The effect of silica dispersion on physical properties.

Rubber mixing has traditionally been optimized by testing and relating the physical properties to mixing conditions. A larger number of studies has focused on carbon black dispersion and its effects on robber properties through the use of various rapid techniques, including reflected light with image processing. Recent technology has allowed for the measurement of filler dispersion, providing direct feedback regarding the incorporation and the disagglomeration of carbon black. These methods were insufficient for testing silica dispersion due to limitations in resolution and overexposure on light-colored surfaces. Measurement of silica dispersion has grown in relevance in recent years, while analytical techniques to quantify dispersion remained a challenge. Methods for analyzing silica dispersion directly have been time-consuming and required a unique expertise in complex microscopy. The required use of the silane coupling agent further complicates the relationship between dispersion and physical properties due to its effects on processability and cure. A new, improved method for measuring dispersion with a resolution of 1.5 microns is described. This article investigates the effect of mixing untreated silica filled compounds, as well as silane modified compounds. Filler dispersion is measured by this method and the results are compared to physical properties. The new method provides a rapid way to accurately measure dispersion.

Until recent years, reinforcing fillers in elastomers have primarily included one of the many available grades of carbon black. Particle size, structure and dispersion have been of the greatest importance to understanding the final performance properties of a product. Dispersion involves a process of incorporation and disagglomeration, normally taking place in the internal mixer. Analysis of dispersion in the 3 to 120 micron agglomerate size range, in conjunction with more traditional physical properties testing such as stress/strain, often provided adequate information to both optimize the mixing cycle and provide useful final properties (ref. 1). Carbon black dispersion could be measured quickly and accurately using an optical reflected light instrument, with a CCD video camera and image processing (ref. 2). With this approach, older, more time-consuming optical microscopy could be replaced, allowing for agglomerate frequency histograms and quantitative percent dispersion to be recorded.

Silica has increased in importance due to both its improved performance properties and environmental efficiency. The raw material source for silica is not petroleum-based. High volume rubber products have sought to increase silica consumption. This is especially apparent in tire tread compounds, which are often highly loaded with silica. Advantages such as abrasion, tear, rolling resistance and heat resistance have all been observed (refs. 3 and 4).

Carbon blacks have served as robust reinforcing fillers which can be incorporated into a polymer with relative ease. Often, a one-pass mix of a carbon black compound is used to reduce mixing time. Silica, however, due to its chemical nature, is harder to mix and does not improve physical properties (ref. 5). Silica alone will not bond properly with a polymer, and will even absorb curing agents, resulting in compromised physical properties (ref. 6). This deficit can be overcome by the addition of organosilane coupling agents. The silane coupling agent is bound to the silanol group at the surface of the silica, allowing for increased processability and reinforcement. Silanization requires high temperatures for activation, requiring two or more complete mixing passes, and optimal temperature control to prevent scorch. G' (storage modulus) will decrease until silanization occurs at the elevated temperature of 145[degrees]C (ref. 7). It has been observed that silanization improves properties such as tensile strength and modulus.

Silica and carbon black have similarities in agglomeration, but the causes of agglomeration and the ability to disagglomerate and disperse are slightly different. Carbon black and silica both experience filler-filler interactions through the relatively weak Van der Waal forces, which can be broken by mixing. These forces are strong enough to create agglomerates, resulting in poor dispersion, but are not strong enough to reinforce (ref. 8). In addition to Van der Waal forces, silica agglomerates contain hydrogen bonds, creating significant challenges in breaking small agglomerates through mixing. Agglomerates are broken in three ways during mixing--erosion, rupture and collision--all of which are dependant on the morphology and polymer matrix properties (ref. 9). Erosion can cause disagglomeration from shear forces, and takes place in the polymer matrix. Erosion is the gradual but continuous breaking away of smaller filler fragments from the agglomerate. This phenomenon was shown by V. Collin et al (ref. 10). Rupture takes place when a hydrodynamic stress exceeds the cohesive force of the agglomerate. Rupture, as opposed to erosion, is a sudden breaking of large agglomerates into several smaller agglomerates. Collision of agglomerates is the third mechanism of disagglomeration, and takes place as the viscosity decreases and the agglomerates collide (ref. 9). Rupture is the most common and relevant form of disagglomeration as it relates to silica dispersion. Even though silica has different bonding and dispersion mechanisms than carbon black, a valid measurement of disagglomeration remains important.

