Printer Friendly

Viscoelastic characterization of rubber with a new dynamic mechanical tester.

Rubber has both viscous and elastic responses to deformation. A viscous response is proportional to the rate of deformation while an elastic response is proportional to the amount of deformation. A viscous response is often modeled with a dashpot while an elastic response is modeled with a metal spring. The combination of the two produce the well known Voigt model for rubber viscoelastic behavior (figure 1). This model illustrates the behavior of rubber to deformation. Measuring the elastic and viscous properties helps characterize rubber behavior in processing and final product performance.

The measurement of rubber viscoelastic properties with traditional dynamic mechanical rheological testers (DMRTs) has not been commonly done in the rubber industry (ref. 1). There are several reasons for this fact. This article discusses the various problems in obtaining good viscoelastic test information from rubber samples with currently available DMRTs. A new viscoelastic tester, the RPA 2000 rubber process analyzer will be discussed. It was designed to solve many of the present problems associated with getting good viscoelastic test information from uncured and cured rubber samples.

Problems in viscoelastic measurement of properties

Conventional DMRT rubber testing requires complex sample preparation. Uncured samples are especially difficult to hold in a traditional DMRT test apparatus. Uncured sample thickness is also hard to keep uniform. Cured samples must be made separately for most DMRT testing using special molds to produce the correct dimensions. Many sample preparation procedures require a high degree of operator skill not always available in a rubber factory. Variations in sample preparation and loading procedures can increase the variation in the data. Different procedures for uncured and cured rubber also mean that uncured and cured samples are tested as two different tests instead of one. This increases total test time significantly.

Rubber test samples kept under pressure give more repeatable and reproducible measurements of resistance to deformation (ref. 2). A pressurized sample cavity prevents the formation of porosity in the sample as well as slippage along the die/sample interface which improves repeatability. Good repeatability and reproducability are necessary to expand the DMRT to statistical process control (SPC). Most DMRTs do not confine the test sample under pressure.

Viscoelastic properties of rubber vary with changes in applied strain, frequency and temperature. Testing at several different frequencies, strains and temperatures is usually necessary in order to thoroughly characterize the viscoelastic properties of a rubber sample. The optimal test conditions will vary with the sample. Samples are often tested over several frequencies, strains or temperatures in one test called a sweep. Temperature sweeps with traditional DMRTs can result in very long test times due to the time required to reach a new stable temperature setting. Long test times can limit the application of traditional DMRTs to developmental testing.

The use of computers makes it easier to set up complex DMRT tests and to reduce the time for data reduction. However, many traditional DMRT instruments still require a highly skilled operator to run the test and reduce the data.

The problems listed above were the primary considerations in the design of a new DMRT.

Features of RPA 2000

Ease of sample preparation

The RPA 2000 is a DMRT containing two cone shaped dies (figure 2). Two seal plates with their respective seals complete an enclosed sample cavity. The two dies can separate for ease of sample loading. The only sample preparation required is to ensure that the volume of the loaded sample is between 4 to 6.0 cm3. The sample is placed onto the unit's lower die. The dies are then closed. The upper die comes down onto the sample and forms a compression mold. Sample in excess of that required to fill the sample cavity is extruded into the spew channel. No further instrument adjustment is necessary. When die closure is complete, the sample cavity contains a constant volume of sample and more than the minimum pressure required to prevent slippage under most test conditions. At the end of a test, the dies are opened again and the sample is removed.

Compared to the traditional DMRT, the RPA 2000 die system greatly reduces the time required to prepare, load and remove the sample. This time is similar to the time necessary to operate an MDR 2000 rheometer.

Torque signal generation

There are several different methods used by DMRTs to determine the viscoelastic behavior of rubbery materials. Usually DMRT testing is done with sinusoidal strains. Sinusoidal DMRT testing can be done in compression, tension or shear. Each method has advantages and disadvantages. The RPA 2000 deforms the sample sinusoidally in shear. A special direct drive motor moves the lower die sinusoidally over a range of strains and frequencies. A complete list of strains and frequencies are given in table 1. Note that the high end of the strain range is not available for testing rubber in any other DMRT.

