Mechanism of changes in the mechanical properties of wood due to water adsorption and desorption described by rheological considerations

2.1 Specimen preparation

The wood species used was hinoki (Chamaecyparis obtusa), specifically the heartwood just inside the sapwood. The wood was sourced from sections that were straight in the longitudinal direction and free of cross-grains. Figure 1(a) shows an illustration of the test specimens. The specimens measured 26 mm in the radial direction (R), 3 mm in the tangential direction (T), and 1 mm in the longitudinal direction (L). A total of 28 specimens were consecutively cut in the L direction. The specimens were alternately allocated for dynamic viscoelasticity measurements and moisture content measurements, 14 specimens each. The average annual ring width of the specimens was 0.73 mm, and their oven-dry density was 410 kg/m³.

Figure 1:

Description of humidity tests. (a) Schematic diagram of the sample attached to the forced-vibration dynamic mechanical analyzer. Humidity program over time for (b) dynamic viscosity, (c) adsorption, (d) and desorption experiments. T1: start of humidity change; T2: end of humidity change; T3, T4, and T5: 100, 300, and 600 min after T2, respectively, at constant humidity.

The specimens were extracted using methanol in a Soxhlet extractor for 9 h to remove wood extractives components, and then air-dried in a fume hood for 3 days. After extraction and drying, the specimens were saturated with distilled water at 20 °C for 60 min under alternating decompression at 0.1 MPa. To ensure uniform drying history, the specimens were boiled in distilled water for 1 h while in a saturated state, then cooled in the same beaker to 20 °C. After cooling, the specimens were conditioned in desiccators at 22, 42, 59, and 80 % relative humidity (RH) for three weeks at a constant temperature of 20 °C. The humidity levels were monitored using a resistance molecular membrane sensor (SHA-3151, T&D Corp., Matsumoto, Japan).

2.2 Humidity control system and dynamic viscoelasticity measurements

The humidity in the test chamber was controlled using a water vapor generator (HC9700, Netzsch Japan K.K., Kanagawa, Japan). Dry and water-saturated nitrogen gas were mixed to create an arbitrary humidity level, and the resulting moist gas was introduced into a 70 mL test chamber at a flow rate of 200 mL/min. A capacitive proximity sensor (TI-A, Toplas Engineering Co. Ltd., Tokyo, Japan) was used to measure the humidity within the chamber. To verify the accuracy of the humidity control, humidity changes were monitored without the specimens, with the change rate set between 0.1 and 1.0 % RH/min to ensure that the measured humidity matched the desired program settings.

Dynamic viscoelasticity measurements were performed using a forced vibration dynamic mechanical analysis system (DMA 242E Artemis, Netzsch Japan K.K., Kanagawa, Japan). Measurements were performed at sine wave frequencies of 1, 5, and 10 Hz, with a forced strain of ±5 μm, a span of 10 mm, and a constant chamber temperature of 30 °C (Figure 1a). Before measurement, the stabilization time for the specimens was determined. It was found that after 3 h, the temperature, humidity, and specimen dimensions were stabilized. Thus, a holding time of 6 h was used to ensure stability before measurement.

2.3 Humidity change programs

The humidity change program in Figure 1b was used to study the temporal variations in dynamic viscoelasticity after reaching each humidity level. The programs started with a specific RH value 20, 40, 60, and 80 %, which were changed to 20, 40, 60 %, and 80 % (excluding the initial value). For example, the initial 40 % state was decreased to 20 % and increased to 60 and 80 % RH. Figure 1c and d show additional humidity change programs designed to examine the effects of humidity change history on dynamic viscoelasticity. In all three programs, the humidity-change steps were performed over a standard 60-min transition time. The humidity changes over time in the test chamber, and the dimensional changes (ΔL) of the specimens were measured. Here, ΔL represents the percentage dimensional change relative to the initial one between the chucks at T1, with positive values indicating expansion and negative values indicating contraction.

Figure 2 compares the RH changes in the test chamber, as controlled by the program in Figure 1b, with the ΔL of the specimens measured in the R direction. These results suggest that water molecules were adsorbed or desorbed by the wood as the humidity in the chamber increased or decreased.

