Bio-Based Polyurethane Composites from Macauba Kernel Oil: Part 1, Matrix Synthesis from Glycerol-Based Polyol

Abstract

Polyurethanes are the result of a reaction between an isocyanate and a polyol. The large variety of possible reagents creates many possible polyurethanes to be made, such as soft foams, rigid foams, coatings, and adhesives. This polymer is one of the most produced and consumed polymers in the world with an ever-increasing demand. Despite its usual petrochemical nature, research on bio-based polyurethanes flourishes due to the ease in creating bio-based polyols. This work covers the synthesis of a novel macauba kernel oil polyol by the epoxidation of the oil, followed by a ring-opening reaction of the epoxide with glycerol, used for the preparation of polyurethane foams using different NCO/OH ratios. The FTIR and H¹ results confirm the formation of the epoxide and polyol, and the polymers in all NCO/OH ratios were confirmed by FTIR, showing great similarities between the samples, especially PU 1.0 and PU 1.2. Despite the TGs showing close behaviors for the three samples, their DTGs showed great difference between the samples, with PU 1.0 presenting a regular PU DTG profile with three degradation peaks while the other two sample presented five degradation peaks, indicating a higher crosslinking density in them.

1. Introduction

Every year, around 300 million tons of plastics are produced in the world [1], most of it made from petrochemical raw materials. Plastics, however, due to their petrochemical nature, may take many centuries to degrade, which has become a well-known problem for the environment and health of both animals and plants, pressing industries, people, and governments to search for less harmful alternatives [2,3]. Among the kinds of plastics produced and consumed yearly, polyurethanes (PU) are one of the most versatile, being possible to make into soft foams, rigid foams, films, adhesives, and many other applications. Despite PUs being usually mostly made from petrochemical raw material, they have the advantage of being easily produced from bio-renewable compounds, such as vegetable oils [4].

Vegetable oils are usually seen as promising replacements for petrochemical materials as they are readily available from many different sources, and with variable structural features. Most oils are easy and cheap to obtain and process [5,6]. They are particularly useful for the synthesis of PUs and constitute one of the two main reagents for the synthesis of polyols. There are many different synthetic routes for vegetable oil-based polyols, leading to different PUs [7]. Macauba kernel oil (MKO) is one of the two main oils that can be extracted from the macauba (Acrocomia aculeata) fruit. Macauba is the fruit of the macauba tree, found throughout South America, especially in Brazil. Macauba cultivation is still not as well-studied and optimized as other more common crops, such as soybeans, and is usually done by family farming with the fruits being picked up after they fall from the trees (palms that grow up to 30 m tall) [8]. Macauba oil is rich in short- to medium-chain fatty acids and is mostly used in the biofuel industry, with health benefits being studied due to its content of carotenoids and tocopherols.

Epoxidation is a well-studied chemical modification for vegetable oils, and the epoxidation of MKO was studied in a previous work of the authors where the oil was epoxidized to study the effect it would have on its physical chemical properties [9]. The work has shown an increase of approximately 1.8% in density, 12 °C in T_onset and 74.92% in oxidative stability caused by the epoxidation of the MKO. This paper can be seen as a continuation of that work.

In contrast to MKO, macauba pulp oil (MPU) is more commonly used and studied, probably due to the kernel of the fruit being confined inside a hard endocarp, making it hard to access. While MPU is rich in oleic acid (~69%), MKO is largely made from saturated fatty acids, with lauric acid usually making up from 32.58% to 44.14% of its composition, and oleic acid representing 18.7–36.27% of the fatty acid chains [10,11]. The lower unsaturation index of MKO makes it harder to modify and functionalize than MPU. Despite being usually neglected for oil production, the macauba kernel is actually richer in oil than the pulp (40–50% oil in the kernel versus 28–30% in the pulp) [12]. Additionally, the leftover cake after oil extraction from the kernel is nutritious, containing about 17.73% in proteins and 41.48% in fibers [13], which makes it a higher-quality feed for animals and humans in comparison to pulp cake (4.11% protein and 36.73% fibers) [14]. A more in-depth study of macauba can add value to the fruits, generating more income for the families that make a living from this crop [15].

