Effects of processing and test parameters on compression set.

Compression set testing has been a basic part of evaluating rubber compounds for a very long time, and is one of the comparatively few standard tests that would seem to have a reasonably direct relationship with a particular function, that is, retention of sealing force for products such as o-rings. The normal test specimen is a cylinder or button of nominal 12.5 mm height and 29 mm in diameter, although testing can be done on other types of specimens when necessary. There are two methods described in ASTM D 395; Method A employs a compressed spring to maintain a reasonably constant compressive force of 1.8 kN on the specimen, while in Method B, the specimen is subjected to a constant 25% compressive strain. In both tests, the specimen is subject to the stress or strain for set time periods at some elevated temperature deemed suitable for the compound, after which the specimen is removed from the fixture and allowed to return to room temperature. The specimen height is then measured and compared to its original height, and the percentage of non-recovery is reported as compression set.

The two methods do not provide readily comparable figures. If the button recovers to 90% of its original height in a Method A test, that would be a compression set of 10%. But the same 90% height retention in Method B would be calculated as a set of 40%, because the button would only have recovered 60% of the imposed strain. The two methods have a major difference in how the specimen is subjected to swain. In Method A, a button of a 45 durometer compound will deform much more than one of 75 durometer, so the strain state during the test duration will be very different, and the softer button is likely to take a higher set. In Method B, both buttons are subject to the same strain, although they will experience different stress, and the stiffer button becomes more likely to take the higher set. Method B is by far the more commonly used test.

The general assumption is made that lower compression set indicates a lower stress relaxation characteristic of the elastomer; which means that under a set strain, its sealing force against the containing walls will decline slowly rather than rapidly. This may not be a universally valid assumption, that is, two compounds with the same compression set might still differ significantly in how much their sealing force has changed. An actual test on the rate of change of sealing force with time would be preferable, but so far such tests have not been widely accepted or practiced, since they require significantly more applied technology to use.

It is also assumed that acceleration of the stress relaxation in the elastomer by elevated temperature is a valid technique. Again, this cannot apply for all elevated temperatures, since beyond some point the thermal effects will cause processes to occur in the elastomer that are very different from the mechanisms of stress relaxation and aging at operating temperatures. At that point, the test is no longer about stress relaxation, it becomes mainly or purely a test of heat resistance. Therefore, care must be taken to not use excessively high test temperatures as a means of predicting product performance for extended time periods.

What D 395 does not specify are the molding conditions for the test specimens. Thus, a manufacturer who vulcanizes his product for 15 minutes at 155[degrees]C is perfectly free to cure his test buttons for 20 minutes at 165[degrees]C. For many compounds, subjecting the material to higher heat or a longer cure cycle will produce a different enough distribution of crosslinks to significantly improve compression set. This could be considered a form of cheating on the test, but at present there is no formal restriction to hamper anyone from manipulating the vulcanization process in any way they wish. A recent survey of procedures used by various laboratories resulted in the following list of standard cure procedures for compression set buttons, which is by no means exhaustive:

* Tc90 plus two minutes;

* Tc90 plus five minutes;

* Tc90 plus ten minutes;

* Double Tc90;

* Triple Tc90;

* 20 minutes; and

* 30 minutes.

To add to the complexity of this, sometimes the cure temperature is the same as the production process, and sometimes it is higher, often by 10[degrees]C or more.

The question then arises as to how much effect molding conditions can have on compression set test results. In order to evaluate such effects, an experiment was drawn up which varied the vulcanization times and temperatures. To broaden the enquiry, four different polymer types were chosen, one of which was used with three different cure systems.

Then, to add further to the level of exploration of the testing process, four different test temperatures were used; but all tests were 70 hours in duration for the sake of simplicity.

The main body of the experiment was focused on Method B, but a limited amount of Method A data were also generated for purposes of comparison.

Experimental

The four polymer types were: SBR (three cure systems), EPDM, NBR and CR. These were selected since all have at least reasonable heat resistance and can logically be tested at the same temperatures with no concern about polymer breakdown. All were compounded to nominal 65 A durometer. A conventional sulfur cure was used for the SBR, EPDM and NBR compounds, while the CR was vulcanized using its own particular system. In addition, the SBR compound was cured with a full efficient vulcanization system (EV), as well as a peroxide system.