Figure 2:

Changes in (a) humidity and (b) sample dimensions in the R direction over time during the experiments at each humidity. Filled data points & solid line: adsorption; open data points & dotted line: desorption.

2.4 Moisture content of the specimens

It is hard to measure the weight change of a specimen in the chamber during dynamic viscoelasticity measurements. Therefore, to monitor the moisture content of the specimens during humidity conditioning, the specimens were quickly removed from the chamber, and their weight was measured in a chamber with a humidity change program similar to that used for dynamic viscoelasticity measurements. The oven-dry weight was determined after drying the specimens for 3 days at 30 °C and 0.1 MPa in a reduced-pressure dryer containing phosphorus pentoxide. This weight and those of the wet samples were used to calculate the moisture content.

The average moisture contents of the samples (n = 3) in the test chamber immediately before the humidity change (at T1) were 5.7 ± 0.06 % (20 %RH), 7.5 ± 0.04 % (40 %RH), 11 ± 0.10 % (60 %RH), and 15 ± 0.09 % (80 %RH). The moisture content immediately after the humidity change (T2) was 7.4 % (40 %RH), 10.9 % (60 %RH), and 14 % (80 %RH) during humidification (n = 1–3), and 5.8 % (20 %RH), 7.6 % (40 %RH), and 11 % (60 %RH) during dehumidification (n = 1–3). Comparing the moisture content at T2 and T5 revealed minimal fluctuations, suggesting that the specimens maintained stable moisture levels. Additionally, the weights of the specimens used for dynamic viscoelasticity and moisture content measurements at T5 were comparable, indicating consistent equilibrium moisture content across tests.

3.1 Dynamic viscoelasticity of wood conditioned for three weeks

To gain insights into the changes in mechanical properties related to humidity fluctuations, particularly from the perspective of energy barriers, the mechanical properties at various humidity levels were first examined under a constant temperature of 30 °C. Measurements of the storage modulus (E′), loss modulus (E″), and loss tangent (tanδ) were taken at T1 (as defined in Figure 1b) for specimens conditioned at each humidity level for three weeks. These results (Figure 3), henceforth referred to as the control values, represent the equilibrium values of the specimens conditioned for the longest duration in these experiments.

Figure 3:

Humidity dependence of tanδ, E′, and E″ at each tested frequency at 30 °C for wood samples moisture-conditioned over 3 weeks. These results were used as control values.

The tanδ, E′, and E″ values at 1 Hz presented in Figure 3 were compared with those reported by Becker and Noack (1968). The tanδ value rose when the humidity changed from 20 to 40 %RH and from 60 to 80 %RH, with minimal difference between 40 and 60 %RH. Becker and Noack’s results similarly demonstrated a higher logarithmic damping ratio at a moisture content of 12 % compared to 8 % (From the moisture content measurements in this study, 12 % moisture content corresponds to 60–80 %RH in this study, and 8 % moisture content corresponds to 40–60 %RH in this study). As illustrated in Figure 3, the E′ reduced as RH elevated, with the most significant reduction occurring in the 60–80 %RH range. This pattern is consistent with Becker and Noack’s dynamic torsional modulus measurements and the established relationship between elastic modulus and moisture content (Kajita et al. 1961). The E″ curve shows a convex peak as a function of humidity, with the maximum value at 40 %RH. As the frequency increased, the slope of tanδ became steeper, and the E″ peak shifted to higher humidity levels.

3.2 Comparison of the dynamic viscoelastic properties of wood during humidity changes and wood conditioned for three weeks (control)

Figure 3 illustrates the relationship between humidity and equilibrium conditions of tanδ, E′, and E″, which were used as the standard values of the experimental system. The following discussion focuses on the behavior of tanδ, E′, and E″ during the adsorption and desorption processes relative to the control values. Figures 4 and 5 compare the dynamic viscoelasticity measured during the adsorption and desorption processes, from T1 to T2 in Figure 1b, with the values for control specimens conditioned over a long period. For the adsorption process, the results are presented as relative values normalized to the value at 20 %RH, whereas for the desorption process, the results were normalized to the value at 80 %RH.

