Objective Laboratory Testing of the Processability of Elastomers . I
L. M. WHITE, E. S. EBERS, AND G. E. SHRIVER General Laboratories, United States Rubber Company,Passaic, N. J .
T
age viscosity as determined H E advent of synthetic An analysis was made of the types of adverse behavior e sample with conencountered in compounding and fabricating dry-mixed elastomers, with their En$c!2 instruments euch elastomers on a factory scale, with an attempt to ascribe extreme variation in properas the Mooney viscometer. the Merent modes of behavior to semifundamental propties due to differences both This definition is necemwy since elastomer com oun+ erties of the elastomer and its compounds. On the basis in polymerization methods possibly contain 1003 vanof this analysis, four simple tests-iiller stiffening, comand in chemical composition, ations in viscosity which may pound viscosity, length shrinkage, and rugosity-are prohaa resulted in considerable be important in judging procposed for determining the processing properties of elastoattention being paid by the essing behavior, but which cannot be detected by availamers on a laboratory scale. Extended correlations of rubber technologist t o the ble viscometers.) predictions made from these tests with factory data demprocessing characteristics of 2. The ease of incorporaonstrate their reliability and usefulness in laboratory evaleach new type and modition of the filler (believed to uation of new rubberlike materials. fication. Not infrequently, be closely related to ease of wetting and dispersion, 9) by Much of the background information referred to in this new elastomers, although the polymer. and the following paper (page 770) was accumulated by commercially advantageous 8. The viscosity of the the rubber industry at large working on the governmentin other respects, have been final compound. sponsored rubber program through the Research and Derejected by the fabrication 4. The tendency of the compound to retract &er velopment !%&on, Rubber Reserve Company, and thereplanta because of poor procstrain. fore has been subject to the usual secrecy orders. For essability. Development of 6. The local nonre@;ulWity this reason much of the prior literature has not been pubimproved elastomers has also of flow snd of distribution of lished in the usual channels. However, every effort has beenhampered by the timestrain in the compound during deformation. been made to recognize the contributions, made but not consuming a n d costly released, by other investigatorsby reference to their names factory-scale evaluations reand organizations. quired to determine processThe first three factors are ing quality. involved primarily in the mixThe growing importance of this particular aspect of polymer ing operation. Higher rates of breakdown, &eater e k e of inoorevaluation prompted an analysis of processing difficulties enporation of the iiller, and a lower viscosity of the finished comcountered in mixing, calendering, extrusion, and related operapound, all contribute to the attainment of lower power consump tions. This analysis led to the development of a set of objective tion, shorter mixing, and shorter remilling cycles. laboratory testa which, it is believed, will predict fairly completely The remaining factors, along with compound viscosity, are the factory-scale processing behavior of dry uncured compounds mainly connected with fabrication operations. Reduced retracof elastomers. Ofthe diverse processing difficultiea encountered tion after fabrication streeses are released is conducive to lass dein the fabricating plants, building tack is believed to be the only parture from the desired shape and dimensions, to reduced tearimportant factor not covered by this investigation. ing, and to fewer lamination failures. Better uniformity of flow A number of useful processing testa, centered around suband of distribution of strain resulta in smoother surfaces, in rejective appearance ratings of tubed or milled specimens, speeds of duced tearing (elimination of thin areas), and in better contact extrusion, or plasticities, have been described in prior publications between laminates. Lower compound viscosity reduces work of (7, 16, 17). Although these tats are frequently useful for preforming, leads to reduced temperatures in extrusion and calenderdicting specific types of behavior, there are several objections to ing, and permits more rapid fusion of laminations. Compound their general uaa. First, they measure only a limited number of Viscosity slso is a factor determining retraction and uniformity of the important types of processing behavior, and seoond, the more flow and strsin distribution, but is aeldom the major factor. comprehensive teats are s u b j e c t i v e t h a t is, dependent upon the An extended survey of factory-acale processing evaluations of operator’s opinion to a large degree. experimental Iota of GR-S made it obvious that no single laboratory teat can be expected to give information about all the above FACTORS INVOLVED IN PROCESSING properties., For example, certain properties are often, but not Types of processing behavior from which difficulties arise on always, antithetical (i.e., when one property undergoes a desirable the elastomer fabrication lime are as follows: change the other undergoes an undesirable change). An example 1. Mixin$ operations: high power consumption during mixis found in factory processing operations on experimental G R S in ,lon mixlng time extended remilling time. samples A-1, A-2, and A-3 (3): &trusion and building operations: surface roughness, gage increase, gross shape changes, tearing, excessive working temperature, lamination failure, hand-working difficulties, slow Polymer Mixing Rating Tubing Rating speed of extrusion. a A-1 (beet) A satisfactory set of laboratory tests must give information conA-2 2 A 4 2 1 (beat) d n g all these types of behavior. Analysis of these processing difficultiesindica- that the folOn a typical tire tread formulation the ratings of the three polylowing characteristics of the elastomer and ita unvulcanised mers were observed SB to cornpatability with black in factory mmpounds are important: mixing, and as to the mlative dimensions and surfaoe appearance 1. The initial viscosity and the rate of decrease of viscosity during mixing operations. (By viscosity is meant the usual averof the final s t o c h after tubing as tire treads. Here the polymer
f.
