*Article* **Hydrogel Beads of Amidoximated Starch and Chitosan as Efficient Sorbents for Inorganic and Organic Compounds**

**Diana Felicia Loghin, Melinda Maria Bazarghideanu, Silvia Vasiliu , Stefania Racovita , Marius-Mihai Zaharia , Tudor Vasiliu and Marcela Mihai \***

> Petru Poni Institute of Macromolecular Chemistry, Aleea Grigore Ghica Voda 41A, 700487 Iasi, Romania **\*** Correspondence: marcela.mihai@icmpp.ro

**Abstract:** The synthesis of hydrogel beads involving natural polymers is, nowadays, a leading research area. Among natural polymers, starch and chitosan represent two biomolecules with proof of efficiency and low economic impact in various utilization fields. Therefore, herein, the features of hydrogel beads obtained from chitosan and three sorts of starch (potato, wheat and rise starches), grafted with acrylonitrile and then amidoximated, were deeply investigated for their use as sorbents for heavy metal ions and dyes. The hydrogel beads were prepared by ionotropic gelation/covalent cross-linking of chitosan and functionalized starches. The chemical structure of the hydrogel beads was analyzed by FT-IR spectroscopy; their morphology was revealed by optical and scanning electron microscopies, while the influence of the starch functionalization strategies on the crystallinity changes was evaluated by X-ray diffraction. Molecular dynamics simulations were used to reveal the influence of the grafting reactions and grafted structure on the starch conformation in solution and their interactions with chitosan. The sorption capacity of the hydrogel beads was tested in batch experiments, as a function of the beads' features (synthesis protocol, starch sort) and simulated polluted water, which included heavy metal ions (Cu2+, Co2+, Ni2+ and Zn2+) and small organic molecules (Direct Blue 15 and Congo red).

**Keywords:** grafted starch; ionotropic gelation; covalent cross-linking; molecular dynamics simulation; sorption capacity

#### **1. Introduction**

In the last few years, extensive research has been undertaken to obtain specialized and selective sorbents containing natural polymers as a cheap and environmentally friendly solution for water cleaning. In this respect, polysaccharide-based hydrogels were studied, taking advantage of their low cost, availability, non-toxicity and biodegradability [1,2]. Among the polysaccharides, starch [3–8] and chitosan (CS) [9–15] have attracted the attention of the scientific community due to their physico-chemical characteristics, chemical stability, and excellent selectivity resulting from the presence of chemical reactive groups (hydroxyl, acetamido or amino functions) in polymeric chains. Moreover, these products are abundant, renewable, and biodegradable, and are able to physically and chemically bind to a wide range of molecules [16–22].

A number of studies have shown that polymers containing amidoxime groups have high complex-forming capabilities with metal ions and can be successfully used in metal ion removal from aqueous solutions [23–30]. Amidoxime hydrogel beads of modified alginate and amidoximated synthetic polymers have been successfully synthesized and used as sorbents for dyes, showing selective adsorption towards cationic dyes in the presence of anionic/cationic mixed dyes [31]. Usually, the synthesis of a sorbent with amidoxime groups involves the incorporation of a nitrile group into a polymer matrix, followed by the conversion of the nitrile group into an amidoxime group by treatment with an alkaline solution of hydroxylamine. For instance, sorbents containing amidoxime

**Citation:** Loghin, D.F.; Bazarghideanu, M.M.; Vasiliu, S.; Racovita, S.; Zaharia, M.-M.; Vasiliu, T.; Mihai, M. Hydrogel Beads of Amidoximated Starch and Chitosan as Efficient Sorbents for Inorganic and Organic Compounds. *Gels* **2022**, *8*, 549. https://doi.org/10.3390/ gels8090549

Academic Editor: Hiroyuki Takeno

Received: 12 August 2022 Accepted: 25 August 2022 Published: 30 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

groups have been obtained by reacting acrylonitrile-divinyl benzene copolymer beads with hydroxylamine [32]. The introduction of amidoxime groups into acrylonitrile-grafted cellulose by interaction with hydroxylamine has also been investigated [33].

Starch is an interesting bio-material, due to its abundance and low cost, but has poor mechanical properties and the fact that it is highly hydrophilic. To overcome these drawbacks, chemical modification of starch is usually applied, mainly by grafting reactions. A large range of polymers can be grafted on starch by ring-opening and radical polymerizations of various monomers in order to modulate the properties of the final products [34]. For example, poly(amidoxime) ion exchange resins were synthesized from polacrylonitrile grafted sago starch, with their batch binding capacity for different metal ions being pH dependent [27,35]. In our previous studies, potato starch (PS) was grafted with acrylonitrile (AN) by the redox initiation by Ce4+ ions [36] and then the amidoximated (Ax) derivative was obtained [37]. Ionic composites based on crosslinked CS and amidoximated potato starch were also obtained and used as super-sorbents for copper ions, with reusability up to five sorption/desorption cycles, with no significant decrease in their sorption capacity [36]. Furthermore, the acrylonitryle grafted reaction, with the same Ce4+ ions as the initiator, was tested on three sorts of starch, namely PS, wheat (WS) and rice (RS) starches, followed by obtaining soluble derivatives by hydrolysis [5] The studies showed that the amylose/amylopectin content in starch and the grain size influenced the grafting performance, which reached 13.81%, 9.71% and 8.22% for PS, WS and RS, respectively [5].

Therefore, herein, the features of three sorts of starch (PS, WS and RS) grafted with acrylonitrile (PSgAN, WSgAN and RSgAN) and then amidoximated (PSgAx, WSgAx and RSgAx) were deeply investigated, following the formation of composite hydrogel beads with CS. The current study starts with the premise that the use of starch from different botanical sources in grafting reactions can influence the properties of the obtained materials, and consequently their properties. Thus, we aimed to prepare hydrogel beads by ionotropic gelation and covalent cross-linking of CS and functionalized starches as SgAN or SgAx. The chemical structure of functionalized starches and the obtained hydrogel beads was followed by FT-IR spectroscopy; their morphology was revealed by optical and scanning electron microscopy, while the samples' crystallinity changes in relation to the functionalization strategies was evaluated by X-ray diffraction. Molecular dynamics simulations were used to determine the influence of the grafting reactions and grafted structure on the starch conformation in solution and their interactions with chitosan. The sorption capacity of the beads for Cu2+, Co2+, Ni2+ and Zn2+ ions and for Direct Blue 15 and Congo red dyes was followed in batch experiments, as a function of the beads' features (synthesis protocol, starch sort), and contact time.

#### **2. Results and Discussion**

#### *2.1. Starch Functionalization*

Previous studies showed that the grafting reaction, in the current applied conditions, took place on the amylose part of starch [36] and the amylose contents of the used starch were as follows: 20–21% in PS, 23–30% in WS and 17–30% in RS [5]. Herein, the three types of starch were functionalized by grafting acrylonitrile to the amylose component of starch (SgAN samples), followed by amidoxime transformation (SgAx samples) (see details in Section 4.2). SEM images (Figure 1(aA)) revealed that starch granules have different morphologies. PS has a spherical, elliptical or irregular shape, whereas WS is predominantly spherical shaped, both with a smooth surface; RS particles show polyhedrontype shapes with sharp edges. Figure 1(bA,cA) show that PS and WS granules after AN grafting (SgAN), and also after amidoximation (SgAx), had irregular shapes and sizes, most of them being smaller than the initial granules, whereas RS-based samples show small changes in morphology, mainly with the loss of sharp edges.

**Figure 1.** SEM images (scale bar 20 μm) (**A**), circular equivalent (CE) diameter (**B**) and X-ray diffractograms (**C**) of initial (**a**) and grafted AN (**b**) and Ax (**c**) starches.

The particle size analysis of native starches (Figure 1(aB)) shows that PS has the largest granule size (mean diameter, Dm = 15.58 μm), with WS and RS being characterized by lower particle sizes (Dm = 10.1 μm and 8.7 μm) and similar particle size distribution (with a shoulder for larger sizes suggesting particle aggregation), which was almost unimodal for each starch type. After AN grafting, the size distribution (Figure 1(bB)) undergoes SEM observation (Figure 1(bA)), and shows that the highest grain fragmentation is observed for the PSgAN and WSgAN, with smaller particle aggregation being found for RSgAN. After amidoximation, the samples show a population of particles with a mean diameter of about 3 μm (Figure 1(cB)) and the aggregation is characteristic also for the RS grains. Moreover, the size and morphology modifications observed in Figure 1A,B can be explained by the selectivity of the grafting and amidoximation reactions to the amylose part of the starch grains and the amylopectin part was partially removed during the starch grain functionalization process [36]. Thus, during the gelatinization process, which took place in the first reaction step, the amorphous amylose part of the grains becomes available to the Ce4+ ions, which interacts with the -OH groups located at the C2 and C3 carbon atoms.

The powder XRD diffraction analyses of PS, WS and RS and grafted starch are presented in Figure 1C. Generally, the X-ray diffraction results of starch beads are classified as A, B or C type, and depend upon the double-helical amylopectin chain arrangement. The A type pattern is a result of close-packed arrangements with a water molecule connection between the double helix, the B type is open hexagonal packing with water in the central cavity and type C is quite similar to the A pattern, except for the appearance of the peak around 5◦ [38]. The X-ray diffractogram of PS revealed a typical B-type pattern [39], with a strong reflection peak (100) at around 17◦, relatively low intensity peaks at around 5◦ and 22◦ and shoulders around 15◦ and 24◦ (2θ). The diffractograms of WS and RS were practically identical, with the strongest peaks at approximatively 15◦, 17◦ and 23◦ (2θ). The A-type pattern of WS and RS can be proved by the appearance of the shoulder at around 18◦ (2θ), a signal that is characteristic to this starch type. After being grafted with

AN (Figure 1(bC)), the characteristic diffraction peaks of native starch disappeared in the XRD patterns with the appearance of a new intense peak around 17◦ (2θ), which can be ascribed to the structure modification following the AN grafting. As compared to native and AN-grafted starch, the XRD spectra of SgAx displayed a typical V-type crystalline structure [40] with a wide peak at around 20◦ (2θ). This wide peak can be attributed to the hydroxylamine functionalization reactions, which lead to the partially destroyed crystallinity of starch. The XRD results also confirmed our previous studies [36], where it was shown by 1H-NMR studies that the grafting reaction takes place mainly at the amylose component of starch, the amorphous component of starch molecules.

Molecular dynamics (MD) simulations were performed in order to visualize the conformation modifications of the starch (amylose) backbone after AN grafting reaction and its further amidoxime functionalization. The different types of patterns observed in starch crystallization are determined mainly by the interaction between two starch molecules. This is why the three simulated systems contained two identical starch molecules (amylose, SgAN and SgAx) solvated in water (see details in Section 4.5). Figure 2 depicts the initial conformation of the system and the final structure obtained after 200 ns of simulation.

