Filler-filler and filler-polymer interactions as a function of in-rubber carbon black dispersion.

At the beginning of this century, the manufacturing of pneumatic tires was revolutionized by the usage of particulate fillers. Initially to cut costs, these practices very soon imparted improved tire properties. At that time, it was essentially a drastic improvement of wear properties: tire life.

Since these pioneering days, a lot of technology has flown through the robber industry, and at the end of this century, high tech polymers and fillers are currently used in modern tires. These tires are required to have many qualities, i.e., good treadwear, low rolling resistance and superior traction, to mention only the most stringent requirements. These three properties are sometimes assigned to a so-called magic triangle, referring to the concept that if one improves one of these parameters, the others worsen.

In this article, by studying tire compound materials interactions, it is shown that this magic may be broken.

Raw materials for tires

Tires are manufactured by using quite a large number of ingredients, two of which are certainly among the most important, both by their effect on tire properties and volume of materials:

* Polymers (elastomers); and

* particulate fillers (carbon black, silica).

The compounds obtained by mixing these ingredients are, therefore, viscoelastic in nature.

The theory of viscoelasticity, at least in its linear regime, is well understood, modeled and described. It is worth mentioning that in the overall frequency and temperature range in which a tire operates, the elastomer is in its robbery plateau. However, at the interface between the road and the tire tread, at high speed, one can imagine that a very small volume of tread material may be subjected to very high frequency. Therefore, the elastomer may well be much closer to its glassy state. This point is discussed in more detail later. It can be considered that the elastomers used in tire compounds exhibit a quasilinear viscoelasticity behavior (refs. 1 and 2).

For these elastomers, the mechanical response to dynamic strain energy input may be described by the well-known complex modulus G*([Omega]) expressed as a function of an elastic G'([Omega]) and a loss modulus G"([Omega]). ([Omega] = 2[Pi] f, f = frequency).

Carbon black

A large amount of work has been done to characterize, describe and explain carbon black. The reader is referred to the author's review published in 1996 (ref. 3), as well as to the outstanding contributions of others (ref. 4).

In summary, the aggregates which vary in size within and among grades are in general geometrically characterized by a strong spatial mass distribution anisotropy (refs. 5 and 6). The surface of the carbon black essentially exhibits two types of carbon atoms arrangement - turbostratic crystallites and "amorphous," nonplanar carbon domains.

These crystallites and the amorphous carbon have been described and it is generally admitted that the surface's most active sites are located at the crystallite edges as normally accepted in material science.

On the other hand, the modern furnace type carbon blacks used in today's tire industry do not exhibit a large amount of surface groups, if any, and experiments determining the heat of immersion (ref. 8) and the author's results (ref. 3) lead to the conclusion, that in general, furnace grade carbon blacks are not microporous.

Silica

Silica is another reinforcing filler which has become of greater importance for several years. Unlike carbon black, silica' surfaces are covered with silanol groups (Si-OH). These groups have a very strong tendency to form hydrogen bonding between themselves, resulting in a strong agglomeration of silica entities.

This strong aggregation, having a negative effect on final compound properties, necessitates chemical treatment of silica surfaces by silanes in order to allow a good filler dispersion. Adding to the fact that special compounding techniques are required, the usage of silica is more expensive, but has proven to challenge the properties obtainable with more conventional compounding techniques using carbon black.

Tire tread compounds

In order to obtain a tire compound, polymer, filler and sometimes oil are mixed together along with curatives and antidegradant chemicals. The object of this article is to determine the role of the interactions due to the filler and the polymer on the compound viscoelastic properties. It is by understanding these interactions that progress has been made and will be made to fulfill the tire requirements of the beginning of the new millennium.

To illustrate the influence of the raw material interactions on the properties of tire compounds, studying the tread seems judicious since the three most relevant tire properties (rolling resistance, traction, wear) are tread compound dependent.

The overall deformation of the tread depends on factors like tire inflation, tread design, etc. The amplitude of this deformation is, in general, well below 50% strain; whereas, its frequency given by the fire/wheel rotation speed, is well below 50 Hz, typically around 20 Hz or lower. On the other hand, considering the contact area of the tread compound with the road, the deformation of that area of the tread must occur at much higher frequencies. Table 1 gives some values of frequencies that the tread surface may experience in the tire road contact area. It is also hypothesized that at these very high mechanical frequencies the deformation is very small.

