**3. Results and Discussions**

### *3.1. Cell Wall Component Composition Study of Untreated and Treated Woods by TGA and DSC*

TGA analysis of the decomposition of pure cellulose, xylan and lignin was used to identify the temperature ranges that will be used for the cell well composition analysis of untreated and treated woods. Figure 4 shows the thermogravimetry (TG) and the differential thermogravimetry (DTG) curves vs. temperature for long and medium cellulose fibers, xylans from beech and birch wood (representing hemicelluloses) and lignin (alkali). The DTG curves show a distinct primary peak for each model component. Xylan from birchwood and xylan from beechwood decomposed over the range of about 200 to 300 ◦C with peak maximums at 269 and 274 ◦C, respectively. The long and medium cellulose fibers mainly decomposed over the range of about 300 to 360 ◦C with peak maximums at 330 and 328 ◦C, respectively. Lignin slowly decomposed over the range of about 200 to 800 ◦C with a peak maximum at 353 ◦C. The hemicelluloses are the most thermally sensitive of the main biomass components. Yang et al. studied pyrolysis of hemicellulose (xylan from birchwood), cellulose fiber and lignin (alkali) and reported that xylan decomposed at 220–315 ◦C with peak maximums at 268 ◦C, cellulose fiber at 315–400 ◦C with peak maximums at 355 ◦C, and lignin at 150–900 ◦C [43]. Our results showed a lower decomposition temperature due to our lower heating rate of 2 ◦C/min versus their 10 ◦C/min. The peak maximums of celluloses measured by both TGA and DSC were comparable. The peak maximums observed for the xylans and lignin by TGA were slightly lower than those measured by DSC but still fell in similar decomposition temperature ranges.

(**a**)  **4.**

*Cont*.

**Figure**

**Figure 4.** (**a**) Thermogravimetry (TG); (**b**) Differential thermogravimetry (DTG) curves versus temperature for the pure model components (xylan/hemicelluloses, cellulose, and lignin) of biomass cell wall components.

Percent weight for each cell wall chemical component was calculated by weight loss in designated temperature ranges that correlate to the cell wall components from TGA analysis [34]. Based on DTG curves of the model components (Figure 4b), the designated temperature ranges for calculating the percent weight of cell wall chemical components were selected. The ranges are hemicellulose between 200 and 300 ◦C, cellulose between 300 and 360 ◦C, and lignin between 360 and 800 ◦C. Percent weight (dry basis) for the component was calculated by the weight loss in the temperature range of the component over the total weight loss in all designated temperature ranges (200–800 ◦C). The weight loss in each temperature range was calculated by area in the range of DTG curve for the samples. In this study, the area was adjusted based on the distribution of the model lignin in the designated temperature ranges. Table 1 lists the calculated cell wall chemical components of the untreated and treated woods at 220, 260, 300, 350, 450 and 550 ◦C. The results clearly show the changes in cell wall chemical components with thermal treatment temperature. Hemicellulose contents of the untreated and torrefied woods at 220 , 260 ,and 300 ◦C are 24.3% 21.2%, 10.1% and 7.4%, respectively, showing that hemicellulose was slightly decomposed by 220 ◦C, mildly by 260 ◦C, and significantly by 300 ◦C. Cellulose was slightly decomposed for the torrefied wood at 300 ◦C and more severely for the pyrolyzed wood at 350 ◦C. The samples pyrolyzed at 450 and 550 ◦C have only lignin with both hemicellulose and cellulose decomposed. The cell wall composition of the torrefied woods correspond with the light, mild and severe torrefaction as classified by Chen et al. [15] and are similar with the results reported by Chen et al. [28] and Mafu et al. [29]. Mafu et al. tested the hardwood cell wall compositions of untreated and torrefied at 260 ◦C samples using the chemical method. Hemicellulose, cellulose and lignin contents of the untreated and torrefied samples were 11.4% and 1.2%, 56.9% and 46.7%, 15.7% and 16.0%, respectively. Most of hemicellulose for the torrefied wood was decomposed while cellulose and lignin were slightly impacted or not changed.


