Mechanically Adaptive Polyrotaxane Interlayers for Low‐Pressure High Energy Density Sulfide‐Based All‐Solid‐State Batteries

ABSTRACT

All‐solid‐state batteries (ASSBs) employing lithium (Li) metal anodes or an anode‐less configuration, despite their superior energy density, suffer from performance degradation under low stack pressure, hindering their practical application. To address this, we design a mechanically adaptive anode interface that leverages an elastic polymer incorporating mechanically interlocked polyrotaxane (PR). This interface synergistically combines the elastic resilience—derived from the unique ring‐sliding motion of PR—with indium fluoride (InF₃), which undergoes spontaneous conversion to form a chemically stable interface. This approach enables robust cycling stability and reliable operation under commercially relevant conditions (25°C, 0.8 MPa), even in an anode‐less configuration (N/P = 0), thus demonstrating the potential of mechanically interlocked molecular architectures for maintaining void‐free interfaces in low‐pressure ASSBs with high energy densities.

A mechanically adaptive interface enhances the stability and cycling performance of lithium metal all‐solid‐state batteries, particularly under low pressure. The interface design synergistically integrates indium fluoride with polyrotaxane, where the inorganic component effectively suppresses lithium dendrite formation and the organic component mitigates interfacial void generation.

1. Introduction

The rapid commercialization of electric vehicles (EVs) has driven intensive research into next‐generation battery technologies, with all‐solid‐state batteries (ASSBs) emerging as a promising alternative to conventional lithium (Li)‐ion batteries (LIBs) [1, 2, 3, 4]. By replacing flammable liquid electrolytes with solid‐state electrolytes (SEs), ASSBs hold great promise for advancing both battery safety and performance [5, 6, 7]. Among the various SE compositions, sulfide‐based compositions—particularly argyrodite structures such as Li₆PS₅X (LPSX, where X = Cl, Br)—stand out for their high ionic conductivity and ability to form stable interfaces without cracking under pressure, thanks to their intrinsically soft and ductile nature [8, 9, 10].

Beyond safety, SEs also open the door to practical use of high‐capacity metallic anodes, such as Li metal, by serving as robust physical barriers against dendrite growth, although this has been challenged in recent years [11, 12]. As the EV industry pushes toward higher energy density, researchers are exploring advanced anode concepts [13, 14]—including silicon (Si) [15, 16, 17], Li metal [18, 19, 20], and even anode‐less configurations [21, 22, 23, 24]—that deliver large theoretical capacities, but introduce new interfacial challenges. In particular, Li metal tends to grow filament‐like dendrites [25, 26], which destabilize the interface, increase the resistance, and accelerate performance degradation [20, 27].

To mitigate these critical issues, protective layer strategies that incorporate the implementation of various coatings on the Li metal anode [28, 29] or lithiophilic seeds such as silver (Ag) and magnesium (Mg) for anode‐less configurations have been explored [30, 31, 32]. Previously, these lithiophilic interlayers have been cast on the current collector using inactive polymers such as poly(vinylidene fluoride) (PVDF) [22]. A key limitation of these systems is that they often need high‐stack pressure, elevated temperatures, or both, a major industrial bottleneck for the commercialization of ASSBs. Moreover, the substantial volume changes during Li plating/stripping induce mechanical deformation at the anode interface, a problem exacerbated under low stack pressures or when employing mechanically rigid protective layers [33]. These limitations highlight the need for self‑healing, mechanically adaptive interfaces that maintain stability under practical, low‑pressure ASSB operating conditions.

In this study, we address this low‑pressure operation challenge by introducing a mechanically adaptive composite interface that incorporates highly elastic polyrotaxanes (PRs) at the Li metal interface. PRs consist of a polymer threaded with rings that can freely slide on the backbone. Upon crosslinking these rings, the mechanically interlocked architecture endows the system with a pulley effect, dramatically improving elastic recovery [34]. Crosslinked PR‐based materials have previously been applied in batteries for their outstanding physical properties, such as binders for Si anodes [35], as solid polymers [36] or quasi‐solid polymer electrolytes [37], and as hybrid ceramic polymer electrolytes [38]. However, polymer electrolytes typically suffer from a loss of ionic conductivity, severely limiting the current densities that can be used.

