of the AMI‑SEI outer layer show concentric rings indexed to LiF, Li2O, Li3N, and other inorganic phases, confirming the inorganic-dominated composition of the AMI-SEI (Fig. 4d, e). The ADF-STEM provides a more detailed view of the AMI-SEI outer layer structure, confirming the presence of LiF-rich regions (Supplementary Fig. 33). Characterization of SEI structure and properties. a–c Cryo-TEM images of Li morphology after the formation process at 0.1 C and 30 °C (1 C = 150 mAh g–1). in (a and c) AMI electrolyte (Scale bar of insert: 10 nm), b OIP. d, e SAED pattern of the different regions at the surface of growth lithium shown in Fig. 4c (Scale bar: 5 1/nm, the dotted lines represent the diffraction rings of each substance.). f, g Cryo-TEM images of Li surface morphology with (f) OIP and (g) AMI after 30 plating/stripping cycles at 0.1 C and 30 °C. h Inverse FFT images of AMI-SEI after 30 plating/stripping cycles at 0.1 C and 30 °C (Scale bar: 2 nm). i The activation energy for ion diffusion in the SEI layer. j Tafel plots and corresponding exchange current of AMI and OIP. k Stress distribution and corresponding average DMT modulus of AMI-SEI and OIP-SEI. To assess the structural stability of the SEI layer, the lithium morphology after cycles in different electrolytes was characterized in detail. It is observed that in the OIP electrolyte, the structure of the interfacial passivation layer is difficult to discern (Fig. 4f and Supplementary Fig. 32g). This is most likely due to the non-uniform lithium deposition/stripping behavior, which severely compromises the fragile OIP-SEI layer in the control case. Conversely, the OI-IO AMI-SEI, with its dense outer layer structure, retains its original morphology even after multiple lithium plating-stripping cycles (Fig. 4g). By calibrating the lattice near the surface from the Cryo-TEM image (Supplementary Fig. 34) and measuring the lattice spacing in the reverse fast Fourier transform (FFT) images (Fig. 4h), we identify a significant presence of inorganic components such as LiF (111), Li2S (200), Li2O, and Li3N (110) on the electrode surface of the AMI-SEI after charge-discharge cycling suggesting that the inorganic multiphase outer layer is beneficial for improving the interfacial resistance to strain and suppressing side reactions. Furthermore, comprehensive electrochemical characterization was performed to evaluate the advantages of the OI-IO SEI in terms of Li+ diffusion, interfacial reaction kinetics, and mechanical stability. Temperature-dependent electrochemical impedance spectroscopy (EIS) analyses quantified the activation energy (Ea) for Li+ migration across SEI layers. The EIS curves at varying temperatures (ranging from 30 °C to 60 °C) are presented in Supplementary Fig. 35. The Li+ diffusion and charge transfer impedance (Rin) of OI-IO SEI is consistently lower than that of the OIP-SEI across all temperatures. Notably, the AMI demonstrates lower interfacial impedance (51.3 Ω) than baseline electrolytes (63.9 Ω), most likely attributable to its inverted bilayer configuration that effectively alleviates parasitic interfacial reactions. In the baseline electrolyte, the SEI layer is predominantly amorphous, thicker, and exhibits higher interfacial impedance (Supplementary Fig. 36a, b). Arrhenius analysis of temperature-dependent Ea values confirms the SEI’s lower diffusion barrier (14.7 kJ mol–1/0.15 eV vs. 18.6 kJ mol–1 /0.19 eV for OIP-SEI), mechanistically explaining its enhanced Li+ transport efficiency (Fig. 4i). To further resolve the kinetic contributions, distribution of relaxation times (DRT) analysis on the EIS data was performed (Supplementary Fig. 37 and Supplementary Table 2). the time constant τ1 ( ~ 10–5 s) is attributed to contact resistance associated with the electrolyte resistance. The peaks at τ2 (10–4–10–3 s) and τ3 (10–3–10–2 s) correspond to Li+ transport through the SEI layer. Two peaks between 10–2 and 10–1 s (τ4, τ5) are assigned to charge transfer resistance, while the peak at τ6 (10–1–10 s) reflects mass transport processes within the solid electrode. Notably, the peak areas of τ2, τ3, τ4, and τ5 are all smaller in the AMI compared to the OIP and liquid electrolyte cases, indicating faster Li+ transport through the SEI and more rapid charge exchange at the current collector surface. DRT analysis of EIS data acquired at different temperatures consistently show smaller τ2, τ3, τ4, and τ5 values for the AMI. Additionally, the interfacial charge‑transfer kinetics were also examined via Tafel analysis derived from cyclic voltammetry (CV), with exchange current densities (j0) calculated from polarization curves (Fig. 4j and Supplementary Fig. 36c, d). The exchange current densities for the liquid-SEI, OIP-SEI, and AMI-SEI are 0.60, 3.14, and 4.91 mA cm–2, respectively, indicating the OI-IO architecture’s fast charge-transfer capability, and ensuring sufficient lithium flux for homogeneous deposition. Notably, the batteries employing the ex situ Ag-plating+OIP electrolyte (EAO) and those subjected to SEI modulation after full polymerization exhibited j0 values of 3.89 and 4.09 mA cm–2, respectively (Supplementary Figs. 38, 39). Both values are lower than that of the AMI cell, underscoring the advantage in ion diffusion kinetics provided by the OI-IO SEI constructed via the controlled-release function of the AMI. Another enhancement provided by the AMI initiator to the SEI is the improved mechanical properties. Atomic force microscopy (AFM) cross-sectional height profile images confirmed the presence of protrusions on the surface of the AMI current collector compared to the OIP system, indicating an effective distribution of Ag particles on the copper foil surface (Supplementary Fig. 40). This regularity may facilitate the equable nucleation and growth of lithium on the surface. The Derjaguin-Muller-Toporov (DMT) modulus obtained from AFM mechanical analysis quantifies the mechanical property of OI-IO SEI (Fig. 4k and Supplementary Fig. 36e). The average DMT modulus of AMI-SEI layer is 5.4 GPa, higher than that in the OIP-SEI (2.3 GPa) and the base-SEI (1.1 GPa). This mechanical reinforcement originates from the inorganic-dominated outer SEI layer functioning as a strain-resistant shield during repeated lithium deposition/stripping cycles. To further verify the advantages of this mechanical property, additional nanoindentation measurements were performed. The results further show that the Young’s modulus of OI-IO SEI is much higher than that of OIP-SEI, and the overall stress uniformity is stronger than that of the control case. The values calculated from the stress curves are E(OI-IO) = 5.43 GPa and E(OIP-SEI) = 2.27 GPa (Supplementary Fig. 41). The OI-IO SEI exhibits competitive performance compared to other engineered SEI structures reported in the literature (Supplementary Table 3). Mechanism of generating OI-IO SEI layer To further elucidate the mechanism underlying the formation of the OI-IO SEI structure, density functional theory (DFT) calculations, in situ Raman characterization and molecular dynamics (MD) simulation were employed to probe the evolving electrode-electrolyte interfacial microenvironment. The DFT calculation results indicate that the adsorption energy between AMI-modified substrate and TFSI- is –3.19 eV (Fig. 5a and Supplementary Table 1), substantially exceeding the copper substrate’s anion adsorption capacity (–2.27 eV), suggesting that AMI induces the TFSI- anchoring adsorption effect. To elucidate the reaction kinetics of AMI additive-promoted OI-IO SEI formation, we systematically investigated two representative LiTFSI decomposition pathways on distinct substrates (Fig. 5b and Supplementary Figs. 42–46). Based on the XRD results (Supplementary Fig. 6), Cu(111) and Ag(111) were selected as the exposed crystal planes of the research substrate. The Gibbs free energy profiles for all reaction steps were calculated, tracking the evolution from TFSI- to Li3N and LiF. The calculation process is based on the conventional reactive decomposition pathway of LiTFSI to generate various inorganic species, and ensures the conservation of the number of atoms throughout the process34,35 (Supplementary Figs. 45, 46). Notably, the first decomposition pathway on the Cu (111) facet exhibits kinetically hindered endothermic reduction steps. Specifically, the LiNSO→NS transition (Step 5 → 6) requires a highly positive energy barrier of 2.91 eV (Fig. 5b), which impedes LiTFSI decomposition and consequently delays OI-IO SEI formation. In stark contrast, all reaction steps on the Ag substrate proceed via spontaneous exothermic processes with negative Gibbs free energy values. Critically, the same LiNSO→NS step (Step 5 → 6) releases energy (-2.84 eV), eliminating the rate-limiting endothermic barrier observed on Cu collector. This difference highlights the catalytic activity of the AMI-modified substrate in facilitating LiTFSI decomposition and OI-IO SEI generation. Further analysis of the LiF formation pathway reveals enhanced decomposition driving forces on Ag substrate. The LiF‑formation steps (LiSO2CF3 → LiCF3, Step 2 → 3, –5.75 eV; LiCF3 → LiCF2, Step 3 → 4, –5.41 eV; LiCF2 → LiCF, Step 4 → 5, –4.42 eV) exhibit more negative Gibbs free energies on Ag than on Cu, further confirming its role in accelerating the formation of an LiF‑rich inorganic outer layer (Supplementary Fig. 42). These results collectively demonstrate that interfacial modification optimizes reaction pathways of lithium salts and enables decomposition kinetics, thereby revealing the function of AMI in SEI engineering at the atomic-level. Consequently, with the onset of in situ polymerization, an increasing number of anions gradually infiltrate the electrode-electrolyte interface, leading to the formation of an inorganic outer layer dominated by LiF, Li2O, and Li3N, which is built upon the organic inner layer generated by the DOL reaction. The disparities in the structure and composition of the SEI induced by AMI, OIP initiator, and EAO are noteworthy (Supplementary Fig. 44). Mechanism of generating OI-IO SEI layer. a Calculated adsorption energies of TFSI- with Ag and Cu substrates. b The Gibbs free energy diagrams of one of the decomposition reaction pathways (from LiTFSI to -CF3, Li3N, Li2O