Preparation and Characterization of Heat-Treated Douglas Fir Wood with Core–Shell Structure

Abstract

Wood heat treatment has been generally accepted as an effective wood modification technology as it improves the dimensional stability and biological durability of wood. However, the property improvements are obtained at the cost of reduced mechanical properties. In this study, heat-treated Douglas fir wood with thermally modified shell and unchanged inner core was prepared by surface heating to make possible the broader applications of heat-treated wood. Colour comparison, dynamic vapor sorption and dynamic mechanical analysis were performed to characterize the physico-mechanical performance of the shell and core of the treated wood. The results reveal a darkened, hydrophobic and rigid shell and a barely changed inner core. X-ray photoelectron spectroscopy shows pronounced degradation of polysaccharides in the surface layer. In contrast, the relative contents of different carbon components in the core layer are similar to that of the untreated wood, indicating the reason for the performance difference between the shell and the core of the treated wood. The initial wood moisture content plays an important role in controlling the temperature gap between the core and the shell during treatment and can be used as a key parameter to adjust the structure and performance of the heat-treated wood.

1. Introduction

Wood is a raw material widely used from prehistoric times to modern day. Its long-lasting popularity is derived from its natural origin, universal availability, good machining performance and extraordinary versatility. However, as a hygroscopic biomaterial, wood is prone to deform or degrade when serving in unfavourable environments. Various modification technologies have been explored to improve the performance of wood, among which heat treatment is a generally accepted and commercialized procedure [1]. The treatment, commonly carried out in an over 160 °C inert environment, enables wood improved dimensional stability and biological durability. The property modification is more significant when higher treatment temperature, relative humidity (RH) and longer duration are applied [2,3]. It has been reported that the treatment of Chinese fir (Cunninghamia lanceolata) in silicone oil at 160 °C reduces the radial and tangential swelling coefficients by 49% and 34%, respectively [4]. For the same species, treatment in 230 °C steam improves the dimensional stability by 73% for heartwood and 71% for sapwood [5]. In terms of decay resistance, the durability grade of “durable” or “very durable” according to CEN/TS 15083-2 [6] is obtained when the treatment temperature is above 215 °C [7]. However, the performance improvement of heat-treated wood is achieved at the cost of strength loss, as high temperature also causes the thermo-degradation or reconstruction of wood cell wall components [8,9,10,11]. The treatment of poplar (Populous tomentosa) wood at 200 °C for 3 h shows a decrease in modulus of rupture (MOR) by 29%, and for the treatment at 230 °C, 5 h, the MOR is down by as much as 54% [12]. Some mechanical properties may be less susceptible to heat treatment, but it can be concluded that most mechanical properties deteriorate with the increase in treatment temperature, especially when it is above 200 °C [13,14]. For this reason, heat-treated wood is used only in non-structural applications.

Many attempts have been made to address the side effect of wood heat treatment. Some tried to improve the mechanical properties of heat-treated wood by combining heat treatment with wood densification [15,16]. Another idea is to combine heat treatment with chemical impregnation, and various chemical agents, including boron compounds [17,18,19], melamine-urea-formaldehyde [20] and carnauba wax/organoclay emulsion [21], have been tried with encouraging results. Combined treatments may also further enhance the stability or durability of wood compared to conventional heat treatment [22], but they make the treatment process more complex and trigger concerns over set recovery of the densified lumber or environmental impacts of the chemical-added lumber.

In this study, a core–shell structured heat-treated wood is prepared by surface heating with the intention to develop a thermally modified wood with hydrophobic shell and unchanged inner core. Studies on heat-treated wood with uneven cross-sectional structure are pretty limited. Efforts have been made to develop heat-treated wood with unevenly distributed density over a cross section by chamber heating, but operational difficulties arose as the samples would not be placed in the chamber until the treatment temperature was reached [23]. Cermak et al. [24] and Kymalainen et al. [25,26] later suggested a surface modification method by hot plate heating. The prepared boards exhibited a surface-substrate structure mainly used for cladding or decking. Additionally, applying hot plate heating, this study tries to develop a sandwich-like core–shell structure and the treatment is on dimension lumber mainly for load-bearing applications. The main challenge for the establishment of such a structure is to keep the core temperature low enough to avoid major cell wall chemical changes while the shell is subjected to prolonged heat treatment. For thick lumber, the possible surface check caused by the tensile stress in the surface layer should also be avoided. As part of our work on wood heat treatment, this paper focuses on the effective establishment of the core–shell structure by investigating the temperature evolution of the sample core and shell during the treatment process and their after-treatment property characteristics.