See table 1 for the formulation details. The compounds were formulated to be as similar as possible, using the same types of carbon black at as close to the same loading as would achieve equal hardness, with the same very basic antidegradant protection, the same process aid, etc. The goal was to ensure that compression set resistance would be primarily a function of the polymer type, rather than any of the other possible compounding variables, such as reinforcement and plasticizer. The use of three very different cure systems in the SBR was to explore how that factor alone might affect properties.

The two temperatures chosen for vulcanization were 150 and 165[degrees]C. Cure times were Tc90, Tc90 plus five minutes, and Tc90 plus 10 minutes. These were arrived at by educated guess, as Tc90 is a very minimal cure and an additional 10 minutes is ample for extra time, with the point midway between as possibly sufficient. The four test temperatures were room temperature, 70, 100 and 125[degrees]C. Room temperature testing provides information on how much of the stress relaxation takes place without acceleration, while the increasing temperatures of the other tests will clarify how much acceleration and/or change in the results is accomplished by heat.

Thus, the pattern for the vulcanization times and the test temperatures was as shown in table 2, which was repeated for each of the two cure temperatures and each of the six compounds. This was a full factorial designed experiment using one factor at two levels (cure temperature), one at three (cure time), one at four (test temperature) and one at six (compound) for a total of 144 runs (each using two buttons).

In addition, Method A testing was done using three compounds (all SBR), one cure time (Tc90 plus 10'), one molding temperature (165[degrees]C) and all four testing temperatures.

Standard processing and physical tests were performed using the following methods:

* MDR rheometry--ASTM D 5289;

* Hardness--ASTM D 2240;

* Tensile, elongation and modulus--ASTM D 412, method A;

* Heat aging--ASTM D 573; and

* Compression set--ASTM D 395, methods A and B

Rheometer and basic physical test data are shown in table 3. Method B compression set data are in table

4: Method A data are in table 5.

Results and discussion

Method B results

The substantial amount of data generated requires a variety of approaches to digest and analyze. Percentage of set in Method B ranges from under 10% to over 60%, a very broad range that indicates clearly that the experimental factors have major effects on the test results.

The simplest form of examination is to simply compare average sets organized according to the major factors alone. For instance, a decrease in set as molding time increases would be expected. The respective averages of the Tc90, plus-5 and plus-10 molding times for all compounds at both molding temperatures and at all the test temperatures were 35.4%, 27.8% and 24.8%. This confirms the anticipated result. Further, the larger drop in set going from the least time to the median time indicates that Tc90 is, in fact, a marginal cure time.

Just how differently the compounds would react to set testing is not precisely predictable, but it is entirely predictable that they would show some contrasts, and they do. Overall sets by compound were: SBR--30.2%, EV cure--12.0%, peroxide cure--17.3%, EPDM--39.6%, NBR--44.0% and CR--33.1%. The effect of alternate cure systems on the SBR is fairly dramatic, and the superior performance of the EV system, compared to the peroxide system, is interesting. (Inclusion of a coagent in the peroxide system might improve its compression set resistance to the level of the EV system.) The contrasts between the polymer types when using comparable cure systems is more subtle, although still meaningful.

Set would be expected to increase with the test temperature, and the grand averages by increasing test temperature were: 14.2%, 20.0%, 36.6% and 46.6%; again confirming the normal model for the process. However, it is noteworthy that room temperature set is about 70% of the set achieved at 70[degrees]C, a temperature almost 50[degrees]C warmer. Clearly, the stress relaxation mechanism operates at a very detectable level, even without thermal acceleration. On the other hand, the more than doubling of the set figures going from 70[degrees]C to 125[degrees]C, again about a 50-degree differential, demonstrates that increasing heat can have a very marked accelerating effect.

