and no weather checking in the grooves, such as is normally encountered with diene rubbers. These results would indicate promise for rubbers of this type in tires. Acknowledgment
The authors thank C. R. Mitchell for his assistance in the experimental polymerization work. literature Cited
(1) Amberg: L. O., Robinson, A. E., IND.ENG.CHEM.53, 368-70 (1961). (2) Bryan, C. E., private communication. (3) Carrick, W. L., J . Am. Chem. Soc. 80, 6455 (1958). (4) Carrick. 1%‘. L., Chasar, .4.G.. Smith, J. J., Ibid., 82, 5319 11960). (51 Hampton, R. R., Research Center, U. S. Rubber Co., Wayne: N. J., unpublished work. (6) Kelly, R. J. (to U. S. Rubber Co.), Belg. Patent 597,592
(9) Lukach, C. A., Spurlin, H. M., Olsen, S. G. (to Hercules Powder Co.), Ibid., 583,040 (1959). (10) Mazzanti, G., Valvassori, A.. Pajaro. G., Chim. Ind. (Milan) 39, 743 (1957). (11) Ibid., p. 825. (12) Miller, R., Carr, C. I.; “High Temperature Protectivr System for Saturated Rubbers,” Meeting-in-Miniature, North Jersey Section ACS, Jan. 30, 1961, (13) Natta, G., Crespi, G., Bruzzone, M., Preprint, Meeting of German Rubber Society, West Berlin, Oct. 4-8, 1960. (14) Natta, G.: Crespi, G., DiGiulio, E.. Ballini, G., Bruzzone, M., Rubber and Plastics Age 42, 53 (1961). (15) Natta, G., Mazzanti, G.. Boschi. G.: Belg. Patent 553,655 11957). I (16) Natta, G., Mazzanti. G.. Valvassori, A . , Sartori, G.: Chim. Ind. (Milan) 40, 717 (1958). (17) Natta. G.. Valvassori. A . Mazzanti. G., Sartori., G.., Ibid.. 40,896 (i958j. (18) Phillips, G. W.. Carrick. I$-. L.. J . Am. Chem. Soc. 84, 920 (1962). (19) Stoner, A. W., U. S. Rubber Co.. Research Center: TVayne, N. J., unpublished work. \ - - -
~
(1 961) \--
RECEIVED for review May 2, 1962 ACCEPTEDJuly 10, 1962
1 -
(7) Luiach, C. A., Olefin Copolymer Symposium, Brooklyn Polytechnic Inst., April 1961. (8) Lukach, C. A,, Olsen, S. G., Spurlin, H. M. (to Hercules Powder Co.). Belg. Patent 583,039(1959).
Division of Petroleum Chemistry, 140th Meeting, ACS, Chicago, Ill., September 1961.
TRIPOLYMERS FROM BUTYL RUBBER POLYM ER I ZAT ION L. S. M I N C K L E R , J R . , A N D A.
B. S M A L L
Chemicals Research Diuision, Esso Research and Engzneerzng Co , Linden, N . J . Various diolefins in the butyl rubber polymerization system have been investigated in the interest of a better understanding of ionic polymerization mechanisms and improved butyl elastomers. Cyclic diolefins are particularly interesting because of the inherent ozone resistance offered by their use. However, previous attempts to use cyclic diolefins have resulted in unacceptable products due to process and product quality problems. The use of certain other diolefins with a cyclic diolefin allows the advantages of each to b e combined in a new high quality ozone-resistant butyl-type rubber. This article deals specifically with the low temperature cationic polymerization of cyclopentadiene-isoprene-isobutylene systems from the standpoints of reaction control, polymer composition, and polymerization mechanism. The resulting butyl tripolymer possesses excellent vulcanizate properties and exceptional ozone resistance. A mechanism for ozone resistance is discussed.
