[CONTRIBUTION FROM
THE
INSTITUTE OF TECHNOLOGY RESEARCH LABORATORY MASSACHUSETTS OF ORGANIC CHEMISTRY, No. 2891
A New Interpretation of the Cellulose-Water Adsorption Isotherm and Data Concerning the Effect of Swelling and Drying on the Colloidal Surface of CeUulose1e2 BY A. G. ASSAF,R. H.HAASAND C. B. PURVES
The great afiinity of dry cellulose for water is lites and that the absorption is a function of the today attributed to the hydrogen bonding amorphous portion. It is well known1o that capacitya of the alcoholic groups because pro- cellulose acquires an equilibrium moisture congressive masking of the latter by esterification tent when kept for a sufficiently long time in a progressively reduces the amount of moisture humid atmosphere and that the adsorption isotherm is of the sigmoid type illustrated in Fig. 1, absorbed from standaxd, humid s~rroundings.~J~~ Although immersion in heavy water replaces plots 1 and 11. Flexed curves are also obtained the hydrogen in all alcoholic groups throughout by plotting the absorbed moisture, or the relathe mass of cellulose by deuterium: X-rays and tive humidity, against other physical properties optical dataQ strongly suggest that the water such as the rigidity,lla the electrical conductivmolecule as a whole fails to penetrate the crystal- ity,6912capacitance and power factor.la Moreover, similar relationships exist between the amount absorbed and the specific heat14 and apparent specific volume16of the absorbed water, the heats of wetting by water, aqueous alkali and methanol,12-16the selective absorption of water, or of solute, from various sol~tions’~ and the absorption of gases1SJQ and alcohols.6bJo-21 fill the results are consistent with the view that the first avid adsorption of moisture (to region B, Fig. 1) is caused by the strong hydrogen bonding of single molecules to each of all the alcoholic groups accessible to water in the amorphous portion of the cellulose. The intermediate, nearly linear portions of plots I and 11 (region B t o region D) correspond either to the formation of multilayers that are held with diminishing tenacitya or to the filling up of the finer c a p i l l a r i e ~ . ’ ~ ~ ~ ~ There is agreement to the effect that the steep, final portion of the adsorption isotherm (from region D) a t high humidities owes its shape to the swelling of the cellulose and to the condensation of “free” or “unbomd” water in the coarser 10 20 30 40 50 60 70 capillaries.2a On this interpretation, the mois% Relative humidity. Fig. 1.-Moisture adsorption isotherms: Cellulose I at ture content at B provides a measure of the 2 5 O , 0 ; Cellulose I1 at 20°, 0;Cellulose I1 at 2 5 O , x; amount of cellulose, or of the extent of its colPlot IV is Urquhart and Williims’1o data for unsmollen (10) Urquhart and Williams, J. T8.t. Inst., 16, T569 (1924). cotton at 80’. (1) Presented in part a t the February, 1943, meeting in Cumberland, Maryland, of the Western Maryland Section of the American Chemical Society. (2) Abstracted from Theses submitted to the Massachusetts In. stitute of Technology in partial fulfillment of the requirements for the degrees of Doctor of Philosophy (A. G. A.. May, 1942) and of Bachelor of Science (R. H. H., December, 1942). (3) Babbitt, Cdn. J . Research, %Ob,143 (1942), gave several literature references. See also the review by Barrer, J. Soc Dyers aiid Colourisfs, 67, 264 (1941). (4) Urquhart, J . Test. Insf., 20, T125 (1929) ( 5 ) Sheppard and Newsome (a) Ind. Eng. Chcm., 26, 285 (1934): (b) J. Phys. Chcm., 86, 2306 (1932). (6) Walker, (a) Teztilc Research, 7 , 229, 289 (1937), (b) J . T c x f . Inst., 24, T145 (1933); (c) J . APDlicd Phys., 8, 261 (1937). (7) Champetier and Viallard, ComDf.rend., 906, 1387 (1937). (8) Katz, Physik. 2.. !26, 321 (1924); Trans. Faraday Soc , 29, 279
(1933).
