Compounding effects on physical properties and rubber-metal bonding.

Many experienced rubber technologists believe that the capacity of an elastomeric compound to bond to substrates can be affected by some ingredients, and most are well aware that different primer/covercoat combinations will provide bonds of different strength and durability. However, very little data have been published showing effects of compounding ingredients on bonding, or how ingredients and bonding systems interact.

In rubber compounding, it is a well known principle that, "you can never change just one property," so the variation of the chosen ingredients was fully expected to have effects on many other properties. The only concern was to keep all the compounds within a sufficiently narrow range of hardness as the simplest means of keeping comparisons between them at least reasonably valid. Just how the properties, other than bonding, might change was also of major interest in this study.

Assuming the hypotheses that certain ingredients might have effects on bonding, and that there could be different responses to those ingredients, depending on the bonding system used, it was decided to choose a polymer commonly used in rubber-metal components, and to make a series of compounds differing only in comparatively minor ways. The polymer was natural rubber (NR), reinforced with a standard carbon black to a durometer A hardness of 35-40. The formulation is shown in table 1, with the first four ingredients being constants, and the remaining tour changing within the ranges shown.

The types of ingredients evaluated for effects on compound properties and bonding included plasticizers, antidegradants and curatives. Cure type (conventional vs. semi-EV) was also evaluated. The first three of these categories were set up to have four members each, while the last category has only the two levels. A complete factorial design would call for 128 compound variations to be mixed and tested, but a special design (a Hyper-Graeco-Latin Square) with only 16 runs was adapted to serve the purpose. In HGLS designs, the capacity to diagnose and quantify interactions is sacrificed for the sake of efficiency, but all main effects can be analyzed validly.

The plasticizers and antidegradants were deliberately used at reasonably generous levels to facilitate detection of any likely effects from their presence. In addition, high and low levels of sulfur were used to represent standard versus semi-EV curing methods, with each curative adjusted appropriately to the amount of sulfur so as to keep the amount of crosslinking at least roughly equivalent for all the cure systems. Since vulcanization chemistry is complex, even using molar equivalents of vulcanizing agents does not guarantee equal crosslink concentration; adjustments to each sulfur/accelerator ratio were made to keep hardness roughly equivalent, as a concession to common compounding practice. Initial work indicated that the plasticizers and accelerators all differed slightly in their effects on hardness, so adjustments of a few phr of plasticizer were made to level out the effects, and fractional adjustments of accelerators were made to the same end.

All batches were mixed using the same lots of raw materials and the same mixing procedure. Test pieces were made and underwent testing the same way and, in general, scientific rigor was applied to all stages of data generation. ASTM test methods were used wherever applicable.

Discussion

Physical properties

The experiment was set up in a 16-run design, with three categories of discrete variables at four levels and one (sulfur) at two levels. The categorical variables are shown in table 2.

The actual pattern of the experiment for all the independent variables is shown in table 3, with their concentrations displayed in parts per hundred rubber (phr).

The durometer readings of all the batches fell in the range of 37 [+ or -] 3, which confirms the adjusted levels of the plasticizers and cure systems as being reasonably appropriate to maintaining constant hardness in the compound series.

Table 4 presents standard physical properties of the batches, including changes after heat aging for 70 hrs. @ 100[degrees]C, as well as dynamic modulus and loss factor in shear (10% strain, 10 Hz). The testing was performed using a Lord test procedure and double lap shear test specimens.

Although durometer hardness values were distributed more or less within normal scatter for that test, the other results were not. For example, elongation varied from 535 to 655%, compression set from 17.3 to 43.6%, and tear from 29.1 to 64.8 N/mm. Since these levels of variation fall well outside normal scatter, clearly the compounding factors had some effects on properties other than hardness, as had been expected.

Since ultimate elongation is one of the defining characteristics of an elastomer, that property will be examined first. In figure 1, the two factors that affect elongation are shown. (Curative identified on the x-axis, plasticizers labeled in the chart body.)