Figure 4:

Comparison of tanδ, relative E′, and relative E″ values of wood measured under adsorption conditions with the control curves. Note: Relative values are based on values at 20 % RH.

Figure 5:

Comparison of tanδ, relative E′, and relative E″ values of wood measured under desorption conditions with the control curves. Note: Relative values are based on values at 80 % RH.

During the adsorption process (Figure 4), relative E′ was higher than the control for humidity values above 60 %RH at all frequencies. However, the differences in tanδ and relative E″ between the control and the test specimens were dependent on both humidity and frequency. In the humidity range of 20–40 %RH, tanδ and relative E″ were equal to or slightly higher than the control values at all frequencies. In contrast, at humidities above 40 %RH, the relative E″ values were significantly different from the control at all frequencies. Tanδ also exhibited notable differences from the control at 1 Hz in the humidity range above 40 %RH; however, at 5 Hz and 10 Hz, the values were nearly identical to the control.

During the desorption process (Figure 5), the tanδ values were greater than the control values at all frequencies, while the relative E′ values were lower than the control values at all frequencies. In contrast, relative E″ fluctuated with humidity, exhibiting values that were equal to or slightly higher than those of the control.

3.3 Rate of stabilization over time at each humidity after adsorption and desorption

In order to further analyse the results shown in Figures 4 and 5, the temporal stabilization of the molecular state of the wood must be considered, starting from immediately after adsorption and desorption in the horny layer. In previous research, tanδ/tanδ was used to show the extent to which an arbitrary unstable state is distant from a stable state (Miyoshi et al. 2020). Figure 6 compares the extent to which tanδ at different times (T2, T3, T4) approaches tanδ at T5, as calculated by the following formula:

Figure 6:

Degree of stabilization at various humidity levels after humidity change. Note: tanδ _i /tanδ_T5 (i = T2, T3, T4) percentages are based on T5 (600 min after T2), indicating how close the values are to that measured at the stable T5 state after changes in humidity.

When this value is close to 100, the difference between the numerator and denominator is small, indicating a high degree of similarity to tanδ at time T5. In addition, tanδ_i/tanδ_T5 represents the ratio of tanδ at each humidity to tanδ_T5 (the loss tangent when the wood itself is relatively stable), which is close to the net molecular momentum at each time point. Therefore, the decrease in tanδ _i /tanδ_T5 can reflect the behavior of the unstable microstructure, which stabilizes over time following adsorption and desorption due to the molecular mobility of the wood at that humidity and time.

Figure 6 shows the decline in tanδ _i /tanδ_T5 over time following adsorption. At 1 Hz, the ratio decreased progressively from T2 to T4 for all humidity levels (40–80 %RH). The most significant decrease occurred in the 20 %RH → 80 %RH transition, followed by the 20 %RH → 60 %RH and 20 %RH → 40 %RH transitions. When comparing the decrease in tanδ _i /tanδ_T5 at 20 %RH → 80 %RH across different frequencies, the greatest decline was observed at 1 Hz, while at 5 and 10 Hz, the ratio remained only slightly below 100 %. Similarly, for the 20 %RH → 60 %RH and 20 %RH → 40 %RH transitions, a decreasing trend was observed at 1 Hz, but the values at 5 and 10 Hz remained slightly higher than 100 %. In contrast, the decrease in tanδ _i /tanδ_T5 over time following desorption exhibited a different trend to that observed for the adsorption process. In all cases, a decline was observed from T2 to T4, with a greater decrease as the range of humidity change increased.

Furthermore, the changes in tanδ _i /tanδ_T5 for different RH values but a fixed change in humidity of 20 %RH were examined (Figure 7). The largest decreases in tanδ _i /tanδ_T5 from T2 to T4 were observed at 1 Hz for the humidity changes of 40 % ↔ 60 %RH, 60 % ↔ 80 %RH, and 20 % ↔ 40 %RH, respectively. At 5 and 10 Hz, the decreasing trend followed the order of 60 % ↔ 80 %RH, 40 % ↔ 60 %RH, and 20 % ↔ 40 %RH, respectively. Additionally, comparisons were made between transitions with the same humidity change range and final humidity levels, such as 20 % → 40 %RH and 60 % → 40 %RH, as well as 40 % → 60 %RH and 80 % → 60 %RH. At 1 Hz, tanδ _i /tanδ_T5 followed the order T2 → T3 → T4.

