Fully bio-based aliphatic thermoset polyesters via self-catalyzed self-condensation of multifunctional epoxy monomers directly extracted from natural sources

EFA was retrieved according to previously described methods in the literature with minor changes to optimize the yields as described below. The retrieval of EFA was divided into two steps where the first step was a removal of extractives performed according to a method described in the literature.30 The residue of the outer birch bark powder was left to dry: first in air for 24 h and then in a vacuum oven at 50°C for 24 h. The second step was conducted by a modified hydrolysis process previously described in the literature31,32 where the dry ground outer birch bark was weighed into a round-bottom flask and impregnated by reduced pressure with 0.8 M NaOH solution in the ratio of 1 g bark per 10 mL solution. The round-bottom flask was then immersed into a 100°C oil bath and refluxed under continuous stirring. After 1 h, the round-bottom flask was allowed to cool down to room temperature and the solid residues was separated and removed by centrifugation. The obtained supernatant was acidified with a 5% H₂SO₄ solution to pH 5.7, and a precipitate was formed. The received precipitate was then separated from the solution by centrifugation and a clay-like crude was obtained. The crude product was subjected to 4 washing steps with 100 mL deionized water on each step. Crude and water were separated by centrifugation each time. Finally, in order to purify the EFA monomer, the crude was recrystallized from toluene. A slightly yellow-colored powder was obtained with a yield of 60–70%. ¹H NMR (400 MHz, CDCl₃, δ): 3.65 (2H, t, –CH ₂ –OH, J = 6.59 Hz), 2.90 (2H, bs, –CH–O–CH– epoxide), 2.35 (2H, t, –CH ₂ –CO–OH, J = 7.45 Hz), 1.66–1.26 (26H, m, –CH ₂ –).

The EMO monomer was obtained by a transesterification reaction according to procedures found in the literature.33 Twenty grams of epoxidized linseed oil was dissolved in 248 mL of 0.02 M NaOH in methanol. The solution was immersed in a 60°C oil bath and refluxed for 1 h. The formed methyl esters were then extracted 4 times in an n-heptane/deionized water system (1:1). The organic phase was then dried with MgSO₄ and the solvent was evaporated by rotary evaporator. The obtained methyl esters had a yield of 70.5 wt%. Thin-layer chromatography showed the presence of 4 different substances having R _f values of 0.44, 0.20, 0.04, and 0 in 10/90 (vol%) ethyl acetate/n-heptane. The methyl ester mixture was then purified via column chromatography using silica gel and an n-heptane/EtOAc gradient elution. ¹H NMR proved that the substance having R _f value of 0.20 was EMO monomer. ¹H NMR (400 MHz, CDCl₃, δ): 3.66 (3H, s, –O–CH ₃ ), 2.89 (2H, bs, –CH–O–CH– epoxide), 2.30 (2H, t, –CH ₂ –CO–, J = 7.5 Hz), 1.62 (2H, p, –CH₂–CH ₂ –CH₂–CO– J = 7.1 Hz), 1.54–1.20 (26H, m, –CH ₂ –).

The other substances were confirmed to be methyl stearate (R _f 0.44), methyl ester with 2 epoxides (R _f 0.04) and methyl ester with 3 epoxides (R _f 0). These methyl esters were not used in this study.

A series of model experiments were performed to evaluate the different possible reaction pathways of EFA at elevated temperature. Different combinations of LA, LOH, and, EMO were evaluated with respect to intrinsic reactivities at elevated temperatures. Table 1 presents the different compositions of the test reactions.

The monomers were weighed into a 5-mL round-bottom flask containing a magnetic stirrer and then sealed with a septum. The round-bottom flask was then purged with argon (g) for approximately 20 min before being immersed in an oil bath at 150°C for 24 h. Samples for NMR analysis were collected before and after each reaction. One reaction (Table 1, entry 5) containing only EFA was polymerized for 2 h. The reaction gelled after 50 min, and NMR samples were retrieved every 10 min up to 50 min before gelation occurred.

All reactions were also studied using real-time FTIR analysis monitoring chemical changes with time on specimens at set temperatures (120, 135, or 150°C) for 60 min. Reaction no. 3 containing LA and LOH was also run for 120 min, and reaction no. 5 containing only EFA was also run for 180 min. DSC analysis on crosslinked EFA was performed on specimens retrieved from crosslinking in the round-bottom flask.

Adhesion properties of EFA were evaluated on different substrates. Two pieces of each substrate (glass, aluminum, steel, Teflon and wood) were glued together to form an overlap joint of approximately 4 cm². All substrates were cleaned with acetone and ethanol before use. Approximately 10 mg (2.5 mg cm⁻²) of EFA was applied onto a preheated substrate (150°C) and then clamped together with the other substrate before it was placed in the oven for 1 h at 150°C.

All NMR spectra were recorded on a Bruker AM 400 at 400 MHz. Deuterated chloroform (CDCl₃) was used as a solvent. The residual CDCl₃ solvent peak was used as reference (δ = 7.26 ppm singlet for ¹H NMR and δ = 77.16 ppm as the central line of the triplet for ¹³C NMR).