The upper die is attached to a torque transducer. In this way, noise from the lower die drive system is eliminated. The torque transducer is calibrated using a torsion spring mounted between the two dies while in the open position.

Figure 3 shows the resulting signal produced by a viscoelastic material such as rubber when subjected to a sinusoidal strain. The signal response will also be sinusoidal (G*). However, this signal will not be in-phase with the applied strain which means that the peak signal (G*) does not occur at the peak strain. Note that the peak signal (G*) occurs just before the peak strain is reached. The resulting torque signal can be separated into two sinusoidal signals 90 degrees out-of-phase with each other. G' is the modulus in-phase with the strain and represents the elastic modulus. G" is the signal 90 degrees out-of-phase with the strain and represents the viscous modulus. G* can be calculated from (G'2 + G"2)1/2. Figure 3 also shows the location for G' and G". Values for G' and G" are calculated from a Fourier Transform of the complex modulus G*. The transform also improves the signal to noise ratio for G' and G" measurements compared to single point measurements at maximum strain for G', and at maximum rate (zero strain) for G". The above method represents a direct way to determine the viscoelastic properties of a sample.

G*, G' and G" are calculated from the torque measurements S*, S' and S" respectively. The conversion is done by using appropriate factors. The torque is multiplied by the factor which depends on the die configuration and then divided by the strain:

G' = (S' x factor)/strain and G" = (S" x factor)/strain An additional property measured by DMRTs is Tan [delta]. It is defined as S"/S' (which is equal to G"/G') or the ratio of viscous torque to elastic torque.

Temperature control

Each die is directly heated using a low mass heater operating with its own individual digital temperature controller. Temperature resolution is to the nearest 0.1[degrees]C. Changes in temperature are very quick. Increasing temperature typically takes about one second per 1[degrees]C plus some temperature stabilization time. A forced air cooling system speeds up the time to reduce temperature. The temperature can be reduced from 190[degrees]C to 50[degrees]C in about five minutes plus some stabilization time.

Subtests and sequence of subtests

There are four different types of subtests which can be programmed in an RPA 2000. One or more subtests make up a complete test (a sequence of subtests). These subtests are:

* Frequency sweep - strain and temperature are kept constant while the frequency is varied in a preprogrammed way. S' and S" are recorded at each frequency.

* Strain sweep - frequency and temperature are kept constant while the strain is varied in a preprogrammed way. S' and S" are recorded at each strain.

* Temperature sweep - frequency and strain are kept constant while the temperature is varied in a preprogrammed way. S' and S" are recorded at each strain.

* Temperature sweep - frequency and strain are kept constant while the temperature is varied in a preprogrammed way. S' and S" are recorded at each temperature.

* Cure - frequency, strain and temperature are fixed. A test time is entered. S' and S" torque are recorded continuously during the test time. Minimum S', maximum S' and cure times are calculated. This subtest can also be used to determine material stability.

Each sample can be programmed for testing with two or more of these subtests in a sequence. Combinations of subtests can be programmed in a large number of different ways. Uncured compounds can be tested before cure, during cure and after cure in one test. In this way, a rubber compound can be characterized for processability, cure and final product properties in a single test. Each test can take place at a different temperature.

Each subtest contains criteria to prevent the collection of invalid data. These criteria ensure that the sample is actually at the test temperature and that the data are valid. These criteria were established to handle many rubber samples. However, these criteria are also programmable to increase flexibility in testing.

In compound development work, these subtests can be programmed with many different test conditions to fully characterize a material. However, a typical factory quality control department needs only those test conditions necessary for process control. Using fewer tests or test conditions will significantly speed up testing for quality control. This will result in very short test times. Both development and quality control test sequences are stored in the microcomputer under simple test names and recalled as needed.

