The influence of carbon black morphology and pellet properties on macro-dispersion.
In this article, a brief review of some performance problems associated with poor dispersion quality is provided. Two methods for objectively and quantitatively assessing dispersion quality, in terms of undispersed materials greater than about 10 microns in size, are then described. The methods described are then used to characterize the effects of black morphology and pellet strength properties on dispersion quality.
Complaints attributed to dispersion quality
Common complaints in the rubber industry which are often related to dispersion problems can be classified into four major categories: product performance, surface defects, surface appearance and dispersion efficiency. These categories together with some possible carbon black performance defects leading to them are summarized in table 1. As discussed below, some of these complaints do not, necessarily, represent dispersion issues.
Table 1 - common carbon black related dispersion problems Customer complaints Possible carbon Types of carbon black attributes black problems Performance Grit Cleanliness Undispersed black Pellet quality Colloidal properties Colloidal properties Defect Grit Cleanliness (Isolated event) Undispersed black Pellet quality Colloidal properties Appearance Grit Cleanliness (Global event) Undispersed black Pellet quality Colloidal properties Colloidal properties Dispersion efficiency Fines Pellet quality Pellet strength Pellet quality Pellet size Pellet quality Colloidal properties Colloidal properties
The functional performance and durability of a carbon black containing rubber formulation, such as tensile strength, fatigue life and wear resistance, are affected substantially by dispersion quality (refs. 9-11). The possible carbon black attributes which may degrade product performance are its grit content (here, defined as hard, undispersible contaminants such as coke balls, inorganic oxides and metallic particles), presence of hard-to-disperse agglomerates (undispersed black) or poor choice of morphological characteristics (surface area and structure). Strictly, the last characteristic does not represent a dispersion issue.
The possible carbon black related causes of surface defects on finished products are mainly grit and undispersed black. Grit is a contaminant and, hence, represents a cleanliness rather than a dispersion problem. Undispersed black, on the other hand, can cause visible defects and, as implied by the terms employed, is associated with poor dispersion quality. This, as will be discussed, results from a deficiency in the mixing operation. Eliminating the presence of surface defects is of critical importance in molded thin parts for functional reasons and in extruded profiles for both aesthetic and functional reasons.
Undesirably rough appearing surfaces can be caused by grit, undispersed carbon black, sub-optimal carbon black loadings and/or low carbon black structure and surface area. The preferred black loading depends on black morphological properties. Sub-optimal loadings can result in high compound extrusion shrinkage (die swell) which can then give a rough surface (localized shrinkage). In general, higher loadings of low structure, low area blacks are required more than their high structure, high area counterparts. In addition, at given loading, the shear stresses generated during the mixing operation in formulations containing blacks with the former set of characteristics are smaller than those generated with the latter blacks. Under such conditions, the shearing stresses generated for compounds containing low structure, low area carbon blacks may be insufficient to attain adequate dispersion quality which results in the presence of undispersed black.
The mixing operations themselves have a direct impact on mixing efficiency and, hence, on dispersion quality. For the same mixing equipment and mixing conditions, dispersion efficiency is affected by both carbon black morphology and carbon black pellet properties such as fines level and pellet strength.
Carbon black dispersion mechanism
When carbon black is made in a reactor, it is in a "fluffy" form that is difficult to transport and handle during usage. Therefore, carbon black is normally densified or pelletized in order to improve bulk handling characteristics. On the other hand, the mixing operation reverses the agglomeration process by breaking up the pellets and uniformly distributing the individual particles throughout the compound using the hydrodynamic and mechanical forces generated in the mixer.
There are many mathematical models in the literature (refs. 12-15) that attempt to describe the dispersion mechanism of carbon black in various media such as ink, plastics and rubber systems. Although the mathematical formulations can be different, the fundamental concepts are similar. In general, two of the most widely accepted models describing dispersion mechanisms are the cleavage and erosion models. In reality, both mechanisms probably occur simultaneously during the mixing operation, depending on how the pellets or agglomerates respond to the hydrodynamic forces generated in the mixer.
