Turkish River Gum (Eucalyptus camuldensis Dehn.) wood was used in this study. The logs were obtained from Mersin, Turkey and cut into boards measuring 150 mm by 500 mm by 18 mm. Two experimental wood panels were obtained from per log. All of the experimental boards were placed in a climate chamber adjusted 20±2°C and 65%±5 relative humidity (RH).

Thermal modification of the eucalyptus wood boards was performed using a laboratory type hot-press located in the Wood Composites Manufacturing Laboratory at the Department of Forest Products Engineering, Istanbul University. Experimental design of the thermal treatments conducted with wood boards is shown in Table 1. Prior to the heat treatments, the modified and control samples were conditioned at 20±2 ºC and 65±5% RH in a climate chamber for three weeks.

Table 1:
Experimental design of the thermally modified wood boards.

The chipped eucalyptus wood boards were grinded in a Wiley mill. Prior to the chemical analysis, the samples were screened to particle size range ranging from 40 to 100 mesh T-257 cm-85 method (TAPPI 1992a). After this, hot water, 1% NaOH, ethanol-cyclohexane, and, ethanol solubilities were determined according to TAPPI Test Methods T-204 om-88 (TAPPI 1992b), T-207 om-88 (TAPPI 1992c), T-212 om-88 (TAPPI 1992d). The amount of klason lignin was estimated using 72% sulfuric acid and 40% hydrobromic acid by Runkel and Wilke (1951). Acid soluble lignin was determined by using a UV/Vis-spectrophotometer according to TAPPI UM 250 (TAPPI 1985). Holocellulose was prepared using sodium chlorite solution and acetic acid at 75ºC according to procedure used by Jayme (1942) and Wise et al. (1946). Amount of α-cellulose was determined according to T-203 cm-99 method (TAPPI 1983). Methanol:water extraction of the samples was performed as described previously (Balaban and Ucar 2001).

For FTIR analysis, the finely ground wood samples were dried in a hot oven at 50°C for 24 hours. The changes in the chemical structure of the ground samples were investigated by using an attenuated total reflectance with fourier transform infrared spectroscopy (ATR-FTIR, Bruker, Tensor 37). The spectra were recorded in the wavenumber range from 400-4000 cm^-1 and 32 scans per sample with a resolution of 4 cm^-1 were used. The evaluation of the spectra were carried out using the Bruker OPUS software.

The solubility values of the thermally modified wood boards are shown in Table 2. The hot water solubility values of the wood boards were decreased with the thermal compression. The solubility of 1% NaOH of the samples increases with rising hot-press pressure and temperature. It could be due by the increase in low molecular weight carbohydrates, hemicellulose and, cellulose degradation products in the samples. With thermal effect, solubility of lignin contributes to this increase too (Korotkova et al. 2015). With thermal effect, lignin starts to degrade but at a lower rate than other component of the wood. Besides that, there was no significant difference between S3 and S4 sample, which showed that increase in the press pressure has limited effect on the carbohydrate degradation. These results were in agreement with Ates et al. (2009), Batista et al. (2016b) and Neiva et al. (2015).

Amount of non-polar extractives in samples were determined with successive ethanol-cyclohexane and ethanol extractions. Ethanol-cyclohexane and ethanol solubility values were 2,6% and 4,1% respectively (Table 2.) The table shows an increase in ethanol-cyclohexane solubility values of the samples with rising treatment pressure and hot- press temperature in comparison with control sample. Brito et al. (2008) observed a similar result. However, in contrast to Brito et al. (2008), ethanol solubility values of the samples have a significant decrease in comparison with control sample. The reason for this difference can be the method of thermal modification.

Methanol-water solubility values (5,1%) in the control samples are given in Table 2. Wanxi et al. (2015) determined the methanol solubility as 5,8% for the untreated sample of same species. Significant reduction was observed in the methanol-water solubility of the modified wood samples with increasing thermal modification temperature. This decline may be due to the auto-condensation reactions of phenolic compounds of the wood by the thermal effect (Rakić et al. 2004, Pizzi 1994). It has also been shown that the increase in pressure applied to the modification has a limited effect on the methanol-water solubility.

Table 2:
Solubility values of the thermal modified wood boards.

The results of lignin content of unmodified- and thermally modified wood boards are shown in Table 3. Amount of residual lignin shown in second column of the table is higher than expected values for eucalyptus wood. The reason for this is that presence high amount of the phenolic compounds in the wood (Cadahía et al. 1997). The phenolic compounds could be condensed and remain as the klason lignin in acidic conditions, it is believable that the klason lignin content is somewhat overestimated, if the wood material was prepared without an additional 1% sodium hydroxide extraction (Gonultas and Ucar 2013, Hafizoglu et al. 1997).The lignin contents determined after successive ethanol/cyclohexane, ethanol and 1% sodium hydroxide extractions would be considered as true lignin values. Table 3 shows amount of residual lignin value after 1% NaOH extraction in third column. In addition, acid soluble lignin and total lignin contents of samples are given in Table 3. The klason lignin value was increased significantly with increasing temperature and pressure effect. The relative increase of the klason lignin content in the samples was due to decreased amount of cellulose and hemicellulose caused by thermal degradation and being condensed the degradation product of carbohydrates such as aldehydes with lignin and other phenolic compounds (Tjeerdsma et al. 1998, Esteves et al. 2013). These results were in agreement with Alén et al. (2002), Ates et al. (2009), Boonstra and Tjeerdsma (2006), Brito et al. (2008) Fengel and Wegener (1984). The content of acid soluble lignin remained relatively constant in all samples besides the sample S3.

Table 3:
Lignin content of the thermally modified wood boards.

Total lignin percentages, yields of holocellulose and amount of α- cellulose in the samples were compiled in Table 4. The yields of holocellulose in the control sample was determined as 71,5 while this value was 69,4 for Eucalyptus grandis wood (Batista et al. 2016a). The holocellulose content of S3 sample was decreased significantly because due to their amorphous and branched polymeric structure especially hemicelluloses are less stable against the effect of the thermal degradation when compared to the other polymeric wood constituents (i.e., lignin and cellulose) (Fengel and Wegener 1984). The cleavage of glycosidic bonds of hemicelluloses is facilitated by deacetylation reactions causing the formation of acetic acid (Windeisen et al. 2007). After thermal treatment, most of the xylan is degraded (Fengel and Wegener 1984). The lower content of hemicelluloses might play a role on the better natural durability of the wood (Windeisen et al. 2007). On the other hand, the crystal structures of cellulose molecules are more resistant to thermal effects (Fengel and Wegener 1984). The thermal resistency of cellulose can be seen in Table 4.; only a relatively small decrease in the yields of α-cellulose was observed when compared to the decrease in holocellulose yields. The findings obtained from this study are supported by a study of Batista et al. (2016a).

Table 4:
Cell wall components of the thermal modified wood boards.

The IR spectra of the thermally modified or unmodified eucalyptus wood panels are shown in Figure 1.

Figure 1:
FTIR spectra of untreated and thermal modified wood boards.

The observed bands of the thermally modified or unmodified eucalyptus wood panels are presented in Table 5.

Table 5:
Main bands of infrared spectrum of the modified samples.

The OH tension which was determined at 3334 cm-1. This peak corresponds to hydrogen bonded OH group that water and all main wood components contain (Kocaefe et al. 2008). In the thermally modified samples, some of the OH groups are replaced and this peak is reduced (Esteves et al. 2013, Mattos et al. 2015, Missio et al. 2015, Tjeerdsma and Militz 2005). The C-H stretching peak at 2909 cm^-1 was increased and shifted to 2894 cm^-1 with thermal modification process. This broadening may be due to oxidation and hydrolysis of acetyl groups from hemicelluloses (Esteves et al. 2013). The C=O stretching vibration was determined at 1731 cm ^-1. This peak indicates carbonyl groups in the xylan. When this peak was examined further, the peak shrinked and shifted to small wavenumbers with increasing temperature and pressure of the thermal modification process. Especially, this peak was shifted to a lower frequency (1721 cm^-1) compared to that of control sample, while at the same time the lowest holocellulose value (55,9%) in the S3 sample was determined. For FTIR analysis, similar results were described by Boonstra and Tjeerdsma (2006), Esteves et al. (2013), Kocaefe et al. (2008), Mattos et al. (2015), Missio et al. (2015), Tjeerdsma and Militz(2005), Peng et al. (2015), Yildiz et al. (2007). In the samples, the characteristic aromatic ring stretching vibration is observed at 1506 cm^-1. Cleavage of the aliphatic side chains in lignin and the formation of new cross-links in lignin with condensation reactions reduces the water absorption properties of the samples (Boonstra and Tjeerdsma 2006, Kocaefe et al. 2008). That peak increased and shifted to a large wavenumbers with rising the temperature and pressure of the thermal process (Esteves et al. 2013, Kocaefe et al. 2008, Missio et al. 2015, Müller et al. 2009, Tjeerdsma and Militz2005). The peak of aromatic ring stretching was shifted to 1511 cm^-1 and 1509 cm^-1 for S3 and S4 sample, respectively. For the S1 and S2 samples, the peak wavelength did not change. Similarly, results of total lignin values in the S3 (38,7%) and S4 samples (32,9%) was significantly increased compared to the control sample (29,6%) while the S1 (31,1%) and S2 (29,8%) samples had slight increase. The peak at 1424 cm^-1 was C-H deformation in the lignin and cellulose. This peak was shifted to 1421 cm^-1 and decreased with maximum pressure of the thermal process. This decrease was also obtained by Akgul et al. (2007), Ates et al. (2009), Esteves et al. (2013), Mattos et al. (2015), Müller et al. (2009), and Yildiz and Gümüşkaya(2007). The C-O vibration of the syringyl ring units was determined at 1324 cm^-1 (Esteves et al. 2013, Kocaefe et al. 2008, Müller et al. 2009). With condensation reactions in the lignin units, this peak increased and shifted to small wavenumbers. While the lowest total lignin value was determined in the S3 sample, the peak wavelength shifted to 1320 cm-1 too. In the infrared spectrum of samples, the C-O carbonyl band in lignin and xylan units was observed at 1229 cm^-1. This peak appeared to decrease and broadened to small wavenumbers for thermally modified wood samples (Esteves et al. 2013, Kocaefe et al. 2008, Mattos et al. 2015, Missio et al. 2015, Müller et al. 2009, Peng et al. 2015). After thermal modification, an increase defined in the peak at 1104 cm^-1 due to OH association of cellulose units. Especially, the increase in S1 and S3 samples may be due to formation of new alcohol and ester groups after the modification (Ates et al. 2009, Kocaefe et al. 2008, Mattos et al. 2015, Missio et al. 2015). The absorption peak at 897 cm^-1 indicated the asymmetric deformation of cellulose and hemicellulose units (Esteves et al. 2013, Kocaefe et al. 2008, Müller et al. 2009). There was a decrease in this peak accompanied with increasing pressure and temperature of process, especially due to depolymerization of the hemicellulose units.

^♠Corresponding author: oktay.gonultas@btu.edu.tr