Influence of carbon black morphology and surface activity on vulcanizate properties.

In two recent papers (refs. 1 and 2) Herd and coworkers have applied transmission electron microscope/automated image analysis (TEM/AIA) techniques to compare the aggregate morphological properties of commercial carbon blacks. TEM/AIA offers distinct advantages over colloidal methods such as dibutyl phthalate absorption (DBPA) and nitrogen surface area ([N.sub.2]SA) because of its ability to provide distributional information and measurements that pertain to specific vehicle systems such as rubber. New image skeletonization techniques (ref. 2) now make it possible to measure the branching in carbon black aggregates and enable better shape classification based on pattern recognition techniques.

The surface activity of carbon blacks has been characterized by Ayala and coworkers (refs. 3-6) using a variety of different methods. These include bound rubber, hydrogen content, oxygen pickup, moisture adsorption, SIMS analyses, inverse gas chromatography (IGC) and calorimetry. Based on the supporting data from these techniques, a carbon black-elastomer interaction parameter (I) has been developed from stress-strain and dynamic modulus measurements:

I = [sigma]/[eta] (1)

where [sigma] is the slope of the stress-strain curve in a reasonably linear region (e.g. 100-300% elongation range) and [eta] is a carbon black networking factor based on dynamic elastic modulus:

[eta] = E' (1% DSA)/E' (25% DSA). (2)

The concept of networking was established by Payne (refs. 7-8) to explain the augmentation of dynamic modulus from high to low strain amplitudes. Networking increases with higher carbon black surface area]structure and reduced surface activity, in conjunction with increased loading (ref. 6). The use of the [eta] term in the interaction parameter produces values which are more independent of surface area and structure.

In this article, TEM/AIA measurements of carbon black morphological properties have been correlated with rubber property variations at different levels of the interaction parameter. The primary objectives of this work were:

* To assess the potential of in-rubber aggregate shape measurements for predicting performance variations and;

* To observe the relative influence of different carbon black variables on rubber properties.

Experimental

Carbon black sampling

The sampling of carbon blacks was selected to provide a designed experiment based on systematic variations in particle size, structure and surface activity. These samples represent the same data base employed by Ayala et al (ref. 6). The colloidal and TEM/AIA properties for these carbon blacks are listed in table 1. Surface activity was varied (lowered) by heat treating the carbon blacks in an inert atmosphere (nitrogen) at temperatures of 1,100 and 1,500 degrees C.

[N.sub.2]SA and DBPA were measured for all 15 samples. The changes with partial graphitization were relatively insignificant for both of these properties. The compressed DBPA and TEM/AIA measurements (EMSA, V'/V) were carried out for the untreated carbon blacks only. EMSA is a specific surface area derived from the particle size distribution. Dispersions are achieved by high shear mixing (two-roll mill) in cellulose acetate butyrate (ASTM D 3849 Method D) which minimizes differences in aggregation. For the most part, the EMSA values were similar to the [N.sub.2]SA results, the exception being N231 which gave higher EMSA.

V'/V is a measure of aggregate absorptivity (occlusion), employing the equivalent sphere model of Medalia (ref. 9). This parameter is a dimensionless number which measures the within-aggregate void volume relative to the volume of carbon in the aggregate. The V'/V measurements were based on ASTM D 3849, Method C (50 phr carbon black loading in SBR-1500).

The interaction parameter measurements are based on a 45 phr loading in SBR-1500 using the results of Ayala, et al (ref. 6). Measurements were also carried out at 30, 60, 75 and 90 phr. At the higher loadings, however, the wetting of the carbon black aggregates by the polymer becomes more of a problem with the smaller particle size grades. Therefore, the networking factor tends to have too great an influence on I. The best measurements were obtained at low to intermediate carbon black loadings where adequate wetting by the polymer was achieved for all samples. The values for the 45 phr loading ([I.sub.45]) indicated the best overall results based on their relationship to other measures of surface activity, as well as their sensitivity to changes in rubber properties.