**Table 1.** Calculated cell wall chemical component of untreated and treated woods at each treatment temperature based on thermogravimetric analyzer (TGA) tests.

The decomposition of cellulose, xylan, and lignin were endothermic in the temperature range up to 600 ◦C as shown by the DSC results. The order of the amount of energy (heat of decomposition) required from high to low is: Cellulose > xylan > lignin. Decomposition temperature ranges and peak temperatures for cellulose, xylan and lignin obtained from DSC tests are comparable to TGA. During the DSC tests, the amount of energy absorbed (heat of decomposition) for treated samples at 350, 450 and 550 ◦C was much less than observed for the lower temperature treated woods at 220, 260 and 300 ◦C since at 350 ◦C and above, the xylans/hemicelluloses and much of the cellulose, which has the highest heat of decomposition of all the cell wall components of biomass, have already been decomposed. For the sample treated at 220 ◦C, some hemicellulose is decomposed. For the samples treated at 260 and 300 ◦C, hemicellulose is mostly decomposed. Both hemicellulose and cellulose are decomposed for the samples treated at 450 and 550 ◦C.

Decomposition of the cell wall components of the cherry wood during the thermal treatments can be used to explain the cause of the grindability improvement after the thermal treatments (grindability test results in Section 3.2). Wood strength results primarily from the fiber content since wood is basically a series of tubular fibers or cells cemented together [32,44]. The cell walls have three main regions: the middle lamella (ML) that acts as adhesion between two or more cells, the primary wall (P) and the secondary wall in its three layers (S1, S2,and S3) (Figure 5a) [44]. These regions have different thicknesses and various quantities of the major chemical components hemicellulose, cellulose, and lignin. For example, for Scotch pine wood, the S2 layer has the highest cellulose content (32.7%) in the cell wall with lesser quantities of hemicellulose (18.4%) and lignin (9.1%) [32]. The S2 layer is the thickest in the secondary wall and in the cell wall overall. From this respect, the S2 layer is mainly responsible for the overall properties of the cell wall. For the S2 layer, a proposed ultrastructural model described the arrangements of hemicellulose, cellulose and lignin and is illustrated in Figure 5b [32]. Lignin and hemicellulose act as a matrix for cellulose in the S2 layer. The lignin in the S2 layer is evenly distributed throughout the layer. Thin, low molecular weight hemicellulose channels bond cellulose microfibrils on their radial faces. Sheaths of hemicellulose are along the boundary area of the cellulose microfibrils. Thermal treatments decompose the hemicellulose resulting in interwall cracks in the S2 layer and weakening of the cell wall. The resulting weakening leads to improved biomass grindability. For the samples treated at 260 and 300 ◦C, hemicellulose was mostly decomposed resulting in a weakening of the cell walls and, subsequently, an improvement in grindability.

**Figure 5.** (**a**) Illustration of three main regions in the cell wall: the middle lamella (ML), the primary wall (P), and the three layers of the secondary wall (S1, S2 and S3) [44]; (**b**) Proposed ultrastructural model of the arrangemen<sup>t</sup> of hemicellulose, cellulose and lignin in the S2 layer [32].

Torrefaction decomposes most of the hemicellulose. Therefore, gasification of torrefied wood should result in improve syngas quality since hemicellulose and lignin may produces a higher tar concentration in the syngas compared to cellulose [6]. This benefit is also confirmed by gasification of torrefied biomass pellets which resulted in both improved energy efficiency and syngas quality [15,23]. The lignin contents of treated woods increased as the treatment temperature increased (Table 1). The high heating values of lignin are higher than the cellulose and hemicellulose due to higher degree of oxidation of the cellulose and hemicellulose [4]. The biochar with higher lignin content has the high heating values that is desired combustion property. On other hand, the air-steam gasification conversions of lignin (52.8%) was much lower than cellulose (97.9%) and hemicellulose (92.2%) [34]. Therefore, the production method of biochar used for gasification should be considered.