**Figure 2.** Snapshots depicting the starting and final structure of the 3 simulated systems: (**A**) amylose from starch, (**B**) SgAN and (**C**) SgAx. The starch molecules are colored in red and blue, the AN side chains are colored in green and the Ax side chains are colored in silver. Water has been omitted for clarity.

The amylose starch molecules start with separated conformation, with a distance between the two molecules of >20 Å, to eliminate any bias in the interaction that could take place. After 200 ns, it can be observed in Figure 2A,B that the unfunctionalized and the SgAN molecules are interacting with each other, while the SgAx molecules remain separate. The unfunctionalized starch molecules are wrapped around each other in a way that is similar to a double-helical structure (Figure 2A). Similar observations were found in the literature [41], which also showed that this association could have two main crystal forms, A and B, as already observed for our starch samples in Figure 1C. The AN grafted starch behaves in a similar manner to the previous system, with the main starch chains wrapped around each other and the side chains exposed to the exterior (Figure 2B). The Ax functionalized starch behaves differently to the other investigated starch, in that the molecules remain isolated for the entire simulation (Figure 2C). The structures obtained in the MD simulation corroborate well with the XRD result (Figure 1C). In detail, in the case of the unfunctionalized starch molecules, the amylopectin crystalline part of the starch, and also the amylose part, can interact with each other and assemble in structures that are precursors to the A or B type patterns. In the case of the SgAN, although the amylose main chains interact with each other in a similar way to the previous system, the side chains located at the exterior and the absence of amylopectin leads to a decrease in crystallinity. Thus, the reflection peak (100) at around 17◦ observed in all SgAN samples (Figure 1(bC)) can be ascribed to the remaining organization of the starch main chains while the AN side chains determine a loose structure. In the case of SgAx, the MD simulation clearly depicts the V-type pattern of single starch molecules in an extended conformation.

#### *2.2. Hydrogel Beads of Functionalized Starch and Chitosan*

The composite hydrogel beads formation follows two routes, which include using SgAN or SgAx as precursors and obtaining three types of composite beads, CS/SgAN, Cs/SgAN-Ax and Cs/SgAx (see details in Section 4.3). The beads are sphere shaped, as observed in the optical images of wet beads, and are porous, as shown by the SEM images of the surface of lyophilized beads (Figure 3).

**Figure 3.** Optical (scale bar 5 mm) and SEM (scale bar 20 μm) images of composite beads.

The optical images in Figure 3 show some color changes as a function of the composite bead synthesis pathway: the beads obtained with SgAx keep the yellow color of amidoximated starches, whereas after post-amidoximation translucid beads were observed, with a slight yellow tinge. In addition, the beads' size in swelled form is not influenced by the synthesis pathway, as is the surface morphology observed in the SEM images. Thus, CS/SgAN beads have uniform pore distribution at the beads' surface, with thin walls. After bead amidoximation (CS/SgAN-Ax samples), the porosity slowly decreased and the pore walls becomes thicker. A different morphology of the composite beads was obtained when SgAx was employed in the bead formation process, i.e., a smaller size of the pores was obtained for all the investigated samples, most probably since the -NH2 groups in SgAx could also be crosslinked by epichlorohydrin (ECH), as is the case for the similar groups in CS; thus, a double crosslinked network is formed.

The samples' porosity in the swollen state was estimated by measuring 50 pores in each SEM image from Figure 3; the resulted mean values of area, perimeter, aspect ratio (ratio of major/minor axis of pores) and Feret diameter (the longest distance between any two points along the selection boundary) are included in Table 1.

**Table 1.** Pore size analysis as area, perimeter, Feret diameter and aspect ratio (mean values ± standard deviation).


Thus, the obtained values of the aspect ratio are higher than 1.00 (circles), meaning that the pores have elongated shapes. In addition, the Feret diameter values showed that the larger (the higher diameter) and the more irregular shaped (with higher aspect ratio values) pores were obtained when WS was used in hydrogel bead synthesis and were smaller with RS. The same trend was observed after bead amidoximation (SgAN-Ax samples). Larger amylose contents in RS (up to 30%) favor better grafting reactions and resulted in its better embedment into the hydrogel beads, leading to smaller pore sizes. PS and WS, with almost the same amylose content (up to 20 and 23–30%, respectively) formed hydrogel beads with larger pores. The amidoximated starch directly used in bead formation led to smaller size pores, as the starch had a very small influence on their porosity. In this synthesis route, the SgAx participation to crosslinking along with CS, and also the elongated form of SgAx both contributed to obtaining homogeneous beads with very close pore sizes, irrespective of the starch source. This observation is also supported by the MD simulations of the Ax-grafted starch that showed the side chains extended in the solvent, which were completely free to interact with ECH (Figure 2). To further support this observation, MD simulations of two systems containing two starch molecules of SgAN or SgAx and one CS molecule were performed (Figure 4).

The MD simulations showed that CS is able to interact with the SgAN molecules, coming into contact with both the AN side chains and the amylose backbone (detail of Figure 4A), while in the case of SgAx molecules, no contact between the CS an SgAx can be observed for the entire simulation (Figure 4B). This is due to the fact that both molecule types have a positive charge and repel each other.

The hydrogel beads' elemental composition was followed by EDAX analysis and Table 2 shows the obtained results both at their surface and in the sections. The values were compared with the values calculated taking into account the ratio between the grafted starch in the synthesis process and CS (1:4) and ECH (1:0.37), considering, for functionalized starches, the mean grafting of three units to each monomer unit.

**Figure 4.** Snapshots depicting the starting and final structure of the simulated systems containing CS and (**A**) SgAN and (**B**) SgAx. The starch molecules are colored in red and blue, the AN side chains are colored in green, the Ax side chains are colored in silver and the chitosan is colored in orange. Water has been omitted for clarity.


**Table 2.** The C/N and C/O atomic ratios calculated and determined by EDAX for composite beads.

Thus, the C/N atomic ratios are higher than the calculated values both at the surface and inside the beads. The lower N content in the beads can be ascribed to a synergy of factors, mainly because of the lower contribution of the AN-grafted side chains in starch and/or a higher contribution of the CS chains inside the beads. By considering the small length of the grafting chains, we may assume that the grafted chains can be hindered by the macromolecule conformation and, as observed in the MD simulation, with the SgAx behaving as elongated macromolecules, with easily accessible functional groups. This is also related to the contribution of CS to the formation of the beads, with the partial crosslinking of SgAx decreasing the amount of ECH available to crosslink the CS chains and the free chains being most probably partially removed during the final washing steps. The above-mentioned observations are sustained by the C/O atomic ratios, which are higher than the calculated ones. The comparison of the C/N and C/O atomic ratios at the surface and inside the beads revealed that there are some differences in the elemental composition and that inside the beads, the atomic ratio values are smaller than outside but are the closest to the calculated values.

Further proof of the samples' chemical structure was obtained by FTIR and is shown in Figures S1–S3 (supplementary information), where the characteristic peak of AN is observed just in the CS/SgAN samples and the characteristic peaks for starch and CS are found in all the spectra. Nevertheless, as shown in previous studies [36], the formation of amidoxime can be demonstrated in FTIR at about 1650 cm−<sup>1</sup> or 933 cm<sup>−</sup>1, assigned to the C=N and N-O bonds in oximes, respectively. Therefore, the 1800–1500 cm−<sup>1</sup> area was selected and the peak deconvolution was performed for both types of amidoximated beads, CS/SgAN-Ax and CS/SgAX (Figure 5, Table S1).

**Figure 5.** Deconvolution of 1800–1500 cm−<sup>1</sup> region of FTIR spectra of CS/amidoximated starch beads obtained by the two procedures.

Thus, the region 1800–1500 cm−<sup>1</sup> of the three spectra of the CS/SgAx samples is deconvoluted in four individual peaks, irrespective of starch nature, which can be ascribed as follows: 1733–1720 cm−<sup>1</sup> to carbonyl groups C=O, resulting from the grafting reaction and anhydro glucose ring opening, 1654–1650 cm−<sup>1</sup> to C=N in the oxime groups (the highest contribution), 1591–1595 cm−<sup>1</sup> to -NH2 groups on both CS and modified starch (with the smaller contribution) and 1557–1559 cm−<sup>1</sup> to vibrations of the protonated amine group. The same region in the FTIR spectra of the composite beads obtained by the amidoximation of the already formed CS/SgAN beads (namely CS/SgAN-Ax samples) demonstrated five individual peaks, with the peak ascribed to C=N in the oxime groups being shifted to about 1660 cm−<sup>1</sup> and the new peak at about 1630 cm−<sup>1</sup> being ascribed to the -OH groups bending. Furthermore, for these samples, the contribution of the peak at 1660 cm−<sup>1</sup> diminished and the contribution of that at about 1597 cm−<sup>1</sup> increased, suggesting a higher amount of free amino groups obtained by the post bead formation amidoximation reaction groups, which are evidently not involved in the crosslinking reactions with ECH.

The modifications in the starch/CS crystallinity after hydrogel bead synthesis was followed by X-ray diffraction (Figure 6).

**Figure 6.** X-ray diffractograms of CS/SgAx, SgAN and SgAN-AX composite beads as compared to bare CS.

The chitosan sample shows diffraction peaks at about 9◦ and 20◦ (2θ), which could be found in all the samples after hydrogel bead formation, as shown in Figure 6. In addition, the diffraction peak at 17◦ (2θ), characteristic for grafted starch (Figure 1(bC,cC)), was observed in the CS/SgAN samples and was also found after their amidoximation for beads obtained with PSgAN. For the samples obtained with the other AN-grafted starch (RS and WS) and that with SgAx, the peak at 17◦ (2θ) was not observed, with the peak at about 9◦ being evident mainly in CS/RSgAx beads. To conclude, the X-ray diffraction analysis showed that the samples prepared by the two methods, either using SgAN and beads post amidoximation or by using SgAx, still contain the functionalized starch structures, suggesting good incorporation of functionalized starch dispersion into the beads.

#### *2.3. Swelling Behavior of Hydrogel Beads*

The swelling behavior represents an important characteristic of a sorbent in its capacity to retain different ions or small molecules. Therefore, the swelling degree was evaluated as a function of time and pH (Figure 7). As shown in Figure 7a, in distilled water with a pH of about 6, the swelling equilibrium was reached in almost five hours for all the beads, irrespective of the sort of starch or the method applied in bead synthesis. Furthermore, there were almost no differences between the swelling degree values for the beads prepared with SgAx compared to those obtained with SgAN and those amidoximated. There are small differences between the swelling of the samples prepared with different sorts of starch, which can be ascribed to the differences in the starch grain sizes (Figure 1), with the largest PS grains resulting in a larger swelling capacity, as compared to the other starch sorts. Nevertheless, the swelling is also connected to the amylose/amylopectin ratio, which is almost 20/80 in PS, 23/77 in WS and 30/70 in RS [5]. Thus, for the lower grain size of WS and RS, the larger amylose content in RS results in a larger swelling capacity compared to that of the WS-based hydrogel beads.