2. Materials and Methods

2.1. Wood Heat Treatment

Douglas fir (Pseudotsuga menziesii) dimension lumber was provided by Jiangsu High Hope Arser Co., Ltd. The lumber was imported from Canada and the grade was select structural. Four blocks 400 mm long, 100 mm wide and 100 mm thick were cut from the same piece of lumber. The initial moisture content (MC) of the blocks was 14.1% and two of them were further dried to 4.6% to evaluate the influence of MC on the treatment process. Before the treatment, both ends of the samples were sealed by resin and aluminium foils to minimize the heat and mass transfer in the longitudinal direction, and three 50 mm-deep holes were drilled on one side face of each sample, as illustrated in Figure 1. K-type thermocouples were embedded in the holes to monitor the surface and core layer temperatures during the experiments.

Heat treatment was performed using a hot press (Dahua Machining & Manufacturing Co., QD42, Kunshan, China). The temperature of the hot plates was set at 200 °C. When the plate temperature reached 150 °C, the samples were loaded and heated until the core layer temperature reached 140 °C. The duration of the treatment varied from 380 min to 480 min depending on the initial moisture contents of the samples. During the heat treatment, the samples were slightly pressed between the plates so that effective heat conduction was ensured while no surface densification developed. The sample temperature evolution was recorded by an Omega RDXL12SD temperature recorder. All the treated samples were stored at 20 °C, 65% RH before further experimental use.

2.2. Wood Colour Measurement

The colour parameters of both the sample surface layer and core layer (obtained after ripping the sample from the middle, Figure 2) were measured using a chroma meter (Konica Minolta, CR-400, Tokyo, Japan). For each kind of layer, twelve evenly distributed points were measured, and the data were statistically analysed.

The CIELab system was used for colour evaluation, where a three-dimensional space is built with the vertical coordinate axis for lightness (L*) and two horizontal axes for colour (a* and b*). The L* axis ranges from black (0) to white (100); the a* axis extends from green (minus values) to red (positive values), and the b* axis from blue (minus values) to yellow (positive values). A colour can thus be quantitatively described by the three coordinates in the space.

Colour difference (ΔE*) between the treated and the control sample was calculated according to Equation (1):

$$\mathsf{\Delta }{E}^{*}={\left(\mathsf{\Delta }{L}^{*2}+\mathsf{\Delta }{a}^{*2}+\mathsf{\Delta }{b}^{*2}\right)}^{1/2}$$

(1)

2.3. Dynamic Vapor Sorption (DVS)

Small sections were taken from different samples, as illustrated by Figure 2. They were then cut by a microtome into slices weighing 4 mg each for dynamic vapour sorption tests. The slice-shaped samples were chosen because their weights are well below the maximum sample weight limit of the DVS apparatus (Surface Measurement Systems, Intrinsic, London, UK) and the shape keeps the original cell wall structure so that the hygroscopic behaviour of the samples can be better evaluated. The relative humidity (RH) range of the sorption cycle was between 0% and 98% and the RH step lengths are shown in

Table 1. Relative humidity range and step lengths adopted in the dynamic water vapor sorption cycle.

Relative Humidity/%	Step Length/%
0~10	5
10~90	10
90~95	5
95~98	3

. The samples were tested at 25 °C and were considered to have reached equilibrium when the rate of mass change was less than 0.002% min⁻¹ for 10 min at each RH.

2.4. Dynamic Mechanical Analysis (DMA)

Wood strips with the dimension of 35 mm × 10 mm × 4 mm (L × R × T) were taken from different samples according to Figure 2. Since the equilibrium moisture content (EMC) of heat-treated wood is lower than the untreated wood in the same environment, all the samples were dried to oven dry status to eliminate the influence of MC on the results. A Q800 dynamic mechanical analyser (TA Instrument, New Castle, DE, USA) was used to measure the storage modulus (E’) and loss tangent (tan δ) of the samples from −100 °C to 300 °C at a heating rate of 3 °C min⁻¹. The tests were performed in dual cantilever mode at a frequency of 1 Hz and displacement amplitude of 20 µm.

