Stabilization of hygro-deformation and interaction between earlywood and latewood in Chinese fir

2.1 Materials

Three samples of air-dried Chinese fir wood (Cunninghamia lanceolata [Lamb.] Hook.) were obtained with dimensions of 5 mm × 3 mm × 0.5 mm (radial × tangential × longitudinal, R × T × L), each containing the intact 17th growth ring. These samples were water-saturated and polished with a cryomicrotome HM 560 (Microm, Germany) removing 2-μm thin transverse slices, and were then fully equilibrated over distilled water for over 4 weeks. Following sorption and hygro-deformation experiments, the samples were mechanically divided into isolated EW (∼4 × 3 × 0.5 mm³, R × T × L) and LW (∼0.5 × 3 × 0.5 mm³, R × T × L) regions. Figure 1 illustrates the detailed sampling procedure, a typical sample with the intact growth ring and the isolated EW and LW regions were shown in step (i) for hygro-deformation and cell wall proportion experiments (step ii). Additionally, sections measuring 3 × 0.01 × 10 mm³ (R × T × L) and 0.01 × 3 × 10 mm³ (R × T × L) were also cut from both EW and LW regions for step (iii) – MFA determination. The average air-dried density of the samples was approximately 370 kg/m³.

Figure 1:

Sampling procedure and three major characterizations: (i) shrinking and swelling, (ii) cell wall proportion, and (iii) microfibril angle.

2.2 Sorption and hygro-deformation experiment

Before conducting the hygro-deformation tests, the surface evenness of the samples was assessed using a digital microscope (VHX-7000, Keyence Company, Japan) to ensure they were free of warping, thereby avoiding any interference with subsequent in-plane deformation measurements. The desorption, adsorption and corresponding shrinking and swelling experiments followed the methodology outlined by Zhan et al. (2023). Successively, the intact growth ring samples and isolated LW and EW samples underwent the desorption and adsorption processes. Relative humidity (RH) levels of 97, 75, 58, 33, or 0 % were maintained using saturated salt solution (K₂SO₄, NaCl, MgCl₂ or NaBr) or P₂O₅ in a sealed container. An electronic analytical balance (Mettler Toledo ME204, ± 0.1 mg) was placed in the sealed container for measuring mass change, i.e. MC, of the wood sample. At each humidity level, MC values were recorded periodically at 0, 3, 6, 9, 12, 24, 36, 48, 60, 72, 96, 120 and 144 h. Cross-sectional morphology was imaged using the focal length adjusted digital microscope without removing the samples from the sealed container, thereby preventing any MC variation during observation (Figure 1). The microscope was equipped with a 12.22-megapixel complementary metal–oxide–semiconductor image sensor, providing a resolution of 6,144 × 4,608 pixels. Three samples each of the intact growth ring and isolated LW or EW samples were replicated to determine average MC and hygro-deformation values. A representative sample was selected to display in-plane strain distribution.

The hygro-deformation was assessed using DIC analysis (VIC-2D software, Correlated Solution Inc., United States), with the following steps (Garcia et al. 2020a, b; Zhan et al. 2023): (i) selecting the area of interest (AOI): approximately 15 mm² (2,567 × 3,679 pixels²) for the intact growth ring sample; (ii) dividing the AOI into evenly spaced virtual grids (subsets: 113), step size selection (28) and predefining the correlation criterion, interpolation, thresholding and post-processing to calculate the full-field strain; and (iii) extracting strain data from 2D contour plot through lines across the growth ring. The strains in the X (ε _TT) and Y (ε _RR) directions represented tangential and radial deformations, respectively. Shrinking or swelling strain was determined by comparing images of the equilibrated sample at 97 or 0 % RH as a reference, following the definition used for wood shrinking or swelling (Peng et al. 2012). Shrinking strain was treated as negative and swelling strain as positive to depict dimensional changes during moisture desorption and adsorption.