Rubber compounding

Carbon black loadings were varied at 30, 45, 75 and 90 phr in the SBR-1500 formulation (100 phr SBR-1500; carbon black at 30, 45, 60, 75 and 90 phr; zinc oxide 5 phr; stearic acid 2 phr; sulfur 2 phr; and 2 phr benzothiazyl disulfide, not oil extended). Mixing was carried out in a BR internal mixer. Total mixing times were based on the amount of energy required to obtain a dispersion index of 95 or higher (ref. 10). Across each carbon black grade/loading combination (including the partially graphitized samples), mixing energies were kept constant. All compounds were cured at optimum times (based on [t.sub.90]) according to rheometer curves.

Rubber testing

Stress-strain data were obtained using ASTM D 412-83 (Method A) employing an automated United Testing System EVI. Shore hardness was measured by ASTM D 2240-86. Dynamic properties were measured in compression (ref. 11 ) using an Instron Model 1332 dynamic tester. The specimens are Goodrich Flexometer Cylinders (ASTM D 623-78, Method A) which were tested at 24 degrees C, 14.5% mean strain and a frequency of 10 Hz. All compounds were tested in duplicate over a double strain amplitude (DSA) range of 1 to 25%.

Property correlations

The full data base contains 75 rubber compounds which contained the following carbon black variables: Loading - five levels (30, 45, 60, 75 and 90 phr); particle size - three levels; structure - three levels; surface activity - three levels.

All data were analyzed by stepwise multiple linear regression. The terms in the equations included first and second order, along with two-way first order interactions.

Results and discussion

CB properties which influence elastomer reinforcement Dannenberg (ref. 12) reviewed the effects of filler-elastomer interaction on reinforcement. He listed a total of 13 molecular processes which have been employed to explain filler reinforcement of elastomers, but concluded that there is still no satisfactory general theory. Nevertheless, it is well known that carbon black properties such as particle size (surface area), structure and surface reactivity all have a major influence on specific aspects of reinforcement.

Structure - Medalia (ref. 9) and Kraus (ref. 13) have related the influence of structure (DBPA) on rubber stiffness properties (e.g. modulus, viscosity) to the occlusion of polymer within the internal voids of the carbon black aggregates. Part of this occluded polymer is shielded from stress and acts as an additional loading of carbon black.

The effective volume fraction, [phi]', of a carbon black in a rubber compound can be estimated from DBPA using the following equation developed by Medalia (ref. 9):

[phi]'= [phi] [(1 + 0.02139 DBP)]/1.46 (3)

where [phi] is the actual carbon black volume fraction and DBP is the dibutylphthalate absorption in [cm.sup.3]/100g. The [phi]' value expressed in this manner includes both the actual volume fraction and the total amount of polymer occluded within the aggregate void volume. Medalia (ref. 14) has also determined that only about 50% of this occluded polymer is effectively shielded from stress. Based on this hypothesis, the effective volume fraction of carbon black ([V.sub.eff]) may be expressed as:

[V.sub.eff] = 0.5 ([phi]'- [phi]) + [phi] (4)

[V.sub.eff] is associated with the hydrodynamic and strain amplification elastomer reinforcement mechanisms relating to viscosity and modulus, and has shown a good correlation with these properties across a wide range of carbon black types and loadings (refs. 15-16).

Particle size (surface area) - Boonstra (ref. 17) has expressed the particle size/surface area contribution to elastomer reinforcement in terms of the total interfacial surface area, [S.sub.T], between the carbon black and the polymer. This function may be calculated as:

[S.sub.T], [m.sup.2]/[cm.sup.3] =[phi] * p * (SA) (5)

where [phi] is the carbon black volume fraction, p is the carbon black density in g/[cm.sup.3] and SA is the specific surface area in [m.sup.2]/g. The interfacial surface area governs the stress distribution across the carbon black-polymer interface and is highly related to hysteresis (ref. 18) and strength properties. [S.sub.T] is the primary determinant for tensile strength across different carbon black grades but the within-grade maximum value is governed by [V.sub.eff] (refs. 15-16).

