Hygro-thermal coupled swelling in Chinese fir wood

2.1 Materials

Wood blocks were prepared from the sapwood regions of 35-year-old Chinese fir [C. lanceolata (Lamb.) Hook.]. After felling, wood blocks were slowly air-dried to avoid high-temperature or rapid drying. Specimens with dimensions of 6 × 4 × 0.5 mm³ (radial × tangential × longitudinal), containing the entire 17th growth ring, were prepared for swelling test. The surface of the samples was polished by removing 2-μm-thin slices with a cryomicrotome HM 560 (Microm, Germany). Nine specimens were prepared and dried in a sealed container over P₂O₅ at 25 °C for more than 9 weeks until a constant mass was achieved. MC of the dried specimens was approximately 0.6 %, with an air-dry density of 370 kg/m³.

2.2 Sorption isotherm

Sorption isotherms were conducted on ca. 30 mg chips by a dynamic vapor sorption apparatus (DVS Advantage 2, Surface Measurement Systems Ltd., UK). The temperature was preset at 25, 35 or 45 °C. The preprogrammed RH was firstly decreased to 0 to determine the initial dry weight of the specimen, then increased by 10 % each step, until reaching 97 % RH, and then decreased to 0 in the reverse sequence. At each RH level, the equilibration time was dependent on the weight change of the specimen. Once the weight change rate was <0.002 %/min, the RH was changed to the next humidity step. Three replications were performed for each temperature.

After obtaining the isotherms (Figure 2a), RH values corresponding to specific MCs (4.9 % and 9.3 %) at each temperature were determined. These RH values were used in subsequent swelling experiments to maintain consistent MC across temperatures. Specifically, the RHs corresponding to 4.9 % MC at the three temperatures were 30.0 %, 30.7 % and 31.1 % RH. The RHs corresponding to 9.3 % MC at the three temperatures were 60.0 %, 62.2 % and 63.1 % RH (Figure 2b). The increase in the required RH with increasing temperature indicated that, under identical RH conditions, the values of MC decreased as elevated temperature. This temperature dependence was more evident at higher RH levels, consistent with the trends observed in the sorption isotherms (Figure 2a). It should be noted that the differences among temperatures were relatively small, as the analysis focused on a narrow MC range within the hygroscopic region. Within this low MC interval, the sensitivity of MC to temperature variations was limited compared with studies covering broader humidity ranges or higher MC levels.

2.3 Swelling measurement under hygro-thermal conditions

Hygro-thermal swelling was performed inside a 1,200 × 900 × 200 mm³ humidity chamber (Figure 1). The controlled RH condition was provided by a modular humidity generator 32 (MHG 32) (ProUmid, Germany, sensitivity ± 0.1 % RH). The chamber contained a 1,100 × 800 mm² heating platform (DB100-2P, JOANLAB, China, sensitivity ± 0.1 °C) to set different temperature conditions. The temperature and humidity sensors were used to ensure that the temperature and humidity conditions in the humidity chamber were maintained at set conditions. The dried specimens were placed on the 25 °C heating platform and equilibrated at 0 %, 30 % and 60 % RH for 2 weeks, corresponding to MCs of 0.6 %, 4.9 % and 9.3 % MC, respectively. For each MC, RH was adjusted according to the target temperature: e.g., for 4.9 % MC, RH was set to 30.7 % at 35 °C and 31.1 % at 45 °C. The testing temperatures (25–45 °C) were chosen to approach the softening temperature of hemicelluloses under moisture plasticization while remaining below the typical softening range of moist lignin (Börcsök and Pásztory 2021). Referring to the setting of Goli et al. (2019), the progress was identified to be 1 °C step every 600 s with an increase of 0.3 % RH. The swelling of specimens with different MCs were conducted with reference to the humidity conditions obtained based on the sorption isotherm (Figure 2b). All images were captured after equilibrating for 48 h at each temperature and humidity condition, and three specimens were replicated at each MC.

Figure 1:

Experiment setup of hygro-thermal coupled swelling measurement.

Figure 2:

(a) Adsorption isotherms of Chinese fir and (b) RHs settings for controlled hygro-thermal environments.

