Plasticizers in Vinvl Chloride Resins J
MIGRATION OF PLASTICIZER H. M. QUACKENBOS, JR. Bakelite Co., a Division of Union of Carbide and Carbon Corp., New York 17, N . Y .
I
S PRACTICE, a plasticizer must stay in place; otherwise, the plasticized compound stiffens and becomes less desirable. The permanence of a plasticized vinyl resin is usually established in terms of the weight lost when samples are exposed to various conditions in use-warm air, water, oil, etc. With improving technology, the variety of test conditions has become wider and testing programs tend to be involved and expensive. Testing might be simplified and made more efficient if the mechanism of migration could be established. If the relation between the behavior of a plasticized resin and the properties of the plasticizer itself were direct and simple, testing time and expense could be saved. The most fruitful approach t o the problem of migration has been to apply the principles that underlie the transfer of matter; these principles have been used successfully in analyzing the drying of wood and clay. The migrating plasticizer has been considered to meet two resistances: in moving from the interior of the vinyl sheet to the surface, and in escaping from the surface into the surrounding medium. The idea of two resistances has been used in a small way in considering loss from a vinyl sheet into air (IO)and into high vacuum ( 5 ) . I n this paper, the principles have been applied to a wider range of media and many of the factors that control migration have been revealed.
TABLE I. PLASTICIZERS USED Plasticizer Chemical Designation Composition R-2H Polyester
'rws
Thio ester
8N8
2,2'-(2-Ethylhexamido) diethyl di(2ethylhexoate) Polyethylene glycol di(2-ethylhexoate) Di- @-ethylhexyl) phthalate
4G0 DOP 3G0
Triethylene glycol di2-ethylhexoate
TOE'
Tri- (2-ethylhexyl) phosphate
CC-55
Dioctyl ester of dicarboxylic aaid
A-26
Di-2-ethylhexyl adipate
DHP
Di(n-hexyl) phthalate
HRE
Aromatic hydrocarbon
P a r t s b y Weight
2GB
Diethylene glycol dibenzoate
100 3 1 40 to 82
140
Cresyl diphenyl phosphate
DBP 160
Dibutyl phthalate Butyl benzyl phthalate
HB-40
Partially hydrogenated isomeric terphenyls Aromatic hydrocarbon
MATERIALS
Srinyl compounds were made by blending resin, plasticizer, stabilizer, and lubricant. The following formulation, with slight variations in the type and amount of stabilizer and lubricant, covers all the compounds.
Trade Name Supplier Flexol Carbide a n d Carbon Plasticizer Chemicals Co. a DiR-2H vision of Union carbide a n d Carbon Corp., 30 East 42nd St., Kew York 17, N. Y. Flexol Carbide a n d Carbon Plasticizer Chemicals Co. TWS Flexol Carbide a n d Carbon Plasticizer Chemicals Co. 8N8 Flexol Carbide a n d Carbon Plasticizer Chemicals Co. 4G0 Flexol Carbide and Carbon Plasticizer Chemicals Co. DOP Flexol Carbide and Carbon Plasticizer Chemicals Co. 3G0 Flex01 Carbide a n d Carbon Plasticizer Chemicals Co. TOF Flexol Carbide and Carbon Plasticizer Chemicals Co. CC-55 Flexol Carbide, a n d Carbon Plasticizer Chemicals Co. A-26 Flexol Carbide and Carbon Plasticizer Chemicals Co. DHP Flexol Carbide a n d Carbon Plasticizer Chemicals Co.
HRE
Ingredient Vinyl resin (copolymer of 95% vinyl chloride, 5% vinyl acetate) Basic lead silicate Calcium stearate Plasticizer
The plasticizer content is expressed here in parts by weight, per hundred parts of resin (phr). The various plasticizers used, together with their chemical compositions and characteristics of vapor pressure, are given in Tables I and 11. The blend of ingredients was fluxed on a two-roll mill a t 150' C., and then calendered on a small four-roll calender a t 150" to 160' C. to give a thickness of either 0.004 or 0.020 inch. Samples for testing were cut from these films or sheets. Properties were established for samples of the same plasticizer used in the formulation. The mechanical properties of the film a t room temperature vary greatly with the percentage of plasticizer present, and depend to some extent on the particular plasticizer used. Table I11 gives sample values for vinyl resin containing two amounts of dioctyl phthalate.
