Per-O-acylation of xylan at room temperature in dimethylsulfoxide/N-methylimidazole

Xylan (isolated from sugarcane bagasse) with xylose content of over 85 % was purchased from Yuan-Ye Biological Technology Co., Ltd. (Shanghai, China). Aa, Pa, Ba, Ac, Pc and Bc with a purity of 99 % were provided by Sigma-Aldrich Co., LLC (Shanghai, China). DMSO, NMI and other chemicals were of analytical-reagent grade from Guangzhou Chemical Reagent Factory (Guangdong, China). DMSO was dried over CaH₂ for 12 h and subsequently distilled under reduced pressure before use. Other chemicals were directly used without further purification.

The xylan esters, denoted as XyAa, XyPa, XyBa, XyAc, XyPc and XyBc, were homogeneously prepared in DMSO/NMI at room temperature with esterifying agents Aa, Pa, Ba, Ac, Pc and Bc, respectively, according the following procedure. Dried xylan (0.11 g, 1.67 mmol of hydroxyl groups in xylan) was suspended in 10 mL of DMSO and 5 mL of NMI. The mixture was agitated with magnetic stirring at 100 °C for about 1 h under nitrogen atmosphere to obtain xylan solution. Then the mixture was cooled to room temperature, and the esterifying agents were added into the xylan solution, as listed in Table 1. After the required time, the mixture was purified by dialysis (cutoff = 3000 g mol⁻¹) against water to remove DMSO, NMI, unreacted esterifying agents and by-products, and then freeze dried. The products were weighted to determine the percentage of weight increase, expressed by WPG, according to the following equation:where WPG is the weight percent gain of xylan after esterification, m1 (g) is the weight of xylan regenerated in DMSO/NMI, and m2 (g) is the weight of modified xylan samples.

The solubility of regenerated and acylated xylan in DMSO, dimethylformamide (DMF) and chloroform (CHCl₃) was investigated. A certain amount of sample (10 mg) was added in 0.5 mL solvent and stirred at room temperature for 12 h. After the required time, the results were recorded.

The functional groups and chemical structure of unmodified xylan and xylan esters were examined by FT-TR spectroscopy using a Bruker spectrophotometer (Tensor 27, Germany) from a KBr disc containing 1 % (w/w) finely ground samples in the range of 4000–400 cm⁻¹.

The ¹H-NMR, ¹³C-NMR and HSQC spectra of unmodified xylan, regenerated xylan and xylan esters were recorded from 40 mg samples in 0.5 mL DMSO-d ₆ on a Bruker Avance III 600 M spectrometer (Germany) with a 5 mm multinuclear probe. For the ¹H-NMR analysis, the detailed collecting and processing parameters were listed as follows: number of scans, 16; receiver gain, 31; acquisition time, 2.7263 s; relaxation delay, 1.0 s; pulse width, 12.5 s; spectrometer frequency, 600.17 MHz; and spectral width, 12,019.2 Hz. For the ¹³C-NMR analysis, the detailed collecting and processing parameters were listed as follows: number of scans, 5000; receiver gain, 187; acquisition time, 0.9088 s; relaxation delay, 2.0 s; pulse width, 11.5 s; spectrometer frequency, 150.91 MHz; and spectral width, 36,057.7 Hz. For the HSQC analysis, the detailed collecting and processing parameters were listed as follows: number of scans, 16; receiver gain, 187; acquisition time, 0.1420 s; relaxation delay, 2.0 s; pulse width, 12.5 s; spectrometer frequency, 600.17/150.91 MHz; and spectral width, 7211.5/24,875.6 Hz.

The thermal stability of samples was performed by using thermogravimetric analysis (TGA) and derivative thermogravimetry (DTG) on a Q500 thermogravimetric analyzer (TA, USA). To eliminate the effect of water, all samples were dried at 105 °C for 12 h before characterization. The apparatus was continually flushed with nitrogen. The sample between 9 and 11 mg was heated from 30 to 600 °C at a heating rate of 10 °C/min.