The last factor was molding temperature, and the averages of all compounds at the 150[degrees]C versus 165[degrees]C molding temperatures were not significantly different from each other. This indicates that when the molding times are related to the Tc90 at the given temperature, roughly equal states of cure are achieved, as theory would predict. (This may not necessarily apply across an extreme contrast in temperature range, such as 40[degrees]C or more.)

Interestingly, there were no correlations between room temperature sets and any of the heat-accelerated sets. This suggests that the primary mechanism of stress relaxation at room temperature may not actually be affected by heat, even though heat clearly has the effect of increasing set.

The question then can be asked as to whether heat resistance plays a major part in compression set resistance. If so, then a correlation between standard heat resistance testing and set should exist. However, the very best correlation between any heat resistance and set tests is between change in M-100 from test slabs and the set of button specimens cured at 165[degrees]C for Tc90 plus 5; and that correlation has an R-squared of only 55%. By contrast, the correlation between change in M-100 and change in elongation for the compound has an R-squared of 90%.

Thus, it appears that Method B set at room temperature is a result of one stress relaxation mechanism, but a different mechanism takes over as higher temperatures become a major factor. And further, the heat-related mechanism that affects set is not closely related to the mechanism of simple heat aging of test slabs.

All of this is, in a sense, unfortunate, because it indicates strongly that acceleration of compression set testing through the use of elevated temperatures reveals little, if anything, about room temperature set; and because it also indicates that compounding to improve heat resistance will have little, if any, effect on elevated temperature compression set.

A detailed analysis of the data reveals two significant contrasts. While every compound displays lower set at the longest cure time, the differences in set with increasing cure time are very small for all the SBR compounds, and then increase significantly in the order of EPDM, then CR, then NBR. This difference must relate to some combination of the polymer's inherent properties and crosslink structure.

Also, average set increases for all compounds as the test temperature increases, but the three compounds whose cure systems are the full sulfur type show a much greater increase than the two alternate SBR cures and the CR cure. This contrast is readily relatable to the sensitivity of polysulfidic crosslinks.

Method A results

The three SBR compounds were used in this comparison, cured at the single time-temperature combination of Tc90 plus ten minutes at 165[degrees]C. Table 5 shows the same trends between the cure systems as was observed in Method B tests; with the full sulfur system highest in set and the EV system lowest. The set figures are much lower for Method A, which is due to the way in which they are calculated, using the original height as the base. If the very approximate conversion of the Method A set to a scale similar to Method B is made by the simple expedient of multiplying by 4 (the ratio of the 0.5 nominal height of a button to the Method B deformation of 0.125), then the figures become very comparable. For example, the room temperature Method B sets of the three compounds cured the same way were 10.24, 8.66 and 15.04%, and the 4x values for Method A are 10.36, 7.92 and 12.16%. (This means the actual permanent deformation of the buttons is reasonably comparable.)

This parallelism in level of permanent deformation would not necessarily be expected, given the differences in the methods as explained earlier. However, table 6 provides the actual dimensions of samples in one of the Method A tests, at 125[degrees]C. The initial deflection under load for the buttons is about 0.355 inches, not that far from the Method B imposed deflection of 0.375; so for these particular compounds at about 65 durometer A hardness, the initial strain state of the specimens in Method A is at least in the same ballpark as in Method B. This makes the roughly comparable results in actual permanent deformation more understandable. If the compounds had been 85 durometer or 45 durometer, then the initial strain state would have been appreciably different, and the final set would also have been different.

The change in deflection from the beginning of the test until its completion after 70 hours at temperature was not really large, although the tendency for the sample to grow in height due to thermal expansion may have had an effect there. As a minor additional bit of data, the samples were re-mounted in the fixtures under load a day later, and the average deflection then was only slightly less than it had been at the first mounting (table 6), even though the set taken during the test made the sample height significantly less before it was re-mounted. This may relate to the non-recoverable deformation that rubber samples show when first subjected to strain, the well-known Mullins effect.

The EV cured specimens deflected least in the fixture at the end of the oven cycle, which is a clear indication that they maintained a counter-pressure to the spring force better than either of the other compound variations. So in this case, the lower permanent set of those specimens did correlate with a higher retained sealing force under the conditions of the test.