;
267
INDUSTRIAL AND ENGINEERING CHEMISTRY
768
TABLEI. CORRELATION OF LABORATORY WITH FACTORY RESULTS Filler stjffenv;n~cb
B-2 B-3 Group I1 (GR-5 type)
c-2 c-1
,E[:]
Laboratory ResulW Com-
tE2it$;.
Factory Reaultsb Compound-
I:[;!
A-2 A-3 Group IV (GR-5 type)d D-1 D-2
%I I : { ! %
E-2 E3 Group VI (Perbunan type)e F- 1
F-2
Et;]
;“si1
Group VI1 (Perbunan type)’ 0-1 G-2 The meaaured value ia given aa the first number; the qualitative rating baaed on these values i a given in parentheses. b Numbers in parentheses are literature citations. e Mooney viscositiee measured at 212O F. with large rotor. d These shrinkage rneaaurementa were made on stock calendered at 100’ nstead of 1130’ F. a Them stocks contained more softener and had longer niilling than the stooks of GR-S type polymers.
I:{;:
%I
which was mixed with the greatest facility was extruded with the greatest difficulty, and vice versa. Garvey and co-workers (7,14) also found that a GR-S compound may be very difficult to mix but can still be tubed satisfactorily. It should be emphasized, however, that such behavior is not universal but occurs frequently enough so that two or more different types of laboratory h t s are needed to obtain a full picture of processing Characteristics. For convenience, processing difficulties are here divided into two groups: those peculiar to compounding (mixing) and those involved in fabrication (calendering, extrusion, and lamination). Such a division is consistent with both factory production and the frequently contrasting behavior of a polymer in compounding and in fabrication. SELECTION OF LABORATORY TESTS
The principal aim in facCOMPOUNDING CHARACTERISTICS. tory compounding operations is to obtain the finished compound a t a specified viscosity in minimum time with minimum power consumption. Hence, some system of viscosity measurements a p pears logical as a laboratory test for mixing characteristics. A complete viscosity history of the mixing cycle would be ideal, and would, it is believed, give considerable practical information about mixing behavior. The area under such a viscosity-time curve will increase whenever the processing behavior during mixing is unfavorable. High initial polymer viscosity raises the entire curve, low rate of breakdown raises all but the initial portion, and slow incorporation of the filler raises the maximum and subsequent portions. As the area under the curve increwes, the power consumption must increase. A s a feasible compromise with the ideal, viscosities taken at several select points along the curve when using a fixed mixing cycle seem satisfactory. A successful testing procedure for studying experimental lots of GRS and Perbunan involves obtaining the viscosity of the polymer (V),adding filler (50 parts by weight of EPC black per 100 of elastomer) over 10 minutes, postmilling
Vol. 37, No. 8
10 minutes, obtaining the viscosity of the filler master batch (V’), adding softeners and curatives during 5 minutes, and obtaining the compound viscosity (V”). Since elastomers differ markedly (8)in temperature coefficientsof viscosity, these viscosities should ideally be measured over the entire working temperature range; or, for practical control of a specific mixing and fabrication cycle, viscosities V and V’ should be measured a t the average mixing temperature, and V” should be measured at the specific forming temperature$ (temperatures of extrusion, lamination, molding, etc.). Compounding was carried out on a 6 X 12 inch cold-roll mill and viscosities were obtained on the Mooney disk plastometer (11). Other methods of determining the viscosity could be used-for example, the Williams parallel-plate plastometer (18) or the Mooney rotating-cylinder rheometer (IS). Master batch viscosity V’ anticipates the order of magnitude of compound viscosity V” and, furthermore, is more sensitive to the fundamental mixing properties-rate of breakdown and ease of incorporation of the filler-as the softener has not been added at this point. Therefore, viscosity V’ is considered the more important of the two in characterizing processing behavior during mixing. Since most elastomers are made to a specified initial viscosity (45-55 Mooney viscosity for GR-S),it is usually desirable to make processing comparisons a t an equivalent viscosity. Using this viscosity method of determining relative compound behavior, the difference (V’ - V) between the master batch viscosity V’ and the initial viscosity V is computed, a measurement here called “filler stiffening”. The measurement (V’ - V ) corrects the slight differences that might exist between the initial viscosities, and measures the combined effectsof rate of viscosity reduction and ease of incorporation of the filler, the two important fundamental factors influencingover-all mixing characteristics. High numerical values for (V’ - V ) are indicative of poor mixing properties and low values of desirable behavior. For