The cellulose used was Microcrystal Cellulose (MCC) powder (OurChem, DP = 235). Special grade trifluoroacetic anhydride (TFAA) and stearic acid (SA) were used for the acylation of cellulose. Reagent grade ethanol and other reagents were used without being further purified.

CTs was synthesized following Morooka’s method [15] with some modifications. TFAA and stearic acid (mole ratio of fatty acid to TFAA was 1.1) were mixed together in a 250 mL three-neck flask equipped with a condenser and stirred mildly at 50 °C for 20 min to form the mixed acid anhydride. Pre-dried MCC powder was added to this solution and stirred mildly at 50 °C for a certain time. During the reaction, chloroform was added as dissolvent. The reaction mixtures were poured into an excess amount of ethanol, and the precipitates were filtered and washed with methanol (50 mL per gram of MCC) three times. Then, Soxhlet extraction with ethanol was carried out for 24 h. Finally, the polymers were dried at 50 °C for 24 h under vacuum to yield products.

The esterification process for CTs preparation involves two steps according to the process used for cellulose acylates [15]. The first stage involves the formation of mixed acid anhydride. The reaction takes place easily at 50 °C. ca. 20 min was enough to complete the first stage. The mixture becomes a pale-yellow color. In the second stage, esterification stage, pre-dried MCC was added into the former solution, and a heterogeneous system was obtained. However, with the development of reaction, a phase transfer can be seen, a homogeneous system gradually formed. The color did not significantly change. Table 1 shows the conditions of synthesis of CTs and the DS of products. The reactions that take place during these stages and the procedure used for the synthesis of CTs are presented in Figure 1. The difference in DS value between sample 1 and sample 3 is 0.15; the increase of DS is due to the addition of chloroform. Research results show that when the degree of substitution of LCCEs is greater than 1.4, the ester can dissolve in chloroform. Thus, the addition of chloroform can dissolve the reactants (except cellulose) and product, make the viscosity reduced and benefit for the contacting of cellulose and mixed acid anhydride. Toluene also has the same effect. The addition of solvents can save reagents effectively, and make a nearly quantitative reaction between cellulose and mixed anhydride reality. The reagents consumed were less than those in reference [15].

A comparative FT-IR analysis of MCC (a), cellulose state (DS = 2.45, b) and CTs (c) clearly shows the relative efficiency of acylation (Figure 2). For cellulose sterate (Figure 2b) and CTs (Figure 2c), a decrease in the intensity and a shift of the characteristic band for hydroxyl groups at 3448 cm⁻¹ were observed relative to the spectra of the MCC (Figure 1a), indicating that a number of OH groups were substituted. In addition to this phenomenon, we also observed an increase in the intensities of the characteristic bands of CH alkyl bonds at 2800–2900 cm⁻¹ (CH anti-symmetric and symmetric stretching of CH₂ and CH₃). There were two new bands in the spectrum: the first band appeared at 1744 cm⁻¹ (C=O stretching), corresponding to the vibration of carbonyl ester groups, and the second band appeared at 722 cm⁻¹ (CH₂ rocking) which is characteristic of at least four linearly connected CH₂ groups.

CTs has a good solubility in CDCl₃, and the ¹H-NMR spectrum in CDCl₃ is shown in Figure 3. The well-resolved spectrum shows a separate signal for all the AGU protons in the region δ = 3.46–5.08 ppm. The protons on the aliphatic chain appear at 0.88 ppm (t, 3H, C18), 1.25–1.54 ppm (m, broad peaks, 28H, C4–C17), 1.61 ppm (m, 2H, C3), and 2.16–2.35 ppm(m, broad peaks, 2H, C2), respectively. The DS of 2.99, calculated from the ¹H-NMR spectrum of the CTs, confirms the fully-esterification of cellulose.

Figure 4 shows dynamic TG curves of MCC and CTs at heating rate of 10 K/min in Ar. The thermal degradation of MCC consists of a series of degradation reaction, such as dehydration at ca. 100 °C, this weight loss was mainly due to the evaporation of bonding water on MCC and the thermal pyrolysis of the cellulose skeleton at ~300 °C. However, there was very little hydroxyl on CTs’ molecular chain, which means that CTs was hydrophobic. Therefore, there has not significant weight-loss at ca.100 °C. The temperature of α equal to 5% for CTs was 320 °C, while that for MCC was only 296 °C. The temperature corresponding to maximum degradation rates (T_p) for CTs was 371 °C, while that for MCC was 344.6 °C. Therefore, thermal stability of CTs is superior to that of MCC. Obviously, the presence of the substituted groups on the cellulose skeleton is expected to influence the thermo-stability. The results can be explained from the difference between the structure of MCC and CTs. The long fatty ester groups on the cellulose skeleton can be in a regular arrangement and form a new ordered structure; even the side chain can form a new crystal structure, so as to improve the thermal stability of CTs. Side-chain crystallization of long n-alkyl side chains on poly(n-alkyl thiophenes), which has stiff backbones, had been found by Hsu et al. [20]. Sealey et al. demonstrated favorable conformation for side-chain crystallization of cellulose eicosanoate (C20) with computer simulation [16]. The char residue of CTs is less than that of MCC. Compared with MCC, stearic acid ester occupies the major proportion in the same quality of CTs, while the ester groups degraded and easily produced volatiles; therefore the char residue of CTs is lower than that of MCC. In summary, the thermal stability of CTs was significantly superior to that of MCC.