and S) on different substrates. Inset: intermediate structure of LiTFSI decomposition process on Ag (111) facet. Raman spectra of TFSI- in the early stage of plating process (top) and in the later stage of stripping process (bottom) at 30 °C in (c) AMI and (d) OIP electrolytes. e In situ Raman curves during plating/stripping process at 30 °C in AMI electrolyte. In situ Raman spectra of TFSI- during plating/stripping process 30 °C in (f) AMI and (g) OIP electrolytes (1 C = 150 mAh g–1, and the red box lines represent the free DOL signal). h Snapshot of the simulation box from MD simulations. i, j Radial distribution functions (RDFs, denoted as g(r)) and the coordination number (CN) between Li+ and other components in AMI electrolyte (r: distance). Atoms are color-coded as follows: H (white), Li (green), C (brown), N (pink), O (red), F (cyan), S (yellow), Cu (blue), and Ag(gray). To precisely modulate interfacial microenvironments, time-resolved Raman spectroscopy was utilized to monitor in situ polymerization rate. The spectral evolution reveals that during the 1–3 h polymerization window, the electrolyte maintains a hybrid composition of polymeric species and residual DOL monomers while preserving dynamic fluidity (Supplementary Fig. 47). In situ Raman also reveals that a polymerization degree of approximately 20% represents a critical optimal window (after 1 h). At this stage, the consumption of DOL monomers allows a significant influx of TFSI- anions into the Li+ solvation sheath (Supplementary Figs. 48, 49a). Concurrently, the deposited Ag nanoparticles exhibit strong adsorption towards TFSI-. The polymerization process and the catalytic effect of silver synergistically generate an anion-rich solvation structure, thereby driving the preferential decomposition of TFSI-. During the in situ polymerization process, the concentration of TFSI- in the lithium-ion solvated structure of the AMI electrolyte gradually increases, exceeding that in the liquid, OIP and EAO systems (Supplementary Figs. 49b and 50). Subsequently, in situ Raman spectroscopy provides molecular-level resolution in characterizing lithium-ion solvation structures. In the AMI electrolyte, the participation ratios of TFSI- ions in the solvation structure reached 95.1% in the early stage of plating process and 95.4% in the later stage of stripping process, respectively, surpassing the corresponding values of 87.2% and 89.6% observed in the OIP electrolyte (Fig. 5c–g). The AMI electrolyte exhibits plating/stripping stability (Fig. 5e), with Raman analysis revealing a prominent 840 cm–1 peak corresponding to pDOL chains, while the characteristic 940 cm–1 DOL monomer signature remained absent. And the content of TFSI- in the solvated structure has always been dominant (Fig. 5f). The persistent dominance of TFSI- in the solvation structure further demonstrates that the AMI-induced anion‑anchoring effect enhances the self‑healing capability of the OI-IO SEI, thereby stabilizing the gel electrolyte over the entire operating voltage window. Conversely, OIP electrolytes display dynamic free DOL signatures (~730 cm–1) with cycling-dependent emergence-vanishing-attenuation behavior (Fig. 5g), suggesting interfacial instability arising from suboptimal OIP-SEI properties that induce non-equilibrium polymerization-depolymerization cascades, ultimately depleting both electrolyte components and lithium reservoirs. In addition, inherent lithium plating/stripping irreversibility in liquid electrolytes maintains perpetual solvation structure reorganization (Supplementary Fig. 51). This results in unpredictable composition and structure of SEI layer, which renders the charge/discharge process more irreversible, creating a vicious cycle. Moreover, in situ Raman results during plating/stripping process show that the EAO also exhibits significant instability, as indicated by dynamically fluctuating DOL monomer peaks (Supplementary Fig. 52). This suggests that the EAO still cannot maintain the stability of the electrolyte and the SEI. The tuning effect of AMI on the solvation sheath was also analyzed using force field-based MD simulations. The simulation system, containing DOL, pDOL, TFSI-, Li+, and FEC components, modeled the hybrid solid-liquid phase environment during SEI formation with 20% polymerization degree after one hour (Fig. 5h and Supplementary Fig. 53). Radial distribution function (RDF) analysis reveals significant component accumulation within 2.2 Å of Li+ ions in the AMI electrolyte, consistent with reported solvation patterns36 (Fig. 5i). Notably, the coordination number (CN) of TFSI- surrounding lithium ions is the highest at 1.94, suggesting that anions predominantly influence the solvation behavior (Fig. 5j). Additionally, the CN of DOL has decreased compared to its state prior to polymerization (Supplementary Fig. 53b, c), revealing that anions preferentially participate in the formation of the outer layer of AMI-SEI during the middle-to-late stages of in situ polymerization. Therefore, the synergistic interaction between the AMI optimization of electrolyte solvation structures and the modified substrate’s