ADSORPTION COMPRESSION ON CELLULOSE AND WOOD. I DENSITYMEASUREMENTS IN BENZENE ALFRED J. STAMM
AND
R. M. SEBORG
Forest Products Laboratory,' Forest Service, U. S. Department of Agriculture Received June 14, 1934 INTRODUCTION
The apparent density of an adsorbent material with a large extent of internal surface varies with the nature of the displaced medium used for making the measurements. Harkins and Ewing (9) attributed the phenomenon in the case of charcoal to varying degrees of adsorption compression of the displaced liquid in the capillary structure. They obtained the greatest densities in the most compressible liquids. The size of the molecules of the displaced liquid and the ratio of their surface tensions to viscosities showed no correlation with the densities. They concluded that the differences in the density values could not be due to differences in penetration. Lamb and Coolidge (13) calculated compressive force values as high as 37,000 atmospheres from their heats of adsorption measurements on charcoal, by assuming that all of the heat evolved is due to compression of the liquid. Cude and Hulett ( 5 ) made measurements of the density of charcoal in various liquids which they injected into the charcoal under pressure to facilitate penetration, and Howard and Hulett (10) made measurements in helium gas which they showed to be nonadsorbed on the charcoal. The helium gas gave quite similar values to those obtained in the various liquids, thus indicating that any adsorption compression taking place in the liquids would have to be much less than the values predicted by Lamb and Coolidge. Davidson (6) determined the specific volume of cotton in helium gas, water, and several organic liquids. Non-polar organic liquids gave the highest values, the helium gas values were but slightly less, and the water values were still smaller. He concluded that the helium gas gives the true values and that the slightly higher values in the non-polar organic liquids are due t o incomplete penetration, while the low values in water are due to adsorption compression. He calculated values for the average compression of the sorbed water of approximately 2000 atmospheres on 1 Maintained a t Madison, Wisconsin, by the Forest Service of the U. S. Department of Agriculture, in cooperation with the University of Wisconsin.
133
134
ALFRED J. STAMM A N D R . M. SEBORG
the basis of the compression taking place entirely in the water. This assumption seems justified, especially in the case of cellulose, because of the fact that x-ray measurements show no change in the x-ray diagrams on swelling of the cellulose in water (11, 12). Similar measurements on wood have been made a t the Forest Products Laboratory (16). Helium gas gave intermediate density values between those obtained in water and in non-polar organic liquids as in the case of Davidson’s measurements, but the helium gas values were nearer to the water values than to the non-polar liquid values. I n the light of more recent findings these helium gas values appear high. Measurements in this medium are now being repeated. A series of organic liquids gave the greatest density values in the most polar liquids which cause swelling and the lowest density values in the non-polar liquids which do not cause swelling. No relationship between the density values and the compressibilities of the liquids was obtained. Water, a slightly compressible liquid, gave the greatest density. These results for hydrophilic cellulose may appear to be in conflict with those of Harkins and Ewing (9) for hydrophobic carbon, but this is by no means the case. The internal surface of contact for non-swelling carbon is constant except for possible slight variations in penetration, while in the case of cellulose and wood, which are subject to swelling, it may vary from several hundredfold to a thousandfold. For example, the estimated internal swollen surface of cellulose and wood varies from 3 X lo6 to lo7 square centimeters per gram (14), while the coarse capillary surface of unswollen wood (total lumen area) varies from 1to 5 X 10s square centimeter per gram. The extent of the surface on which compression can take place, rather than the magnitude of the compression per unit of surface, thus appears to be the prime factor in causing the variation in the density values of swelling solids obtained in different liquids. Filby and Maass (7) have made measurements in helium gas of the density of cellulose with various amounts of adsorbed water present by an ingenious combined gas expansion and moisture sorption method. From the few measurements which they reported they drew several very farreaching conclusions : First, that the compression takes place only within the first 8 to 10 per cent of moisture adsorbed; second, that a constant maximum compression is reached at moisture content values below about 4 per cent; and third, that the initial compression amounts to 0.6 cc. per gram of water adsorbed. An externally applied compressive force of about 100,000 atmospheres would be required to give this compression. The first two conclusions are not in harmony with moisture contentrelative vapor pressure data or heats of swelling data (17). Vapor pressure depressions and differential heats of swelling are obtained at all moisture content values up to the fiber-saturation point, which varies from
ADSORPTION COMPRESSION ON CELLULOSE AND W O O D
135