To benefit from the mechanical properties of PRs without sacrificing the overall performance of the system, we propose to use a PR‐crosslinked poly(acrylic acid) (PR‐PAA) as a binder for InF₃ particles. InF₃ undergoes a spontaneous conversion reaction upon contact with Li metal, forming a Li–In alloy and lithium fluoride (LiF). Integrating PRs enables autonomous void restoration via the unique ring‐sliding mechanism of PR‐PAA, thereby supporting repeated Li plating/stripping under low stack pressure. The resulting synergistic interlayer, InF₃(PR‑PAA), delivers remarkable ASSB performance by retaining 82.2% of the initial capacity after 300 cycles and sustaining stable operation in a pouch‐cell configuration under commercially relevant conditions (25°C, 0.8 MPa, and an N/P ratio of 0).

2. Results and Discussion

2.1. Protective Layer Design With PRs

For commercial applications, such as in EVs, the applied stack pressure needs to be below 2 MPa [33]. However, most currently reported Li metal ASSBs operate under high stack pressures (i.e., exceeding 20 MPa), particularly when using a thin Li metal anode or anode‐less configurations [39, 40]. Critically, in low‐pressure environments, interfacial voids emerge during Li stripping [41, 42]. In this regard, the integration of highly elastic polymers could offer a mechanism to accommodate substantial volume changes to eliminate interfacial discontinuities during electrochemical cycling [43, 44]. Covering the Li metal anode with a mechanically adaptive protective layer, which effectively prevents Li dendrites from penetrating the electrolyte, alongside a highly elastic polymeric binder, allows the performance of Li metal ASSBs to be significantly enhanced, particularly under challenging conditions with minimal external pressure.

PR‐PAA was synthesized to serve as a mechanically adaptive binder for the interlayer. The design leverages the ring‐sliding architecture of PRs to impart elasticity, while poly(acrylic acid) provides chemical functionality for particle adhesion and interface stability. We synthesized a slide‐ring PR composed of a poly(ethylene glycol) (PEG) thread with α‐cyclodextrin (α‐CD) rings, which were subsequently crosslinked with PAA (Figure 1a). The PR was synthesized using a threading‐followed‐by‐stoppering method (see Figure S1 and the Experimental Section) [35, 45]. In this process, α‐CDs were threaded onto an acid‐functionalized PEG chain and capped with 1‐adamantylamine to prevent dethreading. To enhance the solubility, the hydroxyl groups of the α‐CDs were modified with propylene oxide by introducing 2‐hydroxypropyl moieties. The PR‐PAA was obtained by cross‐linking the α‐CD rings with PAA via ester bond formation using a 1,1'‐carbonyldiimidazole (CDI) cross‐coupling reaction (Figures S2–S5).

FIGURE 1.

Chemical structure of the PR‐PAA and schematic illustration of the molecular pulley effect. (a) Chemical structure of the rotaxane‐based polymer. (b) Schematic illustrations of ring motions within the PRs in the polymer. (c) Depiction of the InF₃(PR‐PAA) adaptive interlayer. (d, e) Nanoindentation profiles of the InF₃ layer with the (d) PVDF and (e) PR‐PAA polymers. (f) Elasticity parameters derived from the nanoindentation results.

Unlike covalently crosslinked polymers, the mechanically interlocked structure of PR allows the crosslinks to move freely along the PEG threads (Figure 1b). When stretched, the sliding motion functions analogously to a pulley system to mitigate the accumulation of tension within the polymer network and provide a restoring force upon stress release. The incorporation of 5 wt% PR into the PAA matrix yields PR‐PAA, a highly stretchable and elastic polymer network. In its pristine state, the functionalized α‐CD rings are randomly distributed along the PEG threads. During Li plating, the ability of the α‐CD rings to slide along the PEG chains can accommodate the volume expansion at the anode interface, where it preserves physical contact and prevents mechanical failure. Conversely, during Li stripping, the entropic forces generated by the uneven distribution of the α‐CD rings upon Li plating can alleviate mechanical stress (Figure 1c).

