effect is not unusual and has been observed with other components on temperature cycling. Summary
Pressure-sensitive ferrite-core transformers have been potted in various casting compounds and inductance values have been measured at room temperature and a t temperatures down to -76’ F. Pressures produced within these same compounds have been measured by embedding the bulbs of mercury thermometers which had previously been pressure-calibrated. Inductance change and pressure buildup as a result of potting, even with a rigid resin, are low or negligible for compounds gelled at room temperature, provided exotherm is low. O n postcuring for 2 hours at 185’ F., inductance changes and pressures of considerable magnitude are obtained with rigid and some semirigid casting compounds. Contrary to the fact that the addition of mineral filler reduces the thermal coefficient of expansion, the inductance loss and pressure buildu p are greater for filled rigid and semirigid epoxy compounds than for the corresponding unfilled compounds. The increased stiffness and rigidity of the filled compounds are responsible for these effects. As the temperature is reduced, pressure builds u p uniformly in the rigid resins but inductance change corrected for temperature effects on unpotted units remains relatively constant. The flexible casting compounds have little or no effect on inductance and develop no pressure a t room temperature. But inductance changes and pressure buildup occur with some of these materials at low temperature, and these changes are associated with low temperature stiffening and embrittlement. The semirigid epoxies which require an elevated temperature for gelling and curing seriously affect inductance at room
temperature but develop no pressure at room temperature on embedded thermometers. Pressure develops as the temperature is reduced but the filled compound behaves as the filled rigid epoxy and the corrected inductance does not decrease as the pressure builds up. The unfilled semirigid compound, however, shows correlation between pressure buildup and inductance loss. Ac knowledgmenl
The author thanks R. C. Harting and F. J. Kasper for supplying the transformers used in this work and for discussions held from time to time on this subject. literature Cited
Bush, A. J., Mod. Plastics 35,143 (February 1958). Clarke, TY. J., Bell Telephone Laboratories, private communication, March 1955. Dallimore, G., Stucki, F., Kasper, D., SPEJ 20, 544 (June 1964). Dewey, G. H., Outwater, J. O., Mod. Plastics 37, 142 (February 1960). Hanson, W. M., Tuzinski, J. R., Proceedings of Second Electrical Insulation Conference, 1959, AIEE Conference Paper 5067. Isleifson, R. E.,Swanson, F. D., Mod. Plastics 43, 123 (November 1965). Johnson, L. I.,Ryan, R. J., Proceedings of Sixth Electrical Insulation Conference, September 1965, IEEE Conference Paper 32C3-43. Rausch, J. M., Bell Telephone Laboratories, private communication, 1956. Samson, R. N., Lesnick, J. P., Mod. Plastics 35, 150 (February 1958). RECEIVED for review October 28, 1966 ACCEPTED March 28, 1967 Division of Organic Coatings and Plastic Chemistry, l52nd Meeting, ACS, New York, N. Y.,September 1966.
N EW FLAM E-RETARDA N T POLY ESTER SYSTEM BASED ON 2,3=BIS(ETHYLENE CARBOXY)-1,4,5,6,7,7-H EXACH LO RO BICYCLOC2.2.11-5-HEPTENE R. C .
