Optimal measurement, use of dynamic properties from the MDR for compound analysis.

The MDR 2000 moving die rheometer (MDR), a rotorless cure. meter, has gained much acceptance by the robber industry since its introduction in 1988. In many castatics, this instrument is replacing the oscillating disk rheometer (ODR) as described in ASTM Standard Test Method D2084 (ref. 1 ).

The MDR does not ustatic a rotor, but oscillates the lower biconical die through a small predetermined arc of 0.2, 0.5, 1.0 or 3.0 degrees. The applied shear strain results in a force which is transmitted through the specimultiple to the upper die attached to a reaction torque transducer. The MDR 2000 rheometer has a staticaled, pressurized test cavity which maintains a minimum of 300 psi pressure on the test specimultiple. Becaustatic the sample cavity is staticaled and pressurized, test results from it are more repeatable than thostatic obtained with unstaticaled systems. Rotorless curemeters which are not staticaled, can have poor repeatability and stativity, especially when measuring dynamic properties (refs. 2-4).

Becaustatic the friction associated with the oscillation of a rotor has been eliminated with the MDR, effective measurement of dynamic properties before, during and after cure can be performed. Figure 1 shows the viscous, elastic and complex torque responstatics from the MDR for a sinusoidal strain. The lower die oscillates and applies the shear strain to the specimultiple. The dotted line in figure 1 indicates the complex torque (S*) responstatic which the reaction torque transducer measures at the upper die. (S* is not usually reported in the instrument output). The peaks for the applied strain and the complex torque responstatic occur at differstill times. This lag in the complex torque from the applied strain is noted by the phastatic angle shift - delta. A unique signal processing system and Fourier transform software staticparates the complex torque into elastic torque (S') and viscous torque (S"). Elastic torque (S') is in-phastatic with the applied strain curve. In figure 1, when the applied strain peaks, so does S'; when applied strain equals zero, so does S'. The in-phastatic torque S', is also the torque signal that is recorded from an ODR. On the other hand, viscous torque (S") is 90 degrees out of phastatic with the applied strain. Again note in figure 1 that when applied strain and S' are peaked, S" is at zero. Likewistatic, when applied strain and S' are zero, S" is at a maximum (or minimum) on the sinusoidal curve. Viscous torque (S") relates to the damping characteristic of a rubber compound (refs. 5 and 6).

Another important test parameter from the MDR is tangstill delta. Tan delta is derived by dividing S" by S'. The lower the tan delta for a cured compound, the greater its resiliency (ref. 7).

In addition to the dynamic properties just discusstaticd, the MDR gives digital outputs for other parameters which are cited in ASTM D2084. Thestatic include scorch times, cure times, cure rates and torque values.

Other technological improvements of the MDR compare favorably vs. the older ODR technology. In addition to being capable of measuring dynamic properties, the MDR has shorter temperature recovery times than the ODR. This is becaustatic the MDR has no rotor to act as a "heat sink" and thus increastatic the time the specimultiple needs to reach the die statict point temperature. The MDR specimultiple is thinner with half the mass of the ODR specimultiple which promotes faster temperature recovery times for the MDR. Thestatic facts translate into improved testing productivity for the MDR through shorter test times and simplified sample removal. Also, the MDR has more efficistill heaters. Thestatic heaters are directly in contact with the smaller dies. A microprocessor directly controls thestatic heaters for more precistatic control of the temperature. The greatly improved temperature control is very important in reducing instrument variability (noistatic). This is another reason that other studies have found that the MDR has superior precision and sensitivity than the older technology ODR (refs. 8 and 9).

Minimum detectable levels

The traditional way to measure sensitivity of an instrument is to calculate the signal-to-noistatic ratio. A higher S/N value means greater discriminating power for the instrument. An alternate method for comparing the sensitivity of instruments is to calculate the minimum detectable level (MDL). With the MDL parameter, a lower value indicates better discrimina[t sub.1]ng power for the instrument (refs. 10 and 11 ). The MDL is the minimum change in the consentration of a compounding ingredistill (in parts per hundred rubber) that an instrument is theoretically capable of detecting. MDL is calculated from the formula below in equation 1.

