The comparison of the FTIR spectra of the mixture of the neat cardanol-based epoxy resin (EC) and DBEDT and the same mixture subjected to a heat treatment at 150°C for 24 h (EC-DBEDT_T), shown in Fig. 2A, served to confirm the occurrence of the cross-linking reaction. The above curing temperature was chosen to promote the cross-linking but limit the thermal decomposition of the material. In both spectra, the presence of the boronic acid ester is evidenced by the stretching of the B-O and the B-O-B bonds at 1211 cm− 1 and 655 cm− 1, respectively, belonging to DBEDT [30]. In addition, the signals at 2925 cm− 1 and 2853 cm− 1 can be attributed to the asymmetric and symmetric stretching of -Csp3-H bonds, respectively. Apart from the bands mentioned above, which are present in both the neat and the treated mixture, some differences are visible. A first observation refers to the broadened band in the thermally treated mixture, in the range between about 3100 cm− 1 and 3700 cm− 1. This signal, associated with the stretching of hydroxyl groups, can be attributed to the formation of hydroxyl functionalities originating from the reaction between thiol groups and epoxides (Fig. 2B) [27–29, 31].
Moreover, it was found that the signal visible in the neat mixture at 910 cm− 1, which can be attributed to the epoxide groups [32], disappears in the treated EC-DBEDT (see insert of Fig. 2). The above finding is also indicative of the occurrence of the reaction between EC and DBEDT under the applied conditions, which is shown in Fig. 2B.
To support the formation of a three-dimensional network, a gel fraction test was performed in anhydrous toluene after checking the solubility of the starting reagents in this solvent. Figure 2C shows the photos of the starting sample and of the material after 24 h in contact with the solvent. It is worth mentioning that this test allowed the evaluation of gel content (GF) and swelling ratio (SR), which were 99% and 188%, respectively. These findings, together with the results of FT-IR characterization, prove the effective cross-linking of the material, owing to the reaction between EC and DBEDT (Fig. 2B). It is worth underling that variable GF were previosuly reported for epoxy-based vitrimers. For example, vitrimers prepared from sebacic acid and diglycidyl ether of bisphenol A, with zinc acetylacetone as catalyst for the exchange
reaction, reached a GF of about 80% at a temperature and time comparable with those used for the preparation of our material, i.e., 160°C and 24 h [33]. In the case of epoxidized soybean oil derived vitrimers, prepared using 4,4′-dithiodiphenylamine as a curing agent and a curing time of 20 h and 160°C of curing temperature, a GF of only 54% was obtained [34]. In this case, the gel fraction could only be increased by increasing the temperature and curing time. Nevertheless, at 180°C and applying a curing time of about 28 h, a maximum value of about 80% was achieved. Comparing these results with the GF value obtained in our work, it can be concluded that the crosslinking system used is extremely effective and allows complete cross-linking without the need to further increase the temperature. This limits problems related to the stability of the material and to secondary reactions, and makes the preparation procedure more easily scalable. Moreover, it can be deduced from the photo of the swollen sample that the high degree of swelling is associated with the retention of the shape of the sample, which is a typical feature of vitrimer systems [4].
Table 1
Thermal properties of the EC and EC_DBEDT_T.
Sample code | Tg [°C] | Tonset [°C] | Tmax1 [°C] | Tmax2 [°C] | Residual weight [%] |
EC | - | 277 | 350 | 439 | 0 |
EC-DBEDT_T | 45 | 304 | 364 | 438 | 15 |
Tonset and Tmax indicate the onset of the degradation temperature at a weight loss of 5% and the maximum rate of degradation temperature, respectively. |
The glass transition temperature of the cured mixture was studied by comparing the DSC traces of neat EC and EC-DBEDT_T (Fig. 3a and Table 1). In contrast to the starting cardanol-based epoxy sample, the cured mixture exhibited a glass transition temperature (Tg) of 45°C, reflecting the relatively high mobility of alkylic chains in cardanol. This finding confirms the formation of macromolecular chains during the cross-linking reaction.