Surface activity - The degree of polymer interaction with the carbon black surface represents a third primary property which strongly influences elastomer reinforcement. Reasonably strong carbon black-polymer adhesion is required to restrict molecular mobility in the interfacial region. Without this type of restriction the polymer is free to slip away from the surface, which greatly reduces the influence of carbon black structure on modulus development. Variations in carbon black-polymer adhesion also affect [S.sub.T]. At a given surface area level, weak adhesion increases the level of hysteresis properties such as tan 6 and loss compliance. Failure properties are also affected, but the changes are not as great as the reduction in high strain modulus (e.g. 100 to 300% elongation).

General reinforcement model

The use of [V.sub.eff] and [S.sub.T], nonconjunction with [I.sub.45] to measure surface activity, provides a relatively simple model for studying carbon black reinforcement variations across the different grades. [V.sub.eff] can be determined from either DBPA or compressed DBPA using equations 3 and 4. [S.sub.T] can be calculated from equation 5 using nitrogen surface area, or any of the commonly used surface area measurements such as CTAB or iodine number. The values for the interaction parameters, I, can be determined from the ASTM SBR formulation (D 3191). However, optimum cures using a rheometer are required.

In this article, we have employed the [I.sub.45] values of Ayala et al (ref. 6, table 1) which are applicable to the 15 different carbon black samples in this data base. The 45 phr carbon black loading in SBR-1500 is quite similar to the ASTM formulation (50 phr), which would be expected to give a similar overall pattern of results.

The [V.sub.eff] and [S.sub.T] values have been based on the TEM/AIA measurements for V'/V and EMSA which are listed in table 1. The TEM/AIA data have been utilized because they account for the relative aggregate breakdown in rubber, as well as the contribution of particle size distribution to the surface area. In a broad based study such as this, the colloidal measurements ([N.sub.2]SA, DBPA) give comparable results. In the long term, however, it will be desirable to study the individual carbon black properties in more detail using within-grade comparison or across-grade comparisons at fixed loadings. The TEM/AIA measurements offer greater sensitivity in these more comprehensive studies and are more technically correct because they measure aggregate breakdown. This is particularly important in separating the relative influence on aggregate morphology and surface activity to specific rubber properties.

Aggregate breakdown in rubber - The recent TEM/AIA measurements of Herd, et al (ref. 2) have indicated that aggregate breakdown in rubber increases with increasing DBPA and diminishing NSA as shown by the following expression:

Breakdown = 30.5 (DBPA)/(NSA) - 0.605

(NSA) - 0.0011 [(DBPA).sup.2] - 59.8 r = 0.995, [S.sub.E] =0.022 (6)

Here, breakdown is expressed as the fractional change in V'/V from the dry state to rubber, divided by the dry state values. This equation applies to the five untreated carbon blacks listed in table 1. There greater breakdown indicated for the coarser carbon black grades confirms the earlier findings of Gessler (ref. 19).

The V'/V values in table 1 were converted to [V.sub.eff] employing the relationships established by Medalia (refs. 9 and 14).

Heat treat carbon blacks - Previous studies (ref. 20) have shown that carbon blacks drop to a minimum level of elastomer reactivity when heat treated to 1,400-1,500 degrees C in a thermal induction furnace. Once this level has been reached, further treatments at higher temperatures do not significantly influence rubber properties. Four of the carbon blacks in the present study have reached this condition. These are the 1,500 degrees C treated samples for N121, N330, N650 ad N630 which are compared in table 2 (ref. 6) by means of four different parameters relating to surface activity. These measurements are hydrogen content, oxygen content, equilibrium moisture adsorption and [I.sub.45], all of which are at extremely low levels relative to the untreated controls. Note that the N231 sample is not listed in table 2. This is because it was not graphitized on the same level as the other four samples, i.e., its minimum hydrogen content was about 1,000 ppm and the [I.sub.45] value only dropped to 1.26.

If one assumes that there are no significant surface activity variations among the four heat-treated carbon blacks in table 2, then the modulus levels for the vulcanizates containing these samples should be dependent on [V.sub.eff] alone. Figure 1 plots 200% modulus versus [V.sub.eff] for the compounds containing the untreated samples at the five loadings. Considerable scatter is apparent at the higher loadings because of differences in surface activity. In particular, the surface of coarser carbon blacks is more easily accessible to the polymer at the higher loadings. Therefore they develop higher 200% modulus for a given level of [V.sub.eff].