S/V-c
Flexol Plasticizer 2GB Santicizer 140
Carbide a n d Carbon Chemicals Co. Monsanto Chemical Co. 1700 South 2nd St.,'St. Louis 4, 110.
... Santicizer 160 Plasticizer HB-40 Sovaloid C
Monsanto
co.
Chemical
Monsanto Chemical co. Socony-Vacuum Oil Co., Inc., 126 B oadway, Xew York 4,
N. Y.
KP-504
A R C D
E
Bis(dimethylbenzy1) carbonate Dioctyl sebacate Blkylated aromatic by. drocarbon Phthalate of mixed octsnols Adipate of mixed aliphatic alcohols Polyester
Plasticizer KP-504
Ohio Apex, Inc., Nitro W . Va.
HOW PLASTICIZER MIGRATES
If the moving plasticizer meets two resistances, the rate of travel (or rate of loss from the vinyl film) depends upon their sum. Similarly, for a given voltage, the flow of current through two electrical resistances in series depends on their sum. However, to make any sort of analysis, it is necessary to measure the individual resistances in order to determine the contribution
each makes to the sum. For example, if a vinyl sheet is exposed to warm air, the resistance of the sheet itself and the resistance of the stagnant film of air in contact with the sheet must be measured. Some idea of the effect of this second resistance alone is given
1335
INDUSTRIAL AND ENGINEERING CHEMISTRY
1336
TABLE 11. CHARACTERISTICS OF PLASTICIZERS Vapor Pressure D a t a a Designation R2H TRS 8N8 4G0 140 DOP 3G0 TOF cc-55 A-26 DHP HRE DBP
s/v-c HB-40
A
BY-1
X m . H g X 10-3 a t 9 8 O C. Very low
hIol. Wt. High 500 b 484 447 340 390 403 435
Latent heat, cal./g. mole
B
27; 800 21,700 22,800 24,500 24,200 21 ,700 23,800 23,500 23,600 22,400 21,800 20,300 23,200 20,800 28,100 19,500
12:33 8.97 10.08 11.20 11.25 10.04 11.27 11 .I1 11.26 10.64 10.61 10.43 12.20 10.74 12.90 9.99
0.086
0.162 0.4800 0.740 0.960 1,750 1.800 2.100 2.460 3.100 6.300 30.000 36. O O O C 42.000 0,230 34.000
3ii 334 250b
278 3006 2308 426 400b
Viscosity, Cps a t 20' C. 21,000 64 139 25 45 82 16 14 30 11 30 2500
..
Vol. 46, No. 6
cise. Considering the similarity of the two curves and the a g r e e ment between the 63 and the 49% by volume of plasticizer, it appears that the resistance of the vinyl sheet is not a factor-Le., it is so low compared with the resistance to evaporation that its effect can be disregarded.
