2748 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 45, No. 12

CONCLUSIONS

The strong dependence of ultrasonic propagation parameters on the extent of cure of phenolic molding resins suggests that they may be used as criteria for the extent of polymerization of phenolic and other thermosetting molding resins. Measurement of these parameters can be made while the resin is undergoing cure in a typical compression mold, thus providing instantaneous informa- tion regarding the progress of cure. The technique also seems capable of providing useful indication of the rate of cure. This method, however, will not furnish absolute information relative to the extent or rate of reaction, but may be calibrated against any technique which might yield such information.

1 0 0 0 / T * K

Figure 16. Relation of Maximum Slope of Attenuation to Cure Temperature

Data plotted against reciprocal of absolute temperature

tained from the Arrhenius equation, using the rate of change of viscosity of polymer solutions, reacting under alkaline conditions (pH 7.0 to 8.5) a t a series of different temperatures, as an index for the rate of reaction.

Viscosity Changes

LITERATURE CITED

(1) Adams, J. H., Bakelite Co., Union Carbide &- Carbon Corp.,

( 2 ) Auerbach, V., Bakelite Co., Union Carbide & Carbon Corp.,

(3) Green, R. B., Barrett. Division, Allied Chemical & Dye Corp.,

(4) Mikhailov, I. G., and Gurevich, S. R., Zhur. Ekspt l . i Teort. Fiz. ,

(5) Nielson, L. E., Am. SOC. Testing Materials Bull., 165, 48-52

( 6 ) Xolle, A. W,, and Rlowry, S. L., J . Acoust. SOC. Am., 20, 432

( 7 ) Sofer, G. A., and Hauser, E. A., J . P o l y n w Sci.. 8, 611 (1952).

R E C E I V F ~ for reriew March 30, 1953. ACCEPTED August 15, 1963. Work descrihed is one aspect of the research which constit,uted a doctor of science thesis in the Department of Chemical Enginearing a t the Massa- chusetts Institute of Technology, Cambridge, llass., under the supervision of E. A. Hauser. Work sponsored by the Plastics Group, Manufacturing Chemists’ .4ssociation, Washington, D. C., under the general direction of Albert G. H. Dietz. professor of structural engineering and director of the Plastics Research Laboratory, RIassachusetts Institute of Technology, Cam- bridge, Mass.

personal communication, July 24, 1952.

personal communication, Dee. 3, 1952.

personal communication, Dee. 22, 1952.

19, 193-201 (1949).

(1950).

(1948).

in Thermosetting I

Resins D. I. MARSHALL

Development Laboratories, Bakelite Co., Division of Union Carbide 6% Carbon Corp., Bound Brook, N. J .

OLID thermosetting resins have been manufactured and S used for many years for molding material and bonding appli- cations in which performance is related to viscosity and reaction speed. Therefore, viscosity-temperature relations up to fabricat- ing temperatures and viscosity-time curves in the fabricating temperature range are of primary concern to the thermosetting plastics industry. However, serious attempts to measure the prop- erties effectively have been announced only recently. The stand- ard test methods used by the industry, such as the hot-plate gel time test, have been useful for production control but have not yielded fundamental data.

Recently published work on flow properties of phenolic resins has included three papers dealing with the viscosity of the Novo- lak (nonsetting) type of resin (2, 4, &). The work included no results on reactive phenolic resins, however. Two methods have been announced for measuring the viscosity of thermosetting resins and following viscosit,y change during reaction, but data had not appeared a t this writing. The methods include a rotat- ing shearing disk viscosity method by Sontag ( 7 ) and an ultra- sonic viscosity method by Roth and Rich (6).

This paper describes the procedure and presents results ob- tained using a third method of measuring rapid viscosity changes -namely, the parallel plate plastometer method. Results ob-

tained on several phenolic resins and a silicone resin are presented. The method is not continuous and is not applicable to materials departing appreciably from Newtonian behavior. However, it has the very desirable features of using layers of resin thin enough for rapid temperature equilibrium, no cleaning problem, and an extremely wide viscosity range.