strengthened affinity for TFSI- facilitates the development of a LiF-enriched inorganic outer layer, which effectively safeguards the electrode interface. High-performance initially anode-free lithium batteries The assembled AFLMBs were aged for at least 6 h, utilizing the slow-release ability of the Ag-based initiator to promote the complete polymerization of the electrolyte (Fig. 6a and Supplementary Fig. 54a). Due to the low solubility of AMI additives in the electrolyte, they cannot directly initiate the ring-opening polymerization of DOL (Supplementary Fig 54g). However, as Ag+ ions are electrochemically reduced and deposited as Ag nanoparticles on the Cu foil, this reduction process is accompanied by the release of corresponding Lewis acidic anions into the electrolyte, driving controlled polymerization. Therefore, the initiation effect of AMI requires synergistic regulation by both anions and cations. This dependence is distinctly different from that of traditional readily soluble initiators (such as LiDFOB/LiPF6) and is a key process for regulating SEI37. This controlled activation process achieves an average lithium-ion transfer number (tLi+) of 0.71 (Supplementary Fig. 54b). The polymerization mechanisms are also confirmed by complementary Nuclear Magnetic Resonance (NMR) and Fourier Transform infrared spectroscopy (FTIR) analyses (Supplementary Fig. 54c, d), with Raman results aligning with these findings. DFT simulations reveal the AMI-modified substrate demonstrates strong interfacial adsorption for quasi-solid-state electrolyte components, which improves compatibility between the cured electrolyte and electrode. AMI can also improve the wettability of the electrolyte on the current collector surface, thereby ensuring a dense solid-solid contact after in situ polymerization. As shown in Supplementary Fig. 55, in the AMI electrolyte, the average contact angle between the electrolyte and the current collector is 34.9° after Ag deposition but before the electrolyte is fully polymerized. In contrast, the average contact angle between the OIP electrolyte and the current collector reaches 50.7°. Cross-sectional SEM and FIB-SEM characterization show the AMI system maintains intimate interfacial contact before and after lithium deposition, whereas the OIP electrolyte exhibits multiple interfacial discontinuities (Fig. 6b and Supplementary Fig. 54e, f). It is worth emphasizing that the 6-h aging process does not affect the manufacturing efficiency of battery materials. Firstly, this aging process can be seamlessly integrated into existing battery production processes. From a practical production perspective, 6 h of aging does not necessarily create a bottleneck. Storage or formation after battery assembly is a common and necessary practice in the battery industry. Furthermore, gentle heating can further shorten the process time. In situ polymerization to optimize interfacial contact for high-performance initially anode-free batteries. a Mechanism of ring-open polymerization of DOL induced by AMI. Atoms are color-coded as follows: H (white), Li (green), C (gray), O (red), F (cyan), P(blue). b Contact between electrolyte and current collector after in situ polymerization with AMI and OIP (Scale bar: 1 μm). c,d Comparison of plating/stripping performance and curves of Li | |Li symmetrical cells at 1 mA cm–2 and 1 mAh cm–2. e, f Comparison of cycle performance and average CE of Li | |Cu cells at 0.5 and 2 mA cm–2 (0.5 and 2 mAh cm–2) after formation process. g Comparison of capacity retention (CR) of Cu | |LFP full cells at 0.1 C (1 C = 150 mA g–1) after formation process. h Full coin cell performance with high active material loading (12.6 mg cm–2, 0.5 C, N/P = 0.8) and under anode-less conditions (0.3 C, N/P = 0.7). i Full cell performance of Cu | |LFP cells under metal lean conditions (0.5 C, N/P = 1). j Comparison of cycling performance of Cu | |NCM811 full coin cells at 0.5 C (1 C = 180 mA g–1) after formation process. (All displayed capacities are discharge capacities.) k Cycling performance of initially anode-free Cu | |NCM811 pouch cell at 0.1 C. The inserts include digital photo of the pouch cell. The integrated strategy implementation enables AMI-based cells to achieve substantially improved electrochemical performance across multiple configurations. Lithium symmetric cells with AMI electrolyte demonstrate lithium plating/stripping stability exceeding 2500 h at 1 mA cm–2, outperforming OIP, EAO and liquid electrolyte counterparts by over eightfold (Fig. 6c and Supplementary Figs. 56b, 57), while sustaining low voltage hysteresis between 25 and 36 mV (Fig. 6d), indicative of uniform lithium deposition/stripping enabled by the AMI. The lithium symmetric cell with AMI also exhibits good performance repeatability, and the EIS spectra after cycling shows no significant increase in Rin, indicating that OI-IO SEI maintains stability during long-term cycling (Supplementary Fig. 58a, b). At elevated current density (2 mA cm–2), AMI electrolytes maintain nearly double the operational lifespan compared to OIP electrolytes (Supplementary Fig. 56a). Lithium symmetric cells