/
I
I
1
45
113 20 152
Vapor pressure, P, is related t o absolute temperature, T ,b y equation: log P = B - . L / 4 . 5 8 T , where L is latent heat of vaporization in calories per gram mole, and B is a constant. b Approximate values. C As these liquids are mixtures, only a n average value can be given. 0
~
0
TABLE111. RIECHANICAL PROPERTIES O F PLASTICIZED S'INYL FILMS COXTAIKING DI-~-OCTY PHTHALATE L Property Tensile strengJh, lb./aq. inch Ultimate tensile elongation, % Tensile elongation a t 1000 lb./sq. inch stress, % Hardness, Durometer A Density, grams per cc. a
47 P H R Q 3740 210 24 87 1 26
in Figure 1. Loss of plasticizer into air from a piece of glass cloth wetted with the plasticizer is slow and the loss from a vinyl sheet is even slower. One possible explanation is that the moving plasticizer meets the additional resistance of the vinyl sheet. The wetted cloth presents to air a surface t h a t is 100% liquid; the vinyl film, on the other hand, presents a surface that is 4970 liquid, as it contains 38% of plasticizer A-26 by weight or 49% by volume. 0.001
N
'5
: W
d 9 I-
s
c 1
TIME,
80
HOURS
4
Loss of Plasticizer into Air
At 58' C. and under vacuum of 2 X 10-4 mm. of mercury from glass cloth wetted with A-26 and from vinyl sheet 0.018 inch thick containing 65 parts of A-26 per 100 parts of resin. Area of both sides used i n calculating loss i n grams per square centimeter
99 61 1.21
Parts per hundred of resin.
400
Figure 2.
70 PHRa 2250 305
2 TIME, HOURS
Further evidence about the reeistance of the vinyl film l e given by repeating the experiment of Figure 1 under a piessure of l o M 4mm. of mercury (illustrated in Figure 2 ) . The rate of IOES for the wetted cloth rises about 1000-fold. However, the vinyl sheet does not show the same increase and does not follow along the dotted line in Figure 2. For a short time it almost keeps up with the dotted line, but later it falls more and more behind. Evidently, the resistance of the vinyl sheet is coming into plav, and indeed, this resistance becomes the more important. This is illustrated by replotting the data of Figure 2 in the form of plas+icizer loss versus the square root of time, as in Figure 3. ilfter a period of initial uncertainty, the line becomes straight. Such linearity is typical of loss controlled by movement or diffusion through the solid. The resistance of the vinyl sheet is such that it permits a loss of 970 in 3.5 hours (Figure 2 ) under high vacuum A more rapid loss is not possible, even though removal from the surface might be much higher, as ahown by the dotted line. At atmospheric pressure, the loss is very slow (Figure l ) , amounting to about 3% in 800 hours. The vinyl resistance would allon7 a much higher loss, but the loss is governed by the high resistance to evaporation into the medium. Between the two extremes of Figures 1 and 2, it is conceivable that the pressure might be adjusted so that both resistances Tvere important, rather than only one. However, under all practical conditions loss into oil, water, alcohol, etc., seems t o fall into one of the extreme classes in which one resistance alone is controlling.
Figure 1. Loss of Plasticizer into Air
VINYL RESISTANCE CONTROLLING
A t 58" C. from glass cloth wetted with A-26 and from vinyl sheet 0.018 inch thick containing 65 parts of A-26 per 100 parts of resin. Area of both sides used i n calculating loss i n grams per square ccntimeter
When the resistance of the vinyl sheet is controlling (or diffusion is controlling), loss is proportional to the square root of time (Figure 3), according t o Equation 1 ( I ) .
x
Q
I t is significant that the curves for wetted cloth and vinyl sheet in Figure 1 are very similar. There is a n initial rapid loss followed by a linear period The initial period in both curves can be assumed t o represent the loss of a inore volatile ingredient. After that, the linear phase reflects evaporation from a pure liquid. The ratio of the b o slopes in Figure 1 is 63%-i.e., the vinyl sheet is 63% wetted. This percentage, of course, is not too pre-
where Q
= 2.26
dg
grams lost in time t , hours
S = total grams of plasticizer in film
D
= diffusion constant, sq. em. per hour L = thickness of fiIm, cm.
100
x
Q/S represents the per cent loss
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1954
EXAMPLE CALCULATIOK USINGFIGURE 3. Q = 9 grams when = 1.83 hours (or actually 1.38 hours, allowing for the intercept of 0.45). S = 38 grams, L =.0.0180 inch or 0.0457 em.
Substitution of these values in Equation 1 gives a value for D of 1.67 x 10-6 sq. em. per hour.