Glycerol (or propane-1,2,3-triol) is a tri-functional alcohol that is present in all vegetable oils as the backbone of their triglyceride molecules and can be easily obtained as a byproduct of the transesterification of vegetable oils during the production of biodiesel. The rapid increase in demand for biofuels, caused by the rise in environmental awareness, created such a big surge in glycerol production that its usual applications, such as cosmetics, animal feed, fermentation, and food packaging cannot keep up with it. In 2021, approximately 2.8 billion liters of glycerol were produced by only five of the biggest biodiesel producers in the world [16]. Global glycerol production will most likely keep on increasing for the foreseeable future, partly due to governmental efforts to popularize biodiesel (e.g., the PROBIODIESEL program in Brazil, which mandates that ever-increasing proportions of biodiesel should be mixed with regular diesel, starting in 2008 with 3% and aiming at 20% by 2030). Despite the already mentioned supply of glycerol having already reached the order of 2.8 billion liters in 2021, it is estimated that the demand for glycerol in 2025 will be only about 4 million tons. This drastic difference can lead to massive price drops [17]. Other than the environmental interest in using glycerol for this reaction, using glycerol to open the epoxide ring makes it possible to add three -OH for each unsaturation, in contrast to only two obtained when using other mono-functional nucleophiles. Therefore, the use of glycerol as a ring-opening agent promotes the conversion of an oil with a low degree of unsaturation, such as macauba oil, into a suitable polyol for polyurethane synthesis.

The excessive glycerol supplies sparked research interest in its valorization. Ben et al. [16] reviewed the pros and cons of using glycerol as a plasticizer in starch films for food packaging, noting how the hydrophilic nature of the molecules interrupts hydrogen bonds between starch molecules, working as a lubricant between them. Calderon et al. [18] used glycerol acetate and coconut oil to synthesize thermoplastic PUs. While the integration of glycerol in polymers by itself is not new, using glycerol as an epoxide ring opener, the simple and straightforward way to produce polyols used in this work, is still novel. Polyol can be produced from a vegetable oil by the epoxidation of its oil, followed by the opening of the epoxide ring by water in a one-step reaction. The work has for objective the use of polyol synthesized by ring-opening reaction of epoxidized macauba kernel oil (MKO), employing glycerol as the ring-opening agent for the production of a novel polyurethane to be used as the polymeric matrix of thermally treated wood composites. The composites will be presented and discussed in a separate article. The PUs produced in this work, despite having different NCO/OH ratios, presented similar results in relation to the analysis by Fourier-transform infrared spectroscopy (FTIR), thermogravimetric (TG) and dynamic-mechanical analysis (DMA), especially the PU 1.0 and 1.2 samples. Their derivative thermogravimetric (DTG) curves, however, were quite different, with PU 1.0 showing a DTG profile with the usual three degradation peaks of PUs (urethane bond, rigid segments and soft segments) [19], while PU 0.8 and PU 1.2 presented five degradation peaks, which were likely caused by the uneven quantities of NCO and OH creating a higher density of crosslinking in these samples, which results in segments of intermediate hardness and softness degrading at different temperatures.

2. Materials and Methods

2.1. Materials

MKO was purchased from Central do Cerrado^TM, Brasília, Brazil. Formic acid 85%, hydrogen peroxide, sodium carbonate and sodium hydroxide were procured from Dinânimca^TM, all from São Paulo, Brazil. Glycerol was obtained from Synth^TM, while boron trifluoride etherate (BF₃·OEt₂) was acquired from Fluka, Germany. Chloroform-D 99.8% and sodium carbonate were purchased from Sigma-Aldrich, St. Louis, MO, USA. 4,4′-diphenylmethane diisocyanate (MDI) was procured from Dow Chemical Brazil. Sodium chloride was obtained from Brazil, as a commercial salt from the Cisne brand.