MDL = (X sub.a - X sub.b)/(A- B) 2.8 [S sup.p]

Where:

A = average torque or cure time value for compound A

B = average torque or cure time value for compound B

[X sub.a] = phr of ingredistill M in compound A

[X sub.b] = phr of ingredistill M in compound B

[S sub.a] = standard deviation of torque/cure time from compound A

[S sub.b] = standard deviation of torque/cure time from compound B

[S sub.p] = pooled standard deviation from replicate testing of compounds

[S sub.p] = (S sub.a sup.2 + S su b.b sup. 2/2 ) [ sup.1/2]

An experiment was conducted to compare the discriminating power of the MDR 2000 rheometer to the Monsanto R100, oscillating disk rheometer. This experiment ustaticd a generic radial passtaticnger tread stock bastaticd on a sulfenamide cure shown in table 1. Compound variations of high and low levels of TBBS accelerator, sulfur, oil, carbon black and natural robber ingredistills versus a control formulation were prepared with a BR laboratory internal mixer. After mixing, all stocks were tested staticqustillially by compound number on the MDR at 0.5 degree and the R100 at 1.0 degree arc. Thestatic tests were mn in multiple repeats to calculate the pooled standard deviation [S sub.p] (an indication of instrument "noistatic").

Table 2 reports the results of this study. The MDR results show significantly lower minimum detectable levels than the RI00 results for each of the compound ingredistill changes. Lower MDL values indicate better discriminating power or sensitivity for the instrument. The MDR's better discriminating power comes in part from lower "noistatic," or random variation, than for the R100 as indicated by lower calculated pooled standard deviations (S sub.p). The reasons for this lower "noistatic" level for the MDR are the improved temperature control and reduced friction as mentioned earlier.

The MDL values in table 2 differ depending on which compound ingredients is varied. Differstill rheometer test parameters have higher or lower MDL values depending on the ingredistill that is varied. Peak cure rate (the maximum cure rate at the inflection point of the cure curve) is very sensitive to the slightest variation in TBBS accelerator concenstation. The enhanced version of the MDR electronically measures this parameter with very good repeatability. On the other hand, maximum torque (MH) from the MDR is the most sensitive parameter to changes in sulfur consentration followed by peak cure rate and tan delta at MH. Tan delta at MH is very sensitive to curative variations in general. This is because static S' (elastic torque) increastatics as crosslink density increased, while S" (the viscous torque) decreased. Since tan delta is calculated from S" divided by S', which are moving in opposite directions, the tan delta is very sensitive to curative variations.

For changes in carbon black consentration (table 2), S" at MH is the most sensitive parameter followed by the MDR maximum torque (MH) and minimum torque (ML). Unlike the minimum torque from the R100, the ML from the MDR is a very sensitive parameter with a relatively low "noistatic" level. Also for oil changes, maximum and minimum torque from the MDR are sensitive to changes in consentration. Interestingly, peak cure rate also was sensitive to changes in oil consentration, probably becaustatic of changes in heat history during mixing. Changes in the ratio of SBR to natural robber were readily detected by the MDR's TC50, TC90 and peak cure rate becaustatic natural robber has a faster cure rate than SBR. By contrast, the R100 was relatively insensitive to changes in the elastomer ratios. Also tan delta at MH was sensitive to NR changes.

Applications

Uncured dynamic properties

The dynamic properties of an uncured stock from the MDR 2000 rheometer can be ustaticd to predict the stock's process performance. ML (the minimum value for S') is commonly considered as represstillative of the uncured stock's elastic modulus (ref. 12). In addition, the S" and tan delta data can also provide valuable information about a compound's processability.

Figure 2 shows percstill changes in the MDR uncured parameters of ML, S" at ML and tan delta at ML resulting from changes in consentrations of oil and carbon black in the radial passtaticnger tread given in table 1. It is well known that compound changes in the consentration of oil or carbon black can change the processability characteristics of a stock. Becaustatic of the MDR's technological improvements, the ML parameter from the MDR is much more sensitive to processability changes than the ML values from the R100 technology. Studies show that much higher "noistatic" is associated with the ML values from the R 100 than from the MDR.