0
I
2
TIME,V"EX Figure 3. Plasticizer Loss per Square Root of Time From vinyl sheet 0.018 inch thick containing 63 parts of A-26 per 100 parts of resin
The relation with the squaye root of time is a necessary, but not exclusive, condition for diffusion. This has been established by varying the vacuum in a n experiment like the one illustrated to l o w 3 in Figure 2. When the pressure was changed from mm. of mercury, rate of loss became somewhat slower from the wetted cloth, but the rate of loss from the vinyl sheet remained the same. The latter loss was still proportional to the square root of time, and diffusion still controlled. At a pressure of 10-2 mm. of mercury, loss was even slower from the wetted cloth. For the vinyl sheet, loss was still proportional to the square root of time, but the apparent diffusion constant was several times smaller than the true value-Le., diffusion no longer was controlling. Loss does not continue indefinitely linear as the square root of time (Figure 3). I n theory, the line of Figure 3 should start falling away from the linear when half of the plasticizer has escaped (1). I n practice, it is usual to observe an earlier departure. B s a general rule for plasticizer contents between 25 and 40%, loss cannot be expected to continue linear in root time past 10 to
2 .a
3.0
4
3.2
1331
12%. However, this does not limit the usefulness of the dif-* fusion constant, because losses above 10% are not of interest here. A vinyl film that has lost 10% of its weight has changed so much in character that such a loss represents failure. This idea is used throughout this paper. DIFFUSIONCONSTANTAND PRACTICE. &4 vinyl product is rarely, if ever, exposed to high vacuum in practical use. However, rapid loss from the surface and, hence, diffusion may be possible under the following conditions and materials which a vinyl film may meet in practical use: (1) mineral oil, (2) 5% soap solution, and (3) rub-off. Unfortunately, there is no direct way of telling whether diffusion is controlling. One cannot use wetted cloth as with high vacuum. However, a n estimate of loss from the surface is possible with mineral oil. The diffusion constant for plasticizer moving through mineral oil can be calculated from the Stokes-Einstein equation. The rate of travel estimated using this calculated diffusion constant is high enough so that loss from the vinyl specimen is controlled by the rate a t which the plasticizer can diffuse through the plastic. Similar estimates for soap solution and absorbent powder are not possible because there are no established equations. With all three media, loss from a vinyl film is proportional to the square root of time. This suggests that control of loss by movement through the plastic is being approached. Apparent diffusion constants calculated for the three media in comparison with the results obtained by using high vacuum are shown in Figure 4. The values from the several sources are in good agreement, and demonstrate that diffusion may govern the loss for a variety of practical conditions. Further confirmation is given by the agreement between diffusion constants for high vacuum and mineral oil over a range of temperature for six plasticizers. Figure 5 illustrates this effect for one of the plasticizers. Results for rub-off, 5% soap solution, and mineral oil have been compared a t one temperature (50" C.) for numerous plasticizers, and the agreement is generally good. Sometimes the result for 5% soap solution is low, which suggests limited plasticizer solubility. A liquid may enter a vinyl compound and modify its properties, or cause a gain in weight which falsifies the observed loss. The agreement among diffusion constants stated above suggests that these effects are small with 5% soap solution when the thickness of the vinyl compound is 0.004 inch, and with mineral oil when the thicknesses are 0.004 and 0.020 inch. Measurement of the density of several compounds after immersion in mineral oil showed that any swelling was insignificant. The
