Optimization of high performance properties.

Cast polyurethane technology represents an almost ideal combination of elastomer performance, processing latitude and economics. As with any technology, however, pressures from competitive technologies and a demanding marketplace require manufacturers to seek continuous improvements. In order for cast polyurethane elastomer markets to grow and provide the expanding commercial opportunities that suppliers, processors and end-users need, continuing efforts to improve the perforinance, processability and cost/performance ratios of cast polyurethane elastomers are required. One area in which polyurethane materials have exhibited significant potential and in which opportunities continue to expand is in dynamic and high temperature applications.

For this reason, we initiated an investigation of structure property relationships with the express goal of improving the performance of one class of cast elastomers in these demanding application areas. In order to put this work in context, it is necessary to first review the current state of high performance polyurethane technology and the opportunities which exist. The rationale for the approach pursued in this work will then be discussed, followed by a description of the optimization techniques employed, and a review of the preliminary laboratory results which have identified the best candidates for field testing.

Currently, a recognized leader in the area of dynamic polyurethane elastomer performance is the Vulkollan elastomer system, a product of Bayer, which is prepared by the reaction of naphthalene diisocyanate (NDI) with one of several adipate ester polyols to produce a prepolymer which can then be chain extended with short chain diols such as butane diol (figure 1).

The various combinations are distinguished by designations in which the weight percent of NDI in the formulations is given. As an example, V27, the workhorse composition which produces a 93A elastomer, contains 27% NDI. This network has been remarkably successful in high load and high temperature applications.

There are, however, inherent drawbacks with the use of the Vulkollan system. First, NDI does not produce shelf stable prepolymers, which requires that processors react monomeric isocyanate and polyol on site in precise ratios and under controlled temperature conditions with a considerable investment in equipment and training. The curative/crosslinker must be added to the newly made resin in precise ratio within a brief time period. Gel times are rapid, so machine processing is essential. Depending upon part geometry, cure schedules of varying complexity must be employed. Thick profiles require complex schedules and must be annealed by several cycles of 110[degrees]C for 12-24 hours followed by 1-5 days at room temperature. Shrinkage can be substantial, resulting in part cracking when complex geometries or long dimensions are involved. Because of these drawbacks, there is a limited number of processors capable of making these elastomers despite the obvious advantages which can be obtained with these systems. Added complications in the processing of NDI based systems include the health and safety concerns associated with isocyanate. In addition to the processing difficulties, a further drawback is the expense of the total system. Recent increases in the price of NDI have put increased cost pressures on processors and end users of these systems

Alternative systems

What alternatives exist? Dynamic performance improvements can be obtained by the use of PPDI systems, but these materials have unfavorable cost/performance characteristics, processing drawbacks similar to NDI, and, they pose health and safety concerns with respect to the free monomer. Whether PPDI systems make significant,commercial inroads remains to be seen. In addition to dynarnic deficiencies, conventional MDI systems have traditionally not provided the high temperature stability required in high-speed, high-load applications although recent developments may make these systems worth reexamining. Conventional TDI-prepolymeramine systems have also proven deficient although the use of low monomer, low oligomer prepolymers has greatly improved the performance of MBOCA-cured elastomers. Even greater improvements have been reported with the use of MCDEA, a product of Lonza which is marketed under the name Lonzacure (figure 2).

This curative is similar in structure to MBOCA but is thought to be free of the health and safety concerns which are sometimes associated with that material. Due to its low melting point (88-90[degrees]C), processing will be straightforward for most TDI processors. Disadvantages include the shorter pot-life relative to MBOCA, which requires machine processing in the higher durometer ranges; and the pricing of the overall system, which is intermediate between that of the NDI/PPDI based systems and MDI/TDI-MBOCA systems.

Despite the drawbacks, the potential advantages of MCDEA-TDI systems over alternative materials prompted an exploration of the full capabilities of MCDEA cured systems in the 93-95 Shore A durometer range. The initial phase of this effort was an optimization program which aimed at finding formulations and processing conditions that would produce the best balance of static and dynamic properties in laboratory tests and were amenable to field testing. One question that had to be resolved first was which factors were to be explored and which properties are appropriate measures of performance.

