Fully recyclable, catalyst-free, highly adhesive, and resilient poly(β-amino esters) covalent adaptable network-based solid polymer electrolytes for lithium metal batteries

Abstract

Despite the widespread adoption of lithium-ion batteries, ensuring the safety, durability, and sustainability of current systems is still a critical challenge. Solid polymer electrolytes (SPEs) have emerged as a safer and more durable alternative to the conventional liquid lithium-ion-battery electrolytes, offering additional benefits such as flexibility, ease of thin-film processing, and mechanical stability. To achieve high lithium-ion conductivity, good interfacial adhesion, and mechanical integrity, crosslinked rubbery polymers with low glass transition temperatures are often used as SPEs. However, their permanent crosslinks make them difficult to reprocess and recycle. To address these limitations, poly(β-amino ester) (PBAE)-based covalent adaptable networks (CANs) are prepared in this work as fully recyclable, catalyst-free, highly adhesive, and resilient SPEs. The adhesive and dynamic bond exchange characteristics, along with the lithium-ion conductivity of the PBAE CANs with varying crosslink densities, are systematically investigated. The obtained PBAE-CAN-based SPEs exhibit exceptional adhesive properties, recyclability, and an ionic conductivity of approximately 10⁻⁶ S cm⁻¹ at room temperature. This conductivity can be further increased by an order of magnitude with the addition of a plasticizer. Long-term performance tests conducted at room temperature demonstrate stable operation for over 1000 h without internal short circuits, attributed to the excellent creep recovery of the SPE at temperatures below the topology-freezing transition temperature where significant dynamic bond exchange begins to occur. Furthermore, full cell tests using LiFePO₄ (LFP)‖Li configurations demonstrate the practical viability of the electrolyte, exhibiting stable rate performance and excellent capacity retention even after cycling at high C-rates. To further highlight its sustainability, the SPE is successfully reprocessed, allowing its smooth reuse. Furthermore, eco-friendly depolymerization and recovery of the lithium salt from the used PBAE CAN-based SPE are also demonstrated.

---

Chae Bin Kim

Chae Bin Kim is an Associate Professor at the Department of Polymer Science and Engineering at Pusan National University (PNU), South Korea. His research focuses on interfacial phenomena in polymers, including covalent adaptable networks, recyclable composites, functional adhesives, and thin films/coatings. He earned his PhD in Chemical Engineering from the University of Texas at Austin in 2016 and completed a postdoctoral appointment at the Korea Institute of Science and Technology (KIST) before joining PNU.

---

Introduction

The rapid growth of portable electronics, electric vehicles, and renewable energy systems has fueled an increasing demand for advanced energy storage technologies. Lithium metal batteries (LMBs) are a leading candidate technology for meeting this demand owing to their high energy density, long cycle life, and reliability.^1–3 However, the conventional liquid electrolytes used in LMBs pose significant safety risks owing to their flammability and leakage which can lead to catastrophic failures. With increasing shift of the energy storage industry landscape toward higher efficiency and sustainability, the development of solid-state electrolytes (SSEs) has gained significant research attention. Among the different kinds of lithium-ion conductive SSEs, solid polymer electrolytes (SPEs) offer a promising solution by combining flexibility and stable mechanical properties with enhanced safety, absence of leakage, light weight, and ease of processing.^4–6

Crosslinked rubbery polymers play a crucial role in the development of high-performance SPEs by offering a unique combination of lithium-ion conductivity and mechanical stability.⁷ For these materials, the glass transition temperature (T_g) is below room temperature, which ensures a soft and flexible matrix at operating temperatures, thereby promoting the segmental motion of polymer chains and facilitates lithium-ion transport. This demonstrates the strong relationship between ionic conductivity and T_g. Additionally, rubbery polymers can provide mechanical stability through crosslinking, maintaining structural integrity under operational stresses while suppressing the growth of lithium dendrites.^8,9 The inherent elasticity and enhanced chain mobility of rubbery polymers further improve their conformability to electrode surfaces, thereby reducing interfacial resistance and enhancing overall battery performance.⁹ This combination of properties makes crosslinked rubbery polymers ideal candidates for achieving the delicate balance of properties required in SPEs, enabling safer, more efficient, and durable LMBs.

Unfortunately, rubbery polymers have the disadvantage of being difficult to reprocess or recycle after curing owing to their permanent covalent bonds.^10,11 Covalent adaptable networks (CANs) can be used to overcome this limitation and enable recycling of rubbery polymers.^12–14 CANs possess crosslinked structures similar to those of thermosetting polymers but allow for reprocessing/recycling through dynamic-bond-exchange reactions. These exchange reactions can be triggered by external stimuli such as light, pH changes, and most commonly, heat. The dynamic-bond-exchange reactions can be activated above the topology-freezing transition temperature (T_v). Above T_v, CANs become malleable and can be processed with heat and pressure. Conversely, below T_v, CANs exhibit extremely slow or dormant bond-exchange reactions, showing behavior similar to that of conventional thermosetting polymers.

Various efforts have been made to develop CAN-based SPEs and demonstrate repairing, reprocessing, and recycling.^10,15–19 Furthermore, some studies have reported that CANs could enhance long-term performance by inhibiting lithium dendrite growth and providing strong adhesion to electrodes. Wang et al. reported that a CAN-based SPE with dynamic imine bonds effectively inhibited lithium dendrite growth and exhibited good interfacial compatibility with electrodes owing to solid-state plasticity.¹⁵ The dynamic bond exchange enhanced electrolyte flowability and enabled self-healing during lithium deposition. Bao and coworkers achieved high mechanical resilience by using a dual covalent network combined with dynamic hydrogen bonds, demonstrating robust long-term stability without sacrificing room-temperature ionic conductivity.²⁰ Yin et al. developed an elastomeric SPE with dual-bond crosslinking, combining chemically and mechanically reversible bonds, to achieve a balance of high ionic conductivity, elastic resilience, and recyclability.¹⁰ Among the approaches mentioned above, dual- or multi-network strategies that incorporate both permanent and dynamic crosslinks have effectively enhanced mechanical properties without sacrificing ionic conductivity. However, in previous studies, the multi-network strategy was apparently necessary owing to the continuous activity of dynamic bonds even at room temperature or during battery operation. Based on these observations, we postulated that a CAN with a low T_g but with a sufficiently high T_v would exhibit adequate resilience and adhesion during operation with dormant bond exchange reactions (high T_v) while maintaining enhanced ionic conductivity (low T_g). Additionally, such a CAN can be reprocessed and recycled on demand upon heating to a temperature above T_v.

Following this approach, a series of poly(β-amino esters) (PBAE) CANs with sufficiently high T_v (>100 °C) but low T_g (below room temperature) were prepared via the aza-Michael addition reaction²¹ between glycerol 1,3-diglycerolate diacrylate (TGDA) and poly(propylene glycol)bis(2-aminopropyl ether) (PEA). Fig. 1 schematically illustrates the basic concept of the present work. PBAE CANs were chosen because they can undergo dual dynamic bond exchanges through transesterification and dynamic aza-Michael reaction.¹² Dynamic-bond-exchange reactions are facilitated more strongly by the use of dual or multiple dynamic networks compared with the use of a single dynamic network, enabling faster processing at moderate temperatures.²² Additionally, β-amino esters^12,23–25 and β-hydroxyl groups²³ in PBAE enable dynamic exchange under mild conditions even without the use of external catalysts. This feature is crucial for the long-term stability of the materials because external catalysts are often corrosive and can pose challenges with regard to compatibility and oxidative degradation.^26,27 Furthermore, the propylene oxide (PO) groups within the backbone structure of the PBAE CANs create an amorphous solid, avoiding the formation of crystalline domains typically observed in poly(ethylene oxide) (PEO) that hinder ionic conductivity.¹⁹ Additionally, the presence of additional hydroxyl groups enhances the dissociation of Li⁺ and its corresponding anion while reducing anion diffusion, thereby lowering local energy barriers and improving lithium-ion conductivity.²⁸ The crosslink density, T_g, and T_v were adjusted by using PEA with different molecular weights (230, 2000, and 4000 g mol⁻¹). The effects of the crosslink density, T_g, and T_v on the ionic conductivity, adhesion, and CAN properties were systematically investigated. Using the optimized electrolyte system, stable operation was achieved in both lithium symmetric and full cells. Furthermore, the reprocessability and recyclability of the electrolyte were verified; thus, this study provides valuable insights into the development of recyclable batteries and sustainable energy storage technologies.

Fig. 1 (Middle) Schematic representation of the current work using PBAE-CAN-based SPEs with dual dynamic bonds (transesterification and dynamic aza-Michael addition). (Left) By varying the crosslink (or bond-exchange-site) density, PBAE CAN with a sufficiently high T_v (>100 °C) but low T_g (below room temperature) can be obtained, maintaining resilience while enhancing adhesion. (Right) The PBAE-CAN-based SPEs were reprocessable, and lithium salt recovery from the used SPEs was also demonstrated. The purity of the recovered lithium salt was confirmed by ¹³C-NMR and ATR-FTIR analyses.