3.4
x 1000
Figure 4. Diffusion Constants for Oil, Soapy Water, and Dry Powder Compared with those obtained using high vacuum. Compound contained 47 parts of DOP per 100 parts of resin and was 0.004 inch thick
& x 1000 Figure 5 . Diffusion Constants for Mineral Oil and High Vacuum over Range of Temperature Compound contained 65 parts of A-26 per 100 parts of resin and was 0.018 inch thick
I t\: D U S T R I A L A N D E N G IN E E R I N G C H E M I S T R Y
1338
2.8
26
,
27
30
$7X I 0 0 0
28
I
Vol. 46. No. 6
29
30
x 1000
Figure 6 . Diffusion Coiistanes for ,PHinera1 for Variety of PlasticiAer Systems
Figure 7. Effect of BlasticiLer Content on Diffusion Cora.itaiit Determined by High
411 compounds tested contained 5 i to 65 parts of plasticizer per 100 parts Qfresin and thiclmess of each sanip!e was about 0.020 inch
Figure, indicate parts of DQF per 100 parts of resin in y i n v i rornpound 0.001 inch tlir \ r P O L 3 D S A N D FROM
GLASSCLOTHWETTEDWITH PLASTICIZER Samples 2 X 2 inch parallel t o air steam Air velocity 43 feet per minute
Ratp froin ~i~~~~ ~Per Cent Intercept& Cloth 4 mil 20 mil
~
~
l - i n y l , To of T h a t loin Cloth
~
Plasticizer PHR Hr.-1 X 10-6 4 mil 20 mil 0.5 24.8 5G 80 59 1.8 0.9 TOF 27.2 62 80 0 0 A-26 40 0.2 0 DOP 57 0.3 2 5 71 83 0.1 I .o GO 77 65 0.5 0.7 0.6 140 a Intercept expressed as per cent of plasticizer present, as in Table IV. b Rate calculated using area of both sides ( 8 square inches).
There is a close similarity among the three system-wetted cloth, 4 m i l compound, and 20-mil compound. An initially rapid loss is followxl by a linear period. The rapid loss probablq represents the escape of a more volatile ingredient of the plasticizer, while the linear period is due to t,he evaporation of plasticizer from a pure liquid. The vinyl materials behave as surfaces partially wetted with plasticizer. I n Table VI the intercepts for the wetted cloth arid for the two thicknesses of vinyl samples are in fair agreement,, considering
Vol. 46, No. 6
their intrinsic lack of precision. With plasticizers D O P and '4-26, there is moderate agreement between the intercepts of Table VI and those of Table V (for moving water), which are very unprecise. Generally, the four plasticizers of Table VI contain only about 1% of volatile impurity. The rat,es for the linear period of Table VI have some unexpected features. With a plastic 0.004 inch thick the rate averages 65% of the rate for wetted cloth. The average value at a thiekness of 0.020 inch is SOYo. These are distinctly higher than the range of volume per cent of plasticizer present, (35 to 49%), which might be expected to agree v i t h the ratio of slopes. A further ananioly is t,hat the rate is not quite independent of thickness, as expected for a process controlled by the surface resistance. However, these unexpected features are apparently only minor f l a w in a st,rong argument that the plastic behaves as a surface partially wetted with plasticizer. For exposure a t 58" C., the intercepts have values similar t o those in Table VI, and the comparative rates average 48 and 58% for thicknesses of 0.004 and 0.020 inch, respectively. These comparative rates are in moderate agreement with the volume percentages of plasticizer. As shonn in Figure 11, the plasticizer loss continued linear in time until it exceeded 10% of the weight of the sample 0.004 inch thick and about 7% of the weight of the sample 0.020 inch thick. Samples 0.004 inch thick containing other plast,icizers were also exposed to moving air at) 98" C. An intercept less than 1% of the weight of the plasticizer and a linear region tailing off for a loss exceeding 10% of the sample were observed for the follon-ing plasticizers: TITS, 8x8, CC-5jl D H P , DBP, a mixture of R-2H and DOP, and a mixture of DOP and HRE. A slou-ly curving line (loss decreasing