SORPTION BEHAVIOR OF FOUR TROPICAL WOODS USING A DYNAMIC VAPOR SORPTION STANDARD ANALYSIS SYSTEM

Sorption behavior of four tropical woods coming from Cameroon is studied using a Dynamic Vapor Sorption-Intrinsic apparatus. Sorption isotherms at 20°C and 40°C of Triplochiton scleroxylon (obeche), Entandrophragma cylindricum (sapele), Sterculia rhinopetala (lotofa) and Terminalia superba (frake also called limba) are compared to those in the literature with methods using saturated salts for lotofa and frake with satisfactory results. Experimental values are fitted using two sorption isotherm models. Sorption hysteresis is greatest in sapele and least in frake. When temperature increases, differences in the hysteresis of sorption among the woods decreases. Hystereses of all studied woods ranged from 0,5 to 3%. Our studied woods have lower equilibrium moisture content probably caused by the higher values of extractives content. The water sorbed molecules on multilayer during both adsorption and desorption at 20°C and 40°C is greater in the case of frake and lower in the case of lotofa. The water sorbed molecules on multilayer of sapele is greater than those of obeche. The predominance of multilayer over monolayer of our woods is from 0,2 to 0,4 relative humidity in adsorption phase, and from 0,3 to 0,4 relative humidity in desorption phase.


INTRODUCTION
It is well known that tropical countries have a lot of woods species. For example, Cameroonian forests have more than 300 woods species. These woods are used in many domains such as plywood, decorative veneers, interior joinery, furniture and packaging (Fernández et al. 2014).
The sorption isotherms are essential properties in the analysis and design of various biological material processes such as preservation, drying (Salin 2011), storing, packaging and mixing. It can be used to determine the net isosteric heat of sorption (Ouafia et al. 2015), for selecting appropriate storage conditions and packing systems that optimize or maximize the mechanical stability of biological material.
Deliiski (2011) presents wood sorption isotherms, but no specific data is given on each species. Jannot et al. (2006) use GAB and BET models to plot desorption isotherms and estimate specific areas of doussié, moabi, ebony, obeche and iroko at 20, 30, 40, 50, 60°C. In the same temperatures previously cited, Bonoma and Simo-Tagne (2005) use Henderson and Chung-Pfost models to explain the evolutions of equilibrium water content in the desorption phases of ebony and obeche with relative humidity. Simo-Tagne et al. (2011) present a correlation between equilibrium water content in function of temperature and relative humidity and estimate the isosteric heat of sorption of obeche and ebony using GAB model. Also, they studied the Soret effect on obeche for various temperatures. Fernández et al. (2014) determine the sorption isotherms and thermodynamic properties of frake and obeche at 15, 35 and 50°C.
In the literature, several authors use the standard static gravimetric method to determine the sorption isotherms (Ouafia et al. 2015). Description of this method is available on Jannot (2008). This method is so long to obtain an equilibrium state of the sample on a given conditions. Applied on wood material, the samples with 1mm of thickness are often used and 30 to 40 days are necessary to obtain one equilibrium state (Jannot et al. 2006, Fernández et al. 2014, Esteban et al. 2004, De Palacios et al. 2012. Simo-Tagne et al. (2016) have presented the experimental EMC, mass diffusivity coefficients and air permeability of these woods. Concerning the EMC, discussion between values obtained with DVS apparatus and others methods, the monolayer saturation moisture content, the modeling of the monomolecular water sorbed and the polymolecular water sorbed are not studied in Simo-Tagne et al. (2016).
The objectives of present work are: (1)To use DVS apparatus to determine and analyze the sorption isotherms (adsorption and desorption) of Triplochiton scleroxylon (obeche), Entandrophragma cylindricum (sapele), Sterculia rhinopetala (lotofa) and Terminalia superba (frake), four tropical woods most commercialized. (2)To deduce the hysteresis of sorption of each studied wood. (3)To study the behavior of water molecules in the scale of pore and (4) to compare the values obtained in this work by those obtained in the literature.

Materials Preparation
All tests presented here have been done in the Studies and Research on Wood Material Laboratory (LERMAB) of University of Lorraine, located in Epinal, France.
As Engelund et al. (2010) and Jalaludin et al. (2009), isotherms were performed using a dynamic gravimetric water sorption analyzer from Surface Measurement Systems (DVS-Intrinsic) ( Figure  1). Specimens were collected from natural forests of Douala Cameroon and from mature log. The specimens used were very fine wood-chips cut with a microtome blade to obtain samples on each board with dimensions approximately 1x0,5x2mm.

