Introduction

Living systems utilize masterfully designed molecular systems to maintain biological activities. For example, sea cucumbers are interesting because some species can switch their body into a liquiform state when removed from seawater but revert to their original state following re-immersion in seawater1. One possible mechanism underlying this biofunction is the crosslinking and de-crosslinking of collagen microfibrils in the extracellular matrix by the peptides tensilin and softenin2. The body of a sea cucumber is hard when collagen microfibrils are crosslinked with tensilin but becomes soft upon interaction with softenin and subsequent de-crosslinking. Sea cucumber-inspired materials with tuneable stiffness have been reported3, however, isothermal reversible liquid–nonliquid conversion (IRLNC) with artificial soft materials remains challenging.

Thermally reversible solid–liquid phase transition typically relies on molecular ordering and disordering is a common process for most organic molecules. The control of molecular ordering by photoisomerization of azobenzene-containing molecules has been described, leading to isothermal crystal melting and recrystallization in response to UV4,5,6,7,8,9,10,11,12,13. Although such commendable studies achieved solid–liquid transition with azobenzene-containing crystalline and liquid crystalline small molecules, these molecules do not exhibit tuneable elasticity. In contrast, the diverse range of polymers synthesized to date exhibit a wide range of tailored properties. However, attaining solid–liquid phase transition using the aforementioned mechanism seems difficult even with highly crystalline polymers because polymers are inherently a mixture of crystalline and amorphous structures. Alternatively, liquid–nonliquid conversion relying on changes in the glass transition temperature (T g) of side-chain azobenzene-containing polymers has been reported14, 15. When trans to cis photoisomerization was conducted by irradiating UV to a CH2Cl2 solution of the trans azobenzene-containing polymers having T gs above room temperature, the obtained cis azopolymers after the removal of CH2Cl2 showed T g below room temperature, attaining liquid and nonliquid states, respectively, with the same polymers. Although such formidable systems have open the door for a repeatable photochemical molding technology, development of a smart polymer system to attain IRLNC with an alternative facile and versatile mechanism remains unexplored.

Progress in polymer chemistry allows us to synthesize various nonlinear polymers. Nonlinear polymers are expected to facilitate the realization of a methodology for tuning properties of polymers without changing their chemical composition or molecular weight16. Among a wide variety of nonlinear polymers, one basic network architecture comprises star-like components. Various network organic and polymeric materials are produced by reactions between star-shaped molecules, for example, the synthesis of silica by a reaction between tetraethylorthosilicates17. In most cases, the star-shaped starting molecules in the liquid state are converted to the corresponding solid network material, but such network formation reactions are in general irreversible. To address this, we focused on a repeatable molecular architectural transformation (MAT) strategy18,19,20,21,22,23. Various repeatable MATs, for example, linear–star18, linear–cyclic19,20,21,22, three-armed star–network23 architectures have appeared and, moreover, MAT of amphiphilic block copolymers into star and comb polymers has recently been achieved by Diels–Alder reaction24. Of various candidates exhibiting repeatable cleavage and reformation of covalent bonds25 utilized as self-healing materials26, we focused on hexaarylbiimidazole (HABI)27 as the linking point. HABI can be synthesized by the oxidation of 2,4,5-triphenylimidazoles (lophines) to generate 2,4,5-triphenylimidazoryl radicals (TPIRs) and the subsequent coupling reaction between the generated TPIRs. The TPIR is stable even under the existence of oxygen but reacts with other TPIR to generate HABI28. The covalent bond between the two imidazoles in HABI is cleavable by UV irradiation, leading to the production of a pair of TPIRs, which can recouple with each other.

Here we demonstrate a solvent-free IRLNC of poly(n-butyl acrylate) (PBA) and poly(dimethyl siloxane) (PDMS) by photo-triggered repeatable MAT between star and network architectures utilizing the HABI chemistry.