These concepts, in simplified form, are summarized in table 2. Filler size reduction, by either cleavage or erosion mechanisms, will take place as long as the hydrodynamic forces (shear and/or elongational forces) generated during mixing are greater than the interaggregate attractive forces. The magnitude of the hydrodynamic forces depends on the product of compound viscosity and the shear rate generated by mixer. The shear rate is determined by the mixing equipment and mixing conditions, and is independent of the type/amount of carbon black or rubber. Compound viscosity is affected by many factors including polymer type, mixing conditions, carbon black loading and carbon black morphology.
Table 2 Carbon black dispersion mechanism Mixing shear force > pellet strength Mixing shear force = Viscosity x shear rate Shear rate = f (mixing equipment, mixing conditions) Viscosity = f (polymer, mixing conditions, carbon black loading, carbon black colloidal properties) Pellet strength = f (interaggregate strength, binder) Interaggregate strength = f (pelletization process, carbon black collioidal properties)
Pellet strength depends on the attractive forces between carbon black aggregates and, if used, with the binder type and level. In general, the pellet strength is dependent, in a complex manner, on the pelletizer type, pelletization conditions and carbon black morphology. If the hydrodynamic forces generated in the mixer are held constant, the only important factor which can affect dispersibility is pellet strength.
In this study, natural rubber was compounded with the requisite amounts of carbon black in a BR internal mixer. The compounds were cured at 160 [degrees] C for 20 minutes, then were microtomed for dispersion measurement. EPDM based formulations were first internally mixed and then extruded. The uncured extruded EPDM tapes were measured on TSDES for surface defects evaluation. The cured EPDM tapes were evaluated using 3D profilometry for surface roughness characterization.
Measuring macro-dispersion by image analysis
A commercial image analyzer was used as a tool to measure macro-dispersion. In the development of a quantitative macro-dispersion test, a knowledge of the critical cut-off size is of importance. In most instances, it is defects larger than about 10 microns in size, consisting of undispersed black, grit and other contaminants, which affect both visual and functional performance. The approach taken here is to characterize defects on a surface (generated by microtoming, extrusion or cutting) greater than 10 microns in size by number and by area per unit area examined using an image analysis procedure. In other words, dispersion quality was assessed by evaluating the following parameters:
Undispersed counts/[cm.sup.2] = [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
% Undispersed area (%) = [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where Am = total sample surface examined; [N.sub.i] = number of defects with size D; [D.sub.i] = diameter of circle having the same area of the defect (equivalent circle diameter).
The undispersed counts per unit area and percent undispersed area were evaluated using automated image analysis procedures. Furthermore, two histograms, undispersed counts/[cm.sup.2] vs. size and percent undispersed area vs. size, can also be generated to provide the detailed information on dispersion quality. Three methods, depending on how a surface was presented for image analysis, were used to characterize dispersion quality. These are discussed in turn.
Macro-dispersion in NR was assessed using microtomed samples. Typically, five randomly selected optical images were taken of the microtomed sample for image analysis (ref. 16). Knife marks were removed using a numerical filtering technique. Areas containing wrinkles and other microtoming defects were not used in the analysis. In this work up to 10 images were analyzed to enhance test precision.
Extruded tape surfaces
Evaluation of dispersion quality for EPDM compounds has been a major challenge in the rubber industry. EPDM is widely used in extrusion, hose and roofing applications, for which surface quality and extrusion appearance are critical. Microtome samples are very difficult to prepare for those EPDM compounds which typically contain high carbon black and oil loadings. Due to the different nature of surface roughness and defects, they must be characterized using different techniques. Surface roughness can be measured using a stylus type profilometer, which will be discussed in more detail later. The other technique utilizesimage analysis to analyze surface defects.
A large area of the sample has to be examined in order to obtain statistically meaningful results. We have developed a method using image analysis techniques to continuously examine the surface of extruded EPDM tapes. The system is termed "tape surface defects evaluation system" or TSDES. In the system, the tape is continuously driven by a set of nip rolls, which apply enough tension to the tape to maintain a flat surface. A reel of non-stretched plastic film is used as the carrier. To insure uniform lighting of the field of inspection, the tape is illuminated by a ring light attached to the end of the camera lens. A video camera constantly scans the tape and transmits the signal, first to a TV monitor, and then to the image analysis system. The custom designed software processes and enhances the grabbed images. By using numerical filtering techniques, the die lines are removed, the background is smoothed out, and only the surface defects are enhanced for analysis. This cycle is then repeated for a pre-determined number of times (usually 50 picture frames are analyzed). Although not used here, the system can also measure both black and white defects (ref. 1).
