In situ observation of shrinking and swelling of normal and compression Chinese fir wood at the tissue, cell and cell wall level

Normal and compression wood of Chinese fir were selected from plantation trees grown in Zhejiang, China. Samples were obtained from air-dried logs. Samples with dimensions of 6 × 0.8 × 0.5 mm³ (radial × tangential × longitudinal) were cut for the shrinking/swelling tests. Within the cross section of the samples, a region containing the whole transition from early- to latewood within the 11th annual ring was present (Fig. 1). The surface of the samples was prepared by removing 1–2-µm thin slices with a cryomicrotome HM 560 (Microm, Germany). Three normal wood and three compression wood samples were prepared. All the samples were soaked in distilled water for 7 days to achieve a water-saturated state. In addition, 20-μm-thick sections (cross, radial and tangential) and 150-μm-thick tangential sections were prepared for anatomical characterization and MFA measurement. After MFA measurement, the sections were cut into chips (1 × 0.8 × 0.5 mm³, radial × tangential × longitudinal) for desorption and absorption isothermals.

The MFAs were determined by a wide-angle X-ray scattering experiment in a Bruker nanostar device. The 150-μm-thick tangential sections were exposed to Cu Kα X-ray radiation with a wavelength of 1.54 Angstöm. The distance between the sample and the Vantec 2000 detector was ~ 5 mm. The recorded 2D diffraction pattern was integrated, and the cellulose MFA was evaluated by the azimuthally scattering intensity distribution of the cellulose 200 reflection (Lichtenegger et al. 1997). Three replicates for each location were performed.

The microscopic structure of the samples was studied with a digital microscope (Keyence VHX-S550E, Keyence Company, Japan) equipped with a universal objective (VH-Z100UR) and operated in the transmission mode.

Desorption and absorption isothermal experiments were conducted on ca. 30 mg water-saturated wood chips by means of a dynamic vapor sorption apparatus (DVS Advantage 2, Surface Measurement Systems Ltd., UK). The temperature was set at 25 °C. The preprogrammed sequence of RH was identical to the shrinking and swelling experiment (95 → 90 → 75 → 50 → 25 → 0 → 25 → 50 → 75 → 90 → 95%) in the section Desorption and absorption experiment. At each RH level, the equilibration time was dependent on the weight change of the specimen. Once the weight change ratio was less than 0.0002%/min, the RH changed to the next humidity step. Three replications were performed for normal and compression wood.

Naturally, cell wall proportions increased from early- to latewood. In normal earlywood, the cell wall ratio was as low as 10.6% and increased to 56.5% in latewood. Compression wood, which is known for its larger density (Timell 1986), showed smaller differences between early- and latewood with a cell wall proportion of 24.8% in early- and 44.2% in latewood. The ESEM micrographs furthermore show numerous intercellular spaces (Fig. 1 and 4d) and hint that the S3 is at least partly missing in the compression wood samples (Fig. 1 and 4). Light micrographs showing the helical cavities (Fig. 4b and 4e) confirm this finding. The helical cavities together with checks point toward a high MFA (Barnett and Bonham 2004), which has been confirmed by the analysis of wide-angle X-ray diffraction data showing a MFA of ~ 40° for compression wood and a MFA of ~ 11° for normal wood. Compared with normal wood, compression wood possessed more tracheids and ray cells filled with an orange substance (Fig. 4e) which we assume are tannins due to their similar coloring as found in other tannin cells (Angyalossy et al. 2016).

Figure 5 shows the change in cross-sectional area change of single cells during desorption and absorption. It needs to be mentioned here that the low pressure inside the ESEM might affect area changes compared to experiments performed under ambient pressure conditions. However, a comparison with literature data is difficult since differences in structure have an immediate effect on swelling and shrinkage (Quirk 1984). Despite this uncertainty, comparisons between the studied tissue types are feasible and it is expected that observed trends are also valid under ambient conditions. The total area of the cells shrank as RH decreased and swelled as RH increased. When RH changes, water molecules are desorbed from or adsorbed onto the cell wall surface. Due to its large proportion in the cell wall, the S2 layer has the largest influence on the shrinkage and swelling behavior of wood (Siau 2012). Expectedly, the area changes of cell walls and cells showed similar patterns. The total area change of a cell is the combined change of cell wall and lumen. During the absorption process, the lumen, cell and cell wall swelled with increasing RH. Surprisingly, the lumens of all cell types initially expanded upon drying when RH decreased from 95 to 90% or even to 75% before they started to shrink. Normal latewood showed the most obvious lumen expansion, which reached as much as 1.57%. The observed lumen expansions seem not to be a unique behavior of Chinese fir. Dimensional changes of lumens upon swelling have been a matter of debate already in 1933. Beiser (1933) found that lumens of both early- and latewood expanded upon swelling and discussed in his paper that this was in contrast to the findings of Mörath (1932), who was reported to suggest that lumen sizes decrease upon swelling. This result agrees well with the present observations that lumen areas increase during absorption. At that time, changes of lumen areas upon drying were not studied and in 1942 Stamm and Loughborough suggested that lumen sizes stay constant during moisture changes (Stamm and Loughborough 1942). Later, Quirk (1984) observed expansion of Douglas fir earlywood lumen, while latewood lumens reduced their size upon drying. More recent experiments showed increasing lumen sizes upon drying for Caribbean pine (Ma and Rudolph 2006) and the opposite behavior in Douglas fir early- and latewood (Murata and Masuda 2006). These different findings highlight that it is not surprising to read in recent reviews that dimensional changes of lumens are still a topic of controversy (Arzola-Villegas et al. 2019). What the present results now clearly show are differences in lumen expansion and shrinkage during absorption and desorption, with consistent trends but varying magnitudes. During the initial phase of the desorption process, the inner side (lumen side) of the cell shrinks more easily than the outer cell wall, because the lignified middle lamella is assumed to donate resistance to lateral movement of cell wall (Quirk 1984). Furthermore, it is speculated that drying-induced increase in cell wall stiffness might play an additional role. The size changes of cells are possibly a trade-off of the stretch and restrain among cell wall and lumen, also among different layers in the secondary wall, and the lumen expansion of Chinese fir may contribute to the reduction of the drying stress.