(9) The researches of NHgeli and of Ambronn were reviewed by Steinbrinck, Koltoid-Bciheftc, 93, 6 (1926-1927)
(11) Peirce, ibid., (a) 15, TdOl (1824); (b) M, T133 (1929). (12) Argue and M ~ a s sCan. , J . Rsscarch, 188,156 (1986). (13) Whitehead, “Impregnated Paper Insulation,” Wiley and Sons,New York,N. Y.,1935, pp. 13-36. (14) Shipley, Campbell and Maass, Can. J. Research. lTB, 40 (1939). (16) Maass and Campbell, PvIp PaPer Mag. Canada, 40, 108 (1939). (16) Morrison, Campbell and Maass, Can. J . Research, 16B,168 (1940). 117) Stamm and Hansen, J. Pkys. Cham., a, 209 (1938). (18) Grace and Maass, i b i d . , 86, 3046 (1932). (19) Stamm and Millett, i b i d . , 46, 43 (1941). A literature review is included. (20) Russell, Maass and Campbell, Can. J . Research. M B , 13 (1937). (21) Kanamaru and Chao, Kolloid-Z., 84, 85 (1938). (22) Stamm, “Colloid Chemistry of Cellulosic Materials,” U S . Dept. Agric. Misc. Publ., No. 240 (1936). (23) Marsh and Wood, “An Introduction t o the Chemistry of Cellulose,” Chapman and Hall, London, 2nd Ed., 1941, pp. 29-48 See also Bancroft and Calkin, Tcxfile Research, 4, 371 (1933-1934)
Jan., 1944
EFFECT OF SWELLING AND DRYING ON CELLULOSE SURFACES
loidal surface, that is accessible to water, and similar information can be obtained by studying the adsorption of othsr vapor^.'^ These views have the great merits of being intrinsically reasonable, of correlating a very large mass of information and of being not inconsistent with the work of Emmett and his collaborators on the adsorption of gases by inorganic catalysts.24 Nevertheless, the validity of applying this interpretation to adsorption by cellulose does not appear to be supported by direct, experimental proof. The present research originated in an attempt to remedy this deficiency. As explained in the preaeding article,26 the fraction of the cellulose alcoholic groups wetted by a noma1 ether can be determined by using a thallous ethylate solution and by subsequently methylating the thallium cellulosate produced. The relationship between the percentage wetted and the molecular volume of the normal ether is linear and extrapolation to molecular volumes of 18 and 3526gives the percentages theoretically available for adsorption of vapors of water and nitrogen molecules, respectively. The weights of adsorbed layers were calculated from these amounts by assuming that each available alcoholic group bonded one molecule and the data were compared with those corresponding to the B and D points on the adsorption isotherms of the same cellulose samples. Although alteration in the colloidal surface of the samples during the adsorption of nitrogen and ethers could be assumed to be negligible, many quantitative results c o n b e d the fact that the sorption of water produced changes of great magnitude in unswollen cellulose. This effect was minimized, but not eliminated, in the comparative experiments by employing samples that were already highly swollen.
Experimental Cellulose Samples.-The high grade cotton linters27 were dewaxed, swollen in caustic soda and dried through methanol and benzene with all the precautions previously described.% Sample I contained 1.47% of moisture when examined, but samples I1 and I11 were thoroughly dried over concentrated sulfuric acid at room temperature and atmospheric pressure. Estimation of Accessible Cellulose.-The thallations of approximately 0.3-g. samples of celluloses I, I1 and I11 were carried out with excess thallous ethylate dissolved in normal ethers exactly as already described, and themethylation of the thallium cellulosates was completed in the same way." The methoxyl contents in duplicate or triplicate estimations of cellulose I were 11.5, 10.0 and 9.7% in diethyl, 8.1, 8.1% in dibutyl and 5.5, 7.4 and 6.7% in diamyl ether. Cellulose I11 gave values of 10.7, 12.6, 10.9; 9.7, 9.3, 9.5; and 7.5, 7.5, 7.5% methoxyl, respectively, with thallous ethylate dissolved in the same three (24) Emmett and Brunauer, TEUSJOURNAL, BO, 1553 (1937). (25) AssPf, Haas and Purves, ibid., 66,59 (1944). (26) "International Critical Tables," McGraw-Hill Book Co., Inc.. New York, N . Y.,1926, give 0.808 g./cm.* aa the density of liquid nitrogen at 195.8'. This value was used in the calculation of the molecular volume. (27) The authors wish to thank Dr. H. M. Spurlin and the H u ules Powder Company, Hopewell, Virginia, for the gift of this materiol.