**Figure 7.** Swelling of the composite hydrogel beads as a function of time at pH = 6 (**a**) and different pH values (**b**); inset in (**b**) potentiometric titration curves of SgAx and CS.

Almost similar and constant values were obtained when the pH of the swelling media varied between 4 and 8 (Figure 7b), irrespective of the sort of the starch and the bead preparation procedure. As shown in the inset in Figure 7b, the point of zero charges of CS is located at a pH of about 6.5; below this pH value, in the presence of hydronium ions, the primary amino groups (–NH2) of chitosan can be protonated (–NH3 +). The amidoximated starches show different points of zero charges, for example, at 4.6, 5 and 5.7 for RSgAx, WSgAx and PSgAx, respectively. Before these values, the primary amino groups of amidoximated starch and CS are protonated; therefore, below pH 4, the interpolymeric electrostatic repulsions in the composite beads and also the intramolecular repulsions between the ionized groups lead to an increase in the bead size; consequently, the equilibrium swelling ratio values increase with the pH decrease (swell between 70 and 90% at pH 2). On the contrary, at pH values up to the CS isoelectric point (6.5), ionization of the -OH groups in the amidoxime functional groups occurs, which could be associated with the protonated amino groups, decreasing the beads' swelling capacity. Furthermore, the swelled beads are stable across the whole range of the tested pHs, suggesting that the crosslinking process was effective.

#### *2.4. Sorption of Metal Ions by Hydrogel Beads*

Composite hydrogel beads with amidoximated groups (CS/SgAN-Ax and CS/SgAx) were tested as supports for the uptake of Cu2+, Co2+, Ni2+ and Zn2+ ions from aqueous solutions, in batch experiments. The influence of contact time on the hydrogel beads' sorption capacity for metal ions is represented graphically in Figure 8, and shows that the time required to reach the equilibrium was about four hours for all the studied metal ions. For each metal ion, the hydrogel beads' sorption capacity was influenced by the nature of the starch used to obtain the beads and the manner in which the beads were obtained. Nevertheless, the best sorption capacity for the investigated ions was obtained when CS/PSgAx hydrogel beads were used, with the highest affinity to Cu2+. Furthermore, with some exceptions, the beads obtained with SgAx show better sorption capacities for the investigated ions, as compared to the beads post-amidoximation, following the same trend as that found for the beads' swelling vs. time (Figure 7a). For each sorbent, the beads' affinity for the tested metal ions follows the same order, which is as follows: Cu2+ > Ni2+ > Zn2+ ≈ Co2+.

**Figure 8.** Sorption of Cu2+, Co2+, Ni2+ and Zn2+ onto composite beads based on PS (square), WS (circle) and RS (triangle) and using hydrogel beads CS/SgAx (close symbols) or CS/SgAN-Ax (open symbols).

The sorption capacity is usually influenced by the ions' properties, such as ionic radii, hydrated radius, atomic weight, electronegativity, and others, as already observed in other studies [42–45]. In this study, the ionic radius (Pauling) (Co 0.745 Å, Zn 0.74 Å, Cu 0.73 Å, and Ni 0.69 Å) and hydrated ionic radius (Zn 4.30 Å, Co 4.23 Å, Cu 4.19 Å, and Ni 4.04 Å) [46] represent the influence parameters. Furthermore, as shown in a previous study [47], the ions' coordination with nitrogen ligands is usually arranged in preferential structures, with the octahedral structure being preferred by Co2+ and Ni2+, tetrahedral by Zn2+ and square planar by Cu2+. Nevertheless, the found affinity for ion sorption in noncompetitive conditions towards the amidoximated starch-based beads of Cu2+ > Ni2+ > Zn2+ ≈ Co2+ suggests that most probably, sorption is favored by the lower ionic radius and by the structural arrangements that allow for the smallest energy consumption, with copper ions best fulfilling these conditions.

Another important parameter is the initial concentration, which herein was set to 300 mg/L and corresponds to different molar concentrations, i.e., Zn 4.587 mmol/L, Co 5.093 mmol/L, Cu 4.724 mmol/L, and Ni 5.11 mmol/L. Thus, the molar sorption capacity of the gel beads was calculated as the moles of metal ions sorbed after 360 min per moles of active sites in the hydrogel (Figure 9). Each amylose unit is grafted with a mean of three amidoxime groups and each group has three active sites (Figure 9a) and each chitosan deacetylated unit has one active site (primary amino group) to interact with the metal ions. Thus, taking into account the amount of each component used in hydrogel bead preparation (see Section 4.3) and the number of active sites of each component, the calculated amount of active sites to interact with metal ions per gram of gels was found to be 3.55 mmol/g. As shown in Figure 9b, PSgAx-based hydrogel beads have the highest sorption capacity for all the investigated ions and their higher swelling degree most probably allowed the ions to easily reach the active sites. The similar beads obtained with RS and WS followed the same trend as the swelling degree, assuming a sorption process controlled by diffusion. Nevertheless, even if CS/PSgAN-Ax beads have similar swelling capacity to that of the CS/PSgAx beads (Figure 7), their sorption capacity is lower, irrespective of the sorbed metal ion, suggesting lower active sites are available for the coordination of the metal ions. The hydrogel beads prepared with the other two sorts of starch (RS and WS) show different

behavior for each tested ion and as a function of the bead preparation procedure (using SgAx or post-amidoximated beads). Therefore, the complex sorption process as a function of the beads' features, as well as the metal ions' characteristics, should be carefully and deeper investigated.

**Figure 9.** (**a**) Schematical representation of the three active sites (colored) of amidoxime functional group and (**b**) sorption capacity of hydrogel beads CS/SgAN-Ax or CS/SgAx expressed as moles of metal (Me)/moles of active sites (AS) in hydrogel beads.

#### *2.5. Sorption of Dyes*

Organic dyes are among the major concerns of water pollutants. Congo red (CR) and Direct Blue-15 (DB15) are organic dyes that can be easily dissolved in water, causing difficulties in their removal from contaminated water. Furthermore, they are toxic and carcinogenic, causing various diseases. Both dyes are secondary diazo dyes, with complex aromatic structures that make them non-biodegradable and quite stable. Moreover, they contain anionic sulphonic groups that can electrostatically interact with the protonated amino groups in the composite hydrogel beads. The sorption capacity of the hydrogel beads with amidoxime groups for both CR and DB15 is represented in Figure 10. The time required for dye adsorption was about six hours, as can be observed in the Figure 10, with higher values being obtained when CR was sorbed as compared to DB15. These differences are influenced by several factors, such as the origin of the starch, the bead synthesis method and the dyes' characteristics.

**Figure 10.** Sorption of dyes onto composite beads based on PS (square), WS (circle) and RS (triangle) and using CS/SgAx (close symbols) or CS/SgAN-Ax (open symbols) hydrogel beads.

Thus, PS-based hydrogel beads show the best affinity to CR, whereas for DB1, better results were obtained with RS-based hydrogel beads and lower values were found when hydrogel beads with WS functionalized starch were used, irrespective of the dye sorbed. Furthermore, better results were obtained with CS/SgAx than with CS/SgAN-Ax. This correlates with the chemical structure of the hydrogel beads and the amino groups of both CS and amidoximated starch could interact, in specific conditions, with the negatively

charged dye molecules. The influence of the presence of ionic groups on the hydrogel matrix and their interaction with ionizable groups of dyes was also observed in other studies [48] The size of the dye molecules could also influence the hydrogels' sorption capacities, with CR having the molar mass of 696.665 g/mol and DB15 of 992.80 g/mol, and is in direct relation to the hydrogel beads' porosity (Figure 3). Thus, CS/PSgAx and CS/RSgAx show higher pore sizes as compared to CS/WSgAx; therefore, the smaller CR molecules could be sorbed inside them, whereas the sorption of DB15 with larger molecules was hindered.

#### **3. Conclusions**

In this study, the features of three sorts of starch (PS, WS and RS) grafted with acrylonitrile and then amidoximated were deeply investigated, following the formation of composite hydrogel beads with CS, and tested as sorbents for four heavy metal ions (Cu2+, Co2+, Ni2+ and Zn2+) and two dyes (DB15 and CR). The MD simulations show that the AN-functionalized starch behaves in a similar manner to the native one, as the main starch chains wrapped around each other and the side chains were exposed to the exterior, whereas the Ax-functionalized starch molecules remained isolated for the entire simulation. The structures obtained in the MD simulation corroborate well with the XRD result; in the case of the SgAN, the side chains located at the exterior of the wrapped arrangement and the absence of amylopectin leads to a decrease in crystallinity, whereas for SgAx, the V-type pattern of single starch molecules in a coiled conformation is found. The comparison of the C/N and C/O atomic ratios at the surface and inside the beads revealed that there are some differences in the elemental composition and that inside the beads, the atomic ratios values are smaller than outside, but are the closest to the calculated values. FTIR spectra of the CS/SgAx samples were deconvoluted in four individual peaks, irrespective of starch nature, which can be ascribed to both components' functional groups, since the CS/SgAN beads post-amidoximation shows five individual peaks. The new peak and also the variation in the others' intensity suggested a high amount of free amino groups obtained by the post bead formation amidoximation reaction, groups which are obviously not involved in any crosslinked reactions with ECH. The hydrogel beads show good sorption capacities for Cu2+, Co2+, Ni2+ and Zn2+ ions and for Direct Blue 15 and Congo red dyes, with their performances being influenced by the synthesis protocol and starch sort. Future studies must continue with other sorption-related experiments as a function of different parameters (pH, concentration, temperature) to elucidate the sorption mechanism for both inorganic and organic molecules.

#### **4. Materials and Methods**

#### *4.1. Materials*

CS powder, from Sigma-Aldrich, was used as received. The degree of acetylation of 15% and the average molar mass of 385 kDa were determined by the methods previously reported [49]. PS (humidity content < 10%, ash < 0.5%) and WS (humidity content < 15%, ash < 0.5%) were purchased from Fluka and were used as received. RS, epichlorohydrin (ECH), sodium hydroxide, methanol p.a., hydroxylamine chlorohydrate, metal ion salts (CoSO4·7H2O; NiSO4·6H2O; ZnSO4·7H2O and CuSO4·5H2O), Direct Blue 15 (DB15) and Congo red (CR) were purchased from Sigma-Aldrich and were used as received. Acrylonitrile was distilled at about 77 ◦C and kept at a low temperature.

#### *4.2. Starch Functionalization*

The starches grafted with acrylonitrile (SgAN), obtained using potato (PSgAN), wheat (WSgAN), or rise (WSgAN) starch, were synthesized and characterized as described in [5], using Ce(SO4)2 as the initiator in H2SO4 0.4 M, at 27 ◦C, under stirring for 1 h, and then separated in methanol. The polyamidoxime-grafted starch (SgAx) samples were obtained by analogous reactions of nitrile groups of SgAN, using hydroxylamine in an alkaline medium [36]. SgAN was introduced in a two-necked flask, which was equipped with a

stirrer and condenser placed in a thermostatic water bath. Then, the hydroxylamine solution was added to the flask, and the reaction was carried out under stirring for 5 h at 70 ◦C, and then for 24 h at room temperature without stirring. After completion of the reaction, the SgAx was separated from the solution by filtration and washed intensively with methanol:water (80:20, *v*/*v*) solution. The same procedure was applied for all the types of starch. The AN average grafted length of three AN/starch structural units was determined from the 1H-NMR spectra of SgAN [5] as the ratio of the integral AN characteristic peak at 3.1 ppm (ascribed to protons from CH groups) and starch characteristic peak located at 3.65 ppm (attributed to hydrogen at C2, C3 and C5 atoms).