Table 1 - tread surface deformation frequencies

Speed      Distance between     Frequency
[km/h]     aspirates [mm]         [MHz]

  30            0.01              2.23
  30            0.05              0.45
  40            0.01              2.68
  40            0.05              0.54

In summary, a tread compound on a rolling tire is subjected to different modes of deformation (ref. 9).

* The whole tread is deformed at relatively low frequency in a strain range well below 50%.

* A very small volume of the tread surface is subjected at the tire/road contact area to very low strain deformation at very high frequencies (MHz).

This clearly indicates that two modes of strain energy input have to be considered to study the tread compound viscoelastic behavior:

* Low frequency over a strain range extending from 0.1% to 50% strain; and

* high frequency (MHz) and low strain.

Because the well known WLF (ref. 1) relationship which holds true for unfilled compound is not readily applicable to filled elastomers, a high frequency low strain acoustic spectrometer was developed for this purpose (ref. 9).

Tread compound viscoelasticity

Low frequency

The piece of equipment used to obtain the results presented here, for strains below 10%, is a Rheometrics SIV, and for strains above 10%, a simple (sandwich) shear device was adapted to a servo-hydraulic tensile tester Instron 8501. Typical experiments were run as strain sweep at 1 Hz and room temperature.

High frequency

A high frequency acoustic spectrometer was built and was described elsewhere (ref. 9). A typical experiment consists of studying the acoustic energy absorbed by a set of rubber specimens of variable thickness in order to determine the characteristic longitudinal absorption coefficient [[Alpha].sub.L] of the material.

Two types of tests at high frequencies are reported here:

* Room temperature at 1 MHz; and

* temperature sweep around the polymer Tg at 1 MHz.

Evaluated rubber compounds

In general, the compound formulations used for these studies were derived from the D3191 ASTM model formulation. Polymer types, carbon black loadings and curing time were varied.

Depending on the studied parameters, both carbon black filled and unfilled compounds were tested in order to evaluate the role of the filler on the filler-filler and filler-polymer interactions. Variations of the crosslinking density were also studied to evaluate its role to the above mentioned interactions, as well as its influence on the polymer-polymer interaction.

Results - low frequency (strain sweep)

Strain range: 0.1% - 10%

Filled (50 phr carbon black) and unfilled compounds all cured at 160 [degrees] C, [t.sub.90] were tested at room temperature over a strain range from 0.1 to 10% strain. It is well known that the linear relationship existing between the torque and the strain for unfilled compounds over the strain range studied does not hold true for the filled compound. Indeed, the departure from the linearity between torque and strain occurs at very low strain when carbon black is present. This indicates that the addition of a filler like carbon black in a linear viscoelastic elastomer results in a non-linear viscoelastic compound.

Figure 1 presents the same results expressed in terms of the modulus (G') (ref. 10) of the tested material as a function of strain [Gamma].

[Figure 1 ILLUSTRATION OMITTED]

The following can be concluded from these initial general results:

* The unfilled compound moduli are strain independent;

* the non-linear filled compound is characterized by a much higher complex modulus than the unfilled compound; and

* if the elastic modulus of the filled compound decreases monotonically with increasing strain, the corresponding loss modulus G" always presents a maximum between 1 and 2% strain (ref. 10).

Both these variations with strain have to be available, and a general representation, called <G-plot>, is proposed as a general expression of the parametric representation G' = [f.sub.1]([Gamma]), G" = [f.sub.2]([Gamma]).

Figure 2 represents such a <G-plot> where a set of coordinates (G', G") correspond to a given strain (i.e., 0.1, 0.5, 1%).

[Figure 2 ILLUSTRATION OMITTED]

This representation will be used throughout this article to describe the low frequency viscoelastic behavior of rubber compounds.

This type of curve is absolutely general for all grades of furnace carbon black. Figure 3 represents a collection of <G-plot> obtained with different carbon black grades in the ASTM D3191 SBR compound formulation.