InF₃‐based materials spontaneously react with Li metal to form a Li‐In alloy and LiF at the interface [46], both of which contribute to suppressing dendritic growth. The higher reduction potential of the Li‐In alloy compared to that of Li metal also reduces undesirable electrolyte reduction at the interface. Simultaneously, in situ generated LiF, as a result of its high interfacial energy with Li metal, serves as a robust barrier to protect the SE against dendrite penetration [47, 48]. Building on this concept, the PR‐based ring‐slide polymer—capable of significantly lowering the required external stack pressure—was integrated into the InF₃‐based protective layer. Together, the InF₃(PR‐PAA) interface enables low‐pressure operation through a synergistic mechanism: InF₃ chemically suppresses Li dendrite growth, minimizing void initiation during stripping, while the PR‐PAA binder exploits the entropic restorative force of its molecular‐pulley architecture to drive elasticity—preserving interfacial contact without high external stack pressure.

To probe the impact of PR‐PAA on the mechanical properties of the InF₃ protective layer, we performed nanoindentation analysis (Figure 1d–f). For comparison, PVDF, a widely used polymeric binder for electrodes and protective layers, was employed as a control. Thus, two different protective layers were prepared by mixing InF₃ powder (Figure S6) with each of the respective polymers (PR‐PAA or PVDF, 4 wt%) to form a homogeneous slurry, which was applied to stainless‐steel (SUS) substrates. The results revealed that the elasticity of the PR‐PAA‐based layer was superior to that of the PVDF, as demonstrated by the elastic recovery during the unloading phase, where the indentation depth was effectively restored (Figures S7 and S8). This effect was further validated by analyzing the depth profile over time (Figure S9). The elastic recovery ratio, representing the percentage of depth recovered during unloading [49], was 6.3 times higher for the PR‐based layer compared to the PVDF‐based one. Furthermore, integration of the nanoindentation load‐displacement profiles enabled quantification of both the elastic and plastic work (Figure S10). The elastic‐to‐plastic work ratio (E/P ratio) for the PR‐based layer was 5.1 times higher than for the PVDF. Detailed formulas and numerical data are provided in Table S1. These findings demonstrate that PR‐PAA imparts flexibility to the protective layer relative to PVDF, a critical step toward more resilient interfaces.

2.2. Assessment of Li Symmetric Cell

The InF₃ protective layer was applied to the interface of the Li metal anode of an all‐solid‐state cell for electrochemical evaluation. X‐ray diffraction (XRD) confirmed the spontaneous formation of LiF and Li‐In alloys through a conversion reaction (Figure 2a,b). This reaction did not involve any electrochemical processes and occurred simply by resting the cell for 6 h after fabrication, demonstrating the favorable conversion properties of InF₃. Critical current density (CCD) measurements were then performed using a Li‖Li symmetrical cell with Li₆PS₅Cl (LPSCl) SE. The current density was initially set at 0.1 mA cm⁻² and increased in steps of 0.1 mA cm⁻², followed by larger increments of 0.2, 0.3, and 0.5 mA cm⁻² (Figure 2c). The electrochemical tests were performed under stringent conditions, with each step maintained for 1 h, and both the current density and capacity continuously increased over time. The measurements, at two different stack pressures—15 MPa (Figure 2d−f) and 2 MPa (Figures 2g,h and S11)—revealed that at 15 MPa, the InF₃ layer significantly enhanced the CCD compared to that of the bare cell. Even though the overpotential was notably lower with the PR‐PAA, indicating a less resistive interface, the CCD remained unchanged at 2.3 mA cm⁻² for both polymers.

FIGURE 2.

XRD analysis and critical current density (CCD) measurement of Li‖Li symmetric cells. (a, b) XRD patterns of the InF₃ layer (a) before and (b) after cell fabrication. (c) Current density profile for CCD evaluation of the Li‖Li symmetric cell. Voltage versus time profiles of the CCD measurements of the Li‖Li symmetric cells at 15 MPa for (d) bare, (e) InF₃(PVDF), (f) InF₃(PR‐PAA), and at 2 MPa for (g) InF₃(PVDF), (h) InF₃(PR‐PAA).