2.5. X-ray Photoelectron Spectroscopy (XPS)

Samples with the dimension of 5 mm × 5 mm × 1 mm (L × R × T) prepared according to Figure 2 were analysed by an X-ray photoelectron spectrometer (Kratos, AXIS Ultra DLD, Manchester, UK). The analyses were performed in a lower than 7 × 10⁻⁸ Pa environment using an achromatic Al Kα source (1486.6  eV) operating at 150 W. High resolution XPS spectra of C1s peak between 280 eV and 295 eV were obtained with an energy channel step width of 0.1 eV and pass energy of 40 eV. Four subpeaks were obtained from the C1s peak by curve fitting using the XPS Peak 4.1 software.

3. Results and Discussion

3.1. Sample Temperature Evolution during the Treatments

As shown in Figure 3, the sample with the initial MC of 14.1% (sample MC14) and the sample with the initial MC of 4.6% (sample MC4) exhibited almost the same temperature rising mode for the surface layers during the heat treatment. It took around 400 min for both surface layers to reach 180 °C, a temperature generally believed enough to modify wood physico-mechanical properties. By contrast, the core layers showed temperature evolution modes different from each other. When the temperature was below 100 °C, the two temperature curves overlapped with each other. Over that point, however, sample MC4 exhibited an almost linear temperature rising rate, while sample MC14 experienced a temperature stagnation at around 100 °C. As a result, it took the core layer of sample MC4 380 min to reach 140 °C, at which no pronounced chemical changes are supposed to occur in the wood cell wall [13]. For the core layer of sample MC14, the time to reach the same temperature was 480 min.

The core-layer temperature stagnation was observed in studies where wood with a large cross-sectional size was heated. It was suggested that this is the result of drying-caused water evaporation and will last longer with higher initial wood MC [27,28]. According to this, the continuous temperature rise of sample MC4 may be attributed to its much lower initial MC. The difference core layer temperature rising modes suggest that by setting the initial wood MC level, the temperature gap between the surface layer and the core layer during the treatment could be adjusted, as could the essential treatment time for the surface layer to be thermally modified. Since the degree of wood modification is closely related to the treatment temperature and duration, the adjustable temperature gap and the essential treatment time means that the structure and performance of the core and the shell of the treated wood could be tailored according to application demand. Nevertheless, it should be noted that the choice of initial MC is not unlimited because too high MC will cause greater internal stress and higher risk of defects such as warping and checking. Furthermore, moisture will penetrate from the surface to the core driven by inward vapor gradient during the hot plate heating process [29]. All these factors make the control of sample temperature profile rather complex and deserving of more comprehensive study.

3.2. Colour and Appearance

The heat treatment darkened the sample surface layers from light yellow to dark brown, while the core layers remained almost the same as the control sample (Figure 4). The darkening penetrates around 5–6 mm beneath the sample surfaces. For sample MC4, the colour difference between the surface layer and the control sample is 30.7, while the difference between the core layer and the control sample is only 3.4. Similar results can be found in sample MC14 (Figure 5). On each surface, only a small variation in colour parameters exists between different measuring points. Taking the surface layer of sample MC4 as an example, the standard deviations of L*, a* and b* are, respectively, 3.16, 0.55 and 0.92, while the corresponding means are 41.05, 7.83 and 14.00. Such colour uniformity can also be found on the surface layer of sample MC14 (Figure 5) and suggests that the surfaces of the treated samples have been evenly treated.

Darkened colour is the typical visual change in wood after heat treatment [30]. It has been generally agreed that the degree of colour change can be used to indicate the heat treatment intensity and is closely related to the chemical changes in wood cell wall components and property alterations to the treated wood [31,32]. Therefore, the colour comparison indicates that the surface layers underwent significant structural changes while the structure of the core layers was much less influenced, and that the performance of the two layers were distinguished from each other.

Smooth sample surfaces were obtained after the treatments, and the appearance kept intact after around one-year exposure to the ambient environment. It seems that no pronounced internal stress was developed by the heating process and the fluctuation of environmental conditions.

3.3. Hygroscopicity

The DVS analysis shows significant differences between the surface layer and core layer of the heat-treated samples with respect to sorption behaviour. The sorption isotherms of the core layers almost overlap the isotherm of the control sample in both adsorption branch and desorption branch (Figure 6a,c), meaning that the core layer of the treated samples exhibits similar sorption behaviour to the control sample. By contrast, the sorption isotherms of the surface layers are below that of the control sample due to the reduction in EMC (Figure 6b,d). Reduced EMC is the typical performance change for thermally modified wood, and the EMC reduction level in this study is comparable to those with similar treatment temperatures [33,34]. Such similarity shows that the sorption behaviour of the surface layers is close to traditional heat-treated wood, and hydrophobic shells have been formed on the treated samples. Since moisture mainly penetrates into the wood from the surface, the hydrophobic shell will improve the dimensional stability of the treated wood as a whole.