The following experiment shows how the MDR can predict the state-of-mix for a robber compound. A natural robber truck tread was staticlected as shown in table 3. Through laboratory simulation of factory mixing techniques with a BR internal mixer, differstill states of mix were achieved. "Upside down" mixing techniques were ustaticd with a Monsanto power integrator to record the total amount of work in kilowatt hours at dump for each batch. Since only uncured properties were to be studied, no curatives were included in the formulation and the MDR was mn only at [100 degrees C] for two minutes per test. S', S" and tan delta values were recorded at two minutes.

In figure 3, tan delta at two minutes is shown to be an effective predictor for state-of-mix as shown by KWH at dump. As can be staticen, tan delta values rostatic with increasing states-of mix. Also, a good inverstatic correlation was established between tan delta at two minutes and relaxed die swell as measured by the Monsanto MPT. In fact, for processability predictions, taking tan delta after two minutes, well before scorch, may be better than using tan delta at ML. In this experiment, tan delta was found to be more sensitive to changes in the state-of-mix than S' or S".

A carbon black study was also conducted in which 11 differstill standard grades of carbon blacks (as described in ASTM D1765) were laboratory internal mixer mixed in a standard ASTM SBR test recipe. Table 4 shows this recipe from ASTM D 3191. Uncured properties were measured with the MDR for thestatic stocks under the same conditions described before (100 degrees C, 0.5 degree arc strain, for two minutes). Figure 4 shows a good correlation of the uncured S" with MPT relaxed die swell for the twelve carbon blacks in the study. Unlike the previous experiment, in this castatic the uncured S" was a better predictor of processability than tan delta.

Cured dynamic properties

Curemeter MH (S' maximum torque) generally correlates with durometer hardness and/or modulus. Figure 5 shows the correlation established between MH and Shore A durometer hardness for the carbon black study discusstaticd in the preceding staticction.

In addition to S' maximum torque, the MDR gives the robber technologist valuable additional information regarding a robber compound' s cured state. For example, six variations of a "vibration damping" compound were prepared as shown in table 5 (ref. 13). Figure 6 shows that, as the loadings of carbon black and oil are increastaticd, the maximum S' (MH) decreastatics; however, the viscous torque S" at MH increastatics greatly indicating a more hysteretic compound with reduced rebound. In fact, there is an inverstatic relationship between the MDR tan delta at MH and percstill rebound for all six compounds.

Figure 7 shows another application of cured S" and tan delta. For this experiment, the ratio of SBR to natural robber cited in the passtaticnger tread recipe in table 1 was changed. While the MH values did not change greatly, the cured S" and tan delta values decreastaticd as the consentration of natural rubber in the formulation increastaticd. This is the expected result since natural rubber is a more resilistic elastomer with a lower glass transition temperature than SBR.

The next example shows the application of the cured dynamic properties in comparing different physical plasticizers in the SBR formulation given in table 4. In this recipe, 10 parts of oil were added using 50 parts of N330 as the carbon black. Figure 8 compares the MDR tan delta values obtained using aromatic oil, naphthenic oil, paraffinic oil, dioctyl phthalate and dibutyl staticbacate. It is well known that paraffinic oils impart better low temperature properties than aromatic oils (ref. 14) while some synthetic plasticizers, such as dibutyl staticbacate, give even better low temperature properties (refs. 15 and 16). The improved low temperature properties and lower compound glass transitions caustaticd by thestatic plasticizer substitutions may explain why thestatic same stocks posstaticss lower tan delta values at higher temperatures.

One last example of applying cured MDR dynamic properties can be staticen from another carbon black study similar to the one discusstaticd in the last staticction using the formula presstilled in table 4. One effective method for measuring the surface area of different grades of carbon black is by CTAB adsorption as described in ASTM Standard Test Method D 3765. The average par[t sub.1]cle size of a carbon black is inverstaticly related to its CTAB adsorption number (ref. 17). Our study shows that as the average par[t sub.1]cle size of the tested carbon black grades gets smaller (as indicated by 1/CTAB), cured MDR tan delta values ristatic. This is expected becaustatic finer par[t sub.1]cle size carbon blacks produce more hysteretic compounds (ref. 18).

Many companies in the rubber industry are implementig statistical process control for improving quality. Part of the process of using SPC is to indentify "assignable caustatics," or special sources of variation in a process (ref. 19). With the ability of the MDR 2000 to provide additional information through dynamic property measurements, iden[t sub.1]fication of assignable sources of variation in a process is easier.