Among the different cell types (normal or compression wood, latewood, transition wood or earlywood), their area changes showed similar trends but different values which depended particularly on the cell wall thickness. For both normal and compression wood, the shrinking (from 95 to 0% RH) and swelling (from 0 to 95% RH) ratios of cells decreased from latewood to transition wood and earlywood, due to decreasing cell wall proportions. Regardless of the sampling location, the area changes of cell walls were always larger than those of cells in both desorption and absorption processes. In the RH range from 50 to 90%, the area changes of cell and cell wall behaved quasi-linear in both desorption and absorption processes, which was also reported in a previous study (Murata and Masuda 2006).

In Fig. 6, the relative areas of normal and compression latewood are shown and were compared among tissue, cell and cell wall levels. During the desorption process, the relative area change of tissue is between cell and cell wall for all RH condition. At 0% RH, RA was 0.876, 0.881 and 0.875 for tissue, cell and cell wall level in normal latewood and 0.915, 0.922 and 0.908 for tissue, cell and cell wall level in compression latewood. These results reflect the interdependencies between cell wall, cell and tissue. Wood is a sophisticated system, which has the abilities to fine-tune its structure at various scales (Fratzl and Weinkamer 2007). At the cell level, a cell can be seen as the compound of cell wall and lumen and its size change is the sum of lumen and multi-layered cell wall changes. Although the role of the lumen as a hollow space is often restricted to density-related aspects, the relative humidity in this empty space is directly connected to cell wall swelling and shrinkage as the observed dimensional changes show. Of course, only the cross section of a bulk sample could be investigated, but due to the 12-h long equilibration times it is assumed that lumens of closed cells would have also reached the same relative humidity. At the tissue level, the tracheids are attached to each other by middle lamella. As the MC changes, the middle lamellae are the connection that allows simultaneous deformation of adjacent cells (at least in small samples or sample areas). However, middle lamellae also contribute to the shrinking or swelling, and according to Nečesaný (1966), the hygroscopic deformations of highly lignified middle lamellae are larger than those of cells. This could explain why dimensional changes at the tissue level are larger than on the cell level.

Figure 6 shows that the area change in compression wood was less than in normal wood at all the three levels. This is not surprising and can be explained by the higher MFA (40° in compression wood vs. 10° in earlywood) which causes reduced swelling and shrinkage in the transverse direction and increased swelling in the longitudinal direction (Burgert et al. 2007). When comparing RA of the desorption and the subsequent absorption, differences at the same RH condition have been observed. This swelling hysteresis is indicated by shaded areas in Fig. 6. The swelling hysteresis is associated with the sorption hysteresis, which will be described in detail in the section Swelling hysteresis.

A comparison of early- and latewood swelling is shown in Fig. 7. The trend that RA_cell changes less upon humidity changes than RA_tissue and RA_{cell wall} was observed for both desorption and absorption processes. During the desorption process, the difference between RA_cell and RA_tissue was larger than that during absorption, especially at 25, 50 or 75% RH. When RH was 50%, the value “RA_cell—RA_tissue” was 0.025 and 0.010, respectively, in desorption and absorption of the normal latewood. Besides, RA changes at high humidity changes are more pronounced during desorption than absorption. It is speculated that this is caused by a simultaneous increase in cell wall and lumen area during absorption, while the initial phase of the desorption process is characterized by a lumen expansion and a more pronounced cell wall shrinkage (Fig. 5).

Naturally, the area changes of the earlywood tissue, cell or cell wall were less than in latewood (Fig. 7), due to the smaller cell wall proportion. For the tissue level, both shrinking (95 → 0% RH) and swelling (0 → 95% RH) ratios of earlywood were more than 10% smaller than in latewood. In addition, the swelling hysteresis was also smaller.