Structural and processing factors

A unique characteristic of Vulkollan is the combination of good dynamic properties due to highly structured hard block domains combined with the toughness typical of elastomers containing polyester soft blocks. One obvious question which can be asked, therefore, is the effect of backbone type (ether vs. ester) on the properties of MCDEA cured resins. A second question is the benefit of low-monomer, low-oligomer resins in combination with MCDEA. These prepolymers have been shown to be dynamically superior to conventional prepolymers when MBOCA is the crosslinker (ref. 1). Cure ratio (the ratio of crosslinker to isocyanate), which affects the crosslink density of the elastomer network and impacts performance, must also be considered. The processing factors which are of interest are the cure temperature and cure time, both of which have been shown in other systems to have an effect on properties.

Responses

The properties, or response factors, which are of interest fall into two groups; static (low frequency strain rate) properties and dynamic (high frequency strain rate) properties. The standard static properties of importance are the well known ASTM type tests: tensile-strength, elongation, tear strength, and of particular interest, rebound, abrasion resistance and compression set. The appropriate measures of dynamic performance are not so obvious. If one systematically considers the steps which lead to failures in dynamic applications, however, the problem becomes more tractable. Failure in dynamic applications invariably results from a buildup of heat which leads to degradation of the polymer matrix and failure due to chunking (tear failure), blow out (melting) or debonding. The discreet operations are infusion of energy, conversion of this energy to heat, and the degradation of the elastomer due to heat.

How is energy infused into the system? Although thermal energy from the surroundings can be a factor, the major entry is via deflection of the specimen under load. It should be obvious that formulations of comparable hardness which deflect less under similar loads should perform best in dynamic applications. An alternative way of ranking materials is to measure the force necessary to produce a given deflection and those formulations requiring the highest compressive stress should be more resistant to heat buildup. Since ten percent deflection represents the maximum practical deflection in most applications, the stress necessary to accomplish this deformation is used here as a figure of merit for comparing compositions. Compression deflection tests are performed by cycling test buttons three times between 5% and 25% deflection at 5% ncrements and measuring the stress values on the third cycle. A typical deflection experiment is shown in figure 3.

Having assessed energy input, the process, by which mechanical energy is converted to heat must be considered. The techniques of dynamic mechanical analysis (DMA) have proven useful in predicting the tendency of materials to build up heat under load (ref. 2). The tangent delta (tan [sigma]) of a material, which is the ratio of the loss modulus to the storage modulus, is of use in predicting energy absorption when the maximum strain is known. A related parameter, the loss compliance, which is also a function of the loss and storage moduli, is more useful when the maximum stress is known. The area under the loss compliance curve should be an indicator of the tendency of the material to convert mechanical energy to heat. One potential complication, however, is that most DMA tests are performed at a single frequency. Since frequency can vary in many applications, it is important to remember that time and temperature can be related mathematically (ref. 3) and that high frequency perturbations are equivalent to lower frequency perturbations performed at reduced temperature. Simply put, the loss compliance curve shifts to higher temperature as the frequency increases. Consequently, it is difficult to predict the performance of materials in applications where the frequency varies greatly without regarding the entire temperature spectrum. In this work we examined the area under the stress strain curve from 0 to 150[degrees]C and from -78 to 150[degrees]C. These calculations were performed using Simpson's rule. A typical graph with the areas of interest is shown in figure 4.

The final parameter to consider is the ability of the material to operate at elevated temperatures. It has been shown that the precipitous decline of the storage modulus curve in the dynamic mechanical spectrum which marks the end of the "rubbery plateau" corresponds to the melting point of the matrix hard segment. Materials with higher hard segment melt points, therefore, should have higher operational temperature ranges. A typical DSC trace for a polyester polyol resin cured with MCDEA is shown in figure 5. Obvious features of the trace are the glass transition temperature (Tg) at 34[degrees]C and the melting point of the hard segment which occurs at 239[degrees]C.