[Figure 3 ILLUSTRATION OMITTED]

It has been shown that all these <G-plot> are affined (ref. 10). Indeed, regardless of the type of carbon black and the polymer, a master <<G-plot>> can be used to characterize the viscoelastic nature of all carbon black loaded compounds at a given temperature in the rubbery plateau of the elastomer used. Indeed, each carbon black, within a given system, exhibits a <G-plot> which can be derived from the master <<G-plot>> using two orthogonal translations of magnitude [a.sub.i] and [b.sub.i], characteristic of the carbon black. This is an indication that the phenomena responsible for the shape of the <G-plot> are identical for all carbon blacks, regardless of the polymer. Obviously, the magnitude of these phenomena as well as [a.sub.i] and [b.sub.i] is carbon black and polymer dependent (ref. 10).

As hypothesized earlier (ref. 11), the authors (ref. 10) attribute this behavior to the formation of a network by the carbon black itself interpenetrating the polymeric network. It was also shown (ref. 10) that the filler network, when tested at a very low strain (0.1%) can be disrupted by thermal motion. In fact, the carbon black network exhibits a typical temperature dependence with a well marked transition point of the In G' vs. 1/T plot around 90 to 100 [degrees] C. This indicates that the sample passes from a state characterized by a continuous filler network to another in which the network effects vanish due to thermal motions.

Strain above 10%

As shown elsewhere (ref. 12), simple shear sample geometry strain sweeps (0.5 to 50% strain) at 1 Hz and room temperature were conducted. The results of such an experiment, shown in figure 4, indicate that:

* From 0.5 to around 10% strain, the <G-plot> behavior observed is identical to the one obtained in a rotational shear experiment; and

* around 10%, a change of behavior is observed and is a general phenomenon regardless of the type of polymer or carbon black. After 10% strain, the linear relationship between G' and G" always extrapolates to the origin of the axis (G' = G" = 0). This indicates that the type of mechanism responsible for this behavior is unchanged over this studied strain range.

[Figure 4 ILLUSTRATION OMITTED]

The following is proposed to explain the linearity of G" versus G' at strains above 10%. The percolated carbon black network is disrupted and the filler forms a set of subnetworks. These subnetworks are composed initially of a large number of aggregates. However, the separation of the subnetwork is such that they no longer form a percolated structure within the polymeric matrix. It is further hypothesized that upon additional strain energy input these subnetworks are in mm subdivided into smaller sub-subnetworks, etc. (ref. 13).

The way the network breaks into subnetworks, as well as the subnetwork size, must be aggregate morphology dependent.

If one accepts this proposed subnetwork concept, the dissipation of energy can be explained (ref. 14) by the transfer of momentum through fractal interfaces.

A generalized hyperscaling theory (ref. 15) where energy is not only dissipated at an interface, but also on its vicinity, coupled with a generalized theory of diffusion, allows one to evaluate the mechanism of transfer of this momentum throughout the sample fractal interfaces via a fractionary Brownian movement. This concept was further evaluated by Heinrich (ref. 16).

This last observation and all other above mentioned results point to the primordial filler-filler interactions and not to polymer-filler interactions.

High frequencies

As mentioned previously, the acoustic spectrometer is capable of testing compounds in a domain of temperatures comprised between 173 and 340 [degrees] K in the MHz range.

High frequency - room temperature

Setting the spectrometer at 1 MHz and 303 [degrees] K, the following studies were undertaken:

* Effect of carbon black type; and

* effect of polymer types.

The change of carbon black type at a given loading (50 phr) does not characteristically affect the acoustic absorption coefficient [[Alpha].sub.L]. This coefficient varying from .198 to .200 represents only an overall 4.5% variation; whereas, the carbon black samples ranging from carcass (N660) to a tread (N 110) are widely different in size.

High frequencies - low temperature

A temperature sweep at 1 MHz using different polymers shows a shift in Tg of about 50 [degrees] when compared to low frequency results (figure 5).

