Alternate approach to study carbon black.

The traditional testing procedures to study the role of carbon black in rubber compounds as well as to characterize it are ambiguous and often lack precise scientific definition.

The use of advanced technologies and concepts has provided a means of overcoming some of these shortcomings and provided new tools to better understand the fundamental nature of the filler.

In this article it will be discussed how Raman spectroscopy, x-ray scattering and fractal concepts were employed to better characterize the carbon black surface. As an example of the usage of these findings it will also be shown that the discrepancy between the results of different "adsorption" techniques can be explained in terms of methodology deficiencies rather than differences in carbon black adsorption behavior.

Carbon black characterization

Conventional parameters

Carbon black is composed of particles (ranging from 10 to 100 nm) solidly fused together to form aggregates which vary in size from 50 to several hundred nanometers. The aggregates are known to associate with one another forming agglomerates which unlike the aggregates are breakable when conventional mechanical mixing energy is used to disperse the compound components. The conventional ways of characterizing carbon black rely mainly on two parameters, namely the "specific surface area" and the so-called "structure."

Specific surface area

This technically well-defined parameter, by far the best reinforcement predictor among a multitude of other attempts to characterize carbon black, has in particular the advantage of characterizing a "fixed" entity of the material. Indeed, the specific surface area, since neither the particle nor the aggregate are usually broken down during mixing, characterizes an "invariant" parameter of the material. During the production process the specific surface of carbon black is "fixed" (geometric factor), as is its intrinsic pseudo-crystalline structure (physicochemical aspect). Given the relationship between a large specific surface area and a small particle size (geometrical aspect) a relationship has been shown to exist[ref. 1] between the "crystallite stacking height" and particle dimension. This finding confers to the specific surface area a physicochemical aspect unique to this filler.

A more recent investigation [ref. 2] using both Raman and WAXS has shown that all carbon blacks obtained using the furnace process possess similar [L.sub.a], the crystallite "crosssection" parameter. In the same study it was shown that by graphitization at 3000[degrees]K the specific surface area was decreasing for all samples whereas no shape or geometrical change could be observed by TEM. One can conclude, therefore, that the adsorption of nitrogen (BET-type experiment) might give an indication of the density of active sites (crystallite edge) on the surface of the sample rather than a "specific surface area."

Indeed, since heat treatment does not alter the geometrical dimension of the particles the change in the "adsorption" of nitrogen cannot be explained according to the general relation

[S.sub.o] = 3/[rho].R

where [S.sub.o] the specific surface area is expressed as a function of the specific gravity [rho] of the carbon black and R the "radius" of the particles. [This relationship is certainly a good approximation since STM (scanning tunneling microscopy) [ref. 3] has revealed that the particles are not interpenetrating but tangent to each other.]

The change of [S.sub.o] with graphitization must therefore have some other explanation. The results of Raman and x-ray studies, which allow a determination of [L.sub.a], the crystallite width perpendicular to the "c" axis, may explain the cause for the decrease in nitrogen adsorption by graphitization. Indeed graphitization increased significantly [L.sub.a] which within the same overall geometry has to be translated by an increase of crystallite size indicating a coalescence of the crystallite by thermal energy. Since the crystallites are larger, with the overall particle surface remaining constant, a decrease of the number of crystallite has to occur. Furthermore, the crystallite edges, areas of high conduction electron density, are preferential adsorption sites and therefore, if the carbon black through heat treatment coalesces its crystallites, fewer crystallite boundaries are present and less preferential adsorption sites are available which may explain the decrease in nitrogen adsorption.

The same reasoning may explain why different grades of carbon black have different amounts of nitrogen adsorbed. Indeed since the same [L.sub.a] was found for all grades of carbon black produced, it indicates that all carbon blacks have the same average size crystallites tiling the surface of a particle. It can be easily shown, if one accepts the TEM measurements of the radius of particles that the larger the radius the larger the amount of crystallite surface per particle [N.sub.c],p, i.e.:

For N110 [arrow right] [N.sub.c],p = 200 (radius = 20 nm)

For N660 [arrow right] [N.sub.c],p = 3,200 (radius = 80 nm) Therefore, in one particle of larger radius where more crystallite edges are available the adsorption of nitrogen per particle has to increase.

On the other hand, the amount of particles per gram of carbon black decreases with increasing particle radius. It can be easily demonstrated that one gram of carbon black containing small radius particles exhibits a larger number of crystallites when compared to a sample containing large radius particles. If [N.sub.c] g is the number of surface crystallites per gram of carbon black, it can be shown that, for example:

For N110 [arrow right] [N.sub.c],g = [3.10.sup.18]

For N660 [arrow right] [N.sub.c],g = [5.10.sup.17] which may explain the results of nitrogen adsorption.

