704
INDUSTRIAL AND ENGINEERING CHEMISTRY
isopropyl alcohol and the nitromethane appear in the ternary azeotrope as 62 and 32 per cent, respectively, together with 6 per cent water. However, in the binary mixture, after the separation of the saturated ammonium nitrate solution, these components are present as approximately 66 per cent isopropyl alcohol and 34 per cent nitromethene. Since this amount of alcohol is less than the percentage present in the minimum-boiling isopropyl alcohol-nitromethane C. B. M., it is possible t o obtain pure nitromethane by batch distillation of the recovered binary mixture. The condensate from this distillation process will have the composition of the nitromethane-isopropyl alcohol azeotrope.
Vol. 34, No. 6
Bibliography (1) Baker, E. M., Chaddock, R. E., Lindaay, B. A., and Werner, R. C., IXD.ENG.CHEM.,31, 1263 (1939). (2) Baker, E. M.,Hubbard, R. O., Huguet, J. H., and Mickalowski, L. L.,Ibid., 31, 1280 (1939). (3) Carey, J. L., and Lewis, W. K., Ibid., 24, 882 (1932). (4) Carveth, H . R., J . Phys. Chem., 3, 193 (1899). (5) Fowler, A. R., and Hunt, H., IND. ENQ.CHEM.,33, 90 (1941). (6) Lebo, R.B., J . Am. Chem. SOC.,43, 1005 (1921). (7) Leoat, M., 2. anorg. Chem., 186, 119 (1930). (8) Othmer, D. F., IND.
[email protected],ANAL.ED., 4, 232 (1932). (9) Timmermans, Z . , Phys. Chem., 58, 29 (1907). (10) Washburn, E. W., in International Critical Tables, Vol. 111, pp. 33, 35, 219, 318 (1928).
Loss of Plasticizers from
Polvvinvl Chloride Plastics J
J
in Vacuum H. A. LIEBHAFSKY, A. L. MARSHALL, AND FRANK H. VERHOEIC‘ General Electric Company, Schenectady, N. Y.
PON being molded, polyvinyl chloride, an amorphous white powder, yields a hard, brittle, and therefore relatively useless, material. But if a “plasticizer” (usually a liquid) is properly added, a soft, flexible plastic is formed. How polyvinyl chloride and plasticizer interact to produce this change is not well known; the extent, conceivably even the type, of interaction may vary from case to case. A study of the way in which plasticizers leave a plastic can shed sonie light on the problem. Because the situation is complex, simple experimental conditions are highly desirable. Heating the plastics in sufficiently high vacuum ensures that no plasticizer molecule, having once evaporated, will ever return to the plastic surface; and that diffusion of the plasticizer through the plastic will eventually determine the rate a t which plasticizer is lost, which makes it possible t o obtain the corresponding diffusion constants from measurements of the rate of loss. Such measurements are reported for the plasticizers tricresyl phosphate, dibutyl phthalate, and dibenzyl sebacate. The results have been treated on the assumption that the surface concentration of plasticizer drops instantly to zero a t the beginning of each experiment and remains zero thereafter. This treatment is applicable until some 30 per cent of the plasticizer has been lost, whereupon complicating factors enter whose effect becomes more pronounced as the loss of plasticizer proceeds.
U
Preparation of Plastics Plastics are usually prepared by milling, which “strains”
the material so that marked dimension changes may occur on 1 Present
address, Ohio State University, Columbus, Ohio.
subsequent heating. Our plastics were prepared by the method of R. M. Fuoss of this laboratory, in which there is no milling and the composition of the produc1,s is accurately known. Weighed quantities of polyvinyl chloride and of plasticizer were thoroughly mixed by grinding for 15 minutes in a glass mortar after they had been stirred together for 5 minutes in a small beaker. The wet, muddy mixture that resulted was transferred to a tall-form beaker and heated for 6 minutes in a liquid bath near 90” C. During this heating, the material changes gradually to a dry, rubbery powder, and the risk of subsequent stray losses is thus decreased. After a second thorough grinding in the mortar, the mixture was transferred to a steam-jacketed positive mold, 4 inches in diameter, to which was attached a vacuum line for the removal of entrapped air. Heat and pressure were applied according to the following schedule: Compression was begun when the temperature of the mold had reached 110’ @.; the full pressure, 2500 pounds per square inch (176 kg. per sq. cm.), was reached 10 minutes later, a t which time the temperature had risen to 145’ C. Pressure and temperature were then maintained for 5 minutes, whereupon the mold was cooled and removed from the press. The molded disk and all scraps were carefully collected and weighed. The disk was finally aged for 1to 1.5 hours in an oven a t 104” C. Except for a slight peripheral shrinkage in the region that had been nearest the vacuum line attached to the mold, our plastics did not change dimensions during aging. The composition of one sample was verified by a chloride determination: found, 40.46 per cent polyvinyl chloride; calculated (with corrections), 40.39 per cent. These losses were allowed for in calculating the composition of an 11-gram sample: 20-30 mg. (presumably of the initial composition) left in the mortar; 10-100 mg. (plasticizer only) lost during molding; 20-30 mg. (plasticizer only) lost during aging.