with time) was observed with plasticizer 4 G 0 and with DOP mixed with plasticizers BN-1, c, or H E 40. Probably 4GO, BN-1, C, and HB-40 are somewhat heterogeneous in character. TThile it is desirable to describe loss in terms of an intercept and a rate, it is more practical and simpler to express the result as a "life"-i.e., time necessary to lose 10% of its original weight. For routine purposes, an excessive number of observations are needed t o establish rate and intercept, and thus extrapolate to life. I t is simpler to run the test until the sample loses 4 to 10% of its n-eight,, and then estimate life by proportion from the t,ime of test. The intercept is small and the rate is ell defined, even with the heterogeneous plasticizers like 4GO. Thus, the result is fairly accurate. The analogy between loss into air from a vinyl sample and loss by evaporation suggests a consideration of the empirical equation ( 6 )that describes the evaporation of a liquid into air. The first question is whether the equation is reliable and gives calculated values of evaporation rates that agree with those experimenhlly ohserved. The equat>ioncan be written as follows when air flow is streamlined (as in ~this case): ~ ~ , ~
where E =
K
=
Pi
=
p
=
;If*
= = = =
L: L T
evaporation rat,e, gram c m . 7 h - 1 factor involving molecular volumes of air and of vapor of liquid saturated vapor pressure of liquid at, temperature being considered, atmospheres partial pressure of vapor of liquid in air approaching liquid aurfare (in the author's experiments, p = 0) molecular neight of liquid velocity of [Lii-, em. sec.-l lrngth of surface parallel to flow, em. absolute temperature, K.
This formula is not, dimensionally consistent, various conversion factors being combined in the 2.67. The factor I< i,3 independent of temperature and is defined by:
~
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1954
1341
V,, Vb = molecular volume of air and vapor respectively, cc. per gram-mole illa,f i f b = molecular weights of air and vapor. respectively For the evaporation of dioctyl phthalate in the oven at 58' C. the following hold:
L
7.6 cm. (sample 3 X 3 inches), I' = 10 em. sec.-l, 7' = 331OX., p = 0 P, = 1.8 X 10-5 mm. of mercury = 2.4 X atmosphere, V , = 29.9 = 517 (calculated from the atomic volumes of the eleT7b ments, 6), M a= 28.8, l l l b = 391. Hence K = 0.0135 and E = 0.15 X 10-8 gram em. - B hr.-I =
Reynolds number can be calculated as about 400 = =
I n the oven a t 98" C., L
li
T P,
5.1 em. (sample 2 X 2 inchcs) 21.5 cm. see.-'
371°K. 126 X 10-8atmosphere and the other values are the same as rat 58" C. Hence. E is 13.7 X 10-6 gram cm. --2 hr. -I. =
=
These calculated rates compare as follows with observed values hr.-1 The agreement between observed and calculated values is not good and a wide disagreement exists in the ratio of values, This disagreement in ratio means that the equation is not effective in predicting the result of changing temperature, velocity of air, and size of sample. These failures of the equation may arise because it is based on results for volatile liquids like water evaporating from fairly large surfaces. Conditions of nearly nonvolatile liquids and small areas are different.
for dioctyl phthalate, where values are in loGgrams
Observed A t 58" C . 2 At 98" C. 3. Ratio 2 to 1
Calculated 0.15 13.7 91
0.4G
1.
9.5 21
However, the equation may have some use in suggesting a correlation when as many variables as possible are fixed. For example, in the oven test a t 98' C . for various plasticizer systems, the only variables are the vapor pressure Pi, the molecular weight, Ma, and factor K . The maximum variation in K?/Jb was estimated from its values for two plasticizers a t the extremes of molecular weight: dibutyl phthalate and 8N8 with molecular weights of 278 and 484, respectively, and molecular volumes that can be calculated as 355 and 708, respectively, Values of
TABLE VII.