### *3.2. Physical Properties of Untreated and Treated Woods*

### 3.2.1. Physic Properties of Color, Weight, Dimensions, and Bulk Density

Table 2 shows the physical properties of untreated versus treated wood at 220, 260, 300, 350, 450 and 550 ◦C. Thermal treatments were found to darken the wood samples, as expected. The samples changed from the light brown of the untreated wood to dark cinnamon after the 220 ◦C treatment and essentially black after the 260 ◦C treatment. The darkening of the wood might be due to losses of arabinose and xylose that are converted to chocolate-brown colored furfural [32]. This agreed with the TGA test results on chemical composition changes of treated woods. Some hemicellulose in the 220 ◦C treated wood, most hemicellulose in the samples treated at 260 and 300 ◦C, and all hemicellulose in the samples treated at 350, 450 and 550 ◦C was decomposed.

The dimensions of the wood were reduced in all directions and shrinkage increased as treatment temperature increased (Table 2). The shrinkage observed is a result of the chemical reactions associated with cell wall component decomposition and subsequent structural changes during the thermal treatments. In the axial/longitudinal direction (height), the shrinkage for the 220 and 260 ◦C treated samples were close to zero; shrinkage started at 300 ◦C (0.41%) and increased significantly at 350 ◦C (6.48%). Our results are similar to Davidsson and Pettersson who reported that the longitudinal shrinkage of birch wood particles was close to zero below 300 ◦C and 5% at 350 ◦C. The longitudinal shrinkage of wood is a result of decomposition of cellulose in the microfibrils in the S2 layer of the cell

walls according to the microfibril dominance theory suggested [45]. Our TGA test results agreed with the axial shrinkage. Cellulose in the 220 ◦C and 260 ◦C treated woods did not decompose during the thermal treatment. Some cellulose in the 300 ◦C treated wood decomposed and cellulose in the 350 ◦C treated wood decomposed completely during these thermal treatments.


**Table 2.** Physical property changes of untreated and treated woods at each treatment temperature.

The shrinkages along radial (width) and tangential directions (depth) were similar in behavior and were different from the axial shrinkage. The shrinkages in these two directions started at low temperature, approximately 220 ◦C, primarily due to hemicellulose decomposition (Table 1). A large increase in the percent shrinkage was observed between 300 and 350 ◦C due to cellulose decomposition. Both the radial and tangential direction shrinkages were more than the axial direction. This may be because decomposition of hemicellulose and lignin mainly contributed to the radial and tangential shrinkages. Since cellulose is arranged in fibrils mainly in the axial directions and both hemicellulose and lignin bond the cellulose fibrils together, hemicellulose decomposition, primarily at low temperature, and lignin decomposition, in all temperature ranges, with a peak at a temperature higher than the cellulose decomposition temperature, is primarily responsible for the observed shrinkage (Figure 4). The order of shrinkages from high to low was width > depth > height. Based on the ultrastructural model (Figure 5b), the thin hemicellulose channels bond cellulose microfibrils on their radial faces and the hemicellulose decompositions results in the shrinkage in the radial direction.

The shrinkage was caused by the decomposition of cell wall components that resulted in loss of a portion of the mass from the biomass. Regression analyses of the shrinkages (S%) along radial, tangential and longitudinal direction as a function of the weight loss (w %) during the thermal pretreatments were carried out on experimental data. Formulas and R-squared (R2) of the regression analysis are listed in Equations (3)–(5). The formulas show good agreemen<sup>t</sup> with experimental data.

For radial direction/width,

$$\text{Sr} = 2.4533 \text{e}^{0.0349 \text{w}}, \text{R}^2 = 0.9911 \tag{3}$$

For tangential direction/depth,

$$\text{St} = 0.5201 \text{e}^{0.0522 \text{w}}, \text{ R}^2 = 0.9955 \tag{4}$$

For tangential direction/Height,

$$\text{Sl} = 0.0085e^{0.103\text{w}}, \text{ R}^2 = 0.9538 \tag{5}$$

The variations in shrinkage in the three sample reference directions are important for modeling studies of the wood thermal pretreatment. The shrinkage affects heat transfer to the particle and gas flow within the particle [25].