2.3 Hygro-deformation difference between isolated and intact LW/EW

The hygro-deformation differences were calculated as follows:

where Δε _RR and Δε _TT represent the radial and tangential hygro-deformation differences, respectively. The subscripts “gr” and “i” denote measurements taken from samples within the intact growth ring and after isolation, respectively. Specifically, the hygro-deformation differences were evaluated for both the desorption process from 97 % RH to 0 and the adsorption process from 0 to 97 % RH.

2.4 Cell wall proportion

After hygro-deformation experiments, the isolated EW and LW were placed in an environmental scanning electron microscope (FEI Quanta FEG 600, FEI Company, USA) for determination of cell wall proportion. The cross-section images of EW and LW were obtained at an acceleration voltage of 4 kV in a low-vacuum mode. The cell wall proportion was quantified by calculating the ratio of cell wall thickness to cell thickness for both radial and tangential directions separately (Figure 1, step ii). For each direction, measurements were taken from 20 cells in either EW or LW to ensure replicability.

2.5 MFA

The 10-μm-thick section was mounted on a glass slide with a drop of deionized water, and the coverslip was sealed with nail polish to prevent water evaporation. MFA determination employed polarized confocal Raman microscopy (Xplora HR Evolution, Horiba Jobin Yvon, Japan) following the methods by Gierlinger et al. (2009) and Zhang et al. (2023). The imaging employed a 100× oil immersion lens (NA 1.40) and a 532 nm laser with 100 % intensity. A 600 mm grating was used with a wavenumber range of 100–3,900 cm⁻¹ and an integration time of 20 s. Raman spectra were obtained by rotating a half-wave plate in 5° increments to vary laser polarization, capturing measurements at 10° interval across a full 360° polarization range to analyze Raman intensity as a function of polarization angle. MFAs on both the tangential and radials wall were evaluated with five replicates each in EW and LW.

2.6 Statistical analysis

The statistical software, SPSS version 17.0 (SPSS Inc, Chicago, IL, USA) was used for data analysis. Significant effects of sampling location (EW or LW) and RH level on hygro-deformation (difference) stress were analyzed by Duncan’s multiple comparison test (p = 0.05).

3.1 Hygro-deformation during desorption and adsorption

The hygro-deformation behavior of Chinese fir during desorption (Figures 2 and 3) and adsorption (Figures 4 and 5) is presented. Figures 2 and 4 display the full-field distributions of radial strains (ε _RR), while Figures 3 and 5 show tangential strains (ε _TT). In Figures 2b, 3b, 4b, and b5b, the evolution of strain over time is shown alongside the corresponding changes in MC. MC represents the average value for the intact growth ring sample, with the sampling locations for LW, EW and restrained EW indicated in the bottom-right subfigure of Figure 2a. Previous studies (Patera et al. 2018; Zhan et al. 2023) attributed the larger hygro-deformation of restrained EW at the growth ring interface to the mechanical restraint imposed by the previous LW on EW. Throughout both the desorption and adsorption processes, LW consistently exhibited greater dimensional changes compared to both restrained EW and unrestrained EW, regardless of humidity levels. Specifically, when comparing dimensions at 97 % RH to those at RH of 0 after desorption, the values of ε _RR and ε _TT for LW were −3.0 % and −4.2 %, respectively, while the corresponding values for EW were −1.1 % and −3.3 %. After fully adsorption from RH of 0 back to 97 %, ε _RR reached 2.82 % for LW and 1.04 % for EW, while ε _TT values were 3.64 % and 2.78 %, respectively.

Figure 2:

Changes in shrinking in radial direction (ε _RR) with stepwise RH decrements (97 % → 75 % → 58 % → 33 % → 0): full-field distribution of ε _RR (a) and its time-dependent value along with MC variation (b). The values of ε _RR shown in (b) were averaged from the sampling positions indicated in the bottom-right subfigure of (a). Full-field distributions of ε _RR in (a) after 144 h desorption were cited from Zhan et al. (2023).

Figure 3:

Changes in shrinking in tangential direction (ε _TT) with stepwise RH decrements (97 % → 75 % → 58 % → 33 % → 0): full-field distribution of ε _TT (a) and its time-dependent value along with MC variation (b). Sampling positions in (b), refer to Figure 2. Full-field distributions of ε _TT in (a) after 144 h desorption were cited from Zhan et al. (2023).