The thermal stability of EC-DBEDT_T and neat EC was studied by TGA analysis at a heating rate of 10°C/min under nitrogen atmosphere. The TGA curves of the analyzed samples are shown in Fig. 3b, while the results are summarized in Table 1, where the onset decomposition temperature (Tonset) and the temperature of the maximum rate of decomposition (Tmax) as well as the residual weight are also given. Both EC and EC-DBEDT_T showed two degradation steps characterized by different Tmax. As previously reported in the literature, the first weight loss can be attributed to the breakdown of the main aliphatic chain of EC, while the second corresponds to the degradation of the phenyl rings contained in the system [35, 36]. It was found that the material cross-linking slightly affects the Tonsent and Tmax1, corresponding to the first degration step, Tonset going from 277°C to 304°C and Tmax1 from 350°C to 364°C for EC and EC-DEBDT_T, respectively. While the starting epoxy did not show any residue at high temperatures, it is relevant to underline that the degradation of the crosslinked material leads to the formation of a significant residual weight (about 15%). This suggests a partial charring of the polymer, in competition with its volatilization, possibly related to the action of the boronic compound, as boron is a well-known charring promoter, which is expected to positively affect the performance in the reaction to a flame exposure [37, 38]. TGA analysis was also performed under oxidizing atmosphere, in air, and the results are shown in the Supporting Information (Figure S1). The TGA profile of EC shows limited charring upon the main decomposition step, yielding approx 20% residue at 400°C, subsequently oxidised leading to negligible residue above 600°C. On the other hand, EC-DBEDT_T decomposes yielding 40% of charred residue at 400°C, which is then progressively decomposed by oxidation at higher temperature, still retaining approx. 15% of the initial weight at 750°C. The porous char residue recovered after TGA in air was analyzed by means of FT-IR (Figure S2). The presence of several bands at 3190, 1406, 1192, 705 and 638 cm− 1 in the spectrum, which can be attributed to the stretching of B-OH bonds and tri- and tetracoordinated B-O units, is consistent with the presence of boron oxide [39–41], thus supporting the role of boron compounds in the organization of a stable charred residue.
Viscoelastic properties for EC-DEBDT_T were studied as a function of temperature by dynamic mechanical analysis and oscillatory rheology to investigate the relaxation dynamics.
DMTA (Fig. 4) exhibits the peak of main chain relaxation (Tα) at approx. 56°C, taken as the maximum of tan δ, in fair agreement with the glass transition temperature measured in DSC. Above the main relaxation, a rubbery state is observed, with storage modulus in the range of a few MPa. Indeed, the rubbery state does not produce a stable plateau, but rather a slowly decreasing curve, ranging between approx. 4 MPa at 80°C and 2.5 MPa at 180°C. This decreasing trend appears to be related to the elongation of the sample, which is higher than for expected linear thermal expansion. This suggests that the sample is actually starting to flow under its own weight in the testing condition, which represents a first evidence of its fast relaxation above Tα [22].
To further investigate the viscoelastic properties at higher temperatures, plate-plate oscillatory rheology was carried out. In Fig. 5a, storage modulus (G’) and loss mudulus (G’’) are reported as a function of frequency at different temperatures. The shape of G’ reflects the relaxation of the system at low frequencies, which becomes faster at higher temperatures, as shown by the increasing slope of the G’ plot in the low-end frequency range. Furthermore, the crossover between G’ and G’’ shifts towards higher frequency values with increasing temperature, namely at 0.6, 1.2, 2.6 and 6.6 rad/s for 120, 150, 180 and 210°C, respectively. This confirms the material shifts toward a dominant viscous behavior with increasing temperature.
These observations were further confirmed by the viscosity plots (Fig. 5b), which show a progressively more defined netwonian plateau with increasing temperature.