The overall ratio of 4x Method A set to Method B set changes with increasing test temperature, going from about 1.14 at room temperature to 1.38 at 125[degrees]C. This indicates that the interaction of higher temperature and constant load makes the test more aggressive than the constant deflection Method B technique. However, the consistency of the comparative performance of the three compounds in both the Method A and Method B tests demonstrates that both methods have similar effects on the samples, at least in the case where sample hardness happens to result in an initial Method A deflection somewhere near the imposed deflection of Method B. With much harder or softer compounds, it is extremely likely that a dramatic contrast in results between the two test methods would be seen.

Recommendation

Since the effect of changing cure time on compression set results is, as expected, clearly significant, the present lack in ASTM D395 of any form of specified cure conditions for making test specimens is a weakness in the specification. Therefore, either D395 should be revised to include some inputs on appropriate levels of time and temperature for the molding of the specimens or, at the very least, require the conditions used to make the specimens to be disclosed as part of the test results report.

Conclusions

1. When specimen cure time is related to Tc90, moderately different cure temperatures can be used without a significant effect on test results.

2. Increasing cure time does have the effect of decreasing observed set.

3. Room temperature set and heat-accelerated set are each the result of different stress relaxation mechanisms, and do not correlate well.

4. Elevated temperature effects on accelerating compression set do not correlate with basic heat aging tests on molded slabs.

5. Crosslink type has a major effect on compression set, with polysulfidic crosslinks displaying substantial temperature sensitivity.

6. Polymer types can also have different set characteristics, even when the remaining compounding variables are kept as constant as possible.

This article is based on a paper presented at a meeting of the Rubber Division, ACS (www.rubber.org).

References

(1.) H.J. Jahn and H.H. Betram, "The compression set behavior of nitrile rubber," presented at the ACS Rubber Division Meeting, Cincinnati, OH, Oct 3-6, 1972.

(2.) D.M Chang, "Investigation of the structure-property relationships of improved low compression set nitrile rubbers," presented at the ACS Rubber Division Meeting, Las Vegas, NV, May 20-23, 1980.

by R.J. Del Vecchio, Technical Consulting Services and Ernest Ferro, Jr., Corry Rubber ([e-mail])

Table 1--formulations

                         CS-1    CS-2    CS-3    CS-4    CS-5    CS-6

SBR-1502                  100     100     100
EPDM, Royalene 525                                100
NBR, Nysyn 33-5                                           100
CR, Neoprene W                                                    100
Stearic acid                1       1               1       1       1
Zinc oxide                  5       5       5       5       5       5
TMQ, Agerite Resin D        2       2       2       2       2       2
N-550 black                25      25      25    17.5    24.5   25.25
N-990 black                50      50      50      35      49      50
Proaid AC-1142              2       2       2       2       2       2
Sulfur                    1.8                     1.8     1.3    0.25
Altax (MBTS)                                              1.2
Methyl tuads (TMTD)                                       0.3       1
DPG                               0.2                             1.5
Vanax A (DTDM)                      2
Akrochem BBTS            1.55       2             1.5
Butyl tuads (TBTD)                1.7
Akrochem DC40C (DCP)                      3.5
Total phr              188.35   190.9   187.5   165.8   186.3   188.0

Table 2

Cure time     Test temp.

Tc90                   RT
Tc90         70[degrees]C
Tc90        100[degrees]C
Tc90        125[degrees]C
Plus 5'                RT
Plus 5'                70
Plus 5'               100
Plus 5'               125
Plus 10'               RT
Plus 10'               70
Plus 10'              100
Plus 10'              125

Table 3--rheometer and physical test data

                       CS-1    CS-2    CS-3    CS-4     CS-5    CS-6

                      Rheometer at 165[degrees]C and 150[degrees]C

MH (165[degrees]C)    23.91   20.88   19.18   30.97    16.45   20.24
ML                     1.99    1.99    2.38    2.28     1.74     3.3
Tc90 (seconds)          636     840     681     547       98     203
TS1 (seconds)           183     159    44.4   156.6     34.2    33.6
MH (150[degrees]C)    25.36    22.8   17.01   31.98    17.32   20.61
ML                     2.39    2.25    2.63    2.69     2.04     3.5
Tc90                  1,578   2,269   2,552   1,391      197     458
TS1                     510   424.2     123   447.6     61.2    57.6