The TSDES allows the operator to select any cut-off size larger than 10 microns, which is predetermined prior to evaluation. The percent undispersed area and undispersed counts strongly depend on the cut-off size selected. Since TSDES is a continuous measuring system, the operator can measure as many pictures as needed in order to obtain statistically meaningful results. The running average of percent defect area is shown on the screen all the time, which can help the operator decide whether more pictures are needed to obtain statistically meaningful results.
The reproducibility of TSDES measurement was investigated. Five batches of a rubber compound were mixed at different times to check the batch to batch variation. In addition, five different tapes from batches 3, 4 and 5 were evaluated to check variations between batches. As shown in figure 1, for the undispersed counts/[cm.sup.2], and in figure 2, for the undispersed area, the batch to batch variation is small, while the reproducibility within the batch is excellent in terms of both percent undispersed area and undispersed counts. These results show that TSDES provides reproducible measurements for fresh tapes. In some cases, however, aging of the tapes can produce inconsistent results. This is attributed to the relaxation of the rubber and blooming (migration) of oil and other chemicals. The aging effects are formulation dependent, and care has to be taken when evaluating highly oil loaded EPDM which can frequently cause problems.
[Figures 1-2 ILLUSTRATION OMITTED]
Like microtome samples, cut surfaces can provide information regarding the dispersion quality inside of the compound (ref. 10). A cutting tool has been developed which achieves clean cut surfaces (minimum knife marks) for bulk dispersion evaluations. The cut sample is then put under the ring light and camera of TSDES and the macro-dispersion parameters are automatically calculated. The advantages of using this sampling method are ease of preparation and reduced polymer sensitivity. For example, it can be used for all rubber systems. Photographs of typical cut-surfaces are shown in figure 3. In general, there are no knife marks on the surface to interfere with the measurements. In this study, up to 10 images were analyzed to enhance test precision.
[Figure 3 ILLUSTRATION OMITTED]
Use of s stylus type profilometer to measure surface roughness is not a new concept (refs. 10 and 16). The method is based on tracking a fine stylus over a sample surface and measuring the topography. However, most available profilometers used in the past could only do line scanning and generate a one dimensional roughness profile. The results were interesting and useful, but they could not describe the characteristics of the entire surface.
A commercially available 3D profilometer having a diamond stylus was used to characterize surface roughness. Surface scanning is achieved by two stepping motors (up to one micron resolution) disposed perpendicularly. Dispersion quality was characterized from the surface roughness measurements using [R.sub.a] (center line average) and [R.sub.q] (root mean square average) values defined as:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
[bar] Z is the average height of all measured points;
[Z.sub.i] is the height of point i;
N is the total number of measured points.
In general, as macro-dispersion quality is degraded, higher values of [R.sub.a] and [R.sub.q] are obtained.
Results and discussion
The parameters of undispersed counts per unit area and percent undispersed area can be used to assess dispersion quality at the selected cut-off size. Accordingly, these two parameters can be used to develop the dispersion chart depicted in figure 4, which indicates that as dispersion quality is degraded, both the undispersed counts per unit area and percent undispersed area increase. As shown in the figure, dispersion quality can be improved by reducing the undispersed counts per unit area, the percent undispersed area or both. As will be shown, changes in these parameters often are proportional to each other.
[Figure 4 ILLUSTRATION OMITTED]
Comparisons with the Cabot dispersion rating
The Cabot dispersion chart procedure for rating dispersions has been used for many years to evaluate dispersion quality (refs. 6 and 16). In this method, photomicrographs of a surface are rated against a standard chart. A letter rating, going from A to H, is assigned as the undispersed area progressively increases and a number rating, from 1 to 6, is assigned as the number of defects increases. The rating system employed is summarized in table 3. The method is subjective and semiquantitative.
Table 3 - letter and number ratings of dispersion chart Letter % Dispersion area Number Size (microns) A 100-99.7 1 0-19 B 99.7-99.1 2 19-33 C 99.1-98.2 3 33-57 D 98.2-96.5 4 57-98 E 96.5-93.7 5 98-170 F 93.7-88.6 6 170-295 G 88.6-80.0 H 80.0-0
Photomicrographs of microtomed surfaces rated by the present procedure and the chart method are shown in figures 5 and 6. The software automatically generates the dispersion ratings. Those for cut surfaces are shown in figures 7 and 6. The advantage of the present dispersion rating method is that it is quantitative and operator independent.