June, 1942
INDUSTRIAL AND ENGINEERING CHEMISTRY
705
Experimental Method Rates of weight loss were measured a t several temperatures on the various plastic specimens suspended in vacuum from a calibrated quartz spiral. The principal experimental difficulty was the maintaining of the plastic a t thermostat temperature while the plasticizer, whose vapor pressure was sometimes below one micron, was being condensed at the temperature of liquid nitrogen. Quartz spirals of sensitivities near 0.05 gram per cm. were used; cathetometer readings were taken to 0.01 cm.; for an average experiment these values correspond to 1 / 5 ~of~ the plasticizer initially present. The apparatus in Figure 1 was evolved during preliminary work. The final design and dimensions give rapid pumping and efficient condensation; all parts of the sample see, along vacuum paths, glass surfaces in contact with oil a t thermostat temperatures. As a test of the apparatus, the rate of evaporation of pure plasticizer from a bucket in position X (Figure 1) was measured. The measured rate agreed with that calculated from the area of liquid exposed, and from the vapor pressure of the plasticizer a t thermostat temperature, and an accommodation coefficient of unity. This agreement proves that, for all practical purposes, the plasticizer was at thermostat temperature and that no plasticizer molecules were returning to it. Stefan’s law calculations have shown that this temperature condition should obtain even during experiments a t the highest temperatures. Specimen S, a disk 2.2 cm. in diameter cut out of the plastic prepared as described above, was suspended from a quartz spiral that was subsequently sealed into the apparatus (Figure 1). The apparatus was then connected to the vacuum system and surrounded by an oil thermostat. When temperature equilibrium had been reached, the apparatus was rapidly evacuated, and liquid nitrogen was poured into trap T, fixing the time t = 0. Cathetometer readings between k e d points on the spiral were started before t = 0 and continued at definite intervals to give the experimental data from which diffusion constants were computed. Readings of the vertical diameter of the sample were taken from time to time to measure the shrinkage accompanying the loss of plasticizer. The experiments were ended by admitting air to the system. For each specimen, initial and final measurements (six of each dimension) were madeof thickness and of vertical and horizontal diameters. T h e experiments could not be c o n t i n u e d until all the plasticizer hadevaporated. In several cases when the work was carried far enough, it could be shown that extrapolated rates of weight loss became vanishingly small as FIGURE 1. APPARATUS FOR M~ASURthe total weight ING Loss OF PLASTICIZER IN VACUUM
Many polymers require the addition of a plasticizer to give a useful product. In the hope of learning something about this interaction of polymer and plasticizer, the loss of plasticizer from polyvinyl chloride plastics was investigated for three plasticizers under the simplest possible conditions-namely, in high vacuum at accurately known temperatures. The results show that the diffusion processes involved are complicated by several factors. Nevertheless, the following reasonable picture of the interaction between polyvinyl chloride and plasticizer has evolved: Van der Waals forces between plasticizer and polyvinyl chloride molecules make plastic the polyvinyl chloride by separating from one another the chain molecules of which the latter is composed. As plasticizer is removed, the van der Waals forces among the chains bring them closer together; thus it becomes more difficult for the remaining plasticizer molecules to get out, and the volume additivity apparently characteristic of these plastics is conserved.