#### *4.3. Hydrogel Bead Synthesis*

The composite beads were obtained by two similar methods that have already been published [36], using CS and amidoxime or acrylonitrile-grafted potato, wheat and rise starches, obtaining CS/SgAN and CS/SgAx beads. Shortly after, 0.5 g of SgAx or SgAN were first dispersed in 100 mL solution of CS 2% (*w*/*v*), and then 2 mL ECH was added as the crosslinker. The obtained mixture was dripped (through a syringe pump, ISPLab02) into an aqueous solution of sodium tripolyphosphate of 0.05 M, at room temperature, under gentle stirring (100 rpm). After 4 h, the formed beads were separated and transferred to an aqueous solution of 400 mL 0.1 M NaOH for 2 h, under slow stirring.

The beads prepared using SgAN were subjected to amidoximation by analogous reactions, obtaining CS/SgAN-Ax beads; the CS/SgAN beads and 70 mL hydroxylamine solution were introduced into a two-necked flask equipped with stirrer and condenser, and the reaction was carried out under mild stirring for 5 h at 70 ◦C. Finally, the prepared microspheres (obtained by both methods) were washed with distilled water at a neutral pH, and dried at 104 ◦C (Moisture Analyzers Precisa XM 60-HR, Precisa Gravimetrics AG, Dietikon, Switzerland).

#### *4.4. Characterization Methods*

Information on the surface morphology was evaluated using the Various G4 UC scanning electron microscope (Thermo Scientific, Brno-Cernovice, Czech Republic), whereas ˇ the elemental composition with an energy dispersive X-ray spectroscopy analyzer (Octane Elect Super SDD detector, Ametek, Mahwah, NJ, USA) were analyzed by SEM. The investigations were carried out on samples sputtered with 10 nm platinum (Leica EM ACE200 Sputter, Leica, Wetzlar, Germany) in a high vacuum mode, using secondary electrons (Everhart-Thornley detector, FEI Company, Brno, Czech Republic) and concentric backscattered detectors. The samples' porosity in the swollen state was estimated by selecting 50 pores in each SEM image using ImageJ 1.48v analyzing software (LOCI, University of Wisconsin-Madison, Madison, WI, USA) [50], measuring the area, perimeter, aspect ratio and Feret diameter of each one.

Optical images were obtained with a Nikon D3300 HDSLR camera, with an AF-P DX NIKKOR 18–55 mm f/3.5–5.6 G VR lens.

FTIR analysis of the grafted samples was performed using a Bruker FT-IR spectrometer. Each spectrum was scanned in the range 400–4000 cm<sup>−</sup>1, using 45 scans with a resolution of 4 cm−<sup>1</sup> by the KBr pellet technique, using 2 mg of each sample. The spectra deconvolution was carried out using the Opus 4.7v software (Universität Stuttgart, Stuttgart, Germany).

The native starch, grafted starched (with AN and Ax) and CS-based composite beads were examined by X-ray diffraction by a Rigaku Miniflex 600 diffractometer (Rigaku, Tokyo, Japan), using CuKα-emissions in the angular range 3–60◦ (2θ), with a scanning step of 0.01◦ and a recording rate of 2◦/min.

The diameter (circular equivalent) of native and AN and Ax-grafted starched was determined by a Morphologi G3SE particle characterization system (Malvern Instruments, Malvern, UK). The samples were carefully spread out over a glass plate. To evaluate the hydrogel beads' diameter, only full shaped particles, which were non-aggregated, were measured.

Potentiometric titrations were carried out with a particle charge detector (PCD 03, Mütek, Germany). The pH variation between 3 and 10 was achieved using 0.1 and 0.01 M solutions of NaOH and HCl, respectively. The potential measurements were carried out using 1 mg beads in 10 mL Millipore water, at room temperature.

#### *4.5. Molecular Dynamics Simulation*

Three systems containing two starch molecules solvated in water and two systems containing two starch molecules and one chitosan molecule were built using the Amber-Tools 18 software (University of California, San Francisco, CA, USA) [51]. Each starch molecule contained 26 repetitive units of linear starch (amylose), and the grafted starch had three units of acrylonitrile or amidoxime grafted to each starch monomer. The chitosan molecule had 26 chitosan monomer units with 15% acetylation. The partial atomic charges were obtained by RESP using the R.E.D.-III.5 tools software [52]. The MD simulations were performed using the GAFF2 forcefield [53] for the starch molecules, the GLYCAM forcefield for the chitosan and the TIP3P water model [54] for the solvent and ions. GROMACS 2020 [55] was used to run the simulation with a temperature set at 300 K, by using V-rescale temperature coupling with a time constant of 0.5 ps. The pressure was controlled by the Parrinello–Rahman barostat and isotropic pressure coupling, with a constant time of 2.0 ps and compressibility of 4.5 × 10<sup>−</sup>5. Each simulation had a length of 200 ns, the composition of simulated systems being introduced in Table 3.

**Table 3.** Molecular dynamic composition of the simulated systems.


#### *4.6. Composite Beads' Swelling Behavior*

The swelling behavior of the composite beads was studied using the conventional gravimetric procedure, by immersing the dry samples for different time intervals in Millipore water with different pHs, at 25 ◦C. Swollen composite hydrogel beads were weighed at predetermined intervals, after wiping the excess surface liquid using filter paper. To calculate the swelling ratio, SR, the three measurements made for each sample and the mean data were used in Equation (1).

$$\text{SR} = (\text{W}\_{\text{t}} - \text{W}\_{\text{0}}) / \text{W}\_{\text{0} \prime} \,\text{g} / \text{g} \,\tag{1}$$

where W0 and Wt are the weights of the samples in the dried state and in the swollen state at time t, respectively.

#### *4.7. Sorption Experiments*

Sorption experiments were conducted using glass bottles containing 0.1 g of the sorbents and 20 mL solution of metal ions or dyes, with an initial pH of about 6, initial concentration of metal ions of 300 mg/L and initial concentration of the dye solutions of 3 × 10−<sup>5</sup> M. The glass bottles were placed on a slow-moving platform shaker (LCD Digital Linear Shaker, SK-L330-Pro, DLAB Scientific Inc., Los Angeles, CA, USA) for 5–6 h. The concentrations of metal ions in the solution, before and after sorption, were analyzed using an atomic adsorption spectrometer (AAS) (ContrAA 800 Spectrometer Analytik Jena, Jena, Germany). The concentration of dyes, before and after sorption, was determined using a UV–Vis spectrophotometer (UV–Vis SPEKOL 1300, Analytik Jena, Jena, Germany), based on the calibration curves determined at the specific wavelength of 497 nm for CR and

597 nm for DR15. The sorption capacity (SC) of the hydrogel beads was calculated with Equations (2) and (3).

$$\text{SC} = (\text{Cs V}) / \text{m} / \text{m} / \text{g} \tag{2}$$

where Cs = sorbed concentration of ions or dyes [g/L], V = volume of the sorption solution and m = weight of used sample [g].

$$\text{SC}\_{\text{M}} = \text{C}\_{\text{sM}} / \text{M}\_{\text{AS}} \tag{3}$$

where Cs = sorbed concentration of ions after 360 min (mol Me/g hydrogel beads), and M = active sites (AS) of hydrogel beads (mol AS/g hydrogel beads).

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/gels8090549/s1, Figure S1: FTIR spectra of CS/SgAN; Figure S2: FTIR spectra of CS/SgAN-Ax; Figure S3: FTIR spectra of CS/SgAx; Table S1: FTIR spectra deconvolution detailed in the region 1800–1500 cm<sup>−</sup>1.

**Author Contributions:** Conceptualization, D.F.L. and M.M.; methodology, D.F.L., S.V. and S.R.; software, T.V.; validation, D.F.L., T.V. and M.-M.Z.; formal analysis, D.F.L. and M.M.; investigation, D.F.L., M.M.B. and M.-M.Z.; resources, D.F.L. and M.M.; data curation, D.F.L. and M.-M.Z.; writing—original draft preparation, D.F.L., M.M.B, T.V. and M.M.; writing—review and editing, M.M.; visualization, M.M.; supervision, M.M.; project administration, D.F.L. and M.M.; funding acquisition, D.F.L. and M.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants of the Ministry of Research and Innovation, CNCS-UEFSCDI, project number PN-III-P1.1.PD-2016-1313 and project number PN-III-P4-ID-PCE-2020-1541, within PNCDI III.

**Acknowledgments:** This work was supported by the research infrastructure developed through the European Social Fund for Regional Development, Competitiveness Operational Programme 2014–2020, Axis 1, Action: 1.1.3, Project "Infra SupraChem Lab-Center for Advanced Research in Supramolecular Chemistry" (contract 339/390015/25.02.2021, cod MySMIS: 108983).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Fengdi Li, Giao T. M. Nguyen, Cédric Vancaeyzeele , Frédéric Vidal and Cédric Plesse \***

Laboratory of Physicochemistry of Polymers and Interfaces, CY Cergy Paris Université, 5 Mail Gay Lussac, 95000 Neuville sur Oise, France; fengdi.li@cyu.fr (F.L.); tran-minh-giao.nguyen@cyu.fr (G.T.M.N.); cedric.vancaeyzeele@cyu.fr (C.V.); frederic.vidal@cyu.fr (F.V.)

**\*** Correspondence: cedric.plesse@cyu.fr

**Abstract:** Ionogels are solid polymer gel networks loaded with ionic liquid (IL) percolating throughout each other, giving rise to ionically conducting solid electrolytes. They combine the mechanical properties of polymer networks with the ionic conductivity, non-volatility, and non-flammability of ILs. In the frame of their applications in electrochemical-based flexible electronics, ionogels are usually subjected to repeated deformation, making them susceptible to damage. It appears critical to devise a simple and effective strategy to improve their durability and lifespan by imparting them with healing ability through vitrimer chemistry. In this work, we report the original in situ synthesis of polythioether (PTE)-based vitrimer ionogels using fast photopolymerization through thiol-acrylate Michael addition. PTE-based vitrimer was prepared with a constant amount of the trithiol crosslinker and varied proportions of static dithiol spacers and dynamic chain extender BDB containing dynamic exchangeable boronic ester groups. The dynamic ionogels were prepared using 50 wt% of either 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl) imide or 1-Ethyl-3-methylimidazolium trifluoromethanesulfonate, both of which were selected for their high ionic conductivity. They are completely amorphous (*Tg* below −30 ◦C), suggesting they can be used at low temperatures. They are stretchable with an elongation at break around 60%, soft with Young's modulus between 0.4 and 0.6 MPa, and they have high ionic conductivities for solid state electrolytes in the order of 10−<sup>4</sup> S·cm−<sup>1</sup> at room temperature. They display dynamic properties typical of the vitrimer network, such as stress relaxation and healing, retained despite the large quantity of IL. The design concept illustrated in this work further enlarges the library of vitrimer ionogels and could potentially open a new path for the development of more sustainable, flexible electrochemical-based electronics with extended service life through repair or reprocessing.