To enable testing at the desired low stack pressure, a spring was integrated into the pellet‐based cell assembly (Figure S12). By taking the spring constant and compression depth into account, pressures were controlled with high precision, enabling increments as small as ∼0.01 MPa to be achieved. This configuration was also employed in subsequent full‐cell evaluations conducted at 0.8 MPa. Testing at 2 MPa in the bare and both polymer/InF₃‐based cells revealed increased overpotential and signs of contact loss toward the end of stripping, as indicated by the rising voltage. Despite these effects, the CCD values for the bare and InF₃ (PVDF) cells remained comparable to those at high stack pressure, 1.2 and 2.5 mA cm⁻², respectively, since dendrite growth is generally suppressed under low‐pressure conditions [50]. In sharp contrast, cells employing the PR‐PAA binder achieved a significantly higher CCD of 3.9 mA cm⁻² at the same pressure, benefiting from reduced dendritic growth and improved interfacial contact. This result underscores that the slide‐ring effect of the PR‐PAA is beneficial for improving the mechanical properties and electrochemical performance under low‐pressure conditions.

To further elucidate the effect of elasticity afforded by PR‐PAA, we prepared a control protective layer using a PR/PAA physical blend. This composite contained the same weight ratio of PAA and PR as PR‐PAA, but without crosslinking; as such, it cannot exhibit the PR pulley effect. The CCD measured at 2 MPa for the PR/PAA blend was 2.0 mA cm⁻² (Figure S13), nearly identical to that of the PVDF‐based interface. This result demonstrates that simple physical blending of components is insufficient to provide the necessary mechanical resilience.

The Li‖Li symmetric cell with InF₃(PR‐PAA), operated at a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻² over extended periods (Figure 3a,b), outperformed both the bare configuration and the cell with InF₃(PVDF). The performance difference was especially pronounced under low‐pressure conditions of 2 MPa and was consistent with the trends observed in the CCD evaluations. Cross‐sectional scanning electron microscopy (SEM) images of the Li‖Li symmetrical cells after cycling revealed distinct morphological differences (Figures 3c−h and S14). The polymer/InF₃‐based cells were analyzed after 200 cycles, whereas the bare cell was examined after only 10 cycles because it short‐circuited early on in the measurement. The bare configuration (Figure 3c,f) without a protective layer resulted in significant Li dendrite penetration and void formation at the interface. In contrast, the InF₃ protective layer with both elastic and inelastic binders effectively mitigated these issues at a high stack pressure of 15 MPa. However, for the InF₃(PVDF) configuration (Figure 3d,g), contact loss at the interface was still apparent at 2 MPa, and horizontal voids began to emerge, suggesting that the negative stress resulting from Li stripping significantly contributed to contact loss with the protective layer. In comparison, the InF₃(PR‐PAA) layer (Figure 3e,h) maintained interfacial contact even at 2 MPa. This result verifies the effectiveness of the PR‐PAA system in forming a mechanically adaptive interface at low stack pressure.

FIGURE 3.

Li‖Li symmetric cell evaluations and cross‐sectional SEM images after cycling. (a, b) Li‖Li symmetric cell evaluation with a current density of 1 mA cm⁻² and a capacity of 1 mAh cm⁻². (c−h) Cross‐sectional SEM images of the Li‖Li symmetric cells after cycling at (c−e) 15 MPa and (f−h) 2 MPa stack pressure.

To further investigate interfacial morphology, we employed x‐ray computed tomography (x‐ray CT) (Figure S15). This technique exploits the x‐ray transmittance contrast between solid components and voids, enabling precise identification of crack formation and location. After Li‖Li symmetric cell operation at 2 MPa, 3D tomographic analysis of the interface revealed distinct voids in the PVDF‐based electrode, consistent with SEM observations. In contrast, the PR‐PAA‐based interface remained intact with no mechanical failure, confirming superior mechanical stability in 3D.