Experimental

Materials

TGDA and PEA (M_n ∼ 230, 2000, 4000), lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) used as the lithium salt, and tetraethylene glycol dimethyl ether (TGDE) used as the plasticizer were purchased from Merck (South Korea). Lithium metal (200 μm thick) for anode preparation was supplied by HONJO Metal Co., Ltd (Japan). Tetrahydrofuran (THF, 99.5%) was obtained from Samchun Pure Chemical Co., Ltd (South Korea). LiFePO₄ (LFP) as the active material, HSV900 poly(vinylidene fluoride) (PVDF) as the binder, conductive acetylene black (AC) as the conductive additive, and N-methyl-2-pyrrolidone (NMP) as the solvent were all purchased from MTI Corporation (South Korea). All chemicals were used as received without further purification. The SUS304 substrate was purchased from POSCO (South Korea), and carbon-coated aluminum foil was also obtained from MTI Corporation (South Korea).

PBAE CAN synthesis

PBAE CANs (denoted as PBAE_X, where X represents the molecular weight of PEA) were synthesized by reacting PEA and TGDA in a molar ratio of 1 : 2 via aza-Michael addition. A mixture of PEA and TGDA was blended for 5 min at 2000 rpm using a THINKY mixer (AR-100, Thinky Corporation, Japan). Then, the resulting mixture was placed in a disk-shaped stainless-steel mold and cured at 120 °C using a heating press (QMESYS, QM900S, South Korea) at 15 MPa for 1, 4, and 6 h for PBAE_230, PBAE_2000, and PBAE_4000, respectively, unless otherwise noted. Curing was conducted in an ambient air atmosphere.

Material characterization

Attenuated total reflection-Fourier transform infrared (ATR-FTIR) measurements were performed using a Jasco FT/IR-4X spectrometer (Japan). The absorbance spectrum was measured from 4000 to 650 cm⁻¹ with 32 scans at a resolution of 8 cm⁻¹. All ATR-FTIR absorbance spectra were normalized by using the area under the invariant C=O peak near 1720 cm⁻¹. The gel content (GC) was characterized by comparing the mass changes in the samples with curing time at 120 °C in an ambient air atmosphere. For the GC measurements, the sample was soaked in THF for 24 h at room temperature. THF was selected because it is a good solvent for both PEA and TGDA. Then, the remaining gel sample was filtered and vacuum-dried at room temperature for 24 h, followed by additional heating at 60 °C for 24 h. The GC was calculated according to

GC (%) = m_f/m_i × 100 (1)

where m_i and m_f are the initial and final masses of the sample, respectively.

Differential scanning calorimetry (DSC) measurements were conducted in a nitrogen atmosphere using a DSC 25 instrument (TA Instruments, USA). The sample (approximately 10 mg) was added to an aluminum pan. The process involves heating from 40 to 150 °C, cooling down to −80 °C, and then reheating to 150 °C. The heating rate was 5 °C min⁻¹ and T_g was reported based on the second heating cycle. Thermogravimetric analysis (TGA) was performed using a Q50 instrument (TA Instruments, USA). The sample (approximately 10 mg) was added to a platinum pan and the measurements were performed in the temperature range of 40–800 °C, with a heating rate of 5 °C min⁻¹.

To conduct the amplitude sweep and temperature sweep, a rheometer was used with an MCR 702e instrument (Anton Paar GmbH, Austria). The specimen size was measured from a circular sample with a diameter of 25 mm. A parallel plate with a diameter of 25 mm was utilized for geometry measurements. Amplitude sweep experiments were conducted using an axial force of 5 N, a strain range from 0.01% to 10%, an angular frequency (ω) of 1 rad s⁻¹, and a temperature of 150 °C. Temperature sweep experiments were conducted in an ambient air atmosphere in the temperature range of −80–150 °C at a ramping rate of 3 °C min⁻¹, with an axial force of 5 N, frequency of 1 Hz, and strain of 0.01% for PBAE_230 and of 0.1% for PBAE_2000 and PBAE_4000. To conduct the stress relaxation test, a rheometer was used with a Discovery HR 20 instrument (TA Instruments, USA). The specimen, with a circular shape and a diameter of 25 mm, was measured using a 25 mm parallel-plate configuration to assess its geometry. Stress relaxation experiments were conducted with an axial force of 5 N and a strain of 1%. The relaxation time τ of PBAE CANs was calculated using the stretched Maxwell equation (eqn (2)).

G/G₀ = exp(−t/τ)^β (2)

where G, G₀, t, and β are the shear stress, maximum (or initial) stress, time, and stretch exponent, respectively.

Adhesive property characterization

The adhesive properties (or lap shear strengths) were evaluated using a universal testing machine (UTM) (QM100TM, QMESYS, South Korea). A layer of the monomer mixture with a length, width and thickness of 25.4 mm, 12.7 mm, and 100 μm, respectively, was applied to a SUS304 plate using a doctor blade and cured following the above-described PBAE CAN synthesis procedure. The lap shear test was performed at an extensional speed of 1.27 mm min⁻¹. The UTM's load cell is rated at 200 kgf, and measurements were performed until fracture occurred. Water contact angle experiments were conducted using a contact angle goniometer (PHEONIX-300, SEO, South Korea). PBAE CAN films were heated in a flat position on a glass substrate. The contact angle was measured by placing a droplet of deionized water with a volume of 20 μL. Once the droplet was formed, it was allowed to stabilize for approximately 10 s, and then the contact angle was measured.

Characterization of the PBAE-CAN-based SPE

To evaluate the lithium-ion conductivities of the PBAE CANs, SPEs based on PBAE CANs were prepared and labeled as CSPE_X, where X denotes the molecular weight of PEA. The same monomers used for PBAE CANs (TGDA and PEA) were combined with the LiTFSI salt. LiTFSI, at 50 wt% relative to the CAN matrix, was dissolved in THF to create a homogeneous solution. The solution was heated at 100 °C under vacuum for 1 h to evaporate the THF. Subsequently, CSPE_230 and CSPE_2000 were fully cured by heating at 100 °C for 1 h, whereas the highly viscous CSPE_4000 required extended curing at 100 °C for 24 h. To enhance Li⁺ conductivity, the PBAE-CAN-based gel polymer electrolyte (denoted as CGPE_2000) was prepared by curing TGDA and PEA (M_n ∼2000) with a liquid electrolyte consisting of 1 M LiTFSI in a TGDE plasticizer. For the preparation of CGPE_2000, LiTFSI was dissolved in TGDE, which served as both a co-solvent and a plasticizer. The TGDE content was fixed at 23.5 wt% relative to the total formulation (PBAE + LiTFSI + TGDE), and no additional organic solvent was used. The polymer matrix constituted 70% of the total mass, with the liquid electrolyte constituting the remaining 30%. Then, the mixture was fully cured at 100 °C for 2 h.

To measure the rheological properties of the polymer electrolyte, experiments were carried out using a rheometer (Discovery HR-2 instrument, TA Instruments, USA) with an 8 mm parallel plate geometry. The dynamic oscillation frequency test was also performed at a constant strain of 0.1% within the linear viscoelastic region, over a frequency range of 0.1–100 rad s⁻¹. Compression creep experiments were conducted using a rheometer (MCR 702e, Anton Paar GmbH, Austria). The specimens were circular with a diameter of 25 mm. A 25 mm parallel plate was used as the measuring geometry. The tests were performed at a compression velocity of 10 μm s⁻¹ for 1 min at room temperature (RT), and at 60 and 80 °C. The ionic conductivity (σ_DC) of the CSPE and CGPE samples was tested by electrochemical impedance spectroscopy (EIS) using a BioLogic VSP-300 Potentiostat (BioLogic, France) with a BioLogic CESH leak-tight sample holder and a BioLogic ITS temperature-controlled system (BioLogic, France) in the frequency range from 10 MHz to 0.1 Hz. To conduct the EIS experiment, the cell was assembled in the configuration of stainless steel (SS)|electrolyte|SS. The ionic conductivities (σ_DC) of CSPEs and CGPE were calculated as

where R, l, and A are the resistance, thickness, and area of the sample, respectively.

The lithium transference number (t_Li⁺) of CGPE_2000 was determined using the Bruce–Vincent method^29,30 and calculated according to

t_Li⁺ = I_S(ΔV − I₀R₀)/I₀(ΔV − I_SR_S) (4)

where I₀ is the initial state current, I_S is the steady-state current, R₀ is the initial resistance, R_S is the steady-state resistance, and ΔV is the polarization potential for a Li|CGPE_2000|Li symmetric cell at ΔV = 10 mV. All tests were performed at room temperature.