Results

Photo-triggered solvent-free IRLNC by repeatable MAT of polymer materials

We envisioned that nonliquid network and liquid-star-shaped polymers can be toggled by external stimuli (Fig. 1a ) and designed a repeatable MAT system relies on the cleavage and reformation of the imidazole–imidazole covalent bond in HABI (Fig. 1b). We first focused on the repeatable MAT of PBA (Fig. 1c), a common noncrystalline acrylate polymer with low T g that is in the liquid state when its molecular weight is low. Four-armed star-shaped PBAs with bromide end groups (star-PBA Br s) were synthesized by atom transfer radical polymerization of n-butyl acrylate initiated from pentaerythritol tetrakis(2-bromoisobutyrate) (1) as the tetra-functional initiator (Supplementary Fig. 1a). Examination of the dependence of molecular weight on the state of the polymerized product revealed that an increase in the total molecular weight above 5500 Da, which corresponds to a degree of polymerization per arm (DPn(arm)) of 8, resulted in solidification of star-PBA Br (Supplementary Table 1). A compound with four lophines (2), which is a model small molecule for a star polymer with a DPn(arm) of 0, was also synthesized (Supplementary Fig. 2). Compound 2 was a crystalline solid, and thus it was suggested that attaining IRLNC with PBA at least requires a DPn(arm) between 1–7. In the present study, star-PBA Br with a DPn(arm) of 3 was selected for further study (Supplementary Table 1, entry 1). The four bromide end groups of star-PBA Br were converted to azide groups by reacting with NaN3 and the subsequent Huisgen reaction with an ethynyl derivative of lophine (3) afforded a four-armed star-shaped PBA with lophine end groups (star-PBA Lophine ) (Supplementary Fig. 1a). The resulting star-PBA Lophine was thoroughly characterized by 1H NMR (Supplementary Fig. 3a), GPC (Supplementary Fig. 4), and MALDI-TOF mass spectrometry (Supplementary Fig. 5). Oxidation reaction of the lophines followed by a coupling reaction of the generated TPIRs provided a network PBA (network-PBA HABI ) (Supplementary Fig. 1a). The synthesized network-PBA HABI was a nonliquid material poorly soluble in common organic solvents. The 1H NMR spectrum of the CDCl3 soluble part of network-PBA HABI showed that the complex signals in the aromatic region (Supplementary Fig. 3b), which did not appear before the reaction (Supplementary Fig. 3a). These signals are due to differences in the type of imidazole–imidazole covalent bond in HABI29, 30. Star-PBA Lophine and network-PBA HABI were further subjected to 1H NMR analyses using DMSO-d 6 to determine conversion yield from TPIRs to HABIs. By comparing the integration of the –NH signal at 12.5 p.p.m. appeared before (Supplementary Fig. 6a) and after the reaction (Supplementary Fig. 6b), conversion yield from TPIRs to HABIs was determined to be ca. 60%.

Fig. 1
figure 1

Design of star–network repeatable MAT system with PBA. Schematic representation of a star–network repeatable MAT, b cleavage and reformation of the covalent bond in HABI, and c repeatable MAT between network-PBA HABI and star-PBA TPIR based on UV-triggered bond cleavage and recoupling

When UV (λ = 365 nm) was irradiated to the produced network-PBA HABI under isothermal condition at 50 °C, the nonliquid network-PBA HABI (Fig. 2a) liquefied to afford a viscous material (Fig. 2b). This change was complete within 1 min (Supplementary Movie 1) and the material reverted to the nonliquid state within 3 min after terminating UV irradiation. Comparison of the ESR spectra before and after UV irradiation for 1 min directly demonstrated cleavage of the imidazole–imidazole covalent bond in HABI to produce a pair of TPIRs (Fig. 2c, red curve). The g-tensor was measured to be 2.003, which exactly corresponded to the reported values for TPIR31, 32. Comparison of the ESR spectra before and after UV irradiation also showed a 10-fold increase in the amount of TPIR after UV irradiation (Fig. 2c, black and red curves). After terminating UV irradiation for 3 min, the generated TPIRs recoupled with each other, resulting in a decrease in the relative signal intensity (Fig. 2c, blue curve). The presence of a small amount of TPIRs before and after UV irradiation suggests that the retarded recoupling of TPIRs in the later stage of network formation reaction and that difficulty in diffusion for polymer chains in the network matrix. Nevertheless, the TPIRs and their recoupling reaction indeed functioned as an IRLNC system.