V A P O R PRESSURE OF PL.4STICIZERS .4SD
PLASTICIZED COMPOUNDS
Plasticizer TWS 8N8
A 140 DOP TOF
PHR GS fiR
70 65 57 59 40 70 70
98" c
Hours to Lose 10% a t 98' C from 0 004InchSampie 920 590 350
LIFE OF
Vapor Pressuie, h I m Hg X 10-3 0 086 0 162 0 23
140
0 74 0 96
88 38
1 80
A-26 34 2 46 CC-52 30 2 1 DHP 31 3 1 8.2 70 4G0 0 48 DBP 70 2.9 30 00 82 206 0 24b DOP-R-2H, 1 to 3 DOP-HRE, 1 t o 2 72 21.0 2 8: 61 6.0 DOP-HB-40,1 to 2 IS 6 DOP-B1C'-l, 1 t o 2 64 6.1 12 o c DOP-R/V-C, 1 to 2 F7 8.5 12 6C DOP-R-H, 1 to 1 72 112 0 48d a Test run until loss was 4 to l o % , and hours to lose 10% was found by proportion. b 1/4 X vapor pressure of DOP (vapor pressure of R2H considered zero). 1/3 X vapor pressure of second plasticizer 2/3 X vapor pressure of
n n_. p. d 1/2 X vapor preesure of DOP.
+
TIME FOR 10% LOSS, HOURS
Figure 12. Vapor Pressure of Plasticizer at 98" C. us. Loss from Vinyl Compound 0.004 Inch Thick in Oven at 98' C. Solid circles represent mixed plasticizers
KMb are then 4.3 for dibutyl phthalate and 5.7 for 8 x 8 and the range for all the plasticizers of Table I can be inferred to be 5.0 rt0.7. This variation is only &14% and is negligible compared with the range in vapor pressures. It follows that there should be a correlation between the vapor pregsure of the plasticizer and loss of plasticizer from a vinyl sample a t 98' C. The correlation appears in Table VI1 and Figure 12. The ordinates of Figure 12 are made logarithmic in order to encompass the wide range of variables. The correlation is excellent, except for 4GO. The correlation coefficient is 0.99 without 4G0, and 0.98 with it. The slope of the correlation line is close to 1.0, as expected. Systems of mixed plasticizers can be included by proportioning vapor pressure according to concentration (Table VII). A uord on the precision (95% confidence limits) of the two variables may be in order. The vapor pressure value is good to about &10%, For a given time in the oven a t 98" C. the loss in weight as an average of four determinations had a precision of about =k 10%. Vapor pressures were determined by the effusion method, with an important modification that allowed a great reduction in time. The plasticizers were evaluated in their original condition, with no attempt a t purification by redistillation. Before observations were taken, 3% or more of the plasticizer was volatilized, and the results hold for the bulk of the plasticizer (Table 11). The importance of vapor pressure has been known for some time. Reed (8) first suggested that plasticizers boiling under 200" C. a t 4 mm. of mercury would be too volatile for satisfactory use. Small (IO)undercovered a good deal of evidence that oven loss is controlled by surface resistance. For a few plasticizers, his correlation between loss and vapor pressure was poor, for reasons that his extremely brief paper does not explain. TESTS ON PLASTICIZER
Such properties of a plasticizer as solubility and vapor pressure are quickly determined. Tests on a plasticized vinyl compound, however, are much more lengthy. The ideal would be realized, then, if tests on the plasticizer itself could be used to predict the performance of the plasticized vinyl compound. The ideal is only partially achieved.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1342 Property of Plasticizer Viscosity Vapor pressure Solubility in vorious media
Prediction of Performance of Vinyl Compound Rough guide to diffusion constant Excellent index of oven loss Ii solubility is high, diffusion coiistaiit will govern loss, unless loss is accelerated by absorption of medium in vinyl compound. Ii solubility is lorn (a3 in water), loss will probably be 1077, but unpredictable
The properties of the plasticizer are not a conclusive index of performance of the plasticized compound. They can be valuable in shoving the synthesizer of plasticizers that it is desirable t o have high viscosity, low vapor pressure, and l o ~ vsolubilitl--. Furthermore, they can be used to eliminate useless wduaiioiifor example, if the vapor p r e ~ s u r eof the plastic can be ruled out.