**Keywords:** ionogel; vitrimer; polythioether; solid electrolyte; self-healing

#### **1. Introduction**

Ionogels belong to the general class of polymer gels, which may be regarded as solid and liquid phases that percolate throughout each other. They are termed ionically conducting membranes when the ionic liquid (IL) is loaded within a polymer gel network. Such materials are more precisely achieved by in situ polymerization of the gel network in the presence of ILs or by swelling a polymer network with ILs [1]. Ionogels combine the mechanical properties of crosslinked polymer networks with the ionic conductivity, nonvolatility, and non-flammability of ILs [1–6]. The remarkable physicochemical properties of ionogels make them promising candidates for applications in flexible electronics. Our group has previously reported the synthesis of such ionogels based on polythioether (PTE) networks and ILs for use as solid electrolytes in electrochemical devices [1,4]. More precisely, these ionogels were obtained from the reaction of multifunctional thiols on diacrylate using thiol-ene Michael addition chemistry in the presence of ILs. The thiol-ene Michael addition is a reaction that involves a base- or nucleophile-catalyzed addition of a thiolate

**Citation:** Li, F.; Nguyen, G.T.M.; Vancaeyzeele, C.; Vidal, F.; Plesse, C. Photopolymerizable Ionogel with Healable Properties Based on Dioxaborolane Vitrimer Chemistry. *Gels* **2022**, *8*, 381. https://doi.org/ 10.3390/gels8060381

Academic Editor: Viorel-Puiu Paun

Received: 19 May 2022 Accepted: 9 June 2022 Published: 15 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

anion to electron-deficient alkenes such as maleimides, vinyl sulfones, acrylates, and methacrylates [7]. The sole difference between base- and nucleophile-catalyzed reactions lies in the way the thiolate anion is generated. Classified as a 'click' chemistry, the thiol-ene Michael addition is rapid, highly efficient, generates no by-product, and exhibits a nearly ideal 1: 1 stoichiometric reactivity [8,9]. Owing to the stoichiometric reactivity of the thiolene Michael addition, a fine-tuning of the surface functionality and mechanical properties of ionogels was possible [1].

In the frame of their applications in electrochemical-based flexible electronics, ionogels are usually subjected to repeated deformation, making them susceptible to damage. Thus, it is critical to devise a simple and effective strategy to improve their durability and lifespan. Imparting ionogels with healing ability seems to be a promising approach because of their capability to repair mechanically induced damage. Hydrogen bonds, ionic bonds, and metal–ligand coordination have all been explored to develop healable ionogels [10–13]. Even though ionic and hydrogen bonds have been shown to demonstrate effective selfhealing properties, these physical networks are generally vulnerable to heat, water, and other polar solvents. Another strategy to endow materials with self-healing capability is to introduce reversible covalent bonds within a chemically crosslinked network. The fabrication of dynamic reversible polymer networks has become a popular strategy, notably by introducing exchangeable chemical bonds into polymer networks, which are known as covalent adaptable networks (CANs). CANs are further divided into dissociative and associative mechanisms based on the intrinsic mechanism of the bond-exchange reaction [14]. In dissociative CANs, bonds are first broken and then reformed in response to external stimuli, such as heat or light [14–16]. Healable ionogels have been reported using dissociative bond-exchange reactions [17,18]. However, dissociative CANs allow topological rearrangements due to a sudden viscosity drop and uncrosslinking, which is a drawback in applications requiring stability toward solvents, easy shaping, and a welding process. CANs that rely on an associative bond exchange reaction are characterized by a constant crosslink density [19]. As the bond cleavage is accompanied by the simultaneous formation of a new crosslink, such systems can change their topology with no loss of connectivity, making such networks permanent and insoluble. More specifically, in 2011, the term 'vitrimers' was introduced by Leibler et al. for thermally activated associative CANs [20]. A few groups have reported vitrimer ionogels. Healable and reprocessable gelatine ionogels based on the reversible exchange of imine bonds have been designed for flexible supercapacitors [21]. Xu et al. reported polyurethane (PU) ionogels that can be readily healed at room temperature and restore their original performance owing to the dynamic boronic ester crosslinker used in the polymer network [22]. The exchange reactions between boronic ester linkages do not generally need a catalyst, initiator, or elevated temperature, and the activation energy of this exchange is relatively low. Another team also showcased healable and recyclable boronic ester-based ionogels using 1-butyl-3 methylimidazolium tetrafluoro-borate [23].

The goal of this work was to synthesize an original ionically conducting polythioetherbased vitrimer based on ionogels using fast photopolymerization (Figure 1). The ionogels' mechanical properties can be fine-tuned thanks to the 1:1 stoichiometric reactivity of thiolacrylate Michael addition. Either 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl) imide (EMIM TFSI) or 1-Ethyl-3-methylimidazolium trifluoromethanesulfonate (EMIM Triflate) was chosen as ionic liquid in these ionogels due to their high ionic conductivity. While keeping the 1:1 stoichiometric ratio between thiol and acrylate groups, and the amount of the crosslinker constant, the choice of the thiol spacers is varied between dynamic dithiol-containing boronic ester spacer and non-dynamic dithiol spacer to wisely control the dynamic properties. The thermal properties, mechanical properties, and ionic conducting behavior of these ionogels were studied to highlight their wide working temperature range, good mechanical properties, and polymer electrolyte propensity. The dynamic properties of these solid-state electrolytes were examined with healing and stress relaxation experiments.

This work further enlarges the library of vitrimer ionogels and provides one simple and effective method to develop healable ionogels with satisfying ionic conductivity.

**Figure 1.** Chemical composition and illustration of the PTE-BDB-IL dynamic ionogels based on boronic ester exchange reaction. We used a 1:1 stoichiometric ratio of acrylate and thiols (1.0 diacrylate:0.5 dithiol:0.5 trithiol). Samples were prepared in different proportions between the flexible spacer DT and the boronic ester dynamic spacer BDB, which is capable of a fast bond exchange reaction. Fifty wt% of either EMIM TFSI or EMIM Triflate compared to the total weight is used to prepare PTE-BDB-IL dynamic ionogels.

#### **2. Results and Discussion**

We have previously reported the preparation and characterization of PTE-based ionogels using thiol-ene Michael addition between a mixture of different thiol and acrylate functional groups [1]. The resulting soft and stretchable ionogels can withstand repeated use and considerably large deformation without failure, making them potential candidates for use in the development of wearable and stretchable electronic devices. Therefore, in this study, similar compositions of ionogel were used as a starting point for the preparation of PTE-based dynamic ionogels. That is, using poly(ethylene glycol) diacrylate (PEGDA) as an electron-deficient partner, 1,4-Butanediol Bis(thioglycolate) (dithiol, DT), trimethylolpropane tris(3-mercoptopropianate) (trithiol, TT) as a thiol partner with their a functional group molar ratio of 100:50:50 in the presence of 50 wt% ionic liquid relative to the total mixture weight.

#### *2.1. Synthesis and Characterization of PTE-BDB Dynamic Networks*

In this study, we use a novel dithiol bearing boronic ester groups (2,2 -(1,4-Phenylene) bis[4-mercaptan-1,3,2-dioxaborolane], BDB, Figure 1) to introduce exchangeable bonds into our PTE-based ionogel. This dithiol has recently been reported to be used as a dynamic crosslinker with pendant vinyl groups of styrene-butadiene rubber chains via a thermally initiated thiol-ene "click" reaction [24]. In this frame, BDB was synthesized following the protocol reported by Chen et al. [24]. Dynamic PTE-BDB networks were prepared by photopolymerization with the presence of 1 wt% of photobase generator (PBG vs the total weight of precursor mixture). While keeping the 1:1 stoichiometric ratio between thiol and acrylate groups, and the amount of trithiol crosslinker (0.5 molar ratio of thiol groups of TT), the choice of the thiol spacers (0.5 molar ratio, by reactive bonds) is varied between dynamic BDB and static DT to control the dynamic properties. Three systems

were prepared: (i) non-vitrimer system PTE-BDB0 with 50 mol% of thiols from the DT static spacer (i.e., 100% of dithiol spacers being static DT); (ii) partial vitrimer system (PTE-BDB25) with 0.25 molar ratio of BDB and 0.25 molar ratio of DT static spacer; (iii) full vitrimer system PTE-BDB50 with 50 mol% of thiols from BDB dynamic spacer (i.e., 100% of dithiol spacer being the dynamic BDB). The chemical structures of all chemicals are illustrated in Figure 1. The composition and properties of all samples are reported in Table 1.


**Table 1.** Compositions and characterizations of PTE-BDB networks.

Rheological studies were carried out on PTE-BDB samples (containing 0 mol%, 25 mol% or 50 mol% of dynamic spacers) by pouring the precursor mixtures into the rheometer followed by an in-situ photopolymerization at 30 ◦C. The storage modulus (G ) and loss modulus (G") were recorded as a function of time (Figure 2a). The gel points, where G' and G" curves intersected, were reached within 30 s for the PTE-BDB25 and PTE-BDB50 samples and within 20 s for the PTE-BDB0 sample, thanks to the fast thiol-ene Michael addition. PTE-BDB25 and PTE-BDB50 samples reached a G plateau of about 400 kPa within 5 min, while PTE-BDB0 reached a G plateau of 800 kPa within 3 min. The slower polymerization kinetics indicate that the thiol functional groups of the BDB dynamic spacer are less reactive than the thiol groups of the DT static spacer. This observation is in agreement with the increased steric hindrance of the BDB chain extender, which would slow the chain transfer step of thiol-ene Michael addition [25]. The lower storage modulus could imply less efficient incorporation of the more rigid BDB crosslinker.