replicate mixing of a single lot of GR-S, (V’ - V ) showed a coefficient of variation (ratio of standard deviation to the mean) of 10%. (The coefficients of variation in this paper are based on not less than fifteen replicated tests, mixing and testing being done on the same day with the same materials, equipment, and operators.) A similar quantity (V” - V ) ,employing the final viscosity, correlates well with (V’ - V ) but shows greatly reduced sensitivity, the coefficient of variation being 29%. Coefficient of breakdown (17),defined as the ratio of viscosity reduction in a fixed milling time to the original viscosity, gives information about molecular weight reduction but not about ease of incorporation of the filler. Furthermore, the reproducibility of the test is rather poor as the coefficient of variation is 50%. Banding time ( l b ) , the time necessary for the raw elastomer to form a hole-free band with milling, has been used with fair success in some laboratories for estimating general processability This test might be construed as giving some indication of both mixing (breakdown rate) and fabrication properties (retraction and uniformity of flow). The principal objection, however, is that this test depends on the operator’s opinion for the end point. According to our results, this test also shows poor reproducibility; the coefficient of variation obtained by the same operator is 25%, and by different operators, greater than 50%. FABRICATION CHARACTERISTICS. As pointed out, three p r o p erties of the elastomer compound appear to be largely responsible for the behavior of the compound during factory-scale fabrication operations-namely, the ability of the system to maintain the imposed gross shape after @tressis released, to flow uniformly so as to eliminate surface roughness, and to flow under a minimum of stress a t commercial rates of shear. To circumscribe these properties, measurements must be made of gross retraction (shrinkage) after stress is released, of roughness of the surface of the retracted sample, and of the viscosity of the sample. Of necessity commercial equipment operates at high rates of shearthat ie, at stresses far above the yield point of the material.
TABLE 11. Group I Black No. 1 Black No. 2 Black No. 3
769
INDUSTRIAL AND ENGINEERING CHEMISTRY
August, 1945
CORRELATIONS FOR COMPOUNDING VARIATIONS Factory Rasultab Laboratory Results" RugosComFabriV' V V" %La& ity pounding cation
-
6'1)
:{;;
Group 111 lack content) 20parts80. 6 13(1) 40 parts No 5
%#
M(1)
... ... ...
22(2) 4%
48 1)
];/:
44 4)
$1
2(8) 1 3
%I/
68 3 ) .
1 2
4
3
2
1 60partsNo:6 8 0 ~ a r t a N o . 6 73 4 7 1 3) 4 3 0 The measured value ia g&en as the first number and the qualitative ratin as the second fi ure. b fiumbers in parenkjleees are literature citations, and ratings in group 111 are baaed on general factory-soale experience.
cosity or the tubing speed of the compound. However, from the viewpoint of routine testing, the tubing method of making the test specimen has two inherent difficulties: First, any estimates of roughness and irregularities of wufsce, important characteristics with respect to fabrication, are subjective; second, a small laboratory tuber is not easily operated at a constant extrusion rate and a constant temperature. Another possible method for obtaining q o s a retraction after strain is the Williams recovery test (18). In a few exploratory comparisons it has shown good correlation with both laboratory and factory-scale observations of length shrinkage. However, the test has not been shown to give information concerning rugosity, a property which probably commands more universal attentinn in the practical field than length shrinkage. CORRELATION O F WW)RAIY)RY AND FACTORY DATA
Therefore, viscosities at commercial rates of shear are of more value than yield points in this discussion. A satisfactory, simple set of objective tests which successfully depicts large-scale fabrication behavior of an elastomer compound consists of obtaining compound viscosity V", of determining the extent of with-grain retraction (length shrinkage) of a thin calendered sheet, and of objectively measuring the rugosity (roughne=) of the surface of the retracted calender sheet. Simple and reproducible means of making the sample for the length shrinkage and rugosity measurements involves banding the compound at 0.030-0.040inch roll gage on the middle roll of a two-roll, even-roll speed calender (additional rolls dropped from contact with the sample) while maintaining the roll temperature within 2-3' F. of some fixed value (160' F.). The stock is marked at convenient lengths (25 cm.) on the back of the middle roll near the nip by a notched wheel running on the stock. On either side of the marking wheel at 5-cm. distances are placed rotary knives which cut the marked sheet to 10-om. widths. After constant temperature is attained in the working stock, six marked sections are