Specimen deformation was recorded using a digital microscope VHX-7000 (Keyence, Japan). Images were analyzed with VIC-2D software (Correlated Solution Inc., USA) following the scattering pattern and subset selection method of Zhan et al. (2023) to obtain full-field swelling (ε). Radial (ε _yy ) and tangential (ε _xx ) swelling were calculated relative to 0.6 % MC at 25 °C. The strain compositions at a series of hygro-thermal conditions were illustrated in Figure 3. Specifically, the radial ( ε y y H ) and tangential ( ε x x H ) hygroscopic swelling were determined from the strain difference between specimens at higher MCs (4.9 and 9.3 %) and the reference (0.6 %) at 25 °C. The radial ( ε y y T ) and tangential ( ε x x T ) thermal expansion were calculated from specimens at 0.6 % MC at elevated temperature. Hygro-thermal coupled swelling (Δε _yy and Δε _xx ) was obtained by subtracting the corresponding thermal expansion and hygroscopic swelling from total measured. Δε _yy and Δε _xx were determined as:

Figure 3:

Schematic of strain composition at a series of hygro-thermal conditions.

The radial ( α rad T i ) and tangential ( α tan T i ) thermal expansion coefficients at different MCs are calculated based on ε _yy and ε _xx at a series of hygro-thermal conditions, which were determined as:

where T _i is certain test temperature (35 °C and 45 °C), T₀ is the initial temperature (25 °C), ε x x T i and ε y y T i are the swelling strain at T _i , ε x x T 0 and ε y y T 0 are the swelling strain at T₀.

The radial ( β rad T i ) and tangential ( β tan T i ) swelling coefficients at different temperatures are calculated based on ε _yy and ε _xx at a series of hygro-thermal conditions, which were determined as:

where MC _i is certain test MC (4.9 % and 9.3 %), MC₀ is the initial MC (0.6 %), ε x x MC i , ε y y MC i are the swelling strain at MC _i , ε x x MC 0 and ε y y MC 0 are the swelling strain at MC₀.

2.4 Calculation of off-axis swelling strain

Referring to the method proposed by Wang et al. (2025) for calculating the off-axis swelling, off-axis thermal expansion ( ε θ T ), hygroscopic swelling ( ε θ H ), and hygro-thermal coupled swelling (Δε _θ ) were calculated at given off-axis angle θ:

2.5 Cell wall proportion

Cell wall proportion in earlywood and latewood was quantified at representative locations using cross-sectional images acquired using the digital microscope. ImageJ (https://imagej.nih.gov/ij/) was used for image analysis. For each tracheid, the cell wall proportion was calculated as the ratio of double wall thickness to the corresponding cell dimension, and it was determined separately for the tangential and radial directions. At each location, 50 tracheids were evaluated to ensure statistical reliability.

2.6 FTIR spectroscopy

Powders of earlywood and latewood were each prepared and mixed with KBr at a ratio of 1:70 (by weight) and pressed into pellets for infrared analysis. Spectra were collected using an infrared spectrometer (VERTEX 80 V, Bruker, Germany). The measurement range was 400–4,000 cm⁻¹ with a resolution of 2 cm⁻¹ and 64 scans. Each measurement was repeated three times for both earlywood and latewood, and one representative spectrum from each set was presented.

2.7 Raman spectroscopy

Earlywood and latewood cross sections (30 μm thick) were mounted on glass slides, sealed with deionized water and nail polish, and analyzed using a confocal Raman microscope (Xplora plus Horiba, Japan). A 50× oil-immersion objective lens (NA = 1.40) was employed, with a 532 nm laser operating at 10 % in-tensity and an integration time of 1 s. The grating was set at 600 mm-covering a spectral range of 100–3,900 cm⁻¹, and the scanning area was sampled with a step size of 0.3 μm across the selected regions of interest. Raman imaging and spectral data were processed by LabSpec6 software for fluorescence background deduction, baseline correction, and data smoothing. Measurements were performed in triplicate for both the earlywood and latewood. One representative spectrum from each set was presented.