Compared to sample MC4, the EMC difference between the surface layer of sample MC14 and the control sample is more significant, as illustrated by Figure 6 and numerically demonstrated by

Table 2. EMC of the surface layer of the heat-treated samples expressed as the percentage of that of the control sample at 4 relative humidity levels.

Relative Humidity (%)	Sample MC4 (%)		Sample MC14 (%)
	Adsorption	Desorption	Adsorption	Desorption
10	99	93	95	73
30	100	98	91	82
60	99	93	88	85
90	90	89	83	83

. The difference may be attributed to the longer treatment time for sample MC14. As shown in Figure 3, the MC14 was treated for 510 min, during which the surface layer temperature was kept over 160 °C for 400 min. Meanwhile, the treatment time for sample MC4 was 400 min and the shell temperature was over 160 °C for only 280 min. According to the European standard on thermally modified timber [35], 160 °C is the threshold temperature for wood thermal modification. This means the surface layer of MC14 underwent longer essential treatment time, and since prolonged treatment time induces greater property modification of the heat-treated wood [12,36], the surface layer of sample MC14 becomes more hydrophilic.

3.4. Dynamic Mechanical Analysis

The storage modulus of a material is related to its reversible response to a force applied to it and refers to the material’s stiffness. According to Figure 7, the storage moduli of all the samples decrease as temperature rises, which is in agreement with other studies on viscoelastic properties of wood [37,38,39] and is attributed to the intensified thermal motion of wood molecules at higher temperatures [37]. A dramatic storage modulus drop can be observed when the temperature is over 200 °C. Such modulus loss is supposed to be caused by the softening of wood amorphous components as the glass transition points of the non-crystalline part of cellulose, hemicelluloses and lignin in oven dry wood are 200~250 °C, 150~220 °C and ~205 °C, respectively [40], and the pronounced relaxation process over 200 °C featuring in Figure 8 gives direct support for it.

It can be found in Figure 7 that the storage modulus curves of the surface layer samples are well above that of the control sample, while the curves of the core layer samples are closer to that of the control sample. The higher storage moduli indicate the formation of rigid shell on the heat-treated samples. Increased rigidity has been observed by many in heat-treated wood and suggested to be the combined effects of wood cell wall component changes during the heat treatment [13,41,42]. One major contributor is the thermal degradation of hemicelluloses, which causes the relatively flexible inter-molecular bonds replaced by more rigid bonds. Furthermore, the cross-linking of lignin during heat treatment also increases the rigidity of the cell wall matrix materials and reduces the molecular mobility even further. In contrast, the core layers show similar dynamic mechanical behaviour to the control sample.

The dynamic mechanical behaviour difference between the core layer and the surface layer of the heat-treated samples is also illustrated by Figure 8. Two relaxation processes can be observed in the figure: a relatively minor one peaking at around −50 °C (β relaxation) and the pronounced one at around 260 °C (α relaxation). The first relaxation is accompanied by slight storage modulus loss (Figure 7) and is considered to be related only to the motion of side chains of wood molecules, while the second one is caused by the main-chain movements and marks the softening of the amorphous components of the wood cell wall [39].

For the surface layers, the β relaxation lasts over a broader temperature range than the control sample and causes the two loss tangent curves cross each other at around 50 °C. Their α peaks are narrower and shift to slightly higher temperatures (Figure 8b,d). These changes obviously indicate the chemical changes of the surface layer samples. The change in shape in the β peak is related to the thermal breakdown of hemicelluloses, the main cell wall components with side chains. The shift in α peak, on the other hand, is influenced by the content and structure of lignin, since it has been reported that a high content of methoxyl group in lignin lowers the softening temperature [43], and that demethoxylation of lignin occurs during wood heat treatment and a more condensed lignin structure is achieved as a result [44].

The loss tangent curves of the core layers, in contrast, show β relaxation temperatures similar to the control sample. However, their α relaxation is not only much different from the control sample but also different from each other (Figure 8a,c). Such a difference implies that the softening behaviour of the core layers has also been influenced by the heat treatment, although the core layer temperature did not surpass the threshold for thermal modification during treatment. Given that the two core layer samples also show different softening behaviour from each other, original sample difference may be another factor contributing to it.