Fig. 2
figure 2

IRLNC experiment with network-PBA HABI network-PBAHABI. Photographs of network-PBA HABI a before and b after UV irradiation. c ESR spectra of network-PBA HABI before (black) and during (red) UV irradiation for 1 min, and 1 min after terminating UV irradiation (blue). d Time-dependent plots of the amounts of TPIR in a THF solution of network-PBA HABI (10 mg/ml). The amount of TPIR was determined by recording A 601 every 5 s. The sample was irradiated with UV and not irradiated with UV during times corresponding to the pink and light blue regions, respectively. The value of A 601 before UV irradiation was defined as 0% and the averaged A 601 during UV irradiation was defined as 100%. e Time-dependent plots of absorbance at 601 nm for network-PBA HABI in THF. Initial network-PBA HABI concentration was 1 mg/ml (black circle), 5 mg/ml (blue triangle), and 10 mg/ml (red diamond), respectively. The starting (ON) and terminating (OFF) times of UV irradiation were indicated with dashed lines

The mechanism of the reaction was investigated using three concentrations (1, 5, and 10 mg/ml) of network-PBA HABI dispersions in THF. UV irradiation to the dispersions completely dissolved network-PBA HABI in THF and the solution became blue. The blue color disappeared after terminating UV irradiation for several minutes. These color changes were repeatable (Supplementary Fig. 7) and could be quantitatively monitored by absorbance at the wavelength of 601 nm (A 601) of the TPIR molecules33 by comparing the UV–vis spectra of star-PBA TPIR and network-PBA HABI (Supplementary Fig. 8). Time-dependent plots of the amount of TPIRs obtained using the network-PBA HABI concentration of 10 mg/ml demonstrated repetitive increases and decreases in the amount of TPIRs by turning the UV lamp on and off and most of these changes occurred within 5 s (Fig. 2d). Although a small amount of TPIRs remained after turning the UV lamp off for 80 s, ~95% of the TPIRs recoupled within 5 s. The production of TPIRs was essentially instantaneous irrespective of concentration, but two different series of TPIR recoupling reactions were evident from the decreases in A 601 (Fig. 2e). The existence of these two trends in the decrease of absorbance at A 601 is likely due to a difference in the reaction rate between inter- and intra-molecular recoupling of TPIRs. The 10 mg/ml solution showed the apparently higher rate for the first sudden decrease in A 601 compared with the other concentrations tested (Fig. 2e). A higher concentration in general allows faster intermolecular reaction and thus the first sudden decrease in A 601 likely corresponds to intermolecular recoupling of the TPIRs. The rate of the subsequent second decreasing trend in A 601 (2–10 min) is apparently slower than the first one yet is nonetheless remarkably faster than those of recoupling reactions of typical TPIRs (>1 h)33, 34. The coupling reaction of TPIRs in dilute solution is slow because the TPIR molecules are far apart. However, similar to the bridged imidazole dimers reported by Abe et al.35, the four TPIRs in the single polymer chain likely caused accelerated intramolecular recoupling of star-PBA TPIR s, resulting in a faster reaction rate than those of reported ones33, 34. Concentration-dependent GPC measurements were performed to test this hypothesis. Comparison of the GPC chromatograms of THF solutions containing different concentrations of network-PBA HABI prepared by irradiating UV to the corresponding dispersions showed products due to the intermolecular reaction at concentrations above 5 mg/ml, and these network-PBA HABI -derived fractions increased as the concentration increased (Fig. 3a). On the other hand, the peak molecular weights (M ps) shifted toward lower molecular weight region compared with star-PBA lophine , where the M p of star-PBA lophine is indicated by a dashed line in Fig. 3a. A reduction in hydrodynamic volume is a well-known effect of the formation of intramolecularly linked polymer products, for example, cyclic and eight-shaped polymers16. These results support our hypothesis regarding inter- and intra-molecular reactions and higher concentrations increased the ratio of network-PBA HABI (Fig. 3b).

Fig. 3
figure 3

Concentration-dependent repeatable MAT. a GPC traces of the products after UV irradiation to THF dispersions (1–100 mg/ml) of network-PBA HABI . b Schematic illustration of the dependence of the network density on concentration

