Chemically recyclable and room-temperature self-healable siloxane-imine benzoxazine vitrimers for soft magnetic actuation

Abstract

Designing multifunctional vitrimers that combine ambient self-healing, magnetic responsiveness, reprocessability and high thermal stability is gaining importance for sustainable soft materials. Herein, we report a siloxane-imine-bridged benzoxazine-crosslinked vitrimer prepared from a bio-based vanillin-furfurylamine benzoxazine (Vfa) and amine-functional polysiloxane, yielding a low-viscosity imine-functional prepolymer with intact benzoxazine rings. Thermal ring-opening polymerization (ROP) affords a soft vitrimer network with high thermal robustness (T_max ∼ 550–560 °C). Stress-relaxation measurements reveal rapid exchange with an E_a of 23 kJ mol⁻¹ and an extrapolated topology-freezing temperature (T_v) well below T_g, consistent with fast network rearrangement. Incorporation of MNPs (20 wt%) reduces the E_a to 11.4 kJ mol⁻¹ and introduces field-addressable magneto-responsive behaviour, including reversible storage-modulus switching and shear-thinning magnetorheology. The networks show hydrophobic surfaces (water contact angle 116–130°), high creep recovery (∼93%), hydrolytic stability (≥8 days), and solvent-selective dissolution enabling recasting and reshaping. Macroscopic scratch healing occurs at room temperature within 8 h for the magnetic vitrimer while it required 24 h for the neat vitrimer, with tensile testing confirming repeatable self-healing over three cycles with average healing efficiency of 92% and 88%, respectively. Finally, reinforcement with recycled carbon-fibre felt enhances mechanical strength while preserving vitrimer-based reprocessability, enabling a sustainable circular route for soft magnetic actuators and chemically recyclable composites.

---

Introduction

Development of sustainable polymeric materials that simultaneously deliver durability, repairability, and reconfigurability is gaining pivotal focus in emerging fields such as soft robotics, flexible electronics, and adaptive structures. While intrinsic self-healing networks have been proposed to mitigate damage accumulation and extend service lifetimes, most systems still suffer from trade-offs between mechanical robustness, healing efficiency, thermal stability and processing conditions.^1–6 In particular, achieving room-temperature healing in mechanically stable networks without external catalysts or complex stimuli remains largely unresolved.

Thermoset polymers continue to dominate applications requiring high stiffness, thermal stability, and chemical resistance; however, their permanently crosslinked architectures fundamentally limit repairability and recyclability. Polybenzoxazines (PBZs) are especially attractive due to their excellent thermo-mechanical performance, low shrinkage, and catalyst-free curing via thermally induced ring-opening polymerization (ROP) of benzoxazine monomers.^7,8 Additionally, the growing use of bio-derived phenols and amines has improved the sustainability profile of polymers especially PBZs.^9–14 Nevertheless, strategies aimed at overcoming the intrinsic chemical irreversibility of PBZ networks are rapidly emerging and hold strong promise for enabling truly circular material lifecycles.^15–17

Covalent adaptable networks (CANs), and particularly vitrimers, provide a conceptual route to overcome these limitations by enabling network rearrangement through dynamic covalent bond exchange while exhibiting dimensional stability.^18–22 Vitrimers often rely on catalyst-mediated reactions, elevated activation temperatures, or specific chemistries^23–25 that may be incompatible with soft, energy-efficient, or device-integrated applications. Current demands are for vitrimer designs that combine catalyst-free synthesis,²⁶ low-temperature dynamics, long-term mechanical integrity and improved thermal stability.

Dynamic imine chemistry has emerged as a promising candidate to address these challenges, owing to its exchange mechanisms (transimination and imine metathesis) that can proceed under relatively mild and catalyst-free conditions.^27–33 Recent studies have demonstrated imine-crosslinked PBZ vitrimers with reprocessability, welding, and shape-memory behavior.^34–37 However, these systems are typically optimized for structural performance or thermal reprocessing, rather than for soft, stimuli-responsive applications. Moreover, the integration of imine dynamics into compliant, low glass transition temperature (T_g) matrices capable of room-temperature healing and functional actuation remains underexplored.

Simultaneously, the design of stimuli-responsive soft materials capable of remote and programmable actuation particularly via magnetic fields, has attracted significant attention. While polysiloxane-imine chemistry offers desirable attributes such as flexibility, low T_g, and excellent environmental stability,^38–42 its integration into dynamic covalent networks that balance mechanical compliance with functional responsiveness is still limited. Furthermore, although magnetic nanoparticle incorporation (e.g., Fe₃O₄) enables magnetorheological behavior and actuation,^43–49 the interplay between particle–matrix interactions and dynamic bond exchange remains insufficiently understood, particularly in imine-based vitrimers.

From a sustainability perspective, carbon fibre (CF)-reinforced vitrimer matrices have been proposed as an alternative, achieving simultaneous reinforcement, reprocessability, and fibre recoverability without compromising dynamic functionality and are a growing area of interest.^50–54

Here, we hypothesize that integrating dynamic imine exchange within a siloxane-modified benzoxazine vitrimer network can simultaneously enable (i) room-temperature self-healing, (ii) catalyst-free reprocessability, (iii) magnetic-field-driven actuation, with good mechanical integrity and (iv) very high thermal stability. To test this hypothesis, a bio-derived vanillin-furfurylamine benzoxazine (Vfa) is combined with poly[dimethylsiloxane-co-(3-aminopropyl)methylsiloxane] to construct an imine-functional, benzoxazine-reactive network. Subsequent ROP establishes a thermally robust yet flexible matrix, wherein imine exchange governs stress relaxation, healing, and solvent-assisted reprocessing. To impart functionality, Fe₃O₄ nanoparticles (MNPs) are incorporated to enable magnetorheological response and remote actuation, while recycled CF (rCF) felt is introduced to enhance mechanical performance and demonstrate compatibility with closed-loop composite design (Fig. 1). This work establishes a multifunctional design framework that bridges high-performance PBZ chemistry with soft, and dynamic vitrimer chemistry, advancing the development of sustainable and reconfigurable soft actuators.

Fig. 1 Preparation of (a) imine-functionalized siloxane vitrimer [poly(Si_P0-Vfa)] with/without MNPs and the respective composite. Schematic representation of the room-temperature synthesis highlighting (b) dynamic imine exchange reactions. Digital images of the respective cured samples, recycled carbon fibre-reinforced dog bone shaped composites, and magneto-responsive actuation.

Experimental

Materials

Poly[dimethylsiloxane-co-(3-aminopropyl)methylsiloxane], eq wt. 4400 amine (abbreviated as Si_P0), vanillin (abbreviated as V, ReagentPlus® 99%), furfurylamine (abbreviated as fa, 99%) and paraformaldehyde were purchased from Sigma-Aldrich. Methanol (LR grade, 99%), chloroform (LR grade, 99%), tetrahydrofuran (LR grade, 99%), and ethyl acetate (99%) were purchased from Fisher Scientific. Ethanol, sodium sulfate (Na₂SO₄), sodium hydroxide (NaOH) and hydrochloric acid (HCl) were bought from Chemlabs. Ferric chloride hexahydrate (FeCl₃·6H₂O), ferrous chloride tetrahydrate (FeCl₂·4H₂O), and ammonium hydroxide (NH₄OH, 28%) were bought from Alfa Aesar and chloroform (CDCl₃ 99.8%) was from Eurisotop. All the reagents were used as received, without further purification.

Solventless synthesis of the vanillin-furfuryl amine benzoxazine (Vfa) monomer⁵⁵. Vanillin (V, 0.03 mol, 5.00 g) was added to a pre-stirred mixture of furfurylamine (fa, 0.03 mol, 2.7 mL) and paraformaldehyde (0.06 mol, 1.82 g) at ambient temperature. The mixture was stirred for additional 15 min before transferring to a preheated oil bath at 80 °C and stirred for 4 h. Upon completion of the reaction, the mixture was allowed to cool to room temperature, and then ethyl acetate (100 mL) and aqueous sodium hydroxide (1 N, 50 mL) were added. The organic phase was separated and successively washed with distilled water (3 × 50 mL). The organic layer was dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure. The resulting crude solid was recrystallized from ethanol to yield Vfa as a white compound (yield 85%). ¹H NMR (400 MHz, CDCl₃, δ ppm): 3.93 (s, 2H, N–CH₂–fa), 3.95 (s, 3H, –OCH₃), 4.08 (s, 2H, N–CH₂–Ar), 5.08 (s, 2H, O–CH₂–N), 6.25 (d, 1H, =CH–CH=), 6.34 (t, 1H, =CH–CH=), 7.14 (s, 1H, Ar–H), 7.30 (s, 1H, Ar–H), 7.41 (d, 1H, CH=CH–O), 9.80 (s, 1H, –CHO); ¹³C-NMR (100 MHz, CDCl₃, δ ppm): 48.49 (N–CH₂–fa), 48.95 (N–CH₂–Ar), 56.21 (–OCH₃), 83.44 (O–CH₂–N), 108.17, 109.41, 110.41, 119.87, 124.88, 129.35,142.91, 148.64, 149.28, 151.10, 190.97 (–CHO). HRMS (ESI interface-positive ions): [M + H + H₂O − CH₂O]⁺ 262.1074 (theoretical) 262.1086 (calculated). FTIR-ATR (diamond crystal/ν cm⁻¹): 3140, 3118, 3004, 2843, 2806, 2752, 2715, 1684, 1226, 1142, 1126, 1010, and 949.

Synthesis of magnetic nanoparticles (Fe₃O₄, abbreviated as MNPs)⁵⁶. A mixture of FeCl₃·6H₂O (2.8 g, 10.35 mmol) and FeCl₂·4H₂O (1.0 g, 5.02 mmol) were dissolved in DI water (50 mL) and stirred at 400 rpm at 80 °C for 1 h under a continuous N₂ flow. Aqueous NH₄OH (23% w/v, 6 mL) was rapidly added to attain a basic pH, and the mixture was stirred vigorously at 1000 rpm for additional 1 h. The suspension was allowed to cool to room temperature, and the MNPs were collected by filtration and washed thoroughly with DI water.

Preparation of the imine prepolymer (Si_P0-Vfa) and polymer [poly(Si_P0-Vfa)]. Siloxane amine (Si_P0, 0.048 mmol) was added to a solution of Vfa (0.12 mmol), dissolved in a minimal amount of chloroform/EtOH (5 mL, 4 : 1 v/v), and the mixture was stirred at room temperature for 12 h. The solvent was removed under vacuum to give a yellow viscous imine functionalized prepolymer. The prepolymer was transferred to a PTFE mold (40 × 10 × 1 mm) and oven cured at a heating rate of 10 °C h⁻¹ till 220 °C to obtain a soft and flexible polymer film.

Fabrication of poly(Si_P0-Vfa_Fex%) nanocomposite films. To a stirring solution of Si_P0-Vfa (5 g) in CHCl₃ (20 mL), MNPs (15 wt% and 20 wt%) were added and kept in a bath sonicator for 1 h to uniformly disperse the MNPs. The resulting black colored dispersion was gradually heated on a hot-plate to 180 °C with intermittent manual stirring. This mixture was poured into a PTFE mold (40 × 10 × 1 mm) and oven cured at a heating rate of 10 °C h⁻¹ till 220 °C.

Solvent stability of polymers. Polymer samples (10 mg) were immersed separately in various solvents (CHCl₃, THF, DMSO, MeOH, acetone, 2-MeTHF, MEK, MeCN, hexane, 1N HCl, 1N NaOH and H₂O) and kept undisturbed for 8 days at 25 °C. Samples were removed from the solvents, gently dried with a Kimwipe, and were weighed to determine the mass change resulting from swelling or dissolution (eqn (1)).where m₁ denotes the initial mass and m₂ represents the remaining mass after 8 days.

Reprocessing, self-healing and reshaping polymer networks. The self-healing behavior of the polymer films at room temperature was evaluated by introducing a controlled cut using a razor blade, followed by bringing the freshly exposed surfaces into intimate contact to allow autonomous healing. This procedure was repeated for multiple healing cycles to assess reproducibility and durability of the healing response.

For reprocessing and reshaping, the vitrimer samples were immersed in 2-MeTHF/MEK under stirring at 50 °C for 4 h, with intermittent bath sonication to facilitate dissolution. The resulting homogeneous polymer solution was cast into dog-bone-shaped molds. After solvent evaporation under ambient conditions, the recast films were annealed at 80 °C to obtain reprocessed specimens with a uniform morphology.

Recycled carbon fiber-reinforced vitrimer (rCF-Si_P0-Vfa). The fibers were recovered from composite scrap waste through a pyrolysis-based recycling process. The resulting raw pyrolyzed fibers were subsequently processed via carding, layering, and needle-felting to produce semifinished nonwoven fabrics. This felt was supplied by Gen 2 Carbon Ltd (Bilston, United Kingdom) and possessed an areal density of 200 g m⁻². A single ply of the fabric was cut to dimensions of 150 × 200 mm² and placed into a mold of matching size. Subsequently, the low-viscosity Si_P0-Vfa resin (100 g) was dispensed onto the fabric and uniformly distributed using a spatula to ensure complete and homogeneous fiber wet-out across the entire surface. The final composite exhibited a fiber weight fraction (W_f) of 5.6%. Curing was carried out in a convection oven (Vötsch VTL 100/150, Vötsch Industrietechnik GmbH, Reiskirchen, Germany) following the specified polymer curing protocol, as mentioned above. After curing, the composite panels were trimmed to the required dimensions using a punch cutter.

Characterization. Nuclear magnetic resonance (¹H and ¹³C NMR) spectra were recorded on a Bruker AVANCE III 400 MHz spectrometer. Chemical shifts (δ, ppm) are referenced to residual solvent signals (CDCl₃: δ_H = 7.26 ppm; δ_C = 77.2 ppm) and splitting patterns are denoted as s (singlet), d (doublet), t (triplet), and m (multiplet). Samples were prepared in CDCl₃ (10 mg in 0.5 mL), and spectra were acquired at 0–5 °C unless stated otherwise. High-Resolution Mass Spectra (HRMS) were recorded on a Q-TOF 6540 series in the ESI positive mode. Fourier-transform infrared (FT-IR) spectroscopy was performed on a Nicolet iS20 mid-infrared spectrometer with FTIR transmission measurements conducted as 64 scans at a spectral resolution of 4 cm⁻¹ under a nitrogen atmosphere for the prepolymer prepared on CaF₂ substrates (2.5 × 1 cm²). Gel permeation chromatograms (GPC) were recorded to determine the relative molecular weight of copolymers using the Viscotek Model 305 TDAmax fitted with a Viscotek modular differential refractive index detector (VE 3580 model). Two general mixed columns (T6000M, standard styrene-divinylbenzene copolymer, 300 × 8 mm) were maintained at 35 °C, and tetrahydrofuran (THF) was used as an eluent at a flow rate of 1 mL min⁻¹. The instrument was precalibrated with polystyrene standards, and data were analyzed using the Omnisec software. For measurement, samples were left overnight in tetrahydrofuran (4–5 mg mL⁻¹) and prefiltered through a polytetrafluoroethylene filter (0.2 µm) to remove insoluble particles before injecting into the instrument. Differential scanning calorimetry (DSC) was performed to determine the polymerization behaviour of monomers and prepolymers using a DSC-3 differential scanning calorimeter (Star System, Mettler Toledo). Samples (1 ± 0.2 mg) were placed in hermetically sealed aluminium pans and heated from 25 to 350 °C at 10 °C min⁻¹ under nitrogen (50 mL min⁻¹). The characteristic polymerization temperatures, initiation (T_i) and peak (T_p), and heat of polymerization (ΔH) were determined from the DSC curve. Thermogravimetric analysis (TGA) of monomers, prepolymers and polymers was performed on a Mettler Toledo TGA instrument (TGA2 SF/1100) at a nitrogen flow rate of 50 mL min⁻¹ and a heating rate of 10 °C min⁻¹ over 35–800 °C. The temperature at maximum mass loss (T_max) was determined from the differential thermogravimetry (DTG) trace. Surface morphology of the synthesised magnetic vitrimers was examined using field-emission scanning electron microscopy (FE-SEM, JEOL) coupled with energy-dispersive spectroscopy (EDS). The crystalline structure was investigated by powder X-ray diffraction (PXRD) using a Rigaku SmartLab SE 3 kW diffractometer equipped with Cu-K_α radiation (λ = 0.154 nm). X-ray photoelectron spectroscopy (XPS) was performed on an Omicron Multiprobe surface analyzer with monochromatic Al-K_α radiation (1486.7 eV); spectra were calibrated to the C 1s peak at 284.6 eV and processed using CasaXPS software. Magnetization measurements were performed using a vibrating sample magnetometer (VSM, Microsense ADE-EV9) at room temperature under fields up to 1.5 T. Saturation magnetization (M_s) was obtained by linear extrapolation of magnetization versus the inverse magnetic field. The magnetic strength of the magnet was determined by using a gaussmeter. A custom electromagnet setup was built to generate oscillating magnetic fields for frequency-controlled actuation experiments. Rheological measurements were performed on an MCR302e (Anton Paar) rheometer equipped with a convection temperature device (PTD-220) and a parallel plate geometry having a disposable lower measuring plate (25 mm) sandwiched with the upper plate (8 mm). To determine the glass transition temperature (tan δ) and storage modulus of the polymer film, it was cooled down to room temperature following an oscillatory mode with an amplitude of γ = 0.5% and a frequency of 1 Hz. Crosslinking density (ρ) was calculated using Flory's⁵⁷ equation (rubbery elasticity) (eqn 2).where G′ is the storage modulus at temperature T_g + 30 K, R is the gas constant, and T is the absolute temperature.

Stress relaxation experiments were conducted on the polymer samples at a constant strain of 1% and the relaxation modulus (G) was monitored over 5000 seconds allowing the sample to relax over time at 25, 40, 60 and 80 °C. The relaxation modulus G(t)/G(o) was plotted as a function of time at different temperatures. The relaxation time (τ) was calculated as per the Maxwell exponential equation (eqn 3), and activation energy (E_a) was calculated using the Arrhenius equation (eqn 4).

ln τ = ln τ₀ + E_a/RT (4)

where t is the measurement time and τ is the relaxation time when the value of G(t)/G(o) reaches 1/e, implying that 0.37 or 67% of the stress is relaxed. R is the gas constant (8.314 J K⁻¹ mol⁻¹), T is the measurement temperature (K), and E_a is the activation energy (J). Creep recovery tests were performed for 10 cycles at a constant shear stress of 50 Pa at 25 °C for 10 min after which the stress was removed and the sample was allowed to recover.

For the magnetorheological measurements, a titanium parallel-plate geometry (diameter 20 mm) was employed, with a measurement gap maintained at 1 mm between the upper plate and the lower base plate. In this configuration, the magnetic field was applied perpendicular to the shear flow direction, with field strengths ranging from 0 to 0.6 T, corresponding to coil currents between 0.02 and 3 A. The shear rate was systematically varied over a broad range from 0.01 to 1000 s⁻¹ to capture the full rheological response under different magnetic field intensities. Frequency sweep measurements were performed from 0.1 to 500 rad s⁻¹ at a strain (γ) of 0.1% at 25 °C. Dynamic mechanical analysis (DMA) of the films (neat polymer/composite) was performed on a DMA Q800 (TA Instruments, Hüllhorst, Germany) instrument in tensile mode with a support span of 20 mm. Rectangular specimens (40 × 10 × 1 mm³) were tested over a temperature range of −100 °C to 100 °C, at a heating rate of 2 K min⁻¹ and a frequency of 1 Hz (5 µm). Storage modulus (E′), loss modulus (E″), and damping factor (tan δ) were recorded as a function of temperature. The composite samples were tested in the machine direction of the fabrics. All measurements were evaluated in triplicate (n = 3). Mechanical characterization of the composite films was performed using a tensile test in accordance with ISO 527-2/3, employing a Zwick universal testing machine (ZwickRoell GmbH, Ulm, Germany) equipped with a 1 kN XforceP load cell at a test speed of 1 mm min⁻¹. Specimens (type A2, 150 × 6.5 × 3 mm³) were tested with a sample size of five (n = 5). All specimens were loaded in the machine direction of the needle-punched felt fabrics and the tests were conducted under standard laboratory conditions (23 °C, 50% RH). Water contact angle (WCA) measurements were performed using a goniometer (Theta Flex, Biolin Scientific) and the OneAttension software. Static WCAs were evaluated using a 10 µL droplet of deionized water (Millipore DI). Optical microscopy was used to investigate the self-healing behaviour of the dynamic imine-crosslinked networks. Samples were scratched with a razor blade and imaged using an optical microscope at 10× magnification. Images were recorded at various time intervals at room temperature to see the changes in the crack width. Mechanical testing was conducted on an Instron 3362 universal testing machine fitted with a 10 kN load cell at a crosshead speed of 0.1 mm s⁻¹ for determining self-healing/recycling efficiency of the vitrimers at room temperature using eqn 5.

Healing efficiency (%) = σ_H/σ_V × 100% (5)

where σ_H and σ_V are the maximum tensile strength (MPa) of the healed and virgin samples, respectively.

Results and discussion

Synthesis and characterization of the imine (pre)polymer

Successful synthesis of the aldehyde-functional benzoxazine monomer (Vfa) was confirmed by ¹H and ¹³C NMR spectroscopy through the presence of characteristic oxazine ring signals along with the aldehyde proton (δ 9.8 ppm) and carbon (δ 190.8 ppm) resonances (Fig. S1). The corresponding molecular ion peak observed in mass spectrometry further corroborates the expected molecular structure (Fig. S2).

Building on this, an imine-functional (pre)polymer (Si_P0-Vfa) was synthesized via a catalyst-free Schiff-base condensation between the amine-functional polysiloxane (Si_P0) and Vfa at room temperature, as illustrated in Fig. 2. To achieve controlled incorporation of benzoxazine units and residual amine functionalities, the reaction stoichiometry was systematically varied ((Vfa : Si_P0) = 1 : 7 to 1 : 1.6 equiv.; 1 : 10 to 1 : 30 w/w; Table S1), and the extent of condensation was monitored using NMR and FTIR spectroscopy. Stacked ¹H NMR spectra (Fig. S3) reveal a progressive decrease in the aldehyde proton signal (–CHO, δ ∼9.8 ppm) with increasing Vfa as in the Vfa : Si_P0 feed-in ratio ascribed to the condensation with the primary amine methylene in Si_P0. Concurrently, the emergence and growth of the imine proton signal (–CH=N–, δ ∼8.3 ppm) confirms efficient Schiff-base formation. Quantitative analysis of aldehyde consumption and imine formation (Fig. 2a) indicates near-complete conversion at lower Vfa loadings. At an optimized stoichiometric ratio of 1 : 25 (Vfa : Si_P0), complete consumption of the aldehyde functionality is observed. Furthermore, this is supported by the characteristic silicon environment signal in ²⁹Si NMR (Fig. 2b) confirming successful siloxane amine incorporation in formation of the prepolymer. Additionally, a downfield shift of methylene protons adjacent to the imine linkage (–CH₂–CH=N–) to δ ∼3.54 ppm (Fig. 2c and S4) further supports successful covalent integration of Vfa into the siloxane backbone. Notably, ∼20% residual amine functionalities remain unreacted, in close agreement with theoretical predictions (refer to the SI), indicating controlled and selective Schiff-base formation.

Fig. 2 (a) Percentage conversion of aldehyde to imine with varying w/w% ratios, (b) ²⁹Si NMR spectrum of the prepolymer and (c) stacked ¹H NMR spectra evidencing the formation of the imine prepolymer (Si_P0-Vfa) from its precursors. *Recorded in CDCl₃.

The formation of imine linkages is further substantiated by FTIR spectroscopy (Fig. 3a and S5), which shows the disappearance of the aldehyde C=O stretching band (∼1685 cm⁻¹) and the appearance of a new C=N stretching band at 1640 cm⁻¹. These observations confirm successful condensation between the aldehyde groups of Vfa and amine functionalities of the siloxane backbone, resulting in imine-linked pendant benzoxazine units. Characteristic Si–O–Si stretching vibrations (1250 cm⁻¹) are observed in the expected region, confirming the integrity of the siloxane backbone. However, due to the spectral complexity and dominance of siloxane-related vibrations, the characteristic oxazine ring absorptions of the incorporated Vfa units are partially obscured. Structurally, the flexible siloxane segments are expected to enhance chain mobility, while the benzoxazine moieties serve as latent crosslinking sites via subsequent thermally induced ROP enabling the formation of a dynamic yet mechanically robust network. The reaction scheme illustrating the probable polymeric network structure is presented in Fig. S6.

Fig. 3 Structural and thermal characterization of the prepolymer and vitrimer. (a) FT-IR spectra of poly(Si_P0-Vfa) and its precursors, (b) GPC traces of soluble fractions with respect to the precursors, (c) DSC profile of the imine prepolymer (Si_P0-Vfa), and (d) TGA (DTG, inset) thermograms.

Gel permeation chromatography (GPC) analysis provides further evidence for successful incorporation of Vfa units into the siloxane backbone. Poly(Si_P0-Vfa) exhibits a lower retention volume compared to the individual components, consistent with an increase in hydrodynamic volume and molecular weight arising from benzoxazine grafting along the polysiloxane chain (Fig. 3b). This shift confirms effective chain extension through Schiff-base coupling in the formation of polymer. The preservation of oxazine functionality and its thermal reactivity were evaluated using differential scanning calorimetry (DSC). The Si_P0-Vfa prepolymer displayed a low glass transition temperature (T_g = −20 °C), reflecting the dominance of flexible siloxane segments. Upon heating, the material exhibited a characteristic exothermic peak associated with benzoxazine ROP at a higher temperature (T_p = 250 °C) than pristine Vfa (T_p = 204 °C). The polymerization enthalpy is markedly reduced (ΔH = 56 J g⁻¹ vs. 120 J g⁻¹)⁵⁸ (Fig. 3c), which confirms the successful incorporation of benzoxazine functionalities into the prepolymer. The lower enthalpy further indicates a limited degree of functionalization, constrained by the available amine groups in Si_P0. Notably, the initiation temperature (T_i) of polymerization in Si_P0-Vfa is lower than that of neat Vfa, which is attributed to the initiating the ring-opening process by the residual amine.⁵⁹ This observation highlights an intrinsic advantage of the developed prepolymer, where amines not only facilitate curing without additional catalysts but are also expected to contribute towards dynamic imine exchange.

Thermal stability of the cured networks was evaluated using (differential) thermogravimetric analysis (TGA/DTG). The fully cured poly(Si_P0-Vfa) exhibits significantly enhanced thermal stability, with a higher degradation temperature (T_d5% = 345 °C, T_d10% = 395 °C and T_max = 550 °C), compared to Si_P0 (T_d5% = 105 °C, T_d10% = 158 °C and T_max = 445 °C) and poly(Vfa) (T_d5% = 300 °C, T_d10% = 365 °C and T_max = 385 °C) (Fig. 3d). This improvement reflects the synergistic contribution of the thermally stable siloxane backbone and the crosslinked benzoxazine network. Incorporation of MNPs further improved thermal stability, with poly(Si_P0-Vfa_Fe20%) showing a T_d5% = 380 °C, T_d10% = 415 °C and T_max of ∼560 °C, suggesting additional interfacial interactions between the nanoparticles and the polymer matrix.

Powder X-ray diffraction (PXRD) confirms that the characteristic reflections of Fe₃O₄ are retained within the predominantly amorphous polymer matrix, indicating that structural integrity of nanoparticles is unaffected within the polymer matrix (Fig. S7). Atomic force microscopy (AFM) reveals uniform embedding of MNPs within the polymer film, with 2D and 3D topographical images showing an increase in surface roughness (R_a ≈ 7.5 nm) (Fig. 4a and b), consistent with nanoscale dispersion. Scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM/EDS) further confirms a homogeneous distribution of MNPs throughout the matrix, with no evidence of significant aggregation (Fig. S8). To probe interfacial chemical interactions, X-ray photoelectron spectroscopy (XPS) was employed. The survey spectrum (Fig. 4c) displays characteristic signals corresponding to Si 2p, C 1s, O 1s, and Fe 2p, confirming successful incorporation of MNPs into the polymer network. Deconvoluted O 1s spectra of poly(Si_P0-Vfa_Fe20%) and its precursors (Fig. S9) provide insight into the oxygen binding states to C/Si. Interestingly, a 0.6 eV shift in C–O/C–OH of the pristine polymer (from 533.2 to 532.8 eV) to poly(Si_P0-Vfa_Fe20%) is observed ascribed to bonding interactions of iron centres and phenolic-OH formed from the thermally mediated ROP of tagged-Vfa on siloxane. The XPS peak at 530.5 eV, assigned to the C–O–Fe linkage, verifies the presence of coordinative interactions between polymer-bound oxygen atoms and iron centres, promoting uniform dispersion of the MNPs throughout the matrix. This coordination strengthens the network architecture and contributes to the enhanced thermal stability observed. Moreover, the Fe 2p signal in poly(Si_P0-Vfa_Fe20%) exhibited a shift toward higher binding energy relative to bare MNPs, along with an increase in the unreacted Fe²⁺ : Fe³⁺ ratio in the deconvoluted spectra. These spectral changes further substantiate the formation of interfacial coordination between the dynamic polymer matrix and the embedded nanoparticles.

Fig. 4 AFM images (2D/3D) of the surface topography of cured polymers for a 6 µm² scan area, with surface features (a) without and (b) with MNPs in the polymer matrix. Surface average roughness (R_a) of bare (poly(Si_P0-Vfa: 0.205 nm)) and MNP-bearing poly(Si_P0-Vfa_Fe20%: 7.5 nm) films. (c) Wide survey spectrum of films (inset shows the zoomed-in region).

Rheological measurements

DMTA (Fig. 5a and c) shows a T_g of 0 °C for poly(Si_P0-Vfa) with a storage modulus of 5550 Pa, while poly(Si_P0-Vfa_Fe20%) shows a slightly lower T_g (−7 °C) and a higher storage modulus (6367 Pa), consistent with nanoparticle reinforcement. Using eqn (2), the calculated crosslink densities are 2.203 and 2.587 mol m⁻³, respectively. Stress-relaxation experiments were conducted at 25, 40, 60 and 80 °C (Fig. 5b, d and S10), and the characteristic relaxation time (τ) was extracted at G(t)/G(o) = 1/e. Arrhenius analysis of τ yields an activation energy (E_a) of 21.4 kJ mol⁻¹ for the neat vitrimer (Fig. 5b′) and 11.4 kJ mol⁻¹ for the magnetic vitrimer (Fig. 5d′). At temperatures above T_g, vitrimers show dynamic exchanges. A low activation energy of our vitrimer can be interpreted as a combined contribution of dynamic imine exchange, interfacial coordination, and localized network relaxation rather than bond exchange alone.^60,61

Fig. 5 Rheological measurements. (a–d′) DMTA thermograms, stress-relaxation curves, and Arrhenius plots of poly(Si_P0-Vfa) and poly(Si_P0-Vfa_Fe20%). (e and e′) Creep-recovery tests for poly(Si_P0-Vfa) at 25 °C over 10 cycles.

A two-fold decreased E_a in poly(Si_P0-Vfa_Fe20%) reflects accelerated relaxation arising from nanoparticle–polymer interfacial effects that increase exchange-dynamics lability. The inherent magnetism of the MNPs further contributes by modulating chain mobility and promoting network rearrangement. The topology-freezing temperature (T_v) was estimated using the standard definition based on extrapolation to a melt viscosity of 10¹² Pa s (Fig. S11).⁶² The extrapolated T_v values are −66 °C for poly(Si_P0-Vfa) and −156 °C for poly(Si_P0-Vfa_Fe20%). These values lie well below T_g, indicating that once segmental mobility is activated above T_g, the rate of exchange is not expected to be the limiting factor for network rearrangement under the studied conditions. Details of the Arrhenius fitting and T_v calculations are provided in the SI.

In creep-recovery tests (Fig. 5e and e′), a constant shear stress of 50 Pa was applied at 25 °C. The strain increased to ∼9% under load and rapidly recovered to ∼2.3% upon stress removal, followed by a residual strain of ∼0.7%. This response was reproducible over 10 cycles, corresponding to a recovery of ∼93%, indicating good elastic recovery and limited permanent deformation under the applied stress.

Magnetic response evaluation

Building on this structural and interfacial framework, we next investigated the magnetic response, actuation behaviour, and magnetorheological performances of the MNP-bearing siloxane–imine benzoxazine resin. Magnetic hysteresis measurements were conducted on pristine MNPs and their corresponding composites, poly(Si_P0-Vfa_Fex%) (x = 10 and 20 wt%) (Fig. 6a). As expected, the saturation magnetization (M_s) increases with nanoparticle loading, reaching ∼1.3 emu g⁻¹ for poly(Si_P0-Vfa_Fe10%) and 4.5 emu g⁻¹ for poly(Si_P0-Vfa_Fe20%), compared to ∼25 emu g⁻¹ for pristine MNPs. The reduced M_s values in the composites are attributed to dilution by the non-magnetic polymer matrix. However, an increase in MNPs beyond 20 wt% led to agglomeration. The external magnetic field (H) was aligned at different orientation angles (Φ = 0–120°), resulting in the development of magnetic anisotropy due to oblique-angle deposition (Fig. 6b). Observation of appreciable magnetic response combined with the homogeneous nanoparticle dispersion at 20 wt%, poly(Si_P0-Vfa_Fe20%) was selected for further actuation and magnetorheological studies.

Fig. 6 Magnetic response and magnetorheological behavior of the poly(Si_P0-Vfa_Fe20%) prepolymer and its crosslinked film at room temperature. (a) Magnetic hysteresis loops of MNPs and poly(Si_P0-Vfa_Fex%) (inset shows the zoomed-in hysteresis plot). (b) Hysteresis loops at room temperature along Φ = 0 °, 30 °, 60 °, 90 ° and 120° for poly(Si_P0-Vfa_Fe20%). (c) Magnetic-field-induced bending and angle deflection. (d) Digital images illustrating magnetic actuation and gripping of poly(Si_P0-Vfa_Fe20%). (e) Viscosity of the neat and magnetic prepolymers under different magnetic flux densities, B (T). (f) Storage modulus (G′) as a function of stepwise magnetic fields between 0 and 0.6 T. (g–g″) Frequency-sweep variation of the storage modulus (G′) and loss modulus (G″) of poly(Si_P0-Vfa_Fe20%) under variable magnetic flux (0–0.6 T).

Under an external magnetic field, poly(Si_P0-Vfa_Fe20%) films exhibit clear and reversible actuation behaviour. Exposure to a permanent magnet (75 mT, Fig. S12) induces directional bending and controlled alignment of the films (Fig. 6c; Movies S1 and S2), demonstrating effective magnetic torque generation within the compliant matrix. Furthermore, when subjected to an alternating magnetic field generated using a custom electromagnet setup, the material displays programmable oscillatory motion (Movie S3), highlighting its potential for dynamic actuation. The stroke length of the magnetic film was found to be in the range of 2–3 cm under 75 mT (Fig. S13). The composite also enables magnetically driven gripping (Fig. 6d; Movie S4) and appreciable adhesion to various substrates (Fig. S14), supporting its applicability in soft robotic manipulation and adaptive interfaces. The load bearing capacity under a magnetic actuation of 75 mT was found to be up to 10 g (as shown in Fig. S15), corresponding to an output force of nearly 0.1 N.

To further elucidate the field-dependent viscoelastic behaviour, magnetorheological (MR) measurements were performed on the prepolymer resins. The neat prepolymer (Si_P0-Vfa) exhibits nearly shear-rate-independent viscosity, indicative of newtonian behaviour (Fig. 6e). In contrast, poly(Si_P0-Vfa_Fe20%) shows non-newtonian characteristics, reflecting the influence of particle–matrix interactions. Upon application of a magnetic flux, a substantial increase in viscosity is observed at a representative shear rate, and the viscosity increases from ∼220 mPa s (B = 0 T) to ∼1200 mPa s at 0.6 T, with a plateau reached at B ≥ 0.4 T. This behaviour is consistent with field-induced chaining and structuring of magnetic nanoparticles, which enhance resistance to flow. Magnetic field induced shear thinning behaviour (Fig. S16) as a function of shear rate in the MNP-bearing prepolymer, whereas the pristine prepolymer remains unaffected.

The solid-state magnetorheological response of the cured vitrimers was further investigated through dynamic mechanical analysis. The storage modulus (G′) of poly(Si_P0-Vfa_Fe20%) increases dramatically (nearly up to four orders of magnitude) upon increasing the magnetic flux density from 0 to 0.6 T (Fig. 6f). Importantly, stepwise switching of the magnetic field results in reversibility with synchronous changes in modulus, confirming rapid and repeatable field responsiveness.

Dynamic oscillatory frequency sweeps at 25 °C of poly(Si_P0-Vfa_Fe20%) further revealed the influence of magnetic fields on relaxation dynamics (Fig. 6g–g″). In the absence of a magnetic field (0 T), multiple G′–G″ crossover points are observed, reflecting a broad relaxation spectrum characteristic of vitrimeric networks. Upon application of a magnetic field (0.4 T), the crossover points shift, indicating modification of relaxation processes due to nanoparticle structuring. From the first crossover, the characteristic relaxation time (τ = 1/ω) decreases from ∼9.17 ms (0 T) to ∼7.14 ms (0.4 T), suggesting accelerated stress dissipation under field conditions. No observable crossover point is observed at 0.6 T under the studied range inferring very fast structural dynamics. Expectedly, no magnetic field effect is noticed in the pristine polymer devoid of MNPs (Fig. S17a). Frequency-dependent moduli under different flux densities (Fig. S17b) of poly(Si_P0-Vfa_Fe20%) indicate that this field-induced stiffening persists across the tested frequency range, underscoring the robustness of the magnetically structured network.

Room-temperature self-healing, reshaping, and chemical recycling

To assess self-healing performance, incised films were brought in close proximity, and they were seen to undergo macroscopic healing at room temperature without external stimuli. Notably, poly(Si_P0-Vfa_Fe20%) achieved complete healing within ∼8 h, whereas the neat polymer requires ∼24 h under identical conditions (Fig. 7a, Movie S5). This enhancement in healing kinetics is consistent with earlier rheological observations, where nanoparticle incorporation substantially reduced the E_a for network rearrangement. The vitrimer films also retain mechanical compliance and functional integrity, exhibiting high flexibility, load-bearing capacity (up to 10 g), and reliable shape recovery (Fig. 7b–d). Optical microscopy scratch healing (Fig. S18) confirmed progressive crack closure at room temperature. Tensile testing revealed average self-healing efficiencies of 88% for poly(Si_P0-Vfa) and 92% for poly(Si_P0-Vfa_Fe20%) over three consecutive healing cycles (Fig. 7f–g), highlighting better healing performance of the magnetic vitrimer. These results, together with their efficient and repeatable healing behaviour, highlight their potential for healable soft-material applications.

Fig. 7 Digital images demonstrating (a) room-temperature self-healing, (b) flexibility, (c) load-bearing capability (10 g), and (d) shape recovery of the polymer film. (e) Solvent-assisted recycling followed by recasting and remoulding. (f) Tensile tests of the cut and self-healable film at 25 °C for consecutive healing cycles. (g) Self-healing efficiency of the healed films. (h) Comparison of thermal stability vs. activation energy and healing temperature of poly(Si_P0-Vfa_Fe20%) with the other reported vitrimers (from the literature).

As an initial assessment, water–polymer interactions were evaluated via WCA measurements (Fig. S19a), which indicate appreciable hydrophobicity and effective water repellence. To further probe chemical stability and reprocessability, solvent interaction studies were performed across solvents of varying polarity (Fig. S19b). Both vitrimer networks showed excellent hydrolytic stability, exhibiting negligible mass change even after 8 days (Fig. S20 and Table S2), surpassing previously reported imine-based vitrimers.⁶³ They also retained film integrity in highly polar organic solvents (MeOH and DMSO). Unexpectedly, moderate swelling (∼20%) occurred in the non-polar solvent hexane, likely due to enhanced solvent ingress arising from polarity compatibility with the network. In contrast, substantial mass loss in THF, 2-methyl THF, MEK and CHCl₃ suggests solvent-assisted bond exchange and partial depolymerisation, reducing the gel fraction.^64,65 This selective solubility enables solvent-assisted chemical recycling (Fig. 7e) which was further validated by tensile testing, where the recycled vitrimers exhibited tensile stress comparable to those of the pristine samples with average recycling efficiency of 92% (Fig. S21), demonstrating excellent retention of material performance. This confirms a practical circular reprocessing route without compromising the inherent softness and flexibility of the vitrimer network. Fig. 7h (Table S3) provides a comparative assessment of reported imine-based magnetic vitrimers and the present soft network, highlighting its unique combination of efficient room-temperature healing, low-energy exchange dynamics, and exceptional thermal stability.

Mechanical reinforcement of siloxane-imine vitrimers using recycled carbon fibres

While the siloxane-imine vitrimer exhibits excellent mechanical adaptability and dynamic functionality, its inherently soft nature with a tensile strength of ∼12 kPa and ∼7 kPa for poly(Si_P0-Vfa_Fe20%) and poly(Si_P0-Vfa) respectively as shown in Fig. S22, can restrict its applicability in high load-bearing or structurally demanding environments. To address this limitation without compromising reprocessability, recycled carbon-fibre (rCF) felt (W_f ≈ 5.6 wt%) was incorporated as a reinforcing phase via impregnation with the low-viscosity Si_P0-Vfa prepolymer, followed by in situ curing (Fig. 8a). The selection of rCF nonwovens was based on the aim of reinforcing the polymer isotropically using a randomly oriented, nonwoven fibre architecture. In this context, rCF felt provides a high-performance reinforcement, while simultaneously offering sustainability advantages by reintroducing CF from waste streams into value-added composite applications.

Fig. 8 (a) Schematic representation of the recycling route and reuse of recycled carbon fibres (rCFs). (b) DMA curves of the neat polymer and rCF-reinforced composite. (c) Stress–strain curves of the neat polymer and rCF-reinforced composite.

Dynamic mechanical analysis (DMA) reveals a clear reinforcing effect, with the composite exhibiting a higher storage modulus compared to the neat vitrimer across the entire temperature range investigated (Fig. 8b). This enhancement is attributed to effective stress transfer between the polymer matrix and the fibre network, as well as interfacial confinement of polymer chains in the vicinity of the fibre surface. Such immobilisation is known to restrict segmental mobility locally, contributing to increased stiffness.⁶⁶ Representative stress–strain curves for the composite (Fig. 8c) further highlight the mechanical response of the reinforced system.

The composite exhibits a Young's modulus of 3.1 ± 0.1 MPa, a tensile strength of 125.4 ± 2.3 kPa, and an elongation at break of 17.4 ± 1.8% (n = 5). The curve progression is characterised by a short initial linear regime followed by a nonlinear rise (typical for elastomers)⁶⁷ and a gradual drop after reaching a maximum. The overall bell-shape of the curves and the gradual ductile failure are observed due to the integration of the rCF felt fabric leading to toughening and increased energy dissipation (area under the curve) due to fibre pull-out and progressive failure of the fabric cohesion. However, the general material behaviour is strongly matrix dominated due to low mechanical contribution of the rCF to the composite's mechanical properties.

This can be attributed not only to the low fibre weight fraction and random discontinuous fibre orientation in the felt fabric, but also to the drastically lower polymer stiffness compared to the rCF leading to inefficient load transfer between the fibre and matrix and hence low mechanical contribution of the fibres to the composite properties. The resulting low stiffness of the composite indicates low mechanical interaction of fibres and the matrix.

Especially for the sample containing rCF, the composite film exhibits comparatively low stiffness and strength when compared with other vitrimeric fibre-reinforced composite materials.⁶⁸ However, the introduction of the rCF fabric can maintain the isotropic integrity of the polymer film even at low fibre content. The present material combination however presents a promising solution to increase isotropic performance and extend the applicability of the developed polymer in technical applications such as membranes, sealings, isolators or adaptive and smart materials.

Conclusion

In summary, a bio-based siloxane–imine benzoxazine vitrimer was developed that couples a thermally robust benzoxazine network with highly labile imine exchange linkages. After ring-opening polymerization, the neat vitrimer exhibits T_max ≈ 550 °C and rapid stress relaxation with E_a = 23 kJ mol⁻¹, enabling room-temperature self-healing (healing efficiency ∼90%) and reshaping under mild conditions. Incorporation of 20 wt% MNPs reduced the E_a to 11.4 kJ mol⁻¹ and provides magnetic-field-addressable mechanical tuning and programmable actuation, as evidenced by magnetorheology and bending/gripping demonstrations. The networks showed hydrophobic surfaces, high creep recovery, good hydrolytic stability, and solvent-assisted reprocessing via dissolution and recasting. Finally, reinforcement with recycled carbon-fibre felt improves mechanical performance while retaining processability, highlighting a promising route towards sustainable soft magnetic actuators and chemically recyclable composite components.

Author contributions

Prashansa Gupta and Bhavika Bhatia: conceptualization, methodology, investigation, formal analysis, data curation, visualization, funding acquisition, writing – original draft. Bimlesh Lochab: conceptualization, supervision, funding acquisition, project administration, writing – review & editing. Carbon fiber reinforcement work was performed at IFAM, Bremen. Jan-Marten Sprenger: methodology, investigation, formal analysis, writing – review & editing. Adrian Wolf: analysis. Katharina Koschek: supervision, resources, funding acquisition, writing – review & editing. All authors contributed to the interpretation of the results, critically reviewed the manuscript, and approved the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data generated or analyzed are included within the manuscript and supplementary information (SI) files. Supplementary information is available. See DOI: https://doi.org/10.1039/d6ta02558g.

Acknowledgements

PG and BB acknowledges a PhD fellowship from Shiv Nadar University. BL acknowledges the financial and instrument support from the Shiv Nadar Foundation and Shiv Nadar University. PG acknowledges the MAPEX PhD Incoming Research Grant from the MAPEX Center for Materials and Processes, University of Bremen. KK and JMS acknowledge the project FOREST (Advanced lightweight materials FOR Energy-efficient STructures) and KK also acknowldges the funding support from the European Union's Horizon 2020 research and innovation programme (ID: 101091790).

References

1. G. Chen, S. Wen and Z. Yue, Langmuir, 2022, 38, 1194–1203.
2. Z. Wang, J. Fan, D. He, L. Ren, Z. Hao, R. Sun and X. Zeng, J. Mater. Chem. A, 2022, 10, 21923–21932.
3. S. Bonardd, M. Nandi, J. I. Hernández García, B. Maiti, A. Abramov and D. Díaz Díaz, Chem. Rev., 2023, 123, 736–810.
4. X. Qiu and X. Zhang, J. Polym. Sci., 2024, 62, 3137–3155.
5. D. Sahu, R. K. Sahu, Y. N. Banothu, S. Bhattacharya and H. Pathak, Sens. Actuators, A Phys., 2025, 392, 116695.
6. R. Zhang, M. Lai, H. Bai, H. Ma and L. Han, J. Mater. Chem. C, 2024, 12, 8525–8533.
7. A. Trejo-Machin, A. Adjaoud, L. Puchot, R. Dieden and P. Verge, Eur. Polym. J., 2020, 124, 109468.
8. B. Lochab, M. Monisha, N. Amarnath, P. Sharma, S. Mukherjee and H. Ishida, Polymers, 2021, 13, 1260.
9. P. Froimowicz, C. R. Arza, L. Han and H. Ishida, ChemSusChem, 2016, 9, 1921–1928.
10. F. Fu, M. Shen, D. Wang, H. Liu, X. Xu, S. Shang, Z. Song and J. Song, Ind. Crops Prod., 2022, 188, 115615.
11. X. Wang, X. Yang, C. Zhao, Y. Pi, X. Li, Z. Jia, S. Zhou, J. Zhao, L. Wu and J. Liu, Angew. Chem., Int. Ed., 2023, 62, e202302829.
12. F. Fu, X. Zhou, M. Shen, D. Wang, X. Xu, S. Shang, Z. Song and J. Song, ACS Sustainable Chem. Eng., 2023, 11, 5973–5985.
13. X. Zhou, F. Fu, M. Shen, Q. Li, H. Liu and Z. Song, Adv. Sustain. Syst., 2024, 8, 2300372.
14. P. Gupta, S. K. Rastogi, B. Bhatia, S. Yadav and B. Lochab, ACS Sustainable Chem. Eng., 2025, 13, 4055–4067.
15. A. Trejo-Machin, L. Puchot and P. Verge, Polym. Chem., 2020, 11, 7026–7034.
16. B. Bhatia, P. Gupta, C. Len and B. Lochab, Macromolecules, 2025, 58, 10725–10737.
17. S. Sahu and B. Lochab, ACS Sustainable Chem. Eng., 2024, 12, 7126–7135.
18. Y. Ma, H.-Q. Wang, P.-C. Zhao, F.-Z. Wang and C.-H. Li, J. Mater. Chem. A, 2022, 10, 20804–20812.
19. A. Adjaoud, L. Puchot and P. Verge, Polymer, 2023, 287, 126426.
20. F. Luo, W. Cui, Y. Zou, H. Li and Q. Qian, J. Mater. Chem. A, 2023, 11, 15456–15465.
21. D. Yang, K. Zhao, R. Yang, S. W. Zhou, M. Chen, H. Tian and D. H. Qu, Adv. Mater., 2024, 36, 2403880.
22. J. C. Hughes, J. A. Wilson, N. Hawkins, Y. Zhang, C. Holland and A. T. Slark, ACS Sustainable Chem. Eng., 2025, 13, 18992–19001.
23. X. Lei, Y. Jin, H. Sun and W. Zhang, J. Mater. Chem. A, 2017, 5, 21140–21145.
24. M. Delahaye, J. M. Winne and F. E. Du Prez, J. Am. Chem. Soc., 2019, 141, 15277–15287.
25. T. Debsharma, V. Amfilochiou, A. A. Wróblewska, I. De Baere, W. Van Paepegem and F. E. Du Prez, J. Am. Chem. Soc., 2022, 144, 12280–12289.
26. A. Adjaoud, B. Marcolini, R. Dieden, L. Puchot and P. Verge, J. Am. Chem. Soc., 2024, 146, 13367–13376.
27. R. W. Layer, Chem. Rev., 1963, 63, 489–510.
28. K. Acharyya and P. S. Mukherjee, Angew. Chem., 2019, 131, 8732–8745.
29. Y. Tao, L. Fang, J. Zhou, C. Wang, J. Sun and Q. Fang, ACS Appl. Polym. Mater., 2020, 2, 295–303.
30. Z. Zhang, D. Lei, C. Zhang, Z. Wang, Y. Jin, W. Zhang, X. Liu and J. Sun, Adv. Mater., 2023, 35, 2208619.
31. L. J. Hamernik, W. Guzman and J. S. Wiggins, J. Mater. Chem. A, 2023, 11, 20568–20582.
32. G. P. Carden, M. L. Martins, G. Toleutay, S. Ge, B. Li, S. Zhao and A. P. Sokolov, Macromolecules, 2024, 57, 8621–8631.
33. Y. Li, Q. An, S. Chen, J. Zhang, H. Liu, S. Shao, L. Yin and Z. Wang, Polymer, 2025, 129173.
34. S. Zhang, J. Yi, J. Chen, Y. Li, B. Liu and Z. Lu, ChemSusChem, 2024, 17, e202301708.
35. Y. Qu, X. Lu and Z. Xin, ACS Sustainable Chem. Eng., 2024, 12, 7739–7747.
36. S. Gulyuz and B. Kiskan, Macromolecules, 2024, 57, 2078–2089.
37. A. Wolf, A. Hartwig and K. Koschek, Macromolecules, 2025, 58, 9091–9097.
38. L. Liu, S. Liang, Y. Huang, C. Hu and J. Yang, Chem. Commun., 2017, 53, 12088–12091.
39. P. Wang, L. Yang, B. Dai, Z. Yang, S. Guo, G. Gao, L. Xu, M. Sun, K. Yao and J. Zhu, Eur. Polym. J., 2020, 123, 109382.
40. H. Hu, L. Wang, L. Wang, L. Li and S. Feng, Polym. Chem., 2020, 11, 7721–7728.
41. Y. Shan, S. Liang, X. Mao, J. Lu, L. Liu, Y. Huang and J. Yang, Soft Matter, 2021, 17, 4643–4652.
42. L. Bai, X. Yan, B. Feng and J. Zheng, Compos. B Eng., 2021, 223, 109123.
43. Y. Wang, Q. Guo, G. Su, J. Cao, J. Liu and X. Zhang, Adv. Funct. Mater., 2019, 29, 1906198.
44. Y. Chen, X. Zhao, Y. Li, Z.-Y. Jin, Y. Yang, M.-B. Yang and B. Yin, J. Mater. Chem. C, 2021, 9, 5515–5527.
45. H. Q. Tao, D. W. Yue and C. H. Li, Macromol. Mater. Eng., 2022, 307, 2100649.
46. G. Dong, Q. He and S. Cai, Soft Matter, 2022, 18, 7604–7611.
47. M. Cvek, J. Sevcik, J. Vilcakova, A. Athanassiou and A. Zych, Appl. Mater. Today, 2023, 35, 101997.
48. S. Leungpuangkaew, L. Amornkitbamrung, N. Phetnoi, C. Sapcharoenkun, C. Jubsilp, S. Ekgasit and S. Rimdusit, Adv. Ind. Eng. Polym. Res., 2023, 6, 215–225.
49. X. Zhang, L. Zhang, M. Liu, C. B. Chng, E. P. Y. Ler, J. Zhou, N. Matsuhisa and Y. J. Tan, Sci. Adv., 2025, 11, eadq7441.
50. C. Cui, X. Chen, L. Ma, Q. Zhong, Z. Li, A. Mariappan, Q. Zhang, Y. Cheng, G. He, X. Chen, Z. Dong, L. An and Y. Zhang, ACS Appl. Mater. Interfaces, 2020, 12, 47975–47983.
51. M. Kamble, A. Vashisth, H. Yang, S. Pranompont, C. R. Picu, D. Wang and N. Koratkar, Carbon, 2022, 187, 108–114.
52. J. Hong, Y. Hong, J. Jeong, D. Oh and M. Goh, ACS Sustainable Chem. Eng., 2023, 11, 14112–14123.
53. B. Brettmann, M. A. Fraga, M. Gosecka and N. Stingelin, J. Mater. Chem. A, 2023, 11, 5975–5977.
54. G. Abels, K. Koschek, P. Jones and W. Howarth, RSC Sustain., 2024, 2, 2959–2967.
55. N. K. Sini, J. Bijwe and I. K. Varma, J. Polym. Sci., Part A: Polym. Chem., 2014, 52, 7–11.
56. M. Monisha, N. Yadav, S. B. Srivastava, S. P. Singh and B. Lochab, J. Mater. Chem. A, 2018, 6, 2555–2567.
57. L. W. Hill, Prog. Org. Coat., 1997, 31, 235–243.
58. M. Monisha, N. Yadav and B. Lochab, ACS Sustainable Chem. Eng., 2019, 7, 4473–4485.
59. A. Wolf, L. Pursche, L. Boskamp and K. Koschek, Macromol. Rapid Commun., 2024, 45, 2400557.
60. L. E. Diodati, S. Liu, C. M. Rinaldi-Ramos and B. S. Sumerlin, ACS Appl. Mater. Interfaces, 2023, 15, 32957–32966.
61. A. Legrand and C. Soulié-Ziakovic, Macromolecules, 2016, 49, 5893–5902.
62. D. Montarnal, M. Capelot, F. Tournilhac and L. Leibler, Science, 2011, 334, 965–968.
63. C. Luo, Z. Lei, Y. Mao, X. Shi, W. Zhang and K. Yu, Macromolecules, 2018, 51, 9825–9838.
64. S. K. Schoustra, V. Asadi and M. M. J. Smulders, ACS Appl. Polym. Mater., 2024, 6, 79–89.
65. B. R. Elling and W. R. Dichtel, ACS Cent. Sci., 2020, 6, 1488–1496.
66. J. Zhao, S. Zhang, X. Ke, A. Pan, Q. Zhou, S. Zeng, P. Chen, Y. Xu, W. Nie and Y. Zhou, Compos. Sci. Technol., 2023, 235, 109970.
67. H. Cho, S. Mayer, E. Pöselt, M. Susoff, P. J. In 'T Veld, G. C. Rutledge and M. C. Boyce, Polymer, 2017, 128, 87–99.
68. S. Basak, D. B. Shakyawar, K. K. Samanta and M. Bhowmick, Int. J. Biol. Macromol., 2024, 282, 137280.

---