As filler particles and the resulting agglomerates become smaller in size, dispersion measurement becomes more challenging (ref. 11). Dispersion measurement has often focused on two general sources, including: physical testing for indirect effects of filler dispersion, and direct measurement using high intensity microscopy. While other mechanical means have been successful in the measuring of carbon black dispersion, the requirement to analyze nanoscaled agglomeration prevents methods such as surface roughness, which measures only to 3 microns, from being utilized. Instead, TEM (transmission electron microscopy) and SEM (scanning electron microscopy) provide useful two-dimensional images, which can be processed to give an agglomerate count for particles with sizes as small as 25 nanometers. More recently, AFM (atomic force microscopy) has been used to provide a range of information on the molecular level, including three-dimensional topography, and separation forces, which can identify molecular bonds. The AFM is flexible in many ways that other high-resolution microscopy is not, including ease of sample preparation and its ability to be operated in ambient climates. As with most microscopes, the instruments themselves provide only a powerful "eye," leaving data interpretation to the user. Most of the time, images are generated in contact mode on an AFM. In this mode, individual agglomerates can be measured and the entire image compared to other known samples. The versatility of AFM has made it a preferred tool for many scientists; however, the subjectivity of visual analysis remains problematic.

It has been established that dispersion of silica has an effect on cured physical properties. Studies have also shown that dispersion has a significant influence on uncured rheological properties, and therefore affects viscoelastic characteristics important to processability. Schuster has shown that by increasing mixing time, dispersion is improved until a point at which the rate of change approaches zero (ref. 12). He also showed that Mooney viscosity is continuously lowered as mixing time increases. Unfortunately, it has been difficult to find the precise point when disagglomeration and dispersion do not improve any further, since viscosity continues to decrease. Determination of relevant final agglomerate size is also a challenge, since the size range between micro and macro dispersion is often ignored.

The following experiment will use reflected light microscopy (RLM) with a 1.5 micron resolution to measure dispersion and disagglomeration, and compare results to a series of rheological and final product properties.

The reflected light microscopy method

Reflected light microscopy has become the prominent tool for dispersion analysis because of its ability to rapidly quantify dispersion. A specimen is prepared by cutting the sample with an industrial razor blade as it is being stretched bi-directionally. Vulcanized or unvulcanized samples may be used. The specimen is placed in the apparatus and illuminated with a halogen light at a 30 degree angle. An image is captured with a CCD camera and processed using gray scale thresholding. Hills and valleys that are present on the cut surface appear as shades of gray. These can be counted and analyzed to give frequency/size histograms, percentage of the total area that are covered by agglomerates and a qualitative comparison to reference images. Custom calculations taken from the images are often useful in combining quantitative values with useful laboratory quality control (refs. 1 and 13). Commercially available RLMs have a resolution of three microns (ref. 14). The scanning and image processing from an RLM takes approximately five seconds.

For carbon black dispersion measurement, it has been shown that RLM agglomerate sizes, termed "nodges," represent significantly smaller agglomerates as observed directly from microtomed samples viewed with transmitted light and with sulfur hardened samples analyzed for reflectance (ref. 2). The Leigh Dugmore (ref. 15) formula commonly used for determining percent dispersion used a five micron threshold of relevance when viewing agglomerates directly. When analyzing surface roughness with RLM, a threshold of 23 microns was experimentally determined to relate to the five microns described by Leigh Dugmore.