The weight losses of treated samples at 260 and 300 ◦C were 27.57% and 39.21%, respectively (Table 2). It was mainly caused by hemicellulose decomposition/devolatilization to volatiles along with the cellulose and lignin decomposition (Table 1). These results were comparable to Yan et al. who

reported that mass yields (100 - weight loss%) of pine woods after torrefaction at 275 and 300 ◦C were 74.2% and 60.5%, respectively [46]. Losses of weight and volume increased as treatment temperature increased. Losses of weight and volume of the pyrolyzed wood samples were much higher than those of the torrefied samples. There was a large step in percentage weight loss between 300 and 350 ◦C, which is more than for any other temperature step. This is due to the decomposition of cellulose which the primary component of the wood. At 350 ◦C, the weight loss (62.07%) and volume loss (37.39 %) were much higher than the weight loss (39.21 %) and volume loss (15.11%) at 300 ◦C. The weight losses at the higher temperature treatments of 450 to 550 ◦C increased slightly from 69.70 to 73.22%. Those results were close to Abdullah and Wu who reported that char yields of treated mallee wood at 300 and 450 ◦C were ~56% and ~27%, respectively [10]. The weight loss during low temperature pyrolysis was much higher than observed during torrefaction.

The measured bulk density of the untreated cherry wood was 540 kg/m<sup>3</sup> (0.54 g/cm3) which falls in the typical range [16], The densities of the treated wood samples decreased with increasing temperature over the temperature range of 220 to 350 ◦C but changed little over the range of 350 to 550 ◦C. The bulk density of wood is dependent upon cellular diameters and wall thickness [47]. During the thermal treatments, weight loss and size reduction caused by the decomposition of chemical components in the cell wall resulted in the density changes. Weight losses higher than volume losses resulted in the density reduction with increasing temperatures. The low density of treated wood may be addressed through densification (pelletizing for example) to improve the handling, transportation, storage and conversion of the wood [14,15].

Regression analyses of the weight loss (w%), volume loss (v%) and density (d g/cm3) as a function of the treatment temperature (T ◦C) were carried out on experimental data. Formulas and R-squared (R2) of regression were listed in Equations (6)–(8). The formulas show good agreemen<sup>t</sup> with experimental data.

For weight loss,

$$\mathbf{w} = -0.0007\mathbf{T}^2 + 0.7451\mathbf{T} - 114.31/\mathbf{R}^2 = 0.9788\tag{6}$$

For volume loss,

$$\mathbf{v} = -0.0004\mathbf{T}^2 + 0.4607\mathbf{T} - 81.733\mathbf{, R}^2 = 0.9555\tag{7}$$

For density,

$$\mathbf{d} = 0.000003 \mathbf{T}^2 - 0.0026 \mathbf{1} \mathbf{T} + 0.9146, \mathbf{R}^2 = 0.9774 \tag{8}$$

### 3.2.2. Grindbility of Untreated and Treated Woods

The untreated and treated at 220 ◦C samples were incompletely milled resulting in a fraction of those materials not being able to pass the bottom screen (0.8 mm) of the hammer mill. The untreated wood was ground three times and during grinding it was occasionally stuck in the mill between hammer-grinding stators and hammer-rotors. The ground samples were sieved into four size fractions: >500 μm, 500–212 μm, 212–106 μm and <106 μm. Table 3 lists the particle size distribution of the milled untreated and treated woods. The grindability of treated woods improved as the percentage of particles with the lower size fractions increased. The treated wood at 260 and 300 ◦C had 23.5% and 55.5% passing through 212 μm, respectively, compared to only 11% of the untreated wood. The results are similar to Aria et al. who reported that the percentage of particles less than 150 μm was double that of untreated wood for wood treated at 240 ◦C for 0.5 h and ground.

From the cell well composition study (Table 1), hemicellulose contents of the treated woods at 260 ◦C (10%) and 300 ◦C (7%) were lower that of the untreated wood (24%). Wood is essentially a series of elongated tubular fibers or cells and cemented together [31,32]. Hemicellulose acts as a matrix for the cellulose along with lignin. The decomposed hemicellulose resulted in the cell walls weakening and subsequently improved grindability as shown by the increase in the percentage of small particles versus the untreated wood.