Lithium stripping/plating cycling tests were conducted under ambient conditions using Li‖Li symmetric cells. The tests were performed by applying alternating current densities ranging from 0.01 to 0.05 mA cm⁻² with 1 h for each charge and discharge step, and then from 0.01 to 0.1 mA cm⁻² with 0.25 h per step with CGPE_2000. The lithium symmetric cell using CGPE_2000 (5 : 5) was tested at current densities ranging from 0.01 to 0.1 mA cm⁻², with 1 h for each charge and discharge step. The Li|CGPE_2000|LFP cell rate performance was measured at different rates of 0.1, 0.2, 0.5, 1, and 2C between 2.5 and 4.2 V at 298 K.

Density functional theory (DFT) calculations were performed using the Gaussian 09 software package with the B3LYP functional and the 6-311G(d,p) basis set. The binding energy (E_b) was calculated as

ΔE_b = E_A,B − E_A − E_B (5)

where A and B represent the components and ΔE_b is the binding energy between A and B. E_A,B is the DFT total energy of the corresponding system, and E_A and E_B are the energies of A and B, respectively.

LiTFSI extraction from the used PBAE CAN-based SPE

First, CSPE_2000 was swollen in THF and then vacuum-filtered to separate LiTFSI dissolved in THF from the PBAE CANs. The filtrate contained LiTFSI along with trace amounts of soluble PBAE components. To isolate LiTFSI, the filtrate was heated at 330 °C in ambient air using an oven. During this process, most organic components underwent oxidative degradation, leaving LiTFSI with a trace of carbonaceous residues. The resultant sample was thoroughly washed several times with deionized water and then filtered. The final filtrate contained only LiTFSI dissolved in water. By evaporating the water, pure LiTFSI was recovered. The recovered LiTFSI was analyzed using ATR-FTIR and carbon nuclear magnetic resonance (¹³C-NMR). ATR-FTIR analysis was performed using a Jasco FT/IR-4X spectrometer (Japan). All ATR-FTIR absorbance spectra were normalized to the area under the invariant S=O peak near 1240 cm⁻¹. ¹³C-NMR measurements were conducted using a Bruker AVANCE NEO 500 spectrometer (Germany), with both LiTFSI and extracted LiTFSI dissolved in D₂O solvent.

Results and discussion

Synthesis of the PBAE CANs

PBAE CANs were synthesized through the aza-Michael addition reaction between TGDA and PEA in a molar ratio of 2 : 1 (Fig. 2a). To investigate the effect of the crosslink density, three different molecular weights of PEA (230, 2000, and 4000 g mol⁻¹) were used. The liquid monomers were thoroughly mixed for 5 min at room temperature using a THINKY mixer, and then cured at 120 °C under a heating press for 1, 4, and 6 h for the PEA molecular weights of 230, 2000, and 4000 g mol⁻¹, respectively. The resulting PBAE CANs were denoted as PBAE_X, where X is the molecular weight of PEA. The aforementioned curing conditions were determined using GC, rheometer, and ATR-FTIR measurements, with the results presented in Fig. 2b, c and S1, respectively. The GC was measured as a function of curing time at 120 °C using THF as the solvent. As shown in Fig. 2b, the gel fraction of PBAE CANs increased upon increasing the curing time, reaching a plateau after curing for 1, 4, and 6 h at 120 °C for PBAE_230, 2000, and 4000, respectively. As shown in Fig. S1a–c, upon heating the monomer mixtures at 120 °C, the intensities of the characteristic peaks of the TGDA and PEA monomers (specifically, the C=C peak of TGDA at 1630 cm⁻¹ and the N–H bending peak of PEA at 1550 cm⁻¹) decreased monotonically. A significant reduction in these C=C and N–H peaks was observed after approximately 1, 4, and 6 h of heating at 120 °C for PBAE_230, PBAE_2000, and PBAE_4000, respectively, consistent with the GC results shown in Fig. 2b. Therefore, fully cured PBAE CANs were prepared by heating the monomer mixture at 120 °C for 1, 4, and 6 h, respectively, unless otherwise noted.

Fig. 2 (a) A schematic illustrating the synthesis of PBAE CANs via aza-Michael addition between PEA and TGDA in a molar ratio of 1 : 2 using PEA with three different molecular weights (230, 2000, and 4000 g mol⁻¹) at 120 °C. The resulting PBAE CANs contained β-amino ester groups and β-hydroxyl groups as highlighted by the red solid box and blue dashed box, respectively. These CANs were denoted as PBAE_X, where X represents the molecular weight of PEA. (b) GC measured as a function of curing time at 120 °C for the PBAE CANs, expressed in wt%. (c) Rheometer temperature sweep results of the storage modulus (G′) for the PBAE CANs. (d) Differential scanning calorimetry (DSC) thermograms of the fully cured PBAE CANs during the second heating scan. The mid-T_g values for PBAE_230, PBAE_2000, and PBAE_4000 were determined to be 4, −57, and −65 °C, respectively. (e) TGA thermograms of the PBAE CANs. The T_d5% values were 248, 281, and 301 °C for PBAE_230, PBAE_2000, and PBAE_4000, respectively. For panels (b–e), red squares, yellow circles, and blue triangles correspond to PBAE_230, PBAE_2000, and PBAE_4000, respectively.

The observation of a rubbery plateau in the storage modulus provided further evidence for the formation of the highly crosslinked CANs (Fig. 2c). Notably, the PBAE CAN prepared using a higher molecular weight of PEA exhibited a lower rubbery plateau, indicating a lower crosslink density. The decline in the storage modulus (G′) for all CAN specimens at temperatures above 120 °C is attributed to the reduction in the crosslink density due to the retro-aza-Michael reaction. The T_gs were determined to be 4 °C, −57 °C, and −65 °C for PBAE_230, PBAE_2000, and PBAE_4000, respectively (Fig. 2d). Additionally, lower T_g was observed with increasing PEA molecular weight. Thermal stability of PBAE CANs was characterized by thermogravimetric analysis (TGA) measurements (Fig. 2e), revealing that the temperature at which the mass loss is 5% (T_d5%) was 248, 281, and 301 °C for PBAE_230, 2000, and 4000, respectively. The degradation temperature showed a slight but monotonic increase with increasing PEA molecular weight. To avoid thermal degradation, all further experiments were performed at temperatures below T_d5%.

Dynamic-bond-exchange characteristics of PBAE CANs

PBAE CANs are known to undergo dynamic-bond-exchange reactions through transesterification and dynamic aza-Michael exchange reactions (Fig. 3a).^12,24 The β-amino esters^12,23–25 and β-hydroxyl groups²³ act as internal catalysts, whereas the neighboring hydroxyl groups further accelerate the transesterification through the neighboring-group-participation effect, resulting in catalyst-free transesterification.²⁴ As a result, PBAE CANs can be reprocessed into various shapes using corresponding molds through successive cycles of heat pressing at 120 °C and application of a pressure of 5 MPa, without the need for an external catalyst (Fig. 3b for PBAE_230, and Fig. S2a and b for PBAE_2000 and PBAE_4000, respectively). Notably, no significant changes in either the chemical structure or T_g were observed after two cycles of reprocessing, as shown by the ATR-FTIR and DSC results (Fig. S3) obtained after each reprocessing cycle.

Fig. 3 (a) A schematic illustrating dynamic aza-Michael exchange and transesterification in PBAE CANs. (b) Photographs demonstrating the catalyst-free reprocessability of PBAE_230 over two cycles. Heat pressing between each cycle was performed at 120 °C and 5 MPa for 1 h. (c–e) Relaxation times extracted by fitting the stress relaxation at various temperatures (Fig. S4) to the stretched Maxwell equation (eqn (2)) for PBAE_230, PBAE_2000, and PBAE_4000, respectively. The blue dashed line corresponds to the WLF fit (eqn (7)), and the red solid line corresponds to the Arrhenius fit (eqn (6)). Fitting parameters for these lines can be found in Table S2. A crossover between the WLF and Arrhenius relationships was observed at T_v (=93, 103, and 95 °C) for PBAE_230, PBAE_2000, and PBAE_4000, respectively. Relaxation times below and above the T_v are shown by squares and circles, respectively. (f) Activation energy was estimated from the Arrhenius fitting of the relationship between the relaxation time and temperature for temperatures above T_v. In the panels, red squares and solid lines, yellow circles and dashed lines, and blue triangles and dotted lines correspond to PBAE_230, PBAE_2000, and PBAE_4000, respectively. The numbers given in parentheses in the legend indicate the E_a value of the corresponding PBAE CAN.