Figure 4:

Changes in swelling in radial direction (ε _RR) with stepwise RH increments (0 → 33 % → 58 % → 75 % → 97 %): full-field distribution of ε _RR (a) and its time-dependent value along with MC variation (b). Sampling positions and arrows in (b), refer to Figure 2. Full-field distributions of ε _RR in (a) after 144 h adsorption were cited from Zhan et al. (2023).

Figure 5:

Changes in swelling in tangential direction (ε _TT) with stepwise RH increments (0 → 33 % → 58 % → 75 % → 97 %): full-field distribution of ε _TT (a) and its time-dependent value along with MC variation (b). Sampling positions in (b), refer to Figure 2. Full-field distributions of ε _TT in (a) after 144 h adsorption were cited from Zhan et al. (2023).

Notably, these values were lower than those reported for Chinese fir in the literature (Cheng 1985; Jiang and Lu 2012; Yao et al. 2017), as well as for other coniferous species with comparable densities (Perré and Huber 2007; Thybring and Fredriksson 2023; Watanabe et al. 1998). The variation in hygro-deformation of wood was closely related to density and MFA, and also influenced by the content of lignin and hydrophilic extractives (Eder et al. 2020; Jankowska et al. 2017). In addition, the origin of hygro-deformation variability should also consider structural differences within the wood; even two trees of the same species can exhibit significant differences in EW shrinking (Almeida et al. 2014). In terms of anisotropy, LW show a lower ε _TT/ε _RR ratio (1.4) than EW (3.0) when RH decreased from 97 % to 0, indicating that EW was more anisotropic. Both ratios were higher than those reported in the literature, where LW is generally considered to be nearly isotropic (Patera et al. 2018; Perré and Huber 2007).

The dimensional changes after full stabilization as a function of difference of MC (ΔMC) are shown in Figure 6. It could be found that the dimensional changes exhibited a quasi-linear relationship with ΔMC, regardless of anatomical tissue type or direction. Notably, the radial deformation (ε _RR) (Figure 6a and c) differed significantly between LW and (restrained) EW, whereas the differences in tangential deformation (ε _TT) were relatively minor (Figure 6b and d). This reduction variation in ε _TT may be attributed to the dominant influence of LW, which constrained the tangential deformation of adjacent EW, resulting in a more uniform overall response across the growth ring (Ouyang et al. 2022).

Figure 6:

Relationships between strain ε _RR (a, c) and ε _TT (b, d) and difference of MC (ΔMC) during desorption (a, b) and adsorption (c, d) processes.

3.2 Stabilization of hygro-deformation and MC

Based on the results in Figures 2–5, it was observed that ε _RR and ε _TT, along with MC, changed rapidly at the start of the desorption or adsorption before stabilizing, expectedly. However, the stabilized periods differed between dimensional changes and MC. To precisely evaluate the stabilizations of dimension and MC, the variations of dimension and MC with logarithmic time were plotted in Figures 7–10. Taken desorption at 0 humidity as an example, MC tended to be constant after 48 h (indicated as the purple arrows, bottom subfigure in Figures 7g and 8g), whereas ε _RR and ε _TT stabilized after 108 h (black arrows in Figures 7g and 8g), similar results could be seen during adsorption (Figures 9 and 10). The desorption or adsorption process involved not only sorption equilibrium but also a time-dependent macromolecular relaxation of the wood polymers. As water molecules were absorbed, hemicelluloses and other amorphous polymers within the cell wall softened and transitioned toward a rubbery state. This softening facilitated stress relaxation, enabling internal polymer conformation, which contributed to a delayed dimensional response even when MC remained constant (Brémaud and Gril 2021; Hunt and Gril 1996; Uehara et al. 2025). Consequently, shrinking or swelling did not perfectly track changes in MC; instead, the internal polymer relaxation led to hysteresis and a time lag in dimensional adjustment.