2.2. Epoxide Synthesis

The first step for the epoxide synthesis was the determination of molar mass and unsaturation index of the vegetable oil that is going to be used. The reagent quantities used for the epoxidation of MKO were calculated based on unsaturation per mole of vegetable oil, as determined by the integration of the peaks in ¹H nuclear magnetic resonance spectroscopy (¹H NMR) spectrum for olefinic protons (5.4 ppm), methylene protons in the glyceryl group (4.2 ppm), α-methylene protons adjacent to the carbonyl carbon (2.3 ppm), allyl methylene protons (2.1 ppm), β-methylene protons from the carbonyl carbon (2.6 ppm), methylene protons on saturated carbons (2.25 ppm), terminal methyl protons (0.85 ppm) and bisallylic hydrogens (2.79 ppm). It is worth noting that the chemical shifts listed are approximate in quantitative terms [20,21]. The reagent quantities listed on

Table 1. Quantity of reagents for the epoxidation reaction.

	MKO (g)	Formic Acid (mL)	H2O2 (mL)
Quantity	55.55	53.57	200.19

for epoxide synthesis are empirical values derived from previous work [9].

For the synthesis of epoxidized MKO, formic acid and MKO were added to a 500 mL round-bottom flask. Hydrogen peroxide was subsequently added dropwise into the reaction mixture under mechanical agitation. The temperature was kept at 60–65 °C for 2 h. The reaction mixture was then transferred to a separatory funnel, where the aqueous phase is removed, followed by two extractions with a saturated NaCl solution approximately the same volume of the original aqueous phase of the mixture. The reaction mixture was then neutralized with a 5% (m/v) Na₂CO₃ solution, usually talking about 9 mL, and after removing the aqueous phase, the product was filtered to remove impurities and dried in a rotary evaporator. Finally, the product was washed with deionized water to remove residuals salts. Each extraction step was allowed to decant overnight to improve separation results.

2.3. Polyol Synthesis

Epoxidized MKO and glycerin, at 1:1 proportion (40 mL of each) were refluxed in a 2-neck round bottom flask under slow magnetic agitation. BF₃·OEt₂ (10% v/v of the epoxide) was then slowly added dropwise into the reaction mixture over the course of 20 min through the second neck of the flask in the proportion of 10% the volume of epoxide. The dropwise, slow addition ensures the reaction proceeds in a controlled fashion. Since each drop of BF₃·OEt₂ reacts readily with the other reaction contents, there is no build-up in its concentration, therefore allowing for a smooth synthesis. After BF₃·OEt₂ is added, the reaction mixture’s viscosity increases significantly. At this point, diethyl ether is added to reduce the viscosity and ensure proper mixing of all components for the duration of the reaction. The reaction mixture was agitated at room temperature for 5 h and 40 min. The agitation speed was slowly adjusted until the reaction mixture homogenizes. The reagent proportion used in this reaction was 1:1:2 (epoxidized MKO–glycerin–diethyl ether). After a homogeneous mixture was obtained, it was transferred to a separatory funnel and decanted overnight, following neutralization with a 10% (m/v) NaOH solution. Subsequently, 2 extractions were carried out with a saturated NaCl solution approximately the same volume as the glycerol added for the reaction, followed by filtration and water evaporation.

2.4. Polyol Hydroxyl Index

The hydroxyl index (OH index) of the polyols was determined according to the Standard Test Method for Hydroxyl Value of Fatty Oils and Acids, ASTM Standard D1193 [22] Specification for Reagent Water, an acetylation of the polyol with acetic anhydride followed by a titration with potassium hydroxide. The method was based on the acetylation of the polyol 9 g to 11 g by pyridine-acetic anhydride solution, and 25 mL of neutralized n-butyl alcohol, which were mixed for 30 min under water reflux. The mixture then was titrated with 0.5 M alcoholic KOH using phenolphthalein indicator. The measurements were made in triplicate, resulting in an OH index of 85.57 mgKOH·g⁻¹ for the GlyMKO.

2.5. Polyurethane Synthesis

For the synthesis of polyurethanes (PUs), polyol and MDI were mixed in 0.8, 1.0, and 1.2 NCO/OH ratios. The reagent quantities for each PU are shown in

Table 2. Quantities of reagents for the PU synthesis.