For the treated wood at 350 ◦C, the percentage that passed through the 106 μm sieve increased to 16.7% compared to 3% of the treated wood at 300 ◦C. However, further treatment at higher temperatures did not further improve the grindability. The results are similar to Abdullah and Wu (2009) who reported that grindability of treated wood at 300 ◦C was drastically improved but increasing temperature to 500 ◦C resulted in only a small additional improvement in the grindability [10]. Therefore the torrefaction of wood can improve the wood grindability. The torrefied wood may be directly fed with coal for co-firing in a PC boiler and for co-gasification in an entrained-flow gasifier that require fine particle size of fuels.


**Table 3.** Particle size distribution of the milled untreated and treated woods.

### *3.3. Proximate Analysis and High Heating Value of Untreated and Treated Woods*

The proximate analysis of untreated and treated woods reveals the changes in moisture, volatile matter, fixed carbon and ash after thermal pretreatment. The moisture content of the untreated wood sample used in this study was low at 6% and similar to the 6.5% of reported for oak wood [16]. Table 3 shows the proximate analysis results of untreated and treated woods at 220, 260, 300, 350, 450 and 550 ◦C. The VM of untreated wood is high (88.1%), while its FC (11.5%) and ash (0.4%) are low. These untreated wood proximate analysis results are similar to that reported for beech wood with VM 84.2%, FC 15.5% and ash 0.3% [14]. The thermally treated woods were decomposed and light volatiles lost so the VM in the treated woods decreased and the FC increased with increasing temperature. The FC of pyrolyzed woods at 350, 450 and 550 ◦C are 54.3 to 83.4% and much higher than those of the torrefied woods at 220, 260 and 300 ◦C 13.9 to 24.9%. The pyrolyzed woods has high fuel quality, similar to coal which has mean FC 43.9% [48].

The calculated HHVs of untreated and treated woods in weight (MJ/kg) and volumetric value (MJ/m3) were listed in Table 4. The HHVs of torrefied woods at 220, 260 and 300 ◦C were 18 to 22 MJ/kg and in the range of 16–29 MJ/kg of torrefied biomass [15]. The heating values (by weight) of torrefied woods at 260 and 300 ◦C increased 6.1% and 15.9% compared to the untreated wood. The torrefied wood with improved fuel quality may be utilized as solid fuel with higher process efficiency. The pyrolyzed woods at 350, 450 and 550 ◦C were 25 to 32 MJ/kg and close to the mean HHV of coal of 25 MJ/kg [48]. Therefore, the pyrolyzed woods are compatible for utilization with coal. The HHVs of the treated woods by weight increased with increasing treated temperature due to increasing weight loss due to devolatilization hemicellulose and cellulose. But the HHVs of the treated woods in a volumetric basis decreased because weight losses were higher than volume losses. These results agree with those reported [17]. Pelletizing can significantly increase the volumetric energy density of the treated biomass facilitating transport and storage, leading to savings in logistics [9].

Fuel ratio is defined as a ratio of fixed carbon to volatile matter (FC/VM) and is an important solid fuel property [5]. From the proximate analysis results (Table 4), fuel ratio of the treated woods increased as the volatiles decreased due to the thermal treatment. The fuel ratios of the torrefied woods were 0.16–0.34 and lower than lignite coal 0.85 [5]. Those of the biochars at 350, 450 and 550 ◦C were 1.2, 2.8 and 5.6, respectively. The biochar from the wood treated at 350 ◦C (1.2) was close to that of high volatile bituminous coal (0.92) [19] and bituminous coal 1.56 [20]. So the low temperature pyrolysis improves wood combustion properties but the torrefaction did not [12]. The biochars were more suitable than raw biomas for co-firing with coal and increased thermal conversion efficiency [5,24].


**Table 4.** Proximate analysis and high heating value changes of untreated and treated woods at each treatment temperature.