[FIGURE 1 OMITTED]

The type of cure might have been expected to show an effect on elongation, with the conventional (higher sulfur) system providing more elongation, but there was no detectable difference in results by sulfur level. The aromatic and naphthenic oils provide higher elongation, while amongst the curatives the TMTD is clearly lowest and ZnDMC highest in their effect.

The 100% tensile modulus (M100) is a more sensitive measure of compound stiffness than durometer readings, and does reveal significant contrasts between the compounds. The principal factor affecting M100 is curative type, as illustrated in figure 2. (There is about a 90% probability that the two antiozo-nants tend to reduce M100 slightly as compared to the antioxidants; in the interests of brevity, this kind of smaller secondary effect will not be illustrated.) Here it can be observed that the curatives fall into two groups, with the CBS and TMTD producing higher modulus, and the MBTS and ZnDMC lower. At this point, the question of whether or not the ZnDMC system has produced full cure might be considered, given the lower modulus and higher elongation already noted.

[FIGURE 2 OMITTED]

However, compression set (22 hrs./100[degrees]C) can be considered a very good indicator of cure state, and examining that property reveals important trends. Two factors are significant in their effect on set, both of which are shown in figure 3.

[FIGURE 3 OMITTED]

As might have been predicted for vulcanization systems of this type, higher sulfur always results in higher set. The consistent pattern of results by curative type at both high and low sulfur confirms the real differences in their effects, with the TMTD and ZnDMC both showing improved resistance to set. The ZnDMC did not show the high compression set that would be expected from an incompletely vulcanized material. Although this system may provide fewer crosslinks than the others (hence the higher elongation and lower modulus), there are ample crosslinks of a stable nature to afford excellent compression set resistance.

Although the tensile strength of a good quality rubber compound is seldom related to its functionality, and is actually a function of both material stiffness and ultimate elongation, analysis of this property is still of interest. There is statistical significance to the variation of tensile strength by curative and by antidegradant, and figure 4 illustrates the effects. There is a gradation of strength, with CBS at the upper end and MBTS at the lower. The antiozonants detract from strength by almost 10%, compared to the antioxidants.

[FIGURE 4 OMITTED]

Tear strength is a complex response, which does not always correlate well with other physical properties. In this case, ASTM D624, Method C was used, and a fairly clear relationship (correlation coefficient of 0.72) with tensile strength was observed. The same kind of effects by curative and antidegradant applied.

Even though, as stated above, the compounding was done with the goal of reaching a consistent level of durometer hardness, the statistical analysis of the hardness readings was performed to see what might be revealed. Although the range of [+ or -] 3 durometer points is close to normal measurement scatter, the power of a good designed experiment will sometimes make it possible to identify even comparatively subtle effects on a response. In this instance that proved to be the case, since two factors were shown to have significant effects on hardness. This demonstrates how even very competent compounding practice is still subject to subtle effects not easily detected without statistically valid means of analysis. Figure 5 shows once again that the antiozonants act differently than the antioxidants, lowering hardness by over two points.

[FIGURE 5 OMITTED]

Since the cure state of the compounds, by the criterion of compression set, was not affected by the antidegradants, it does not seem this softening effect is related to crosslink density. Possibly the large aromatic structures of the para-phenylene diamines simply have a plasticizing effect.

The other factor influencing hardness was the curative type, which is quite consistent with all the other effects already noted for that class of ingredient.

Given the major contrasts in the antidegradant systems used, some variation in heat resistance was to be expected. Loss of elongation is a commonly used criterion for effects of heat aging, and the range of decline, from -9 to -47%, was ample to confirm differences between compounds. Surprisingly, the analysis could not establish any contrasting effects from the antidegradants. Instead, it was the cure system type that had by far the greatest effect (average loss of -13% for semi-EV cures, versus -32% for conventional cures). Lesser contrasts between accelerators also contributed (figure 6). Possibly the substantial (but not unrealistic) level of antidegradant was such that it leveled off any contrasts in effectivity between them, at least, in regard to loss of elongation.