[Figures 5-7 ILLUSTRATION OMITTED]
Relationship between undispersed area and undispersed counts
In general, and as already mentioned, the undispersed counts and the undispersed area are often proportional. This effect is shown in figure 9 where the dispersion quality of a series of NR compounds containing different types of carbon blacks at two levels of carbon black loadings, 40 and 80 pier, were evaluated. The ratings for each sample and the letter ranges are also included in the figure. A linear relationship between undispersed area and counts gives a reasonable representation of the data at each loading. Such a relationship is empirical and not always observed.
[Figures 9 ILLUSTRATION OMITTED]
3D profilometry of EPDM tapes
Surface roughness and surface imperfections (defects) are of major concern in EPDM based products. The dispersibility of carbon black can have a profound effect on surface characteristics. The topography attained with a conventional and improved black in an extruded EPDM tape is shown in figure 10. The superior surface attained with the improved black, as shown quantitatively by the reductions in the values of [R.sub.a] and [R.sub.b], is apparent.
[Figure 10 ILLUSTRATION OMITTED]
Carbon black pellet property and morphology effects
Studies have shown that both carbon black pellet properties and carbon black morphology affect dispersibility (refs. 18 and 19). The effect of pellet properties on dispersibility was investigated using TSDES. For these purposes, a standard carbon black, having a dibutyl phthalate, DBP, absorption value of 132 cc/100 g carbon and a surface area of 53 [m.sup.2]/g, was pelletized under widely differing conditions and binder levels so as to form soft, weak and hard, strong pellets. The soft pellets were designated as sample S and the hard ones as sample H. The samples were dispersed in EPDM under identical conditions and black loadings at several mixing times and then extruded. The percent undispersed area and undispersed counts/[cm.sup.2] were determined for defects greater than 10 microns in size. The results obtained are depicted in figure 11. As expected, as the mixing time increases, dispersion quality improves for both sets of samples. The dispersion quality attained with the soft pellets (sample S) at the same mixing time is superior to that obtained with the hard pellets (sample H). The histograms of defect size distribution after a four minute mixing time for both sets of samples are shown in figure 12. They demonstrate that the undispersed counts/[cm.sup.2] of the compound produced using the harder pellets is higher over the whole defect size range. The effects of progressively increasing pellet strength (or hardness) was also investigated by 3D profilometry. In this case, soft, intermediate and hard pellets were studied. The products were dispersed in EPDM under identical mixing conditions and black loadings and then extruded. The topography and [R.sub.a] and [R.sub.q] values of the surfaces of the resulting tapes are shown in figure 13. The data demonstrate that surface quality is degraded as pellet hardness is increased.
[Figures 11-13 ILLUSTRATION OMITTED]
From the above results, it is clear that the carbon black dispersibility can be affected by pellet strength in a complicated manner, even for carbon black with the same carbon black morphology.
TSDES was used to investigate the dispersion quality attained with commercially formed pellets having different morphologies. As summarized in table 4, the carbon blacks studied had surface areas ranging from 25 to 55 [m.sup.2]/g and DBP values ranging from 105 to 135 cc/100 g carbon. The pelletized blacks were mixed in EPDM for four, six, eight and ten minutes. To minimize the effects of differing shearing stresses during the dispersion process, the carbon black loadings in the formulations were adjusted so that the compound viscosities for all the mixtures were similar. The percent undispersed area, undispersed counts/[cm.sup.2] relationships were measured using a 10-micron cut-off, a 40-micron cutoff and a 50-micron cut-off. A comparison of the data indicates that relative dispersion quality depends on the cut-off size selected. However, since the samples were obtained from different plants using somewhat different pelletizing conditions, it is difficult to separate morphological and pelletizing effects.