[Figure 5 ILLUSTRATION OMITTED]

Studies of materials interactions

Using the two frequency domains described above, compounding variations have allowed better understanding of the role of the three following interactions:

* Polymer/polymer;

* filler/polymer; and

* filler/filler.

The following compounding variations have been made in order to assess the role of the key ingredients and their interactions.

Polymer variations

Polymers covering a wide range of microstructure, unsaturation and chemical composition were studied:

* Butyl;

* BIMS (brominated isobutylene co-para-methyl styrene);

* EPDM; and

* S-SBR (solution SBR).

Carbon black variations

Besides variations of carbon black grades presented in a previous part of this article, a variation in the aggregate degree of spatial anisotropy at constant nitrogen adsorption value was evaluated.

Compounding/process variations

Variations in mixing time directly related to the degree of carbon black dispersion were used in order to assess the role of filler-filler network density and also on the filler-polymer interactions. The amount of the latter presumably will increase in quantity along with a better dispersion.

On the other hand, different vulcanization times to vary the crosslink density and/or types were also studied in order to evaluate the influence of the polymer network density on the measured properties.

Results

Polymer variation

Strain sweep experiments at 1 Hz, 30 [degrees] C using the above-mentioned polymers loaded with carbon black show typical <G-plot> responses, in spite of the drastic differences in the degree of unsaturation of the polymer. Figure 6 shows the variation of [[Delta]G'.sub.max] for all the above mentioned compounds. [[Delta]G'.sub.max] represents (% change) the difference in elastic modulus for filled vs. unfilled compounds.

[Figure 6 ILLUSTRATION OMITTED]

In all these cases, by adding carbon black, an increase of more than 95% in [G'.sub.max] is observed, regardless of the polymer type.

The same compounds filled and unfilled were tested at room temperature at 1 MHz by acoustic spectrometry. The results, figure 7, clearly indicate that on the opposite to the low frequency testing, it is the elastomer which plays the most important role as far as the absolute value of the attenuation coefficient is concerned.

[Figure 7 ILLUSTRATION OMITTED]

From the two sets of experiments, it can be concluded that at low frequency, the filler network, hence the filler-filler interactions, are essential, whereas at high frequency, it is the polymer net, hence the polymer-polymer interactions, which are primordial.

Carbon black variations

The degree of spatial anisotropy of carbon black aggregates may be indirectly estimated by the compressed DBP (CDBP) test. The higher the CDBP at a given nitrogen adsorption number, the more branched the aggregate; hence, a more spatial isotropic geometry. Using samples of different CDBP in the vicinity of a nitrogen adsorption number equivalent to 80 [m.sup.2]/g, strain sweeps in the 0.1 to 10% and 0.5 to 40% strain at 1 Hz and 30 [degrees] C were made.

Figure 8 summarizes these results, showing:

* A systematic decrease of G" with increasing CDBP; and

* the high CDBP sample exhibits a significantly lower drop in G' with increasing strain.

[Figure 8 ILLUSTRATION OMITTED]

Experiments from 0.5% to 40% strain show exactly the same trend. Comparing a set of highly branched aggregates to a set of much less branched aggregates, it can easily be postulated that the highly branched aggregates will form a much less dense network and therefore the corresponding subnetworks will be smaller, but in greater quantity. This hypothesis of network and subnetwork size, and therefore quantity, can explain the behavior of carbon black of similar specific surface area, but different CDBP.

From these observations it appears that once again the filler-filler interactions are the primordial ones.

Using the same samples for high frequency acoustic spectrometry, it was shown that the highly branched aggregate confers a slightly higher attenuation coefficient. In this experiment, the strain being extremely low, the filler network is not disrupted and the increase in attenuation is possibly due to a better dispersion of the filler. This would enhance the quantity of filler-polymer interactions, resulting in additional immobilization of polymeric segments explaining the higher attenuation.

Compounding process variations

Mixing time

The variation of mixing time was used to achieve different degrees of dispersion. Low strain, low frequency testing clearly indicates, as shown in figure 9, that a better dispersion of the filler (a more homogeneous distribution of carbon black in the rubber matrix), in other words, a less dense network, reduces the reinforcing effect of the filler. Imparting fewer filler-filler interactions, a better filler dispersion means more surface accessible to the polymer for interactions with the filler, and therefore the decrease of reinforcement indicates that the polymer-filler interactions play a secondary role, if any, in the low frequency region. This observation is absolutely general, regardless of the type of carbon black or polymers.