Homogeneous esterification of hemicelluloses had already been achieved (Sun et al. 2000; Belmokaddem et al. 2011; Zhang et al. 2011b). However, the required reaction conditions were complicated, e.g. with high reaction temperature or catalyst. In the present study, the esterification of xylan was carried out in DMSO/NMI at room temperature without additional catalyst to simplify the reaction procedures. The optimum conditions in terms of the molar ratio of reagents to AXU (from 0.5:1 to 8:1) and the kinds of reagents, including Aa, Pa, Ba, Ac, Pc and Bc, were investigated to obtain appropriate reaction efficiency (Table 1; Fig. 1). To better view the reactivity of reagents, the molar ratio of reagents in relation to DS and WPG is illustrated by graphical relation in Fig. 2.

Figure 2a illustrates the effect of the molar ratio of anhydrides to AXU on the DS of xylan esters (XyAa, XyPa and XyBa). Apparently, an increase of the molar ratio of anhydrides to AXU from 0.5:1 to 8:1 resulted in an increment in DS from 0.37 to 1.98 (XyAa), 0.41 to 1.95 (XyPa) and 0.41 to 1.96 (XyBa), respectively, indicating the favorable effect on xylan acylation. These results were probably due to the increased molecule collisions of acylation reagents with reactive hydroxyl groups in AXU. Theoretically, two hydroxyl groups are available to substitute at the position of C-2 and C-3 in each AXU of xylan, and the maximum DS is 2. Therefore, these results indicated that the per-O-acylation of xylan has readily accomplished in DMSO/NMI system at room temperature with the optimum molar ratio of anhydrides to AXU 8:1. More importantly, NMI is a derivative of imidazole with a slightly more basic N-methyl group as indicated by the pK _a’s of the conjugate acids of 7.0 and 7.4 (Zhang et al. 2012). This means that NMI can act as a solvent, a base and an excellent acylation catalyst in DMSO/NMI solvent system (Lu and Ralph 2003; Liu et al. 2007; Zhang et al. 2012). Because the esterification of xylan with anhydride is acid- or base-catalyzed reaction, NMI shows promise in accelerating the esterification reaction. In comparison with DS, the WPG value displayed a similar trend as shown in Fig. 2b. An increase of the molar ratio of anhydrides to AXU from 0.5:1 to 8:1 resulted in an improvement of the WPG of XyAa, AyPa and XyBa from 16.29 to 72.93, 21.78 to 78.74 and 31.39 and 86.88 %, respectively.

Figure 2c, d show the influence of the molar ratio of acyl chlorides to AXU on the DS and WPG of xylan esters (XyAc, XyPc and XyBc). In general, DS of xylan esters gradually increased, then tapered and decreased after reaching the maximum value when increasing acyl chlorides dosages. From Fig. 2c, DS of XyAc, XyPc and XyBc increased from 0.34 to 1.77, 0.32 to 1.76 and 0.22 to 1.76, respectively, with increasing acyl chlorides dosages from 0.5:1 to 4:1. After reaching its optimum value, the DS of xylan esters decreased slightly with increasing acyl chlorides dosages from 4:1 to 8:1. These results indicated that higher molar ratio of acyl chlorides to AXU resulted in a decreased DS. This decrease in DS was probably due to the severe degradation and hydrolysis of the acylated xylan at the presence of the by-product hydrochloric acid (Zhang et al. 2012). In Fig. 2d, WPG of XyAc, XyPc and XyBc decreased from 0.99 to −28.62, 6.71 to −15.48 and 15.87 to −0.81 %, respectively, with the increase of the molar ratio of acyl chlorides from 0.5:1 to 2:1. This decrement was probably due to the degradation and hydrolysis of xylan esters at the presence of hydrochloric acid. Further increasing the reagents dosage from 2:1 to 4:1, and to 8:1 led to an increase in the WPG of XyAc, XyPc and XyBc from −28.62 to −4.97, −15.48 to 5.71 and −0.81 to 14.27 %, respectively, followed by a slight decrease. Therefore, these results showed that the condition that gave the maximum DS of XyAc, XyPc and XyBc was not same as that for the highest WPG. Due to the serious degradation of xylan esters caused by hydrochloric acid, the optimum conditions was selected mainly based on the highest DS. Therefore, the optimum molar ratio of acyl chlorides to AXU was 4:1.

