Dispersion measurements using video techniques.

The measurement and analysis of carbon black dispersion has been an important focus in the development and support of sulfur cured rubber products. Various microscopy and analytical methods, including the Leigh-Dugmore technique, have succeeded in giving both qualitative and quantitative dispersion results, taking into account the percentages of undispersed agglomerates. More recently, researchers have used CCD video cameras and image processing to more quickly and objectively analyze filler dispersion. The two methods this article will focus on are a modified AMEDA method and a reflected light automatic dispersion tester. The first method uses a sample preparation system involving sulfur hardening and polishing requiring extensively trained technicians and long sample preparation time. The reflected light method uses a razor blade cut specimen, and can be tested on cured and uncured rubber in a matter of minutes. A comparison of results from multiple compounds using both approaches will be described, as well as the value of rapid testing and interpretation of dispersion data. This article provides a positive step toward the resolution of two of the major challenges in dispersion testing, including magnification requirements and fast quantitative measurements. Both of these have been made possible because of technological developments in CCD cameras and image processing, which provide the resolution, detail and accuracy necessary.

Background

Modern dispersion testing methods'

Beginning in 1909, rubber and carbon black manufacturers have tested for dispersion in order to optimize mixing and to better understand the mixing performance properties of compounds and rubber products. These first researchers started with a torn rubber sample and evaluated it with the naked eye. With the advent of electron microscopy in the early 1950s, the focus shifted to analyzing filler aggregates in the sub micron range (in order to closely recognize filler morphology and structure) as well as to studying in the micron region (in order to more generally characterize a compound). The methods used for studying aggregates are SEM and TEM. Both of these methods are applicable on a nanometer scale. More commonly used in quality control are optical methods, micro roughness, interferometric microscopy (ref. 1) and CCD cameras with image processing. Though standards have been adopted, a wide range of test methods is presently being used. In addition to the analysis at various magnifications, many methods remain subjective, time consuming and qualitative. Leigh-Dugmore in 1956 expressed the concern that electron microscopy is "unwieldy" and that other visual methods of inspecting dispersion were not quantitative (ref. 2). The situation remained unchanged for the next four decades. Gerspacher, et. al., of Sid Richardson explained in 1997 that while optical microscopy could be used effectively for quality control, all known methods failed to give useful results on the aggregate level (ref. 3). With electron microscopy, however, he recognized the cost, time and difficulty involved, noting that optical methods were desirable for analysis of agglomerates. Richmond stated (ref. 4) that, "dispersion is considered adequate when 95% of the agglomerates are below 10 microns." Although the threshold is slightly different from the five micron threshold designated by Leigh-Dugmore, it is a similar approach, and one that has been widely accepted in recent years. This distinction has led to further developments in techniques to analyze dispersion in the micron range.

ASTM D2663 method B defines the agglomerate count method which uses the Leigh-Dugmore formula and the microtoming sample preparation process (ref. 5). The formula is:

Dispersion, % = 100 - SU/L

Where: U = number of graticule squares that are at least half filled with carbon black; L = volume percentage of carbon black in the compound; and S = swelling factor of the specimen based on area.

Eliminating subjectivity has required the use of image processing. Kudance of BF Goodrich used Leigh-Dunmore's transmission microscopy approach to calculate undispersed carbon black in order to give a quantitative dispersion rating. Kudance developed AMEDA, which used a computer to automatically calculate percent dispersion (ref. 6). Analysis could occur very quickly, even though sample preparation could be time consuming.

By the 1980s, it was clear that two fundamental approaches to evaluating dispersion in the micron range were being utilized, including AMEDA and the comparison of magnified surfaces to reference photographs. Even more importantly, the development of digital photography and image processing enabled two further analytical advances, automatic comparison and rapid quantitative ratings.

Reference comparisons

In many rubber factories, visual comparison to reference photographs remains a common, and in some ways effective, method for measuring dispersion. The Phillips scale (ref. 7), with its ten reference photographs, has been used when comparing a carbon black filled compound at 30x magnification (see ASTM D2663, Method A). In the 1980s, Persson developed a system where a razor cut specimen could be placed on a split screen and compared side by side with the Phillips references (ref. 8). This helped in standardizing the simple sample preparation, and provided uniform lighting, a previous problem in Phillips comparisons. The ISO Method 11345 addressed the use of reflected light and reference photographs (ref. 9). The 30x magnification was an advantage over the naked eye in terms of ability to detect agglomerates in the 50 to 120 micron range, and, in many ways, was ideal in detecting roughness that created performance problems for rubber manufacturers. These problems included defects in extrusions, hoses and parts with rubber to metal adhesion (ref. 10). The reflected light method had the disadvantage of requiring the subjectivity of the operator. The method did, however, provide a standardized set of procedures and a group of references.