During growth, trees manifest a wide variability of adaptable parameters of cell wall organization (MFA, cell size and cell wall proportion) to optimize water transport and mechanical support (Burgert et al. 2007; Eder et al. 2008; Reiterer et al. 2001). These adaptable parameters affect the shrinking/swelling and mechanical properties of wood. To explain the roles of MFA and cell wall ratio, the dimensional changes (shrinking and swelling ratios) are presented as a function of MFA and cell wall ratio, respectively, in Fig. 8 and 9. In Fig. 8, it is obvious that the shrinking ratio of normal latewood is higher than normal earlywood, which can be related to the higher cell wall proportion in the cross section. The significantly higher MFA of compression wood is clearly reflected in smaller dimensional changes of the cell wall. Due to the observed swelling hysteresis (Fig. 6), swelling ratios were less obvious than shrinking ratios for all samples (Fig. 8).

While significant differences in tissue swelling and shrinkage have been found between normal early- and latewood (p < 0.05), no statistically significant differences existed between early- and latewood of compression wood. The reason for this is twofold: (i) The higher MFA of compression wood leads to smaller area changes, and the differences in cell wall ratios of early- and latewood are smaller than in normal wood. Additionally, anatomical features of compression wood could play a role. This could include the presence of helical cavities in the tracheid walls (Fig. 4b and 4e) absent in normal wood (Fig. 4a and 4c). In addition, voids existed in corners among almost round cross-sectional structure of tracheid (Fig. 4d).

Equilibrium MCs of normal and compression wood are shown in Fig. 10a. Independent of whether the equilibrium moisture content is measured in absorption or desorption, MC of the compression wood at a given humidity level was higher than the normal wood. The higher MC of the compression wood may be related to the greater deposition of hemicelluloses and lignin between microfibrils (Peng et al. 2019b; Plomion et al. 2001) and more tannin in tracheids (Fig. 4e). In this study, the sorption hysteresis was determined by the desorption isothermal (initiated from the water-saturated condition), and its value increased with increasing MC, which was different from the previous studies by using the scanning isothermal (Fredriksson and Thybring 2018; Hill et al. 2010; Xie et al. 2011). At given RHs (25, 50 and 75%), sorption hysteresis of the compression wood was higher than the normal wood.

Based on results of MC (Fig. 10a) and area change at the cell wall level (Fig. 6), the relations are displayed in Fig. 10b. For both normal and compression wood, the relation was linear below a MC of 15%. Similar findings were reported in previous studies (Ma and Rudolph 2006; Murata and Masuda 2006). Moreover, changes in relative area during desorption and subsequent absorption coincided at low MC ranges (less than 10%), because adsorbed water molecule connected with cell wall surface directly. With further increasing MC, “mobile water” in small voids of the cell wall (Zelinka et al. 2012) had less impact on shrinking/swelling of cell wall polymers (swelling hysteresis in Fig. 10b). The relation between “sorption hysteresis” and “swelling hysteresis” was further constructed (Fig. 10c). These two hystereses were positively correlated. As a function of sorption hysteresis, swelling hysteresis of normal wood was significantly higher than compression wood (Fig. 10c). This result points to an interesting interrelation between sorption and swelling hysteresis. The sorption hysteresis is considered to be related to the mechanics of shrinkage and swelling of the cell walls during changes in MC (Engelund et al. 2013; Hill et al. 2012). Commonly, more lignin content would cause more hysteresis in wood cell wall (Derome et al. 2018; Kulasinski et al. 2015b). Compression wood of Chinese fir has more lignin content than normal wood (Peng et al. 2019b), which is consistent with the sorption hysteresis magnitude observed in Fig. 10a and b.

Due to higher MFA values, values of shrinking and swelling of compression wood were lower compared to normal wood, which at least partially explain the smaller swelling hysteresis. In addition, the absorption or desorption process results in the expansion or contraction of microcapillaries in the viscoelastic wood cell wall. The helical cavities in compression wood give rise to free movements of the S2 layer to some extent and also influence the size changes of microcapillaries. It should be pointed out here that the in situ shrinking and swelling tests and the desorption isothermal tests were carried out at 6 and 25 °C, respectively. These temperature differences as well as the above-mentioned differences in pressure were not considered in the interpretation, and it cannot be excluded that additional effects arise from them. We also need to consider that the swelling hysteresis was evaluated only in the transverse 2D-plane, and the dimensional changes in the longitudinal direction were not considered, while the sorption experiments were performed on the whole 3D wood material. Burgert et al. (2007) revealed that normal and compression wood showed contrary dimensional change in the longitudinal direction. Hence, a comprehensive consideration of the swelling hysteresis, especially for compression wood, should also consider the deformation in both transverse and longitudinal directions. Nevertheless, the results presented allow the comparison of compression and normal wood swelling in the 2D transverse direction and the sorption hysteresis for the same material.