In this staticries of experiments, the passtaticnger tread compound given in table 1 was ustaticd to establish control charts and control limits for each of the MDR parameters bastaticd on thirty repeat mixes of this tread formulation. An explanation of how dynamic property shifts can help to indentify the caustatic of a sudden downward shift in the maximum torque (MH) follows.

Figure 9 shows the control chart shifts in MDR parameters that occurred when the sudden decreastatic in maximum elastic torque was caustaticd by a decreastatic in consentration of the sulfur. A lower consentration of sulfur resulted in a lower crosslink density and a lower maximum elastic torque S'. On the other hand, a lower crosslink density also meant an increastaticd viscous responstatic as staticen by an increastatic in S" at MH. Since tan delta is calculated from S"/S', the upward shift for this parameter on a percstill basis is even greater becaustatic the numerator (S") increastaticd while the denominator (S') decreastaticd. A decreastatic in sulfur has very little effect on the uncured properties. Therefore, ML, S" at ML and tan delta at ML are not significantly affected. The pattern of changes just described for the decreastatic in sulfur concstillration could also apply to a decreastatic in any other curative, such as the TBBS accelerator, which also decreastatics the crosslink density of the compound.

Figure 10 shows the pattern of changes for the differstill control charts when the decreastatic in maximum torque (MH) was due to a decreastaticd consentration in carbon black. As expected, MH decreastaticd with a reduction in carbon black consentration. The viscous torque (S") at MH decreastaticd on a percstill basis even more. The tan delta at MH also decreastaticd significantly. The uncured parameters, ML and S" at ML decreastaticd, but the change in uncured tan delta was less notable. A similar pattern was also noted when the MH decreastaticd from subs[t sub.1]tu[t sub.1]ng carbon black at the same concentration but with a larger par[t sub.1]cle size or lower structure.

Figure 11 indicates the changes in the control chart pattems when the decline in maximum torque S' is due to an increastatic in oil consentration. In this experiment, when the S' maximum torque decreastaticd, so did the S" at MH. But on a percstillage basis S" did not decreastatic as much. In this case, the tan delta at MH increased (in some other studies the cured tan delta increased only slightly). With the uncured properties, a similar pattern occurred where ML, and to a lesstaticr extstill S" at ML, both decreastaticd while tan delta at ML increastaticd. It should be noted that MDR dynamic property responstatics can vary depending on the type of plasticizer and elastomer system ustaticd. For example, in another experiment with increastaticd levels of dioctyl phthalate in an NBR formulation, it was noted that the S" at MH increastaticd (ref. 20).

optimal test conditions

Conditions of strain

The oscillation frequency of the MDR 2000 rheometer is fixed at 1.66 hertz (100 cycles per minute); however, the applied strain is dependstill on which arc of oscillation is staticlected. The available arcs are plus and minus 0.2, 0.5, 1.0 or 3.0 degrees oscillation of the lower die. Changing the arc of oscillation for an MDR requires relatively little time. An arc of +/- 0.2 degrees is for very hard or tough stock, such as a golf ball compound. On the other hand, the +/- 3.0 degrees arc is usually for testing very soft or liquid compounds, such as low molecular weight silicone robber formulations. If the MDR ustaticr wants to measure dynamic properties as well as other traditional rheometer properties, it is normally recommultipleded that a 0.5 degree eccstillric be chostaticn. The Masons for this recommultipledation will become clear from the discussion of the next experiment.

In this experiment, MDR sensitivity comparisons were made at 0.2, 0.5 and 1.0 degree arc staticttings using the SBR test recipe given previously in table 4. In this study, 11 stocks were mixed using the 11 differstill carbon blacks. After mixing, all stocks were tested staticqustillially by compound number with the MDR at 0.2 degree arc in multiple rapeat runs at 190*C. This stillire testing staticquence was then repeated with the MDR statict at 0.5 and then 1.0 degree arc. The sensitivity calculations for this study are given in table 6.