Primary particles of carbon black and silica are nanomaterials. Relevant agglomerates for carbon black, however, have been shown to be over five microns in size. In the studies of Noordermeer (ref. 5), a silica agglomerate of 500 nanometers was shown to be relevant to dispersion and disagglomeration. Applying the same Nodge theory as described above, an RLM with resolution of 1.5 microns was used as the threshold for silica dispersion.

Experimental

A study was done by first mixing a masterbatch. Table 1 is the description of the equipment used for the study, as well as the formulations for the masterbatch and the final mix. The masterbatch includes all ingredients other than the silica and silane. This guarantees that surface roughness, as seen by RLM, is primarily the result of silica agglomeration only.

For the final batches two different types of precipitated silica were used, including: the Ultrasil 7000 highly dispersible grade granulates (HD) and the VN 3 precipitated silica granulates (ND). Both of these have roughly the same structure and particle size. Only the production process differs. Both silicas were mixed with and without organosilane.

For all final compounds, 10 mixes of various times were done, ranging from one minute to 13 minutes. Because the masterbatch included accelerators, precautions were made to keep temperatures low to avoid scorch.

A variety of common tests was conducted on each final batch for comparison, including: Mooney viscosity, stress/ strain and durometer. A more detailed set of tests on the uncured compound and the cured compound was conducted on the rheoTECH MDPT, dynamic mechanical rheological tester (DMRT). Strain sweeps were run, and storage modulus (G') recorded. All testing conditions can be seen in table 2.

The RLM used for calculating % dispersion was the disperGRADER 1000NT with an extended magnification to resolve nodge sizes of 1.5 microns and greater.

Results

In considering dispersion using RLM, disagglomeration is measured by analyzing a set of test samples to see the area not covered by nodges. As can be expected, as mixing time increases, agglomerates rupture and the total percent dispersion increases for all of the compounds studied. Disagglomeration of silica greater than 1.5 microns does, however, level off at approximately nine minutes, suggesting that dispersion is completed. A comparison of the HD highly dispersible silica with the addition oforganosilane, and the compound which is the ND normal silica grade, also with organosilane, is shown in figure 1. It can be seen that the highly dispersible compound with silane does indeed achieve its maximum level of dispersion in a faster time than the normal grade silica with silane. It is also clear, however, that the maximum level of dispersion is the same for both, suggesting that the similarity in structure of the two silicas leads to the same degree of disagglomeration.

[FIGURE 1 OMITTED]

A look at dispersion results of all of the compounds for the effect of silane on dispersion is shown in figure 1. The lack of clear distinction between the silane activated compounds and those without silane suggests that silica disagglomeration is caused by the mixing shear forces, and unaffected by the addition of a silane coupling agent. This is not to disregard the possibility that incorporation may occur at slightly different rates, as the RLM method strictly measures the amount of disagglomeration.

When evaluating the various compounds with traditional tensile tests, the most obvious result is the clear distinction between the compounds with silane and those without silane. This is not surprising, as silica without silane fails to create strong filler-polymer bonds, as well as absorbs curative ingredients, making for an incomplete cure. The difference in rate of disagglomeration of the silane-containing compounds that was observed in the RLM comparisons was not present in the tensile tests. Results are shown in figure 2. These results suggest that tensile testing alone does not give a complete picture of dispersion.

[FIGURE 2 OMITTED]

Stress at 200% elongation behaved as expected and is seen in figure 3. In a comparison of the compounds with and without organosilane, a large difference is observed, similar to tensile. There is, however, a noticeable difference between the highly dispersible and normal silica, which indicates that incorporation may occur faster, much as was shown in the RLM results. The primary distinguishing characteristic of all of the traditional tests was a continual deterioration with longer mixing times, whereas the RLM dispersion leveled off at a certain point. A similar characteristic applies to tear strength, as shown in figure 4. Durometer hardness results are shown in figure 5. As can be expected, those without silane have lower hardness. The hardness test did not distinguish between the two grades of silica. For cured properties then, it is no longer important to look at the compounds without silane, as they are consistently lower.