Vol. 46, No. 6
matcrial in air at any temperature, In pirticular, n ested in room temperature. The line of Figure 1 2 has the cqu:ition Life = 0.0YQjvapor pressure
('4)
~ h e r life c is tho tiiiic in hours for a 1 of 10% when tlie thiclincss ii 0.004 inr!i. Thc vapor pressuw 25" c. n-as calculatcd from the data of T a l h 11: and dubstitiitcd in Eyuiition -1 to gi at 25" C. Thc rmults appear in T:ihlc 1-111. Somc of ilie cxtraorcharily long lives are probably misleading, hecause th(. materi:tl It-ould fail first Cor other reason:;, such :is deleterious off t. Tne chief \-slue of Tabl:: 1-111 is in shon-ing tlmt t>.:It room tcmpc~ratui'eis no piohlvni n-it11 ronini ikc dioctyl phthalate. E v c th:: ~ morc rolaiilc 111 likc A-20 :ire vi rj. good.
TESTS OS VINYL CO\IPQUND
The user of vinyl compouiids is like an architcci. The arcliitect is not interested in the load that will buclile a floor and cause a catastrophe; he n-ants to l i n o ~ rthe maximum safe load. Similarly, a loss in m i g h t of 20 t o 3053, from a vinyl compound is catastrophic. The user is concerned by a loss of 10% (or usually one quarter to one third of the plasticizer), when the niaterial is becoming useless. If the value of 10% could he accepted, D could be put on one basis and the result always expressed as a life-i.e., the t,inie t o lose 10%. The extrapolation from t,he actual test result of a certain loss in a certain time n-ould be made according to the square root of time, if diffusion were controlling, or in proportion to time, if surface resistance n-ere controlling. To nialre this extrapolation valid, t,he time of testing would have t o be adjusted to alloir a loss of 4 to 1070, for reasons already discussed. Such a system is much superior to the old method of reporting a, per cent loss from a certain time n-hich wns fiscd for, say, loss to water, but niiglit be different for air. h valid method of extrapolating the percentage to longer times n-as, of course, not k n o ~ - n ; usually extrapolation was by proportionality. Furthermore, t,he fixing of one time for all plasticizers in a given t,est meant that the result n-as occasionally nithout, mc.aning. l'lnsticizers have such a wide range of properties that, at one extremc, the loss vas low aiid uiiprecise (0 to 1%) and might be made up largely of an impurity not characteristic of the bulk of the plasticizer. At tlie other extreme, the loss was extremely high (20 to 30'%). With the new system, it is possiblc to specify several standard times for each teat, so that a time may be chosen for each plasticizer to allow a loss of 4 to 10%. DETAILS OF TESTS. Two points may be noted. The oven test is bett,er made a t 98" C. than at, say, 58' C., because the result is obtained much more rapidly; and it may be desired to establish diffusion constant by only one of the four methods (high vacuum, mineral oil, rub-off, and 5% soap). The vacuuni method is the most rigorous, because it is possible t o check that the surface loss is high enough for diffusion to control. Disadvantages are the high cost of the equipment and t,he limited range of temperat,ure, usually only above 60" C., for which the method is applicable. The restriction in temperature arises from the fact that diffusion constant varies slowly with temperature i n contrast with the steep variation of surface loss, vr-hich is connected with vapor pressure. The determination of diffusion constant from loss of weight in mineral oil is an acceptable, simple altcrnatire, provided the plasticizer is a t least moderately soluble. Effects at,tending t,he entry of oil int,o t,hevinyl compounds usually seem small. GEUERA LIZATION S
For the correlation between vapor pressure and oven loss a t
98" C., in Figure 12, it is possible to predict the life of a vinyl
Plasticizer
The predictcd lives of Tablc 1-111 are, o l c o u i ~ oapproximate ~ 1~cc:tuscthere mag- be error in extrapolating the data for vapor pressure determined in the r a n g of G O 3 t o i20" C . dorrii t o 25' C. and air fion and size of snniplc affcct rate o l lo