NCO/OH Ratio	Polyol (g)	MDI (g)
0.8	10	1.71
1.0	10	2.14
1.2	10	2.57

. This ratio was calculated using the previously determined free NCO index for aromatic isocyanates ASTM D 5155-96 [23] on the available MDI and the OH index in the polyol. The polyol was firstly added in a Teflon beaker, followed by the NCO. The reagents were then mixed by mechanical stirring for one min. The mixture was poured into Teflon molds and then left to cure in an oven pre-heated at 65 °C for 24 h. The final products were identified and characterized.

2.6. Fourier-Transform Infrared Spectroscopy

Fourier-transform infrared (FTIR) spectroscopy analyses were performed on a Varian 640 spectrometer equipped with an attenuated total reflectance (ATR) accessory. The spectra reported correspond to an average of 16 scans with a spectral window from 400 cm⁻¹ to 4000 cm⁻¹ with a normal resolution of 4 cm⁻¹. The presented spectra have been normalized by the 2900 cm⁻¹ peak and all spectra were analyzed with Origin 9.0 software.

2.7. Thermogravimetric Analysis

The PU samples from 5 mg to 10 mg (aiming near 7 mg) were analyzed by thermogravimetric analysis (TGA) and its derivative (DTG) under synthetic air atmosphere on a NETZSCH STA 449F3 analyzer. The TGA analysis temperature profile is as follows: raising the temperature from room temperature to 103 °C, 20 °C·min⁻¹, followed by a 30 min isothermal, after which the temperature is raised to 1000 °C at the same temperature rate. The temperature is then decreased to room temperature at −50 °C·min⁻¹. The whole process happens under 100 mL·min⁻¹ synthetic air atmosphere.

2.8. Dynamic Mechanical Analysis

Dynamic mechanical analysis (DMA) was performed on a Q800 (TA Instruments, New Castle, DE, USA) using a three-point bend fixture. Specimens of size 5.0 mm × 9.0 mm × 5.0 mm (length × width × thickness) were cut and initially cooled to 0 °C (in preliminary tests, some DMA specimens broke before the start of the experiments when temperatures below 0 °C were employed, limiting the temperature range used), followed by heating to 150 °C at a heating rate of 3 °C·min⁻¹ under an iso-strain mode with an amplitude of 14 µm and a frequency of 1 Hz. Two replicates of each sample were tested individually, and the results disclosed herein represent the average of the two trials. The crosslink density (ε) was calculated based on the rubber elasticity theory according to Equation (1) [24,25], where E’ is the storage modulus at temperature T and R is the gas constant.

$$\mathsf{\epsilon }={\mathrm{E}}^{\prime }{\left(\mathrm{RT}\right)}^{-1}$$

(1)

3. Results and Discussion

3.1. Vegetable Oil Molar Mass and Degree of Unsaturation

Table 3. Properties of MKO.

	Molar Mass	Unsaturation Index
MKO	660.03 g·mol−1	0.85

shows the results for the calculation of unsaturation index based on a procedure well described in the literature [20,21]. The low molar mass and unsaturation index calculated can be explained by the most significant fatty acid in MKO being lauric acid, a short (12 carbons) and saturated fatty acid, which makes up approximately 40% of the oil’s fatty acid composition. The most important fatty acid in MKO is oleic acid, a monosaturated acid that makes up approximately 23% of the oil composition.

Table 4. Main fatty acids present in MKO [10,11].

Fatty Acid (Cx:y) a	Percentage (%)
Lauric acid (C12:0)	32.58–44.14
Oleic acid (C18:1)	18.70–36.27
Myristic acid (C14:0)	8.45–13.60
Palmitic acid (C16:0)	6.57–9.20

^a Cx:y corresponds to the length and unsaturation of each specific fatty acid, where “x” is the number of carbon atoms in the fatty acid chain, and “y” is the number of carbon-carbon double bonds present in the fatty acid.

displays the main fatty acids present in MKO [10,11].

3.2. Hydroxyl Index

The average OH index, calculated for the polyol from epoxidized MKO (EPMKO) and glycerol-based polyol (GlyMKO), was 85.57 mgKOH·g⁻¹, a relatively low value, in line with the low content of unsaturated fatty acids present in MKO, as shown in Table 4. Since the final hydroxyl groups originate from the ring-opening of epoxide groups created on carbon-carbon double bonds in the triglyceride, it is expected that a lower degree of unsaturation results in a lower OH index.