[FIGURES 3-5 OMITTED]

Using the DMRT to run a variable strain test for cured dynamic properties, it is easy to distinguish the two grades of silica. Figure 6 shows the results at 50% strain. Storage modulus (G') is significantly higher for the shorter mixes on the highly dispersible compound, while the results converge for the long mixes. These results support the RLM findings. It is possible that early in the mixing process the highly dispersible silica develops polymer-filler bonds in a shorter time than normal grade silica. By the time the longer mixing has occurred, large agglomerates have ruptured, and the remaining agglomerates are strongly cohesive.

[FIGURE 6 OMITTED]

The previous discussion of cured properties gives an understanding of product performance and how it relates to dispersion using RLM. In addition to final properties, it is also relevant to explore the effect on flow and processability, as can be seen from a number of uncured compound tests. The standard Mooney viscosity test shows a large difference in viscosity between the compounds with and without silane, and can be seen in figure 7. The compounds with silane are significantly lower, since silane acts as a process aid. The results are also similar to RLM dispersion, in that viscosity tends to level off with longer mixes. Comparing the compounds with silane did not show significant differences between the two silicas for Mooney viscosity. When using the more sensitive DMRT on the uncured compounds, however, G' for the highly dispersible silica is higher for the short mix times and converges at longer mix times (figure 8). This is likely caused by the same phenomena as cured G'.

[FIGURES 7&8 OMITTED]

All of the tests discussed provide an indication of important compound properties, which will help in the prediction of final product performance. These tests are also cumulative, and can indicate variations in weighing, mixing and raw material. Only the RLM is a direct measurement of dispersion, and can be used to optimize compound mixing.

Conclusion

When considering a carbon black filled compound, dispersion has a direct effect and is predictive of many physical properties. In a silica reinforced compound, however, physical and processing properties are not only related to dispersion, but also to the effect of the organosilane coupling agent. Cured and uncured compound properties have been shown to bc dependent on mixing. To fully optimize mixing and understand compound properties, then, a variety of tests may need to be performed. An understanding of how properties change as mixing is increased is crucial in achieving optimal properties. Evaluation of percent dispersion gives useful information regarding disagglomeration, and clearly indicates when a material has achieved complete dispersion. RLM dispersion measurement provides a rapid way to directly measure the relevant property of silica dispersion that cannot be determined by any other method.

References

(1.) J. Putman and R. Samples, "An improved method for measuring surface roughness to indicate filler dispersion," presented at the Rubber Division, ACS, October 2002.

(2.) M. Putman and F. Welsh, "Dispersion measurements using video techniques, "Rubber World, January 2005, pp. 21-26.

(3.) S. Wolff, U. Gorl, M.J. Wang and W. Wolff, "Silica-based tread compounds, "European Rubber Journal, January 1994, pp. 16-18.

(4.) P.K. Pal and S.K. De, "Effect of reinforcing silica on vulcanization, network structure and technical properties of natural rubber," Rubber Chemistry and Technology, vol. 55, June 1982, pp. 1,171-1,173.

(5.) W. Dierkes, J.W.M. Noordermeer, "Mixing of silica compounds: The chemistry of the silanization reaction as determined by the type of coupling agent and process parameters, " European Union Satpro Publication, February 2005. pp. 45-50.

(6.) A. Ansarifa J.W.R. Nuhawan, T. Nanapoolsin and M. Song, "Reinforcing effect of silica and silane fillers on the properties of some natural rubber vulcanizates," Rubber Chemistry and Technology, vol. 76, June 2003, pp. 1,290-1,310.

(7.) A. Limper and K. U. Kelting, "Mixing of silica compounds: Higher silanization efficiency by adapted process and machine design, " European Union Satpro Publication, February 2005, pp. 16-35.