were also deposited/stripped at different temperatures to determine if a short circuit had occurred. If soft shorts (dominated by electronic conduction) were present, the overpotential would be largely insensitive to temperature change. In contrast, our results show a clear and significant decrease in voltage hysteresis with increasing temperature at a current density of 1 mA cm–2: 20 °C: ~50 mV; 30 °C: ~36 mV; 40 °C: ~27 mV (Supplementary Fig. 59). Moreover, when the test temperature was reduced from 40 to 20 °C, the voltage hysteresis of the lithium symmetric battery recovered to around 51 mV. This strong negative correlation is characteristic of an interface where charge transfer is governed by ionic transport kinetics, not electronic shunting. The thermally activated reduction in polarization strongly indicates the absence of significant soft shorts. Distribution of relaxation times (DRT) analysis of the cycled lithium symmetric cell reveals the presence of concentration polarization and electrochemical polarization, which also indicates that the cell did not have a short circuit (Supplementary Fig. 58c, d). To evaluate lithium cycling reversibility, Li | |Cu cells were assembled to assess the CE, and the AMI half-cell demonstrates the capability for reversible cycling for more than 1200 cycles at 0.5 mA cm–2, which is 4 and 20 times greater than the OIP and baseline Li | |Cu half-cells, respectively (Fig. 6e and Supplementary Fig. 60). AMI half-cell can also cycle more than 450 times at 1 mA cm–2 (Supplementary Fig. 61). The AMI cell achieves 99.8% average CE at 2 mA cm–2 (Fig. 6f), providing guarantee for the reversible cycling of the initially anode-free full battery. Compared to recent literature, the cycle number and CE of Li | |Cu cells with OI-IO SEI are competitive (Supplementary Fig. 62, and detailed data can be found in Supplementary Table 4) and outperform cells using ex situ deposited silver current collectors (Supplementary Fig. 63). This performance is also evident in the Cu | |LFP full cell. After formation process, while control cells exhibit rapid capacity fading after just 30 cycles, the AMI-based full cell retains nearly 90% of its capacity after 200 cycles (Fig. 6g, Supplementary Figs. 64, 65). Particularly concerning, liquid electrolyte-based Cu | |LFP batteries suffer severe capacity collapse (<20% retention) within merely 10 cycles (Supplementary Fig. 66). The cycle stability of the AMI full cell is also higher than that of the EAO cell, indicating that it is necessary to combine Ag-modified Cu foil and OI-IO SEI to regulate the interface in order to reduce the consumption of active lithium during the charging and discharging process (Supplementary Fig. 67). Meanwhile, the optimized AMI concentration (0.1 wt.%) achieves the highest capacity retention (Supplementary Fig. 68). Compared with recent works, the performance of AMI additives is competitive (Supplementary Fig. 69, and Supplementary Table 5). After charge/discharge cycles, the full cells were disassembled in the fully charged state (SOC = 100%), and the Cu foils were retrieved for morphological analysis (Supplementary Fig. 70). The Li deposit in the AMI case exhibits a dense and uniform morphology without visible gaps or irregular deposits. In contrast, the OIP displays abundant dendritic/rod-like structures on the current collector surface. Moreover, the AMI system demonstrates good scalability with high mass‑loading positive electrodes (12.6 mg cm–2). At a 0.8 N/P capacity ratio and 0.5 C rate, the cell delivers areal capacities exceeding 1.7 mAh cm–2 and maintains stable operation over 300 cycles (Fig. 6h, Supplementary Note 2). Under stricter conditions (0.7 N/P ratio, 0.3 C), the discharge capacity after 300 cycles exceeds 140 mAh g–1 without capacity decay (Fig. 6h). In addition, at an N/P ratio of 1 and a charge/discharge rate of 0.5 C, the capacity retention rate remains above 90% after 1100 cycles (Fig. 6i). In situ polymerization enhances the electrochemical stability window of the electrolyte. LSV reveals that the AMI electrolyte exhibits a decomposition current of only 1.39 μA cm–2 at 4.3 V (vs. Li/Li+), lower than the 13.3 μA cm–2 observed for the baseline electrolyte (Supplementary Fig. 71). Consequently, by exerting its integration effects, AMI enables stable cycling of ether-based electrolytes in high-nickel electrode systems. Gr | |NCM811 pouch cell (~1 Ah) maintains stable cycling performance, with a capacity retention > 90% and a CE > 99.9% after 1000 cycles (Supplementary Fig. 72). And the Cu | |NCM811 cells for over 150 stable cycles (Fig. 6j), in contrast, the OIP electrolyte suffers rapid failure after only 35 stable cycles due to inhomogeneous deposition/stripping on the bare Cu surface. The baseline electrolyte system, constrained by its limited electrochemical window, fails to achieve effective charge/discharge cycling. To validate the practical implementation of AMI additives, cycling performance was assessed in high-capacity Cu | |NCM811 pouch cells (2 Ah-level). The synergistic integration of surface engineering, SEI refinement, and in situ polymerization