lost approached the weight of plasticizer initially present. Additional evidence follows to show that the weight loss was due principally to loss of plasticizer, I n several experiments with tricresyl phosphate plastics, the condensate in the liquid nitrogen trap was examined with these results: The weight of each condensate equaled the weight lost by the corresponding sample; the condensate had the refractive index reported for tricresyl phosphate; hydrolysis of the condensate produced the amount of phenol to be expected for tricresyl phosphate; chloride was absent or negligible. [The vapor pressure of hydrogen chloride (2) is high enough even at the boiling point of liquid nitrogen to cast some doubt on whether all the hydrogen chloride formed by pyrolysis was condensed in our trap.] Taken together these results show that the pyrolysis of polyvinyl chloride did not distort the measured weight losses during the early stages of our experiments, although darkening of the specimens showed that some pyrolysis was always occurring. To disclose its extent in an extreme case, a 0.5-gram disk containing 0.2 gram of plasticizer was heated in vacuum to a weight of 0.35 gram; according t o calculations based on the exact weight, its polyvinyl chloride content should have increased from 59.0 to 84.2 per cent if only plasticizer had been lost. Analysis of the heated specimen gave 82.5 per cent polyvinyl chloride, indicating some loss of hydrogen chloride formed by pyrolysis.
Results The results most readily interpreted are values of the dimensionless ratio E/co, in which co is the concentration of plasticizer in the specimen a t t = 0, and E is the average concentration a t t = t. The elongation proportional to co of the quartz spiral can be calculated from the initial elonga-
INDUSTRIAL A N D ENGINEERING CHEMISTRY
706
The construction of this scale and the approximations involved are described elsewhere (3). The reference line applies for the conditions D = 1, a = 2. If the slope of an experimental line differs from that of the reference line, these conditions are not both satisfied. D for such a line can be calculated a t any E/co by Equation 2A, t being the corresponding abscissa for the experimental plot ; 7,that for the reference line. The results are summarized in the Tables I and 11. For any experiment the tabulated D value applies until the critical value, ( Z / c ~ ) ~ ~ iof t . the , concentration ratio is reached; AV,,,,. is the volume shrin'hge calculated from the measured dimension changes; and the last column of each table gives the quotient of this decrement by A v o s i c d . , the shrinkage obtained by dividing the density of the plasticizer into the weight loss determined on an analytical balance.
13% ,774 ,724 ,681
,643 ,609 ,570 .549 ,522 .496 A72 ,449
,427 ,406
.37 ,3668 .350
I"
,333 .317
,302 .Le8 274 ,261 ,248
.e36 225 214
,204 .I94 .I84
0.1l50)
0.2(100)
0.31150)
0.41200)
0.6L3001
0.5(250)
0.71350)
T lt,lNMINUTES)
FIGURE 2.
REFERENCE LINEAND DETAILED EXPERIhfENTAL RESULTS
tion and the weight percentage of plasticizer; the former elongation diminished by the contraction of the spiral between t = 0 and t = t gives the elongation proportional t o E; Z/CO is then obtained as the quotient of two lengths. The individual E / c o values for two experiments are given in Figure 2. Detailed results for all experiments cannot be given. In general, the values lie on straight lines in plots like Figure 2 until some 30 per cent of the plasticizer has been lost; thereafter they lie above this line, and the divergence increases with increasing values of t. Occasionally initial irregularities similar to that in experiment 2, Figure 2, occurred; they arise because the transition in the apparatus from atmospheric pressure and thermostat temperature to high vacuum with the trap a t liquid nitrogen temperatures cannot be made instantaneously a t t = 0. These irregularities were ignored in drawing straight lines for the computation of diffusion constants. The fundamental Fourier equation as it relates to linear diffusion in the x direction of a substance a t volume concentration, c, is: &/at = Dd%/bx2
RESULTSFOR TRICRESYL PHOSPHATE TABLE I. EXPERIMENTAL PLASTICS A.
Temp,, Expt. Conon., C . No. wt.% 145 15 19.8 12 29.7 14 39.8 13 49.8 5 59.5 125 9 29.7 3 59.5
B. T:mp., C. 126
a
For a large slab a cm. thick (where a is the maximum value of x), initially of uniform concentration CO, whose two large plane faces are brought a t t = 0 to c = 0 and then maintained in this condition, Equation 1 gives the distribution function,
D?r2t/a2 = d / 4
(2.N
An average value, c/co, of the concentration ratio is obtained by integrating Equation 2 with respect to x and dividing by the thickness : B/CO
= (l/a)
(c/co)dx = (8/7r2)
(e-g
lo*
