Dynamic testing as a quality control tool for EPDM polymers and compounds.

Mooney viscosity has been the primary quality control test for the production of elastomers for over 50 years. In recent years, more sophisticated methods have become available for evaluating the viscosity and elasticity of polymers and compounds (ref. 1).

Dynamic mechanical testing, in particular, has been shown to be useful for distinguishing between polymers having the same mooney viscosity but differing in elasticity (refs. 2 and 3). New instrument designs have allowed extension of this technique to polymer compounds.

The purpose of this article is to show how dynamic testing can be used as a very sensitive quality control tool not only for EPDM polymers, but for compounds as well. In the first part we will discuss how polymer variations, such as molecular weight distribution and branching, and mixing, affect the dynamic properties of both polymer and compound. In the second part, we will show how black and oil content, along with mixing time, affect the dynamic properties of the compound. The dynamic properties, in mm, affect the rate of carbon black dispersion.

Experimental

Polymer preparation

Polymers were prepared using a Ziegler-Natta catalyst system. Catalyst and monomer feeds were adjusted to obtain the correct polymer properties. Molecular weight distribution was controlled through variations in catalyst and by the use of hydrogen. The catalyst was deactivated through the addition of alcohol, and the cement was washed with water, stabilized with antioxidant, flocced with steam, and dried.

Three polymers were prepared to very near the same Mooney viscosity. One polymer was a copolymer, the second was a terpolymer having a low content of 5-ethylidene-2-norbornene (ENB), and the third contained dicyclopentadiene. Table 1 gives an analysis of the polymers, which will be designated polymers A, B and C, respectively. Compounds of these polymers will be designated compounds A, B and C, respectively.

Table 1 - raw polymer properties

                                      A         B          C

ML 1+4@ 125 [degrees] C             61.3      63.9       60.0
E/P, (wt. ratio) D-3900             51/49     69/31      58/42
Diene, D-6047                         -      2.3 ENB    3.2 DCP
[M.sub.w](10[5]), GPC, PS equiv.    4.14      4.38       4.77
[M.sub.n](10[5])                    1.95      1.59       1.11
[M.sub.w]/[M.sub.n]                 2.12      2.75       4.31
%>10[6]                             6.48      8.90       12.9
%<10[5]                             11.2      15.3       22.8
Branching (from tan [Delta])        none       low       high

Compound preparation

Compounds were mixed in a Brabender Prep Center mixer using a Banbury mixing head having a capacity of 65 [cm.sup.3]. The mixing chamber's thermostat was set at 65 [degrees] C using a circulating oil bath. The rotor speed was 50 rpm, and the ram pressure .28 Mpa. The instrument is equipped with torque and batch temperature recorders which can monitor these variables. The fill factor for the mixer was kept at 71.5% for all mixes. We used a compound similar to ASTM D 3568 (#1) for our standard compound, containing 80 phr of N650 carbon black, and 50 phr of Sunpar 2280 oil. The levels were subsequently varied in an SED, along with the mixing time. The terpolymer compounds all contained a cure package having 5 phr of zinc oxide, 1 phr of stearic acid, 2 phr of Delac NS, 0.5 phr of Tuex and 1.5 phr of sulfur. The copolymer compounds contained 5 phr of zinc oxide and 8 phr of Vulcup 40 KE.

Rheological measurements

Dynamic mechanical measurements were carried out using an Alpha Technologies RPA 2000 dynamic tester. A 1 [degree] arc (14% strain) was used for polymer measurements, and a 0.5 [degrees] arc for compounds. Ten frequencies were determined over the range 2 to 2,000 CPM. A five minute preheat (cure cycle at 2 CPM) was used prior to taking measurements in order to allow the sample to relax and come to temperature prior to testing. Measurements were carried out at 100 [degrees] C on compounds, and at 50, 100, 150 and 200 [degrees] C on the polymers. Two parameters were used to characterize the dynamic properties, [Eta](*), the complex dynamic viscosity, and tangent delta.

Dispersion index measurements

Dispersion index measurements were carried out on a Federal Instruments dispersion analyzer according to ASTM D 2663, Method C. The constants A and B were published values (ref. 4).

Results and discussion

Polymer data

Analytical data obtained on the three polymers used in this study are shown in table 1. The polymers have very close to the same Mooney viscosity, but differ in molecular weight distribution and branching. Polymer A is essentially linear in structure, and has a narrow molecular weight distribution, while polymer B has a low but measurable degree of branching, and is slightly broader in molecular weight distribution than polymer A. Polymer C has a relatively high degree of branching, and has a broad molecular weight distribution.

Dynamic data were obtained on these polymers as a function of temperature and frequency, as discussed earlier. The data are summarized in figures 1-6.