enables stable operation for over 50 times after formation process with minimal gas evolution (Supplementary Figs. 73,74), while achieving energy densities approaching 467 Wh kg–1, enabling it to small drone flight operations exceeding 30 min (Fig. 6k and Supplementary Fig. 75). Discussion In summary, we propose a straightforward dynamic integrated interface engineering strategy for the development of quasi-solid-state AFLMBs by incorporating a controlled-release AMI into the electrolyte. The additive is composed of Ag+ paired with anions capable of generating Lewis acids, and establishes effective electrode-electrolyte interfaces through anion-triggered in situ polymerization while boosting the lithiophilicity of current collector via cation-modified deposition. Most importantly, by leveraging the sustained-release function and anion anchoring effect of AMI, a SEI layer can be constructed with an organic-rich inner and inorganic-rich outer (OI-IO) inverted structure that possesses strong mechanical properties and stable transport capacity. The optimized SEI facilitates high reversibility in lithium plating and stripping. The initially anode-free Cu | |LFP full cell demonstrates 90% capacity retention over 200 cycles, while high-loading positive electrodes (>12 mg cm–2) under lithium-limited conditions (N/P = 0.8) maintain stable operation for 300 cycles. The assembled Cu | |NCM811 pouch cell achieves a capacity of up to 2 Ah-level and a specific energy of 467 Wh kg–1. This resource-efficient approach establishes a viable pathway for high-energy-density solid-state AFLMBs. Methods Materials The lithium metal was obtained from China Energy Lithium Co., Ltd. NCM811, LFP, lithium bis(trifluromethanesulfonyl)imide (LiTFSI, 99.9%), DOL (99.9%), and Fluoroethylene carbonate (FEC, 99.9%), Separators (Celgard 2400) were procured from Kelude Corporation. Ethylene glycol dibutyl ether, AMI initiator (silver hexafluorophosphate, 99.9%), and OIP initiators (lithium hexafluorophosphate, 99.9%) were used as the additives and sourced from Aladdin Co., Ltd., with all materials incorporated into the electrolytes within an argon-filled glovebox maintaining strict atmospheric control (H2O < 0.1 ppm, O2 < 0.1 ppm). An ex situ deposited silver layer was used for comparison with AMI to highlight the advantages of in situ current collector modification. Ex situ silver layers of different thicknesses were obtained by magnetron sputtering, and the results were compared with those obtained by the in situ integration strategy. Regulation of silver particles morphology and construction of OI-IO SEI By adjusting the concentration of additive and the deposition potential of electrode surface, the plating of different silver particle morphology on the surface of copper collector was realized. The Cu foil in the coin cell has a diameter of 12 mm or 16 mm. The concentration of silver initiator ranges from 0.1 wt.% to 1 wt.%, and the potentiostatic deposition potentials are −0.5, –1, –2, and −3V (vs. SHE). The SEI layer construction of OI-IO structure occurs in the process of in situ polymerization. By applying constant current, the generating potential of SEI layer is controlled between –0.1 V and 1.0 V (vs. Li/Li+), and the cycle is more than 20 times, All of the above processes were carried out at the temperature of 30 °C. Electrochemical measurement Electrochemical measurements of Li | |Cu cells Two-electrode measurements were conducted using CR2032 coin cells. The lithium metal sheets were pre-cut to a diameter of 12 mm (Electrolyte: 40 μL). The Li | |Cu cells were employed to evaluate cycling stability, conduct electrochemical analyses, and observe lithium deposition morphological changes. Both the conventional test and the Aurbach test were applied to the CE tests. For the Aurbach testing step: lithium is pre-deposited to 10 or 20 mAh cm−2, and subsequently stripped to a cutoff voltage of 0.7 V; lithium is then deposited again at a current density of 1 or 2 mA cm–2 for 10 h to establish a lithium storage layer; lithium is then deposited and stripped down with capacities of 1 or 2 mAh cm–2 at 1 or 2 mA cm–2, and cycled 30 times; finally, the remaining lithium is stripped to 0.7 V. For comprehensive electrochemical characterization (EIS and LSV), Li with a capacity of 0.5 mAh cm–2 or 1.0 mAh cm–2 was pre-deposited on Cu sheets at a current density of 0.1 mA cm–2 to form Cu@Li electrodes. The cycling test for symmetric batteries was performed at current densities of 1.0 mA cm–2 or 2.0 mA cm–2. Electrochemical measurements (EIS, CV) utilized a CHI660E/760E workstation (Shanghai Chenhua). EIS spectra were performed applying a potentiostatic signal with an amplitude of 10 mV (quiet time: 2 s), over a frequency range of 0.1–105 Hz frequencies at 30–60 °C averaging 1 cycles 12 points per decade. Before the EIS test, an open-circuit potential test was conducted to maintain the steady-state potential for 10 min. CV scans operated at 0.5 mV s–1 within a –0.2 to 0.2 V window. Full cells assembly and electrochemical measurements Cu current collectors with different pre-lithiation levels were used to assemble Cu ||LFP coin cells, and were employed to evaluate rate capability at 0.1 C, 0.3 C, and 0.5 C. For initially anode-free lithium metal cells (N/P = 0), the current collector modification, SEI optimization and in situ polymerization are completed in one step after battery assembly. For anode-less lithium metal cells, a small amount of lithium is pre-deposited before in situ polymerization (N/P < 1). For the CR2032 coin-type initially anode-free cells, the positive electrode active material coating density is about 5 mg cm–2, and the electrolyte volume used is 40 μL. All cathode materials used PVDF as a binder (6 wt.