Figure 7:

MC and ε _RR as a function of logarithmic desorption time (a, c, e, g), and corresponding evolutions of ε _RR versus MC (b, d, f, h). RH level: 75 % (a, b), 58 % (c, d), 33 % (e, f) and 0 (g, h). The purple and black arrows in (g) indicated the stabilized duration of MC and dimension, respectively.

Figure 8:

MC and ε _TT as a function of logarithmic desorption time (a, c, e, g), and corresponding evolutions of ε _TT versus MC (b, d, f, h). RH level: 75 % (a, b), 58 % (c, d), 33 % (e, f) and 0 (g, h). The purple and black arrows in (g), refer to Figure 7.

Figure 9:

MC and ε _RR as a function of logarithmic adsorption time (a, c, e, g), and corresponding evolutions of ε _RR versus MC (b, d, f, h). RH level: 33 % (a, b), 58 % (c, d), 75 % (e, f) and 97 % (g, h). The purple and black arrows in (g), refer to Figure 7.

Figure 10:

MC and ε _TT as a function of logarithmic adsorption time (a, c, e, g), and corresponding evolutions of ε _TT versus MC (b, d, f, h). RH level: 33 % (a, b), 58 % (c, d), 75 % (e, f) and 97 % (g, h). The purple and black arrows in (g), refer to Figure 7.

In this study, MC and strains were not monitored continuously, but rather measured at several specific time points. Therefore, the exact time required for stabilization among MC and strains could not be determined. Nevertheless, the available results allowed for a comparative analysis of the differences in the stabilization behavior between MC and strains. These delayed stabilization of ε _RR and ε _TT relative to MC contrasts with findings from the literature (Ma et al. 2010; Nopens et al. 2019; Ouyang et al. 2022). Ouyang et al. (2022) observed that the size of Catalpa bungei samples stabilized immediately once MC reached equilibrium, using real-time dynamic vapor sorption combined with a digital microscope. Besides, Ma et al. (2010) reported that the dimensional change of Picea sitchensis Carr. occurred more rapidly than MC change. However, the stabilization periods of dimension and MC reported in this study were not comparable with published literature due to variations in environmental conditions. Since maintaining constant humidity was challenging when temperature fluctuated, various methods were used to control RH, such as saturated salt solutions in this study, and humidity generators in the reported literature. Besides, this non-simultaneous behavior could be associated with variations in sample size and the characterization of length scales as well. The sample size used in this study (5 × 3 × 0.5 mm³, R × T × L) was smaller than those reported in the literature: 20 × 20 × 1.5 mm³ (R × T × L) (Nopens et al. 2019), and 20 × 20 × 4 mm³ (R × T × L) (Ma et al. 2010). When examined sample sizes across multiple growth rings, the observed length scale was typically restricted to the macro scale. Ouyang et al. (2022) reported using samples of 8 × 4 × 4 mm³ (R × T × L), but measured dimensional change across a series of cells from one vessel to another. According to Rafsanjani et al. (2014), the hygro-deformation of wood cell wall was substantially larger than the one of wood tissues with the growth ring and wood samples made of several growth rings, because the interactions between cell wall layers, cell wall and middle lamella, and between cells contributed to deformation at the cell wall and cellular scales (Zhan et al. 2023). Benefiting from the full-field hygro-deformation provided by DIC, the analysis in this study was able to focus on the dimensional changes within a few cells.

The relationships between strains and MC during desorption (right side of Figures 7 and 8) and adsorption (right side of Figures 9 and 10) were constructed. In these figures, the strain exhibited a trend of initially rapid and then slower change with decreasing (or increasing) MC. On the one hand, the desorption (or adsorption) of water molecules from (or onto) the polar groups of wood components directly caused changes in cell wall dimensions, corresponding to the initial rapid phase of strain variation with MC. Additionally, the hygro-deformation of wood was influenced by the state of moisture as well. During sorption process, moisture gradually returned to its equilibration state. This equilibration dependent on interaction with the chemical constituents of the wood cell wall (Nakano 1994). As moisture equilibrated, changes in the size and quantity of nanopores within the wood cell wall occurred (Shi and Avramidis 2019), potentially causing slight dimensional deformations. To assess the impact of moisture state on dimensional changes, it was recommended to record real-time variations in MC and dimensions using dynamic vapor sorption and high-resolution microscopy. Additionally, further investigations could be performed to understand how RH levels affect the stabilization of dimensions, as the relationship between RH and moisture presence remained unclear (Li and Ma 2022; Telkki et al. 2013).