PC graphical user interface

The RPA 2000 computer display uses the OS/2 Presentation Manager environment for user-friendly operation. This system provides a menu bar at the top plus mouse control. Such a system can be run with just a mouse so that extensive commands need not be memorized.

Repeatability and reproducibility

The RPA 2000 has the same dies as the MDR 2000. However, there are many differences in the design of both instruments. The effect of these differences on data had to be determined before testing any other materials. Comparable results would ensure the validity of the data. Five RPA 2000s were tested with a standard SBR formulation (table 2). Five replicates were tested on each instrument. The same tests were also done on five MDR 2000s for comparison. The results were used to determine the repeatability and reproducibility (instrument to instrument).

Table 3 summarizes the results. There is no significant difference in the repeatability and reproducibility data shown in table 3 between the RPA 2000 and the MDR 2000. The average torque values were reduced slightly on the RPA 2000 compared with the MDR 2000. Also, the T90 values were reduced slightly on the RPA 2000. Since the RPA 2000 is not intended to do the same type of testing as the MDR 2000, this is not important as long as the instrument differences are known. These data prove that the RPA 2000, which performs frequency and strain sweeps, is as repeatable and reproducible as the MDR 2000 rheometer.

Characterization of polymers

The viscoelastic properties of raw polymers can be characterized at many different test conditions on the RPA 2000. Figures 4 to 7 show results using various test conditions for four distinctly different polymers tested on the RPA 2000. Figure 4 shows the G' (elastic modulus) versus frequency. Note that the ranking of the polymers (lowest G' to highest G') varies with frequency. Tests such as the Mooney viscosity are done only at a single low shear rate and cannot provide information at different shear rates. The RPA 2000 is able to show the behavior of raw polymers at different shear rates through combinations of frequencies and strains. Figure 5 shows the G" (viscous modulus) versus frequency. Note that the G" curves do not match the G' result in figure 4. This indicates that the G' result cannot predict the G" value and that both are providing different information about the polymer.

Figure 6 gives S' versus strain for the same four polymers. The high strains show some very unusual curve shapes for each of the four polymers tested. The strains which exceed about 50% are a unique feature of the RPA 2000. The unusual curve shapes at high strains are thought to be due to the power law behavior of rubber materials. Figure 7 contains G' versus strain. The data in figure 7 are the same as those used in figure 6 except that S' has been converted to G'. Note that now the low strains show the greatest amount of separation among the polymers. S' shows better separation among the polymers at high strains.

The above tests show that the frequency and strain of the test change the values for G' (or S') and G" (or S") significantly. In addition to frequency and strain, temperature also affects the viscoelastic properties of raw polymers. Figure 8 shows the effect of temperature on S" versus frequency for an EPDM polymer. The EPDM raw polymer was tested at 70[degrees], 100[degrees] and 120[degrees]C. As the temperature was reduced, the frequency at which the maximum S" value occurred was reduced. This shift in frequency is usually referred to as the WLF shift (ref. 3).

Although final Mooney viscosity can correlate to average molecular weight, there are many other polymer characteristics which relate to rheological behavior that cannot be measured by the Mooney viscometer. Some of these polymer variations include the following:

* Molecular weight distribution (MWD).

* Monomer ratios for copolymers or terpolymers (such as percent ENB for EPDM).

The test methods presently available to measure these polymer characteristics are very laborious and/or time consuming and do not lend themselves well to a quality control environment. The new RPA 2000 has the potential for detecting variations in these polymer characteristics by the selection of the proper test conditions.

A study was made using the RPA 2000 in testing two different EPDMs. Both EPDMs had similar average molecular weights but one had a narrow molecular weight distribution (MWD) and the other had a medium MWD. A frequency sweep result for G" shows that G" values are reduced as the MWD becomes narrow. The difference in G" increases with increasing frequency.

Another pair of EPDMs had medium and broad MWDs. At lower frequencies the broad MWD had a greater G' value. The results became reversed at higher frequencies where the G' was greater for the medium MWD.

Three different EPDMs were tested which had similar MW and MWD. The greatest difference among them was the percent ENB. The results of G" versus frequency testing for all three EPDMs showed that high amounts of ENB appear to reduce the G" value.