many years, this laboratory has been vitally interested in the copolymerization of various diolefins in the butyl rubber polymerization system. Although only 1 to 3% of the diolefin is incorporated in a polymer that is basically polyisobutylene, it is essential to its vulcanization. Thomas and Sparks (7) are responsible for this discovery, which opened the door to the butyl rubber industry. Their investigations of the low temperature cationic copolymerization of various diolefins with isobutylene indicated that isoprene was the diolefin of choice, although several other conjugated diolefins were also effective. Consequently, isoprene is used exclusively today in the synthesis of butyl rubber. Investigations of different diolefins in the butyl rubber polymerization system have continued in the interest of a better understanding of ionic polymerization mechanisms and improved butyl elastomers. Cyclic diolefins are particularly interesting because of the inherent ozone resistance offered by their use. However, previous attempts to use cyclic diolefins such as cyclopentadiene, cyclohexadiene, and fulvenes have not OR
216
I&EC P R O D U C T RESEARCH A N D DEVELOPMENT
resulted in acceptable products. because of process and product quality problems. The purpose of this paper is to define these problems, explain their origin, and provide solutions in the form of improved butyl elastomers produced with cyclic diolefins. The discussion is limited primarily to cyclopentadiene as a prototype for the various cyclic diolefins investigated in both copolymer and tripolymer systems. All of these led to ozoneresistant products, but each had its unique characteristics. The others are discussed in more detail in a continuing series of articles on the butyl rubber polymerization system. Experimental
Batch polymerization reactions (6) were carried out in a 2-liter copper reactor submerged in liquid ethylene which maintained the temperature a t -100’ C. (Figure 1). The reactor was provided with internal agitation and a nitrogen atmosphere. The 1200-ml. charge consisted of 7570 by volume of methyl chloride and 25% of isobutylene-diolefin feed. This was polymerized by a water-white catalyst solution con-
n
lo(&,
I -C
0.25% AIC13
Nl TROG E N
1
I
-C
a
-+-
I
I
E-3 (Cyclopentadiene) ... .
1
6-3 (Isoprene) H-3 (Cyclohexadiene) D - 3 (Djmethvlbutadiene)
1
I
-
I N CH?CI
EXIT
LIQUID ETHYLENE -100°C.
*-
!A*-* 01 0
Figure 1
.
Low temperature polymerization reactor
taining 0.20 to 0.25y0 aluminum chloride dissolved in methyl chloride. The catalyst solution (-78O C.) was added as a jet stream a t approximately 10 ml. per minute onto the vigorously stirred surface of the chilled reaction mixture. Including a short induction period, the polymerizations were completed in 5 to 10 minutes. Termination was accomplished by the addition of 10 ml. of 2-propanol at the desired conversion level. The butyl polymer product was obtained as a fine slurry which was recovered in 2-propanol. After flashing off the methyl chloride, the product was kneaded in hot 2-propanol. The resulting snow-white crumb was dried under vacuum and submitted for analysis. Polymer unsaturation and molecular weight were determined by the iodine-mercuric acetate method (4, 5) and diisobutylene solution viscosity ( 3 ) ,respectively. Catalyst Preparation. An excess of aluminum chloride was washed three times with portions of methyl chloride and then boiled for 15 minutes in methyl chloride to obtain a saturated solution (approximately 1%). This was diluted to the desired concentration with the same solvent. The final catalyst concentration was determined by sodium hydroxide titration of the hydrolyzed solution. Hydrolysis was best accomplished by the addition of 1 / 5 volume of 2-propanol, followed by an excess of water. Materials. Commercial grades of methyl chloride and isobutylene were used after treatment to ensure reasonable purity. T h e methyl chloride was passed through scrubbers containing anhydrous calcium chloride, 96 to 98% sulfuric acid, and sodium hydroxide flake, in that order. T h e isobutylene was passed through anhydrous calcium chloride and barium oxide, respectively. Both were then condensed under dry nitrogen in glass containers cooled to -78 C. in dry ice. Pure grades of diolefin monomers were carefully redistilled and stored a t -60' C. until used. The one exception was cyclopentadiene, which was obtained by cracking dicyclopentadiene prior to redistillation. The catalyst was prepared from J. 'T'. Baker's C.P. grade granular anhydrous aluminum chloride. Standard compounding and vulcanization techniques were utilized in the evaluation of these elastomers. ,411 compounding was done on a cold rubber mill. Press cures were run a t 307' F. and the resulting pads were cut into dumbbells 0.25 X 0.075 inch in cross section for use in physical testing. Ozone resistance was determined by exposure of these dumbbells a t 50% extension to 0.271, ozone. Results and Discussion
I n batch butyl polymerization systems cyclopentadiene possesses some unique characteristics among cyclic diolefins that have led to process control and polymer quality problems. The most important of these factors are: Heterogeneous and gelled copolymers are obtained. Reaction control is difficult. Molecular weight poisoning is severe, particularly a t lower conversions. I'ulcanizate properties are relatively poor.