(8.) M.L. Kralevisch and J. Koenig, "FTIR analysis of silica-filled natural rubber, "Rubber Chemistry and Technology, vol. 71, September 1997, pp. 300-309.

(9.) O. Seyvet and P. Nevard, "Collision-induced dispersion of agglomerate suspensions in shear flow, "Journal of Applied Polymer Science, vol. 78, November 2000, pp. 1,130-1,133.

(10.) V. Collin and E. Peuvrel-Disdier, "Dispersion mechanisms of carbon black in elastomers, "European Union Rotor Project Publication, February 2005, pp. 155-173.

(11.) D. Bielinski, P. Glab, O. Dobrowolski and L. Slusarski, "Dispersion of fillers and its effect on properties of rubber," European Union Rotor Project Publication, February 2005, pp. 324-346.

(12.) R.H. Schuster, "Dispersion of fillers--a decisive factor for rubber processing and the performance of elastomers," European Union Rotor Project Publication, February 2005, pp. 185-205.

(13.) S. Otto, O. Randl, O. Gonclaves and B. Cantaloube, "New reference value for the description of filler dispersion with the DisperGrader 1000 NT," Kauschuk Gummi Kunststoffe, Vol. 58, January 2005. pp. 390-392.

(14.) L.O. Anderson, S. Persson and L. Skoog, "Dispersion control," Tire Technology International, June 1998. 15. C.H. Leigh Dugmore, Rubber Chemistry and Technology, vol. 29, 1956, pp. 1,303-1,308.

Table 1--formulation and mixing instructions

Masterbatch formula        Masterbatch mixing specification
                    phr

SBR 1778          103.00          Mixer: 75 L, Kneader tilt
                                  mixer
BR                 25.00          Rotor speed: 25 rpm
ZnO                 2.50          Mixing procedure for
                                  masterbatch
Stearic acid        2.00   (1)    Add SBR and BR at time "0"
BBTS                1.00   (2)    Lower RAM
RM sulfur           1.50   (3)    Add remaining ingredients
                                  at time: 1'
T102                5.00   (4)    Drop at 100[degrees]C

Final formula                     Final mixing specification
Masterbatch       140.00          Mixer: Farrell, #B, 1.5 L
                                  lab mixer
Tetrasulfide        6.86          Rotor speed: 50 rpm
coupling
agent'
Naphthenic oil     5.00           Mixing procedure for final
Silica*        60.00 (l)   (1)    Add 1/2 MB, remaining
                                  ingredients, 2nd 1/2 MB
                                  at time "0"
                           (2)    Lower RAM
                           (3)    Sweep at 0.5 min.
Total          211.86 (4)  (4)    Drop at specified times
                                  of 1, 2, 3, 4,
                                  5, 6, 7, 9, 11, 13 min.

Compound          HD w/     HD w/o      ND w/      ND w/o
  variations     silane     silane     silane      silane
Masterbatch      140.00     140.00     140.00      140.00
HD silica         60.00      60.00         --          --
ND silica            --         --      60.00       60.00
Coupling           6.86       0.00       6.86        0.00
agent

Note: HD silica: Degussa Ultrasil 7000; ND silica: Degussa
Ultrasil VN3; Silane: PolyOne DCS-4S

Table 2--testing conditions

Mooney viscosity tests:

ASTM D1646

Temperature:                   100[degrees]C
Measurement:                   MD +4
  Uncured dynamic properties
        ASTM                   D6204
Static preheat:                4 min. @ 100[degrees]C
        Test conditions:
Temperature:                   100[degrees]C
Frequency:                     1 Hz
Strain:                        2%

Press cured specimens:

ASTM D3182

Time:                          20 min.
Temperature:                   171[degrees]C
  Cured dynamic properties
        ASTM                   D6601
Static preheat:                20 min. @ 171 [degrees]C
        Test conditions:
Temperature:                   100[degrees]C
Frequency:                     1 Hz
Strain:                        50%