Taken together, the results demonstrate that the desired IRLNC relying on repeatable MAT has been achieved and well-defined star polymers allowed detailed analysis of the mechanism underlying the present repeatable MAT. There are several reports describing polymers with lophine side chains31, 36 that characterize in detail their side-chain oxidation and UV responses; however, to achieve IRLNC with such polymers is difficult because of their inherent solidity. On the other hand, the molecular weight range of star-PBA Lophine and star-PBA TPIR compatible with maintaining the liquid state is limited, and the temperature required to demonstrate IRLNC with bulk network-PBA HABI was relatively high (vide ante). Wherein, we next focused on the IRLNC of PDMS. Four-armed star PDMSs with hydrosilane end groups (star-PDMS SiH ) were synthesized by organocatalytic ring-opening polymerization of hexamethylcyclotrisiloxane37 initiated from pentaerythritol and terminated with chlorodimethylsilane (Supplementary Fig. 1b). Subsequent hydrosilylation with 4-allyloxy benzaldehyde using Karstedt’s catalyst afforded star-shaped PDMSs with aldehyde end groups (PDMS Aldehyde s). Star-shaped PDMSs with lophine end groups (star-PDMS lophine s) were then synthesized by the reaction of star-PDMS Aldehyde with benzil using Debus–Radziszewski imidazole synthesis conditions; subsequent oxidation to generate TPIRs followed by their coupling reaction afforded the corresponding network polymer (network-PDMS HABI ) (Supplementary Fig. 1b). The star-shaped PDMSs were designed to have core pentaerythritol-derived branch points similar to those of star-shaped PBAs to minimize the effect of the branch points on porperties. Comparison of the GPC chromatograms of star-PDMS SiH (Supplementary Fig. 9a), star-PDMS Aldehyde (Supplementary Fig. 9b), star-PDMS Lophine (Supplementary Fig. 9c) confirmed that the above-described sequential reactions did not cause degradation reactions such as cleavage of the relatively weak Si–O–C linkages. Although the molecular weight distributions of these star-shaped PDMSs were broad in comparison with those of star-shaped PBAs (Supplementary Fig. 4), the targeted star-PDMS Lophine was well-characterized by 1H NMR (Supplementary Fig. 10a). Under the range of experimental conditions tested, the inherent liquid nature of PDMS and relatively long polymer chains prevented solidification of star-PDMS Lophine despite the terminal four lophines (Supplementary Table 2). The subsequent network formation by oxidizing the terminal lophines afforded nonliquid network-PDMS HABI . The 1H NMR spectrum of CDCl3 soluble part of network-PDMS HABI showed changes of signals at the aromatic region (Supplementary Fig. 10b) similar to those recognized for network-PBA HABI . The conversion from TPIRs to HABIs was also determined by comparing 1H NMR spectra of star-PDMS Lophine (Supplementary Fig. 11a) and acetone-d 6 soluble part of network-PDMS HABI (Supplementary Fig. 11b). The conversion yield was calculated to be ca. 70% from these spectra. The GPC chromatogram of the THF soluble part of network-PDMS HABI showed a shift of the trace toward a higher molecular weight region (Supplementary Fig. 9d) compared with that of star-PDMS Lophine (Supplementary Fig. 9c), demonstrating that the TPIRs in the generated star-PDMS TPIR reacted intermolecularly with each other. Importantly, no lower-molecular-weight fractions were observed in the GPC chromatogram of network-PDMS HABI , suggesting that the Si–O–C linkages near the junction points are stable under the reaction conditions tested and even under the existence of TPIRs.

We further attempted IRLNC based on repeatable MAT between network-PDMS HABI and star-PDMS TPIR (Fig. 4a) by UV irradiation to nonliquid network-PDMS HABI deposited on the surface of a vial (Fig. 4b). Significantly, room temperature (25 °C) IRLNC was achieved (Supplementary Movie 2). The network-PDMS HABI only at the area exposed to UV liquefied and flowed down (Fig. 4c). The slight blue coloration and liquefaction of the material suggested production of star-PDMS TPIR s. The liquid state due to star-PDMS TPIR was readily reverted by terminating UV irradiation and nonliquid network-PDMS HABI reformed on the glass surface (Fig. 4d). Remarkably, the present IRLNC completed within a few minutes (Supplementary Movie 2). Furthermore, we have succeeded in applying network-PDMS HABI to a photochemically metamorphosing material (Supplementary Movie 3). Thus, a plastic sheet with circle, triangle, and hexagonal shaped templates was fixed on a slide glass and the network-PDMS HABI was deposited on a surface of these templates (Fig. 4e). The nonliquid network-PDMS HABI on the templates liquefied upon UV irradiation and the liquefied material, that is, star-PDMS TPIR spread within the templates (Fig. 4f). After terminating UV irradiation and removing the plastic sheet with the templates, the reverted network-PDMS HABI s with circle, triangle, and hexagonal shapes were formed on the slide glass (Fig. 4g). This demonstrates that the present molecular system has a potential for the repeatable photochemical molding technology14 with the different mechanism from conventional ones.