I
20
1
I
40 60 i6 CONVERSION
I
80
100
Figure 2. Unsaturation of per cent diolefin-isobutylene copolymers with per cent diolefin utilization indicated
Many of the other cyclic diolefins investigated. such as cyclohexadiene, did not give similar process problems but gave polymers with poor vulcanizate properties. Although cyclopentadiene was more difficult to work with, the better understanding of its properties was one of the most interesting and rewarding features of this work. Butyl Copolymers. This investigation of cyclopentadiene as a comonomer in butyl synthesis soon showed that its greater reactivity relative to isobutylene was largely responsible for its undesirable behavior (Figure 2). Throughout the remainder of this paper, B (for isoprene) and E (for cyclopentadiene), followed by a number indicating weight per cent of each based on total reactants. are used to identify the various polymerization feeds. T h e balance of the reactive feed is isobutylene. A plot of mole per cent unsaturation in the polymers us. conversion shows clearly that cyclopentadiene copolymerizes with isobutylene considerably faster than most other diolefins and, indeed, faster than isobutylene itself. For example, a t 25% total monomer conversion, 70%, of the cyclopentadiene has reacted and a t 50% total monomer conversion. 93% of the cyclopentadiene has reacted. The other diolefins shown behave differently. as indicated by their relatively flat curves, and copolymerize slightly more slowly than isobutylene. For example, at 50% total monomer conversion, only 30% of the isoprene is utilized. These curves explain process control problems rather simply, if we keep in mind that isobutylene alone polymerizes almost instantaneously, and that all diolefins are more or less poisons in the butyl polymerization system. Diolefins tend to slow doLvn the over-all polymerization rate and therefore prevent a runaway reaction. Hence, reaction control problems should have been expected in the copolymerization of small amounts of cyclopentadiene with isobutylene, since, as shown, cyclopentadiene copolymerizes faster. Consequently, its concentration is rapidly depleted and runaway reactions begin a t about 50% total monomer conversion, with the production of essentially pure polyisobutylene. The rapid copolymerization of cyclopentadiene relative to isobutylene also explains the presence of gel, since the tendency to form gel increases at higher polymer unsaturations. Sufficiently high unsaturations were obtained in the copolymer formed early in the reaction to account for this. Furthermore. complete removal of gel from the product by a simple solutionfiltration process sharply decreased unsaturation. This uneven distribution of unsaturation and the resulting heterogeneity contribute to the poor vulcanizate properties. VOL. 1
NO. 3
SEPTEMBER
1962
217
The individual effects of cyclopentadiene and isoprene on their corresponding copolymer molecular weights are compared in a plot (Figure 3) of viscosity-average molecular weight us. per cent conversion. The severe poisoning of molecular weight by cyclopentadiene, when compared with isoprene, is apparent. The increase in molecular weight with increasing conversion is the result of reduced cyclopentadiene concentration; the reverse is true for isoprene, but to a lesser degree. This is consistent with the relative copolymerization rates of these diolefins as stated above and indicative of the dependence of the various polymer properties on diolefin concentration. Butyl Tripolymers. After reasons for the observed behavior of cyclopentadiene in the butyl polymerization system had been established, methods to overcome these difficulties were considered. Naturally, polymerization control and product uniformity were primary objectives. This suggested an approach that would lead to a better distribution of diolefin during polymerization and in the final product. It was accomplished by the addition of a third reactant, specifically isoprene, to the cyclopentadiene-isobutylene system with the consequent tripolymer formation. This effectively allows the advantages of each type of diolefin to be combined in a single elastomer. Tripolymers of cyclopentadiene, isoprene, and isobutylene proved to be high quality ozone-resistant rubbers. They were prepared from feeds containing 1 to 4% total diolefin (based on total reactants). In almost every combination, reaction control was maintained and high quality products were recovered. No gel could be detected and polymer composition was improved, although uneven distribution of unsaturation persisted. When unsaturation is plotted us. per cent conversion for the tripolymer, a curve similar to that obtained for the cyclopentadiene copolymer results (Figure 4). However, it differs in one important respect: The unsaturation is increased by the amount of isoprene incorporated. This demonstrates, as expected, that these curves are additive. Since the curve for isoprene copolymer (B-1) is relatively flat, its addition does not alter the shape of the cyclopentadiene copolymer curve (E-2) appreciably. The incorporated isoprene simply displaces the cyclopentadiene curve upward by an amount equal to the isoprene concentration. This technique supplies the tripolymer (E-2, B-1) with enough uniform unsaturation to ensure good vulcanization. I t follows that as the cyclopentadiene-isoprene ratio decreases, relative uniformity of unsaturation will improve. Although cyclopentadiene remains a molecular weight poison (Figure 5) in the tripolymer system, the shape of the
y
1200-
0
1
- - - - .-.