While the 30x comparisons remain an important way of determining surface roughness for many types of compounds and products, new optical technology enabled objectivity in dispersion measurements. With the use of a CCD camera and software algorithm, comparisons can be done automatically, eliminating subjectivity, while keeping the rapid test times and easy sample preparation.

By 1997, resolution strength and computing power made it possible to go beyond the traditional 30x comparisons, and look at smaller agglomerates through the higher magnification of 100x. Looking at agglomerates that go down to three microns, it became useful to compare not only one set of reference photographs, but to compare a potentially wider range. Five reference scales, each with magnifications on one end of the one to ten scale, were added to a unit with the reflected light method. This range was comparable to more detailed development work that had been done which looked at agglomerates in the five micron range.

The CCD camera

The charged coupled device (CCD) has become the most common technology used for digital imaging. Since its invention at Bell Laboratories in 1970, researchers began using the CCD camera as a scientific tool long before consumers took advantage of digital cameras. For 10 years, the main users of CCD cameras were professional astronomers, including NASA. They appreciated the high linearity and large dynamic range offered by the CCD. By 2000, the CCD camera was used in more diverse industry applications, including detecting defects on metal parts and distortions in paint of automobiles.

The CCD camera performs four tasks to generate an image: Generate charge; collect charge; transfer charge; and detect charge.

The CCD has photosites arranged in columns and rows. All of the photosites present in a CCD contain a photodiode that converts photons into electrons, thereby creating a charge. The amount of light directly relates to the amount of electricity that is discharged. Light is collected over the entire chip and then transferred to the next cell within the column. This is known as parallel shift. Each row of data is then moved to a separate horizontal charge register. Charge packets are read out serially and sensed by a charge-to-voltage conversion. Once the process of conversion and storage is complete, the voltage is digitized into an intensity value (ref. 11).

Testing was done on three different types of equipment using CCD imaging: ConCarb dispersion system, disperGRADER 30x and disperGRADER 100x. Percent dispersion and qualitative measurements automatically comparing the sample to reference photos were used.

ConCarb dispersion system

The ConCarb method involves a sample preparation, where a sample cylinder of 2.5 centimeters in diameter by 1.25 centimeters was molded in a curing press. The cylinder is placed in molten sulfur for 24 to 48 hours (depending on the recipe) at 155[degrees]C, where it produces a 0.25 centimeter hardened surface layer. The cylinder is polished and lubricated with water to clear away all grindings. The entire process requires three stages of grinding, as well as a final polishing with alumina and a microcloth (ref. 4). The final sample to be analyzed is circular, and divided into four distinct regions. Five random measurements are taken in each region, and at the center of the specimen.

A light source reflects 20% of the light into an eyepiece, and 80% to the camera. The areas that contain well dispersed carbon black are darker in color, while those that have agglomerates reflect more, and are therefore brighter. The system uses thresholding, and evaluates the presence of agglomerates on a scale from white to black. Using a modified Leigh-Dugmore formula, percent dispersion is calculated. In total, the time to prepare, test and analyze a sample from the time it is press cured takes a minimum of 48 hours. Actual analysis takes approximately 30 minutes and is performed by extensively trained technicians.

DisperGRADER

The disperGRADER 30x and 100x provides a faster way of obtaining a similar percent dispersion, as well as an algorithm to compare the sample to reference photographs. The disperGRADER 30x compares the samples to the Phillips Scale. It can use either cured or uncured rubber, with only a slight difference in sample preparation. An uncured sample is taken hot from a mill, having undergone only its normal mixing procedure. It is placed in a sample pressing and cooling unit where it remains for two minutes to remove all porosity. The sample is then pulled bi-directionally approximately 10% while being cut by an industrial razor blade (figure 1) (ref. 12). This last step is similar to the process for preparing a sample of cured rubber, with the cured rubber being pulled less than 2%.

The sample is then placed in front of an aperture (9 mm x 5 mm) at 30x magnification or (4 mm x 3.5 mm) at 100x magnification. An incandescent light at a 30 degree angle allows the CCD camera to pick up bumps in the surface. Using a similar threshold method, it distinguishes those areas that are flat (as black), from bumps (as shades of gray to white). The bumps, or white area, as it is seen after a scan, are caused from the presence of agglomerates. Since the agglomerates are harder than the polymer matrix, they return to the surface after a sample is cut. The disperGRADER only looks at the agglomerates as covered by a sheet of polymer. These agglomerates are referred to as "nodges."