3.5. XPS Analysis

The C1s peak of each sample obtained by the detailed scan is deconvoluted into four sub-peaks according to the chemical status of the four carbon components (

Table 3. Binding energy and binding type of the four carbon components in wood.

Carbon Components	Binding Energy (eV)	Binding Type
C1	284.8	C-H, C-C
C2	286.5	C-OH, C-O-C
C3	288.0	C=O, O-C-O
C4	289.5	O-C=O

). In the wood cell wall, the C1 component consisting of C-H and C-C linkage is attributed to aliphatic and aromatic carbons in lignin and extractives and its contribution to cellulose is negligible [45,46]. The C2 component includes the carbon atoms bonded with one oxygen atom in the form of C-OH or C-O-C, and the C3 component corresponds to the carbon atoms bonded to a carbonyl or two non-carbonyl oxygen atoms in the form of C=O or O-C-O [46]. Both C2 and C3 components are universally distributed in cellulose, hemicelluloses and lignin. The D-galactose, D-glucose and D-mannose, which are the main units of the polysaccharides in softwood, contain 5 C2 atoms and 1 C3 atom in each unit. In lignin, the C-O-C and C-OH (C2) are commonly found, and the C=O group (C3) is the main chromophoric group but is less abundant. In contrast to C2 and C3, the C4 component in the form of O-C=O is mainly found in hemicelluloses, as there exist substituted acetyl groups on the galactoglucomannan backbone of softwood hemicelluloses for every 3 to 4 hexose units [47].

Figure 9 illustrates the high resolution C1s peaks of all the samples with subpeaks separated by deconvolution, and the contribution of the four carbon components to the respective C1s is listed in

Table 4. Contribution of C1, C2, C3 and C4 components to the C1s peak from different samples.

Sample		Percentage (%)
		C1	C2	C3	C4
Control		55.6	28.8	11.5	4.1
MC4	surface layer	67.1	22.4	7.3	3.2
	core layer	57.6	26.2	11.9	4.3
MC14	surface layer	68.8	22.4	5.9	2.9
	core layer	53.8	30.2	11.8	4.2

. Heat treatment changed the percentages of the four carbon components of the surface layer sample. For both sample MC4 and sample MC14, the surface layer shows higher C1 content and lower C2, C3 and C4 contents relative to the core layer. On the other hand, the core layers show similar carbon component percentages to the control sample. Such a difference between the surface layer and the core layer verifies the chemical changes in the surface layers and the relative intact core layer chemical composition suggested in the previous discussion.

Heat treatment causes the degradation of hemicelluloses and the amorphous part of cellulose. For this reason, the relative contents of C2 and C3 decrease in the surface layers. Meanwhile, as lignin is the most thermally stable component in the wood cell wall, its relative content (C1) rises with the fall of polysaccharides content. Since the C4 component is mainly distributed in hemicelluloses, its relative content also decreases in the surface layer samples. For sample MC14, the percentage changes in C1, C2, C3 and C4 in the surface layer relative to the core layer are 27.9%, −25.8%, −49.2% and −29.3%, respectively. For sample MC4, the corresponding changes are 16.5%, −14.9%, −34.8% and −27.9%, which, once again, supports that the extended treatment time made the surface layer of sample MC14 undergo more pronounced modification.

4. Conclusions

Heat-treated Douglas fir lumber with core–shell structure was prepared by heating the lumber using hot plates. A darkened, rigid and hydrophobic shell formed on the treated wood while the appearance and properties of the inner core remained almost unchanged. The DMA and XPS analysis unveiled pronounced degradation of polysaccharides in the wood surface cell wall. In contrast, the relative contents of chemical components in the core layer were similar to that of the control sample, indicating the reason for the appearance and property difference between the surface and the core. The characteristic core–shell structure could be advantageous as the modified shell provides improved properties similar to traditional heat-treated wood and the unchanged core makes the mechanical performance less influenced.

Sample initial MC plays an important role in shaping the structure and performance of the heat-treat wood by controlling the temperature gap between the core and the shell during the treatment. As a result, samples with different initial MC were found to exhibit different extents of property modification. It is therefore expected that the exact performance of the heat-treated wood could be tailored by adjusting the initial wood MC.

Although further investigation is still needed to study the in situ performance of the treated wood, and its performance response to the treatment parameters, this study has presented a method to prepare a novel heat-treated wood with potential load-bearing applications.