cut from the marked band either as individual sections or as groups of two or three. These sections are allowed to shrink freely at room temperature, top roll face up, on a talcked, smooth, level surface for 16 hours (shrinkage is actually 90% complete at the end of 15 minutes for GRS compounds, but the longer time is used because it gives complete shrinkage). As a refinement of this method the specimens could be free-shrunk in a constanttemperature water bath maintained at the calendering temperature. Except for high-melting crystallizable substances, such as balata, this refinement has not been deemed necessary. The sample prepared in this manner is examined for length shrinkage (with-grain retraction) and rugosity. The length shrinkage is calculated as the percentage decrease in length, 100 (Lo - LI)/L. where L. = unshrunk length (25 cm. for the example given) L, = average length between notched wheel marks on fully shrunk sample, cm. The rugosity (roughness) of the top-roll face of the same specimens is determined by the Mooney rugosimeter (I@, an instrument which gives, on the basis of some assumptions of regularity of the surface roughness, the average hill height in millimeters. The coefficients of variation for the three tests on a mix-to-mix replication were found to be 2oJ, for length shrinkage, 14% for rugosity, and 4% for compound viscosity. To obtain such good reproducibility, control over the mixing operations must be exercised, such as fixed times for addition of ingredients and reasonably constant-temperature rolls. As with filler stiffening (V' V ) , high numerical values for length shrinkage, rugosity, and compound viscosity point to poor processing characteristics. Another method of obtaining a complete picture of fabrication characteristics would be the Garvey tubing index test (7), along with measurement of percentage swell of the tubed piece after retraction to equilibrium (I?'), and measurement of either the vis-
-
Four objective laboratory tests on a new polymer-filler stiffening, compound viscosity, length shrinkage, and rugosityhave been postulated as giving a fairly complete picture of the way in which the polymer will behave relative to a well-established control polymer in factory processing. To determine whether the set of tests is giving a qualitatively correct answer, advantage was taken of the extensive facto4 processing information accumulated on many of the new synthetic elastomers produced during the.past two years. In every case cited in Table I, the laboratory tests were made on Sample6 of the identical polymers used in the factory evaluations. According t o common procedure, factory ratings are relative only (1 is best, 2 is next, etc.) and are limited to intragroup comparisons. However, comparisons between groups can be made by use of measured values from laboratory tests. If proper standardization of sample preparation and instrument calibration has been made, comparison of results from different laboratories is also possible. Inspection of the correlations in Table I shows that it is i m p sible to predict in every case the observed factory behavior by any one of the laboratory tests, since a combination is necessary for the most complete evaluation. For example, neither compound viscosity V" nor m e r stiffening will predict the differences among B-1, B-2, and B 3 , or among E-1,E 2 , EX, and E-4. Likewise length shrinkage (% L.S.) will not detect the large processing difference between B-1 and B-2. Rugosity, on the other hand, will not differentiate between E-1 and E-3. Major compounding changes on which factory processing data are available also have been checked by these laboratory tests for correlation purposes. As Table I1 shows, the laboratory testa predict factory behavior in every case. LITERATURE CITED
(1) Bacon, W. E., U. S. Rubber Co., private oommunication, Nov. 17, 1943. (2) Bartell, F. E.,and Herahberger, A., J . Rhsd., 2, 177 (1931). (3) Cuthbertson, Q. R., Miller, R. R., and Temple, J. W., U. 8. Rubber Go., privsts communication, Jan. 81,1944. (4) I b g . , March 30, 1944. (6) Ibid., July 16, 1944.
(6) Drew, P.W., Goodyear Tire & Rubber Co., private communiaation, Feb. 27. 1944. (7) Garvey, B. S., Whitlock, M. H.,and F r m , J. A., IND. ENG. CHBM.,34, 1309 (1943). ( 8 ) Kelsey, R. H., and Dillon, J. A., J . AppliedPhyn., 15,362 (1944). (9) Millard, J., U.8.Rubber Co., private wmmunication, April 26, 1944.
(10) (11) (12) (13) (14)
Miller, R. R., Ibid.,June 1, 1944. Mooney, M., IND.ENG.Cmaa., &AI..
ED.,6 , 147 (1934).
Ib%., 17, Aug., 1946. Mooney, M., Physics, 7, 413 (1936). Schroeder, W. W., B. F. Goodrich Co., private communication, Dec. 16, 1943. (16) Sturgis, B. M., and Vincent, J. R., Div. of Rubber Chem., A.C.S., New York,Oct. 6, 1943. (16) Vanderbilt, B. M., Esso Lab., private communication, July 28, 1944. (17) Vila, G. R., IND. ENQ.CHPY.,36,1113 (1944). (18) Williams, I., Zbid., 16, 362 (1924).