In addition, Fig. 2 compares the influences of reagent species on the DS and WPG of xylan esters. Comparatively, when considering the value of DS and WPG of xylan esters, the reactivity of anhydrides was higher than acyl chlorides, which was probably due to the serious degradation of xylan esters caused by hydrochloric acid by-product. Figure 2 also presents the effects of chain length (C2, C3 and C4) of anhydrides and acyl chlorides on DS and WPG of xylan esters. As shown in Fig. 2a, c, the DS values of xylan esters with anhydrides (Aa, Pa and Ba) and the corresponding acyl chlorides (Ac, Pc and Bc) were nearly equal, indicating the similar reactivity of anhydrides (Aa, Pa and Ba) and the corresponding acyl chlorides (Ac, Pc and Bc) under the selected conditions. Similar results were found in cellulose esters (Ratanakamnuan et al. 2012). In Fig. 2b, d, the noticeable differences of WPG increment of each xylan ester were observed with the increase of chain length of anhydrides and acyl chlorides, probably due to the attachment of acyl substituents (Ratanakamnuan et al. 2012). The longer acyl substituents represented the higher molecular weight of acyl group which can result in higher WPG of xylan derivatives [WPG XyBa (C4) > XyPa (C3) > XyAa (C2), XyBc (C4) > XyPc (C3) > XyAc (C2)].

One of the reasons that limits the application of most polysaccharides is their poor solubility in organic solvents (Fundador et al. 2012b). Esterification is one method to solve this problem, for the introduced hydrophobic acyl group can alter the chemical structure of the polysaccharides.

In this study, the solubility of esterified xylan in various organic solvents was investigated and the results are summarized in Table 1. It was found that the regenerated xylan (re-xylan) was swollen in DMF and insoluble in CHCl₃, probably due to the strong intermolecular hydrogen bonds between xylan chains. As expected, xylan esters with different DS could be easily dissolved in DMSO, providing the further structural characterization of the xylan derivatives with NMR. The solubility of xylan esters in DMF and CHCl₃ increased with an increase of DS. Similar results can also be observed in acylated cellulose (Fundador et al. 2012a; Ratanakamnuan et al. 2012). In general, the introduction of hydrophobic acyl groups into the molecular structure of xylan results in the destruction of hydrogen bonds, which was expected to improve its solubility.

FT-IR spectroscopy is the most convenient method for the elucidation of functional groups and structural features of xylan derivatives. Figure 3 compares the FT-IR spectra of unmodified xylan and xylan esters XyAa7 (DS = 1.87), XyPa23 (DS = 1.83) and XyBa39 (DS = 1.84). As shown in the spectrum of unmodified xylan, a sharp band at 898 cm⁻¹ is characteristic of β-glucosidic linkages between the sugar units (Fang et al. 1999; Belmokaddem et al. 2011). The band at 1045 cm⁻¹ is assigned to C–O stretching in C–O–C linkages. An intense band at 1619 cm⁻¹ corresponding to the adsorbed water is broader because it obscures vibrations of uronic acid carboxyl, which decreased substantially in xylan esters (Kačuráková et al. 2000). The bands at 3419 and 2896 cm⁻¹ are attributed to hydrogen bond in hydroxyl groups (O–H) and the symmetric C–H vibration band, respectively. As can be seen, the spectra of XyAa7 (DS = 1.87), XyPa23 (DS = 1.83) and XyBa39 (DS = 1.84) provide evidence of acylation by the presence of three important typical ester bands at 1752 cm⁻¹ (C=O ester), 1364 cm⁻¹ (–C–CH₃) and 1248 cm⁻¹ (C–O stretching) (Sun and Hughes 1999; Ayoub et al. 2013). The occurrence of peaks at around 2800–3000 cm⁻¹ in the spectra indicates the presence of methyl and methylene C–H stretching of alkyl chains in xylan esters (Belmokaddem et al. 2011). As the carbon chain length of the acyl group increases from three-carbon chain (spectrum of XyPa23) to four-carbon chain (spectrum XyBa39), the signals of methyl and methylene band at around 2800–3000 cm⁻¹ become intense. The O–H stretching vibration of xylan esters drifted from low wave-number domain to 3479 cm⁻¹, probably due to the decrease of hydrogen bonding in xylan after esterification. Moreover, in comparison a drastic decrease was observed in the O–H stretching (3479 cm⁻¹), indicating the high degree of acylation of xylan, which was in accordance with the DS of xylan esters. Similar results were reported in acetylated cellulose (Chen et al. 2014).