Fig. 4
figure 4

IRLNC experiment with network-PDMS HABI . a Repeatable MAT between network-PDMS HABI and star-PDMS TPIR based on UV-triggered bond cleavage and reformation. Photographs of network-PDMS HABI on the vial b before and c during UV irradiation for 3 min 43 s and d immediately after terminating UV irradiation. Photographs of network-PDMS HABI on the templates e before and f during UV irradiation for 9 min 55 s, and g after terminating UV irradiation and removing the template plastic sheet

Rheological aspects of photo-triggered solvent-free IRLNC

Dynamic mechanical analysis (DMA) was performed to directly reveal the change of storage modulus (G′) and loss modulus (G″) in the present IRLNC system. Time-dependent plots of G′ (Fig. 5a) and G″ (Fig. 5b) of network-PDMS HABI demonstrated their repetitive decreases and increases by turning the UV lamp on and off (Fig. 5). The measurements were repeated three times (1st: black circle, 2nd: blue square, and 3rd: red triangle) and all of the measurements showed similar trends, demonstrating the reversibility of the present liquid–nonliquid conversions. More importantly, the temperature change of the material monitored during the measurements was negligible (Fig. 5, green line). DMA analysis of network-PBA HABI under irradiating UV at 50 °C likewise showed repetitive decreases and increases of both of G′ (Supplementary Fig. 12a) and G″ (Supplementary Fig. 12b), indicating that the present repeatable MAT-based IRLNC has a certain level of generality.

Fig. 5
figure 5

Rheological properties upon repetitive UV irradiations. Time-dependent plots of a G‘ and b G“ for network-PDMS HABI upon repeating UV irradiation cycles (1st cycle: black circle, 2nd cycle: blue square, and 3rd cycle: red triangle) with monitoring temperature (green line). The starting (ON) and terminating (OFF) times of UV irradiation were indicated with dashed lines

Meanwhile, DMA analysis of network-PDMS HABI showed higher G″ than G′ irrespective of UV irradiation, which is more typical for liquid materials. From this characteristic and the actual nonliquid appearance of network-PDMS HABI without UV irradiation (Supplementary Movie 2), two reasons can be considered for the present IRLNC. One is high-complex viscosity (η*) of network-PDMS HABI enough to avoid flow of the material. The other is the existence of crossovers with the frequency-dependent curves of G′ and G″. In fact, various polymer materials show frequency dependence of G′ and G″ and whether we should define the materials as liquid or nonliquid relies on a time scale for observing the phenomenon as recognized in the pitch drop experiment started from 1927 and still continues38. Moreover, Sheiko and coworkers reported unique dependence of G′ and G″ on frequency with solvent-free supersoft and superelastic bottlebrushes composed either of PBA or PDMS39, 40. Thus, we further examined frequency dependence of G′ and G″ with network-PDMS HABI . Based on frequency dependent DMA measurements for network-PDMS HABI , it was revealed that the plots of G″ were always greater than those of G′ irrespective of UV irradiation in the experiment range (Fig. 6a) and η*s at the corresponding frequencies were ~150 Pa s (Fig. 6b). Given that η*s of commercial linear PDMS melts were in the range of 1.0–1.5 Pa s at a similar frequency range41, the measured η*s for network-PDMS HABI were roughly hundredfold greater than those of such PDMS melts. The decreased amounts of G′ and G″ at the frequency around 3 Hz did not correspond to that appeared in the temperature-dependent measurement (Fig. 5) likely due to the difference in thickness of the samples. However, the frequency dependent DMA measurements also revealed that the decrease of G′ upon UV irradiation was more prominent than that of G″ (Fig. 6a).

Fig. 6
figure 6

Frequency dependence on rheological properties with and without UV irradiation. a Frequency-dependent plots of G′ and G″ for network-PDMS HABI (G′ without UV: blue triangle, G′ with UV: blue open triangle, G″ without UV: red diamond, G″ with UV: red open square). b Frequency-dependent plots of η* without (black circle) and with (black open circle) UV irradiation

Discussion

IRLNC of polymeric materials relying on repeatable MAT between star and network architectures was demonstrated. The key for the present repeatable MAT was cleavage and reformation of the imidazole–imidazole covalent bond in HABI, which enabled prompt IRLNC in response to UV. The mechanism for the present IRLNC was further examined by DMA studies. Generally, the existence of network structure is primarily responsible for exhibiting elasticity and network-PDMS HABI showed greater shift of G′ than G″ upon UV irradiation. Therefore, the decrease of η* or G′ by UV irradiation presumably caused the photo-triggered solvent-free IRLNC. Moreover, IRLNC methodology is, in principle, general as long as the parent star-shaped polymers are liquid. Therefore, the described repeatable MAT-based IRLNC and photo-triggered repeatable metamorphosis of polymers open the door to soft materials for use in various fields such as recyclable and deformable structural materials and adhesives, including human body-related applications.