                      Tensile slabs cured Tc90 at 165[degrees]C

Tensile, MPa           15.5    13.7    13.0     8.0     12.9    19.1
Elongation %            296     282     404     268      408     228
Durometer A              67    65.9      63    66.3     66.4      69
100% modulus            4.6     4.2     2.8     3.2      2.8     7.3

                      Heat aged 70 hours at 100[degrees]C

Aged tensile loss %    1.37   -2.11   -9.03    3.87    -5.52   -7.79
Elongation loss %     -45.3   -35.5   -2.47   -43.7   -54.41   -21.9
Durometer increase      8.5     5.2     2.8     5.4        5     3.1
M-100 change %        95.76   42.69   -0.24   64.52   105.65   26.47

Table 4--compression set data (method B)

                           CS-1    CS-2    CS-3    CS-4    CS-5    CS-6

Cure time          Test            Percent compression set
@ 150[degrees]C   temp.
                   Deg.
                      C

Cured for Tc90       23   10.54    8.56   13.64   11.11   13.79   39.00
Tc90 + 5             23   10.32    8.59   13.16   11.02   15.41   35.65
Tc90 + 10            23   10.24    8.66   15.04   10.67   15.04   31.27
Cured for Tc90       70   19.49    8.63   16.86   29.76   47.08   20.07
Tc90 + 5             70   16.93    8.17   18.48   25.88   28.57   12.56
Tc90 + 10            70   15.81    7.84   17.69   20.55   23.22   10.51
Cured for Tc90      100   45.63   14.57   16.48   62.44   76.23   45.46
Tc90 + 5            100   40.32   12.20   18.87   56.30   57.42   34.39
Tc90 + 10           100   37.00   10.94   16.72   53.15   48.30   30.81
Cured for Tc90      125   60.08   19.21   30.65   71.15   81.67   55.09
Tc90 + 5            125   53.94   17.58   27.69   67.45   64.05   44.77
Tc90 + 10           125   52.17   15.75   26.81   63.38   53.76   44.58

Cure time          Test            Percent compression set
@ 165[degrees]C    temp
                   Deg.
                      C

Cured for Tc90       23    9.68    6.97    9.31    8.43   20.93   35.94
Tc90 + 5             23    9.44    6.97   10.33    8.39   14.55   21.27
Tc90 + 10            23   10.11    7.00    9.72    8.39   14.54   19.11
Cured for Tc90       70   20.22   10.11   14.13   38.09   64.49   32.57
Tc90 + 5             70   18.35    7.81   14.94   22.18   27.93   12.12
Tc90 + 10            70   16.72    7.78   14.38   17.90   23.27   10.81
Cured for Tc90      100   43.44   18.91   14.39   66.79   83.10   52.73
Tc90 + 5            100   37.89   12.83   12.99   53.56   49.28   30.21
Tc90 + 10           100   31.63   10.15   12.41   47.48   36.39   24.42
Cured for Tc90      125   57.77   22.87   27.82   72.30   90.96   61.30
Tc90 + 5            125   50.98   18.29   23.49   64.36   58.55   47.82
Tc90 + 10           125   45.67   16.40   19.95   59.00   46.79   41.86

Table 5--compression set data (Method A)

SBR compound    Sulfur     EV   Peroxide

Room temp.        2.59   1.98       3.04
70[degrees]C      4.79   2.28       5.15
100[degrees]C     9.07   2.78       4.90
125[degrees]C    14.06   5.16       8.17

Table 6--Method A specimen dimension
(125[degrees]C test)

SBR compound               Sulfur      EV   Peroxide

Original height (inches)    0.502   0.504      0.514
Loaded (room temp.)         0.356   0.355      0.355
70 hours (hot)              0.336   0.348      0.332
Recovered                   0.431   0.478      0.472
Re-loaded (room temp.)      0.346   0.362      0.341