To investigate the catalyst-free dynamic bond exchange characteristics of PBAE CANs, stress relaxation experiments were conducted at temperatures ranging from 70 to 150 °C (Fig. S4). All experiments were performed under 1% strain, which is within the linear viscoelastic region (Fig. S5). The PBAE CANs underwent rapid stress relaxation at 150 °C (the highest temperature tested), with the relaxation process becoming progressively slower as the temperature decreased to 30 °C. At all temperatures, the relaxation data were fitted using the stretched Maxwell model (eqn (2)), providing the values of the fitted relaxation time (τ) and exponent (β) for each temperature. The parameter β is related to the width of the relaxation time distribution. The use of the stretched Maxwell model is appropriate here because all functional groups are found in distinct chemical environments based on their specific network connectivity. The fits showed excellent agreement with the experimental data at all temperatures, and the τ and β values obtained by the fitting are listed in Table S1.

Fig. 3c–e show τ plotted as a function of temperature for PBAE_230, PBAE_2000, and PBAE_4000, respectively. Notably, for all PBAE CANs, an inflection point occurs between 90 and 110 °C with two distinct τ–T correlations observed above and below this temperature range. The rheological properties of CANs, including the relaxation time, are known to follow an Arrhenius-type temperature dependence at temperatures above T_v^24,31 which is the temperature at which the chain topology begins to be actively rearranged by the dynamic-bond-exchange reaction. We attribute the Arrhenius-type temperature dependence of CANs above T_v to stress relaxation at higher temperatures which is governed by the relatively slow dynamic bond exchange rather than by the fast diffusion of chains. Therefore, in this kinetically controlled region, τ should follow the Arrhenius equation given by

τ = A exp(E_a/RT) (6)

where A is the pre-exponential constant, E_a is the activation energy, and R is the ideal gas constant.

However, at temperatures below T_v, the chains exhibit some segmental motion, but this motion is restricted owing to the dormant dynamic bonds, yielding a crosslinked molecular structure without any topology rearrangement. In this region, the relaxation time follows the Williams–Landel–Ferry (WLF) equation (eqn (7)) which governs the relaxation time for classical thermosets.^{13,23,31–33} The WLF equation is given by

log(τ/τ_ref) = −C₁(T − T_ref)/C₂ + (T − T_ref) (7)

where τ is the relaxation time, τ_ref is the reference relaxation time, C₁ and C₂ are empirical constants, T is the temperature, and T_ref is the reference temperature, which is set at 90 °C for PBAE_230 and 70 °C for PBAE_2000 and PBAE_4000. Therefore, in the low-temperature region (i.e., temperatures below 80 °C), τ was fitted using the WLF equation (eqn (7)), whereas in the high-temperature region (i.e., temperatures above 110 °C), the Arrhenius equation (eqn (6)) was used for fitting. Both fits show excellent agreement in all corresponding temperature ranges (Fig. 3c–e). The corresponding fitting parameters are listed in Table S2. These fits unambiguously indicate that for all PBAE CANs, the τ–T behavior is described by using the WLF equation owing to the diffusion-controlled relaxation (or slow diffusion without dynamic bond exchange) at temperatures below T_v, transitioning to the kinetically controlled relaxation (or fast diffusion and relatively slow dynamic bond exchange) at temperatures above T_v. Thus, the crossover between the WLF and Arrhenius relationships^31,34–36 in Fig. 3c–e was defined as T_v (93, 103, and 95 °C for PBAE_230, PBAE_2000, and PBAE_4000), respectively. We note that the reprocessing temperature (or 120 °C) in Fig. 3b and S2 is well above T_v.

Notably, the values of T_v, stress relaxation time at temperatures above T_v, and the extracted E_a are similar across all PBAE CANs regardless of the molecular weight of the PEA, as shown in Fig. 3f. This similarity is attributed to the fact that at temperatures significantly above T_g, at least T_g + 89 °C, the polymer networks operate in a kinetically controlled regime dominated by dynamic bond exchange, with diffusion being negligible. Therefore, at these temperatures, PBAE CANs are expected to be governed solely by the kinetic (or dynamic bond exchange) timescale. Furthermore, all PBAE CANs share identical chemical structures and functional groups, differing only in the crosslink density (Fig. 1), resulting in similar values of E_a, τ at temperatures above T_v, and T_v regardless of the PEA molecular weights. However, at lower temperatures, differences in relaxation behavior become more pronounced—particularly in PBAE_230, which exhibits markedly slower stress relaxation compared to PBAE_2000 and PBAE_4000 (Fig. 3c–e and S4). This is attributed to its higher T_g and limited segmental mobility, which retain diffusion-related contributions and result in slower relaxation. In contrast, the latter two benefit from lower T_g values (Fig. 2d) and slightly higher sol fractions (Fig. 2b), both of which enhance chain mobility and promote exchange-controlled dynamics with less diffusion-related contributions.

Adhesive properties of PBAE CANs

Rubbery polymer electrolytes offer significant advantages over inorganic electrolytes, particularly due to their seamless adhesion to electrodes and interfaces without the presence of voids. This characteristic is critical for ensuring optimal interfacial contact and enhancing the overall battery performance under operational conditions. Therefore, to further illustrate the versatility of PBAE CANs as polymer electrolytes, their adhesive properties were examined. The lap shear test was performed to compare and analyze the dependence of the adhesion of the PBAE CANs on the PEA molecular weight. PBAE_2000 demonstrated the highest lap shear strength (1.9 MPa), whereas PBAE_230 (1.4 MPa) and PBAE_4000 (0.3 MPa) exhibited progressively lower lap shear strength (Fig. 4a). Unlike conventional thermoset polymers, PBAE CANs were recyclable and could restore their original adhesive properties through bond-exchange-induced re-adhesion achieved by the application of slight pressure and heat above the T_v (Fig. 4b). PBAE_2000, which exhibited the highest lap shear strength, was used for this test.

Fig. 4 (a) Lap shear strength of PBAE CANs. The lap shear strengths of the PBAE_230, 2000, and 4000 CANs were determined to be 1.4, 1.9, and 0.3 MPa, respectively. Error bars indicate the standard deviation of the results. (b) Representative stress–strain curves of the PBAE_2000 CAN before (PBAE_2000 virgin) and after recycling (PBAE_2000 recycled). The recycled specimen was measured by reattaching the failed surfaces at 120 °C and 0.5 MPa for 1 h. (c) Water contact angles on the cured PBAE_230, 2000, and 4000 CAN films were measured to be 23.5° ± 0.5°, 44.1° ± 0.3°, and 70.9° ± 0.3°, respectively. Errors indicate the standard deviation of the results. (d) Photographs showing (top) the uncured PBAE monomers on the SUS plate and (bottom) the cured adhesive after failure. PBAE_230 adhesive shows adhesive failure, whereas PBAE_2000 and PBAE_4000 adhesives show cohesive failure. (e) Photograph showing the PBAE_2000 adhesive bearing weights of 55.2 kg. In the panels, red, yellow, and blue correspond to PBAE_230, PBAE_2000, and PBAE_4000, respectively.

To explain the variation in the lap shear strength among the three PBAE CANs, water contact angles for the PBAE CANs were measured. The water contact angle for the cured PBAE_230 was 23.5° ± 0.5°, whereas PBAE_2000 and PBAE_4000 exhibited water contact angles of 44.1° ± 0.3° and 70.9° ± 0.3°, respectively (Fig. 4c). The high surface energy of PBAE_230 implies that the surface tension of the corresponding monomers is relatively high prior to curing, resulting in poor coatability on the SUS substrate (as shown in the leftmost image in the top row of Fig. 4d). However, PBAE_230 exhibited the highest T_g and storage modulus (Fig. 2c and d). As a result, adhesive failure was observed due to the relatively strong cohesion, as evidenced by the leftmost image in the bottom row of Fig. 4d. By contrast, PBAE_4000 exhibited the lowest surface energy (or the highest water contact angle), resulting in excellent coatability on the substrate (as shown in the rightmost image in the top row of Fig. 4d) but also the lowest cohesion, as evidenced by its lowest T_g and storage modulus (Fig. 2c and d). Consequently, cohesive failure was observed owing to the relatively strong adhesion force between the substrate and the PBAE CAN, as shown in the rightmost image in the bottom row of Fig. 4d. The PBAE CAN with an intermediate PEA molecular weight, namely PBAE_2000, exhibited cohesive failure while also displaying excellent coatability and moderate cohesion. Therefore, the strongest lap shear strength among the examined PBAE CANs was obtained for PBAE_2000. As demonstrated in Fig. 4e, PBAE_2000 could bear a weight of 55.2 kg, which is 184 000 times its own weight.

Ionic conductivity and electrochemical stability of SPEs based on PBAE CANs

Finally, the ionic conductivity, mechanical modulus, and electrochemical properties of the PBAE CAN-based SPEs were analyzed. To identify the optimal LiTFSI concentration for ion transport, a series of electrolytes with varying salt contents (30–70 wt%) were prepared and tested. As shown in Fig. S6, the ion conductivity exhibits a nonmonotonic dependence on salt concentration, reaching a maximum at 50 wt% LiTFSI. This behavior is attributed to the competing effects of increasing charge carrier density and decreasing segmental mobility due to ion pairing and coordination, consistent with previous observations in similar systems.³⁷ The corresponding [PO]/[Li⁺] ratio in this composition was calculated and is provided in Fig. S6 for reference. Fig. 5a shows the ionic conductivity of PBAE CAN-based solid polymer electrolytes (denoted as CSPE_X, where X is the molecular weight of PEA), with 50 wt% LiTFSI incorporated into each CAN matrix (PBAE_230, PBAE_2000, and PBAE_4000) at room temperature. The ionic conductivity (σ_DC) of CSPE_230 was 1.2 × 10⁻⁷ S cm⁻¹, whereas that of CSPE_4000 was 4.6 × 10⁻⁷ S cm⁻¹. By contrast, CSPE_2000 exhibited the highest ionic conductivity among the three examined electrolytes, reaching a σ_DC of 3.7 × 10⁻⁶ S cm⁻¹ at 25 °C. The oxygen atoms in the propylene oxide (PO) units of the PEA chain interact with the lithium cations, promoting the dissociation of lithium salts and enabling lithium-ion migration through ion–dipole interactions, facilitating lithium-ion transport via the ion-hopping mechanism. This mechanism is further assisted by the segmental motion of the polymer chains at temperatures above the T_g.^38,39 Coupled with the fact that PBAE_2000 exhibited superior adhesive properties compared with the other PBAE CANs (Fig. 4a), CSPE_2000, which has a higher PO content, exhibits enhanced σ_DC relative to that of CSPE_230. The increased PO content promotes more efficient dissociation of lithium cations from the lithium salt, thereby facilitating improved ion transport (Fig. 5b). However, further increasing the PO content in CSPE resulted in a reduction of σ_DC, as observed for CSPE_4000 (which has the highest PO content among the three). This indicates that although a higher [PO]/[Li⁺] ratio may initially facilitate lithium-ion hopping between chains, the enhanced ion–dipole interactions between lithium cations and the numerous oxygen atoms can ultimately reduce the lithium-ion mobility. Furthermore, the increased number of methyl groups in PO further hinders lithium-ion movement, ultimately resulting in lower σ_DC.^40,41

Fig. 5 (a) Room-temperature ion conductivity σDC of CSPE_230, CSPE_2000, and CSPE_4000. (b) Ionic conductivity as a function of the [PO]/[Li+] ratio at room temperature for CSPE_230, CSPE_2000, and CSPE_4000. (c) DSC curves of CSPE_2000 and CGPE_2000. (d) Temperature dependence of ionic conductivity (dark yellow triangles) for CGPE_2000, following Arrhenius (red dashed line) and VTF (blue dashed line) behavior at temperatures above and below the transition temperature Tt = 353 K, respectively. (e) Compression creep experiment for CGPE_2000 at various temperatures. The sample was compressed for 1 min, after which the load was removed. Squares, circles, and triangles correspond to RT, 60, and 80 °C, respectively. (f) Ionic conductivity at 298 K as a function of plasticizer content (Φplasticizer) for CSPEs and CGPE_2000, including literature data. (g) Determination of the lithium transference number for CGPE_2000 using a plot of the current versus time for the Li|CGPE_2000|Li symmetric cell during 10 mV polarization at 298 K; the inset shows the AC impedance spectra of the cell before and after polarization. (h) Molecular configurations of (left) LiTFSI, (middle) TGDA-LiTFSI, and (right) PEA-LiTFSI, where C, H, Li, O, N, S, and F are shown in dark gray, white, purple, red, blue, yellow, and light green, respectively. The binding energy Eb between Li+ and TFSI− and the intermolecular distance dLi+LiTFSI–O between Li+ and O within TFSI− (indicated by arrows) are also shown. (i) Lithium plating/stripping cycling performance of the Li|CGPE_2000|Li symmetric cell at current densities of 0.01 and 0.02 mA cm−2, with each cycle lasting 1 h, followed by a return to 0.01 mA cm−2 (dark yellow curve). After 240 h, the experiment was paused for 1 h (indicated by an arrow) and then resumed (light yellow curve). The insets show magnified views at specified times.

To further increase the σ_DC of the electrolytes, PBAE-CAN-based gel polymer electrolytes (CGPEs) were prepared by blending CSPE with TGDE used as the plasticizer. From this point onward, PBAE_2000 was used as the main matrix, as it exhibited higher ionic conductivity without the plasticizer and superior adhesive properties after CGPE preparation, as shown in Fig. 5b and S7, respectively. In CGPE_2000, the plasticizer/lithium salt mixture (TGDE and LiTFSI) was incorporated into the PBAE_2000 CAN matrix at a weight ratio of 3 : 7. Fig. 5c presents the DSC second-heating curves for CSPE_2000 without TGDE and CGPE_2000 with TGDE. As shown in Fig. 2d, the pristine PBAE_2000 exhibits a T_g of −57 °C. However, when lithium salt is incorporated into the PBAE_2000 polymer matrix to form CSPE_2000, the T_g increased to −35 °C. This increase in the T_g is attributed to the coordination of lithium cations with the polymer backbone. The ionic interactions restrict the mobility of the polymer chains, thereby increasing T_g.¹⁹T_g is a critical factor influencing σ_DC because lithium-ion hopping is facilitated by the segmental motion of the polymer chains. Consequently, the introduction of TGDE as the plasticizer into CSPE_2000 lowered T_g from −35 to −70 °C. TGDE functions as a plasticizer agent, softening the network, reducing T_g, enhancing the segmental motion of the polymer backbone, and promoting greater ion mobility at a given temperature. This leads to improved σ_DC compared with that of the unplasticized system.¹⁰

In addition to σ_DC and T_g, the values of the storage modulus G′ of PBAE_2000 and CGPE_2000 and other reported SPEs are presented in Fig. S8. Although the addition of the plasticizer led to a slight reduction in the mechanical properties at room temperature, CGPE_2000 still maintains a modulus comparable to or slightly higher than that of the conventional SPEs.^42,43 The temperature dependence of ionic conductivity σ_DC(T) for CGPE_2000 was further investigated using electrochemical impedance spectroscopy (EIS). Fig. 5d shows the temperature-dependent ionic conductivity, measured at intervals of 10 °C in the temperature range from 20 to 100 °C. With the addition of the plasticizer, the synthesized CGPE_2000 exhibited an order of magnitude higher ionic conductivity (σ_DC = 4.3 × 10⁻⁵ S cm⁻¹ at 25 °C) compared with the unplasticized CSPE_2000 (σ_DC = 3.7 × 10⁻⁶ S cm⁻¹ at 25 °C). Interestingly, σ_DC(T) of CGPE_2000 exhibits a change in the slope at the conductivity transition temperature T_t = 80–90 °C, shifting from the Vogel–Tamman–Fulcher (VTF) behavior (eqn (8)) at lower temperatures from 20 to 80 °C, to the Arrhenius behavior (eqn (9)) at higher temperatures above 80 °C.

σ_DC = A/T^1/2 exp[−E^VTF_a/R(T − T₀)] (8)

σ_DC = σ₀ exp[−E^Arr_a/RT] (9)

where A and σ₀ are pre-exponential constants at T → ∞, E^VTF_a and E^Arr_a are the VTF and Arrhenius activation energies for ion transport, respectively, R is the ideal gas constant, and T₀ is the Vogel temperature, and the obtained values of these fitting parameters are listed in Table S3. The T_v of CGPE_2000 (Fig. S9a) closely matches its T_t (Fig. 5d), and both temperatures are notably lower than the T_v of PBAE_2000 (Fig. 3d) and CGPE_2000 without LiTFSI (Fig. S9b). This suggests that LiTFSI (or more specifically, the dissociated lithium ion) may facilitate dynamic bond exchange by acting as a Lewis acid.^44–46 The correspondence between T_v and T_t for CGPE_2000 further indicates that the CAN properties are well preserved in this system, and that the activation of dynamic bond exchange is directly linked to the onset of increased ionic conductivity. Interestingly, the T_v and relaxation times above T_v for PBAE_2000 (Fig. 3d) and CGPE_2000 without LiTFSI (Fig. S9b) were nearly identical, indicating that the plasticizer alone does not significantly influence the exchange dynamics. This suggests that, while TGDE may enhance chain mobility, it does not notably affect the bond exchange characteristics. This observation reinforces the earlier conclusion that, in the absence of compounds influencing dynamic bond exchange, the system operates in a kinetically governed regime.

Fig. 5e demonstrates the compression creep recovery of CGPE_2000. Below T_v and T_t (or from RT to 80 °C), deformation induced by compression was fully recoverable upon release of the compressive load. This indicates that recovery can be achieved regardless of the deformation induced by the volumetric expansion and contraction of the electrodes during charging and discharging, as further confirmed by the lithium plating and stripping tests described below. This ensures a conformal contact between the SPE and the electrode, which is crucial for inhibiting lithium dendrite formation.¹⁵ Notably, our CGPE_2000 with 23.5 wt% plasticizer exhibited a higher σ_DC than other reported GPEs with Φ_plasticizer >50 wt% (Fig. 5f).^47–52

Fig. 5g shows the lithium transference number (tLi+) of the synthesized CGPE_2000. tLi+ is a crucial parameter because only mobile lithium ions contribute to the battery's performance during the charge and discharge cycles. Although both lithium cations and TFSI anions contribute to the overall ionic conductivity in CGPE_2000, the observed tLi+ is 0.57, which is higher than that of the conventional PEO-based polymer electrolytes (tLi+ = 0.2–0.5).53,54 In PEO systems, the mobility of the lithium cation is reduced compared with that of its anionic counterpart owing to the coupling of the cation with the Lewis-basic regions of the polymer matrix.55 However, for CGPE_2000, the lithium cation mobility appears to be higher than that of TFSI−. This behavior originates from the coordination of Li+ with the carbonyl groups in TGDA, whereas the polar hydroxyl groups provide additional stabilization through ion–dipole interactions. Furthermore, the oxygen atoms in the PO groups of PEA effectively decrease the Li+–TFSI− binding energy via similar ion–dipole interactions, enhancing ionic dissociation. This results in the formation of a dynamic, percolating network that facilitates lithium-ion transport via the ion-hopping mechanism.28 These combined effects resulted in a higher tLi+ compared with that of the conventional dual-ion systems. To further investigate the ion transport mechanism in our CAN-based electrolytes, DFT calculations were also performed to estimate the binding energy and the distance between Li+ and TFSI−. The optimized structure is illustrated in Fig. 5h, and the binding energy of the optimized LiTFSI was calculated to be −623.66 kJ mol−1, with the distance between the Li+ cation and the oxygen atom of the TFSI− anion (dLi+LiTFSI–O) found to be 1.8 Å. However, when LiTFSI is combined with TGDA, the binding energy decreases to −382.04 kJ mol−1, and the distance between Li+ and the oxygen of the anion (dLi+LiTFSI–O) increases from 1.8 to 1.96 Å, indicating a weaker interaction between the lithium ion and the anion compared with the case of LiTFSI alone. Additionally, when LiTFSI is combined with the PEA chain, the binding energy decreases to −501.81 kJ mol−1, and the Li+–O distance (dLi+LiTFSI–O) increases from 1.8 Å of the ion-paired state of LiTFSI to 1.86 Å. These results demonstrate that the PBAE_2000 polymer matrix effectively weakens the interaction between Li+ and TFSI−, facilitating greater ion dissociation and mobility.

Fig. S10 shows the linear sweep voltammetry data. The experiment was conducted by assembling a Li|CGPE_2000|stainless steel (SS) cell and conducting measurements from 0 to 6 V at a scan rate of 0.5 mV s⁻¹. CGPE_2000 demonstrates oxidative stability without side reactions up to 4.5 V, which is defined as the point where the current density reaches 10 μA cm⁻². To evaluate the long-term cycle stability of our CAN-based electrolyte, the performance of the Li|CGPE_2000|Li symmetric cell was tested at room temperature, as shown in Fig. 5i. The symmetric cell demonstrated stable lithium plating/stripping behavior, with the overvoltages of 45 and 110 mV at the current densities of 0.01 and 0.02 mA cm⁻², respectively, with each cycle lasting 1 h. When the current density was reduced back to 0.01 mA cm⁻², the overvoltage recovered to 45 mV and remained stable for an additional 800 h. After approximately 240 h of continuous cycling, the experiment was paused for 1 h and then resumed at 0.01 mA cm⁻². Remarkably, the cell continued to operate stably, maintaining a low overvoltage of 45 mV for another 760 h. Overall, the Li symmetric cell exhibited no signs of either short-circuiting or overvoltage increase for over 1000 h. Additionally, we conducted Li plating/stripping tests at higher current densities of 0.05 and 0.1 mA cm⁻². Fig. S11a presents the results obtained at current densities of 0.01, 0.025, and 0.05 mA cm⁻² (mA h cm⁻²). As in the previous tests, the experiment was paused at 150 h and resumed after one day. CGPE_2000 maintained stable voltage profiles for a total of 500 h without any sign of short-circuiting. Furthermore, we extended the current density to 0.1 mA cm⁻², with a corresponding areal capacity of 0.25 mA h cm⁻². CGPE_2000 demonstrated robust cycling performance, showing stable operation for up to 240 cycles (120 h) even under this elevated condition (Fig. S11b). These results clearly indicate the excellent electrochemical stability of CGPE_2000 under higher operating conditions. Lastly, when the weight ratio of the polymer matrix to liquid electrolyte was adjusted to 5 : 5 (CGPE_2000 (5 : 5)), the symmetric cell showed even more stable voltage profiles at 0.1 mA cm⁻² (0.1 mA h cm⁻²), maintaining stable operation for up to 700 h without short-circuiting (Fig. S11c). To evaluate electrochemical performance in a full-cell configuration, rate capability tests were conducted using a LFP|CGPE_2000|Li full cell at 0.1C (2 cycles), 0.2C, 0.5C, 1C, and 2C (5 cycles each), as shown in Fig. S12 and S13. CGPE_2000 exhibited discharge capacities of 107 mA h g⁻¹ at 1C and 67 mA h g⁻¹ at 2C. Notably, when the rate was returned to 0.1C, the cell recovered a capacity comparable to that in the initial cycles, demonstrating excellent capacity retention. The remarkable electrochemical stability observed in both Li symmetric and full cells is attributed to the excellent adhesion (Fig. 4a), mechanical resilience (Fig. 5e), and high t_Li⁺ (Fig. 5g) of the SPE, supporting the superior performance of the PBAE-based polymer system. These results demonstrate the stability and suitability of CGPE_2000 for practical applications in lithium-metal batteries, effectively simulating real-life conditions of repeated charging and resting cycles.

Reprocessability of SPEs based on PBAE CANs

Electrolytes based on PBAE CANs possess a unique ability to undergo dynamic-bond-exchange reactions at temperatures above T_v. This property facilitates the reprocessability of the electrolyte while maintaining its functional network. Leveraging this advantage, we successfully reprocessed the utilized CGPE_2000 and conducted a comprehensive analysis of its electrochemical/mechanical performance characteristics, such as ionic conductivity, shear modulus, and long-term cycling, compared with those of the virgin electrolyte. Fig. 6a shows the photographs of the fractured electrolyte (left) and the reprocessed CGPE_2000 (right). The used electrolyte was initially cut into fragments and then subjected to reprocessing under mild conditions at 120 °C for 1 h. The right-hand image confirms the successful reformation of the electrolyte into a cohesive structure after reprocessing. The σ_DC of the reprocessed CGPE_2000 was systematically evaluated across a temperature range of 20–100 °C (in 10 °C increments) and compared with that of virgin CGPE_2000 (Fig. 6b). Remarkably, at all tested temperatures, the reprocessed electrolyte showed ionic conductivity values that were nearly identical to those of the virgin sample (σ^virgin_DC = 4.3 × 10⁻⁵ S cm⁻¹ at 25 °C and σ^reprocessed_DC = 4.0 × 10⁻⁵ S cm⁻¹ at 25 °C). The T_t, which marks the shift from VTF to Arrhenius behavior, also remained unchanged and was equal to 80 °C for both samples. This consistency in T_t confirms that the dynamic polymer network topology was fully preserved during the reprocessing. Further validation was achieved through Arrhenius and VTF fitting analyses. The E_a values obtained from both models showed negligible differences between the virgin and reprocessed samples (E^Arr_a = 62 kJ mol⁻¹ and E^VTF_a = 9.5 kJ mol⁻¹versus E^Arr_a = 67 kJ mol⁻¹ and E^VTF_a = 9.0 kJ mol⁻¹). The obtained fitting parameters are listed in Table S4. These results underscore that the reprocessing protocol did not alter the energy barriers for ion hopping and polymer chain dynamics, which are governed by the CAN's dynamic-bond-exchange characteristics. Additionally, as demonstrated in Fig. S14, the reprocessed CGPE_2000 exhibited a shear modulus of 2.8 × 10⁴ Pa at 2.5 rad s⁻¹ at room temperature, which is nearly identical to that of virgin CGPE_2000. Such retention of mechanical properties underscores the robustness of the reprocessing protocol, confirming that the structural integrity and viscoelastic behavior of the electrolyte remain uncompromised even after reprocessing/recycling.

Fig. 6 (a) Photographs of the fractured CGPE_2000 sample (left) and the reprocessed CGPE_2000 (right). (b) Temperature dependence of ionic conductivity σ_DC(T) for the original CGPE_2000 (dark yellow triangles) and reprocessed CGPE_2000 (yellow squares). Both samples exhibit a similar transition temperature (T_t), where the σ_DC(T) behavior shifts from Arrhenius to VTF, with comparable activation energies obtained from the corresponding fits. (c) Lithium plating/stripping cycling performance of the Li|Li symmetric cell using virgin and reprocessed CGPE_2000 at a current density of 0.01 mA cm⁻² (d) TGA thermograms of CSPE_2000, LiTFSI extracted CSPE_2000 with THF washing, and LiTFSI. (e) ATR-FTIR spectra of virgin and extracted LiTFSI. All ATR-FTIR spectra were normalized by using the area under the invariant S=O peak appearing near 1200 cm⁻¹. (f) ¹³C-NMR spectra of virgin LiTFSI and extracted LiTFSI.

To rigorously assess the electrochemical recyclability and interfacial stability of the CAN-based electrolyte, Li plating/stripping tests were performed using Li|reprocessed CGPE_2000|Li symmetric cells at 25 °C for 500 h at a constant current density of 0.01 mA cm⁻². Following the cycling tests, the cells were disassembled, and the used CGPE_2000 electrolyte was carefully extracted. The recovered electrolyte underwent reprocessing via the thermal-pressure protocol described in Fig. 6a (120 °C, 1 h, and mild pressure), which effectively restored its structural integrity. The reprocessed electrolyte was then reassembled into a fresh symmetric cell and subjected to identical cycling conditions. Fig. 6c compares the voltage hysteresis profiles of the virgin CGPE_2000 (dark yellow curve) and the reprocessed CGPE_2000 (light yellow curve) over 500 h. Both electrolytes exhibited nearly overlapping voltage profiles, with stable polarization voltages and no signs of sudden voltage fluctuations or short-circuiting. This indicates that the reprocessed electrolyte maintains excellent interfacial compatibility with lithium-metal electrodes, even after repeated cycling and reprocessing.

LiTFSI extraction of PBAE CAN-based electrolytes

Recently, the growing demand for batteries has led to increased generation of battery waste. The disposal of used solid electrolytes presents significant environmental and economic challenges, highlighting the importance of recovering and recycling both lithium resources and polymer electrolytes. As mentioned earlier, conventional crosslinked SPEs make recycling particularly difficult. In Fig. S15 and S16, we demonstrate an eco-friendly method to depolymerize PBAE CANs in water. Consequently, hydrolyzing PBAE-based SPEs (CSPE/CGPE) in water, and then separating the LiTFSI salt from the aqueous mixture of hydrolyzed PBAE CANs and salt may be possible.⁵⁶ As an alternative method, we also demonstrated an efficient process for LiTFSI extraction from the PBAE-CAN-based SPEs. First, CSPE_2000 was swollen in THF and then vacuum-filtered to separate the LiTFSI dissolved in THF from the PBAE CAN matrix. The filtrate contained LiTFSI along with a trace amount of soluble PBAE components. As shown in the TGA thermogram (Fig. 6d), CSPE_2000 exhibited two-step thermal degradation at 277 and 389 °C, corresponding to the degradation of PBAE_2000 and LiTFSI salt, respectively. However, after the LiTFSI salt was removed from CSPE_2000 using the above-described method, only thermal degradation of the CAN matrix was observed, indicating the complete extraction of the LiTFSI salt from the CSPE_2000 matrix.

To further isolate the LiTFSI salt, the filtrate was heated at 330 °C in ambient air using an oven. LiTFSI did not thermally degrade up to 350 °C, whereas CSPE_2000 began degrading at 284 °C (Fig. 6d). Therefore, heating the filtrate at 330 °C yielded mostly LiTFSI salt with some carbonaceous residues. The resultant sample was thoroughly washed several times with deionized water and then filtered. The final clear aqueous filtrate contained only LiTFSI salt dissolved in water. By evaporating the water, pure LiTFSI was successfully recovered. The purity of the recovered LiTFSI was analyzed using ATR-FTIR and ¹³C-NMR (Fig. 6e and f), and all characteristic ATR-FTIR and ¹³C-NMR peaks for the virgin LiTFSI salt were observed with no changes in the recovered LiTFSI. This extraction method demonstrates the feasibility of effective separation of the LiTFSI salt from the PBAE-CAN-based SPEs and highlights the potential for the reuse of high-purity lithium salts. Additionally, the PBAE CAN matrix can be recovered by simply evaporating THF from the first retentate, enabling its reuse through thermal reprocessing.

Conclusion

To address the safety and sustainability challenges of the current lithium-ion battery systems, PBAE CANs were prepared as fully recyclable, catalyst-free, highly adhesive, and resilient SPEs. To investigate the effects of the crosslink density, T_g, and T_v on the ionic conductivity, adhesion, and CAN properties, PEA with varying molecular weights (230, 2000, and 4000 g mol⁻¹) was used for the PBAE CAN preparation. The presence of β-amino esters and β-hydroxyl groups in the PBAE enabled dynamic exchange under milder conditions, even without the use of external catalysts. A thorough investigation revealed that the PBAE CANs exhibited similar values of T_v, τ at temperatures above T_v, and E_a regardless of the PEA molecular weight (or crosslink density), indicating that these CANs were in the kinetically controlled regime with negligible diffusion timescales at temperatures near and/or above T_v. This result highlights that PBAE CANs with a low T_g ensure a soft and flexible matrix while maintaining mechanical integrity through dormant dynamic bond exchanges at operating temperatures (or below T_v), including room temperature. The PBAE CAN, exhibiting excellent resilience below T_v and T_g below room temperature, further facilitated lithium-ion transport and improved adhesion to electrode surfaces, thereby improving long-term stability and enabling stable rate performance in both Li symmetric and full cells. In addition to the PO-like backbone structure in the PBAE CANs, the presence of additional hydroxyl groups had a positive effect on lithium-ion conductivity. To further highlight the sustainability of the PBAE CANs, the SPE was successfully reprocessed using heat and pressure, allowing for its smooth reuse. Furthermore, eco-friendly depolymerization and lithium salt recovery from the used PBAE CAN-based SPE were also demonstrated.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data that support this study are available from the corresponding author upon reasonable request.

The SI includes ATR-FTIR spectra of PBAE CANs; photographs demonstrating catalyst-free reprocessability; ATR-FTIR spectra and DSC thermograms of PBAE CANs before and after two reprocessing cycles; stress relaxation results and rheometer strain sweep data of PBAE CANs; fitted τ and β parameters from the stretched Maxwell equation, along with WLF and Arrhenius fitting parameters of PBAE CANs; room-temperature ionic conductivity of CSPE_2000 as a function of LiTFSI content; lap shear strength of CGPEs; comparison of storage modulus between neat CAN (PBAE_2000), its plasticized electrolyte (CGPE_2000), and conventional solid polymer electrolytes; VTF and Arrhenius fitting parameters of CGPE_2000; relaxation times at various temperatures for CGPE_2000 and CGPE_2000 without LiTFSI; linear sweep voltammetry of the Li|CGPE_2000|SS cell; lithium plating/stripping cycling performance of the Li|CGPE_2000|Li symmetric cell; rate capabilities and charge–discharge voltage profiles of LFP‖Li full cells with CGPE_2000; VTF and Arrhenius fitting parameters of reprocessed CGPE_2000; comparison of shear modulus and loss modulus between virgin and reprocessed CGPE_2000; photographs showing hydrolysis of PBAE CANs in heated water; and gel fraction of PBAE CANs as a function of soaking time in water. See DOI: https://doi.org/10.1039/d5ta03293h.

Acknowledgements

This study was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean government (MSIT) (RS-2024-00346133, RS-2024-00409589, and NRF-2022R1F1A1069348).

References

1. D. K. Maurya, B. Bazri, P. Srivastava, J.-Y. Huang, Y.-T. Hung, W.-T. Huang, D.-H. Wei and R.-S. Liu, Adv. Energy Mater., 2024, 14, 2402402.
2. Q. Wang, Z. Yao, C. Zhao, T. Verhallen, D. P. Tabor, M. Liu, F. Ooms, F. Kang, A. Aspuru-Guzik, Y.-S. Hu, M. Wagemaker and B. Li, Nat. Commun., 2020, 11, 4188.
3. H. Wu, Y. Xu, X. Ren, B. Liu, M. H. Engelhard, M. S. Ding, P. Z. El-Khoury, L. Zhang, Q. Li, K. Xu, C. Wang, J.-G. Zhang and W. Xu, Adv. Energy Mater., 2019, 9, 1902108.
4. H. Li, Y. Du, Q. Zhang, Y. Zhao and F. Lian, Adv. Energy Mater., 2022, 12, 2103530.
5. P. Ding, S. Zhao, Z. Lin, L. Wu, L. Xu, S. Liu, X. Guo, M. Tang and H. Yu, Nano Energy, 2024, 129, 110006.
6. Z. Song, F. Chen, M. Martinez-Ibañez, W. Feng, M. Forsyth, Z. Zhou, M. Armand and H. Zhang, Nat. Commun., 2023, 14, 4884.
7. R. Khurana, J. L. Schaefer, L. A. Archer and G. W. Coates, J. Am. Chem. Soc., 2014, 136, 7395–7402.
8. J. Hou and G. L. Baker, Chem. Mater., 1998, 10, 3311–3318.
9. J. Li, Y. Cai, H. Wu, Z. Yu, X. Yan, Q. Zhang, T. Z. Gao, K. Liu, X. Jia and Z. Bao, Adv. Energy Mater., 2021, 11, 2003239.
10. L. Yin, P. Zhang, J. Yang, J. Meng, M. Wu and X. Pu, Angew. Chem., Int. Ed., 2024, 63, 202404769.
11. Y.-N. Zhou, H. Yong, R. Guo, K. Wang, Z. Li, W. Hua and D. Zhou, Angew. Chem., Int. Ed., 2024, 63, 202319003.
12. C. Taplan, M. Guerre and F. E. Du Prez, J. Am. Chem. Soc., 2021, 143, 9140–9150.
13. D. Montarnal, M. Capelot, F. Tournilhac and L. Leibler, Science, 2011, 334, 965–968.
14. C. N. Bowman and C. J. Kloxin, Angew. Chem., Int. Ed., 2012, 51, 4272–4274.
15. X. Wang, M. Wang, H. Chen, Y. Zhang, B. Niu, L. Tian and D. Long, Chem. Eng. J., 2024, 496, 153830.
16. B. B. Jing and C. M. Evans, J. Am. Chem. Soc., 2019, 141, 18932–18937.
17. S. Yang, S. Park, S. Kim and S.-K. Kim, Mater. Today Energy, 2024, 45, 101690.
18. Z. Amirsedghi, S.-A. Safavi-Mirmahalleh, H. Roghani-Mamaqani and M. Salami-Kalajahi, J. Energy Storage, 2024, 102, 114175.
19. J. Wang, C. Zhang, W. Ye, X. Zhou and Z. Xue, ACS Appl. Polym. Mater., 2024, 6, 2041–2048.
20. J. Lopez, Y. Sun, D. G. Mackanic, M. Lee, A. M. Foudeh, M.-S. Song, Y. Cui and Z. Bao, Adv. Mater., 2018, 30, 1804142.
21. A. Genest, D. Portinha, E. Fleury and F. Ganachaud, Prog. Polym. Sci., 2017, 72, 61–110.
22. M. Chen, L. Zhou, Y. Wu, X. Zhao and Y. Zhang, ACS Macro Lett., 2019, 8, 255–260.
23. G. Lee, H. Y. Song, S. Choi, C. B. Kim, K. Hyun and S.-k. Ahn, Macromolecules, 2022, 55, 10366–10376.
24. D. Jeon, Y. Yoon, D. Kim, G. Lee, S.-k. Ahn, D. Choi and C. B. Kim, Macromolecules, 2023, 56, 697–706.
25. R. Miao, Y. Ding, J. Liu, J. Liu, Z. Xin and C. Bao, ACS Appl. Polym. Mater., 2024, 6, 13439–13448.
26. K. W. Lim, J.-I. Lee, J. Yang, Y.-K. Kim, H. Y. Jeong, S. Park and H. S. Shin, ACS Appl. Mater. Interfaces, 2014, 6, 6340–6345.
27. B. Scrosati, J. Hassoun and Y.-K. Sun, Energy Environ. Sci., 2011, 4, 3287–3295.
28. D. Yu, D. Troya, A. G. Korovich, J. E. Bostwick, R. H. Colby and L. A. Madsen, ACS Energy Lett., 2023, 8, 1944–1951.
29. L.-Y. Lin and C.-C. Chen, J. Power Sources, 2024, 603, 234236.
30. P. Zhou, X. Zhang, Y. Xiang and K. Liu, Nano Res., 2022, 16, 8055–8071.
31. J. Park, H. Y. Song, S. Choi, S.-k. Ahn, K. Hyun and C. B. Kim, J. Mater. Chem. A, 2022, 10, 6475–6480.
32. Y. He, Thermochim. Acta, 2005, 439, 127–134.
33. A. Shundo, M. Aoki, S. Yamamoto and K. Tanaka, Macromolecules, 2021, 54, 5950–5956.
34. A. Perego and F. Khabaz, Macromolecules, 2020, 53, 8406–8416.
35. Y. Yang, S. Zhang, X. Zhang, L. Gao, Y. Wei and Y. Ji, Nat. Commun., 2019, 10, 3165.
36. Y. Yoon, H. Moon, W. Cho, D. Lee, S. Jeong, J. J. Wie and C. B. Kim, Adv. Mater., 2025, 2503161,  DOI:10.1002/adma.202503161.
37. L. Herbers, V. Küpers, M. Winter and P. Bieker, RSC Adv., 2023, 13, 17947–17958.
38. H. Yang and N. Wu, Energy Sci. Eng., 2022, 10, 1643–1671.
39. C. Wang, B. B. Xu, X. Zhang, W. Sun, J. Chen, H. Pan, M. Yan and Y. Jiang, Small, 2022, 18, 2107064.
40. P. B. Sassmann and O. Weichold, Ionics, 2021, 27, 3857–3867.
41. W. H. Meyer, Adv. Mater., 1998, 10, 439–448.
42. X. Shan, M. Morey, Z. Li, S. Zhao, S. Song, Z. Xiao, H. Feng, S. Gao, G. Li, A. P. Sokolov, E. Ryan, K. Xu, M. Tian, Y. He, H. Yang and P.-F. Cao, ACS Energy Lett., 2022, 7, 4342–4351.
43. P.-F. Cao, B. Li, G. Yang, S. Zhao, J. Townsend, K. Xing, Z. Qiang, K. D. Vogiatzis, A. P. Sokolov, J. Nanda and T. Saito, Macromolecules, 2020, 53, 3591–3601.
44. D. Zheng, D. Qu, X.-Q. Yang, H.-S. Lee and D. Qu, ACS Appl. Mater. Interfaces, 2015, 7, 19923–19929.
45. M. Liu, X. Yang, Q. Sun, T. Wang, R. Pei, X. Yang, Y. Zhao, L. Zhao, G. Frenking and X. Wang, Angew. Chem., Int. Ed., 2023, 62, e202300068.
46. J. L. Self, N. D. Dolinski, M. S. Zayas, J. Read de Alaniz and C. M. Bates, ACS Macro Lett., 2018, 7, 817–821.
47. S. Liang, Q. Chen, U. H. Choi, J. Bartels, N. Bao, J. Runt and R. H. Colby, J. Mater. Chem. A, 2015, 3, 21269–21276.
48. U. H. Choi, S. Liang, Q. Chen, J. Runt and R. H. Colby, ACS Appl. Mater. Interfaces, 2016, 8, 3215–3225.
49. M. A. A. Sunita, B. Sundaresan and S. Jeya, J. Non-Cryst. Solids, 2024, 646, 123243.
50. Z. Xiong, Z. Wang, W. Zhou, Q. Liu, J.-F. Wu, T.-H. Liu, C. Xu and J. Liu, Energy Storage Mater., 2023, 57, 171–179.
51. H. Cheng, C. Yan, R. Orenstein, M. Dirican, S. Wei, N. Subjalearndee and X. Zhang, Adv. Fiber Mater., 2022, 4, 532–546.
52. Z. Wang, L. Shen, S. Deng, P. Cui and X. Yao, Adv. Mater., 2021, 33, 2100353.
53. Z. Xue, D. He and X. Xie, J. Mater. Chem. A, 2015, 3, 19218–19253.
54. Y. Zhai, W. Hou, M. Tao, Z. Wang, Z. Chen, Z. Zeng, X. Liang, P. Paoprasert, Y. Yang, N. Hu and S. Song, Adv. Mater., 2022, 34, 2205560.
55. Y. Zhao, L. Wang, Y. Zhou, Z. Liang, N. Tavajohi, B. Li and T. Li, Adv. Sci., 2021, 8, 2003675.
56. H. Du, Y. Kang, C. Li, Y. Zhao, Y. Tian, J. Lu, Z. Chen, N. Gao, Z. Li, J. Wozny, T. Li, L. Wang, N. Tavajohi, F. Kang and B. Li, Carbon Neutralization, 2023, 2, 416–424.

---

Footnote

† These authors contributed equally to this work.

---