2.8 Crystallinity (CrI) and microfibril angle (MFA)

X-ray diffraction (XRD) was employed to characterize the cellulose crystallinity and average MFA of both earlywood and latewood. Diffraction measurements were performed at ambient conditions using a Bruker D8 Advance diffractometer (Bruker, Germany) equipped with Cu Kα₁ radiation (λ = 1.5406 Å), operated at 40 kV and 30 mA. Diffraction patterns were collected over a 2θ range of 5° ∼ 40° with a step size of 0.02° and a scanning speed of 5°/min. Data analysis focused primarily on the (200) reflection of cellulose, which is oriented parallel to the cellulose chain axis and typically appears at 2θ ≈ 22.5°. The CrI was determined using the peak height method, with three independent replicates for each tissue type, following the approach proposed by Segal et al. (1959):

where I₂₀₀ is the maximum intensity of the (200) peak at 2θ ≈ 22.5°, and I _am is the minimum intensity between the (200) and (110) peaks, typically located near 2θ ≈ 18°.

The tangent plugin was applied to identify tangents at the steepest slopes on each side of the diffraction peak, and their intersections with the x-axis were recorded to calculate the 2T value. MFA was then calculated as 0.6T (Cave 1966).

2.9 Statistical analysis

The statistical software, SPSS (SPSS Statistics 26, IBM, New York, NY, USA), was used for data analysis. Significant effects of thermal expansion and swelling coefficients of wood were analyzed by Duncan’s multiple comparison test (p = 0.05).

3.1 Microstructure and chemical composition

Figure 4a illustrated the intra-ring variation of direction-dependent cell wall proportion along the growth ring. Latewood exhibited lower MFA (12.2°) and higher CrI (55.0 %) than earlywood (Figure 4b). Given the relatively similar cellulose-related characteristics between earlywood and latewood, the differences in moisture-induced deformation, particularly under hygro-thermal conditions, were more likely attributed to variations in the amorphous matrix. FTIR spectra after normalization to the 1,160 cm⁻¹ band were shown in Figure 4c. The band at 1,732 cm⁻¹ was assigned to C=O stretching of hemicelluloses, whereas the lignin-related region was represented by the bands at 1,632 cm⁻¹ and 1,510 cm⁻¹. Semi-quantitative analysis using peak intensity ratios (Figure 4d) showed a higher I₁₇₃₂/I₁₁₆₀ ratio in latewood, indicating a higher relative hemicelluloses content. In contrast, the I₁₆₃₃/I₁₁₆₀ ratio was higher in earlywood, suggesting a higher relative lignin content, which was probably associated with the thinner cell walls and higher proportion of lignin-rich intercellular layers in earlywood (Lanvermann et al. 2013; Liszka et al. 2023). Raman spectra were extracted separately from radial and tangential cell walls in earlywood and latewood (Figure 4e), and spatial distribution of lignin-associated signals are shown in Figure 4f. The spectra confirmed a higher lignin-associated signal in earlywood than in latewood. Furthermore, in both tissues, the radial cell walls exhibited higher lignin contents than the tangential cell walls, indicating preferential lignification in radial tracheid walls (Kato 1968).

Figure 4:

Microstructure and chemical composition of earlywood and latewood: (a) Variation of cell wall proportion in the radial and tangential directions and MFA; (b) XRD patterns; (c) FTIR spectra in the range of 800–1,800 cm⁻¹; (d) peak intensity ratios in FTIR spectra-I₁₇₃₂/I₁₁₆₀ and I₁₆₃₃/I₁₁₆₀; (e) Raman spectra in the range of 1,000–1,700 cm⁻¹; (f) enlarged Raman spectra in the range of 1,500–1,700 cm⁻¹ and corresponding Raman spectroscopic images showing lignin distribution. Scale bars in inset (f): 10 μm.