The Li‖Li symmetrical cells were additionally subjected to chronopotentiometry analysis with a continuous one‐sided current of 1 mA cm⁻² (Figure 4). These analyses were followed by hourly in situ electrochemical impedance spectroscopy (EIS) measurements. The cell lifetimes declined in the following order: InF₃(PR‐PAA) > InF₃(PVDF) > bare cells (Figure 4a,c,e). For the bare cell, the Nyquist plots based on in situ EIS showed a significant increase in resistance over successive measurements, indicating degradation at the Li metal interface (Figure S16). Both the bulk resistance (R _bulk), represented by the x‐intercept, and the charge transfer resistance (R _CT), represented by the diameter of the semicircle, increased substantially [51]. Apart from this, the additional semicircle that appeared in the high‐frequency region of the Nyquist plot from the fifth sequence corresponds to the SEI resistance (R _SEI) (Figure 4b). This result indicates that Li dendrite penetration led to the formation of an additional SEI layer. In contrast, the absence of this additional semicircle from the plots of the cells with the InF₃ protective layer (i.e., negligible R _SEI) confirmed the effective suppression of electrolyte degradation (Figure 4d,f). The continued increase in both R _bulk and R _CT of the PVDF‐based cell as opposed to the minimal increase in the corresponding values of the InF₃(PR‐PAA)‐based one highlights the crucial role of PR‐PAA in maintaining the internal resistance and ensuring overall stability.

FIGURE 4.

In situ EIS during chronopotentiometry of the Li‖Li symmetric cells. Voltage profiles and corresponding Nyquist plots obtained from in situ EIS after 5 charging sequences (5 mAh cm⁻²) for the cells with the (a, b) bare, (c, d) InF₃(PVDF), and (e, f) InF₃(PR‐PAA) interfaces.

2.3. Full‐Cell Evaluations Under Practical Cell Conditions

Subsequently, full‐cell tests were conducted by pairing LiNbO₃‐coated LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) cathodes with Li metal foil‐based anodes with and without the protective layers. Keeping in mind that the energy density can be increased by using a thinner Li metal anode to lower the N/P ratio, cells with an N/P ratio of 2 were tested using a 20 µm‐thick Li metal anode. The full cells incorporating the InF₃(PR‐PAA) layer demonstrated superior performance at a stack pressure of 15 MPa (Figure 5a−d). Although the bare and InF₃(PVDF)‐based cells experienced internal short circuits after 63 and 149 cycles, respectively, the InF₃(PR‐PAA)‐based cell retained 82.2% of its capacity over 300 cycles and operated stably without internal short‐circuiting. Further testing at a significantly reduced stack pressure of 0.8 MPa (Figure 5e) revealed significant degradation in the bare cell without a protective layer, which quickly reached the upper cutoff voltage of 4.2 V and charged only in the constant voltage (CV) region (Figure S17). The InF₃(PVDF) cells also showed signs of early degradation, with the Coulombic efficiency (CE) dropping below 90% after only 20 cycles. In contrast, the InF₃(PR‐PAA) cell maintained 70% of its capacity over 100 cycles without significant degradation, demonstrating its effectiveness under a commercial‐level stack pressure of 0.8 MPa.

FIGURE 5.

Full‐cell evaluations with a thin Li metal anode (20 µm thick) under practical operating conditions. (a−c) Voltage profiles of the full cells at a current density of 0.5C under 15 MPa stack pressure. Cycling performance of the full cells (d) at a current density of 0.5C under a stack pressure of 15 MPa and (e) at a current density of 0.3C under 0.8 MPa stack pressure (1C = 2 mA cm⁻²).