According to previous studies, FT-IR spectra can provide an alternative method for calculating the DS (Samios et al. 1997; Zhang et al. 2011a; Chen et al. 2014). In the present study, a typical example of the detailed relationship between the DS of XyPa and the ratio of the hydroxyl peak (3479 cm⁻¹) height to the carbonyl peak (1752 cm⁻¹) area is presented in Table 2. From the data, the graph of the ratio of the peaks versus the DS was plotted (Fig. 4). This serves as a calibration curve from which the DS of other xylan esters can be calculated once the FT-IR spectrum is available. Thus, it can also be used as an alternative measurement to evaluate the DS of xylan esters obtained from ¹H-NMR spectra.

In the present study, to elucidate the changes of xylan structure after acylation, the samples of unmodified xylan, XyAa7 (DS = 1.87), XyPa23 (DS = 1.83) and XyBa39 (DS = 1.84) were characterized with ¹H-NMR, ¹³C-NMR and HSQC spectroscopies.

Figure 5 illustrates the ¹H-NMR (Left column) and ¹³C-NMR spectra (Right column) of native xylan (A and B), XyAa7 (C and D, DS = 1.87), XyPa23 (E and F, DS = 1.83) and XyBa39 (G and H, DS = 1.84), respectively. In Fig. 5a, the β-(1 → 4)-linked d-xylp units were characterized by the signals at 3.04, 3.32, 3.48, 9.91, 4.09 and 4.34 ppm, which correspond to the protons at H-2, H-5a, H-3, H-4, H-5e and H-1 of AXU in xylan, respectively (Fundador et al. 2012a; Zhang et al. 2015a). The proton signals at 5.29 and 5.41 ppm are assigned to the hydroxyl groups (2-OH and 3-OH) in xylan. In addition, the signals at 3.60, 3.78, 3.91 and 3.41 are attributed to H-2, H-3, H-4 and H-5 in α-l-arabinofuranosyl residues (α-l-Araf) (Wen et al. 2011). The signal at 5.25 ppm in the ¹H-NMR spectrum has been assigned to protons at C-1 of MeGlcA residues. However, the signal corresponding to the protons of the acetyl groups at about 1.90 ppm was not observed, probably for its low content or the isolated method (Fundador et al. 2012b). In Fig. 5c (XyAa7, DS = 1.87), the signal at 1.95 ppm relates to the hydrogen proton in methyl group (–CH ₃) from the attached acyl moieties. This result confirms the successful esterification of xylan. In addition, the signals at 3.26, 3.75, 3.85, 4.51, 4.68 and 4.92 ppm are attributed to H-5a, H-4, H-5e, H-2, H-1 and H-3 of the ring structure of XyAa7, respectively. In terms of XyPa23 (Fig. 5e), the signals assigning to the acyl moieties a and b are located in 0.97 and 2.20 ppm, respectively. For XyBa39 (Fig. 5g), the signals at 0.84, 1.48 and 2.17 ppm are corresponded to the acyl moieties a, b and c, respectively. The signals in the range of 3.00 to 5.50 ppm are attributed to the ring protons in AXU backbone. It should be noted that, the signals corresponding to the protons of the hydroxyl groups attached at C-2 (2-OH) and C-3 (3-OH) in AXU could not be detected in xylan esters, indicating their high DS (Fundador et al. 2012b; Zhang et al. 2015b).