[Figures 1-6 ILLUSTRATION OMITTED]

Figure 1 shows tangent delta as a function of temperature and frequency for polymer A. This polymer is linear, and has a narrow molecular weight distribution. Therefore one would expect it to be very liquid-like in its flow properties. In agreement with this expectation, tangent delta increases rapidly with decreasing frequency and increasing temperature, as expected for a fluid. The viscosity data for this polymer are shown in figure 2, and the viscosity becomes quite Newtonian at high temperature and low frequency.

Corresponding data for polymer B are shown in figures 3 and 4. Tangent delta for this polymer tends to reach a plateau value of about 1.3 at high temperature and low frequency, much lower than for polymer A. The plateau tangent delta is a measure of the level of branching in this material. The viscosity of polymer B does not level off as much as for polymer A at high temperature and low frequency. Therefore, it has much higher viscosity than polymer A under these conditions. At high frequency and low temperature, the viscosities of these two polymers approach one another.

Data for polymer C are shown in figures 5 and 6. The plateau tangent delta is much lower for this polymer, reflecting its higher level of branching. It also has much higher viscosity at lower frequencies and higher temperature. However, polymer C has lower viscosity at higher frequencies and lower temperatures than polymers A and B, because it has lower molecular weight for the same raw Mooney level, again due to branching.

These data reflect the greater entanglements in the more branched samples at low frequency. The data are in agreement with earlier findings (refs. 2 and 3).

These data demonstrate the great utility of dynamic testing as a quality control test, since the test is sensitive to small variations in molecular weight distribution and branching at constant Mooney viscosity. The dynamic test is also very reproducible, another requirement of a QC test, as shown in figure 7 for a standard EPDM similar to polymer B (ref. 5).

[Figure 7 ILLUSTRATION OMITTED]

In the following sections, we will discuss the application of this test to compounds.

Compound evaluation - effect of rubber

Fewer studies are available on the application of dynamic testing to evaluating EPDM compounds (ref. 6). Studies have been done on other systems (ref. 7). Changes in the amount of oil and black in the compound, as well as the degree of mixing, will obviously affect the dynamic properties, as will any variation in the polymer. We broke the study of these variables into two sections. In the first section we mixed the three polymers in our standard compound recipe for various mixing times, and measured Mooney viscosity, dynamic properties (frequency sweeps at 100 [degrees] C), and dispersion index on all of the compounds. In the second section, we carried out a statistical experimental design in which we varied the black level, the oil level and the mixing time using polymer B.

Data on the dynamic properties of the standard compounds of the three polymers as a function of mixing time are shown in figures 8-11.

[Figures 8-11 ILLUSTRATION OMITTED]

Figure 8 shows tangent delta at low shear rate (0.25 rps, 100 [degrees] C) as a function of mixing time for the compounds A, B and C. The data show that the tangent delta of the compound is mainly determined by the polymer. Compounding has lowered the tangent delta, especially for the copolymer, polymer A. Mixing time has a relatively small effect on tangent delta, mainly at short mixing time. The differences in tangent delta among the three compounds are not as great as for the polymers.

Data on the dynamic viscosity (0.25 rps, 100 [degrees] C) as a function of mixing time are shown in figure 9. The dynamic viscosity shows a significant drop with increased mixing time for both compound B and compound C. Compound A shows little change in dynamic viscosity with mixing time. The reason for this difference is not clear, but is similar to the compound Mooney behavior, which will be discussed later.

High frequency dynamic data on the three compounds as a function of mixing time are shown in figures 10 and 11. The dynamic viscosity drops rapidly with mixing time for compound B and compound C, while showing little change for compound A, similar to the low frequency behavior. However, the ordering of the samples at longer mixing times is different. At low frequency the more branched sample had the lowest tangent delta and the highest viscosity, while at high frequency order is exactly the opposite. Tangent delta at high frequency does not show much change with mixing time for any of the samples.

Dispersion index values obtained on the compounds as a function of mixing time are shown in figure 12. These values represent the percentage of the carbon black that has been dispersed. The dispersion index is higher for compound A, which is based on the more linear copolymer, at short mixing times. At longer mixing times, there is not much difference in dispersion index among the samples.

[Figure 12 ILLUSTRATION OMITTED]

This is surprising since previous studies (ref. 3) have indicated that the more branched samples mix more slowly. This may be compound dependent. The current compound is relatively rich in rubber. The effect of black loading on mixing will be discussed in the next section.

Compound Mooney data for the three polymers as a function of mixing time are shown in figure 13. These data show that the compound Mooney shows little change with mixing time for compound A, and a large drop for compound B and for compound C. This is presumably due to faster incorporation and dispersion of the carbon black for compound A. However, the difference is more pronounced than for the dispersion index, which indicates that incorporation may be more important. Another way of analyzing the Mooney data is to convert it to a mixing index (ref. 3), which is defined as the percent increase in the Mooney at a given mixing time over that of the well mixed compound, defined as the five minute mix for this study.