S L A G E L , G . P. S H U L M A N , A N D F . M . Y O U N G T . L. Daniels Research Center, Archer Daniels Midland Go., Minneapolis, Minn. 55420
RIMARILY
because of strict building codes regarding the use
P of combustible materials in construction, there have been recent efforts (75) to make plastics flame-retardant. The most generally accepted method of flame-retarding polyesters in the trade today is by using highly halogenated dibasic acids and building the retardancy chemically into the structure of the resin. There are currently commercially available two chlorinated monomeric anhydrides based on hexachlorocyclopentadiene which are widely used to make flame-retardant polyesters: chlorendic anhydride (73, 74) (2,3-dicarboxy-1,4,5,6,7,7hexachlorobicyclo [2.2.1]-5-heptene anhydride produced by the Velsicol Chemical Corp. and the Hooker Chemical Corp.) (I), and a newer monomer, Chloran T M (Universal Oil Products Co.’s 2,3-dicarboxy-5,8-endomethylene-5,G,7,8,9,9-hexachloro100
l & E C PRODUCT RESEARCH A N D DEVELOPMENT
1,2,3,4,4a,5,8,8~-0ctahydronaphthalene anhydride) (11) (72). Because of the two carbonyl activated a-hydrogens, also CY to a carbon containing chlorine, on I, resins based on I have less than the desired light and thermal stabilitites and weatherability (70, 77). I t was postulated that if these activated a-hydrogens were removed or replaced by another less reactive species, greater thermal and light stabilities would result. Such was found (72) to be the case with 11. However, because
A new chlorinated dibasic acid, 2,3-bis(ethylene carboxy)-l,4,5,6,7,7-hexachlorobicyclo j2.2.1 1-5-heptene (Ill), was synthesized by oxidative ozonolysis of 1,10,1 1,12,13,13-hexachlorotricyclo [8.2.1 .O2l9]trideca5,l 1 -diene (IV), the mono Diels-Alder adduct of hexachlorocyclopentadiene and cis,cis-cyclooctadiene. A polyester resin system based on this dibasic acid was compared with similar systems based on Chloran TM (trade-mark of Universal Oil Products Co.'s 2,3-dicarboxy-5,8-endomethylene-5,6,7,8,9,9-hexachloro1,2,3,4,4a,5,8,8a-octahydronaphthaIene anhydride) (II), chlorendic anhydride (I), and tetrachlorophthalic anhydride (TCPA; V). The physical properties of the resin system based on Ill were generally superior to those of the corresponding resins based on clorendic anhydride and TCPA. The chief advantage of the new resin system over that based on Chloran TM is toughness or flexibility with retention of good strength properties, especially useful in some applications not requiring reinforced plastics, such as castings, pottings, coatings, and plasticizers.
of the rigid ring structures of both I and 11, their corresponding resins are rigid and brittle. For many types of castings, pottings, or coatings it is desirable to have a more flexible resin system to help prevent crazing, cracking, and brittleness, and still retain good strength properties ( 4 ) . Consequently, we have developed a new chlorinated dibasic acid which provides resins with more flexibility than I or 11, but with other physical properties retained or enhanced as compared to resins based on 11. The new dibasic acid, 2,3-bis(ethylene carboxy)1,4,5,6,7,7-hexachlorobicyclo[2.2.1]-5-heptene (111), is obtained from oxidative ozonolysis of l,lO,l 1712,13,13-hexachlorotricyclo [8.2.1.O2$9]trideca-5,1 1-diene (IV), conveniently prepared from hexachlorocyclopentadiene and cis,&-cyclooctadiene (7, 7). We have compared physical properties of this new resin system with similar resins based on I , 11, and tetrachlorophthalic anhydride (V, TCPA). While it may not seem desirable to compare the resins based on similar compounds I, 11, and 111 with a resin based on a chemically dissimilar com-
m
IP:
P
pound (V), it is significant that many of the desirable physical properties of the latter resin, such as high flexural and tensile strengths, are possessed by the new resin system. Experimental