I
I
I
-
6-3
0
-..- B I
.-I I
I
I
Figure 4. Unsaturation of a tripolymer and corresponding copolymers curve is altered significantly in favor of higher values. This, of course, is the result of replacing some of the cyclopentadiene with isoprene, which reduces the over-all poisoning effect and flattens the curve at higher conversions. Figure 6, constructed from data obtained a t high conversion (90 to 98%), shows the effect of varying amounts of cyclopentadiene and isoprene on tripolymer molecular weights as compared to those of isoprene copolymers. Here X indicates a varying amount of diolefin consistent with the ordinate of the curve. As expected, increasing the cyclopentadiene-isoprene ratio decreases the molecular weight of the tripolymer. By manipulation of the ratio and amounts of the two diolefins, both unsaturation and molecular weight can be controlled over wide limits. In fact, by making use of the additive nature of these polymer systems with respect to unsaturation and employing Figures 4 and 6, one can predict the polymer properties of the product resulting from a given feed. Consequently, a great deal of flexibility is available in the choice of suitable combinations of properties for specific applications. Vulcanizate Properties. When compounded in a variety of butyl rubber formulations, cyclopentadiene-isopreneisobutylene tripolymers gave excellent vulcanizates. However, for purposes of this discussion all results given here were obtained from vulcanizates prepared from the following formulation : Parts by Weight
Ingredient
100
Polymer Carbon black (MPC) Zinc stearate Phenyl-2-naphthylamine Zinc oxide Sulfur Tetramethyl thiuram disulfide
;
ro
1000)
0
50 1 0.2
5
1 1.2
I
,
1
,001
k
r
2 600W 3
2 400J
01
0
0 I
20
1
4
40 60 % CONVERSION
1
80
100
Figure 3. Viscosity-average molecular weight of diolefinisobutylene copolymers 218
I&EC PRODUCT RESEARCH A N D DEVELOPMENT
2
0.
0
1
20
,
I
40 60 70 CONVERSION
Figure 5. Viscosity-average polymer and copolymer
80
-
100
molecular weights of tri-
22001
*n
I
I
I
I
1
d 1000-
1800
1400 -
E- 1,&2 E-1.5, 6-1.5
2
-
-
B-3
0 0
v)
a.