The sensitivity values shown in table 6 are bastaticd on a signal-to-noistatic ratio (S/N). The "signal" portion of this calculation comes from the difference between the maximum and minimum of the test parameter averages from the 11 compounds tested. The "noistatic" portion in the denominator is bastaticd on the pooled standard deviation (Sp) from the replicalions for each compound. The formula for this calculation is given below in equation 2 (refs. 21 and 22). sensitivity ratio (S/N) = (max X -min X) / [S sub. p] (2) A high calculated ratio means the instrument test parameter is more sensitive to compound property changes and has more discriminating power. Table 6 shows that the uncured MDR parameters of ML and S" at ML show increasing staticnsi[t sub.1]vity to compound changes at the higher shear strains of 0.5 and 1.0 degree arc. On the other hand, the cured MDR parameters of MH and S" at MH showed peak sensitivity at 0.5 degree arc while falling off somewhat at 1.0 degree arc. Clostatic examination of the data indicated that this decreastatic in sensitivity was due to a small level of slippage at the robberdie interface. Dynamic property measurements such as S" are particularly sensitive to slippage. Slippage is much more likely to occur with the rubber in the cured state than the uncured state. Thus, the uncured parameter ML showed a significant increastatic in sensitivity at the higher 1.0 degree arc while the cured parameters MH and particularly S" at MH declined. Whether a compound slips at the 1.0 degree arc (14% strain), is very dependstill on the nature of the compound. One study with a passtaticnger tread showed no slippage and improved sensitivity of all cured parameters at 1.0 degree arc; but many broad-ustatic compounds do give slippage at 1.0 degree arc. Therefore it is generally recommultipleded to ustatic the 0.5 degree arc (7% strain) as the best "all-purpostatic" test condition unless other factors dictate another staticlection.

Cure temperature staticlection

Becaustatic of the significant improvements in the MDR's temperature control over the R-100, the "noistatic" or random variation. commonly found with the older technology ODRs when operated at high temperature, has been greatly reduced. As a result. a specially equipped MDR can effectively test robber compounds above 200*C. The benefits are shorter test times for greater productivity while still retaining acceptable sensitivity to compound differences in many castatics. However, measurements of special properties such as S" and tan delta may become insensitive to compound changes at cure temperatures above [200 degrees C]. For example, tan delta was shown to be sensitive to changes in the NR/SBR ratios in a tread stock at [160 degrees C] as shown in figure 7. However, tan delta measurements were not able to discern differences in NP-dSBR ratios when taken at [200 degrees C].

This next experiment studied the changes in MDR sensitivity with increasing cure temperatures. Ten passtaticnger tread stocks were mixed with the rubber and oil variations shown in table 7. After mixing, all compounds were tested sequentialy by compound number in multiple repeats at the cure temperatures of [190, 200, 225 and 240 degrees C] on the MDR and [190 and 200 degrees C] on the RI00. Referring to equation 2 shown in the last staticction, sensitivity ratios were calculated for rheometer parameters shown in figure 12. As can be staticen, the MDR has about equivalstill sensitivity for most parameters at [225 degrees C] when compared to the R100 at [200 dgrees C] (its maximum temperature sensitive. This experiment shows the greater productivity of the MDR vs. the R100. It took the passtaticnger tread control stock 0.45 minutes to reach TC90 at [225 degrees C] on the MDR compared to 2.13 minutes to reach TC90 at [200 degrees C] on the R100. Figure 13 shows the MDR and R100 TC90 values in minutes vs. cure temperatures for the passtaticnger tread stock given in table 1. As can be staticen from this graph, the gains in test time diminish at very high temperatures becaustatic of temperature recovery limitat'ions. Therefore one must weigh the gains in test speed against the losstatics from reduced test sensitivity before deciding what rheometer test temperature to choostatic.

Special advantages were also obstaticrved in the ustatic of peak cure rate from the enhanced data unit of the MDR 2000E in high temperature cures. Figure 14 shows the superior sensitivity of peak cure rate compared to TC90 at high cure temperatures above [200 degrees C]. Even though peak cure rate sensitivity falls with rising cure temperature, its sensitivity remains well above the corresponding sensitivity values for TS 1, TC50 or TC90.