Figure 7:

Change of tanδ _i /tanδ_T5 (i = T2, T3, T4) over time after adsorption/desorption in each humidity range.

3.4 Transition in mechanical properties around the medium humidity (30–60 %RH)

Section 3.1 described the observed peak in E″ around the medium humidity. This peak may represent a turning point that significantly alters the mechanical properties of wood, suggesting that molecular mobility may differ between the low and high humidity ranges, with the medium humidity as the threshold value. To further clarify the behavior of the wood as it transitions through this point during the adsorption and desorption processes, experiments were conducted in which humidity was gradually increased and decreased (Figure 1c) or decreased and increased (Figure 1d). This humidity change program passes through each of the humidity levels of 20, 40, 60, and 80 %RH in a given molecular state and assumes that the molecular state at that time is recorded. For example, in the case of a humidity change program of 20 %RH → 40 %RH → 20 %RH → 60 %RH, molecular motion is activated by water adsorption during the transition from 20 to 40 %RH, and the molecular state during the relaxation process at 40 %RH is “frozen” during desorption from 40 to 20 %RH. It can be predicted that the molecular state at 20 %RH will be reached while maintaining the accumulated unstable state during the adsorption/desorption process of 20 %RH to 40 %RH to 20 %RH and that the molecular motion will be reactivated during the transition from 20 to 60 %RH.

To demonstrate the tracking performance of this humidity change program, Figure 8 shows the changes over time in the RH of the chamber and the ΔL of the test piece. As the specimen expands and contracts in response to the ambient humidity, it is judged to be tracking with little effect from diffusion.

Figure 8:

Time dependence of humidity and dimensional change (ΔL) during the humidity history experiments under adsorption/desorption conditions. Lines and colors follow Figure 2. “Passing humidity” means a process where the humidity increases up to a specific level and then returns to the original value.

Figures 9 and 10 compare the dynamic viscoelasticity of wood during stepwise adsorption and desorption with the control values. Relative values are shown based on 20 %RH for the adsorption process (Figure 9) and on 80 %RH for the desorption process (Figure 10).

Figure 9:

Influence of adsorption humidity range on dynamic viscoelasticity of destabilized wood. Lines and colors follow Figure 2; cross follows Figure 3.

Figure 10:

Influence of desorption humidity range on dynamic viscoelasticity of destabilized wood. Lines and colors follow Figure 2; cross follows Figure 3.

During the adsorption process (Figure 9), tanδ, relative E′, and relative E″ did not significantly deviate from the control values in the 20 → 40 → 20 %RH cycle at all frequencies. Additionally, relative E′ only showed minor deviations from the control values during adsorption in the humidity range above 60 %RH at all frequencies. However, the humidity at which relative E″ deviated from the control varied was frequency dependent. At 1 Hz, relative E″ deviated from the control when the RH exceeded 40 % during the 20 % → 60 %RH transition, after which the values oscillated above the control. At 5 and 10 Hz, relative E″ moved away from the control when the RH exceeded 60 % during the 20 % → 80 %RH transition and continued fluctuating above the control value.

During the desorption process (Figure 10), the relative E′ at all frequencies deviated from the control immediately after desorption and fluctuated between smaller values. Relative E′ also showed larger values during adsorption than desorption, eventually approaching the control. However, the behavior of relative E″ depended on the frequency and humidity. At 1 Hz, relative E″ did not deviate during the 80 % → 60 %RH transition but showed deviations during re-adsorption from 60 to 80 %RH. During desorption from 80 to 40 %RH, the behavior of relative E″ between 80 and 60 %RH showed a distinct difference compared to the re-adsorption cycle (60–80 %RH), where the values were consistently larger than the control and did not overlap with any points obtained during the humidity change history. At 5 and 10 Hz, relative E″ remained close to the control values, but during desorption from 80 %RH to 40 %RH, overlapping behavior with the 80 % → 60 %RH transition was observed, as seen at 1 Hz.