3.3. Reaction Mechanism

The general scheme for the synthesis of polyol in shown in Figure 1, with more detailed mechanisms for each step presented in Figure 2, Figure 3, Figure 4 and Figure 5. Figure 1 shows the progress from triglyceride to epoxide then to the polyol. The one-step epoxide synthesis reaction from the triglyceride begins with the formation of the epoxidizing agent, the performic acid. The proposed mechanism for the peracid formation is shown in Figure 2.

The performic acid is formed by the reaction of formic acid with H₂O₂. Rubio et al. [26] proposed two possible mechanisms for this reaction, called Route A and Route B in Figure 2, both considered equally possible. Route A begins with the nucleophilic attack of the peroxide at the carbonyl’s carbon, resulting in both a positively and a negatively charged oxygen, the charges of which are neutralized by a proton exchange between them, followed by the release of an intramolecular water [26]. Route B starts with an acidic activation of the carbonyl, which is then attached by the peroxide, breaking its double bond. The double bond is then reformed, with OH⁻ acting as leaving group, followed by the loss of H⁺, resulting in the peracid [26]. After the formation of the peracid, the actual epoxidation can take place,; the proposed mechanism is shown in Figure 3.

The epoxide formation begins with the unsaturation of the vegetable oil attacking the hydroxyl oxygen in the peracid, which is activated by an intramolecular hydrogen bond, forming the transition state shown in the figure, which breaks down into the epoxide and a molecule of formic acid that can be used to form one more peracid molecule [27].

With the formation of epoxide, it can undergo a ring-opening reaction to form the polyol. The polyol was synthesized in this work via a two-step reaction. The epoxidation of the oil was followed by epoxide ring-opening with glycerol. The epoxidation reaction is well documented in the literature [26]. The reaction begins with the formation of performic acid, followed by the transfer of one oxygen atom to the fatty acid unsaturation. Epoxide ring-opening by glycerol, while a novelty, likely happens in a similar manner to other epoxide ring-opening reactions described in the literature [7], in which one of the carbons of the epoxide group undergoes a nucleophilic attack in acidic conditions. In this work, glycerol was used as the nucleophile, while BF₃·OEt₂ acts as an acid catalyst. It is believed that the mechanism for this reaction follows the route illustrated in Figure 4, resulting in three new hydroxyl groups for each unsaturation in the oil.

After the purification of the polyol, its reaction with MDI, an isocyanate, leads to the formation of the PU. The proposed reaction mechanism for this reaction is shown in Figure 5.

As shown in Figure 5, the first step of this reaction is the nucleophilic attack of the -OH oxygen at the carbonyl’s carbon, resulting in the breaking of the nitrogen pi bond to carbon, to maintain its tetravalency, followed by the transference of H⁺ from the oxygen to the nitrogen to form the final product [28]. A usual hydroxylation, using water as the epoxide opening agent, leads to a polyol with only two hydroxyls per unsaturation. The method used in this allows the formation of three hydroxyls per unsaturation, which could give the opportunity for vegetable oils with low iodine index, such as MKO, to be used in applications that require larger quantities of hydroxyl. Also, the increased number of -OH groups was obtained using glycerol, a biomolecule that is often seen as a waste product. Although strong catalysts such as BF₃·OEt₂ will have to be used to achieve the desired reaction, it is possible to test other catalysts as this research progresses.

3.4. Fourier-Transform Infrared Spectroscopy

Figure 6 displays the FTIR spectra of MKO, EPMKO, and GlyMKO. The MKO spectrum indicates a small unsaturation index, with hardly visible signals at 3010 cm⁻¹ and 1600 cm⁻¹, corresponding to H-C (sp²) and C=C vibrations, respectively. The two small epoxide peaks at approximately 750 cm⁻¹ and 960 cm⁻¹ in the FTIR spectrum of the EPMKO indicate the successful formation of the epoxide ring [29]. Figure 6 also clearly shows the appearance of an O-H peak at 3500 cm⁻¹, as well as the decrease in the epoxide peaks in the FTIR spectrum of the GlyMKO, indicating that the epoxide ring-opening reaction happened as expected.

Figure 7 represents the normalized FTIR spectra of all PUs synthesized in this work, labeled for their NCO/OH ratio. The N-H peaks (~3700 cm⁻¹) and the C-N peaks (~1500 cm⁻¹) seen in all three spectra have similar intensity, with nearly identical spectra. The similarity between the spectra of the three samples is an indication that the changes in NCO/OH molar ratios investigated did not lead to differences that could be easily observed via FTIR spectroscopy. Although minor changes can be seen in specific areas and peak intensities, most notably at approximately 1500 cm⁻¹ and in the fingerprint region (~300 cm⁻¹), the main signals from the oil (Csp³-H at 2800–3000 cm⁻¹ and C=O at 1750 cm⁻¹) are dominant in all FTIR spectra presented in Figure 7. From a logical standpoint, when NCO/OH > 1.0, the final sample should have an excess of NCO, whereas NCO/OH < 1.0 should result in excess OH. Although not prominent, the peak corresponding to O-H is clearly visible in all spectra included in Figure 7 from approximately 3100 cm⁻¹ to 3600 cm⁻¹. It is possible that, due to its high hydroxyl content, the polyol may have absorbed moisture while exposed to air during filtration, PU synthesis, and/or cure. This residual moisture can react with excess NCO, limiting its detection by FTIR spectroscopy.

In the digital image of the PU 1.2 (Figure 8a), small orange-brown regions can be seen throughout the sample, probably related to a higher density, resulting from the excess NCO present in PU 1.2, providing more bond formation urethanes. In the FTIR spectrum of PU 1.2 (Figure 8b), the absence of the strong peak at 2200 cm⁻¹, associated with the vibrations of the N=C=O bond [30,31], indicates denser regions of PU 1.2, instead of unreacted MDI.

3.5. Proton Nuclear Magnetic Resonance

Figure 9 represents the proton nuclear magnetic resonance (¹H NMR) spectra of MKO, EPMKO, and GlyMKO. The progress of the transformation can be clearly followed from the triglyceride to the polyol. Signals referring to C=C-C-H (2 ppm) and C=C-H (~5.35 ppm) can be seen in the ¹H NMR spectrum of the MKO but are significantly reduced in the ¹H NMR spectra of EPMKO and GlyMKO, indicating that the majority of the carbon-carbon double bonds have been consumed during the epoxidation, confirming the FTIR results. The incomplete conversion of carbon-carbon double bonds to epoxides during epoxidation is fairly common with triglycerides. The appearance of the signals at 1.50 ppm and 2.76 ppm in the ¹H NMR spectrum of the EPMKO, referring to the oxirane ring, shows that the epoxidation reaction was successful [32,33]. Its subsequent disappearance in the ¹H NMR spectrum of the GlyMKO suggests the successful ring-opening reaction. As typically observed with labile hydroxyl groups [34], the OH signal in the ¹H NMR spectrum of GlyMKO is barely seen, although the OH band is clearly seen in the FTIR spectrum of GlyMKO (Figure 2). This shows the successful formation of the polyol. It is worth noting the appearance of a signal at 2.2 ppm after epoxidation (Figure 9). That signal is believed to be the result of a possible side reaction involving the carbonyl of the triglyceride under the acidic conditions employed during the synthesis.

3.6. Thermogravimetric Analysis

The three PU samples prepared in this work were also analyzed by TG (Figure 10a) and DTG (Figure 10b). TG curves show a mostly similar thermal behavior under synthetic air atmosphere during heating.

Table 5. T_onset, T₅ and T₁₀ values for the PUs 0.8, 1.0 and 1.2.

Sample	Tonset	T5	T10
PU 0.8	339.2	272.3	326.8
PU 1.0	332.3	254.6	314.3
PU 1.2	332.7	249.6	313.8

lists the onset temperature (T_onset) and the temperature at which the polymer mass loss reaches 5 wt.% and 10 wt.% (T₅ and T₁₀, respectively).

Besides the initial loss of moisture and volatiles under 100 °C, PUs usually have three decomposition stages under air, beginning with the breaking of urethane bonds, followed by the degradation of hard segments, starting usually at around 250 °C, and then the degradation of soft segments, at around 310 °C, as the temperature increases [19]. Although these stages are not easy to observe in the TG curves, they are evident in the DTG curves, including showing that PU 0.8 and PU 1.2 present four stages of decomposition. That has likely happened because using a glycerol-based polyol allowed more crosslinking between the diisocyanate and the three different hydroxyl groups on the resulting polyol (one primary and one secondary alcohol on the glycerol moiety), as well as a secondary alcohol on the fatty acid backbone (Figure 1), creating a mixture of hard and soft segments with more complex degradation profiles, as previously noted in the literature [35]. The sample of PU 1.0 exhibits a much more regular thermal degradation profile with only three degradation stages and a small shoulder on the third stage, indicating that the uneven quantities of NCO and OH might be related to atypical thermal degradation behavior. The T_onset shows little correlation to the NCO/OH ratio, as seen in the literature [19].

The T_onset value is nearly the same for all three samples, differing only by 0.4 °C. PU 1.2 has the lowest T₅ and T₁₀, exhibiting a slightly lower thermal stability. T_onset represents the temperature at which the most significant thermal degradation step of a sample begins to take place. The higher quantity of hard segments in its polymeric matrix present on the samples with higher NCO/OH ratio, in agreement the appearance of orange-brown regions mentioned in the FTIR discussion, contributes to lowering the T_onset. PU 0.8 exhibits the highest T_onset, T₅, and T₁₀, likely caused by larger number of soft segments, related to the deficit of NCO. This could also explain why samples with higher NCO/OH ratios had lower T₅ and T₁₀. Furthermore, PU 1.2 showed the lowest mass loss at temperatures above 400 °C, probably because it has fewer soft segments to degrade at higher temperatures. The opposite effect was observed for PU 0.8. The PU 1.2 T_onset is probably caused by the bigger quantity of crosslinking in this sample which might have increased its thermal stability enough to compensate for the lower thermal degradation temperature of its hard segments.

3.7. Dynamical Mechanical Analysis

DMA is a method for analyzing the viscoelastic properties of flexible PU foams [36].

Table 6. DMA results of PU samples.

Sample	E’ at 25 °C (MPa)	E’ at 100 °C (MPa)	ε (µmol·cm−3)	Tg 1 (°C)	Tg 2 (°C)
PU 0.8	0.30	0.012	3.87	51	96
PU 1.0	0.36	0.030	9.67	48	91
PU 1.2	0.39	0.024	7.74	56	104

summarizes the DMA results for the PUs obtained in this work, including the storage moduli (E’) at both 25 °C and 100 °C, as well as the glass transition temperature (T_g) as determined from the peak of the tan ∂ curves, and the corresponding crosslink densities for each sample. A slight increase in E’ at 25 °C is observed for samples with increasing NCO/OH ratios, correlating well with the increase in hard segments discussed with the TGA results. At 100 °C, all samples have attained their rubbery plateau, as illustrated in Figure 11a. At this temperature, E’ for PU 1.2 deviates from what seemed to be a growing tendency with the NCO/OH molar ratio. That happens because MDI has only two NCO groups, while the polyol has three OH groups. Therefore, the polyol is the major contributor to the formation of crosslinks. This suggests that the excessive NCO content might be detrimental to the mechanical properties and negatively affect crosslink density. Indeed, the TG results indicated that more soft segments were formed with an excess of NCO. These soft segments may act as plasticizer in the sample, allowing for the dissipation of energy imparted during deformation in the form of chain movement.

The T_gs measured from the peak of the tan ∂ curves do not follow any particular trend. From Figure 11b, the lack of well-defined peaks is indicative of the heterogeneity of the samples, revealing a network of broad composition across the samples tested. The presence of a secondary T_g in all samples indicates the predominance of two major phases in the PUs studied. It is possible that the increase in T_g for NCO/OH molar ratios different than 1.0 corresponds to effects in different segments of the sample. While soft segments are primarily affected by excess NCO (NCO/OH > 1.0), hard segments are primarily affected by excess OH (NCO/OH < 1.0). Therefore, whenever NCO/OH deviates from 1.0, an increase in T_g is observed due to enhancement of either soft or hard segments in the sample. Ultimately, when NCO/OH < 1.0, reaction of the NCO groups is maximized, while NCO/OH > 1.0 primarily affects soft segments by maximizing reactions of the polyol.