%) and Super P (6 wt. %) as a conductive agent, and were equipped with an active lithiation agent (Li2Se, 5 wt.%) in both experimental and control cases. Cu | |NCM811 Pouch cells (64 mm × 55 mm) were used to study initially anode-free full cells. NCM811 (97% active, ~28 mg cm–2) was used as the positive electrode, and bare Cu (6 µm) was used as the current collector. The pouch cells were filled with 3.0 g electrolyte before being vacuum sealed (E/C = ~ 1.3 g Ah–1). The total weight of the assembled initially anode-free pouch cell is about 18-19 g. The full cells were cycled between 3.0 and 4.3 V using a protocol of 0.1 C charge/0.2 C discharge. The total pouch cell capacity was about ~2.3 Ah under these conditions. Gr | |NCM811 pouch cells were used to explore the stability and versatility of this in situ polymerized electrolyte in batteries with different electrode configurations. NCM811 (97% active, ~25 mg cm–2) was used as the positive electrode, and graphite (95.7% active, ~4.9 mg cm–2) was used as the negative electrode. The cells were cycled between 2.75 and 4.2 V using a protocol of 0.3 C charge/0.5 C discharge and under identical cycling conditions with a constant stacking pressure of 1 MPa. All cycling experiments were performed in a temperature-controlled environmental chamber (30.0 ± 0.5 °C). Characterizations The Cryo-TEM analysis was carried out using a Thermo Scientific Glacios microscope operating at 200 kV to investigate the composition and structural characteristics of the SEI. Sample preparation was conducted inside a glovebox, followed by storage under an argon atmosphere utilizing a sealed transfer system. The samples were then rapidly transferred to the preparation and testing chamber. Under liquid nitrogen protection (~ –180 °C), the sealed package was opened, and the samples were continuously maintained under cryogenic conditions during storage, transfer, and measurement. ADF-STEM imaging and EDS elemental mapping were performed on a JEOL ARM200F STEM (JEOL, Tokyo, Japan) equipped with a thermal field emission gun and a probe Cs corrector (CEOS GmbH, Heidelberg, Germany) at an accelerating voltage of 200 kV. In addition, ToF-SIMS measurements (IonTof M6 Hybrid) were conducted to further analyze the SEI composition. Atomic force microscopy (AFM, Dimension ICON) was employed to determine the DMT modulus. In situ Raman spectroscopy was carried out using a 785 nm excitation laser (Invia system). The laser beam was focused near the current collector region to collect Raman signals corresponding to the electrolyte in the interfacial region. All Raman measurements were performed using a sealed electrochemical cell at 25 °C. The morphology of deposited lithium was characterized using a scanning electron microscope (SEM, JSM 7800 F, JEOL) operated at voltages of 10.0 kV or 15.0 kV. After extended cycling, SEM was further used to examine lithium deposition/stripping behavior as well as the structural integrity of the SEI layer. Cross-sectional morphology was analyzed using a focused ion beam–scanning electron microscope (FIB-SEM, FEI Helios G4 UX). Initial coarse milling of the deposited layer was conducted using a Ga ion beam at 30 kV and 1000 pA. For both SEM and FIB-SEM observations, samples were transferred under argon protection using a sealed transfer vessel and promptly introduced into the instruments for vacuum pumping at 25 °C. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Kratos Axis Supra+ system to determine the chemical composition of the plated electrodes. The analysis chamber operated under a vacuum of approximately 2 × 10–7 mbar. An Al Kα X-ray source (1486.6 eV) was used, with a power of 200 W for survey scans and 300 W for high-resolution spectra. Binding energies were calibrated using the C 1 s peak at 284.8 eV. All XPS samples were prepared inside a glovebox under argon atmosphere, mounted on sample holders, and transferred to the instrument through a sealed device at 25 °C. First-principles computation Spin-polarized density functional theory (DFT) calculations were conducted using the Vienna Ab initio Simulation Package (VASP) to explore the interaction between substrate and ions. The exchange-correlation interactions were described by the generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) functional. The projector-augmented wave (PAW) method was adopted to represent the ion-electron interaction, and a plane-wave kinetic energy cutoff of 500 eV was employed for the expansion of electronic wavefunctions. A vacuum