Experimental design

An optimization protocol which has become popular in recent years is the so called "Taguchi" method in which an orthogonal array or a series of balanced experiments are performed (ref. 4). As an example, if three factors or variables were being examined in the transformation of chemical compound A to compound B, the test points represented by the black dots shown in figure 6 would adequately describe the experimental region.

The benefit of this strategy is that a very large experimental space can be probed in a small number of experiments while retaining a high degree of precision. It's difficult to visualize the experimental space in the current work since it is not three dimensional as in figure 6, but rather, five dimensional with three, three-level variables (table 1). The total number of experiments required to evaluate every combination is 108, but by using an orthogonal array, we are able to cover the same experimental space with 18 experiments and two repeat points as a determinant of error. A summary of the array points can be found in table 2. The experiments were performed in a semi-randomized order since the processing was done by machine and complete randomization was not practical.

Table 1 - factors for analysis

Factor           Levels    Values
Backbonetype        2     Ester vs. ether
Resin type          2     Conventional vs. low
oligomer
Cure ratio          3     85%,95%,105%
Cure temperature    3     100[degrees, 115[degrees], 130[degrees]C
Cure time           3     16, 48, 96 hours
Table 2 - summary of experimental points

Run Backbone Resin type     Cure ratio      Cure     Cure time
no.                     [NH.sub.2]-,NCO)  temp.([degrees]C)  hours)
1    Ester    Low oligomer        .85         100         16
2    Ester    Low oligomer        .85         130         96
3    Ester    Low oligomer        1.05        100         48
4    Ester    Low oligomer        1.05        130         48
5    Ester    Conventional        .95         115         48
6    Ester    Conventional        .95         115         48
7    Ester    Conventional        .95         115         96
8    Ester    Conventional        .95         115         16
9    Ether    Low oligomer        .95         100         48
10   Ether    Low oligomer        .95         130         48
11   Ether    Low oligomer        .95         100         96
12   Ether    Low oligomer        .95         130         16
13   Ether    Conventional        .85         115         16
14   Ether    Conventional        .85         115         96
15   Ether    Conventional        1.05        115         48
16   Ether    Conventional        1.05        115         48
17   Ether    Conventional        1.05        130         96
18   Ether    Conventional        .85         loo         48
19   Ether    Low oligomer        1.05        115         16
20   Ester    Conventional        1.05        100         16

Static results

Tensile, tear, resilience and compression set were all performed according to ASTM procedures. The results were analyzed and compression set is reviewed here in order to illustrate the methodology (figure 7). The average value of compression set for all the data points, as measured by ASTM D395, is approximately 29[degrees]C. The effect of changing the backbone type from ester to ether is seen to improve the result slightly (about two percentage points), while changing from a low free monomer resin to a conventional prepolymer increased compression set by roughly the same amount. The well-known effect of cure ratio on permanent deformation is reproduced here and is clearly the most significant factor affecting compression set improvement.

The processing variables affect compression set and increased cure temperature has a significant effect on improving set, while cure time displays a more complicated relationship, with the property changing between 16, 48 and 96 hours. The conclusion is drawn that the optimum compression set will be obtained with a conventional ether resin cured at 85% stoichiometry and 130'C for a period of 96 hours. The major factors affecting compression set, however, are cure ratio and temperature.

Similar analyses were performed for each of the other static properties. The summary of the key factors, in order of importance, can be found in table 3. As might be expected, a single formulation does not produce the optimum response for every property. In general, however, the best overall result is obtained at a 95% cure ratio and a cure schedule of 48 hours at 130[degrees]C, which is a departure from the conditions commonly used with MBOCA cured elastomers. Among the resin structural factors, on average, the ester backbone appears to produce the best overall balance of properties. Low monomer, low oligomer prepolymers are not clearly superior to conventional prepolymers in the properties measured. It should be mentioned, however, that the gel time of conventional resins in this hardness range has been found to be much shorter than low oligomer resins (1.5 minutes vs. 3-5 minutes), and it may serve as an inducement for using the latter materials.