Because the diisocyanate used (MDI) bears only two functional groups while the polyol resulting from the ring-opening reaction of the epoxidized oil with glycerol bears three functional groups (Figure 4), the polyol is the major contributor to crosslinking in the system. It would therefore be expected that a lower NCO/OH molar ratio would result in unreacted hydroxyl groups and therefore a lower crosslink density, leading to lower storage moduli. This relationship is well respected for PU 0.8 and PU 1.0 throughout the temperature range investigated in Figure 11a. In theory, when excess diisocyanate is used (NCO/OH > 1.0), all hydroxyl groups should react, leading to higher crosslink densities and higher storage modulus. This is only observed between 40 °C and 90 °C (Figure 11a). It is possible that, at lower temperatures (before the first T_g), the storage modulus is also affected by hydrogen bonding. Therefore, PU 1.2 (bearing excess NCO and presumably having all -OH groups reacted) exhibits a substantially lower storage modulus than PU 0.8 and PU 1.0 because of its inability to form hydrogen bonds.

4. Conclusions

The unique structure obtained from the ring-opening of epoxidized MKO with glycerol is unprecedented and resulted in PUs with four distinct thermal degradation stages, as shown in Figure 10, instead of the typical three thermal degradation stages observed for other vegetable oil-based polyurethane foams. While the use of glycerol to produce polymers is not new, as demonstrated by the production of polyols, plasticizers, and chain extenders, the use of an epoxide ring-opening reaction to insert glycerol into a triglyceride molecule results in a novel polyol (GlyMKO). This polyol provides new and excellent types of PUs, confirmed by FTIR and ¹H NMR spectroscopies and with unique thermal properties, shown by TG and DTG. The FTIR data showed that both the epoxide formation and the epoxide ring-opening reaction were successful and evidenced by the disappearance of signals associated with the presence of carbon-carbon double bonds in MKO accompanied by the appearance of epoxide peaks in EPMKO, and their subsequent disappearance in GlyMKO along with the appearance of a large OH band in the FTIR spectrum. Those results agree with other reports in the literature. The FTIR spectra also revealed that, although different NCO/OH ratios were used to obtain the PUs, there were no significant differences in the final products, especially between PU 1.0 and PU 1.2. Despite this, PU 1.2 exhibits lumps not present in considerable quantities in the other samples, which correspond to denser regions within the sample. The three samples prepared presented similar thermal behavior under synthetic air atmosphere, showing a difference of a few degrees in T_onset, T₅, and T₁₀, with PU 0.8 having higher temperatures and PU 1.2 lower temperatures, except for T_onset. Notwithstanding that the PUs have similar thermal stability, the samples presented different behaviors in relation to DTG curves. PU 1.0 followed the known degradation profile seen in the literature for PUs, with three well-defined degradation stages. The first stage corresponds to degradation of urethane bonds, followed by rigid segments, and lastly by soft segments. The other two samples (PU 0.8 and PU 1.2) present four degradation stages, likely caused by different levels of rigidity and softness associated with changes in crosslink density due to the uneven quantities of NCO and OH, resulting in complex thermal degradation profiles. The lower temperatures for the rigid segments’ thermal degradation might also explain why PU 1.2 exhibits the lowest T_onset, T₅, and T₁₀, while PU 0.8 showed the highest values. DMA analysis was used to calculate the crosslink density of the samples and help explain their thermo-mechanical behavior. The PUs obtained in this research are excellent candidates as a matrix material for bio-based composites. The reinforcement of these materials with thermally treated wood biomass will soon be investigated by the authors. It is hypothesized that excess NCO in the PUs can react with hydroxyl groups on the surface of wood to enhance matrix-reinforcement interactions and maximize mechanical properties.