3.2 Thermal expansion

Full-field swelling of Chinese fir at MCs of 0.6 %, 4.9 % and 9.3 % was obtained at 25, 35 and 45 °C (Figure 5). Higher temperatures and MCs resulted in greater swelling in both radial (ε _yy ) and tangential (ε _xx ) directions. Taking the specimen with 0.6 % MC at 25 °C as a reference, the thermal expansions ε y y T and ε x x T of wood at 35 °C and 45 °C are shown in Figure 5a₁, a₂, d₁ and d₂. Within the growth ring, ε y y T exhibited a pronounced gradient from earlywood to latewood (Figure 5a₁), while ε x x T showed little intra-ring (Figure 5a₂). In both directions, thermal expansion increased with temperature. At 0.6 % MC, ε y y T and ε x x T at 45 °C were markedly higher than at 35 °C, with larger increase observed in latewood than in earlywood, consistent with its higher cell wall proportion (Figure 4a). Thermal expansion in the radial ( α rad T i ) and tangential ( α tan T i ) directions was calculated for earlywood and latewood across MC levels (Figure 6). Coefficients measured over different temperature intervals showed no significant differences, consistent with the previous findings (Ross 2010). In contrast, both greater α rad T i and α tan T i were obtained at higher MCs. Compared to α rad 45 ℃ and α tan 45 ℃ at the 0.6 % MC, the values at 9.3 % MC were 13.5 and 6.9 times higher in earlywood, and 6.4 and 6.6 times higher in latewood, respectively. Statistical analysis confirmed significant differences between oven-dry and high-MC specimens, but no significant difference was observed in earlywood coefficients between 4.9 % and 9.3 % MC. Thermal expansion in wood was attributed to increased molecular vibrations and expanded intermolecular spacing within the cell wall as a result of temperature elevation (Goli et al. 2019). Meanwhile, a low MFA resulted in high cell wall stiffness and thereby restricting its thermal expansion. As MC increased, the plasticization effect weakened intermolecular constraints and enhanced segmental mobility, thereby leading to higher thermal expansion coefficients.

Figure 5:

2D distribution across the growth ring (a₁–c₂) in one representative specimen and the relationship between swelling strain and relative position in the growth ring (d₁–f₂) averaged from three specimens.

Figure 6:

The (a) radial and (b) tangential thermal expansion coefficients of earlywood and latewood at 0.6, 4.9 and 9.3 % MC. The different capital (latewood) or small (earlywood) letter indicates statistical difference at p < 0.05.

3.3 Hygroscopic swelling

Taking specimen with 0.6 % MC at 25 °C as a reference, the radial ( ε y y H ) and tangential ( ε x x H ) hygroscopic swelling with 4.9 % and 9.3 % MC are shown in the first column of Figure 5. At all MC levels, latewood exhibited greater radial swelling than earlywood. The amorphous matrix, particularly hydrophilic hemicelluloses, dominated moisture adsorption and associated swelling (Fredriksson et al. 2023). Consequently, the greater adsorption sites from more cell wall substances and higher hemicelluloses content in latewood provided the chemical basis for more pronounced moisture-induced transverse swelling. Moreover, radial ( β rad MC i ) and tangential ( β tan MC i ) swelling coefficients of earlywood and latewood were determined at 25, 35 and 45 °C across different MC ranges (Figure 7). Statistical analysis showed no significant differences in β rad MC i and β tan MC i across either different MC ranges or at different temperatures. However, both thermal coefficients and swelling coefficients of the latewood were larger than those of the earlywood, suggesting a greater sensitivity of latewood to hygro-thermal conditions. In general, hygroscopic swelling was 10–20 times larger than thermal expansion. Notably, observation of 2D swelling plots and corresponding thermal expansion/swelling coefficients of specimens revealed that total swelling under hygro-thermal conditions exceeded the sum of individual thermal expansion and hygroscopic swelling, indicating an additional hygro-thermal coupling effect. Furthermore, thermal expansion coefficients of the latewood showed significant differences across all three MC levels. In contrast, there was no significant temperature dependence of swelling coefficients, indicating that moisture might be the dominant factor in hygro-thermal coupled swelling (Salmen 1982; Schuerch 1964; Wolcott and Shutler 2003).

Figure 7:

The (a) radial and (b) tangential swelling coefficients of earlywood and latewood at 25 °C, 35 °C and 45 °C. The different capital (latewood) or small (earlywood) letter indicates statistical difference at p < 0.05.