3.3 Hygro-deformation difference between isolated and intact LW/EW

The full-field hygro-deformation was compared between the intact growth ring samples and isolated LW and EW samples. Figure 11 shows the comparison of the shrinking results. When the intact growth ring sample was divided as two parts as well as LW from the last growth ring was removed, the restraining effect disappeared, specifically leading to reductions in shrinking in EW. These reductions were characterized as Δε _RR (Figure 11c) and Δε _TT (Figure 11d). Besides, there were also reductions of ε _RR and ε _TT in transition wood (TW) from the isolated samples compared to that within the intact growth ring samples. However, no significant variation in ε _RR or ε _TT was found between isolated LW and LW within the intact growth ring.

Figure 11:

Comparison of shrinking results in intact growth ring samples and isolated EW and LW samples when RH changed from 97 % to 0: full-field distribution of ε _RR (a) and ε _TT (b), and corresponding relationships between shrinking strain and relative position in the growth ring (c and d). The results obtained from the intact growth ring samples were referenced from Zhan et al. (2023).

Figure 12 displays the swelling results. Similarly, obvious reductions of ε _RR (a) and ε _TT were observed in EW and TW in the isolated samples. These reductions demonstrated that LW promoted hygro-deformation in both adjacent EW and TW. LW had higher stiffness and more hygro-deformation than EW and TW. In hence, EW and TW were forced to deform with LW during sorption process, attributing to higher hygro-deformation of EW and TW within the intact growth ring. The value of Δε _TT was higher than Δε _RR, regardless of shrinking (Figure 11c and 11d) or swelling (Figure 12c and 12d). During desorption, the values of Δε _TT and Δε _RR were 0.46 and 0.40 %, respectively. The corresponding values were 0.58 and 0.71 % during adsorption. The variations in Δε _TT and Δε _RR were influenced by the interaction between EW and LW. This interaction originated from the structural variation. Particularly, MFA and cell wall thickness are key determinants of wood’s stiffness and hygro-deformation (Eder et al. 2020). Figure 13 provides a detailed overview of the variations in MFA and cell wall proportion in both radial and tangential walls between EW and LW. It was evident that EW exhibited higher MFAs and lower cell wall proportions compared to LW, irrespective of the wall orientation. The greater values of Δε _TT compared to Δε _RR could be explained by the more pronounced difference in MFA within the tangential wall. Specifically, LW had the lowest MFA in the tangential wall (7.6°), while its cell wall proportion (18.3 %) was lower than that in radial wall (22.2 %). It should be noted that the cell wall proportion in this study was lower than that reported in a previous study (Zhan et al. 2021) using the same species, as it was calculated based on a line ratio (Figure 1) rather than an area ratio.

Figure 12:

Comparison of swelling results in intact growth ring samples and isolated EW and LW samples when RH changed from 0 to 97 %: full-field distribution of ε _RR (a) and ε _TT (b), and corresponding relationships between swelling strain and relative position in the growth ring (c and d). The results obtained from the intact growth ring samples were referenced from Zhan et al. (2023).

Figure 13:

Variations of MFA (a) and cell wall proportion (b) of EW and LW in radial and tangential walls. Polar plots in (a) illustrated the intensity variations of the 1,096 cm⁻¹ Raman band as the polarization direction of the incident laser was rotated in 10° steps. Inset in (b) presented the SEM images of LW and EW. Scale bars: 10 μm.

Comparing the difference of strain in intact growth ring and isolated EW and LW is helpful to understand the interaction between EW and LW, and probably to assess the evolution of moisture variation induced stresses. When the interaction stress exceeds transverse tensile strength, the likelihood of cracking in the wood increases significantly (Qu et al. 2025). Future research could further evaluate the elastic and even viscoelastic stresses under varying hygro-thermal conditions which simulate practical drying environments.