Initial Desorption of Reaction Beech Wood Inicijalna desorpcija reakcijskog drva bukovine

• The research aimed to obtain empirical data for modeling the initial desorption in reaction wood from the cross-section of the green beech (Fagus sylvatica L.) log. Firstly, we analyzed the chemical composition, macro and microscopic structure of tension and opposite wood tissue. Then, the Equilibrium Moisture Content (EMC) was measured by the Dynamic Vapor Sorption method during the initial desorption. The used air param - eters were specific for the mild drying schedule of green beech timber (t = 20, 35, and 50 °C, Relative Humidity (RH) ranging from 95 to 0 %). Relationships between the EMC of reaction wood and drying parameters were modeled using the Response Surface Method (RSM). The tests revealed: different hygroscopic properties of tension and opposite wood, the dependence of EMC value on temperature, and differences between EMC values for initial (first) and second desorption. Moreover, it was confirmed that, during initial desorption, the EMCs of reaction wood are significantly higher than reference EMC data. The differences in the EMC value are up to 0.14 kg/kg (for air with RH above 90 %). The presented polynomial model of the initial desorption of reaction beech wood can improve drying schedules for beech sawn timber with a high amount of reaction tissue.


INTRODUCTION 1. UVOD
The equilibrium moisture content (EMC) depends on the wood species and the variation of its structure (Kollmann, 1936;Skaar, 1988).The most frequently used method of measuring the EMC is the procedure based on the Hailwood-Horrobin sorption model (1946).Simpson fitted this model to empirical EMC data provided by the "Forest Products Laboratory" (Simpson, 1973).These data are averaged values for desorption and adsorption and, by this simplification, have limited suitability for modeling a wood water relation in kiln-drying of timber (Wengert, 1976;Langrish and Walker, 1993;Salin, 2011;Glass et al., 2014;Redman et al., 2016).The EMC data is most often applied to relate air parameters and expected drying intensity during kiln-drying of timber.The relation is given by drying schedules for different wood species, usually using a Drying Gradient (DG) concept, which is defined as a ratio of the actual moisture content of timber and EMC values appropriate for drying parameters (Brunner, 1987).The EMC based on the initial desorption isotherms can significantly improve the accuracy of the drying gradient determination (Majka and Olek, 2013).Most published data on the hygroscopic properties of wood are the result of research on the phenomenon of the second desorption, occurring after drying the wood (Spalt, 1958;Weichert, 1963;Böhner, 1996;Ahmet et al., 1999;Jannot et al., 2006;Popper et al., 2009;Popper and Niemz, 2009;Jankowska, 2018).However, the second desorption process differs from the initial (first) desorption.The second desorption of beech wood gives lower EMC than the initial desorption, especially in higher air relative humidity (RH) (Barkas, 1936;Skaar, 1988).
The beech can form up to 25 % of the reaction tissue (Kúdela and Čunderlík, 2012).Reaction wood is a structure of wood tissue that takes the form of compression wood and tension wood.This type of reaction wood induces the desired displacement of the stem towards a more favorable position by tensile force (Côté, 1964;Scurfield G., 1973;Tulik and Jura-Morawiec, 2011;Felten and Sundberg, 2013;Groover, 2016).An indicator of reaction wood in a log is a pith eccentricity.The eccentricity is caused by wider annual increases on the tension side.On a microscopic scale, the reac-tion wood of beech is characterized by lower content of vessels with smaller diameter and length and significantly elongated tracheids with thickened walls.Changes in the S 2 and S 3 cell layers are a typical feature of the microscopic structure of reaction wood in beech (Côté et al., 1969;Wardrop and Davies, 1964).However, the essential factor influencing wood drying may be a non-lignified gelatinous layer (G-layer).Bound water diffusion in the G-layer causes an almost always higher in tension wood than in normal wood, despite similar density values.The fibers of reaction wood have a greater longitudinal shrinkage than normal wood (Scurfield and Wardrop, 1962;Tarmian et al., 2012).The reaction tissue of beech wood dries more slowly and has a higher final moisture content after drying than the opposite and normal tissue (Klement et al., 2019(Klement et al., , 2020)).The typical drying behavior of beech reaction wood is more evident during drying above Fiber Saturation Point (FSP) when liquid-free water is removed (Tarmian et al., 2009).Beech timber produced from logs with a high content of tension tissue shows a greater risk of developing defects during kilndrying (Tarmian and Perré, 2009).As far as the authors are aware, there are no published data describing the relationship between the EMC of reaction wood and air parameters in the range corresponding to the kiln-drying schedules of sawn beech timber.Therefore, the aim of the research was to provide empirical EMC data and use them to develop a model of the initial desorption in the reaction beech wood.