15 to 35 per cent for the different fibrous materials. It thus seems unreasonable that the compression, which is a related property, should not be effective up to these higher moisture content values. F’urther, the moisture content-vapor pressure and the moisture content-entropy change data show inflection points at presumably the transition point between the bound and capillary held water, but they do not indicate a constancy of attractive force of all of the sorption points (17) as in the case of the data of Filby and Maass. The extremely high compression obtained by these investigators would require that the work of adhesion of cellulose and water be seven to eight times the work of cohesion of water. This, too, is a t variance with other data. Bartell and Osterhof (2) and Bartell and Jennings (1)have shown from adhesion tension measurements that the work of adhesion for a solid wetted by water is but 1.00 to 1.05 times the work of cohesion of water. The work of adhesion of mercury and water according to Harkins (8) is only 1.25 times the work of cohesion of water. Because of these irreconcilable results of Filby and Maass and the general unsettled opinions as to the magnitude of adsorption compression, the Forest Products Laboratory undertook this research with the hope of settling this point, which has considerable bearing on the general problem of sorption. EXPERIMENTAL
The measurements of Davidson (6) show that the specific volume of cellulose determined in toluene is only a little over 1 per cent greater than in helium gas. The difference must be due entirely to incomplete penetration by the toluene, as toluene is practically a non-swelling liquid in which the compression can be only 0.01 to 0.001 of that in a swelling liquid. Such a small compression would be outside the range of experimental measurement. It further seemed reasonable to assume that the small amount of void structure of cellulose or wood inaccessible to benzene or toluene would be filled by the first few per cent of water added, so that density measurements in these liquids of cellulose or wood containing appreciable amounts of sorbed water should give results free from the error introduced by incomplete penetration. Only in the case of dry or relatively dry cellulose or wood should an error be introduced by making density measurements in a non-swelling liquid. For these reasons the density measurements of cellulose and wood containing sorbed water were made in benzene, thus greatly simplifying the measurements and increasing their accuracy. Benzene was chosen in preference to toluene for the measurements because it caused no measurable swelling of wood, whereas toluene caused a very slight swelling. Although water is very slightly soluble in benzene, only in the case where the moisture present exceeds the fiber-saturation
136
ALFRED J. STAMM AND R. M. SEBORG
point of cellulose or wood could the benzene become saturated because of the vapor pressure depression caused by the cellulose or wood. The error caused by using dry benzene for the measurements should hence be small and will be shown later to be entirely negligible. The benzene used was a chemically pure benzene, thoroughly dried over phosphorus pentoxide, and then distilled. The adsorbent materials used were Sitka spruce and white spruce heartwood sawdust of 40 to 60 mesh, from which the extractives had been removed by extraction with a mixture of benzene and alcohol, cotton linters alpha-cellulose, a normal spruce sulfite pulp, and the same pulp beaten for 20 hours to a "highly hydrated" condition. Beaten pulp when dried by ordinary methods becomes hard and horny. The water was hence replaced by alcohol and the alcohol, in turn, replaced by ether and then the pulp dried from this latter solvent. By drying in this way practically normal porous texture is obtained, and the sealing up of voids is prevented. The measurements were made in 50-cc. and 100-cc. pycnometer flasks brought to thermal equilibrium in a thermostatic water bath held a t 30°C. f0.02"C. The volume of the pycnometers was obtained from their weight filled with water, and the density of the benzene was determined. The pycnometers were then half filled with cellulose or wood (3 to 10 g.) that had been brought to moisture equilibrium in rooms controlled at 97, 90, 75, 65, and 30 per cent relative humidity and 80°F., or partially dried over calcium chloride, weighed, and covered with anhydrous benzene. The air was removed from the submerged fibrous materials by carefully applying a vacuum and releasing until no sign of air bubbles was obtained up to a vacuum which caused boiling of the cold benzene. Experience showed that the cellulose and the wood could be completely freed from air after a few hours treatment. After bringing the pycnometers containing the adsorbent, water and benzene to thermal equilibrium and volume in the thermostatic bath, they were rapidly weighed. The weights in general could be checked to 0.5 mg. The excess of benzene was then decanted off, and the materials dried a t 105°C. to constant weight. The compression of the water was calculated with the aid of the following equation: ww
Wb - (W - md mo
in which C is the decrease in volume due to compression in cubic centimeters per gram of dry adsorbent material, d, the normal density of the water, d b the density of the benzene, and D the true density of the adsorbent material, ww the weight of sorbed water, ma the weight of the dry adsorbent material and mithe weight of the wet adsorbent material, Wb the weight
137
ADSORPTION COMPRESSION ON CELLULOSE AND WOOD *
of the pycnometer filled with benzene alone, and W the weight of the pycnometer filled with wet adsorbent material and benzene. I n figure 1 are plotted the apparent adsorption compressions for several cellulosic materials calculated on the basis of the density of the dry material as determined in benzene being the true density. The average densities of each of the materials determined in benzene are Sitka spruce 1.449, white spruce 1.459, sulfite pulp 1.529, beaten sulfite pulp 1.543, and cotton 1.549. The integral curves give the relationship between the total compression occurring in the water adsorbed on a gram of cellulosic material and the moisture content. The differential curves obtained by graphically determining the slope of the integral curves give the compression per gram of 0.050
~~