The soluble fractions of all samples were extracted in DCM at 60 ◦C under 100 bar to verify the successful formation of polymer networks. Those of the PTE-BDB25 and PTE-BDB50 samples were 8 wt% and 12.5 wt%, respectively, which were higher than that of the PTE-BDB0 sample without a dynamic spacer (3.9 wt%). BDB accounted for 8 wt% of the total weight of precursors for preparing the PTE-BDB25 sample and 15.7 wt% for the PTE-BDB50 sample. To verify that BDB was successfully incorporated into the polymer networks, extractable contents were examined by 1H NMR. The NMR spectrum of the soluble fractions of the PTE-BDB25 and PIL-BDB50 samples can be found in Figures S2 and S3, respectively. Integrations of peaks originating from precursors were examined, and the ratios between PEGDA, BDB, DT, and TT were calculated. PTE-BDB25 and PTE-BDB50 samples were prepared using PEGDA, BDB, DT, and TT precursors with functional group ratios of 100:25:25:50 and 100:50:0:50, respectively. Therefore, in the case of stoichiometric reactivity, the theoretical ratios of PEGDA/BDB/DT/TT extracted should be 6/1.5/1.5/2 for the PTE-BDB25 sample and 6/3/0/2 for the PTE-BDB50 sample. Based on the calculation, residues of all samples after extraction were found to demonstrate ratios close to the theoretical values, even if the proportion of BDB tended to be progressively overrepresented in the extractable fraction with increasing dynamic crosslinker content (6/1.9/1.1/1.7 for PTE-BDB25 and 6/4.9/0/3 for PTE-BDB50). These results indicate that all thiol and acrylate precursors participated in the polymerization successfully but also confirm the less reactive nature of BDB.

**Figure 2.** (**a**) Rheological studies of PTE-BDB0, PTE-BDB25, and PTE-BDB50 precursor mixtures during in-situ photopolymerization; (**b**) DMA curves of PTE-BDB series samples.

Thermal and mechanical properties: Differential scanning calorimetry (DSC), dynamic mechanical analysis (DMA), and tensile strength tests were conducted to study the thermomechanical properties of these materials and to evaluate how dynamic spacer content impacts these properties. All samples were completely amorphous, and no crystallization or melting was observed (Figure S4). Moreover, these materials demonstrate only one glass transition. The onset *Tg* values are listed in Table 1. There is an evident shift to higher *Tg* values when the aromatic BDB chain extenders replace the more flexible DT spacers, as previously reported [26,27]. The more rigid BDB moieties restrict the segmental chain mobility of the polymer networks, leading to a stronger glass propensity. Figure 2b shows the storage modulus and tan d versus temperature of all samples with different contents of BDB chain extenders. At low temperatures, the storage modulus of the glassy states of all samples was about 2 GPa, and then decreased remarkably when the samples went through an evident and narrow α relaxation. All curves exhibited only one transition, which is in line with the results obtained by DSC. The values of the relaxation temperature Tα values, which correspond to the peak of tan δ versus the temperature curve, are around −23 ◦C. In

addition, the storage moduli of all samples in rubbery states were similar. The influence of BDB spacer loading on the mechanical properties of polymer networks was examined using tensile strength tests (Table 1). By replacing 25 mol% of the DT spacer with the BDB spacer, Young's modulus increased from 1.2 MPa for the PTE-BDB0 sample to 1.5 MPa for the PTE-BDB25 sample. The Young's modulus then slightly decreases to 1.3 MPa for the PTE-BDB50 sample, probably due to the higher soluble fraction, which suggests that crosslinking is less advanced compared to PTE-BDB25. The resulting PTE-BDB50 network, in this case, contains more dangling chains and is therefore softer. The elongation at break values of all systems remained similar at around 70%.

Stress relaxation behavior: Owing to bond exchange reactions of vitrimers at elevated temperatures, these materials demonstrate macroscopic flow and stress relaxation enabled by reversible rearrangement of the crosslinked network, without risking permanent loss of material properties [28]. In contrast, conventional thermosets are resistant at elevated temperatures (but below the degradation temperature) to relaxation under an applied strain due to their stable crosslinked networks [29]. To study the exchange dynamics of boronic ester groups, PTE-BDB samples were subjected to stress relaxation experiments at various temperatures by monitoring the decrease in stress over time at a constant strain of 3%.

Figure 3a compares the relaxation curves of samples containing 0, 25, and 50 mol% of BDB spacers at 80 ◦C, where the relaxation percentage was plotted as a function of time. PTE-BDB0 sample containing no BDB dynamic spacer relaxed very few percentages of the stress before reaching a plateau, demonstrating the typical elastic response of a thermoset. Upon increasing the BDB content and decreasing the fraction of the permanent network, samples demonstrated stress relaxation behavior, with PTE-BDB25 and PTE-BDB50 relaxed about 50% and 65% of the stress, respectively, within 500 s. These results provide convincing evidence of the exchange reaction between boronic ester groups, allowing network rearrangement in the PTE-BDB samples. Figure 3b,c compare the stress relaxation behavior of the PTE-BDB25 and PTE-BDB50 samples at different temperatures, respectively. At 30 ◦C, PTE-BDB25 and PTE-BDB50 rapidly relaxed 5% and 30% of the stress, followed by a stress plateau. By increasing the temperature from 30 ◦C to 140 ◦C, both the relaxation rate and the eventual relaxation extent of the PTE-BDB25 and PTE-BDB50 samples increased. These results indicate that at low temperatures, the boronic ester exchange is limited and the network rearrangement is almost 'frozen'. At elevated temperatures, the exchange reaction is activated and demonstrates its temperature-dependent nature as the relaxation rate increases with temperature. At 140 ◦C, the stress relaxation percentages of PTE-BDB25 and PTE-BDB50 stabilized at 83% and 92%, respectively. Full stress relaxation could not be reached even after 1 h, indicating the probable presence of random non-dynamic TT-PEGDA-TT segments within the polymer networks, leading to materials demonstrating vitrimer-like behaviors [30].

Relaxation time is defined as the time required to relax to 37% (1/e) of initial stress [31]. The relaxation time of the PTE-BDB25 and PTE-BDB50 samples can be found in Table S1. PTE-BDB50 sample with a higher BDB fraction exhibits shorter relation times than the PTE-BDB25 sample at the same temperature. A faster network rearrangement is expected with an increasing number of exchangeable boronic ester linkages. Moreover, the relaxation times exhibited an Arrhenius-like temperature dependence (Figure 3d), indicating the associative exchange mechanism of the boronic ester exchange reaction [24]. Accordingly, the activation energies (*Ea*) were calculated to be 75.6 kJ·mol−<sup>1</sup> and 49.9 kJ·mol−<sup>1</sup> from the slope of the Arrhenius linear fit of PTE-BDB25 and PTE-BDB50, respectively. These values are within the range of 50–90 kJ·mol−<sup>1</sup> reported in the literature, where the same dioxaborolane groups are used as dynamic chain extenders [27,32]. The more available boronic ester groups existing in the PTE-BDB50 network resulted in faster exchange kinetics and consequently lower activation energy of the viscous flow. As another characteristic key for vitrimer materials, the hypothetical topology freezing temperature (*Tv*) is conventionally chosen as the temperature at which the viscosity equals 1012 Pa.s that describes the liquid-

to-solid transition of a glass-forming liquid [14,20]. The hypothetical *Tv* values can be calculated from the relaxation times using the method described in the experimental section. *Tv* of PTE-BDB25 and PTE-BDB50 were calculated to be 11.7 ◦C and −31 ◦C, respectively (Table S1). Despite the low *Tg* and *Tv* values, as indicated by the relaxation experiments, a relaxation stress plateau was observed at 30 ◦C for both materials, confirming their stable topological behavior at room temperature.

**Figure 3.** (**a**) Stress relaxation behaviors of PTE-BDB samples with 0, 25, and 50 mol% of BDB dynamic chain extender at 80 ◦C; (**b**) Stress relaxation curves of PTE-BDB25 sample at different temperatures; (**c**) Stress relaxation curves of PTE-BDB50 sample at different temperatures; (**d**) Arrhenius linear fit of relaxation times of PTE-BDB25 and PTE-BDB50 samples plotted as a function of 1000/T.

Healing behavior: Vitrimers are polymers that can change their topology through dynamic exchange reactions without degenerating the network, maintaining a constant crosslink density [14,20,33–36]. To examine the dynamic exchange of boronic ester groups, the healing abilities of PTE-BDB series samples were studied by cutting film samples into two pieces and stacking them face-to-face. The films were clipped between two glass plates at 120 ◦C for 2 h. Pictures of all samples before and after the healing test are shown in Figure 4. After 2 h, the two pieces of PTE-BDB0 could be easily separated, and no healing effect could be observed. In contrast, the two pieces of the PTE-BDB25 and PTE-BDB50 samples could not be separated without breaking them. However, the two film pieces were not fully fused, indicating incomplete healing, which is consistent with the incomplete stress relaxation described earlier. These results proved that boronic ester exchange is directly responsible for the healing behavior of our materials.

**Figure 4.** Healing behavior of PTE-BDB samples at 120 ◦C for 2 h.

#### *2.2. Synthesis and Characterization of PTE-BDB-IL Dynamic Ionogels*

PTE-BDB25 and PTE-BDB50 samples demonstrated dynamic properties such as stress relaxation and healing thanks to the network reorganization enabled by the exchange reaction of boronic ester bonds. To combine these interesting properties with ionically conducting behavior, ionic liquids were incorporated into these networks. A content of 25 mol% of BDB dynamic spacer was selected because of the higher dimensional stability of the PTE-BDB25 sample. Fifty wt% of either EMIM TFSI or EMIM Triflate was incorporated into the reagent mixture in order to achieve in situ ionogel formation. The chemical formula of the precursors and illustration of the PTE-BDB-IL dynamic ionogels based on the boronic ester exchange reaction are shown in Figure 1. Table 2 compares the compositions and properties of the PTE-BDB-IL ionogels with the PTE-BDB25 sample.


**Table 2.** Compositions and details of the PTE-BDB-IL samples.

Soluble fractions of PTE-BDB25-TFSI50 and PTE-BDB25-TfO50 TfO50 (i.e., PTE sample containing 25 mol% of BDB and 50 wt% of EMIM Triflate) in DCM were 53.8% and 56.3% respectively. Considering that 50 wt% of ILs were used to prepare the ionogels, these results demonstrated that the polymer networks are well formed. The rheological properties of the PTE-BDB25-TFSI50 (i.e., PTE sample containing 25 mol% of BDB and 50 wt% of EMIM TFSI) and PTE-BDB25 precursor mixtures were monitored during photopolymerization and compared to the PTE-BDB25 sample (Figure 5a). As the liquid precursor solution turned into solid crosslinked ionogel when exposed under UV, the liquid-to-solid transition indicated by the cross-over of the G" and G curves was attributed to the gel point [37–39]. The gel points of both samples were reached within 30 s, as in the case of the corresponding single network, demonstrating the fast polymerization kinetics of the thiol-ene Michael addition. G plateaus of 200 kPa for the PTE-BDB25-TFSI50 sample and 120 kPa for the PTE-BDB25-TfO50 sample were reached within 5 min. They are lower than the G plateau of PTE-BDB25 (400 kPa) prepared in bulk, which indicates material softening in the presence of ILs.

**Figure 5.** (**a**) Rheological studies of the PTE-BDB25-TFSI50 and PTE-BDB25-TfO50 precursor mixtures during in situ photopolymerization compared with PTE-BDB25 samples; (**b**) DMA tests of PTE-BDB-IL ionogels and PTE-BDB25 sample.