Materijal
The research material was primarily sawn in February 2020 from a green 105-year-old beech (Fagus sylvatica L.) log with a distinctly eccentric pith.The test tree was selected from a fresh mixed deciduous forest located in the Forest District Rzepedź (in the Subcarpathian province of south-east Poland, close to the border with Slovakia).The 100 mm disc (at breast height) was cut and uncontrolled changes in moisture content were prevented.Two sections of the research material were selected for laboratory tests, located on the cross-section of the trunk on opposite sides of the eccentric pith.The strip axis ran along the line marked by the highest difference in the width of annual increments (Figure 1).The test samples were prepared from two sections of the strip: A -tension wood and B -opposite wood.Both sections included 40 ± 1 and 80 ± 1 annual rings (the sections inner and outer cross-section zones).

Macro and microstructure 2.2. Makroskopska i mikroskopska struktura drva
The annual tree ring widths (s) were measured with an optical device with a computer image analyzer (BEPD-19, BIOtronik, Warsaw, Poland) with a measuring range of 470 mm and measurement uncertainty of 0.01 mm.According to the measurement results, the difference of about three times was found in the average width of annual rings between sections A and B. The average values of the annual ring widths in Section A (tension wood) are 1.40 ± 0.71 mm, and in Section B (opposite wood) 3.19 ± 1.30 mm.The width measurements of all annual rings are presented in Figure 1.
The microtome samples were photographed using a biological microscope with a computer image analyzer (B3 Professional, Motic, Hong-Kong, China).

Chemical composition of wood 2.3. Kemijski sastav drva
Firstly, each prepared wood sample was ground (0.5-1.0 mm fraction, mass ca.50 g) in a Fritsch Pulverisette 15 laboratory mill (Fritsch GmbH, Germany).The cellulose content was measured according to Seifert's method using a mixture of acetylacetone, 1.4-dioxane, and hydrochloric acid to isolate cellulose (Browning, 1966).According to the chlorite method, the holocellulose content was measured using NaClO 2 as a reagent (Browning, 1966).The pentosane content was measured using hydrochloric acid and phloroglucinol according to the TAPPI standard method T 223 cm-01.Acid-insoluble lignin was assessed according to T 222 om-06 standard TAPPI method.The content of extractives soluble in alcohol was measured according to TAPPI standard method T 204 cm-97.All tests were carried out with three replicates for each option of samples.

Sorption experiments 2.4. Eksperimenti sorpcije
Sorption experiments were carried out using a dynamic vapor sorption (DVS) apparatus (DVS Advantage 2 from Surface Measurement Systems, London, UK).The appropriate air RH levels were achieved by mixing dry and saturated air streams.The EMC values for air humidity (RH) in the range of 95 to 0 % were measured.It was assumed that the hygroscopic equilibrium was obtained at a given RH value when the mass change was less than 0.0005 % min -1 for at least 60 min.The procedure was repeated for each RH step and the EMC values were calculated.Samples for measuring the EMC variability of the tension and opposition wood were produced in two stages.In the first stage, four fragments were separated from Section A (tension wood) and B (opposite wood), each containing three annual rings, i.e., 39-41 and 79-81.Then, the prepared fragments with dimensions of 20 mm in the tangential (T) and longitudinal (L) direction were divided in the radial plane into final samples with a thickness of ca. 1 mm.The initial mass of each investigated sample was 12 ± 0.5 mg.
The sorption experiments consisted of air parameters, specific for mild kiln-drying beech sawn wood schledue, which saves the natural color.Three air temperature values were used, i.e. 20, 35, and 50 °C and five relative air humidity (RH) values, i.e. 95, 80, 65, 50 and 35 %.After the parameters were measured during the initial desorption, an additional sorption experiment was performed, consisting of water adsorption and second desorption.An additional experiment compared the EMC values with the available literature data.The list of all variants of air parameters in the sorption experiments is presented in Table 1.The Response Surface Methodology (RMS) was used to generalize the relationship between the EMC for initial desorption and the characteristic air parameters of the kiln-drying schedule (Box and Draper, 2007).The levels of independent variables used for the sorption experiments are presented in Table 2.
According to the following formulas, the independent variables were coded: x 1 = (t -35)/15, and x 2 = (RH -65)/15.The third-order polynomial equation approximated the results: (1 Where y is the predicted response (i.e.equilibrium moisture content for initial desorption), b 0 -b 9 are estimated coefficients.The fitting algorithm (Leven-berg-Marquardt approach) was used to estimate the coefficients of the response models of the initial desorption.Due to the possible linear dependence of the variables, backward stepwise regression was applied to exclude statistically insignificant model parameters (Chatterjee and Hadi, 2013).The experimental input data for RSM modeling are presented in Table 3.The results are supplemented by EMC values calculated from the Hailwood-Horrobin equation as applied by Simpson (1973) to the data from the Forest Products Laboratory (Wood Handbook, 2010).