From a more theoretical standpoint a particle surface can be energetically represented as a surface of inhomogenous energy densities with maximum energies at the crystallite edges. Therefore any gas molecule would be preferentially adsorbed at these sites of maximum energy. The initial adsorption will decrease the site energy which may still be more energetic than the surface of the crystallite itself. This site will again preferentially adsorb gases. Only when all edges have decreased their energy by adsorbing gas molecules to the level of the crystallite surface will the gas molecules be adsorbed on the crystallite surface. Figure 1 schematically represents these possible adsorption sequences clearly highlighting that the monolayer hypothesis suitable to the BET theory is not the rule. Therefore, the nitrogen adsorption on carbon black may indicate a density of energetic sites rather than a true surface area.

In fact, material like amorphous round glass possessing a nitrogen specific surface area identical to carbon black does not exhibit reinforcing properties, therefore, the nitrogen adsorption number for carbon black, the best reinforcing potential predictor, may be more indicative of the density of energetic sites (crystallite edges) rather than of "true surface area." This reasoning supports the hypothesis of a filler network being mainly responsible for reinforcement at low strain energy input.

Structure

The so-called "structure" of carbon black, usually expressed as DBP number, is the volume of dibutylphthalate absorbed by a given amount of carbon black ([cm.sup.3]/100g). This parameter characterizes nothing else than the empty space (void) between randomly packed aggregates, i.e., agglomerates. In spite of the fact that this parameter is well controlled and can be relatively easily adjusted during carbon black production, it does not represent an "invariant" since the agglomerates easily change their shapes and spatial distribution during mixing and thus influence the volume of voids between the aggregates in the final compound. It is therefore not surprising, given a set of conditions, that "structure" may sometimes correlate with processing behavior or some final compound property; however, it is far from being a "universal" parameter.

The same oil absorption test can be run on a carbon black sample previously compacted (crushed DBP). This measurement intuitively should be an indication of the average shape of the aggregates. Indeed, one can admit that the more branched the aggregates are, the more empty volume may exist in a compressed sample. This area of research, using a mathematical approach of descriptive geometry on TEM aggregate images taken under different angles of the electron beam has successfully shown [ref. 4] that the carbon black aggregates have an overall tendency to exhibit some degree of planarity. It has also been shown that this degree of planarity is related to reinforcement. Indeed, since the carbon black networking process in rubber is due to Van der Waals type interaction between the crystallite edges (high concentration of conduction electron) of the particle, and since these types of interactions are distance sensitive, it makes sense that the more branched aggregate should develop less interaction and therefore decrease the network cohesion energy [ref. 5].

Carbon black, a fractal object

It has been shown [ref. 6] that carbon black presents fractal characteristics. Studies by small angle scattering (x-ray, neutron) and/or gas adsorption [refs. 7 and 8], show that all carbon black particles have the same surface fractal dimension. Furthermore, the boundary of carbon black aggregates possesses fractal dimensions which can be related to the networking energy of the filler in an elastomeric compound. This aggregate boundary fractal dimension is grade dependent. The more reinforcing the filler, the larger the boundary fractal dimension [ref. 9].

It has also been shown [ref. 6] that the reinforcement of carbon black loaded compound under high strain deformation can be explained by momentum transfers throughout fractal interfaces.

Review of CTAB results

General results

Figure 2 shows a compilation of more than 500 experimental points comparing nitrogen specific surface (ASTM #D4820) area with CTAB (ASTM #D3165) data.

It is interesting to note that below around 90 [m.sup.2]/g (BET) the correlation between the two procedures exhibits an acceptable linear relationship. Above that value the linearity vanishes and the CTAB value gives lower readings than the nitrogen adsorption parameters. In fact below 80 [m.sup.2]/ g (BET) the CTAB results are slightly higher than for nitrogen.

This fact was for a long time attributed to the presence of small pores undetected by the larger molecule of CTAB when compared to nitrogen. Furthermore, it was sometimes hypothesized that this may be also true with rubber which would ignore the pores and therefore the CTAB results were more descriptive of the rubber-filler interaction. Carbon black porosity for tread or carcass grades produced by the furnace process, according to a general consensus of data [refs. 10 and 11] does not exist. These findings cast doubt about the above mentioned hypothesis and the validity of the CTAB measurements.

Considering the new findings on the exact nature of the carbon black particle, the following discussion helps explain that in fact the discrepancy between CTAB and nitrogen measurements is due to the methodology of the CTAB procedure.

Hypothesis

Based on the latest rubber grade carbon black particle model:

* carbon black particles are not microporous;

* carbon black particles are quasi spherical and tangential to each other in an aggregate (no interpenetrating particles).