Thermal and mechanical properties of the ionogels: The thermal properties of ionogel samples were studied and compared with the PTE-BDB25 sample (Figure 5b). All samples are completely amorphous materials; no crystallization or melting could be seen in the thermograms, even at temperatures lower than the melting point of the pure ionic liquids (Figure S5). These materials displayed only one glass transition, suggesting that these ionogels can be potentially utilized at temperatures as low as −20 ◦C. *Tg* values of −49.4 ◦C and −51.8 ◦C were found for PTE-BDB25-TFSI50 and PTE-BDB25-TfO50, respectively, which were lower than that of PTE-BDB25 (−41 ◦C). Indeed, the plasticizing effect of ILs is responsible for the lower *Tg* values of ionogels. The storage modulus and tan d versus temperature of all samples are compared in Figure 5b. The addition of ILs resulted in a slightly lower storage modulus at both glassy and rubber states and lower Tα values. Tensile tests were conducted on the PTE-BDB25-TFSI50 and PTE-BDB25-TfO50 samples. The values of Young's modulus and elongation at break are listed in Table 2. Compared to the PTE-BDB25 sample, these ionogels demonstrated lower tensile strengths and similar stretchability to the presence of ILs.

Dynamic behavior of the ionogels: PTE-BDB-IL ionogels were also subjected to stress relaxation experiments carried out between 60 and 140 ◦C to study whether the addition of ILs would modify the dynamic properties enabled by the boronic ester groups of the PTE-BDB25 network. Table S2 compares the stress relaxation behaviors of the PTE-BDB25, PTE-BDB25-TFSI50, and PTE-BDB25-TfO50 samples. Stress relaxation curves of the PTE-BDB25-TFSI50 and PTE-BDB25-TfO50 at different temperatures can be found in Figure 6a,b, respectively. In both cases, the relaxation rate increases with temperature, as the relaxation process is essentially controlled by the thermally activated boronic ester exchange reaction, the rate of which increases with temperature. The relaxation behavior of ionogel is represented as Arrhenius-like temperature dependence in Figure 6c. As for the single networks, *Ea* and *Tv* were calculated to be 78.7 kJ·mol−<sup>1</sup> and 0.3 ◦C for PTE-BDB25-TFSI50, 72.0 kJ·mol−<sup>1</sup> and 2.0 ◦C for PTE-BDB25-TfO50, which were comparable to those values of PTE-BDB25 sample (75.6 kJ·mol−<sup>1</sup> and 11.7 ◦C). The similar *Ea* values indicate that the ILs are spectator compounds in regard to the exchange reaction. However, the behavior of PTE-BDB25-TFSI50 seems to deviate from the expected linear behavior. This can be due to experimental error but also to the presence of TFSI- counterion susceptible to form boron-TFSI adduct responsible for a significant decrease of relaxation times, as previously reported in the presence of LiTFSI salt [40]. The *Tv* displayed the same 10 ◦C decrease as observed for the *Tg*. We can assume that this comes from the already mentioned IL plasticizing effect, which promotes the mobility and rearrangement of the polymer chain at low temperature.

**Figure 6.** (**a**) Stress relaxation tests of PTE-BDB25-TFSI50 sample at various temperatures; (**b**) The stress relaxation tests of the PTE-BDB25-TfO50 sample at various temperatures; (**c**) Arrhenius linear plot extracted from relaxation times of the (-) PTE-BDB25-TFSI50, () PTE-BDB25-TFSI50, and (•) PTE-BDB25 samples at different temperatures.

Ionic conducting and healing profiles of the ionogels: After demonstrating that the presence of ILs does not inhibit the boronic ester dynamic exchange. The ionic conductivity behaviors of PTE-BDB-IL ionogels at different temperatures were studied using electrochemical impedance spectroscopy (Figure 7a). At 25 ◦C, PTE-BDB25-TFSI50 and PTE-BDB25- TfO50 demonstrated ionic conductivities of 1.3 × 10−<sup>4</sup> S·cm−<sup>1</sup> and 1.1 × 10−<sup>4</sup> S·cm<sup>−</sup>1, respectively. Such conductivity has been proven to be satisfying for a wide range of flexible electronic applications, such as sensors and solid electrolytes [13]. By increasing the temperature, the ionic conductivities increased as the ion mobility rose, eventually reaching 1.3 × 10−<sup>3</sup> S·cm−<sup>1</sup> and 1.4 × 10−<sup>3</sup> S·cm−<sup>1</sup> at 80 ◦C respectively. The temperature dependence of these ionogels can be described by the Vogel–Tamman–Fulcher (VTF) equation [41–43]:

$$
\sigma = AT^{-\frac{1}{2}} e^{-\frac{E\_g}{R(T-T\_0)}} \tag{1}
$$

where *A* is a temperature-independent constant associated with the number of charge carriers, *Ea* is the pseudo-activation energy related to polymer segmental motion, and *R* stands for the gas constant. *T*<sup>0</sup> is a reference temperature usually correlated with the ideal glass transition temperature at which free volume disappears or at which the configurational entropy of the polymer chain reaches zero. In either scenario, *T*<sup>0</sup> is usually 35 to 50 K below *Tg* [44–46]. The VTF behaviors of all PTE-BDB-IL samples were investigated

with *T*<sup>0</sup> set to *Tg*—50 K. The values of R<sup>2</sup> of the VTF linear fit are above 0.99, showing that the VTF model is suitable for describing the ionic behavior of these materials and that polymer chain segmental mobility plays a role in facilitating ion conduction. The VTF of the two samples are very similar, with the parameter *A* found to be 5.6 S·K1/2·cm−<sup>1</sup> for the PTE-BDB25-TFSI50 sample and 5.7 S·K1/2·cm−<sup>1</sup> for the PTE-BDB25-TfO50 sample, and the *Ea* of ionic conduction of 8.1 kJ·mol−<sup>1</sup> and 8.5 kJ·mol−<sup>1</sup> respectively.

**Figure 7.** (**a**) Ionic conducting behaviors of PTE-BDB25-TFSI50 and PTE-BDB25-TfO50 samples at different temperatures before and after the healing test at 120 ◦C for 2 h; (**b**) Pictures of PTE-BDB25- TFSI50 and PTE-BDB25-TfO50 samples before and after healing.

PTE-BDB-IL ionogels are also expected to display healing properties because of the dynamic exchange of boronic ester bonds. The two IL-loaded PTE-BDB25 sample films were stacked face-to-face and subjected to a healing experiment at 120 ◦C for 2 h. For both samples, the two films could not be separated after 2 h, demonstrating healing behavior similar to that of the PTE-BDB25 sample (Figure 7b). To evaluate the healing efficiency of these ionogels, the cross-sections of the stacked samples were examined by scanning electron microscopy (SEM) after healing. A scar of 263 nm between the two stacked films was observed for the PTE-BDB25-TFSI50 sample (Figure 8a), while a slightly larger scar of 2.6 mm was found in the PTE-BDB25-TfO50 sample (Figure 8b). These results demonstrate the healing abilities of these ionogels thanks to the bond exchange reaction, while the remaining scars are consistent with the incomplete relaxation observed at 120 ◦C. Moreover, the fast relaxation rate in the presence of TFSI- anions may induce a better healing efficiency than triflate anions within the same healing duration. For healable ionogels, the materials must retrieve their ionically conducting behavior after failure. Thus, ionic conductivity measurements were carried out on the samples after the healing experiment. It was found

that the healed samples demonstrated similar behaviors as pristine samples, and the ionic conductivity remained in the order of magnitude (Figure 7a). These results indicate that the topology rearrangement of the polymer electrolytes enabled by boronic ester groups allows the materials to heal and recover their original ionically conducting profile.

**Figure 8.** (**a**) SEM image of PTE-BDB25-TFSI50 healed sample cut transection; (**b**) SEM image of PTE-BDB25-TfO50 healed sample cut transection.

#### **3. Conclusions**

In this work, we demonstrated the in-situ preparation of polythioether-based vitrimer ionogels, taking advantage of the thiol-acrylate Michael addition. In the first step, to select the best polymer network of dynamic ionogels, dynamic PTE-based polymer networks were prepared by keeping the amount of trithiol crosslinker constant. The choice of the dithiol spacers varies between a dynamic chain extender BDB containing boronic ester groups (from 0 to 50 mol% of total thiol functions) and static dithiol to control the dynamic properties of these materials, with relaxation times varying with the composition of the samples from 160 min at 60 ◦C to 36 s at 140 ◦C. These PTE-BDB networks exhibited vitrimer properties, such as healing and stress relaxation, at elevated temperatures, thanks to the boronic ester exchange reaction. In the second part, dynamic ionogels were prepared using 50 wt% of either EMIM TFSI or EMIM Triflate compared to the total weight. The resulting materials are completely amorphous (*Tg* around −50 ◦C), suggesting that these ionogels can be potentially utilized at low temperatures. These ionogels are stretchable with an elongation at break around 60%, soft with Young's modulus between 0.4 to 0.6 MPa, and demonstrated ionic conductivities in the order of 10−<sup>4</sup> S·cm−<sup>1</sup> at room temperature. It has been found that the dynamic properties of these materials, such as stress relaxation (with relaxation time in the same range) and healing, are retained and not significantly modified in the presence of a large quantity of IL. This work further enlarges the library of vitrimer ionogels, and we can envision an easy surface functionalization with either thiol or acrylate groups of these ionogels, thanks to the stoichiometric reactivity of thiol-acrylate Michael addition. The design concept illustrated in this work could potentially open a new path for the development of flexible electrochemical-based electronics with extended service life through repair or reprocessing.

#### **4. Materials and Methods**

#### *4.1. Materials*

The photobase generator (PBG) 2-(9-Oxoxanthen-2-yl)propionic acid 1,5,7-triazabicyclo [4.4.0]dec-5-ene salt, 1,4-Butanediol Bis(thioglycolate) (dithiol, DT), and heptane were purchased from TCI Chemicals (Zwijndrecht, Belgium). Poly(ethylene glycol) diacrylate (PEGDA, Mn = 700 g·mol<sup>−</sup>1), trimethylolpropane tris(3-mercoptopropianate) (trithiol, TT), triethylamine, 1-Ethyl-3-methylimidazolium trifluoromethanesulfonate (EMIM Triflate), and 1-thioglycerol were purchased from Sigma-Aldrich (De Schnelldorf, Germany). Magnesium sulfate heptahydrate (MgSO4.7H2O) was obtained from Acros Organics (Geel, Belgium). Dichloromethane (DCM) was obtained from VMR Chemicals (Fontenay sous Bois, France). 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (EMIM TFSI) was purchased from Solvionic (Toulouse, France). Finally, benzene-1,4-diboronic acid was purchased from Apollo Scientific (Stockport, UK).