All melting points are corrected. The elemental analyses were performed by Huffman Laboratories, Inc., Wheatridge, Colo. Ozone was generated from dry oxygen by a Model T-23 ozonator manufactured by the Welsbach Corp., Philadelphia, Pa. 2,S-Bis(ethy1ene carboxy)-1,4,5,6,7,7-hexachlorobicyclo[ 2.2.11-5-heptene (111). Ozone was bubbled through a mixture of 30 grams (0.079 mole) of the monoadduct ( I V ) ( 7 , 7) in 150 ml. of glacial acetic acid for 71 minutes at a rate of 67.2 mg. per minute (0.099 mole). A solution consisting of 15 ml. of 1M sulfuric acid and 15 ml. of 50% hydrogen peroxide was added to the reaction mixture and the total mixture was stirred overnight. The total mixture was then poured into 300 ml. of water containing 15 ml. of concentrated hydrochloric acid. A white sticky solid precipitated which was washed several times with water, dissolved in acetone, and then reprecipitated with 1S hydrochloric acid. After drying there remained 31.4 grams (89.4%) of white solid which gave a melting point of 225-6' C. when recrystallized four times in glacial acetic acid. pK, = 4.59. Acid number: calculated 252; found, 247. The infrared spectrum (halocarbon mull) shows a strong carbonyl absorption at 1705 cm.-', as well as the peak typical of this ring system at 1600 cm.-l
ANALYSIS.Calculated for C13H12C1~04: C, 35.09; H, 2.72; C1, 47.81. Found: C, 35.18; H, 2.75; C1, 47.64. The diethyl ester was obtained in 73yO yield by refluxing I11 in ethanolic benzene with a drop of concentrated sulfuric acid until \vater no longer azeotroped. Three recrystallizations from aqueous ethanol gave colorless crystals (m.p. 4849"). The infrared spectrum (melt liquid) shows carbonyl absorption at 1725 cm.? ANALYSIS.Calculated for C17H20C1604: C, 40.75; H, 3.99; Cl, 42.47. Found: C, 40.81; H, 3.31; C1, 42.69. The dimethyl ester was synthesized in nearly quantitative yield from the reaction of methanol \vith the diacid chloride (prepared from I11 and thionyl chloride). Four recrystallizations from aqueous ethanol gave colorless crystals (m.p. 56.557.5'). The infrared spectrum (halocarbon mull) shows carbonyl absorption at 1725 cm.-' AKALYSIS.Calculated for CljH1&1604: C, 38.10; H, 3.38; C1, 44.98. Found: C, 38.19; H, 3.48; C1, 44.64. The diamide was made in nearly quantitative yield by bubbling excess anhydrous ammonia into an ethereal solution of the diacid chloride. Recrystallization from chloroform gave colorless crystals showing two irreversible endotherms, melting first at 137', resolidifying, and melting again at 175". The TGA curve shows no weight loss over the temperature range of the endothermic transitions, indicating that the latter are due to a crystalline transition. The infrared spectrum (halocarbon mull) sho\vs N H absorption at 3430, 3340, and 3140 (broad) crn.-l, and carbonyl absorption at 1670 (sh) and 1650 cm.-' ASALYSIS. Calculated for C I S H ~ ~ C ~ ~ c, N ~ 35.24; O~: H. 3.19; C1, 48.01; N, 6.33. Found: C, 35.26; H, 3.38; C1, 47.61; N, 6.61. Synthesis of Resins. In a 2-liter three-necked roundbottomed flask equipped with a stirrer, condenser with a Dean Stark trap, thermowell, and nitrogen inlet was placed a mixture of 2.0 moles of maleic anhydride (1 .0 mole in the case of TCPA) and 4.4 moles of 1,2-propylene glycol. The mixture was heated until the maleic anhydride went into solution. At this point 0.65 gram of a 20% solution of hydroquinone in methyl Cellosolve was added, followed by 2.0 moles of the chlorinated acid or anhydride (3.0 moles in the case of TCPA). The temperature was then raised to and held at 190' to 195' C. until the acid number had reached the recorded value (Table I). The reaction mixture was then cooled to below 150' C. and 0.85 gram of a 20yGmonotertiary solution of butyl hydroquinone in methyl Cellosolve added, followed by 30yG of the total resin weight of styrene. .4fter thorough mixing the mixture was cooled to room temperature.
Table 1.
Property
Properties of Resins Prior to Cure Chloran Compound Chlorendic TM III TCPA Resin Resin Resin Resin
Cl based on starting materials 23.9 Brookfield viscosity, 855 CP . Gardner color 4 .4cid number 27.4 50.6 Hydroxyl number
VOL. 6
21.9
21.3
1390 4 27.4 43.1
865 6 25.8 24.6
NO. 2
23.5 635 4
17.7 6.9
JUNE 1967
101
Table 11.