As the relaxation of a covalent associative network depends on the dynamics of bond exchanges, stress relaxation tests were also carried out and relaxation times (Fig. 6a) were used to calculate the apparent activation energy for the relaxation, according to Arrhenius plots (Fig. 6b). The results reported in Fig. 6a showed very short relaxation times in the range of a few seconds in the temperature range between 70 and 130°C, suggesting bond exchange to be effective for the network relaxation as soon as cooperative movement of the macoromolecules are possible, i.e., just obove the Tg. The calculated activation energy is approx. 48 kJ/mol, which is the same range as the previously reported systems based on the boronic ester methathesis exchange reaction (Fig. 6c) [22, 42]. The observed extremely rapid stress relaxation behavior can be explained considering the high segmental mobility of the macromolecular chains above Tg and the very fast kinetic of the exchange reactions.
Since recyclability is one of the main advantages of vitrimers over conventional epoxy resins, this property was tested for the system developed in this work. Indeed, the above feature meets the recent demand for circular economy, since recycled materials can be reintroduced in the market and reduce both waste and need of raw materials. To demonstrate recyclability, the developed films were first fractured into small pieces, which were subsequently collected in a rectangular-shaped mold, where hot pressing was applied. The test was performed under a pressure of 3 MPa, which is rather low compared to the pressure normally used for tests with similar systems [31]. After hot pressing at 120°C for 15 min, the fractured pieces were merged again into a film (Fig. 7a). This phenomenon indicates that sufficient bond exchange and chain interpenetration had occurred among the broken pieces. In addition, the TGA curves (Fig. 7b) as well as the FTIR spectra (Fig. 7c) for the recycled films coincide well with those of the neat films, indicating that the chemical structures were maintained after the recycling processes.
The biodegradability of cardanol-based vitrimer was tested in the marine environment, by BOD analysis (Fig. 7d), comparing the behavior of EC-DBEDT_T with that of EC and microcrystalline cellulose, a well-known biodegradable material. The biodegradation of EC-DBEDT_T was first detected after 22 days of immersion with sea water and within one month BOD reached a value of 1.3 mg O2/100 mg. Conversely, the biodegradation of the starting cardanol-based epoxy, EC, was detected much earlier (12 days) and showed a higher value of 5.4 mg O2/100 mg at the end of the test (30 days). The initial delay in the biodegradation of the vitrimer described above can be ascribed to its cross-linked structure and the fact that boronic ester bonds must be hydrolyzed before the biodegradation can occur. Nevertheless, it is relevant to underline that after 30 days, the vitrimer sample reached a BOD value of 1.6 mg O2/100 mg, which is similar to that of the reference material, namely the microcrystalline cellulose. The comparison of the behaviour of our material with that of other systems based on cardanol derivatives is not straightforward, since degradation depends specifically on the type of compound used as well as on the degradation procedure applied. For example, the biodegradability of cardonol-based compounds used as plasticizers of poly(vinyl chloride) was found to be higher than that of cardanol when activated sludge was used as microbial biomass for the test [43], proving the tendency of the above-mentioned materials to be easily biodegraded. Moreover, the results obtained with our materials are very interesting as they show that the incorporation of a cardanol-based compound into a dynamic network based on hydrolyzable boronic esters does not limit its ability to decompose under environmental conditions.
Another interesting property of the developed materials that was evaluated was the flammability behavior. In particular, the behavior of the vitrimer was compared with a permanently crosslinked system prepared with a conventional crosslinker, i.e., Pripol 1006. Figure 8 shows snapshots of the test specimens during combustion. These preliminary results indicated a significant difference between the two materials. While the vitrimer film burned without dripping, leaving a significant charred residue on the tongs, the combustion of Pripol-based epoxy lead to a significan dripping with no residue. As dripping is correlated to flame propagation in a real fire scenario, stopping the detachment of inflamed parts of the specimen is indeed interesting and may be attributed to the presence of the boron-based compounds, as demonstrated by TGA and IR analysis, which are known to be potentially active flame retardant for polymers and for epoxy resins in particular [44–47].