During the thermal treatment, the mass loss results in energy loss with respect to the untreated biomass. The treated biomass is a solid fuel so its energy content can be evaluated using energy yield. The energy yield is defined as: energy yield (%) = mass yield (%) \* HHVf/HHVo, where mass yield = 100 − weight loss (%), HHV is high heating value, subscript o and f refer to the untreated and treated biomass, respectively [22]. The energy yields of the torrefied woods were in the range 70.3–86.5% and those of the pyrolyzed woods in the range 41.8–56.0%. With respect to energy yield, the torrefied woods are superior compared to the pyrolyzed woods [12]. The energy loss may be recovered by utilizing the volatiles generated during the pretreatment.

### *3.4. Morphological Study of Untreated and Treated Woods by SEM*

Figure 6 shows the SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm; vertical as images are presented) of the tangential section views of the untreated and the 260 and 550 ◦C treated woods. There are fibers, rays seen in end-view formed by stacks of ray parenchyma cells and pores visible in the tangential section images. Identified morphological structures were seen in all three woods. Figure 7 shows the SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm; vertical as images are presented) of the radial section views of the untreated and the 300 and 550 ◦C treated woods. There are fibers, rays that look like a brick wall crossing in the longitudinal direction and pores visible in the radial section images. Figure 8 shows the SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm) of the cross section views of the untreated and the 300 and 550 ◦C treated woods. There are many pores of various sizes and arrangements, fibers and annual or growth rings visible in the cross section images. Overall, the morphological features of the wood remain intact during the thermal treatments, which is in agreemen<sup>t</sup> with Haas et al. [49] who studied the pyrolysis of poplar wood up to 700 ◦C using microscopy in real-time and Winandy and Rowell [32] who studied the pyrolysis of pine wood at 295 ◦C. Figure 9 shows the SEM images (horizontal field width = 320 μm; scale bar of 100 μm) of cross sections of the untreated and the 350 and 450 ◦C treated samples. The cell walls were thinner after the thermal treatments.

Along the axial direction (Figure 6), the untreated and the 260 ◦C treated samples did not significantly shrink: however, the 550 ◦C treated sample shrank about 15%, which agreed with the physical property tests of sample height (Table 2). Along the tangential direction, the 260 ◦C treated samples shrank slightly; however, the 550 ◦C treated samples shrank 22% compared with the untreated sample, which agreed with the physical property tests of sample depth (Table 2). There was more shrinkage along the tangential direction than the axial direction. Along the radial direction (Figure 7), the samples treated at 300 and 550 ◦C shrank 10 and 29% respectively, compared with the untreated one, which agreed with physical property test of sample width, Table 1.

**Figure 6.** SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm; vertical as images are presented) of the tangential section views of the untreated and the 260 and 550 ◦C treated cherry wood samples. White arrows indicate rays and fibers seen in end-view formed by stacks of ray parenchyma cells, also pores are visible in the tangential section images. Yellow bars indicate shrinkage along the axial direction using reference structures. Red bars mark shrinkage along the tangential direction using reference structures.

**Figure 7.** SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm; vertical as images are presented) of the radial section views of the untreated and the 300 and 550 ◦C treated cherry wood samples. White arrows point out fibers and rays that look like a brick wall crossing in the longitudinal direction, also pores are visible in the radial section images. Yellow bars mark shrinkage along the axial direction (using reference structures). Red bars mark shrinkage along the radial direction (using reference structures).

*Energies* **2018**, *11*, 25

**Figure 8.** SEM images (horizontal field width = 1.28 mm; scale bar of 500 μm) of the cross section views of the untreated and the 300 and 550 ◦C treated cherry wood samples. White arrows point out fibers and pores. Yellow bars mark annual or growth rings and its shrinkage along the radial direction (using reference structures).

**Figure 9.** SEM images (horizontal field width = 320 μm; scale bar of 100 μm) of the cross section views of the untreated and the 350 and 450 ◦C treated cherry wood samples. White bars point out cell walls which were thinner after the thermal treatments.

(**a**) Untreated (**b**) 350 °C (**c**) 450 °C