Test Conditions
The sample is weighted using a digital microbalance (total capacity of 1 g; noise < 1 µg). Approximately 26 mg of each sample were used for this measurement and two tests are made on each species. The sorption cycles applied in this work started from 0% relative humidity (RH) to 90% RH. The dry mass is determined at the end of this plateau. Isotherms were made at 20°C (± 0,02 °C) and 40°C (± 0,02 °C). Samples were maintained at a constant RH level until the weight change per minute (dm/dt) value reached 0,0005% per minute. Figure 2 is an example of evolutions of RH and weight change with the time in the case of sapele at 20°C. We see that this experiment (adsorption and desorption) was made around 4500min. The raw DVS data was first analyzed using the DVS Standard Analysis System. This software permits construction of isotherm and hysteresis curves. For all samples, Equilibrium Moisture Content values were determined using the last fifteen data of each plateau.

Sorption Analysis
The Hailwood-Horrobin sorption model (HH-model) (Hailwood and Horrobin 1946) and the Guggenheim, Anderson and De Boer-Dent model (GAB-model) (Ouafia et al. 2015) are used for describing the envelope curves (desorption and adsorption isotherms from the extreme values 0% and 90% of RH) because the agreement of the measured values with the computed isotherms are usually very good (Fernández et al. 2014, Popper et al. 2009). Mathematical equations of these models are briefly recalled on Table 1.  Table 1. Equations used to model equilibrium desorption and adsorption isotherms.

Names and References
Equilibrium moisture content model HH-model (Hailwood and Horrobin 1946) GAB-model (Ouafia et al. 2015) With: X eq : Equilibrium moisture content (%); X m : Monolayer saturation moisture content (%). The insignificant values of X m parameter may result from the weak presence of the active sorption sites available within the wood cell walls; C: Guggenheim constant (-) which corresponds to the total sorption heat of the first sorbed layer (Alix et al. 2009); K: Constant associated to water sorbed molecules on multi-layer (-) (Alix et al. 2009); RH: Water activity or Relative humidity on a scale of zero to one (-).
A, B and C o are the shape factors. We can use these values to estimate the equilibrium between the activities of absorbed water and water vapor (K d ), the equilibrium between the activity of hydrated wood over the activity of dry wood and absorbed water (K h ) and the apparent molecular weight of wood per sorption site W i (Dieste 2010). We have (Dieste 2010): A low K h implies a decrease in the activity of hydrated wood with respect to both the activities of dry wood and dissolved water, the increase of W i is taken as evidence for a decrease in the number of active sorption sites. The HH model is composed into the adsorbed water or monomolecular water sorbed (X h ) and the absorbed water or polymolecular water sorbed (X d ). We have (Dieste 2010): Monolayer moisture is the water within the cell wall that is associated closely with the OH groups, while polylayer water is when water occurs within the transient microcapillaries but is not intimately associated with the OH groups ).
The sorption isotherm fits were considered valid if the correlation coefficient (R) was superior to 0,990 (Fernandez et al. 2014). The water taken up by the sample via monolayer sorption from RH m to 1RH is given by (Esteban et al. 2006): Where X ref is the moisture content corresponding to RH m that gives the minimum of the derivative of equilibrium moisture content in relation to the RH.

1800
(1 )  Figure 3 presents the measured data obtained in this study at 20°C and 40°C respectively. It is clear that the sorption behavior of tropical species is different and apparatus used does not permit the obtaining an idea on the sorption behavior when water activity is above 90%. But, during the conservation, storage and packaging of finished products, air relative humidity used are below 90%. Both adsorption and desorption phase, obeche has higher equilibrium moisture content (EMC) values compared with others, frake has lower EMC. EMC of sapele is higher than the one of lotofa.