A histogram of frequency as a function of nodges is determined. From that. an area histogram is created and percent dispersion calculated. The disperGRADER uses the following formula:

Drl, dispersion, % = 100- 100UV/(FL)

Where: U = the total area of the nodges greater than 23 [micro]m; V = volume fraction of carbon black in the nodge; F = the area of the specimen examined; and L = the volume of carbon black in the compound.

By changing the fill factor, a more representative rating may be determined, similar to the approach taken by Medalia (ref. 13). He understood that an agglomerate of carbon black is not made fully of carbon black.

Experimental

An experiment was done on compounds with a wide range of dispersion, in order to compare the three methods. The ConCarb method was chosen for its precise ability to analyze agglomeration, and because it is an accepted scientific tool used by Continental Carbon. The disperGRADER 30x comparisons, like all disperGRADER tests, are fast and objective, and compare as well to the long established Phillips scales. The disperGRADER 100x percent dispersion is used because it is quantitative and uses the same calculations as the ConCarb method.

All of the methods used for the experiment involve different approaches to sample preparation, but use the same theory and basic formula used by Leigh-Dugmore. The disperGRADER, using a razor blade to cut the sample, requires no solvent; therefore, the swelling factor is eliminated. There is no new vocabulary or reference when looking at percent dispersion.

The compounds used for the experiment were SBR/BR blends.

Results and discussion

ConCarb percent dispersion and disperGRADER percent Dispersion

The values of the tests are shown in figure 2 for both the ConCarb and the disperGRADER percent dispersion tests. DisperGRADER tests were performed at 100x magnification. A strong correlation coefficient of 0.978 exists, with results ranging from 66% to 98% from the disperGRADER and 54% to 93% for the ConCarb. While both methods give valuable quantitative results, the disperGRADER method was able to be run quickly, allowing for greater productivity.

disperGRADER 30x ISO scale

Results for the comparisons to the ConCarb method are shown in figure 3. A large range is clear from 3.8 to 7.4. It should be noted that these results are qualitative on a scale of 1 to 10.

Summary

In evaluating the comparison between the disperGRADER and ConCarb methods, it was found that by using a Leigh-Dugmore formula and CCD imaging, similar results were able to be generated, regardless of the great difference in sample preparation and test method. These results demonstrated that for this type of analysis, it is no longer necessary to spend days preparing samples, but instead only minutes, testing cured or uncured rubber. While quantitative measurements are desirable for many purposes, a visual image with automatic comparison at the lower magnification of 30x can give information useful for many rubber products. With a simple and objective test, it is now possible to obtain all relevant data on cured or uncured rubber with great speed.

References

(1.) A.P. Smith, T.L. Aybar, R.W. Magee and C.R. Herd, "Carbon black dispersion measurement in rubber vulcanizates via intecferometric microscopy," presented at the Rubber Division, ACS, October 2003.

(2.) C.H. Leigh-Dugmore, Rubber Chem. & Tech., 29, pp. 1,303-1,308, (1956).

(3.) M. Gerspacher, L. Nikiel, H.H. Yang and C.P. O' Farrell, "Reflectometry--a new method Of/measuring filler dispersion," presented at the Rubber Division, ACS, Spring 1997.

(4.) B.R. Richmond, "Carbon black dispersion measurement," presented at the Rubber Division, ACS, Spring 1993.

(5.) ASTM D2663-B, "Standard Test Methods for Carbon Black--Dispersion," American Society for Testing and Materials, West Conshohocken, PA, pp. 401-411 (2001).

(6.) W.M. Hess, "Characterization of dispersions," Rubber Chem. & Tech., 64, p. 387, 1991.

(7.) N.A. Stump, Jr., and H.E. Railsbeck, "Carbon black dispersion--photographic technique and rating system," Rubber World, volume 151, (1964).

(8.) L. Andersson, J. Sunder, S. Persson and L. Nilsson, "Optimizing mixing through filler dispersion control," Rubber World, March 1999.

(9.) ISO 11345. "Rubber--assessment of carbon black dispersion--rapid comparative methods," First Edition 1997.

(10.) L.O. Anderson, S. Persson and L. Skoog, "Dispersion control," Tire Technology International, June 1998.

(11.) O. Graydon, "Scientific CCD cameras," Opto and Laser Europe, February 2003.

(12.) J.B. Putman, M.C. Putman and R. Samples, paper# 19, Rubber Division, ACS, October 2001.

(13.) A.I. Medalia, "Dispersion of carbon black in rubber." Rubber Chem. & Tech., 34-4, pg. 1,134 (1961).