MEASUREMENT OF PARTICLE SIZE FOR XITROCELLULOSE DISPERSION* ____ BY HENRY B. DEVORE** ASD WHEELER P. D A V E T * * *
d water-spreading method of measuring the particle size of the disperse phase of a colloid has already been described by one of u8.l This method uses, in the measurement of the sizes of colloidal particles, the technique described by Langmui? for the measurement of molccular dimensions. A shallow rectangular tray filled with water is provided with an aluminum float extending the width of the tray with gaps a t either side. Glaqs strips 'orcc 'orce
\
Area-
Area
FIG.I
FIG.2
Srhematir Force-Area Curve
Force-Area Curve for Sitrocellulose
sliding on the tray edges and making contact with the water serve to limit the free surface of the water. A small quantity of dispersion is carefully dropped on the water surface upon which it, immediately spreads. The dispersing phase disappears either by solution or by evaporation, leaving a monoparticle film of the disperse phase on the water surface. Jets of air directed along the surface of the water at the ends of the aluminum float prevent the escape of the surface film through these gaps. The surface film is compressed by moving the glass strips toward the aluminum float and the force exerted by the film is measured by a torsion balance attached to the float. h typical force-area curve is shown diagrammatically in Fig. I . The vertical portion of this curve is obtained when the colloid particles are in * The experimental work recorded here was done at the Pennsylvania State College during the summer of 192;. It was supported hy E. I. duPont deNemours k Company, Inc. * * .4t present physicist; Experimental Station, E. I. duPont deNemours 6r Company, Inc., Wilmington, Delaware. * * * Professor of Physical Chemistry, School of Chemistry and Physics, Pennsylvania State College, State College, Pa. m.P. Davey: Srience, 64, 2 j 2 (1926) and Eighth Colloid Symposium (1930). Langmuir: J. Am. Chem. Soc., 39, 1881 (191 j ) .
2 130
HENRY B. DEVORE AND WHEELER P. DAVET
contact with one another and any further decrease in area requires sufficient additional force to pile the particles up two or more deep. It is the area corresponding t o this vertical portion of the curve (A) which is used in the calculation of particle size. I n the case of nitrocellulose on water the vertical part of the curve degenerates to a point at which the curvature of the graph changes sign. A representative curve obtained in this way for a nitrocellulose dispersion is shown in Fig. 2 . In the experiments described here a sample of nitrocotton was dispersed in ethyl acetate. A measured amount of the dispersion was spread on clean water and the ethyl acetate allowed to disappear into the air or into the water. For sufficiently dilute dispersion, the film left on the surface of the water may be assumed to be composed of separate particles of nitrocotton corresponding to the separate particles of the disperse phase. In these experiments we used a medium viscosity nitrocotton containing 12.4% nitrogen. However, since we were interested only in the method of measuring the particle size of the nitrocotton, its technical characteristics were of very little importance. It was sufficient for our purpose to make Time our dispersions always from the same sample and FIG.3 to make up all of the dispersions a t the same time. Time-Area Curve at Constant Force The concentrations used ranged from I X IO-^ gm. per cc. to 7.8 X IO-^ gm. per cc. The films had a tendency to shrink aft,er being spread on the water. This was probably due partly to the slow rate a t which the last traces of solvent were given up by the nitrocellulose. It is quite possible that a part of the shrinkage may be due t o a drying of the nitrocellulose in tiny patches or islands which orient themselves with time to give a closer and closer packing. Such a picture is consistent with the work of Zocher3 who found by optical examination that dried films of nitrocotton show areas whose thickness varies discontinuously after the fashion of soap films. For a given compressive force on the film the area decreased with time as shown in Fig. 3 . It was therefore an essential part of the technique that only the final values of curves like Fig. 3 be used in plotting the force-area curves. The points of inflection on the force-area curves from which the particle sizes were calculated were much more easily determined for the high concentrations than for low. This may be explained on the basis of larger particles since more force would be required to move them from the water surface and pile them up two deep. This would tend to make the determinations for the higher concentrations more accurate than for the lower. The most concentrated dispersion required ten drops (0.25 cc) for its determination and gave a monoparticle film 20 cm. long in a tray 2 3 cm. wide. The most dilute dispersion tried gave a film length of only 4.5 cm. when 1 2 0 drops ( 3 cc.) were used. At lower concentrations the quantity of dispersion was more than sufficient to cover the sur-
b
3
H. Zocher and F. Stiebel: Z. physik. Chem., 147A,401 (1930)
PARTICLE SIZE I N NITROCELLULOSE DISPERSION
2131
face of the water, so that part of the dispersion rested not on water but on a layer of still more of the dispersion. This had the effect of increasing the concentration before a monoparticle film could be obtained. This again leads to the conclusion that the measurements made with the concentrated dispersions were somewhat more reliable. The height of the nitrocotton particles was found to decrease with decreasing concentration from 3.69 X IO-^ cm. for the solution containing 0.001gm. per cc. to 1 . j 4 X 10-7 cm for the solution containing 0.000,0078 gm. per cc. When the heights of the particles were plotted against the logarithms of the concentrations all but three of the points were found to fit
FIG 4 Variation of Particle Height with Logarithm of Concentration a straight line within the limits of experimental error (Fig. 4). Two of these three points which were off the line were measured a t the same concentration. Their average falls almost exactly on the line. I t is obvious that the line of Fig. 4 can not be extrapolated below the height of a single “molecule” of nitrocellulose. At this height it must become horizontal. Although we have no direct evidence to indicate what this limiting height is, our measurements show that it is not in excess of 1.7 X 10-7 cni. It is presumable smaller, since a t this point the curve has not given any indication of becoming horizontal. It is interesting to compare this figure with the values given by Herzog 4. From X-ray diffraction studies he finds for the unit cell of cellulose nitrate the dimensions 10.1, 8.6, and 10.8 X 10-8 cm., and for cellulose 8.60, 7.78, and 1 0 . 2 2 X IO-^ cm. These values are so close to our lowest dimension that it seems reasonable to assume that they are not far from what we would have found if we had been able to use solutions sufficiently dilute to bring the curve of Fig. 4 to a horizontal. It is true that the values which we have measured represent the average size of the dispersed particles. It is also true however, that the particles of the disperse phase must be multiples of an ultimate particle size which can not be smaller than the chemical molecule itself. We feel justified, then, in believing that the ultimate particles in a nitrocellulose dispersion are not greatly different in size from the unit cell of cellulose or of nitrocellulose. R.0. Herzog: J. Phys. Chem., 30,457 (1926); Pulp Paper Mag., Can., 24,697 (1926).