Statistical analysis 2.6. Statistička analiza
The experimental data were analyzed using STA-TISTICA 13.3 software (TIBCO Software Inc., Palo Alto, CA, USA).A one-factor analysis of variance (ANOVA) was performed to determine significant differences between the components of the average content   The microstructure of the investigated samples is shown in Figure 2. Microscopic images presented the features typical of tension and opposite wood.Gelatin fibers are mainly seen in the tension wood section (signed A 40 and A 80 ).The lower content of gelatin fibers in the near pith zone (A 40 ) of the tension wood than the content of gelatin fibers in the outer zone (A 80 ) confirms previous information that asym-metric growth is not always accompanied by the formation of reactive tissue across the width of the tree trunk (Kojs et al., 2012).The microscopic images of the test samples taken from the pith area and peripheral parts of section B (marked B 40 and B 80 ) are characteristic of the structure of the opposite wood.The anatomical elements have smaller lumen with a thicker wall than normal wood.

Kemijski sastav drva
Table 4 summarizes the experimentally measured chemical composition of the tension wood (Section A) and opposition wood (Section B).These data were compared with the literature data for normal wood.The cellulose content is similar, but substances soluble in alcohol content in the tension wood samples (Section A) and opposition wood are ca.twice as high as in normal wood.Moreover, a higher hemicellulose content (in Sections A and B) was found than in normal beech wood.

Sorption isotherms 3.3. Sorpcijske izoterme
The sorption isotherms of the reaction beech wood for the successive phases: initial (first) desorption, adsorption, and second desorption at a temperature of 20 °C are shown in Figure 3.
For all tension and opposite wood samples, it was confirmed as follows: the sorption hysteresis occur-rence (desorption differs from the subsequent adsorption), differences between the first (initial) and second water desorption, the EMC value in the initial desorption was higher than the EMC value in the second desorption in the range of RH air above 70 %.The usefulness of the FPL data, which is presented as an additional isotherm (dot line, Figure 3), was confirmed only for adsorption.For this reason, the FPL data have limited suitability for determining the technological parameters of drying beech lumber with a high amount of "reaction wood".In our research, significantly lower EMC values were observed in the first desorption of  The influence of extractives on the sorption phenomenon was described previously, and it was pointed out that their higher content causes a decrease in EMC (Simón et al., 2015;Jankowska et al., 2016).The results of sorption experiments confirm earlier literature reports that wood containing more substances soluble in alcohol achieved lower EMC, especially when air RH is above 50 % (Hernández, 2007).

Initial desorption modeling 3.4. Modeliranje inicijalne desorpcije
Figure 4 presents a comparison of the EMC response surfaces for initial desorption as observed for reaction green beech wood and the FPL reference EMC data.The maximum difference between the estimated EMC of the reaction beech wood and the reference FPL data for the highest RH values included in the experiments is up to 0.14 kg/kg (for air RH above 90 %).
The responses models (Figure 4) present the differences in the hygroscopic properties of the tension and opposite wood tissues.The most significant differences in the EMC value occur in the air RH range above 70 %.The EMC of the tension wood (Section A) was significantly higher than that of the opposite wood (Section B).Moreover, responses models for the reac-tion tissue show that the EMC values are much more temperature-dependent for the initial desorption than can be calculated using the Simpson procedure, taking into account the FPL data (Simpson, 1973).In extreme cases, an increase in temperature from 20 to 50 °C, for RH near saturation reduces the EMC of the reaction beech wood by even 0.010 to 0.15 kg/kg.The EMC reduction for the same conditions calculated from the FPL data is only 0.025 kg/kg.
Table 5 shows the estimated coefficients of response models developed in this study.Figure 6 shows calculated absolute differences between reaction wood EMC for initial desorption and normal wood EMC according to FPL data (Simpson, 1973).

ZAKLJUČAK
Taken together, these findings demonstrate that: 1.The research results show the possible range of variability of the hygroscopic properties of the raw beech wood containing pathological tissue.The dependence of the hygroscopic properties of the examined wood on the type of pathological tissue (tension wood, opposite wood) and chemical composition was confirmed.The experimental results show lower EMC values of tension beech wood in higher air RH values.The higher extractives content in reaction wood than in normal tissue is the most likely cause of the lower EMC. 2. It was confirmed that the EMC value for the initial desorption is higher than for the second desorption (in the range of RH above 70 %).3. The EMC values for the initial desorption for investigated tissues are much more dependent on the temperature than in the Wood Handbook data, which does not consider the anomalous properties of the reaction wood.4. The calculated EMC value corresponding to the initial desorption can verify kiln-drying schedules for beech sawn timber with a high content of reaction tissue.Applying the developed initial desorption models can significantly improve the accuracy of the drying gradient determination.It can be concluded that there is a potential to improve the efficiency and drying quality of beech kiln-drying.(Simpson, 1973.)

Figure 5
Figure 5 compares the predicted EMC to the experimental EMC.Figure6shows calculated absolute differences between reaction wood EMC for initial desorption and normal wood EMC according to FPL data(Simpson, 1973).

Table 3
dova of the chemical composition of the examined wood samples.The post-hoc HSD Tukey's test was used to test the significance of differences between the average values of the mean.Significance was established at p<0.05.