1000 -
600
0
200 1 Figure 6. tripolymers
2 3 4 5 WEIGHT % DIOLEFIN I N FEED
I
1
ob
6
,
I
I
I
15 30 45 60 CURE TIME, MINUTES AT 307°F
I
75
of
Figure 7. Comparative cure rates for tripolymers and butyl copolymers
In milling and processing operations the tripolymer looked and behaved much like ordinary butyl rubber (the isoprene copolymer). Even the vulcanizate properties were similar, with one important exception-in the area of ozone resistance. The data shown in Table I indicate the magnitude of this improvement for various tripolymers. The 0.2y0 ozone test used to measure ozone resistance is extremely severe. Test dumbbells of natural rubber crack and break in less than a minute, while relatively ozone-resistant butyl rubber (B-3) breaks in approximately 30 minutes. In sharp contrast, tripolymers-in which 50% of the diolefin content is cyclopentadiene-are almost unaffected after 3 days of exposure. At lower concentrations the ozone resistance is diminished, but only a small amount of cyclopentadiene is required to obtain significant effects. For example, 1% cyclopentadiene with 270 isoprene yields a tripolymer 80 times more resistant to ozone than the corresponding butyl rubber and yet with almost equivalent physical properties. Apparently, the ozone resistance of these tripolymers is limited only by the amount of cyclopentadiene incorporated and its ratio relative to isoprene. For comparison, a cyclopentadiene copolymer is also listed in the table to show that while it is unaffected by ozone, its other physical properties are not acceptable.
Although cure rates, in terms of 300% modulus (Table I, Figure 7), for tripolymers are somewhat less than for isoprene copolymers with the same total unsaturation, they are considerably faster than can be accounted for by isoprene content alone. Furthermore, cyclopentadiene copolymers can be vulcanized. I t is evident then, that the incorporated cyclopentadiene contributes to the cross-linking process. However, its uneven distribution within the polymer is probably responsible for cyclopentadiene's reduced effectiveness, relative to isoprene, in terms of tensile properties. Whether or not this factor influences the ozone resistance of tripolymer vulcanizates could not be determined directly from this study, since tripolymers with uniform distribution of cylcopentadiene unsaturation were unavailable. However, over fairly broad limits, unevennes of distribution did not seem to alter ozone resistance greatly, so long as the cyclopentadiene was incorporated in a t least a portion of the polymer. Degree of cure is, of course, important in considering data pertaining to ozone resistance, especially within any given polymer system. The data presented in Table I permit comparisons at equal states of cure; but, when isoprene copolymers are compared with the tripolymers, this consideration is of minor importance relative to the effect of cyclopentadiene it-
Viscosity-average
molecular
weights
Table 1.
Polymer4
B- 1 B-3
Cure Time, Min . at 307' F.
15 40 60 15 40 60
E-2, B-1
15 40 60
E-1 .5, B-1 . 5
15 4n ._
E-1, B-I
60
15 40 GO
E-1, B-2
300% modulus, p.s.i.
305 485 510 590 975 1000 380 600
675 430 750 800 340 475 650 450 800 850 345
Polymer Vulcanizate Properties
Tensile Properties Tensile strength,
p.s.i. 2955 3035 3100 3520 3370 3410 2680 3260 3275 2950 3250 3225 3250 3560 3600 3200 3350 3425 1470
35 40 60 E-3 40 a In per cent isoprene ( B ) and cyclopentadiene ( E ) . See discussion.
Elongtion,
Tensile Properties after 72-Hr. Ozone Exposure Tensile modulus, strength, Elongation:
Orone Break.
800
'L7.S.i.
%
385 650 750 585
2920 2850 2950 2920
830 720 700 725
360 550 675 450
675 2630 2600 2700 1135
285 720 660 610 500
325
1350
1020
p.s.i.
Hr.
915 780 760 775 650 680 880 815
1 83 1 60 0 67 0 53 0 49
>72 >72 >72
875 755 715 890 805 785 870
>72c >72c > 72. >72 >72 >72 >72c
73s ..
40
735 880
4i >72
0.25 X 0.075 inch dumbbells at 50% extension.
VOL.
1
NO. 3
c
Cracking observed.
SEPTEMBER 1962
219
self on ozone resistance. O n the other hand, at low cyclopentadiene content, degree of cure again becomes important. Several factors could be responsible for the protection against ozone attack offered by cyclopentadiene and other cyclic diolefins. Both of the following mechanisms may function, but the first is probably more important.
maintain the surface network intact. The result is that the rate of crack initiation is drastically reduced and is primarily dependent on the relative amounts of the two diene structures. The fact that very small concentrations of these unbreakable cyclopentene structures give large improvements in ozone resistance is evidence for such a mechanism. This rules out the possibility of a simple isoprene dilution effect.