[Figure 9 ILLUSTRATION OMITTED]

On the other hand, at high frequency, as shown in figure 10, a better dispersion increases the attenuation coefficient. As mentioned earlier, an increase of immobilized elastomer segments (these responding to 1 MHz stimuli) due to a larger probability of interaction with more dispersed filler, may explain the observation. Nevertheless, it is to be mentioned that the change in attenuation due to better dispersion is minimal when compared to attenuation characteristics of the polymer, indicating that in high frequency acoustic spectrometry the polymer-polymer interactions are much more important than the filler-polymer interaction.

[Figure 10 ILLUSTRATION OMITTED]

Carbon black loading

It is well established that increasing the carbon black content drastically changes the low frequency strain sweep. This can be explained as for all low frequency testing by the fact that increasing the filler loading is translated by a "stronger" filler network.

The same variations at high frequency also show an increase in the attenuation coefficient, which may be explained by the fact that the more carbon black is present, the more polymer immobilization (filler-polymer interaction) occurs, but again, this increase is minimal compared to the possible variations of the attenuation coefficient by changing the polymer type (polymer-polymer interactions).

Change in curing time

The change in curing time affects only slightly the <G-plot>, whereas it shows a very net effect on high frequency viscoelasticity. Once again, this phenomenon may be explained by an increase in the polymer immobilization by sulfur crosslinks, and therefore this increase is attributed to polymer-polymer interactions.

Conclusions

If the filler-filler and polymer-polymer interactions seem prevailing, the increase of the filler-polymer interaction may benefit the very high frequency behavior of the compound, as well as the filler dispersion during mixing. Since a better carbon black dispersion decreases the filler network impact on hysteresis at low frequency, but increases the filler-polymer interaction, improving the high frequency behavior, one can hypothesize that improving rolling resistance may also improve traction. In that sense, the "magic triangle" may be overcome.

References

(1.) J.D. Ferry, Viscoelastic Properties of Polymers, J. Wiley & Sons, (New York) (1980).

(2.) N.S Tschoegl, The Phenomenological Theory of Linear Viscoelasticity, Springer Verlag, (New York) (1989).

(3.) Gerspacher, M., O'Farrell, C.P., Nikiel, L. and Yang, H.H., Rubber Chem. Technol., 69, 3 (1996).

(4.) Donnet, J.B. and Wang, M.J., Carbon Black 2nd Edition, Marcel Dekker, (New York), (1993).

(5.) Gruber, T.C., Zerda, T.W. and Gerspacher, M., Rubber Chem. and Technol. 67, 280 (1994).

(6.) Rinewalt, R., Payne, L.D., Morgenstern, C.A. and Gerspacher, M., Rubber Chem. Technol. 68, 169 (1995).

(7.) Zerda, T.W., Wei, Xu, Yang, H.H. and Gerspacher, M., ACS Rubber Div. Spring Meeting in Anaheim (1997).

(8.) Donnet, J.B., private communication.

(9.) Gerspacher, M., O'Farrell, C.P., Nikiel, L. and Yang, H.H., Rubber Chem. Technol. 69, 786 (1996).

(10.) Gerspacher, M., Yang, H.H. and Starita, J.M., Carbon black networking, L'Actualite Chimique, March, 1990.

(11.) Medalia, A.I., Rubber Chem. Technol. 51, 437 (1978).

(12.) Gerspacher, M., O'Farrell, C.P., Yang, H.H. and Wampler, W.A., ACS Rubber Div. Spring Meeting Montreal (1996).

(13.) Gerspacher, M. and O'Farrell, C.P., Kautschuk Gummi, Kunststoffe, 45.2 (1992).

(14.) Le Mehaute, A., private communication.

(15.) Le Mehaute, A., Fractal Geometries, CRC Press, London (1991).

(16.) Heinrich, G., et al., Polymer Networks Blend 5(4) 199-204 (1995).