Next, we evaluated full cells using an anode‐less configuration (Figure 6); that is, the electrode employed as the anode was prepared by omitting the Li metal that would typically be applied to the current collector. This configuration, with an N/P ratio of zero, represents the optimal approach for maximizing the energy density and processability [52, 53]. Instead of the Li metal, a 200 nm‐thick Mg film was deposited underneath the InF₃‐based protective layer to promote stable Li plating. Using this electrode, the mechanical self‐healing effect of PR‐PAA was verified by monitoring the cell pressure in real time (Figure 6a,b). Stack pressure (15 MPa) was applied at the beginning of the experiment, and the cell was charged and discharged while maintaining a fixed displacement. This setup (Figure S18) enabled pressure fluctuations to be continuously monitored within a fixed volume to gain real‐time insights into internal mechanical changes. During the initial charging phase, both the InF₃(PVDF) and InF₃(PR‐PAA) cells exhibited the same pressure increase of +3.5 MPa, confirming that the volume expansion due to Li deposition during charging was nearly the same for both polymer types. However, a notable difference emerged during discharge. After the second discharge cycle, the stack pressure in the PVDF‐based cell decreased by –0.5 MPa relative to the initial 15 MPa, whereas that of the PR‐PAA changed by only –0.1 MPa. This post‐cycle pressure drop is directly linked to mechanical degradation, such as void formation at the electrode interface. Once again, these results emphasize the effective self‐healing properties of the PR‐PAA through the mechanical motions of its rings, to effectively mitigate negative stresses during the discharge process.

FIGURE 6.

Real‐time stack pressure monitoring and pouch‐cell evaluations with anode‐less configuration (N/P ratio = 0). (a, b) Stack pressure in the InF₃‐based full cells monitored in real time. (c, d) Performance evaluation of the InF₃(PR‐PAA)‐based full cell in a pouch‐cell configuration. (c) Cycling performances and (d) corresponding voltage profiles at 25°C and 0.8 MPa. (e) Performance comparison with previously reported studies on low‐pressure Li metal ASSBs.

In addition to these pellet‐based assessments, the InF₃(PR‐PAA)‐based anode‐less electrode was also evaluated in a pouch‐cell configuration (333 Wh L⁻¹, Table S2) under practically viable operating conditions (25°C, 0.8 MPa, and an N/P ratio of zero). To investigate initial‐cycle charging behavior, we measured thickness changes at six locations of the pouch cell before and after the first charge (Figure S19). The increase at all sites closely matched the theoretical value for Li metal deposition relative to capacity (5 µm thick per 1 mAh cm⁻²), confirming that Li‐ions extracted from the cathode were effectively deposited within the anode‐less layer. This finding was further supported by post‐charge SEM‐EDS analysis. Both top‐view (Figure S20a) and cross‐sectional (Figure S20b) images show stable Li deposition within and beneath the protective interface layer.

The anode‐less pouch‐cell operated successfully without internal short‐circuiting, demonstrating capacity retention of 85.1% over 100 cycles at a capacity of 16.2 mAh (Figure 6c,d). Compared to previously reported performance metrics for Li metal ASSBs (Figure 6e) [22, 54, 55, 56, 57], this result represents a significant advancement that is aligned with practical requirements in terms of the operating temperature, pressure, and N/P ratio (equally, the thickness of the Li metal anode). These findings indicate that interfacial contact loss at the anode remains the key bottleneck for low‐pressure operation in anode‐less or Li metal ASSBs. Overcoming this requires not only suppressing dendrites but also an intrinsic elasticity, thereby reducing reliance on high stack pressure.

3. Conclusion

In summary, the integration of InF₃ and PR‐PAA into a protective layer for the Li metal anode has enabled Li metal ASSBs to operate stably at room temperature (25°C) with a low N/P ratio (even as low as zero in the anode‐less configuration) and under a low stack pressure (0.8 MPa). The synergistic effects of InF₃—which suppresses Li dendritic growth by undergoing spontaneous conversion into a Li–In alloy and LiF—and the mechanically interlocked PR–PAA polymer network—which imparts elasticity via a slide‐ring mechanism—coherently maintain the structural integrity even under low‐pressure conditions. This approach not only improves the performance of Li metal ASSBs under practically viable conditions but also highlights the usefulness of an advanced molecular structure with mechanically interlocked configurations to keep interfaces void‐free at low stack pressures.

Author Contributions

Jihoon Oh: conceptualization, data curation, visualization, investigation, methodology, formal analysis. Leonie Braks: methodology, resources, validation. Ali Coskun: supervision, writing – review and editing, funding acquisition. Jang Wook Choi: funding acquisition, writing – review and editing, writing – original draft, project administration, resources, supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72660‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available on Zenodo (zenodo.org).