Aggregate surface = [epsilon] particle surface;

* carbon black particles are tiled with crystallites, whose average width [L.sub.a] is equal for all furnace carbon black. The edges of the crystallites are preferential adsorption sites. Therefore the amount of adsorbed species is proportional to the amount of crystallites.

Summary

The results of the measurement of the adsorption of molecular species are in general expressed as a surface per unit mass of carbon black.

Since, according to the above mentioned hypothesis (facts), the edges of the crystallites located at the filler surface seem of great relevance for molecular adsorption, it is necessary to calculate the amount of these crystallites present at the surface of the particle, as well as their number per gram of filler.

The amount of surface crystallites [N.sub.c],g per gram of filler of particle radius [R.sub.i] is calculated according to the following scheme:

* the outer surface Sp of a spherical particle of radius [R.sub.i] is given by: Sp = 4[pi][([R.sub.i]).sup.2]

* the volume Vp of a particle of radius [R.sub.i] is given by: Vp = 4/3 [pi][([R.sub.i]).sup.3];

* if Sc is the cross section of a crystallite the amount of surface crystallite per particle Nc,p is given by: [N.sub.c],P= Sp/Sc;

* since the weight of one particle is given by: [m.sub.p] = ([V.sub.p]. [rho]), where [rho](g/[cm.sup.3]) is the average specific gravity of the filler, the number of particles per gram of filler is equal to: [N.sub.p,g] = 1/[m.sub.p];

* and the number of surface crystallite per gram [N.sub.c,g] is given by: [N.sub.c,g] = [N.sub.p,g].[N.sub.c,p];

* which expressed as a function of the particle radius [R.sub.i,p] and [S.sub.c] is given by: [N.sub.c,g] = 3/([S.sub.c.p],[R.sub.i]). Surface crystallites per gram of carbon black is a hyperbolically decreasing function of the radius of the particle.

Application to the existing CTAB methodology

The ASTM methodology calls for a calibration curve using different weights of a standard carbon black of a given specific surface area [S.sub.O] (typically (IRB3, 83 [m.sup.2]/g), in order to obtain a relationship between the amount of -adsorbed CTAB and total filler surfaces. In general this calibration curve is very good (5 points linearly distributed with [r.sup.2] better than .999). A given mass of the unknown sample is chosen such that its total approximated surface falls within the range of the experimental points of the calibration curve. The obtained results show some satisfactory correlation with nitrogen specific surface area measurements for samples below approximately 100 [m.sup.2]/g. Above this value the linear correlation vanishes. This behavior is not due to the filler but to an artifact of the methodology as is demonstrated below.

The theoretical approach consists in considering the change of the number of crystallite per gram of carbon black with the change of the particle radius. This is obviously proportional to the difference in the number of crystallite per gram of carbon black. Therefore by taking [R.sub.O] as the radius of the reference sample the relative change [DELTA]N in the number of crystallites is proportional to

[DELTA]N [congruent to] 1/([R.sub.O]) - 1/(R)

As it can be seen from figure 3 this difference is minimal for radius above [R.sub.O] but rapidly increases for values of the radius below [R.sub.O].

This is the reason why the relationship between CTAB and N2SA shows the non linear relationship above a certain specific surface area. This threshold is obviously dependent on the reference sample as the above equation indicates.

Conclusion

The usage of advanced techniques and concepts can be of great help for the everyday tasks of the rubber technologist. This article in particular shows how a better understanding of the exact nature of a material can help provide beneficial insight into its behavior in a complex matrix, i.e., rubber formulation. Also, it addresses the need to avoid misleading and redundant measurement techniques.

References

[1.] Gerspacher, M., Lansinger, C., Paper #7, ACS Rubber Div., April 1988. [2.] Gruber, T., Zerda, T, Gerspacher, M., Paper #29, ACS Rubber Div., May 1993. [3.] Donnet, J.B., Custodero, E., Proceedings of the 2nd Int. Conf. on Carbon Black (p. 177). [4.] Gruber, T., Zerda, T., Gerspacher, M., Carbon, 31, 7, 1209. [5.] Gerspacher, M., O'Farrell, C.P., proceedings 2nd Int. Conf. on Carbon Black (p. 319 ). [6.] Gerspacher, M., O'Farrell, C.P., "Carbon black is a fractal object," Elastomerics, April 1991. [7.] Zerda, T.W., Yang, H.H., Gerspacher, M., ACS Rubber Div., May 1991. [8.] Gerspacher, M., Seeger, P.A., LANSCE Progress Report, LA-11933-PR. [9.] Gerspacher, M., O'Farrell, C.P., Kautshuck & Gummi Kunstoffe, 45, 1992. [10.] Brown, W.A., Ashland Carbon Blackboard, Vol. 5, No. 5, 1982. [11.] Stoekli, F., Donnet, J.B., personal communications.