layer of 20 Å was introduced to avoid artificial interactions between periodic images. Brillouin zone sampling was performed using a Γ-centered 5 × 5 × 1 Monkhorst–Pack k-point grid. Structural optimization was carried out until the total energy and atomic forces converged to 1 × 10–6 eV and 0.01 eV Å–1, respectively. Long-range van der Waals interactions were incorporated using the DFT-D correction method38–41. The Gibbs free energy change (ΔG) was evaluated based on the computational hydrogen electrode (CHE) model, as expressed by: where ΔE represents the reaction energy calculated from the total energy difference between adsorbed reactants and products on the catalyst surface, ΔZPE denotes the zero-point energy correction, and ΔS corresponds to the entropy change associated with the reaction. For adsorption energy calculations, a Cu (111) slab model with dimensions of 10.22 Å × 10.22 Å × 19.17 Å containing 48 Cu atoms was constructed. Similarly, the Ag (111) surface (11.56 Å × 11.56 Å × 19.72 Å) consisted of 48 Ag atoms. The adsorption energies (Eads) of the Li and TFSI- on the Cu (111) substrate were calculated as follows42: Where the Esur and ELi+sur are the structural total energies of studied system before and after Li surface absorption, ELi is the system energy of lithium. Molecular dynamics simulation Molecular dynamics (MD) simulations were carried out using the GROMACS 2020.6 package, and structural visualization was performed with VMD software. The initial configurations were generated by constructing cubic simulation boxes containing mixed molecules with periodic boundary conditions using the PACKMOL program. The Amber force field was employed due to its suitability for describing small organic molecules, and corresponding parameters were generated with the SOBTOP program. Before the MD simulations, energy minimization was conducted using the conjugate gradient algorithm (5000 steps with a step size of 0.01 nm) until the convergence criterion was satisfied, defined by a maximum force below 50 kJ mol–1 nm–1. Van der Waals interactions were calculated using a cut-off scheme, while electrostatic interactions were treated with the particle mesh Ewald (PME) method. Both the cut-off radius and PME real-space distance were set to 1.0 nm. Following energy minimization, the system was equilibrated at a pressure of 1.0 bar to obtain a stable density. Temperature and pressure were controlled using the V-rescale thermostat and Berendsen barostat, respectively, with a coupling time constant of 1.0 ps and a compressibility of 4.5 × 10–5 bar–1. Each system was equilibrated for 10 ns with a time step of 0.001 ps. Subsequently, a 50 ns production run was performed, during which the pressure control method was switched to the Parrinello-Rahman barostat for improved accuracy. In addition, bond constraints involving hydrogen atoms were enforced using the LINCS (Linear Constraint Solver) algorithm. COMSOL simulation A simplified two-dimensional model was employed to study lithium deposition behavior on different substrates, each characterized by a distinct density of active nucleation sites. The interfacial electrochemical reactions were described using the Butler-Volmer equation, whereas ion diffusion was modeled according to Fick’s law. Ionic migration under an electric field was governed by the Nernst-Einstein relation. The simulation domain was defined as 50 × 25 µm2 and discretized using a refined triangular mesh. Comparative simulations were conducted on substrates with varying numbers of active nucleation sites. Additional information can be found in Supplementary Note 3. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source Data Author contributions Y.C. and M.S. conceived the concept and the project. Y.C. conducted all experiments, measurements, and calculations. X.Q., Y.Q., Y.L., S.Z., X.L., and S.L. were involved in the discussions. All authors participated in the analysis of the results and reviewed the manuscript. Peer review Peer review information Nature Communications thanks Svetlana Bachbut, Yi-Hung Liu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This work is supported by National Natural Science Foundation of China (Project No. 22408064), Hong Kong Research Grant Council (JLFS/P-602/24, R6011-23F, CRS_HKUST604/22), Otto Poon Center for Climate Resilience and Sustainability, Guangzhou Science and Technology Bureau (2024A03J0609), Guangzhou Nansha Science & Technology Bureau (2024ZD009), and LKS Fellowship from Biological Cryo-EM Center, HKUST. Data availability The data generated in this study are provided in the Supplementary Information/Source Data file. Source data are provided with this paper. All data are available from the corresponding author upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76060-y. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source Data Data Availability Statement The data generated in this study are provided in the Supplementary Information/Source Data file. Source data are provided with this paper. All data are available from the corresponding author upon request. Source data are provided with this paper.