Property
Cure Data and Physical Properties of Clear Casts of Resins Chlorendic Chloran T M Compound III Resin Resin Resin
SPIa cure, 180' F. Total time, sec. Gel time, sec. Peak exotherm, F. SPI cure, 250" F. Total time, sec. Gel time, sec. Peak exotherm. F. Barcol hardness Flexural strength, p s i . Flexural mod., p.s.i. Tensile strength, p.s.i. Elongation, yo Heat distortion, C. Flame retardancy, ASTM D 635-56T Extent of burning, inch a Society of the Plastics Industry.
540 405 410
580 400 391
147 102 469 56-5 8 9,900 663,000 4,500 0.9 114 Self-ext.
158 120 460 54-56 10,400 677,000 4,200 0.8 111 Self-ext.
0
652 345 382 162 120 459 37-39 15,500 550,000 8,900 1.9 82 Self-ext.
0
0
TCPA Resin
649 428 337 176 135 440 51-53 15,600 686,000 8,300 1.6 84 Burning
...
Table 111. Physical Properties of laminates (Three-ply woven glass cloth with 35% resin) Chlorendic Chloran T M Compound III
Property
Flexural strength, p.s.i. Flexural mod., p s i . . Tensile strength, p.s.1. Elongation, yc
Resin
Resin
Resin
22,800 1,651,000 33,700 1.51
44,000 2,809,000 31,000 1.49
31,800 2,347,000 32,100 1.50
T C P A Resin
12,900 3,940,000 47,400 2.1
Results and Discussion
Resin
Table IV. Solvent Attack" yo Weight 7 0 Change in Loss Flex Strength
7-Day Soak in 10% NaOH (25' C.) Chlorendic 0.46 f39.0 Chloran T M 0.25 -29.4 Compound I11 0.61 -5.5 TCPA 0.27 -1.2 7-Day Soak in 37y0 HC1 (25' C.) f3.4 Chlorendic 0.26 +13.8 Chloran T M 0.22 Compound I11 0.21 -15.3 TCPA 0.11 -3.3 7-Day Soak in Skellysolve B (25 C.) Chlorendic 0.34 +11.3 0.34 -23.8 Chloran TM 0.35 -0.7 Compound I11 TCPA 0.19 -4.1
7~ Change in Flex Modulus -
1.9 8.6 9.4 -11.3
-21.0 -16.5 -16.2 -19.5
O
Chlorendic Chloran T M Compound I11 TCPA
3-Hour Boil in Distilled Water 0.33 -34.3 0.30 4-8.7 0.31 -33.2 0.17 -18.1
-19.3 -21.6 -21.6 -20.6 -21 .o -18.6 -20.0 -20.8
3-Hour Boil in l o Yo NaOH 0.64 -27.9 -17.6 -20.4 0.62 -15.9 2.13 -65.3 -25.5 -24.3 0.45 -34.3 Samples dried overnight in oven at 110' C. before being weighed
Chlorendic Chloran T M Compound I11 TCPA and tested.