[Figures 13 ILLUSTRATION OMITTED]

Plots of mixing index versus mixing time for the three compounds are shown in figure 14. As can be seen, the mixing index for compound A shows relatively little change with mixing time, while compound B and compound C show a large drop as mixing progresses. Thus, from this test it appears that compound A, based on the linear copolymer, mixes much faster than the other two materials. However, the dispersion index does not show as big a difference.

[Figure 14 ILLUSTRATION OMITTED]

The dispersion index and the mixing index are compared for the three materials in figure 15. They do show a good correlation, although the correlation may be different for the different polymers. The copolymer, in particular, seems to show much less change in mixing index for a given change in dispersion index. The curve for compound C is shifted relative to compound B due to a continued change in Mooney at long mixing time, with little change in dispersion index.

[Figure 15 ILLUSTRATION OMITTED]

The effect of mixing energy, i.e., the area under the torque-time curve, on the dispersion index is shown in figure 16. It can be seen that the dispersion index correlates very well with the input energy. It appears that compound A gives a little higher dispersion index for a given input energy. The reason for this is not known, but again it could relate to faster incorporation for this polymer.

[Figure 16 ILLUSTRATION OMITTED]

Of course, one would expect that the mixing torque would relate to the compound viscosity. Figure 17 shows a plot of the final torque from each mix versus the dynamic viscosity, [Eta]*, of the mixed compound at a high shear rate, 100 rps. It can be seen that the two are closely correlated, except for one very poorly mixed compound. Compound A has the lowest compound viscosity at short mixing times, but the highest at long mixing times. Compound B is higher in viscosity than compound C at all mixing times.

[Figure 17 ILLUSTRATION OMITTED]

To summarize this section, the dynamic properties of the compounds are a function of the polymer used, and the mixing time. The tangent delta is more affected by the polymer type, and less affected by the mixing time. Thus, the compound based on the high tangent delta polymer has the highest tangent delta at all mixing conditions, and the compound based on the low tan delta polymer has the lowest tan delta.

The dynamic viscosity, [Eta]*, at all frequencies, is more affected by mixing time, and less by the polymer type. The higher the high shear rate viscosity, the higher the energy input in the mix. Dispersion index follows the energy input, with less dependence on the polymer type.

The compound based on the linear copolymer, compound A, shows much less change in viscosity with mixing time than the others. Therefore, it tends to have low energy input at short mixing time, but high energy input at long mixing time. Despite this, this compound has the highest dispersion index at short mixing time. This may be due to faster incorporation of the black.

Effect of black and oil

In the second part of our study, we carried out a statistical experimental design in which we varied the amount of carbon black and the amount of oil in the compound, along with the mixing time. Polymer B was used for these studies. Once again, the responses were the dynamic properties, the Mooney, and the dispersion index, along with the mixing torque and temperature. The center of the design was the standard compound discussed in the first section. The experimental design is shown in table 2, along with the main responses for each of the experiments.

Table 2 - SED variables and responses

Expt.    Black    Oil    Mix time    [Eta]*, .25 rps    Tan [Delta]
   #      phr     phr    minutes        10^5Pa-s         0.25 rps

    1       90     55      2.5            3.24             1.04
    2       90     55      1.5            4.25             0.98
    3       90     45      2.5            3.97             1.03
    4       90     45      1.5            5.52             0.95
    5       70     55      2.5            2.30             1.14
    6       70     55      1.5            3.10             1.09
    7       70     45      2.5            2.97             1.12
    8       70     45      1.5            3.76             1.04
    9      100     50       2             4.62             0.97
   10       60     50       2             2.39             1.09
   11       80     60       2             2.60             1.09
   12       80     40       2             4.17             1.02
   13       80     50       3             2.80             1.13
   14       80     50       1             6.32             0.97
   15       80     50       2             3.14             1.06
   16       80     50       2             3.07             1.07
   17       80     50       2             3.09             1.06

Expt.    [Eta]* 100 rps   Tan [Delta]    %DI           MLI+4
   #        10^3Pa-s        100 rps              @ 100 [degrees] C

    1         6.03           0.46        83.0          75.9
    2         6.82           0.47        61.6          82.6
    3         7.09           0.46        91.9          91.7
    4         8.18           0.45        57.8          95.5
    5         5.24           0.45        75.8          60.8
    6         5.74           0.45        51.1          64.9
    7         6.32           0.44        79.3          73.0
    8         6.69           0.44        57.1          80.3
    9         7.11           0.46        84.6          96.5
   10         5.38           0.43        68.8          61.2
   11         5.24           0.46        80.0          63.9
   12         7.31           0.45        81.3          90.1
   13         6.19           0.49        81.1          73.3
   14         8.55           0.44        46.0          89.7
   15         5.96           0.48        76.4          74.8
   16         5.99           0.46        73.3          73.4
   17         5.99           0.47        76.4          74.2

It should be mentioned that the composition variables are not independent. By expressing them in phr, an increase in oil level has two effects. It changes the composition of the rubber phase, in which the oil is dissolved, and also dilutes the carbon black. An increase in black will not affect the oil level in the robber phase, which remains constant in composition. The black increases relative to the total rubber phase, including oil.