THEORY OF METHOD

The theory of the parallel plate plastometer and the applica- tion of the device to the measurement of the viscosity of Novolak resins have been described by Dienes (3, 3) . Briefly, a test specimen is squeezed between two parallel plane surfaces, causing radial flow, and the distance between the planes, h, is measured as a function of time. In accordance with the theory, a plot of l / h 4 versus time, equivalent to a deformation-time curve, is constructed and the slope determined. The equation for vis- cosity is

8.21 X 10W mV2 1 7 =

where 7 = viscosity, poises; W = load on sample, kg.; V = volume of sample, cc.; ?n = slope of the plot, cm.? sec.-l

December 1953 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 2749

NEWTONIAN BEHAVIOR OF THERMOSETTING RESINS

6

2

0

I

TWO-STAGE RESIN WITH HARDENER 910 POISES

TIME IN SECONDS

Figure 1. Deformation-Time Curves on Phenol- Formaldehyde Resins at 140’ C.

Equation 1 is the basic equation by which viscosity values are determined and used to construct viscosity-temperature curves or viscosity-time curves on thermosetting resins.

I I

10 20 30 40 50 TIME IN SECONOS

Figure 2. Deformation-Time Curve on Silicone Resin at 150’ C. Viscosity 3 X lo6 Poises

The papers previously cited on flow properties of Novolak resins are in agreement on the Newtonian character of these substances. Data on the reactive resins were lacking, however. Experiments were carried out, therefore, to confirm the Newtonian character of some reactive resins. Figures 1 and 2 show the deformation- time curves obtained. The fast deformations recorded in Figure 1 were recorded with a photographic technique, whereas the curve shown in Figure 2 was obtained by reading the dial micrometer at %second intervals. The linear nature of these plots shows the two-stage resin with hardener, the self-hardening resol resin, and the silicone resin to be essentially Newtonian under the condi- . tions of the tests. With most phenolic resin samples this be- havior holds true until the viscosity a t 140” C. has increased by a factor of approximately 100, at which point non-Newtonian behavior commonly appears as a result of increased molecular weight and increased cross linking.

Further indication of whether a material is Newtonian can be had by carrying out repeated parallel plate viscosity tests using different loads and sample sizes. Variations in viscosity results greater than those obtained under constant loads and sample sizes are an indication of anomalous flow properties. Many resins show this behavior in the later portions of the viscosity- time curves, but in the initial portions the method revealed no anomalous flow properties.

PROCEDURE

Some simplification of the viscosity measurement is possible where linear deformation-time plots such as those shown in Figures 1 and 2 are obtained. In this case two points only are required to determine the slope. The curves will be concave toward the time axis if the resin is hardening during the test

Figure 3. Viscosity-Temperature Curve and Viscosity-Time Curve on Resol Phenolic Resin

2750 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y VOI. 45, No. 12

6 i

8. RESOL PHENOLIC RESIN AT 140' C.

%w ,/ C. T~~r,STfiGE--PH_ENOLIC

I /0 D. SILICONE RESIN AT i60° C.

0 40 eo 120 160 200 2 , TIME IN SECONDS

Figure 4. Yiscosity-Time Curves on Three Different Types of Thermosetting Resins

or if the resin has advanced to a point where non-Newtonian flow becomes noticeable. The effect of hardening may be minimized by speeding up the test, but anomalous flow is more troublesome.

6

5

4

3

u) W 9

f 2

F

3' 3

2

I

I I I

120. c. G = RESOL RESIN H * TWO-STAGE RESIN -

SECONDS 100 200 300 400 500 6

140' C. A= RESOL RESIN I = TWO-STAGE RESIN

I I I I 30 60 90 120 150

TIME IN SECONDS

Figure 6. Viscosity-Time Curves at Two Different Temperatures

150 180 0 30 60 90 120

TIME IN SECONDS AT 14OOC.

Figure 5 . Yiscosi ty-Time Curves on Three Different Two Stage Phenolic Resins

When the curvature is not too pronounced, the slope of a secant to the curve, which approximates the slope of the tangent a t a point midway between the two time values, may be used as the measure of viscosity. With resinous materials that already are non-Newtonian in the initial melts, the parallel plate method i9

of very limited usefulness. Its use is further limited to conditions under which foaming is not severe.

Equation 1 was made more convenient by representing the proportionality constant and normalization factor by s ingle quantity, 8'. Equation 1 then becomes

By setting up a chart indicating F values for practical loads and specimen sizes, and a table of Allh4 values for practical intervals of h , the calculations are reduced to a minimum.

The need for cleaning the testing apparatus is eliminated by using 0.001-inch aluminum foil to separate the resin from the metal plates.

The test samples are pressed into pellet form from pulverized or granulated resin. A series of tests is carried out a t different tem- peratures to construct a viscosity-temperature curve, and at different heating times to construct a viscosity-time curve. For the latter, each test pellet is rapidly pressed to a thickness of about 0.8 mm., where effective temperature equilibrium requires ap- proximately 3 seconds' time, and the position is held for the de- sired heating time. A4t the end of the heating time the force is applied to squeeze the sample and a specified interval of thick- ness is timed. The viscosity value is computed and plotted a t the time value obtained by adding the heating period to one half the time required for the viscosity reading. Two viscosity values may be obtained in a single test by timing two thickness intervals. In cases where it is desired to extend a viscosity-time curve into the region of anomalous flow properties it is helpful to plat de-

December 1953 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y 275 1

6 I 5-

4-

TWO-STAGE RESIN K = I . l%WATER - J * .3% WATER

Z 6- P

4 -

RESOL RESIN

B * 2% WATER L* I% WATER

I I I I 0 30 60 90 120 150 180 -

TIME I N SECONDS AT 140. G.

Figure 7. Viscosity-Time Curves at Different Values of iMoisture Content

formation-time curves from readings taken a t several time values. Anomalous flow curves so obtained can be used to draw tangents a t different points, thereby yielding two or more viscosity values per curve. While laborious, this procedure improves accuracy where anomalous flow is encountered.

A sample result combining a viscosity-temperature curve and a viscosity-time curve is shown in Figure 3. This plot serves to illustrate the tremendous viscosity range available. Such a combination curve offers a picture of the viscosity changes that take place when the resin is carried through an idealized melting and curing cycle. In this paper the melting portion is not taken up in detail; rather attention is focused on the hardening portion.

METHOD OF ANALYZING THE VISCOSITY-TIME CURVE

Neither a theoretical nor an empirical relation for fitting the the viscosity-time curves has been found. One has several choices for obtaining useful quantities from the curves, however. The initial viscosity may be selected as a measure of how soft the resin will become. As a measure of reaction speed, the time required for viscosity to change by a factor of 100 was chosen. A third property, which perhaps is of more interest to the fabricator, is the time required to reach a certain viscosity level. Still a fourth property was obtained by plotting the reciprocal viscosity against time and measuring the area under the curve. The integral so obtained, which may be termed the fluidity-time integral (FTI), is further defined as

FTI = Jrn : (3)

It gives a measure of the maximum shear strain theoretically obtainable per unit of stress. In other words, i t is a measure of the amount of flow that will be experienced in an idealized situation.

Applying the analysis to curve A (Figure 3) one finds that t h e resin had an initial viscosity of 80 poises, required 112 seconds for viscosity to increase by a factor of 100, required 141 seconds for viscosity to reach 106 poises, and had a fluidity-time integral of 0.475 sq. cm./dynes (0.475 strain per unit of stress).

RESULTS ON SEVERAL RESINS

Viscosity-time curves are shown in Figures 3 to 7 on several phenol-formaldehyde resins and a silicone resin. The data obtained from the curves are listed in Table I. These results illustrate the wide ranges of viscosity and reaction speed that are found among thermosetting resins. Curves A and B represent base-catalyzed resol type resins. Curves C, E, and F show a, wide range of behavior found in three two-stage resins- Novolak resins with hardener. The hardener for the two-stage resins was hexamethylenetetramine added in the amount of approximately 10% by weight. The differences in initial vis- cosity found among the two-stage resins represent differences in the mole ratio of formaldehyde to phenol, the ratios ranging in this case approximately from 0.80 to 0.86. Curve F represents a fastrcuring, two-stage phenolic resin (1 1.

Curve D is that obtained on a thermosetting silicone resin of the methyl phenyl polysiloxane type catalyzed with 0.25% tri- ethanolamine. The silicone resin was much slower in its reaction a t 150 O C. than any of the phenolic resins a t 140 O C.

.4 - u) W z * n

ai .3- z 0

z - . e -

: I- * . I - - G li

0 I I I I

0 IO 20 30 40 50 60

INCLINED-PLATE FLOW AT 125'C. IN MM.

Figure 8. Correlation between Fluidity-Time Integral and Inclined-Plate Flow

The effect of temperature on the viscosity-time curve is illustrated by the curves shown in Figure 6. The reaction of the two-stage resin (curves H and I ) was more sensitive to tempera- ture than that of the resol (curves A and G ) .

TABLE I. DATA FROM VISCOSITY-TIME CURVES Fluidity-

Time Initial See. to See. to Integral

Resin Viscosity, Increase Reach S Cm 1 Curve Type O C. Poises X 100 106 Poises 8 y n i

A B C D E F a H I J K L

Resol Resol Two-stage Silioone Two-stage Two-stage Resol Two-stage Two-stage Two-stage Two-stage Resol

140 140 140 150 140 140 120 120 140 140 140 140

80 21

320 25tiO 2050

48 600

4200 320 250 170 30

112 92 95

350 93 38

393 ,586 97 95 96 81

141 118 101 250

88 56

420 520 104 111 135 96

0.475 1.73 0.149 0.032 0.020 0,400 0.227 0.056 0.116 0,194 0.239 1.16

2752 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 45, No. 12

Viscosity and reaction speed were both very sensitive to temperature in a compensating manner, but with the examples given, the viscosity effect predominated, resulting in a net increase in flow with increasing temperature (compare curves H and I or curves A and G ) . However, the fact that the tempera- ture sensitivity of viscosity becomes smaller a t higher tempera- tures (6) leads to the conclusion that a temperature of maximum flow exists. This checks with the experience of the thermosetting plastics industry where temperatures of maximum flow are commonly encountered.

The effect of moisture on the viscosity-time curves is shown by Figure 7 . The results indicate that moisture not only plasticized the two-stage resin, but increased its initial reaction rat& The acceleration was not large, however, and later became of no consequence when the wetter sample failed to harden to the degree that the dryer one did. This result brings out the inadequacy of a single-point test for hardening speed, indicating that the complete viscosity-time curve is needed for interpreting flow- behavior. In the case of the resol the effect of moisture was somewhat different, the wetter sample remaining softer throughout. The explanation for the difference probably lies in the fact that water is a reaction product in the resol; therefore, the presence of water may tend to retard the reaction through the influence of an equilibrium effect.

The results obtained at different levels of moisture content indicate that moisture tends to cause the viscosity to level off a t a lower plateau, indicating a lowering of the rigidity of a cured article. High moisture content, therefore, while giving rise to greater flow, may be a disadvantage in applications requiring high rigidity a t the end of the curing cycle.

It is interesting to compare the fluidity-time integral with the inclined-plate flow teet commonly used in the grinding wheel industry. The inclined-plate flow test uses a pellet of resin placed on an inclined glass plate a t 125" C. to determine how far

Moisture plasticizes phenolic resins (a).

the resin will flow, which is a crude approach to the property represented by the fluidity-time integral. Figure 8 shows graphically the degree of correlation obtained between fluidity- time integral a t 140' C. and inclined-plate flow at 125" C. for several phenolic resins covering wide ranges of viscosity and reaction speed. The correlation between the two properties is not bad in spite of the temperature difference and the crudeness of the inclined-plate flow test.

The parallel plate plastometer was found useful for measuring rapid ViscoEity changes. Although it has not been applied to other types of thermosetting resins, it should be applicable except where departure from Newtonian behavior is appreciable or where foaming is severe.

ACKNOV('LEDGMF,YT

The author wishes to acknowledge the assistance and helpful criticisms from his associates a t Bakelite Co., particularly F. D. Dexter, V. E. Meharg, W-. A. Zinzow, and H. M. Quackenbos, Jr.

LITERATURE CITED

(1) Bender, H. L., and Farnham, A. G., U. S. Pa ten t 2,475,587 (July

(2) Dienes, G. J., J . Colloid Sci., 4, 257-64 (1949). (3) Dienes, G. J., and Klemm, H. F.. J . A p p l . Phys . , 17, 4 3 - 7 1

(4) Guzzetti , A. J. , Dienes, G. J., and Alfrey, T., J . Colloid Sci., 5,

( 5 ) Jones, T. T., J . A p p l . Chem., 2, 1 3 4 4 9 (1952). (6) Rqth, W., and Rich, S. R., J . A p p l . Phys., 24, 940-50 (1953). (7) Sontag, L. A,, U. 3. P a t e n t 2,574,715 (Nov. 13, 1951).

12, 1949).

(1946).

202-17 (1950).

RECEIVED for review June 4 , l K 3 . ACCEPTED August 13, 1953. Presented before the Division of Polymer Chemistry at the 124th Jfeeting of the z h E R I C A X CHEMICAL SOCIETY, Chicago, 111.

Rosin Acid-Rubber Master A STUDY OF COMPOUNDING VARIABLES

J. F. SVETLIIL AND R. S. HANRIER Research Disisian, Phillips Petroleum Co., Phillips, Tex.

HE use of relatively large amounts of rosin acids as extenders T for synthetic elastomers of high Mooney viscosity is a com- paratively recent outgrowth of oil extension of similar types of polymers. Earlier work had been carried out on latex master- batching of moderate quantities of rosin soaps (6) . Rosin and other organic acids \?-ere also evaluated by other investigators (8) in amounts up to 9 parts based on the elastomer. These in- vestigations showed that certain vulcanizate properties were im- proved by the addition of rosin, especially tensile strength, heat generation, and processability. Howland et al. (5) reported that an elastomer of high Mooney viscosity masterbatched with up to 33 parts of rosin acids gave vulcanizates having excellent tensile strength, heat generation, tear strength, abrasion resistance, and resistance to flex-crack growth. These properties were observed both before and after accelerated oven aging. These same in- vestigators reported that latex masterbatches gave superior prod- ucts compared to those prepared by addition of the rosin acid during compounding, that masterbatching with black aided in overcoming processing problems of the rubber-rosin acid crumb, and that coagulation with various metallic salts offered little ad- vantage over conventional brine-acid techniques. Other groups active in the development of rosin acid masterbatches have shown

substantially the same trends in improvements from extension with rosin acids (i-4, 7 ) .

Only very limited information has been disclosed for master- batches containing larger amounts of rosin. In view of this, masterbatches containing variable amounts of rosin were studied systematically to establish the maximum amount of this extender that can be utilized without causing serious degradation of vul- canizate properties. It was also believed that higher zinc oxide levels might be desirable, especially in the masterbatches con- taining the larger quantities of rosin acid, in view of the possible formation of zinc rosinate during vulcanization. Accordingly, a systematic stiidy has been made of a series of masterbatches of 150-Mooney GR-S-1500 in which the rosin acid content was varied from 0 to 200 parts per 100 parts of rubber. Zinc oxide levels of 3, 5 , 10, 15, and 20 parts per 100 parts of masterbatch were evaluated at each rosin level.

To establish the optimum compounding formulation for use with a rosin acid-extended elastomer, other compounding vari- ables were investigated. The masterbatch selected for this por- tion of the work contained 100 parts of rosin acid per 100 parts of rubber. The effects of varying the pigments, softener level, and curatives were evaluated.