-
67
ethers. The average methylation corresponding t o each ether is plotted in Fig. 2 against the molecular volume of the ether and extrapolation t o volqmes of 18 and 35, respectively, gave the methylation theoretically corresponding to the penetration of the celluloses by water vapor and nitrogen gas. Ethyl ether values of 10.7, 11.6 and 11.1% methoxyl were the only ones available for cellulose 11, but the line drawn through this point is approximately correct, since the slopes of the plots for highly swollen celluloses do not vary greatly. The difference between samples I1 and 111, although obvious in Fig. 2, was in fact well within the experimental error. As previously explained,= multiplication of the methoxyl percentage by the factor 100/57.4 gave the percentage of cellulose, or of cellulose alcoholic groups, accessible to the penetrant (Fig. 2, Tables I and 11). The product of the latter percentages and the faetor 1.87 X I O 7 expressed the data as accessible colloidal surface in sq. cm. per g. of cellulose.
I 25
€k 20
d
E
x
15 0 V
c (
3 10
4t
2I
6 Et31
3 " L
.31
80 120 160 200 Molecular volume. Fig. 2.--Superlicial thallation-methylation of celluloses I, I1 and 111, with thallous ethylate in normal ethers. Molecular volume of ether plotted against %OCHs and % accessible hydroxyl groups in the celluloses.
40
Sorption of Water Vapor at 20 and 25 ".-The lower portions of twelve small desiccators, kept in a room whose temperature was constant a t 20, or 25O, t o within l o , were filled with aqueous sulfuric acid of various definitely known concentrations. Published tablesP8 gave the relative humidity within each desiccator at either temperature. The cellulosesample, 1.5 t o 1.7 g., held in an open weighing bottle, was kept in the desiccator containing concentrated sulfuric acid until the weight became constant, when the sample was assumed to be quite dry. It was then transferred to desiccators of increasing humidity, and daily weighings, quickly made on a magnetically damped balance, determined the moisture absorbed. The sample remained in each desiccator until weighings made on successive days did not differ by more than 1 mg. Three days at the lower humidities and five t o seven days at the higher ones were required for equilibrium t o be established within this weight limit. The best straight line was drawn through the observed points in the linear, intermediate portion of each adsorption isotherm (Figs. 1 and 4) and the moisture contents at the humidities where the plots commenced t o diverge perceptibly from the extrapolation of this line were observed (Table I).a4 At ordinary temperatures, these humidities were 7 to 10% for the B,and 45 to 50% for the D point. (28) Wilson, Ivd. Eng. Chcm., 18, 326 (1921).
The desorption isotherms were obtained by transferring the cellulose samples through the desiccators in reverse order, but the data are omitted from Fig. 1. Sample I had 26.7% moisture a t 100% relative humidity, when desorption was commenced, and the corresponding figure for sample I1 was 40.4% at 20'. The desorption of I11 and of I1 a t 25' was from equilibrium a t 56% relative humidity, or from just beyond the D point, when the moisture contents were 7.35 and 10.7%, respectively (Figs. 1 and 4). After desorption had been completed by standing over concentrated sulfuric acid, the standard estimation with thallous ethylate in ethyl ether gave methoxyl contents of 0.2, 0.2, 1.4 and 2.5y0 (latter in ethanol) for samples I , I1 at 20°, I1 at 25" and 111, respectively. A value of 0.5yo methoxyl (ethyl ether) was obtained whcn 2.06yomoisture \tilt remained in cellulose I.