Clear casts of the resins were made by pouring the resin mixtures, with 1% benzoyl peroxide added, into a mold consisting of two glass plates separated by a l/s-inch deep rubber gasket in the shape desired and curing for 2 hours a t 66' C. and then 2 hours at 121' C. The resulting casts were cut into test samples. Test results are shown in Tables 11, IV, V, and VII. Laminates were made by pouring 35y0 by weight of the resins used for clear casts onto three plies of KO.181 glass cloth (A-1 100 finish) and then curing at 150' C. in a press at 20 p.s.i. for 5 minutes. The samples were then cut for testing. Test results are shown in Table 111. 102
l&EC PRODUCT RESEARCH A N D DEVELOPMEkT
Properties of Resins Prior to Cure. The four resins under comparison all had nearly the same chlorine content (within 2.6%) and acid number (within 1.6 units except for TCPA) and were all copolymerized with maleic anhydride and propylene glycol. Their properties are listed in Table I. Cure Data and Physical Properties of Clear Casts of Resin. The SPI cure data are comparable for all of the resins, as shown in Table 11. The first real difference in properties occurs in the physical properties of the cured clear casts (Table 11). Because compound I11 is a substituted straightchain dibasic acid (suberic) and the other three are more rigid cyclic acids (anhydrides), one would expect the resin system based on I11 to have a lower hardness value, greater degree of elongation, and lower heat distortion temperature ( 2 ) . The results are as expected. For a more flexible semirigid resin one would expect a higher flexural strength and a lower flexural modulus (between 300,000 and 600,000 p s i . ) than for a rigid polyester (3, 9 ) . Again the results agree with predictions. T h e one result which does not follow the expected pattern is the tensile strength. For less rigid resins one generally expects a decreased tensile strength (2). We were pleasantly surprised to find that the resin based on compound I11 has a higher tensile strength than the corresponding more rigid resins. All resins were self-extinguishing by the ASTM D 635-56T test, except the TCPA-based resin, which was rated burning. This is another evidence that flame retardancy depends not only on the total halogen content of a resin but, perhaps more importantly, on the strength of the carbon-halogen bond. I t is well known (76) that the bond dissociation energy of the aromatic C-C1 bond is greater than that of the aliphatic C-CI bond. I n the present case, then, it is probable that chlorine radicals are not produced at the ignition point of the TCPA resin (see thermal stability of the monomers and resins for other discussion on decomposition). I t is known that halogen radicals inhibit the propagation of flames. Other TCPA resins based o n different formulations, however, have been shown to be self-extinguishing (73).
Table V.
Fadeometer Exposure Time, Hours 0 50 100 200 300
Effect of Fadeometer Exposures on Transmission of lighta through Chlorinated Polyester Resins
Chlorendic Resin Re Rb 70.3 65.2 5.1 60.9 9.4 55.8 14.5 53.3 17.0 I
.
.
Chloran T M Resin Re Rb 70.1 ... 67.6 2.5 65.1 5.0 61.7 8.4 58.1 12.0
Compound 111 Resin Ra Rb 67.4 ... 61.5 5.9 58.4 8.7 57.0 10.4 53.7 13.7
Ra 64.6 44.5 40.0 34.7 30.8
T C P A Resin Rh
...
20.1 24.6 29.9 33.8
aObtained as rejectance values using a rejectance standard of 86.4 for 45’ angle on a Gardner automatic color difference meter, manufactured by Gardner Laboratory, Inc., Bethesda 14, M d . b R for 0 exposure time minus R for each of the other exposure times.
Physical Properties of Laminates. T h e results of making laminates \vith the four resins mentioned above are shown in Table 111. The strength properties do not follow the same pattern shown in the clear casts. Evidently the bonding of the resins to the glass fibers is sufficiently different to cause the different strengths. I t is probable that because of the long chain nature of I11 the polarity is less than that of the similar compound (111, causing a slight decrease in laminate strength, proportionally, over the clear casts. Chemical Resistance of Clear Casts. Table IV shows the weight change and change in flexural properties of the resins after they had been subjected to attack by various solvents. Unexpectedly, the resin based on compound I11 has the greatest susceptibility to attack by base. This difference in base stability, especially between the similar resins based on Chloran and compound 111, is puzzling. One possible explanation may be that the resin based on I11 is more open to base attack because I11 is an open chain compound whereas the other resins are based on cyclic compounds which should provide more hindrance about the carbonyl groups. Resins based on TCPA and chlorendic anhydride show the best resistance toivard Skellysolve B. This is not surprising, since one xvould expect the materials with less hydrocarbon nature to be less susceptible to attack by mixed hexanes (Skellysolve B). Resins based on TCPA and Chloran T M show the best stability toward boiling water. Light Stability of Resin. Light stability studies of the resins (Table V) show the stabilities to be in the order of Chloran > Compound I11 > chlorendic > TCPA. The much lower light stability of TCPA-based resins is to be expected because of the aromaticity of TCPA (5). Thermal Stability of Monomers a n d Resins. Roberts, Haigh, and Rathsack (72) have shown by differential thermal analysis that Chloran T M is considerably more heat stable than chlorendic anhydride. In a similar study we have found that compound I11 is also more heat stable than chlorendic anhydride and comparable to Chloran Thl, as would be expected from structural considerations. Compound I11 actually has a slo\ver rate of decomposition than Chloran T M u p to 270’ C. but a more rapid rate a t higher temperatures (Table VI). I n the case of TCPA one would expect decarboxylation a t a lower temperature both because it is a P-chloro acid (anhydride) (8) and because aromatic acids decarboxylate more readily than aliphatic acids. I n the previous report (72) it was mentioned that the relative thermal stabilities of the monomers appear to carry over to the corresponding polyesters. Such was the case in our work as well (Table VII). The reason for this may be that the polyesters undergo the pyrolytic cis elimination ( 6 ) to give an olefin and the acid; the acid may then undergo decomposition as described above. Again the rate of decomposition of the resin based on I11 is
slower u p to 250’ C. than that based on Chloran T M , but more rapid a t higher temperatures.
Table VI.
Temp., O
c.
105 140 160 180 200
Thermal Stability of Monomer as Per Cent Decomposition (Weight Loss) in Air Chlorendic Chloran Compound Anhydridea TMa IIP TCPAb 0.05 ... ... ... 0.40 ... ... ... 1.15 ... ... ... 2.80 ... ... 0.4 5.70 ... ... 1 .o
270 6.70 6.5 74.3 280 9.0 10.3 ... 290 12.00 16.4 ... 70’ C.per min. in air using Du Pont a 6.4’ C.per min. in air ( 12). 950 thermograaimetric analyzer.
Table VII.
Thermal Stability of Unfilled Cured Resins as Per Cent Decomposition (Weight loss)
(IOo C. per min. in air)
170 180 220
Chlorendic Resin 0.0 0.2 0.4
Chloran TM Resin 0.0 0.2 0.4
Compound 111 Resin 0.0 0.0 0.2
TCPA Resin 0.2 0.7 3.1
350 375 400
67.1 82.0 84.0
24.8 52.3 80.4
45.5 71.8 76.0
68.6 86.5 90.7
Temp., a
c.
Acknowledgment
The authors acknowledge the fine technical assistance of Dorothy J. Woessner, Arnold E. Bloomquist, and Vincent Rogalski. literature Cited ( 1 ) Archer Daniels Midland Co., unpublished results, 1962. ( 2 ) Bjorksten, J., “Polyesters and Their Applications,” pp. 131-8, Reinhold, New York, 1956. ( 3 ) Ibid., p. 138. ( 4 ) Ibid., pp. 157-8. ( 5 ) Ibid., p. 162. ( 6 ) DePuy, C. H., King, R. W., Chem. Revs. 60,431 (1960). 1963,p. 4284. ( 7 ) Fray, G. I., J . Chem. SOC. (8) Gould, E. S., “Mechanism and Structure in Organic Chemistry,” p. 495, Henry Holt, New York, 1959. VOL. 6
NO. 2
JUNE 1 9 6 7
103
(9) Lawrence, J. R., "Polyester Resins," p. 40, Reinhold, New Ynrk. 1960. -
(14) Robitschek, P., Nelson, S. J., Zbid., 48, 1951 (1956). (151 Schmidt. W. G.. Tram. J . PlasticsZnst. 1965. D. 247. (16j Walling,' C., "Free Radicals in- Soluiion," p. 50, Wiley, - I '
(10) Z b k , pp. 58-77.
(11) Parkyn, B., Brit. Plastzcs 32, 29-31, 34 (1959). (12) Roberts, C. W., Haigh, D. H., Rathsack, R. J., J. ApPl. Polymer Sci. 8, 363 (1964). (13) Robitschek, P., Bean, C . T., Znd. Eng. Chem. 46, 1628 (1954).
New York, 1957.
RECEIVED for review September 2, 1966 ACCEPTED April 3, 1967
USE OF DIFFERENTIAL THERMAL ANALYSIS IN STUDYING GLASS TRANSITIONS AND THERMAL DEGRADATION OF POLYSTYRENE A L T O N E . M A R T I N ' A N D H O W A R D F. R A S E Department of Chemical Engineering, The University of Texas, Austin, Tex.
A series of commercial and standard polystyrene samples was studied using a differential thermal analyzer (DTA) designed for microsamples. Glass transition temperatures were measured and pretreatment of samples, reproducibility of observations, effect of heating rates, effect of sample size, effect of volatiles, and effect of molecular weight were studied to evaluate the utility of DTA as a laboratory tool for polymers. The range of samples used was such that the many unusual effects observed with commercial samples of varying history were easily discernible. DTA is also valuable for following thermal decomposition of polymers, as demonstrated by studies on a polystyrene sample.
INCE Keavney
and Eberlin (1960) first reported extensively
S on the utility of differential thermal analysis (DTA) for determining glass transition temperatures, the technique has been widely applied for these and other transitions. Equipment introduced in recent years permits even more rapid and sensitive measurements which have increased the value of DTA as both a qualitative and quantitative tool. The purpose of this study on polystyrene was to demonstrate both the sensitivity and possible uses of a typical modern DTA designed for microsamples and applicable to routine testing. Experimental Details
Apparatus. The DTA apparatus used in this study was manufactured by the Robert L. Stone Co., Austin, Tex. The experimental data were taken using a Model KA-H recorder-controller assembly in conjunction with a Model H-3 subzero unit. Polystyrene Sample Specifications. The polystyrene samples used were obtained from the Dow Chemical Co. and the National Bureau of Standards. Several of the important physical properties which distinguish among the different samples are listed in Table I. The volatile material in each DOWsample is about 0.1 to 0.2% monomer and the remainder is a commercial plasticizer. T h e percentage of volatiles in Styron 683-27 and -28 was determined by thermogravimetric analysis (200' C. for 200 minutes). Other volatile percentages are as reported by the manufacturer. With the exception of PS-2 and PS-3, which were irregularly sized flakes, all other polystyrene samples were cylindrical pellets. Pellets ranged in sized from 1/16-inch diameter by '/16 inch long (Verelite 683) to '/r-inch diameter by "16 inch long (NBSB). Samples were prepared by cutting small pieces from the pellets. Experimental Procedure. To produce a drift-free base line the product of mass and heat capacity should be equal for Present address, Dow Chemical Co., Freeport, Tex. 104
I & E C P R O D U C T RESEARCH A N D DEVELOPMENT
the polymer and reference substance. For this reason a constant mass ratio of 1.2 parts of reference to 1.O part of polymer was maintained. I n testing each sample the sample was annealed by heating at the chosen rate to approximately 40' above the glass transition and then cooled at the same rate to -20' to -40' C. This was followed by heating a t the desired rate '40' or 60' above the glass transition. Heating rates were measured at frequent intervals.
+
Analysis of Thermograms. Three dependent variables, the transition temperature, the slope of the transition peak, and the transition peak height, were chosen for correlating. Because there are several ways to interpret thermograms and only relative values were of interest in this study, a standard method of measuring the above quantities was used. The transition temperature was taken as the point of intersection of the base line and the extrapolation of the deflection of the curve as shown in Figure 1. Peak height was the measured distance in centimeters from the extended base line to the point of greatest deflection immediately after the glass transition. Peak height was always measured at right angles to the edge of the chart paper. Assessment of Experimental Factors and Sample Characteristics
Necessity for Annealing. Annealing of the samples was important in obtaining good thermograms. T h e annealing process consisted of heating the sample through the glass transition region and cooling it once again to ambient. Nonreproducible exotherms and endotherms were recorded for all of the commercial Dow samples during this initial heating. Figure 2 A shows two such peaks obtained with Styron 690. Curve B is the second heating of the same sample. As can be seen, the sharp peaks between 60' and 80' C. have disappeared. Curve C, however, is a different sample of the same polymer, for which no peaks between 60' and 80' C.