Table 3 - factors affecting the static properties
of MCDEA-TDI elastomers (93-95a)

Response             Dominant factors and optimum
                     values

Tensile properties   Backbone type (ester), cure temperature
                     (130[degrees]C), cure time (96 hrs)

Tear properties      Backbone type (ester), cure ratio
                     (105%), cure temperature (130[degrees]C),
                     cure time (48 hrs)

Rebound              Backbone type (ether), cure ratio
                     (95%), resin type (low oligomer)

Abrasion index       Cure ratio (105%)

Compression set      Cure ratio (85%), cure temperature
                     (130[degrees[C), cure time (96 hrs)

Dynamic results

The dynamic properties of the cured elastomers were analyzed in a fashion analogous to that used above. The test parameters of interest are the compressive stress at 10% deflection, the area under the loss compliance curve, and the melting point of the elastomer hard segment. The response diagram for compressive stress at 10% deflection is shown as an example (figure 8). The most significant factor affecting deflection is backbone type, with the ester backbone representing the best overall performance at an optimum cure ratio of 95[degrees]C. It will be remembered that the area under the loss compliance curve is being used as a predictor of the tendency of the polymer matrix to convert mechanical energy to heat. Surprisingly, it was found in this work that esters offer the best performance over the range -78 to +150[degrees]C and that esters and ethers are roughly equivalent in this perfoffnance aspect in the range 0 to +150[degrees]C. Unfortunately, a direct comparison with MBOCA cured elastomers was not possible and it is difficult to say if the result is a real effect of MCDEA or an artifact of the test. What does seem clear, however, is that elevated cure temperatures and extended cure times favorably impact this parameter.

Model tests

In order to check the validity of the prediction that higher temperature and extended cure time improve performance, elastomers were prepared using standard and optimized processing conditions, and the results were compared. The tear, compression set and abrasion changes observed when Airthane PST 90A prepolymer-mcdea is cured at 130[degrees]C for 48 hours are summarized in table 4. From these results, it is clear that significant improvements in tensile, compression set and abrasion properties are realized.

Table 4 - static property improvements of
Airline PST90A prepolymer MCDEA

Property                 100[degrees]C/   130[degrees]C/   [delta]
%
                            16 hrs.            48 hrs
Tensile strength (psi)       6,900             8,285          +20
Tensile tear (pli)             719               785           +9
Split tear (pli)               132               151          +14
Compression set (%)             33                23          -30
Rebound (%)                     44                43           -2
NBS index (%)                  162               219          +35

A similar comparison for other ether and ester based resins in the 93-95A range shown in table 5, where the percentage changes observed with a 130[degrees]C for 48 hour cure cycle are summarized. It should be pointed out that these static property improvements apply equally well to resins which contain higher oligomer levels, although these materials have higher viscosities and monomer levels which may complicate their use.

Table 5 - percentage change in static
properties of MCDEA cured resins using
optimized cure schedule (130[degrees]C/48 hrs)

Property                  Airthane     Cyanaprene    Durathane
                           PET91A         A9          STE90A
Tensile strength (psi)      +20          +16            +24
Tensile tear (pli)          +11          +11             +4
Split tear (pli)             -8          +12            +29
Compression set (%)         -38          -26            -24
Rebound (%)                  -3            0             -6
NBS index (%)               +10          +26            +22

The effect of optimized cure on the dynamic properties have been determined and the impact of elevated temperature and extended cure time on the properties of Airthane PET91A prepolymer are summarized in table 6. In this case, no change is observed in the compressive stress at 10% deflection or in the hard segment melt point. The area under the loss compliance curve decreases slightly, but this improvement may be at the lower level of statistical significance

[TABULAR DATA 6 OMITTED]

The effect of cure change on other systems is shown in table 7. In all cases, high temperature and extended cure time has a slight effect on the stress at 10% deflection, no effect on the hard segment melting point, and a sizable favorable effect on the areas under the loss compliance curves.

Table 7 - percentage change in dynamic properties
of MCDEA cured resins using optimized
cure schedule (130[degrees]C at 48 hrs)

Property                Airthane    Cyanaprene    Durathane
                         PST90A         A9          STE90A
Stress at 10%
deflection (psi)          -9%          -3%            -5%
Hard segment
m.p. ([degree]C           -2%          -2%            -1%
Loss compliance
area
  0 - 150[degrees]C      -31%         -47%           -29%
  78 - 150[degrees]C     -27%         -42%           -21%

The final question is the degree to which these optimized systems can compete with other high performance systems. The comparison of the loss compliance areas of Vulkonan 27 (Shore 93A), Ultracast PE-60 prepolymer (Shore 95A PPDI-PITMEG), and representative MCDEA systems is shown in table 8. Although the optinlized cure protocols used with the MCDEA crosslinked resins resulted in improved loss compliance areas relative to Airthane PET91A and PST90A prepolymers processed in the standard manner (tables 6 and 7), the loss compliance areas for these materials are greater than those of the NDI and PPDI systems. When compared to a MBOCA cured, low oligomer PTMEG resin system (Airthane PET95A) which is known to perform well in dynamic environments, however, the MCDEA candidates show clear improvement. This suggests that improvements relative to the current TDI-amine performance standard can be expected. Additional testing is required to determine the validity of this analysis technique and the magnitude of any performance benefits.

[TABULAR DATA 8 OMITTED]

A further comparison of these systems in the other performance areas is given in table 9. From these results, it is clear that tensile, tear, compression set and rebound properties equivalent to NDI and PPDI systems can be realized. Similar conclusions can be drawn with respect to the compressive stress at 10% deflection and hard segment melting point. It also is worth noting that the hard segment melt point of the MCDEA systems is significantly higher than that of the PPDI material, suggesting that the TDI-amine elastomers may be particularly suited for high temperature applications.

[TABULAR DATA 9 OMITTED]

Conclusions

Several points of interest have been demonstrated. First, the validity of the optimization technique employed has been shown. Predictions based upon these experiments led to demonstrable improvements in several key performance areas. The absolute value of the performance parameters selected were comparable to those of NDI and pPDI candidates in many areas, although significant differences in loss compliance behavior do exist. Whether these differences are important and/or compensated for by improvements in other properties must be determined in the field. Based upon these observations, however, it seems clear that TDI resins cured with MCDEA for extended periods at elevated temperature are potential candidates for dynamic and high temperature applications. The favorable processing and economic characteristics of these systems compared to other materials are additional reasons for interest.

Figure 1 - Vulkollan system

[ILLUSTRATION OMITTED]

Figure 2 - MBOCA and MCDEA structures

[ILLUSTRATION OMITTED]

Figure 3 - compression - deflection experiment

[ILLUSTRATION OMITTED]

Figure 4 - loss compliance area

[ILLUSTRATION OMITTED]

Figure 5 - DSC melting point

[ILLUSTRATION OMITTED]

Figure 6 - orthogonal array for three two-level variables

[ILLUSTRATION OMITTED]

Figure 7 - optimization of compression set

[ILLUSTRATION OMITTED]

Figure 8 - structural and processing effects on compression-deflection

[ILLUSTRATION OMITTED]

References

(1.) J.P. Casey, S.J. Dobkewitch, S.R. Gingras and P.F. Plank, "Airthane perfect prepolymers," presented at UTECH 88, the Hague, the Netherlands. (2.) S.M. Clift, A.L. Clement, J.R. Quay and J.E. Dewhurst, Airthane high performance polyurethane prepolymers," presented at the Polyurethane Manufacturers meeting, 1992. (3.) M.L. Williams, R.F. Landel and J.D. Ferry, J. Am. Chem. Soc. 77, 3701 (1955). (4.) P.J. Ross Taguchi Techniques for Quality Engineering, McGraw Hill, NY, 1988.