The compounds containing the 1,500 degrees C treated samples are compared in figure 2. Here, the scatter in the modulus values due to the surface activity differences have been eliminated and there is an excellent correlation with [V.sub.eff]. Similar correlations were indicated for static modulus at 100% elongation, and for dynamic elastic modulus at high strain amplitudes. The latter relationship is shown in figure 3 for E' at 25% DSA. The data in figures 2 and 3 strongly support the fact that the modulus differences among these heat-treated carbon blacks are related only to [V.sub.eff], as measured in rubber. It can be argued that the measurements in rubber were only carried out for the standard carbon blacks at one loading. Actual breakdown may vary with both heat treatment and loading. Experiments to determine these factors are planned for future studies. In the meantime, however, the V'/V data appear to give a more accurate representation for the relative levels of carbon black structure in these rubber compounds than either DBPA or compressed DBPA. The compressed DBPA values do, however, support the fact that a higher structure, large particle size carbon black breaks down much more than a small particle size grade at the same DBPA level (e.g. N121 versus N650 in table 1).

Tan [delta] is significantly affected by carbon black loading surface area and surface activity. Again, the surface activity contribution is eliminated for the 1,500 degrees C heat treated carbon blacks. At a constant carbon black loading, the level of tan [delta] increases with higher [S.sub.T] and diminishing surface activity. This property correlation will be illustrated for the full data base in the final section.

Rubber property responses

Property response equations were established for five stressstrain measurements (including Shore hardness) and five dynamic measurements based on the variations in [V.sub.eff], [S.sub.T] and [I.sub.45]. These equations are listed in table 3, along with their respective correlation coefficient, standard error and observed property range. The listed variables were all significant at a 95% confidence level.

All of the correlations were reasonably high except for tensile which exhibited more complex variations than most of the other properties. Note that the surface activity term (X3) is always present in the equations as an interaction with either [V.sub.eff] or [S.sub.T] . The first and second order [I.sub.45] terms also appear in some of the equations to improve the fit. However, these terms are never entered alone, which is not the case for [V.sub.eff] and [S.sub.T] . Carbon black surface activity primarily governs the magnitude of the polymer response to variations in the effective loading or interfacial surface area, but can also influence microdispersion and rate of cure.

The response equations are useful in determining property relationships but are difficult to interpret visually.

Static modulus and hardness - The variations in modulus (100 to 300% elongation) were predominantly attributable to [V.sub.eff] and [I.sub.45]. As elongation is increased, however, there is also a surface area contribution. The present studies were limited to comparisons of 100 and 200% modulus because many of the breaking elongations were below 300%.

A three-dimensional graphic representation of the 100% modulus equation is illustrated in figure 4. At low levels of [I.sub.45] there is only a small increase in modulus with increasing [V.sub.eff]. The modulus response becomes markedly steeper as both variables are increased. The plot for 200% modulus (figure 5) is similar but the response is much steeper as the two variables increase. With partial graphitization at 1,500 degrees C (far fight), 200% modulus increases with higher [V.sub.eff] up to a level which is about half of that achieved with high carbon black-elastomer interaction (left side of figure). It would be noted that the equation for 200% modulus (table 3) contains [S.sub.T]. Figure 5 represents the response equation prior to the addition of [S.sub.T] , which exerted a relatively small influence. The r value -- 0.951 without the [S.sub.T] terms.

The response equation for Shore hardness was quite simple and indicates that hardness increases with increasing effective loading ([V.sub.eff]) in conjunction with the interaction of [I.sub.45] and [S.sub.T] . A high loading was chosen to include a broad range of [S.sub.T] . The interfacial surface area can be varied independently of [V.sub.eff] by changing the type of carbon black, but was limited to a maximum of about 75 [m.sup.2]/[cm.sup.3] (N121) in the present experimental design. For each 0.05 unit drop in [V.sub.eff], Shore hardness was lowered by about three units across the observed response surface. However, the total area of the response surface becomes smaller at lower [V.sub.eff] because of the reduction in the [S.sub.T] range.

At low values of either [S.sub.T] of [I.sub.45], changes in the other parameter did not significantly affect hardness. At high values for both variables, however, there was about a 15 unit increase in hardness. This is equivalent to a change in [V.sub.eff] of about 0.25.

Tensile and elongation - Rubber properties that involve catastrophic failure processes are typically more difficult to predict on the basis of carbon black properties. Nevertheless, the graphic representation of the tensile strength equation (figure 6) is quite descriptive. This plot shows tensile as a function of [S.sub.T] and [I.sub.45] at a [V.sub.eff] level of 0.50. Note the dome shape of the response surface. The highest level of tensile is dependent on a combination of relatively high [S.sub.T] and high interaction levels. However, at the highest levels of [I.sub.45] and

there is actually a drop in tensile strength. At very high loadings the inactive samples (partially graphitized) for the finer high DBPA grades gave higher tensile than untreated samples. This was not true for the coarser grades. For example, the 1,500 degrees C sample of N121 gave a tensile strength of 22 MPa at 90 phr which was 4 MPa higher than the untreated control. In contrast, N650 gave about 30% lower tensile for its 1,500 degrees C sample at 90 phr relative to the control. Maximum tensile strength occurred with a [V.sub.eff] level of about 0.30 for all carbon blacks, which is consistent with previous findings (ref. 16). The small particle size grades exhibited well defined tensile maxima while those for the coarser types were quite broad.

The level of carbon black-elastomer interaction exerts a much greater influence on % elongation than [S.sub.T] . Elongation reaches a minimum at the highest levels of [I.sub.45] and [S.sub.T] . The effect is considerably less with decreasing carbon black loading. Again, it should be remembered that the [S.sub.T] range is lower (about half) for [V.sub.eff] = 0.25.

Dynamic modulus - At high strain amplitudes, the response for dynamic elastomer modulus, E', was quite similar to that for 100% static modulus. The overall range for the variations in E' is similar to that for 200% modulus.

At low strain amplitudes, dynamic modulus increases with carbon black surface area because of the networking phenomena (refs. 7 and 8). Carbon black-elastomer interaction continues to exert a significant influence, but not nearly the magnitude observed at high amplitudes. At lower loadings a much flatter response is indicated because of the reduction in networking.

Hysteresis properties - The response for hysteresis at constant strain is shown for the loss modulus E" at 25% DSA in figure 7. The dominant variable is [V.sub.eff], but increasing [S.sub.T] also produces an increase in E". Of interest is the fact that there is no surface activity term in the response equation. Across most of the data base, partial graphitization did not produce any significant change in E". This is probably attributable to the cancelling effects of increased polymer mobility (reduced hysteresis) and the simultaneous increase in carbon black networking and polymer slippage at the surface (higher hysteresis). At the highest carbon black loading (90 phr) there was a slight drop in E" for most of the compounds containing the 1,500 degrees C heat-treated carbon blacks. The exception was N630, which had the lowest levels of [V.sub.eff] and [S.sub.T].

For hysteresis at constant energy input (tan [delta]) all three carbon black variables exerted a significant influence. The response surface for tan [delta] is illustrated in figure 8 with [S.sub.T] and [I.sub.45] being plotted at [V.sub.eff] = 0.50. The values for tan [delta] represent their maximum level, which typically occurs in compression at a DSA in the range of 4 to 5%. The response surface gave a similar pattern at different levels of [V.sub.eff]. A 10% change downward in [V.sub.eff] produced a 15 unit drop in tan [delta] x [10.sup.3]. At low surface activity there is a steep increase in tan [delta] with increasing [S.sub.T] . The surface area influence diminishes greatly at high surface activity, which has been attributed to the reduction in networking (refs. 8, 21 and 22). Payne and Whittaker observed a marked reduction in the tan [delta] peak of carbon black filled HR compounds which were heat treated in the presence of a promoter to improve their microdispersion. Hess and Chirico (refs. 21-22) indicated a similar trend for improved tread grade carbon blacks with higher surface activity. These carbon blacks gave higher treadwear resistance in conjunction with reduced hysteresis.

For hysteresis at constant stress, carbon black-elastomer interaction plays a major role. At [V.sub.eff] = 0.50 there is a very sharp drop in D" with increasing carbon black surface activity. Surface area increases D" up to a limiting value of [S.sub.T] above which the increases in hardness produces a negative effect. At a [V.sub.eff] of 0.25, the response to surface activity is less and the increase in D" with higher [S.sub.T] is greater.

Summary and conclusions

TEM/AIA measurements on carbon black morphology were employed in conjunction with surface activity variations to assess rubber property changes within a designed experiment containing 75 SBR vulcanizates. Rubber property response equations were obtained from stepwise multiple regression analyses using a reinforcement model based on three primary carbon black properties: The effective volume fraction ([V.sub.eff]); the total interfacial surface are with the polymer ([S.sub.T]); and surface activity. In-rubber TEM/AIA measurements were used to determine the polymer occlusion component of [V.sub.eff]. These measurements were found to be superior to colloidal measurement such as DBPA because they account more precisely for aggregate breakdown and distributional variations in the rubber. The use of three-dimensional graphics provided an excellent means of visualizing the relative contribution for the different carbon black properties. [V.sub.eff] and surface activity were the dominant factors associated with increases in static modulus and high strain dynamic elastic modulus. All other properties were also influenced significantly by [S.sub.T] , the magnitude of these effects generally being dependent on the respective [V.sub.eff] and surface activity levels.

References

1. C.R. Herd, G.C. McDonald and W.M. Hess, Rubber Chem. & Technol. 65, 107 (1992).

2. CR. Herd, G.C. McDonald, R.E. Smith and W.M. Hess, Paper presented at a meeting of the ACS Rubber Division in Nashville, TN (November, 1992).

3. J.A. Ayala, W.M. Hess, F.D. Kistler and G.A. Joyce, Rubber Chem. & Technol. 64, 19 (1991).

4. J.A. Ayala, W.M. Hess, A.O. Dotson and G.A. Joyce, Rubber Chem. & Technol. 63, 747 (1990).

5. J.A. Ayala, W.M. Hess and G.A. Joyce, Kautsch. Gummi Kunstst., 44 (5), 424 (1992).

6.. J.A. Ayala, W.M. Hess, G.A. Joyce and K.D. Kistler, Paper presented at a meeting of the A CS Rubber Division in Nashville, TN (November, 1992).

7. A.R. Payne, Rubber Chem. & Technol. 37, 1190 (1964).

8. A.R. Payne and R.E. Whittaker, Rubber Chem. & Technol. 44, 440 (1971).

9. A.I. Medalia, J. Colloid. Interface Sci. 32, 115 (1970).

10. W.M. Hess, R.A. Swor and E.J. Micek, Rubber Chem. & Technol. 57, 959 (1984).

11. J.D. Ulmer, V.E. Chinco, and CE. Scott, Rubber Chem. & Technol. 46, 895 (1973).

12. E.M. Dannenberg, Rubber Chem. & Technol. 895 (1973).

13. G. Krause, Rubber Chem. & Technol. 44, 199 (1971).

14. A.I. Medalia, Rubber Chem. & Technol. 45, 1171 (1972).

15. A.M. Gessler, W.M. Hess and A.I. Medalia, Plast. Rubber Process 3, 4 (1979).

16. G.C. McDonald and W.M. Hess, Rubber Chem. & Technol. 50, 842 (1977).

17. B.B. Boonsira in "Rubber Technology, "M. Morton, Ed., Litton Educational Publishing, Inc. (1973).

18. J.M. Caruthers, R.E. Cohen and A.I. Medalia, Rubber Chem. & Technol. 49, 1076 (1976).

19 A.M. Gessler, Rubber Chem. & Technol. 43, 943 (1970).

20 W.M. Hess, L.C. Ban, F.J. Eckert and V.E. Chirico, Rubber Chem. & Technol. 41, 356 (1968).

21. W.M. Hess, V.E. Chirico, L.L. Ban and J.D. Ulmer, paper presented .at International Rubber Conference, Munich, Germany (1974).

22. W.M. Hess and V.E. Chirico, Proc. Inst. Rubber Ind., 1st European Conf, Brussels, Belgium (1975).

[Tabular Data Omitted]