The first thing that jumps out of these data is that the carbon black dispersion is a function of the black level. Higher black content results in faster dispersion. This can be seen in figure 18, which is a contour plot of dispersion index versus mixing time and carbon black level obtained from an analysis of the data. Higher black level results in a higher dispersion index at a given mixing time. The oil level is not a significant factor.

[Figure 18 ILLUSTRATION OMITTED]

The dynamic properties are also a function of composition and mixing time. The dynamic viscosity, [Eta]*, at low shear rate (0.25 rps) is plotted versus black and oil level at a constant mixing time of three minutes in figure 19, and against black and mixing time in figure 20. Increasing black level increases the viscosity, and increasing oil level lowers the viscosity. Increasing mixing time gives lower viscosity, as discussed in the earlier section. Similar results are found for the high shear rate viscosity ([Eta]* @ 100 rps).

[Figure 19-20 ILLUSTRATION OMITTED]

Results for tangent delta at low shear rate (0 [degrees] 25 rps) showed that increasing mixing time gives higher tangent delta, while increasing black and decreasing oil lower it.

Analysis of the Mooney data showed this variable follows the same trends as [Eta]*.

In the previous section, we plotted the dispersion index versus the mixing energy for the three different compounds. We have repeated this for the SED samples in figure 21. Once again we can see that the dispersion index follows the mixing energy independent of the composition of the compound.

[Figure 21 ILLUSTRATION OMITTED]

The composition affects the viscosity, which affects the mixing energy, which determines the dispersion.

To conclude this section, an SED was conducted in which we varied the black, oil and mixing time. Important conclusions are that the rate of carbon black dispersion increases as the black level of the compound increases. Increasing the black level raises viscosity, and lowers tan delta, as expected, and increasing the oil level decreases viscosity, and raises tan delta. Carbon black dispersion closely follows the mixing energy, which is a function of the compound viscosity.

Conclusions

In this article we have attempted to outline how the different variables in a robber compound, the rubber itself, the black and oil level, and the mixing time, affect the dynamic properties, the Mooney and carbon black dispersion.

First we studied three polymers with variations in molecular weight distribution and branching. These variations have a profound effect on dynamic properties, both [Eta]* and tangent delta. Increasing branching, at constant Mooney, lowers tangent delta and raises viscosity at low shear rate, and has the opposite effect at high shear rate.

These effects are carried over into the compound. Thus, a compound based on the linear copolymer has a higher tan delta (0.25 rps) than a compound based on the two more branched polymers, independent of variations in mixing time or black level in the range we studied. Compound viscosity is strongly affected by mixing time, as well as by the polymer, the black level and the oil level.

Carbon black dispersion closely follows the mixing energy, which is a function of compound viscosity and mixing time. The compound viscosity varies with the polymer, the black and oil level, and the mixing time. Higher black level gives faster carbon black dispersion.

From these data, one can see how useful dynamic data can be as a quality tool. Any variation in the compounding ingredients or the mixing time will show up in the dynamic data. A change in the polymer should show up in the tangent delta, while poor mixing will lead to much higher viscosity.

References

(1.) J.S. Dick and H.A. Pawlowski, paper 37, presented at a meeting of the Rubber Division, American Chemical Society, Nashville, Tennessee, November 3-6, 1992.

(2.) K.P. Beardsley and C.C. Ho, J. Elastomers Plast., 16, 20 (1984).

(3.) K.P. Beardsley and R.W. Tomlinson, Rubber Chem. and Technol., 63, 4, 540 (1990).

(4.) Federal Products Corporation, Dispersion Analyzer Manual, 87.

(5.) K.P. Beardsley and W.A. Wortman, "Dynamic mechanical testing as a measure of the viscoelastic consistency of EPDM produced with vanadium catalysts," presented at the Northeast Regional Rubber and Plastics Exposition, Mahwah, N.J., April 17, 1997.

(6.) K.P. Beardsley and W.A. Wortman, proceedings of the Belgian PRI Meeting, paper XII, April 20, 1994.

(7.) A. Coran, Rubber Chem. and Technol., 65, 85, 1016 (1992).