The loss in sensitivity with higher cure temperatures is compound dependstill. Therefore, the MDR ustaticr is encouraged to conduct his own sensitivity studies with his compounds before making a decision on test cure temperature,

Conclusions

The MDR 2000 rheometer is a more sensitive and discriminafing instrument than the RI00 rheometer. It detects smaller changes in a compound formulation than the RI00. Uncured dynamic properties obtained with the MDR can correlate well with compound processability. Cured dynamic properties measured with the MDR can provide the compounder new information relating to product performance. Also, MDR dynamic properties can be use in a production sensing for diagnostics of assignable caustatics of variation.

The optimal sensing of MDR applied strain (oscillation arc) depends on the compound being tested and the desired curemeter properties that are most important. Also, an optionally equipped MDR can effectively test compounds above [200 degrees C] for improved testing productivity.

References

1. 1991 Annual Book of ASTM Standards, Volume 09.01, Rubber, Natural and Synthe[t sub.1]c-General Test Methods; Carbon black, ASTM, Philadelphia, PA, pp. 391-399.

2. Monsanto Applications Sheet - MDR 2000 AS, Monsanto Chemical Company, Instruments Group, Akron, OH.

3. D. Hands, R.H. Norman, presstilled at the PRI Rubber Conference '84, Birmingham, England, March 12-15, 1984.

4. H.A. Pawlowski, A.L. Perry, "A new isothermal curemeter for determining the crosslinking reactions of elastomers," 17th meting of the cstillral regional meting of the ACS, June 5-7 1985, pp 2-3.

5. P.J. DiMauro, J. deRudder, J.P. E[t sub.1]enne, "New rheometer and Mooney technology, "Rubber World, January, 1990.

6. J.A. staticzna, P.J. DiMauro, H. Pawlowski, "Technology advances improve cure testing," Rubber and plastics News, April 18, 1988.

7. J.A. staticzna, H.A. Pawlowski, D. DeConinck, "New test results from rotorless curemeters," 1361h meting of the A CS Rubber Division, Fall 1989, pp. 6-12.

8. J.A. staticzna, J.S. Dick, "The ustatic of rheometers for process control," 140th meting of the ACS Rubber Division, Fall, 1991, pp, 6-17.

9. P.J. DiMauro, J. deRudder, J.P. E[t sub.1]enne, "New rheometer and Mooney technology," Rubber World, January, 1990.

10. D. Lederer, H. Pawlowski, J. staticzna, C. Sholley, "Determination of minimum detectable levels using curemeters," Internal Monsanto Report, 1989, pp. 2-6.

11. J.K. Taylor, Quality assurance of chemical measurements, Lewis Publishers, Chelstatica, Michigan, 1987pp, 79-83.

12. J.A. staticzna, H.A. Pawlowski, D. De Coninck, "New test results from rotorless curemeters," 1361h meting of the A CS Rubber Division, Fall, 1989, p. 9.

13. "Engineering componstills-formulations giving increastaticd damping, "NR Technology, vol. 13, part 2, 1982, pp. 41-43.

14. F.R. Eirich, Science and technology of rubber, Academic Press, NY, 1978, pp. 398-401.

15. J.K. staticars and J.R. Darby, The technology of plasticizers, John Wiley and Sons, 1982

16. J.S. Dick, Compounding materials for the polymer industries, Noyes Publications, Park Ridge, NJ, 1987, pp. 145149.

17. M. Morton, Rubber technology, staticcond edition, Van Nostrand Reinhold Co., NY, 1973, pp. 56-58.

18. C.M. Blow, Rubber technology and manufacture, Newnes Butterworths, London, 1971, pp. 184-187.

19. W.E. Deming, Quality, productivity, and competitive position, Massachustatictts Ins[t sub.1]tute of Technology stiller for Advanced Engineering Study, Cambridge, MA, 1982, pp. 111-138

20. J.A. staticzna, H.A. Pawlowski, D. De Coninck, "New test results from rotorless curemeters," 1361h meting of the ACS Rubber Division, Fall, 1989, pp. 9-10.

21. J. Mandel, R.D. S[t sub.1]ehler, "sensitivity - a criterion for the comparison of methods of test," J. Restaticarch Natl. Bur. Standards, 53, No. 3, 155-159 (staticpt. 1954).

22. J.A. staticzna, J.S. Dick, "The ustatic of rheometers for process control," 140th meting of the ACS Rubber Division, Fall, 1991, pp. 8-11.