RESULTS AND DISCUSSION
The temperature dependence of the EMC seems to be weak (Figure 3). This influence of temperature is conform to the results obtained by Rodriguez-Ramirez et al. (2011) and Björk and Rasmuson (1995). Also, we noticed that differences on our EMC are probably caused by the differences of extractive content, history of the sample, the wood structure and the part of the sample in the log.  Figure 4a, we see that present results at 20°C are lower than those obtained by Fernández et al. (2014) at 15°C, because EMC decreases when temperature increases (Esteban et al. 2004, Fernández et al. 2014, Jannot et al. 2006, Simo Tagne et al. 2011). But, we see a weak difference between present values and those obtained in the literature (Jannot et al. 2006, Simo-Tagne et al. 2011 on obeche at 20°C in desorption phases. It is probable that the origin, history of each sample used and natural variability from the species are the reason of this difference. Figure 4b gives a good agreement between present results and those obtained in the literature (Jannot et al. 2006, Simo-Tagne et al. 2011 on obeche at 40°C in desorption phases. Figure 5 shows that our results at 20°C are lower than those obtained on the same wood species (frake) at 15°C. Figure 6 presents hysteresis of sorption given by present results. We see that at 20°C, hysteresis of sorption are greater than those at 40°C. Hysteresis increases when relative humidity increases. Tekleyohannes (1995) had obtained the same results. Sapele has higher hysteresis values compared with obeche. Frake has lower levels of sorption. When temperature increases, difference of the sorption hysteresis between our woods decreases. Also, we see that sorption hysteresis of studied woods are ranged from 0,5 to 3%.    Table 2 shows that intensities of correlation are greater than 0,9. Table 3 presents K d , K h and W i .We observe that both adsorption and desorption phases at 20°C and 40°C, obeche has a low equilibrium between the activities of adsorbed water and water vapor (K d ) and the one of frake is the great. The one of sapele is greater than the one of lotofa. Both adsorption and desorption at 20°C and 40°C, lotofa has less accessibility for monolayer water (K h is weaker). Sapele has most accessibility for monolayer water (K h is higher). Both adsorption and desorption, at 20°C, frake has less accessibility for monolayer water compared of obeche, but at 40°C, inverse is presented. Both adsorption and desorption phases at 20°C and 40°C, frake has a great apparent molecular weight of wood per sorption site (W i ) and obeche has the lower. At 20°C, obeche has the great W i than lotofa. K d , K h and W i increase with the temperature in desorption, but decrease with temperature in adsorption. Table 4 confirms that isotherms plotted are the type II sigmoid (because C ˃ 2 according to Esteban et al. (2004) and Fernandez et al. (2014)) and intensities of correlation are greater than 0,9. We saw that monolayer saturation water content is greater in the case of desorption than the case of adsorption. As well as adsorption than desorption, the monolayer saturation moisture content X m is lower in the case of frake and higher in the case of obeche. X m of lotofa is greater than the one of sapele.     Figure 7 presents evolutions of measured and modeled EMC (during desorption and adsorption), also the monolayer sorption (X h ) and polylayer water sorption (X d ) applied on sapele (at 40°C) and on lotofa (at 20°C). The HH and GAB models seem to explain plausibly the sorption of water in wood. The monolayers are always greater in desorption than in adsorption and it is most affected by the sorption hysteresis. For this reason, X m obtained between adsorption and desorption are more different. The polylayers in desorption and in adsorption are near when RH is inferior of 0,7. The monolayers sorption are reached at equilibrium moisture fractions in the range 0,04-0,06 such as announced in the literature (Björk and Rasmuson 1995). Figure 8 shows that the point of inflexion of the isotherm corresponds to a moisture content of 0,04 kg/kg and occurs at 0,3 RH in adsorption at 20°C (Figure 8a). This point of inflexion corresponds to a moisture content of 0,08 kg/kg and occurs at 0,4 RH in desorption at 20°C (Figure 8b). Beyond each of these points, multilayer sorption predominates over monolayer sorption. These findings indicate that, from 0,3 to 1 RH (in adsorption) and from 0,4 to 1 RH (in desorption), the water taken up by the samples via monolayer sorption is minimal, at only is ranged from 0,6 to 3,85% of the total (given in Table 5). Effectively, example given on Figura 8, shows that when EMC is higher than X m , X h versus RH is not constant and vary slowly. At the given temperature, X up is higher in desorption than in adsorption. Applied on the old wood of Pinus sylvestris, Esteban et al. (2006) indicate that in desorption, from 0,316 to 1 RH the water taken up by the sample via monolayer sorption is minimal, at only 0,7% of the total.

CONCLUSIONS
In this work sorption behavior of tropical woods applied on frake, obeche, lotofa and sapele is studied and analyzed using DVS-Intrinsic apparatus. The duration of each adsorption-desorption cycle Sorption behavior of four tropical..: Simo-Tagne et al. is almost 4500min. DVS-Intrinsic apparatus gives satisfactory results on sorption isotherms of studied tropical woods. These sorption isotherms are well fitted with GAB and HH models and the results indicate a type II sigmoid shape for the EMC-RH curves, showing a decrease of the EMC with the increase of temperature, but with a weak influence. Sorption hysteresis most affects the monolayer saturation moisture content, the proof that the cycle desorption-adsorption most affects heat of sorption in the monolayer scale and briefly physical and chemical sorption properties in the scale of the pores. Monolayer saturation moisture content given by X m is always higher than the moisture content corresponding to RH m that gives the minimum of the derivative of EMC in relation to the RH. By growth order of magnitude, active sorption sites available within the wood cell walls are frake, sapele, obeche and lotofa at 20°C; frake, sapele, lotofa and obeche at 40°C. Air temperature, air relative humidity and wood species affect the sorption behavior and sorption hysteresis of the tropical woods. The predominance of multilayer over monolayer of our woods is from 0,3 to 0,4 RH.