#### *4.2. Synthesis of 2,2 -(1,4-Phenylene)-bis[4-mercaptan-1,3,2-dioxaborolane] (BDB)*

The synthesis of dithiol-containing boronic ester 2,2 -(1,4-Phenylene)-bis[4-mercaptan-1,3,2-dioxaborolane] (BDB) was reported by Chen et al. [24]. Benzene-1,4-diboronic acid (3.0 g, 18.1 mmol) and 1-thioglycerol (4.01 g, 37.1 mmol) were dissolved in tetrahydrofuran (80 mL) and water (0.1 mL). Five grams of magnesium sulfate was added to the mixture. After stirring at room temperature for 24 h, the mixture was filtered and concentrated. The resulting solid is purified by repeatedly filtering and washing with abundant heptane, and concentrated to obtain the target compound as white solids (yield 80%). The successful synthesis of BDB was explicitly confirmed by 1H NMR (Figure S1). 1H NMR (400 MHz, CDCl3) δ 7.83 (s, 4H), 4.74 (m, 2H), 4.49 (dd, 2H), 4.18 (dd, 2H), 2.81 (dd, 4H), 1.48 (t, 2H).

#### *4.3. Preparation of PTE-BDB Dynamic Networks*

In a vial, the dithiol-containing boronic ester (BDB) is solubilized with acetone (msolvent = 0.5 mPEGDA) before introducing thiol precursors (TT and/or DT) and the acrylate precursor PEGDA. In parallel, the photobase generator PBG (1 wt% of the total weight of thiol and acrylate precursors) is dissolved in EtOH (50 mg·mL<sup>−</sup>1) and then added into the vial under a light-protected condition. The mixture was poured into a mold consisting of two glass plates separated by a 0.5 mm thick Teflon spacer. Free-standing PTE-BDB films are obtained by curing the precursor solution with a UV curing conveyor system (Primarc UV Technology, Slough, UK, Minicure, Mercury vapor Lamp, UV intensity 100 W·cm<sup>−</sup>2, duration of each scan 4 s). Fifty UV passages were applied for each sample. The acetone was evaporated under vacuum after synthesis at 50 ◦C for 1 day.

#### *4.4. Preparation of PTE-BDB-IL Dynamic Ionogels*

In a vial, the dithiol-containing boronic ester (BDB) is solubilized with acetone (msolvent = 0.5 mPEGDA) before introducing thiol precursors (TT and/or DT), acrylate precursor PEGDA), and either EMIM TFSI or EMIM Triflate (50 wt% vs total weight). In parallel, the photobase generator PBG (1 wt% of the total weight of thiol and acrylate precursors) is dissolved in EtOH (50 mg·mL<sup>−</sup>1) and then added into the vial under a light-protected condition. The mixture was cast into a mold consisting of two glass plates separated by a 0.5 mm thick Teflon spacer. Free-standing PTE-BDB-IL films are obtained by curing the precursor solution with a UV curing conveyor system after 50 scanning passages (Primarc UV Technology, Minicure, Mercury vapor Lamp, UV intensity 100 W·cm<sup>−</sup>2, duration of each scan 4 s). The acetone was evaporated under vacuum after synthesis at 50 ◦C for 1 day. The resulting network was named PTE-BDB*X*-TFSI*Y* or PTE-BDB*X*-TfO*Y* for a PTE network containing X mol% of BDB and Y wt% of EMIM TFSI or EMIM Triflate.

#### *4.5. Methods and Techniques*

Nuclear Magnetic Resonance Spectroscopy (NMR): 1H NMR spectra were recorded at 297 K on a AVANCE 400 spectrometer (Bruker, Karlsruhe, Germany) at 400 MHz and referenced to the residual solvent peaks (1H, δ 7.26 ppm for CDCl3).

Infrared spectroscopy (IR): Attenuated total reflection (ATR)-FT-IR spectroscopy was performed using a Tensor 27 (Bruker, Champs-sur-Marne, France) FT-IR instrument equipped with an ATR accessory unit.

Extractable content: Soxhlet experiments were performed with a BUCHI SpeedExtractor E-914 (Villebon sur Yvette, France). The extractable content was determined by 3 cycles of extraction in DCM at 60 ◦C under 100 bar. Each cycle lasted about 15 min.

Rheology: Rheological measurements were performed with an Anton Paar Physica MCR 301 rheometer (Graz, Austria) equipped with a CTD 450 temperature control device and a plate-plate geometry (Gap 500 μm, diameter 25 mm, plate; polymerization system made from a lower glass plate coupled with a UV lamp 142 mW·cm<sup>−</sup>2). A 1% deformation was imposed at 1 Hz. The storage modulus (G ) and loss modulus (G") were recorded as a function of time. The solution of precursors of materials was put in the rheometer geometry, and measurements began immediately with UV exposure at 30 ◦C.

Thermogravimetric analysis (TGA): TGA experiments were performed in air on a Q50 model (TA Instruments, New Castle, DE, USA) applying a heating rate of 10 ◦C·min−<sup>1</sup> to 600 ◦C.

Differential Scanning Calorimetry (DSC): Glass transitions of the materials were determined by DSC. Sequences of temperature ramps (heating, cooling jump, heating, cooling, heating) in the −80 to 180 ◦C range were performed at 20 ◦C·min−<sup>1</sup> ramping up and 5 ◦C·min−<sup>1</sup> cooling down using a TA Instruments Q100 model (New Castle, USA) equipped with a liquid cooling accessory and calibrated using sapphire and high purity indium metal. All samples were prepared in hermetically sealed pans (5−10 mg/sample) and were referenced to an empty pan. The reported *Tg* values were obtained from the second heating cycle.

Tensile testing: Traction experiments were performed on a Dynamic Mechanical Analyzer instrument (TA Instruments, Q800 model, New Castle, USA) in tensile mode at room temperature. A strain rate of 20%·min−<sup>1</sup> to 500% was applied with an initial strain of 0.05% and a preload force of 0.01 N to obtain stress-strain curves.

Dynamic mechanical analysis (DMA): DMA experiments were conducted on Q800 (TA Instruments, New Castle, USA) in tension mode. Heating ramps were performed from −70 ◦C to 200 ◦C at a constant rate of 3 ◦C·min−<sup>1</sup> with a maximum strain amplitude of 0.05% at a fixed frequency of 1 Hz, and a preload force of 0.01 N.

Stress relaxation measurement: Stress relaxation measurements were carried out on the Q800 at different temperatures. A preload force of 0.01 N and a constant strain of 3% were applied, and stress decay was monitored over time.

Ionic conductivity: Ionic conductivity was measured by electrochemical impedance spectroscopy using a VSP 150 potentiostat (Biologic SA, Grenoble, France). Samples were placed between two gold electrodes and placed in a thermostated cell under an argon atmosphere. Experiments were carried out in a temperature range from 25 to 100 ◦C, in the frequency range from 2 MHz to 1 Hz with a rate of 6 points per decade, and for an oscillation potential of 10 mV. The ionic conductivity *σ (S*·cm<sup>−</sup>1) is calculated using Equation (2):

$$
\sigma = \frac{1}{Z} \frac{d}{S} \tag{2}
$$

where *Z* is the real part of the complex impedance (ohms), *d* the thickness of the sample (cm), and *S* is the sample area (cm2).

Healing test: Sample films were first cut into two pieces and stacked together. The films were protected by Teflon films, pressured by 2 glass plates with clips, and then heated at 120 ◦C for 2 h in the oven.

Scanning electronic microscopy (SEM): The samples were mounted directly on SEM stubs, sputtered with 4 nm of platinum (ACE600, Leica, Wetzlar, Germany), and imaged using a Field Emission Gun Scanning Electron Microscope (GeminiSEM300, Carl Zeiss, Oberkochen, Germany) with an acceleration voltage of 2 keV under a high vacuum. Secondary electrons were collected. Scan speed and line averaging were adjusted during the observation.

Calculation of topology freezing temperature *Tv* and activation energy of the viscous flow *Ea*: Based on Maxwell's model for viscoelastic fluids, the stress relaxation behavior of the vitrimer can be described with Equation (3), where the relaxation time *t* is determined as the time required to relax to 37% (1/e) of the initial stress [31]:

$$\frac{\sigma\_{(t)}}{\sigma\_0} = e^{-\frac{t}{\tau}} \tag{3}$$

For vitrimers, relaxation times reflect associative exchange reactions, and their temperature dependence can be fitted to the Arrhenius equation (Equation (4)) [33,47] :

$$
\pi(T) = \tau\_0 e^{\frac{E\_q}{RT}} \tag{4}
$$

The values of t were then plotted as a function of temperature to determine the activation energy *Ea* of the associative exchange reaction. The topology freezing temperature *Tv* is another key characteristic of vitrimer materials. Conventionally, the hypothetical *Tv* is chosen as the temperature at which the viscosity equals 10<sup>12</sup> Pa·s as this value describes the liquid-to-solid transition of a glass-forming liquid [14,20]. The relation between viscosity *η* and the characteristic relaxation time *τ* can be expressed with the Maxwell relation (Equation (5)) [48]:

$$
\eta = \mathbf{G} \cdot \mathbf{r} = \frac{E'}{2(1+\nu)} \cdot \mathbf{r} \tag{5}
$$

where *G* stands for the shear modulus, *ν* for the Poisson's ratio, and *E* for the storage modulus at the rubbery plateau. Using the Poisson's ratio = 0.5 usually used for rubbers, [33,48] *Tv* is determined by combining Equations (4) and (5).

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/gels8060381/s1. Figure S1: 1H NMR of 2,2 -(1,4-Phenylene)-bis[4 mercaptan-1,3,2-dioxaborolane] (BDB) in CDCl3; Figure S2: 1H NMR of the extractable content of PTE-BDB25 sample in CDCl3; Figure S3: 1H NMR of the extractable content of PTE-BDB50 sample in CDCl3.; Figure S4: DSC curves of thermal cured PTE-BDB series samples.; Figure S5: DSC curves of thermal cured PTE-BDB-IL dynamic ionogels and PTE-BDB25 sample. Table S1: Relaxation times extracted from stress relaxation tests of PTE-BDB samples and their corresponding theoretical *Tv* and *Ea* associated with the boronic ester exchange reaction.; Table S2: Relaxation times extracted

from stress relaxation tests of PTE-BDB-IL samples and their corresponding theoretical *Tv* and *Ea* associated with the boronic ester exchange reaction.

**Author Contributions:** Conceptualization, G.T.M.N., C.V., F.V. and C.P.; Data curation, F.L. and G.T.M.N.; Formal analysis, F.L.; Funding acquisition, G.T.M.N., C.V., F.V. and C.P.; Investigation, F.L., G.T.M.N., C.V., F.V. and C.P.; Methodology, F.L.; Project administration, C.P.; Resources, F.V. and C.P.; Supervision, G.T.M.N., C.V., F.V. and C.P.; Validation, G.T.M.N., C.V. and C.P.; Visu-alization, F.L.; Writing—original draft, F.L.; Writing—review & editing, G.T.M.N., C.V., F.V. and C.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No. 825232. "WEAFING project".

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** The authors would like to thank the Conseil regional île de France for funding the Cerasem project (grant: 15013107) the acquisition of a ZEISS Gemini SEM 300.

**Conflicts of Interest:** There are no conflicts to declare.

#### **References**