Analyses were recorded on a Perkin Elmer Spectrum 2000 FTIR instrument (Norwalk, CT) equipped with a single reflection (ATR: attenuated total reflection) accessory unit (Golden Gate) from Graseby Specac LTD (Kent, England) and a TGS detector using the Golden Gate set-up. Recorded data were processed in Spectrum software from PerkinElmer. RT-FTIR (relative reactivity studies) data were recorded at an optimized scanning rate of 1 scan per 6 s with a resolution of 4.0 cm⁻¹ and were processed on Time Base^® software from Perkin Elmer. The Golden Gate accessory unit was equipped with a temperature control (Specac, Heated Golden Gate Controller). The temperature on the Golden Gate was set to desired temperature and a background scan was made before conducting each experiment. The sample was put directly on the crystal and a plain glass cover slide placed onto the sample. The characteristic peaks from the RT-FTIR measurements were compared with characteristic infrared absorption bands in reference (34).

The thermal analyses were carried out on a Mettler Toledo DSC-1 equipped with Gas Controller GC100. For the analysis, approximately 5–10 mg of each sample were weighed into aluminum crucibles of 100 μL. The data were collected using a heating/cooling rate of 5°C min⁻¹ from −40 to 150°C with 5 min isotherms. All analyses were carried out in nitrogen gas atmosphere of 30 mL min⁻¹. The evaluation of acquired data was done using STARe Excellence Software. The glass transition temperature (T _g) was reported as the midpoint of the heat capacity change and was acquired from the second heating scan.

The epoxide–carboxylic acid reaction (Table 1, entry 1) was studied by both NMR and RT-FTIR. The NMR results showed proof of the ring-opening reaction by the disappearance of epoxide protons and carbon peak g at 2.90 ppm and G at 57.39 ppm, respectively (Fig. 2). It was further seen that two new peaks, i (3.56 ppm) and h (4.80 ppm) on ¹H NMR, and I (72.55 ppm) and H (76.30 ppm) on ¹³C NMR, emerged. These protons were assigned to protons on hydroxyl branched carbon of the fatty acid chain and the ester branched carbon of the fatty acid chain, respectively.35,36

A ring-opening reaction between an acid and epoxide would create a secondary alcohol and an ester.24,25,36 To confirm this, reaction characteristic carbonyl regions for acids (1740–1700 cm⁻¹) and esters (1750–1725 cm⁻¹) were monitored. The reaction was further confirmed by RT-FTIR by the depletion of the absorption band at 1714 cm⁻¹ and an increase in absorption band at 1740 cm⁻¹ which correlates with the free acid and an ester, respectively (Fig. 3). The formed alcohol group was observed by the increase in absorption band at 3530 cm⁻¹. Epoxide compounds have a weak absorption band in the region of 950–750 cm⁻¹. Although a small decrease was seen in that particular area, it was difficult to quantify these changes due to overlapping from other bands in the fingerprint region. The results indicate that the reaction was catalyzed by a protonation of the epoxide ring followed by an attack of a carboxylate anion forming a new ester bond and an alcohol. However, there was no evidence of side reactions such as generation of linear ether groups (e.g., from alcohol attacking a protonated epoxide or cationic homopolymerization of epoxides) on NMR or RT-FTIR. Another possible reaction is the reaction between the secondary hydroxyl formed by the addition of acid to the epoxide reacting further either with the carboxylic acid or the methyl ester. However, the NMR data did not indicate any detectable amount of these reactions occurring under set conditions.

It is well known that hydroxyl groups can perform a nucleophilic attack on epoxides under acidic conditions and form an ether linkage.9 The results from reaction containing hydroxyls and epoxides (Table 1, entry 2) under neutral conditions, however, showed no detectable reaction between the hydroxyl group and the epoxide as studied with NMR or RT-FTIR.

Apart from the protonation of the epoxide, EFA has the possibility of a direct esterification reaction between the acid and the alcohol group. In order to evaluate the possibility of esterification at elevated temperature, a model experiment containing only acid (LA) and alcohol (LOH) was conducted (Table 1, entry 3). The performed experiment suggests that an esterification occurs by the appearance of peaks d′ and p′ at 2.28 ppm and 4.05 ppm, respectively. Characteristic peaks corresponding to the monomers LA (2.35 ppm, peak d) and LOH (3.60 ppm, peak p) both disappear (Fig. 4). The esterification reaction was also confirmed by RT-FTIR by the depletion of the absorption band corresponding to the acid (1715 cm⁻¹) and an increase in the peak corresponding to the ester (1740 cm⁻¹) (Fig. 4).

Comparing the RT-FTIR results of reaction no. 1 and reaction no. 3 (Figs. 3 and 4), it is observed that the ring-opening reaction, adding a carboxylic acid to the epoxide, is faster than the direct esterification reaction. The conversion after 40 min of the ring-opening reaction was over 90% while the esterification only reached approximately 70% after 2 h. Although different in rates, one should note that the differences are within the same order of magnitude.

By knowing the trends from each reaction, it was easier to follow the reaction containing all three different functional groups present, i.e., EMO, LA, and LOH (Table 1, entry 4). As the previous model experiments depicted, a ring-opening reaction and an esterification reaction occurred but this time in parallel with each other, as evidenced by NMR and RT-FTIR (Fig. 5). The relative conversion differences at higher conversions are less than initially foreseen by the tests on reactions no. 1–3. The explanation for this is mainly due to the change in stoichiometry as a function of conversion. A more rapid epoxide–carboxylic acid reaction also leads to a more rapid depletion of the EMO monomer while the relative concentration of LOH increases, thus promoting the direct esterification at higher conversions.