Correlations to other instruments

A common method for determining the validity of a new instrument is to test the same samples on several different instruments. The data from each instrument are then checked to see if it correlates with data from the new instrument. This aids in understanding what the new instrument is actually measuring.

Figure 9 shows a correlation of Mooney viscosity data to the RPA S* torque value at 7% strain, 6 cpm and 100[degrees]C. Each of the 23 data points in figure 9 represent a different raw polymer. The results show a good linear relationship between Mooney viscosity and RPA S* values considering the very wide variety of raw polymers tested. Figure 10 shows the correlation of the slope of the Mooney stress relaxation curve (from a log-log plot of MU versus time) to the RPA Tan [delta] at 28% strain and 6 cpm. The results show a very good correlation considering the wide range of polymer types and grades tested.

Figure 11 shows the correlation of capillary rheometer apparent shear stress at 30 sec-1 to the RPA S' at 698% strain and 6 cpm. The compounds in figure 11 were all made with the standard SBR test recipe ASTM D-3191. Each compound was made with a different carbon black type. Results show a very good correlation. all of the above correlations were made using either raw polymers or uncured compounds. Some RPA correlation studies were also done with tests which use cured stocks. The RPA samples were all cured in situ for these studies while all other instrument samples were made in a separate mold.

Figure 12 shows the correlation of Shore A hardness to RPA S' at 5.6% strain and 6 cpm. The compounds used in this study were the SBR test recipe ASTM D-3191 with the different carbon black types. The results in figure 12 show a good correlation with some scatter as expected considering the variability of the Shore A hardness test. Figure 13 shows the correlation of the Goodrich Flexometer [Delta]T to the RPA Tan [delta] at 2.8% strain and 6 cpm. This correlation looks good considering the variability in the Goodrich Flexometer. Note that in this study, the RPA sample was cured in situ at 160[degrees]C and then the temperature was reduced to 100[degrees]C for the Tan [delta] measurement. This result does show that the RPA can measure hysteresis in cured stocks at test temperatures which are well below the curing temperature.

Summary and conclusion

The RPA 2000 has several distinct advantages over traditional DMRT instruments. There is no special sample preparation. The sample is kept under pressure during a test to prevent porosity or slippage. The ability to program the system allows a sample to be easily tested before cure, during cure and after cure in a single test.

The repeatability and reproducability of the RPA 2000 was shown to be good. The unit could characterize raw polymers with differences in structure using frequency sweeps, strain sweeps or changes in temperature. The data correlated to a variety of instruments testing polymers, uncured compounds and cured compounds. The RPA 2000 is a new dynamicmechanical rheological tester designed to provide dynamic property data on rubber polymers, masterbatches, uncured compounds and cured specimens.


(1.) R.H. Norman and P.S. Johnson, Rubber Chem. Technol. 54, 525-527 (1981).

(2.) D. Hand and R.H. Norman, presented at the PRI Rubber Conference, Birmingham, England, March 12-15 (1984).

(3.) J.D. Ferry, Viscoelastic properties of polymers, John Wiley & sons Inc., 1961, Chapter 11.
COPYRIGHT 1992 Lippincott & Peto, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:RPA 2000
Author:Dick, John
Publication:Rubber World
Date:Jun 1, 1992
Previous Article:High performance precipitated silica equivalent to fumed silica in silicone reinforcement.
Next Article:Park-Ohio Industries'.

Related Articles
Predicting toughness.
Independent testing laboratories directory.
Rubber testing for injection molding.
New rheometers & test instruments emerge at recent shows.
Characterization of physical properties of SBR/carbon black masterbatch.
Effect of a MFA on the processing and mechanical properties of a CB-filled NR compound.
Rubber characterization by applied strain variations using the rubber process analyzer.
Effective processability measurements of acrylonitrile butadiene rubber.
Independent testing laboratories directory.
Effect of carbon black on dynamic properties.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters