.+*
w+
SHRINKING and SWELLING of WOOD Factors affecting the dimension changes of wood resulting from moisture content changes tire discussed. A simple relationship is shown to exist between the shrinkage and the density of small specimens of wood dried under practically stress-free conditions, which indicates that the volumetric shrinkage is practically equal to the volume of water lost below the fiber-saturation point and that the change of fiber cavity dimensions is small. Deviations from this relationship caused by drying stresses are explained. Data for the swelling of wood in aqueous electrolyte solutions and in dry organic liquids are presented and considered from the standpoint of swelling theories. Data on the replacement of water in swollen wood by other liquids are presented.
further in that they exhibit a HE most detcoarse capillary structure that is r i m e n t a1 capable of taking up liquids withD.roDertv of o u t a c c o m p a n y in g swelling wood from the s t a n i p o i n t of as well as the c h a r a c t e r i s t i c structural use is undoubtedly capillary structure of swelling. that of shrinking and swelling. This fact greatly complicates the A large proportion of the degmeasurement of the swelling of radation resulting during the cellulosic materials. The swellcourse of logging, milling, seasoning of gelatin can be followed by ing, and construction processes weight changes alone. This can is either directly or indirectly due be done for cellulosic materials to the dimension changes of the only with the greatest precauwood resu!ting from loss or gain tions, since only the true adsorpof m o i s t u r e . Even after the tion is accompanied by swelling. processing of the m a t e r i a l is For example, a piece of wood c o m p l e t e , fluctuations in the can take up about 30 per cent dimensions of the material conof w a t e r o n the basis of the tinue to occur with changes in w e i g h t of the dry wood with atmospheric conditions. The accompanying swelling. When a r t i f i c i a l m i n i m i z i n g of the shrinking and swelling of wood immersed in water it may take u p f r o m 150 t o 300 per c e n t for structural purposes is thus ALFRED J. STAMM more water in the coarse capilhighly essential in order for wood Forest Products Laboratory, l a r y s t r u c t u r e without any to regain part of its lost markets further swelling taking place. and even to retain its present Madison, Wis. markets. So-called swelling measurements In this paper recent fundahave been made on cellulosic mental findings on the shrinkmaterials by determining the ining and swelling of wood are crease in weight of the material presented to give a background after immersion in various liquids for more applied investigations a n d s o l u t i o n s . T h i s , howon the minimizing of these detrimental properties. There ever, gives a measure of the absorbency and depends upon are three essential characteristics of the phenomenon of swellthe capillary structure of the material, the surface tension of ing: (1) The solid material increases in both dimensions and the liquid, and the ability of the liquid to wet the solid as well weight with an accompanying thermal change as a result of as upon the extent of swelling. It would therefore be prefthe taking up of another phase; (2) the combined phases beerable to make basic swelling measurements of cellulosic macome and remain homogeneous in a microscopic sense; and terials on a volume change basis. This, however, is difficult (3) the cohesion of 1,he solid is reduced but not eliminated. if not impossible to do experimentally because of the internal as well as external dimensional changes. Even in wood Phenomenon of Swelling where the external dimensional changes are easy to determine, the internal changes in the fiber cavities may be very comThe phenomenon of swelling is characteristic of all elastic plicated, thus making possible only approximations of the colloidal materials but differs somewhat for different types of magnitudes of the changes. A study of these dimensional materials. Gelatin, For example, can swell to the point where changes is worth while because it will lead t o a better underthe cohesion of the solid becomes negligible; that is, it shows standing of the mechanism of the swelling of fibrous materials. a gradual transition from a swollen to a dispersed condition. Cellulosic materials, on the other hand, show no such propShrinking and Swelling of Wood erty in ordinary swelling agents. The cohesion of the solid structure is, under practically all conditions, sufficiently The shrinking and swelling of wood as manifested by the great to limit the distention of the solid and imbibing of external dimension changes vary with the species, structure, liquids. Cellulosic materials also show a definitely oriented structural directions, density of the wood, and drying condistructure that gives rise to differences in the dimensional swelltions. On drying green wood, no shrinkage occurs until the ing in the three structural directions, and they lack the ammoisture content is reduced below the fiber-saturation point photeric property that makes gelatin readily responsive to (the moisture content below which the activity of the water p H changes. Cellulosic materials differ from gelatin still becomes less than unity). This varies for most species from
@T
401
402
INDUSTRIAL AND ENGINEERING CHEMISTRY
VOL. 21, NO. 4
content might be expected to hold on a total shrinkage basis over the sorption range if the effect of adsorption compression of water is small or if it changes in a practically linear manner with changes in moisture content. The fact that the linear relationship holds for shrinkage manifested externally indicates that the internal shrinkage into the fiber cavities must be either proportional to the external change or practically negligible. The following data will show that drying of small specimens under carefully controlled conditions causes a practically negligible change in the size of the fiber cavities. IRTERNAL SHRINKINGAND SWELLING.The internal dimensions of the voids in a hypothetical gel, free from stresses and dried under stress-free conditions, would change the same as the dimensions of the bulk of gel that would just fill the voids. Under these conditions shrinking and swelling as transmitted t o the external dimensions would be independent of the bulk density or degree of porosity of the gel. This is not the case for wood, which shows an appreciably greater shrinking and swelling for a greater density, indicating that the external dimension changes are not independent of the volume of the voids. The decrease in the volume of the voids on shrinking is necessarily less than that for the same volume of gel; in fact, the voids may not decrease in size a t all or may even increase during the course of shrinking. This is perhaps to be expected on the basis of the structure of wood fibers, which are made up of bundles of fibrils approximately parallel to the length of the fiber with outer fibril wrappings a t right angles (12). The inner and outer part of a fiber wall will tend to respond differently to moisture content changes, thus necessitating a modification of the direction of the resulting dimension changes. 0 2 4 6 8 IO /2 14 16 18 20 22 24 26 28 The volumetric shrinkage of small sections of wood dried MOI5TURE CON7€NT [PdRCENT OF DRY WEIGH7 O f W O O 4 FIGURE 1. VOLUMETRIC SHRINKAGE OF THINTANGENTIAL under minimum stress conditions was found to be approximately proportional to the bulk density of the sections with SECTIONS OF DOUGLAS FIR HEARTWOOD OF DIFFERENT DENSITIES but slight shifts of the proportionality constant for different species. The proportionality constant is practically equal to the fiber-saturation point. As a first approximation the following relationship seems to hold: between the threadlike micelles of the cellulose fibers which, in general, are oriented parallel to the length of the fibers. The greater shrinkage in the tangential direction than in the s. Pwf* (1) radial is accounted for on the basis of ray cells which have where S. = percentage volumetric shrinkage from Ratd. to oven-dry condition on an external dimension their longitudinal component in the radial direction of the change basis wood, restraining the dimension changes in that direction. p w = bulk density of the wood on a dry weight-green MOISTURECONTENT us. SHRINKING AND SWELLING. The volume basis external dimensional shrinkage of small specimens of wood, fa = fiber-satn. point, in terms of volume of water held per 100 grams dry wood dried under gradual stepwise reduced relative-humidity conditions so as practically to eliminate moisture gradients, has The relationship requires that the volume of the grosser been shown by both Morath (6) and the author (16) to be alcapillaries does not change with changes in moisture content most a direct function of the amount of moisture removed and that the total volumetric swelling is approximately equal from the fiber-saturation point to the oven-dry condition for to the volume of water taken up. all the structural directions. Further data are given in This same relationship holds for swelling as well as for Figure 1 illustrating this linear relationship for the external shrinking. In this case 8, is the percentage external voludimensional volumetric shrinkage of thin tangential sections metric swelling from oven-dry t o the saturated condition, pr of Douglas flr, cut from material with broad annual rings so is the bulk density of the wood on a dry volume and dry that varying proportions of spring wood and summer wood of weight basis, and fa is the same as given in Equation 1. a single ring exist in each section, thus giving adjacent secThis fiber-saturation point on a volume basis is, in general, tions of quite widely varying density. Even data obtained 2 to 5 per cent less than the generally accepted fiber-saturaon considerably larger specimens in which drying moisture tion point on a weight basis and would indicate that the gradients cannot be eliminated, deviate but slightly from the average compression of the sorbed water is of this order of linear relationship. Data (IO) for loblolly pine sapwood magnitude. These results are in keeping with the predicted boards X 51/2 X l / 8 inch) are plotted in Figure 2. Apadsorption compression values obtained from measurements preciable deviations from the linear relationship occur only of the density of wood substance in water and in a nonswella t the high moisture content values. These deviations reing medium for which an average compression of the sorbed sult from the outer part of the material falling below the fiberwater for spruce of 2.5 to 3.5 per cent has been obtained (1'7). saturation point tending to shrink while the average moisture For example, a small transverse section of Sitka spruce with content is still above the fiber-saturation point. This effect a density of 0.401 gave an external dimension volumetric of moisture gradients upon the dimensional shrinkage will be shrinkage of 11.3 per cent (16). The fiber-saturation point considered more a t length later. constant, fa, is thus equal to 28.2. The normal fiber-saturaThe linear relationship between shrinkage and moisture about 27 to 32 per cent of the weight of the dry wood (15). The shrinkage is greatest in the tangential direction (tangent to the annual rings), varying from 5 to 12 per cent. In the radial direction (from the center of the tree to the periphery) it varies from 2 to 7 per cent, and in the longitudinal direction (parallel to the fibers) it is only a small fraction of 1 per cent. The practically negligible longitudinal swelling and shrinking are accounted for on the basis of the swelling taking place only
-
APRIL, 1935
INDUSTRIAL AND ENGINEERING CHEATISTRY
tion point on a weight basis as obtained from electrical conductivity measurements (16) and shrinkage measurements (16)is 29.0. The compression in this case is thus 2.8 per cent. The compression for white pine was found in a similar manner to be 4.2 per cent. Values for the shrinkage of thin tangential sections of Douglas fir of different densities are given in Figure 1 and Table I. The data show that the constant fa obtained from the equation is within the experimental error of the fiber-saturation point expressed on a volume basis. The fiber-saturation point on a weight basis was obtained from the zero shrinkage intercept of the moisture content-shrinkage relationship. Dividing this by the average density of the absorbed water gives the fiber-saturation point on a volume basis. Schwalbe and Beiser (19) found from an analysis of their data for the swelling of wood that the fiber cavity dimensions are practically unchanged when assuming a fiber-saturation point of their wood of 30 per cent, Beiser (1) has also shown from microscopic studies of wood cross sections that the change in the size of the fiber cavities resulting from swelling and shrinking of the wood is small. He found that the fiber cavities of the springwood increase in diameter slightly or do not change a t all upon swelling of the wood, whereas the fiber civities of the summerwood actually decrease slightly in size. Such an effect would be expected as the heavier walled, denser, summerwood fibers would tend to exhibit a greater external swelling than the less dense springwood fibers. The summemood fibers would thus exert an external tension on
403
The author has also obtained data on the permeability of the fiber cavities of wood to air of different relative humidities which indicate that on the average the size of the openings changes only slightly with shrinking and swelling of the wood, Over a complete cycle of relative humidity changes the maximum permeability variation for the heartwood of western hemlock was 4.0 per cent; this value is equivalent to a maximum change of the diameter of the fiber cavities of
TABLEI. VOLUMETRIC SHRINKAGE-DENSITY DA’rA FOR THIN TRANSYEI~SE SECTIONS OF DOUGLAS FIR P.W,
Ex’’giment
Density on Green Vplume Basis Grams per cc.
f., Fiber-Satn. Constant on Vol. Banis From point External of initial Volumetric From shrinkage Shrinkage equation 129.2/1.035) Per cent Cc. per 100 grams
SS.
Av. 2 8 . 1 5
TABLE 11. VOLUMJCTRIC SHRINKAGE-DENSITY DATA LOBLOLLY PINESAPWOOD BOARDS
FOR
fa, Fiber-satn.
Nature of Specimen Av. 50 specimens
Constant on Vol. Basis Estd. PW, S 6, fron. Density External initial on Green Volushrinkage Volume metric From (29.0/ Bmie Shrinkage equation 1 .O3B) G.,/cc. Per cent Cc./lOO oram8 0.471 11.2
Densest specimen Least dense specimen 0 . 36 52 89 Least deviation 0.538 Greatest deviation 0,409
1;:; 13.9 7.7
%]
25.8
18.8
28,0
Average Increase in Fiber Cavity Cross Section Per cent
1.9 5.2
the springwood fibers, and the springwood fibers would exert an external compressive force on the summerwood. The tension force would tend to cause a greater external swelling of the springwood than normal and hence a compensating increase in the size of the fiber cavity, whereas the external compression on the summerwood fibers would tend to reduce their external swelling. This would have to be compensated for by a swelling into the fiber cavities, resulting in a decrease in their dimensions.
FIGURE2. VOLUMETRIC SHRINKAGE OF LOBLOLLY PINESAPWOOD BOARDS OF DIFFERENT DENSITIES
only 1.0 per cent, the mean deviation being less than half of this percentage. Even in the drying of larger specimens of wood, the change in the fiber-cavity dimensions may be relatively small. Newlin and Wilson (8) give the relationship between the external volumetric shrinkage from the green to the oven-dry condition and the density of the mood on a green volume basis for the average of several thousand tests on 121 different species. The shrinkage for the average increases linearly with an increase in density giving an average value off. of 28. The average fiber-saturation point for all of these species is perhaps about 30 per cent on a weight basis. This value corresponds to about 29 cc. per 100 grams of wood on a volume basis. There is thus an indication of the shrinkage being slightly less than that required by the foregoing equation, which means that a slight increase in fiber lumen dimensions may accompany shrinkage. The data of Peck (IO),which are plotted in Figure 2 and the corresponding calculations of which are given in Table 11,illustrate this point still further. If these larger specimens could have been dried under conditions giving negligible moisture gradients and negligible resulting stresses, the moisture content intercept of the extended straight line to zero shrinkage would be a measure of the normal fiber-saturation point on a weight basis. The premature shrinkage caused by the presence of moisture gradients shifts the linear part of the shrinkage curve to lower moisture content values. The fiber-saturation point, instead of being 27 to 28 per cent, is in reality perhaps nearer 29 per cent or, on a volume basis, 28 cc. per 100 grams of wood. The values of f, calculated from the densities are given in Table 11. In all cases they are less than 28. The
404
INDUSTRIAL AND ENGINEERING CHEMISTRY
deviation for the average of all the specimens is 4.2 cc. per 100 grams of wood. In order for equilibrium to be attained, this difference in volume must be accounted for by an internal shrinkage that will cause an increase in the fiber-cavity dimensions. The percentage increase in the cross section of the fiber cavity is given in the last column of Table 11. It was obtained by dividing the difference between the theoretical volumetric shrinkage on an external dimension basis (us-
VOL. 27, NO. 4
The shrinkage of wood under conditions that practically eliminate drying stresses is thus shown to take place almost entirely by changes of the external dimensions. When stresses are set up during the course of the shrinkage, part of the shrinkage is internal and results in an increase in the dimensions of the fiber cavities. The extent of the internal change thus depends on the magnitude of the stresses, A deviation from the shrinkage-density relationship in the opposite direction from that caused by drying stresses and set can result from collapse. This phenomenon takes place when certain species of hardwood which have high normal moisture contents when green are dried below the fiber-saturation point (9). It is a localized caving in of the fibers and rarely if ever occurs uniformly over a specimen. If the shrinkage of a specimen exceeds that called for by the shrinkage-density relationship, then it can be assumed that the effect of collapse exceeds that of drying stresses and set.
Swelling in Aqueous Solutions Wood in concentrated aqueous solutions of salts will swell beyond its normal water-swollen dimensions. Series of saturated salt solutions of chlorides and potassium salts gave the following order of increasing volumetric swelling as transmitted t o the external dimensions (16): K