Methods

Synthesis of network-PBAHABI

Into a flask was prepared an ethyl acetate (10 ml) solution of star-PBA Lophine (170 mg, 0.174 mmol for end groups), and an aqueous solution (10 ml) containing potassium ferricyanide (2.86 g, 8.7 mmol) and potassium hydroxide (488 mg, 8.7 mmol) was added. The mixture was stirred at room temperature for 60 min. The aqueous layer was separated and extracted with ethyl acetate. The combined ethyl acetate phases were washed with water, dried over MgSO4, and evaporated. After drying under reduced pressure, an elastic solid poorly soluble to organic solvents (network-PBA HABI ) was obtained. The yield was 142 mg. 1H NMR (500 MHz, CDCl3) δ p.p.m. 0.88 (s, –CH2CH 3), 1.08 (s, –C(CH 3)2–), 1.21–2.65 (m, –CH 2CH(COOnBu)–, –CH 2CH 2CH3), 4.05 (m, –COOCH 2–), 5.12 (m, triazole–CH2OAr), 5.38 (m, –CH 2CH(COOnBu)–triazole), 6.85–7.93 (–CH2OArH). M n (RI) = 4300, M p (RI) = 7200, M w/M n = 2.34.

Synthesis of network-PDMSHABI

Into a flask was prepared a hexane (10 ml) solution of star-PDMS Lophine (1.35 g, 0.45 mmol for end groups), and an aqueous solution (30 ml) containing potassium ferricyanide (7.41 g, 22.5 mmol) and potassium hydroxide (1.26 mg, 22.5 mmol) was added. The mixture was stirred at room temperature for 60 min. The aqueous layer was separated and extracted with hexane. The combined hexane phases were washed with water, dried over MgSO4, and evaporated. After drying under reduced pressure, an elastic solid poorly soluble to organic solvents (network-PDMS HABI ) was obtained. The yield was 1.32 g. 1H NMR (500 MHz, CDCl3) δ p.p.m. 0.03–0.53 (s, –CH 3), 0.61 (s, –CH 2CH2CH2OAr), 1.75 (s, –CH2CH 2CH2OAr), 3.45–3.89 (m, –CCH 2O–, –CH2CH2CH 2OAr), 7.00–8.22 (m, ArH). M n (RI) = 6100, M p (RI) = 11200, M w/M n = 2.34.

UV irradiation experiments

UV irradiation experiments were performed using a Hamamatsu LIGHTNING CURE LC8 L9588 model with a Hamamatsu A10014-35-0110 light guide. A Hamamatsu A9616-07 filter was used for efficiently transmitting only the UV light around 365 nm (150 mW/cm2). UV irradiation to the bulk network-PBA HABI was performed on the slide glass heated at 50 °C with hot plate. For concentration-dependent experiments, UV was directly irradiated to the quartz cells containing THF dispersions of network-PBA HABI (1, 5, 10, 50, and 100 mg/ml). Further, UV irradiation to the bulk network-PDMS HABI inside the vial was performed at 25 °C.

Rheological analyses

Rheological analyses under the irradiation of UV (365 nm, 150 mW/cm2) were performed on an Anton Paar MCR 302 rheometer equipped with a UV curing system and a Pertier temperature control device. Time-dependent analyses of storage modulus (G′) and loss modulus (G″) were conducted using a parallel plate with a diameter of 12 mm at the frequency of 3 Hz for the samples with the thickness of 0.5 mm. Temperatures of the parallel plate were set to 50 °C for network-PBA HABI and to 25 °C for network-PDMS HABI and were monitored during the measurements. Each time-dependent measurement was conducted for 30 min and then the sample was held sufficient time to reach equilibrium state before starting the following measurements. For frequency-dependent G′ and G″ measurements, the network-PDMS HABI sample with the thickness of 0.12 mm was employed.

Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.