Results and Discussion Suggested Interpretation of the Cellulose Moisture Adsorption Isotherm.-When swelling is neglected, the percentages of the cellulose samples I, 11 and 111 theoretically accessible to water vapor can be read from Fig. 2 and are recorded in Table I, column 2, as percentages of accessible alcoholic groups. If each of the latter is assumed to become hydrogen bonded to one water molecule, the calculated moisture content of the entire mass of cellulose is onethird of this percentage because fully hydrated H ~33.3% O , water (column cellulose, C O H ~ ~ O V ~has 3 ) . The B and D points of inflection in the moisture adsorption-relative humidity plots a t 20 or 25' (Figs. 1 and 4), are given in columns 5 and 6 of Table I. Comparison between columns 3 and TABLE I CORRELATION OF ACCESSIBLE ALCOHOLIC GROUPSWITH THE MOISTUREADSORPTION ISOTHERM OF CELLULOSE Accessiblea
OH HzOb groups, calcd.,
%
Cellulose 1 Cellulose I1
'
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7
1 -
08
Relative pressure, plpo. Pig. 3.--Nitrogen adsorption by an 0.3284 g. cellulose sample a t -195.8'. The observed adsorption, after correction for "dead space" and to N.T.P , is plotted against the relative vapor pressure
Adsorption of Nitrogen at -195.8". -The isotherms for celluloses I and I1 were determined by the method of Brunauer, Emmett and Teller*Q,mwith 0.2-g. samples. Helium was used t o estimate the "dead space" in the apparatus and the data for the adsorption of nitrogen were corrected accordingly. The preliminary evacuation was a t 80". If v is the volume adsorbed a t pressure p , and PO is the saturation pressure a t the same temperature,,,u the volume adsorbed at the iirst inflection of the isotherm (the B point) is given by the relationship _ _ P_ = _
.(A
- p)
BC
pc
Ratio
point, point, Temp.
%
%
D point
25" 2 . 1 6.9 0.30 25" 2.4 7.9 0.30' 20' 3 . 1 9.0 0.34) Cellulose I11 25.6 8.5 20" 2.3 7.4 0.31 Unswollen 20" 2.0 6.0 0.33 .. ... 1.9 5.8 0.33 cottond 25" Data in Fig. 2 extrapolated to molecular volume 18. * Calcd. by assuming that one water molecule was adsorbed by each accessible alcoholic group. From Figs. From Urquhart and Williams' data.1° 1 and 3 . 22.8 24.4
5 shows a t once that the water adsorption calculated from the thallous ethylate-methylation data is far higher than that at the B point. Since the accuracy of the calculated values is thought to be well within 1 1 5 % and lines
c-lp
1
v,c
lo 7.6 8.1
Water adsorbed
+
a,C
po
where C is a constant. Samples I and I1 had v, values of 2.34 cc and 2.83 cc , corrected t o N.T.P. These results correspond to adsorptions of 11.6 cc. and 14.0 cc. per gram of cellulose, or to unimolecular layer surfaces of 53.0 X lo4 sq. cm./g. and (2.6 X lo4 sq. cm./g., respectively, if 17 sq. A. is assumed for the area covered by a nitrogen molecule in the liquid condition. A complete isotherm for a different sample is reproduced (Fig. 3) t o show that the shape is sigmoid and that the above method of calculating area was probably legitimate when applied to cellulose. The area computed Tor this sample was 34.8 X IO* sq. cm./g. 129) Brunauer, Emmett and Teller, THISJOURNAL, 60,309 (1938). 1.30) We wish to thank Dr. W. R. Smith and Mr. F. S. Thornhill, uf the Godfrey Cabot Laboratories, Boston, Massachusetts, for their
great kindness in determining all of the nitrogen isotherms and in calculating the rr\iilt