3.4 Hygro-thermal coupled swelling

The contribution of hygro-thermal coupling to swelling was quantified in earlywood and latewood across different temperatures and MCs (Figure 8). Using ε _yy of specimens with 0.6 % MC at 25 °C as a reference, hygroscopic swelling, thermal expansion and hygro-thermal coupled swelling were isolated. The hygro-thermal coupled swelling increased with both temperature and MC. In latewood with 9.3 % MC at 45 °C, ε _yy consisted of 4.7 % ε y y T , 72.6 % ε y y H and 22.7 % Δε _yy . This indicated that a simple summation of thermal expansion and hygroscopic swelling underestimated actual swelling, which highlighted the importance of considering coupling effects when evaluating wood swelling at hygro-thermal conditions.

Figure 8:

Contributions of hygroscopic swelling, thermal expansion and hygro-thermal coupled swelling in (a, b) earlywood and (c, d) latewood at a series of hygro-thermal conditions.

Figure 9:

2D visualization of swelling strain of (a–d) EW and (e–h) LW at a series of hygro-thermal conditions. (a and e) Specimens with 4.9 % MC at 35 °C; (b and f) specimens with 4.9 % MC at 45 °C; (c and g) specimens with 9.3 % MC at 35 °C; (d and h) specimens with 9.3 % MC at 45 °C.

The hygro-thermal coupling effect was attributed to plasticization of the amorphous matrix, with hemicelluloses playing a dominant role. Bound water weakened intermolecular hydrogen bonding and promoted segmental mobility of amorphous polysaccharide chains (Engelund et al. 2013; Jakes et al. 2019). This plasticization lowered the temperature associated with hemicelluloses relaxations/softening. As a result, heating within the range of 25–45 °C could shift the matrix closer to the relaxation regime (Kelley et al. 1987; Naidjonoka et al. 2020). Once the matrix entered this mobility-enhanced state, chain segments could rearrange into more efficiently packed conformations and reduce free volume. Under these conditions, bound water was more effectively converted into transverse deformation, yielding a value of Δε beyond the linear superposition of thermal expansion and hygroscopic swelling. Tissue-dependent contrasts further supported this matrix-dominated interpretation. Latewood exhibited greater Δε _yy but similar Δε _xx compared to earlywood. At 9.3 % MC at 45 °C, the value of Δε _yy of latewood was nearly two folds higher than that of earlywood. This behavior was consistent with the higher proportion of cell wall and greater hemicelluloses content in latewood, which increased the amount of hydrophilic amorphous that underwent moisture-assisted relaxation.

Transverse visualization of full-field swelling revealed distinct patterns that earlywood exhibited a “peanut shell” pattern, indicating strong anisotropy, whereas latewood displayed a quasi-isotropic, near-circular pattern, which was consistent with previous studies (Derome et al. 2011; Rafsanjani Abbasi 2013; Sun et al. 2024; Wang et al. 2025) (Figure 9). The lower MFA in latewood implied higher stiffness of the cellulose crystallization region and reduced off-axis compliance. Consequently, expansion of the amorphous matrix was more uniformly constrained, leading to quasi-isotropic transverse swelling, whereas earlywood with higher MFA and thinner walls deformed more anisotropically. Critically, through 2D visualization and in-plane analysis, the contribution of hygro-thermal coupled swelling at varying hygro-thermal conditions and off-axis angles was quantified. For instance, hygro-thermal coupled swelling accounted for 22.7 % of the radial swelling of latewood with 9.3 % MC at 45 °C. From an application perspective, separating ε _T , ε _H and Δε under controlled MCs helped avoid underestimation of transverse deformation when temperature and humidity varied simultaneously in service. The growth-ring-resolved strain fields could support more accurate assessment of dimensional-stability in wood and related products used in structural and interior applications exposed to daily and seasonal hygro-thermal cycles. It should be noted that the relative contributions of temperature and moisture to hygro-thermal coupled swelling remain unresolved. Future studies should isolate temperature-driven and moisture-driven swelling mechanisms to advance predictive models for optimizing dimensional stability of wood under hygro-thermal conditions.