[FIGURE 6 OMITTED]

However, change in hardness is also a commonly used measure of aging or heat aging, and analysis of this response gave contrasting and somewhat surprising results. All three factors, other than cure system type, were seen to have meaningful effects (figure 7). The compounds containing antioxidants heat aged much better than those with antiozonants; CBS and TMTD curatives gave better aging; and paraffinic and ester type plasticizers also aged better.

[FIGURE 7 OMITTED]

The factor effects on loss of elongation might have been expected to give at least as good an analysis as increase in hardness, and the conventional high sulfur cured compounds which aged worse in loss of elongation might have been expected to also age worse by increase in hardness. One flaw in this type of design is lack of ability to diagnose interactions between factors, and it is possible that for a few responses the interactions (for instance, between type of curative and conventional vs. semi-EV cure system) may contribute enough to either provide a misleading analysis or raise the scatter level to the point where real effects cannot be separated from the high noise level. On the other hand, the many results that fit perfectly with known trends in compounding (as were seen for compression set) or that appear very consistently for several responses (like the difference in effects of the antiozonants vs. the antioxidants) provide reassurance that, for the most part, the design has made possible a valid and accurate understanding of how the factors affect the compound properties.

Dynamic properties also varied appreciably among the compounds, and the elastic shear modulus (G') was affected by three factors. The conventional cure yielded a slightly reduced average G' of 0.524 MPa as compared to 0.537 MPa for the semi-EV cure. In figure 8, the effects of curative and plasticizer types are contrasted, showing that TMTD cures and aromatic oil plasticizer both delivered increased shear moduli.

[FIGURE 8 OMITTED]

The dynamic loss factor (tangent delta) was also affected by TMTD and aromatic oil, as displayed in figure 9. The aromatic oil appears to increase the elastic modulus and produce an even larger jump in the viscous modulus, thereby increasing tangent delta; while the TMTD primarily increases G', so that tangent delta is decreased. This contrast in behavior is most likely dictated by the TMTD producing more mono- or di-sulfidic crosslinks, which stiffen the polymer matrix but do not contribute to hysteresis.

[FIGURE 9 OMITTED]

Processing effects

The types of curatives would certainly be expected to affect processing differently, and table 5 displays some of the conventional measurements of processing properties.

Merely looking at the Mooney scorch time to a 10-point rise immediately reveals substantial contrasts. In figure 10, it is readily evident that the ZnDMC and TMTD allow much less processing safety, with the MBTS intermediate and the CBS the safest of all. (But the close correlation between time to a 10-point rise and the [Tc.sub.90] indicates that higher processing safety comes at the price of longer cure times.)

[FIGURE 10 OMITTED]

Maximum torque (MH) was affected by the same class of ingredients seen as factors for compound hardness, acting in the same manner. The good correlation between MH and hardness would lead one to expect those results.

Minimum torque (ML) is only a function of plasticizer type, as illustrated in figure 11. The aromatic oil clearly stands out as a less effective internal lubricant for the compound.

[FIGURE 11 OMITTED]

Bonding properties

Since the compounds are close in hardness and none has grossly deviant properties (in the sense of very poor strength or very high compression set, etc.), it is reasonable to examine their comparative bonding capacities.

Each compound was evaluated in both peel and tension/ shear modes (ASTM D429, Methods B and F); and each was bonded using two different adhesive systems, one solvent based and the other aqueous. These systems were:

Type          Primer             Covercoat

Solvent       Chemlok 205        Chemlok 220
Aqueous       Chemlok 8007       Chemlok 8560

In table 6, the bond strengths are shown for the sample pieces that were made using all the combinations of compounds and adhesive systems. The run numbers correspond to the compound compositions, as displayed in table 3.