Table 4 - Carbon morphology parameter of different carbon blacks Carbon black sample I J K L M CTAB ([m.sup.2] /g) 55 55 45 36 25 DBPA (cc/100 9 carbon black) 105 135 120 120 135
The effect of cut-off size on dispersion quality is further illustrated by the histograms in figure 14 which show the relationship between percent undispersed area and defect cutoff size at the four minute mixing time. The histograms show that the distribution of defects differs for the different blacks. For example, the contributions of small defects to the total undispersed area for samples K, L and M are substantial. In contrast, the defects in the size range of 90 to 100 microns make a major contribution to the undispersed area of sample I. Sample J has a relatively uniform distribution of defects. The undispersed counts/[cm.sup.2] in various defect size ranges (greater than 50 microns) for the samples are depicted in figure 15.
[Figures 14-15 ILLUSTRATION OMITTED]
The total percent undispersed area for each black decreases with mixing time. The effect of mixing time on the percent undispersed area as a function of defect size is illustrated for samples M and I by the histograms in figures 16 and 17. The figures show that the undispersed area declines with time but the dominant defect size is insensitive to time. Analogous trends were obtained for the other samples. The reason for the insensitivity of the dominant defect size to mixing time is not known but may be related to the time required for the stronger agglomerates, having comparable strengths, to pass through the more intense shearing zones in the mixer. Although not the subject of this article, the histograms also suggest that rate constants for macro-dispersion can be evaluated
[Figures 16-17 ILLUSTRATION OMITTED]
Two new techniques, based on automated image analysis and on 3D profilometry, have been developed for assessing dispersion quality. The defined dispersion quality parameters, undispersed counts/[cm.sup.2] and percent undispersed area derived from image analysis and [R.sub.a] and [R.sub.q] derived from profilometry, were used to characterize macro-dispersion levels in rubber. The methods are quantitative, reproducible and free of operator bias.
The two methods were used to characterize the surfaces of carbon black filled rubbers. The results demonstrated that dispersibility was affected by pellet hardness, with harder pellets degrading dispersion quality. Carbon black morphology, as determined by surface area and DBP, also affected dispersibility. The relationship between morphology and dispersion quality was dependent on the defect size studied. For defects above about 40 microns in size, dispersibility improved as carbon black DBP was increased and surface area was reduced. Defect area was decreased with increasing mixing times. The dominant defect size, however, was insensitive to mixing time.
[Figure 8 ILLUSTRATION OMITTED]
[1.] W.M. Hess and J.W. Wiedenhaefer, Rubber World 186, no. 6, 15 (1982).
[2.] W.M. Hess, Rubber Chem. Technol. 64, 386 (1991).
[3.] C.H. Leigh-Dugmore, Rubber Chem. Technol 29, 1303 (1956).
[4.] W.F. Ford, J.E. Callan and W.M. Hess, Rubber Age 92, no. 5, 738 (1963).
[5.] G. Cotton and L. J. Murphy, Rubber Chem. Technol. 61, No. 4, 609 (1988).
[6.] All. Medalia and D.F. Walker, Cabot Technical Report RG-124.
[7.] B.R. Richmond, paper no. 68, ACS Rubber Div. Meeting, Denver, May 1993.
[8.] B.R. Richmond, paper no. 158, ACS Rubber Div. Meeting, Orlando, Oct. 1993.
[9.] E.M. Dannenberg, Eur. Rubber J. 167 (1), 25 (1985).
[10.] A.Y. Coran and J.B. Donnet, Rubber Chem. Technol. 65, 973 (1992).
[11.] W.M. Hess, V.Z. Chirico and P. C. Vegvari, Elastomerics, 112, no. 1 (1980).
[12.] S. Shiga and M. Furuta, Rubber Chem. Technol. 58, 1 (1985).
[13.] I. Manas-Zloczawer, A. Nir and Z. Tadmor, Rubber Chem. TechnoL 55, 1250 (1982).
[14.] I. Manas-Zloczower and D. L. Feke, Int. Polymer. Process 4, 3 (1989).
[15.] A.Y. Coran and J.B. Donnet, Rubber Chem. Technol. 65, 998 (1992).
[16.] ASTM D2663 Methods.
[17.] P.C. Vegvari, W.M. Hess and V.Z. Chirico, Rubber Chem. Technol. 51, 817(1978).
[18.] B.B. Boonstra, Rubber Chem. Technol. 50, 194 (1977).
[19.] W.M. Hess, R.A. Swor and E.J. Micek, Rubber Chem. Technol. 57, No. 5, 959 (1984).