Figure 5 shows TG (Figure 5a) and DTG (Figure 5b) curves of CTs corresponding to dynamic experiments carried out at 10, 15, 20, and 25 K/min with argon flow rate 25 mL/min. It is clear from the DTG plots that the T_p values shift upwards with increasing heating rate. An increase of 19 °C in the initial thermal decomposition temperature is measured. Obviously, there is a strong influence of heating rate for the degradation of CTs.

In order to analyze more deeply the degradation mechanism of CTs, it is important that activation energy E and conversion function g(α) are evaluated. The relationship between kinetic parameters can be found by using the mass-loss curves recorded in TG dynamic thermograms. The thermogravimetric curves of CTs heated in Ar at different heating rates are shown in Figure 5.

Two methods, Ozawa [21] (also known as the FWO method) and Kissinger [22], have investigated the thermal degradation kinetic parameters of CTs. The Coats-Redfern method [23] was used to determine the mechanism of the degradation of CTs after the activation energy obtained.

The Ozawa method is an integral method, using multiple heating rates curves to calculate kinetic parameters, and does not involve reaction mechanism hypotheses. According to the Ozawa method, Equation (1) can be obtained. In Equation (1), the ln[F(α)] is constant for the same α, and the plot of lnβ vs. 1000/T should be a straight line, thus, the activation energy can be calculated from the slope, which is −1.0516 E/R.

where β is the heating rate, T is the temperature, A is the pre-exponential factor, α is the conversion, and E is the activation energy.

Some of the Ozawa plots are shown in Figure 6, and all data are summarized in Table 2. Figure 6 shows the data fitted to straight lines, with a correlation coefficient greater than 0.990. These lines are nearly parallel in the conversion range of 0.10 < α < 0.80, thus indicating the applicability of Ozawa method to the present system. As seen from Table 2, when the conversion was less than 10%, the activation energy gradually increased, indicating that weight loss becomes increasingly difficult. However, for conversion in the 10–80% range, values of activation energy can be considered essentially the same, and the average activation energy value was 152.8 ± 8.98 KJ/mol. E does not significantly change for conversions between 0.10 and 0.80, indicating that the degradation of CTs takes place through the cleavage of linkages with similar bond energies.

From Table 2 and Figure 7, we can see that the dependence of the active energy on α can be separated into three distinct regions for the thermal degradation of CTs. The first region is for values of α ≤ 0.1, a small increase in the active energy. The second region is between 0.1 < α < 0.8, where the activation energy can be considered as a constant value. The third region is for α > 0.8, where a rapid increase of the activation energy is observed. The change of activation energy with α indicates that a complex reaction mechanism exists for the thermal degradation of CTs. According to Figure 7, there may be three different mechanisms during the decomposition of CTs. The first mechanism corresponds to the part where small loss appears; the second part, where the substantial mass loss takes place, is owed to the chief decomposition mechanism; the third mechanism corresponds to the final part of degradation; each mechanism presenting different activation energies. This is in good agreement with the results of many aliphatic polyesters [24,25,26,27].

In order to investigate the degradation mechanism of CTs in detail, the Coats and Redfern method was used to calculate the activation energy. The activation energy for every g(α) function listed in Table 3 can be obtained at constant heating rates from fitting of ln[g(α)/T²] vs. 1000/T plots. The Coats–Redfern method can be written as Equation (2):

Comparing the activation energy values in Table 2 and in Table 3, we can see that, at the early degradation stage, namely α < 0.1, the degradation mechanism may match the A2 or A3, and tend towards A3 because the value obtained from A3 is very close to those from the Ozawa method. In the main degradation process, or 0.1 < α < 0.8, the mechanism may match the F2 mechanism well; in the final stage, α > 0.8, the degradation mechanism may match R3.

For comparative purposes, the activation energy in the main degradation stage also was calculated by the Kissinger method. The T_p was applied to calculate the activation energy in this method, though the Kissinger method was suitable to explain the mechanism for the chief thermal degradation region. The Kissinger method has been used in the literature to determine the activation energy of solid-state reactions from plots of the ln(β/T_p²) vs. 1000/T_p in constant heating rate experiments. The activation energy can be determined by the Kissinger method without precise knowledge of the reaction mechanism, using Equation (3):

where β, A, E, and T_p are the known meanings. Figure 8 shows that the activation energy obtained is 159.2 KJ/mol, and the correlation coefficient is 0.9973. The value is slightly higher than that calculated with the Ozawa method; however, it falls within the scope of former’s deviation. Thus, the activation energy calculated from the Kissinger method was consistent with that from the Ozawa method.

For this study, we have used the same conversion values in the whole α range. Table 3 also shows activation energies and correlations for conversions at heating rate 10, 15, 20, and 25 K/min, respectively. At all heating rates, the activation energy values calculated with F2 type (Random nucleation with two nuclei on the individual particle) mechanism for the chief stage (0.1 < α < 0.8) are in better agreement with those obtained using the former two methods. Also, from this table it can be seen that the optimum heating rate value is 20 K/min, at which the activation energy corresponding to F2 mechanism is 159 KJ/mol, very close to 159.2 KJ/mol obtained by Kissinger method. These facts strongly suggest that the solid state thermodegradation mechanism followed by CTs in 0.1 < α < 0.8 is an F2 type. Therefore, the thermal degradation mechanism for CTs may be the combination of A3, F2 and R3.