As with any designed experiment, the range of responses is immediately examined to see if it demonstrates a significant level of variation beyond normal scatter. For the solvent system, the peel values run from 3.99 to 6.11 N/mm, and the Method F values (also known as the buffer test) from 1.10 to 3.08 MPa. The former range is reasonably significant for that test, and the latter is dramatically wide, so there is no question that bond strengths have been affected by the chosen compounding variables.

Statistical analyses of the data reveal that while some of these responses have the same model, others are quite different. The models for each type of adhesive system were examined for both types of test.

When the four sets of bonding data were analyzed separately, the plasticizer factor appeared to be marginally significant sometimes, not quite significant (by the criterion of 90% probability) at others. However, the trend seen was always the same, and in figure 12, the consistent pattern by individual plasticizers is easily observed in figure 12.

[FIGURE 12 OMITTED]

Even though the individual analyses do not provide 95% statistical confidence for any one set of data, the repeated pattern for all the data sets confirms that the contrasts shown are real. Aromatic oil in these NR compounds raises bond strength in both tests, with a downward progression to naphthenic and paraffinic oils, and finally the ester lowest of all. Since there is not a good correlation, in general, between these two methods of bond testing, the consistency of this effect is especially noteworthy.

The consistently higher figures for the solvent based adhesive demonstrate a slight advantage in bonding. Other differences between the solvent and aqueous systems are revealed in that the only other factor affecting the peel strength of the water based adhesive is the cure type, with the semi-EV averaging 4.68 N/mm versus 5.42 for the conventional cure; while the second factor affecting the solvent system bonded peel strength is curative type. Figure 13 shows that for this type of test and adhesive, TMTD cures result in the strongest bonds, and ZnMDC the weakest.

[FIGURE 13 OMITTED]

This contrast, sulfur level affecting aqueous system bonds while curative type affects solvent system bonds, must relate to the particular chemistry of the adhesives deposited on the metal surface by the different products. It should be noted that in no case were the bonds achieved notably deficient in any sense; all these data demonstrate is an appreciable but not overwhelming difference in strength. Other important characteristics (such as long term durability in different environments) might show other contrasts, but there is no present basis for any prediction of them.

Method F stresses bonds in a very different way than Method B; so even though the common effect of plasticizer type has been observed in the data for both tests, other contrasts might be expected from the analysis of the buffer (Method F) data. And, in fact, neither cure system type nor curative type affected buffer strength for either adhesive system. Instead, antidegradant affected both. Figure 14 illustrates the interplay of the effects and also shows that the solvent based adhesive again provides consistently stronger bonds in this type of test.

[FIGURE 14 OMITTED]

The commonalities are that 7PPD always results in lower bond strength and the antioxidant mixture results in higher. Interestingly, the IPPD works much better with the solvent based adhesive than the aqueous, and the very commonly used TMQ antioxidant is comparatively insensitive to these adhesive systems.

These contrasts demonstrate how complex the interaction of compounding ingredients with bonding agents can be. There is at least a suggestion that antiozonants may tend to interfere with bonding, possibly through their tendency to migrate towards surfaces. (However, the 3 phr level used is certainly a possible multiplier of such an effect.)

Summary observations

The main effects of the four categories of compounding variables are:

* Conventional sulfur level (versus semi-EV cure)

--results in worse aging, according to loss of elongation;

--increases compression set;

--decreases dynamic shear modulus slightly; and

--increases bond strength in peel test using aqueous adhesive.

* Accelerator type

--CBS and TMTD give higher rheometer max torque;

--CBS and MBTS have greater scorch safety (time to 10 pt. rise) and longer [T.sub.90];

--CBS and TMTD raise durometer hardness and M100 tensile modulus;

--ZnMDC has highest elongation, TMTD lowest;

--CBS yields highest tensile strength, MBTS lowest;

--ZnMDC has poorest aging, according to loss of elongation;

--MBTS has worst aging, according to increase in hardness;

--CBS gave the best tear strength, ZnMDC the worst;

--TMTD and ZnMDC provide better compression set;

--TMTD gives slightly higher dynamic shear modulus, lower loss factor; and

--TMTD and MBTS increase bond strength in peel test (solvent adhesive).

* Plasticizer type

--bond strength across both tests and types of adhesive declines in the order of aromatic, naphthenic and paraffinic oils, with the ester plasticizer lowest;

--aromatic oil provides the least compound plasticizing (highest rheometer ML);

--aromatic oil yields highest elongation, ester the lowest;

--ester and paraffinic oil age best, according to increase in hardness; and

--aromatic oil increases dynamic shear modulus and loss factor.

* Antidegradant type

--antiozonants reduce maximum rheometer torque, compared to antioxidants;

--antiozonants also reduce durometer and tensile modulus by comparison;

--tensile and tear strengths decrease when antiozonants are used;

--antiozonants increase the durometer change in heat aging; and

--buffer bond strength is decreased by both antiozonants when the aqueous adhesive is used, but with the solvent adhesive only the 7PPD shows the effect.

Conclusions

Even when formulated to maintain similar final compound hardness, comparatively subtle changes in compounding ingredients can have numerous significant effects on a wide range of properties, including standard physicals, processing, aging and bonding tendencies.

Different compound properties are often affected very differently by ingredients, so that generalization about the effects of any given ingredient is often risky.

The type of accelerator used was the single largest influence on all the tested properties, which may be an indicator that the type and distribution of crosslinks is more important than basic crosslink density.

The use of a designed experiment in the investigation permitted analysis of effects whose magnitude would make it difficult or impossible to detect in more limited or unstructured experiments.

Table 1 - basic formula

SMR5-CV60                100.00
Zinc oxide                 5.00
Stearic acid               2.00
N762 black                40.00
Antidegradant              3.00
Plasticizer           22-25 phr
Curative              0.4-5 phr
Sulfur           0.7 or 2.1 phr

Table 2 - categorical compounding ingredients

Antidegradants                Curatives      Plasticizers
IPPD (antiozonant)                 MBTS    Paraffinic oil
IPPD (antiozonant)                 TMTD    Naphthenic oil
TMQ (antioxidant)                   CBS      Aromatic oil
Mixed amines (antioxidant)        ZnDMC             Ester

Table 3 - 16-run design for categorical control factors

Ingredient       Run 1       2       3       4       5

Mixed            3 phr
Amines
7PPD                         3
IPPD                                 3               3
TMQ                                          3
Aromatic oil        22
Paraffinic oil              25                      25
Naphthenic oil                      25
Ester                                       22
Sulfur               0.7     0.7     0.7     0.7     2.1
CBS                  2.5                             0.5
TMTD                         1.5
ZnMDC                                5.0
MBTS                                         3.5

Ingredient           6       7       8       9      10     11

Mixed                        3
Amines
7PPD                                 3                      3
IPPD                                                 3
TMQ                  3                       3
Aromatic oil        22                                     22
Paraffinic oil
Naphthenic oil                      25      25
Ester                       22                      22
Sulfur               2.1     2.1     2.1     0.7     0.7    0.7
CBS                                          2.5
TMTD                 0.4                             1.5
ZnMDC                        0.5                            5.0
MBTS                                 1.0

Ingredient          12      13      14      15      16

Mixed                3               3
Amines
7PPD                         3
IPPD                                                 3
TMQ                                          3
Aromatic oil                                        22
Paraffinic oil      25                      25
Naphthenic oil                      25
Ester                       22
Sulfur               0.7     2.1     2.1     2.1     2.1
CBS                          0.5
TMTD                                 0.4
ZnMDC                                        0.5
MBTS                 3.5                             1.0

Table 4 - standard properties

Run     Duro. A    Tensile    Elong.    M100
       hardness      MPa         %      MPa

  1         38      23.79       635     1.03
  2         37      18.52       570     0.93
  3         34      19.07       655     0.79
  4         36      19.48       605     0.90
  5         36      20.41       610     0.93
  6         40      22.59       625     1.07
  7         38      18.59       615     0.90
  8         35      18.62       645     0.86
  9         38      22.24       620     1.03
 10         38      17.97       535     1.03
 11         34      19.48       655     0.83
 12         39      19.62       575     0.93
 13         38      21.07       595     1.03
 14         40      21.03       580     1.17
 15         35      19.45       635     0.93
 16         36      17.07       610     0.93

Run      Tear      Comp.     Duro.     % Tens.
         N/mm        set     chng.     chng.

  1       64.8       33.7         4     -8.4
  2       36.9       17.3         4     -5.2
  3       29.8       25.0         6     -8.7
  4       38.3       29.2         2     -7.8
  5       37.8       43.6         6    -49.7
  6       45.2       27.1         1    -44.6
  7       37.8       29.0         0    -71.4
  8       29.1       38.8        12    -13.7
  9       47.6       32.0         4     -5.3
 10       43.8       18.1         4     -3.6
 11       34.8       21.9         8     -6.5
 12       48.0       27.2         3     -9.5
 13       46.9       38.7         3    -42.1
 14       41.1       24.8         1    -62.6
 15       34.8       28.0         2    -57.4
 16       37.3       40.4        11      2.4

Run    % Elng.        G'       Tan
        chng.        MPa     delta

  1       -9.4      0.586     0.105
  2      -10.5      0.524     0.076
  3      -19.8      0.505     0.100
  4      -12.4      0.512     0.093
  5      -33.6      0.498     0.090
  6      -25.6      0.594     0.109
  7      -47.2      0.506     0.088
  8      -30.2      0.481     0.093
  9      -12.9      0.534     0.089
 10      -12.1      0.570     0.074
 11      -17.6      0.557     0.109
 12      -10.4      0.507     0.083
 13      -22.7      0.518     0.079
 14      -43.1      0.554     0.083
 15      -40.9      0.498     0.097
 16      -16.4      0.539     0.115

Table 5 - processing characteristics

Run     MDR       MDR     [Tc.sub.90]     Mooney -
         ML        MH           min.     minutes to
                                        10 pt. rise

  1     0.89      6.20          9.97            8.2
  2     0.63      4.77          2.74            3.2
  3     0.80      4.13          3.45            3.6
  4     0.70      5.37          8.61            5.6
  5     0.71      5.45          9.51            6.7
  6     0.87      6.21          3.69            4.0
  7     0.79      5.44          2.62            2.3
  8     0.76      4.71          7.28            4.6
  9     0.62      5.98          9.41            9.0
 10     0.80      5.63          3.19            3.7
 11     0.86      4.54          2.44            3.2
 12     0.69      5.50          8.78            5.6
 13     0.74      6.02          7.95            8.0
 14     0.72      6.09          4.04            4.0
 15     0.74      5.20          3.10            3.1
 16     0.84      5.38          7.33            7.3

Table 6 - bond test results

                 (Peel)             (Tension/shear)
           Method B test data      Method F test data
Run      205/220     8007/8560     205/220     8007/8560

  1        5.83         5.30         3.08         2.10
  2        5.69         4.78         1.49         1.65
  3        5.60         4.17         2.76         1.55
  4        4.64         4.59         1.94         1.74
  5        5.20         5.36         1.54         0.94
  6        5.93         5.43         2.68         2.08
  7        3.99         5.01         2.32         1.63
  8        5.90         5.55         1.89         1.78
  9        4.97         5.04         1.60         1.99
 10        5.09         4.62         2.10         1.33
 11        4.92         4.39         1.96         1.40
 12        4.97         4.59         2.34         1.92
 13        4.94         4.67         1.10         1.03
 14        5.81         5.76         2.99         1.73
 15        5.18         5.18         2.28         2.00
 16        6.11         6.44         2.34         1.72

by R.J. Del Vecchio, Technical Consulting Services