Breaking the double bond by reaction with ozone does not break the chain, as is the case with isoprene.
Conclusions
The cyclopentadiene structure may act as an antiozona,nt by reacting preferentially with the ozone. Buckley and Robison (2) showed that ozone degradation of elastomers requires the creation of new surface through cracking under strain. This cracking occurs when ozone attacks surface double bonds, resulting in chain cleavage. The stored energy in the chains is released as they move apart under strain to initiate the crack which grows as ozone attacks newly formed surfaces of highly localized strain. Therefore, protection of the surface to prevent the formation and growth (7) of cracks is of primary importance. In application of this concept to an explanation of the tripolymer behavior, two basic considerations are : The cyclopentadiene component takes part in cross linking; and ozone attack does not break the chain at this point. Therefore, even though the isoprene links are broken, the remaining cyclic components and associated cross links are sufficient to
The causes of process control and product quality problems in the copolymerization of cyclopentadiene and isobutylene were determined. The process was improved by addition of isoprene. with the resulting formation of a cyclopentadieneisoprene-isobutylene tripolymer. This tripolymer combines the advantages of each diolefin in a new butyl-type elastomer which exhibits excellent vulcanizate properties and outstanding ozone resistance. The tripolymer owes its ozone resistance to the cyclic component’s resistance to cleavage and participation in cross linking. literature Cited
(I) Braden, M., Gent, A. N., Kautschuk und Gummi 14, WT157 (1961).
(2) Buckley, D. J., Robison, S. B., J . Polymer Scz. 19, 145 (1956). (3) Flory, P. J., “Principles of Polymer Chemistry,” p. 309, Cornell University Press, Ithaca, N. Y . , 1953. (4) Gallo, S. G., LViese, H. K., Nelson, J. F., IND.ENG.CHEM. 40,
1277 (1948). (5) McNall, L. R., Eby, L. T., Anal. Chem. 29, 951 (1957). ( 6 ) Rehner, J., Jr., Zapp. R. L., Sparks, W. J., J . Polymer Sci. 11, 21 (1953). (7) Thomas, R. M.. Sparks, W. J., “Synthetic Rubber,” G. S. Whitby, ed., Chap. 24, \Viley, New York, 1954. RECEIVED for review May 3, 1962 ACCEPTEDJuly 17, 1962 Division of Rubber Chemistry, ACS. Boston, Mass., .4pril 1962.
FACTORS CONTROLLING THICKENING
OF CARBOXYLATED LATEX B Y POLYACRYLATE THICKENER W
.
W , W
H ITE,
Naugatuck Chemical Division, United StateJ Rubber Go., Naugatuck, Conn.
Statistical methods of experimental design and analyses are particula ly applicable to studies of synthetic latices where a number of factors and their interactions are important. A study of factors controlling the viscosity response of a carboxylated latex to a polyacrylate thickener demonstrates that surface area (particle size) of the latex particles is the largest factor in controlling viscosity build-up. The smaller the latex particles, the higher the viscosity at a given level of thickener. Electrolytes and emulsifiers reduce viscosity, but their effects are small. The viscosity increase with thickener is not linear, and its rate wiih added thickener changes with the colloidal and chemical characteristics of the latex. Increases in the amount
of alkaline electrolyte in the polymerization recipe give inordinately large increases in the final particle size of the latex with consequent decrease in latex thickening. as a raw material has the outstanding characteristic of good fluidity with consequent ease of handling and processing. However, in many applications, such as paint, paper, and textile coatings and carpet backsizings, it is necessary to control the viscosity of the final compound a t a relatively high level. The desired viscosity varies with each application, and this is frequently controlled by the addition of high molecular weighr water-soluble thickeners ( 6 ) . The ATEX
220
I & E C P R O D U C T RESEARCH A N D DEVELOPMENT
viscosity response to a given thickener varies with the chemical and colloidal composition of the latex being used (7). Therefore, it is important to undersrand which factors affect the viscosity build-up in a latex and how these can be controlled during the latex manufacture. A statistically designed experimenr was conducted to study the thickening of a carboxylated styrene-butadiene latex system treated with a polyacrylate thickener. Carboxylated