On the basis of the typical proton signals from xylan esters, the DS of xylan esters (Table 1) was calculated according to the following equation (Belmokaddem et al. 2011):where DS is the degree of substitution of xylan esters, I_a is the integral area of the resonances of the methyl protons at a, I_H1 and I_H4 are the integral areas of anomeric proton assigned to H-1 and H-4 of AXU.

Figure 5 also lists the ¹³C-NMR spectra of xylan and xylan esters. In ¹³C-NMR spectrum of unmodified xylan (Fig. 5b), the five major signals at 63.7, 73.2, 74.5, 75.8 and 102.1 ppm are attributed to C-5, C-2, C-3, C-4 and C-1 of xylan, respectively. Similar findings were observed with other xylan sources (Wen et al. 2011; Fundador et al. 2012b; Vega et al. 2012). However, the signals attributing to acetyl groups, MeGlcA and other sugar residues were not detected, probably due to their low content in xylan. The ¹³C-NMR spectra of the xylan esters are displayed in Fig. 5d, f, h. In Fig. 5d (XyAa7, DS = 1.87), the signals for the ring structure of xylan were detected. In addition, the signals at 26.6 and 172.9 ppm originate from the –CH₃ (methyl, a) and C=O (carbonyl, b) in the acyl moieties. In Fig. 5f (XyPa23, DS = 1.83), the signals at 9.4, 27.2 and 172.8 ppm are attributed to a (–CH₃, methyl), b (–CH₂–, methylene) and c (C=O, carbonyl) of the acyl moieties, respectively (Fundador et al. 2012b). In case of Fig. 5h (XyBa39, DS = 1.84), the signals at 171.8 (C=O, d), 35.2 (–CH₂–, methylene, c), 18.3 (–CH₂–, methylene, b) and 13.6 ppm (methyl, –CH₃, a) in acyl moieties of esters side chains are all observed. These signals further supported the data, indicating the successful esterification of xylan.

HSQC is an efficient tool for qualitative and quantitative characterization of chemical structures, and can provide more detailed information about the signals which was overlapped in ¹H- and ¹³C-NMR. To better understand the chemical structure of xylan and xylan esters, Fig. 6 represents the HSQC spectra of unmodified xylan (A), regenerated xylan (B), XyPc31 (C, DS = 1.76) and XyBa39 (D, DS = 1.84), and the different correlations from xylan backbone, sugar residues and esters side chains are exhibited in Red, Green and Blue, respectively. In Fig. 6a, the six strong correlations (presented in Red) at δ_C/δ_H 102.1/4.26, 73.2/3.02, 74.5/3.25, 75.8/3.48, 63.7/3.85 and 63.7/3.13 ppm are related to C₁–H₁, C₂–H₂, C₃–H₃, C₄–H₄, C_5e–H_5e and C_5a–H_5a, respectively, in xylan backbone. The four correlations (presented in Green) at δ_C/δ_H 78.3/3.60, 80.8/3.78, 86.4/3.91 and 62.2/3.41 ppm are attributed to C₂–H₂, C₃–H₃, C₄–H₄ and C₅–H₅ in α-l-Araf (Wen et al. 2011). The presence of MeGlcA was characterized by the weak correlation at δ_C/δ_H 107.7/5.25 ppm corresponding to C₁–H₁ methoxy proton (presented in Green) (Belmokaddem et al. 2011). As shown in the HSQC spectrum of regenerated xylan (Fig. 6b), the major signals corresponded to xylan backbone, α-l-Araf and MeGlcA were also detected, indicating that DMSO/NMI was capable of dissolving xylan without apparent degradation. As shown in Fig. 6c (XyPc31, DS = 1.76), the correlations which are found at δ_C/δ_H 9.1/0.98 and 27.3/2.21 ppm arise from the protons in the acyl moieties specifically from the methyl protons (–CH₃, C_a–H_a) and methylene (–CH₂–, C_b–H_b), respectively. These results confirm the successful esterification of xylan in DMSO/NMI. Moreover, the six strong correlations at δ_C/δ_H of 100.3/4.70, 71.0/4.54, 72.3/4.97, 76.1/3.76, 62.7/3.27 and 62.7/3.80 ppm are assigned to C₁–H₁, C₂–H₂, C₃–H₃, C₄–H₄, C_5a–H_5a and C_5e–H_5e in AXU, respectively. In Fig. 6d, the correlations at δ_C/δ_H 13.7/0.87, 18.1/1.50 and 35.9/2.18 ppm are corresponded to C_a–H_a (–CH₃), C_b–H_b (–CH₂–) and C_c–H_c (–CH₂–) in acyl moieties of XyBa39. The correlations for the ring structure of XyBa39 were also clearly detected. Apparently, more esters were attached to C₂ position than to C₃ position in AXU backbone. The results of integrated resonances indicated that 50.76/49.24 % (XyBa39, DS = 1.84) and 51.23/48.77 % (XyPc31, DS = 1.76) were attached to C₂ and C₃ positions of AXU, respectively. Note that, the correlations arising from MeGlcA and α-l-Araf residues were not detected in XyPc31 and XyBa39. This is probably due to the formation of carboxylic acid or hydrochloric acid by-products in esterification reactions resulting in the decrease of these sugar residues in xylan esters, which could not be observed in HSQC spectra. In addition, in comparison with xylan derivatives with low DS (Zhang et al. 2014), the correlations for substituted C₂–H₂ and C₃–H₃ could not be detected in xylan esters with high DS, indicating the reliability of DS calculation on the basis of ¹H-NMR.

TGA/DTG analysis provide information on the thermal decomposition profiles of derivatives, which can be used to follow the physiochemical changes that occur during chemical modification (El-Sayed and Mostafa 2014). Moreover, thermal stability of derivatives is an important parameter in exploring its further applications (Yan et al. 2009). Therefore, TGA/DTG was applied to investigate the effects of DS and esterifying reagents on the thermal stability of xylan esters. Figure 7 shows a comparison of the thermal properties of xylan and xylan esters (XyPa17, XyPa23, XyPc25 and XyPc31) with different DS. As shown in TGA curves, the onset of degradation takes place at about 150, 160, 180, 200, 220 °C for XyPc25, XyPc31, xylan, XyPa17 and XyPa23, respectively. The temperature at which there is 10 % w/w weight loss is about 200, 210, 230, 240 and 250 °C for XyPc25, XyPc31, xylan, XyPa17 and XyPa23, respectively. As can be seen, the thermal stability is improved by acylation with anhydrides while decreased with acyl chlorides. The high thermal stability of XyPa is brought about by the decrease in the number of remaining hydroxyl groups after esterification (Aburto et al. 1999; Fundador et al. 2012a). Similar findings were observed with other xylan esters (Belmokaddem et al. 2011; Fundador et al. 2012b). In previous study, the thermal stability of xylan esters with decanoyl chloride and lauroyl chloride was higher than unmodified xylan (Fundador et al. 2012b). However, in the present study, the thermal stability of xylan decreased after the esterification with acyl chloride, which was probably due to the shorter alkyl chain length and the serious degradation of xylan esters for the formation of hydrochloric acid (Fundador et al. 2012b). At the end of the analysis, it was noted that xylan esters had lower residual weight compared to unmodified xylan, probably due to the decreased contents of inorganic salts after the esterification in DMSO/NMI. Similar findings were observed with acetylated xylans in DMF (Belmokaddem et al. 2011).

Figure 7 also represents the DTG curves of samples. The DTG curves show the weight loss rates, and DTG_max presents the maximum degradation rate to compare the thermal stability between samples (Nadji et al. 2009). In DTG curves, the DTG_max are determined at about 270, 275, 320, 250 and 265 °C for the unmodified xylan, XyPa17, XyPa23, XyPc25 and XyPc31, respectively, which implied that the thermal stability of xylan increased after esterification with anhydride, while decreased with acyl chloride. This is probably for the serious degradation of xylan esters in the presence of hydrochloric acid by-product. Generally, esterification is an effective method to improve the thermal processability of xylan, which could broaden its applications, like blend with other polymers to produce thermoplastics.