Improving fatigue resistance with neodymium polybutadiene.

Today's car industry demands increased tire diversification and improved overall performance. This challenge is answered by enhanced tire design and new improved compounds. As a consequence the synthetic rubber industry is facing ever increasing requests to supply specialty polymers; either new polymers or improvements on already existing types, such as 1,4 cis polybutadienes.

Early studies by Enichem of diolefin polymerization using Ziegler-Natta catalyst systems based on rare earth elements (ref. 1 ) made it possible to synthesize a polybutadiene having both very high chain linearity and 1,4 cis content.

Cis contents as high as 98% were achieved using a neodymium based catalyst in an aliphatic solvent. The industrial production of this type of polybutadiene, known as Europrene Neocis, began in 1984.

High tacticity and chain linearity, i.e. high stereoregularity, are of paramount importance in 1,4 cis polybutadiene. This is because the ultimate properties and fatigue life of the vulcanized compounds are strongly related to the phenomenon of strain-induced crystallization.

The relationships between micro/macro structure of neodymium polybutadiene and ultimate properties and fatigue resistance of 100% polybutadiene compounds have been already reported (ref: 2) and the occurrence of a strain sensitive crystallization mechanism has been found (ref.3).

Further studies have been addressed towards the behavior of neodymium polybutadiene in industrial compounds.

This article deals with the effect of polybutadiene micro/macro structure on fatigue resistance and crack growth in NR/BR sidewall compounds.

Experimental

Materials

Samples of commercial high cis polybutadienes manufactured using Ziegler-Natta catalyst based on titanium and cobalt organometallic complexes have been compared to polybutadiene prepared using the neodymium based catalyst. These polymers will be referred to as Ti-BR, Co-BR and NdBR respectively.

Their physical-chemical properties are shown in table 1 . It should be noted that microstructure has been determined by a Perkin-Elmer model 1760 FTR spectrometer and calculated according to Morero method (ref.4).

[TABULAR DATA 1 OMITTED]

Melting temperature, Tm, has been assessed via DSC using a DSC30 model Mettler calorimeter. Molecular weights have been calculated from GPC chromatograms performed in THF at 23[degrees]C using a Millipore-Waters model 712 Wisp (Lichrogel columns array: PS 40000, PS 40000, PS 4000, PS 400, PS 40) and using the following Mark-Houwink constants:

K = 0.00()457 a = 0.693

The branching Index, G, evaluated according to the Drott-Mendelson method (ref. 5), is the ratio between the experimental intrinsic viscosity and the theoretical one derived from GPC chromatograms assuming all molecules to be linear. Therefore G values close to 1 mean linear polymers, whereas the lower the G value the higher the macromolecular branching.

Compounding and curing

The sidewall compound recipe evaluated in this work has been obtained using a statistical method based on design of experiments in order to analyze the effect of compound components on performances.

A three independent variable central design was adopted. Independent variables were carbon black, processing aids and polybutadiene contents, while dependent variables were rheological, mechanical and dynamic properties of compounds. Five levels for each independent variable were considered and 20 compounds were evaluated (six of which were repeated center point) instead of the theoretical 125 compounds considering all the combinations of variables.

Relationships between independent variables and compound properties (dependent variables) have been analyzed by regression equations. Adequacy of fit was assessed by "F-test" and "lack of fit" parameters.

Afterwards, the Harrington method (ref. 6) and a Derringer computer program (ref. 7) were used to tailor compound properties. Each dependent variable was given a desirability function di, defined using weight, range of validity and shape parameters. Various di were combined into a composite desirability D (weighted geometric mean), so that the tailor made formulation shown in table 2 was obtained. The following curing conditions were used:

* Curing temperature: 151[degrees]C;

* Curing times: - t95 for fatigue to failure test specimens and crack growth test specimens; - t9s x 2 for dynamic characterization specimens ([phi] = 17 mm, h = 25 mm cylinders).

Table 2 - sidewall compound optimized recipe
                                                          phr
Natural rubber                                            50
Polybutadiene                                             50
Carbon black HAF 375                                      45
Zinc oxide                                                 4
Octadecanoic acid                                          2
N-(1,3 dimethyl butyl) -N,- phenyl-p-phenylenediamine      3
Poly-2,2,4-trimethyl - 1,2 - dihydroquinoline            1.5
Wax                                                        2
Aromatic oil                                               4
Processing aids                                          4.2
TBBS                                                     0.8
Sulfur                                                     2

Testing

To assess fatigue resistance the Fatigue to Failure Monsanto Tester has been used. Unnotched specimens of all compounds have been tested both in air (23[degrees]C, applied strain amplitude = 1.36, frequency = 1 Hz) and in an ozone atmosphere (50 pphm, 50[degrees]C, applied strain amplitude = 0.45, frequency = 1 Hz).

Experimental data for each sidewall compound were collected on a large number of specimens in order to take account of the statistical nature of rubber failure. The data were processed according to the Weibull probability distribution function (table 3) which has already been proved capable to describe compound behavior within tight confidence limits (ref.2).

Table 3 - sidewall compounds: average fitting
          of Weibull distribution function
             F (x) = 1 - exp [- [(X/b).sup.a]]
                   a          b        [r.sup.2]
Nd - BR          1.98       60.1        0.97
Co - BR          2.09       53.2        0.98
Ti - BR          2.12       50.9        0.96

To analyze crack propagation a procedure has been defined in order to perform laboratory tests at strain, temperature and frequency levels which simulate the normal sidewall service conditions. These are: strain in the range 10-30%, temperature about 50[degrees]C (level for normal highway running) tire deformation frequency range 20-40 Hz.

Crack growth phenomenon was analyzed according to the fracture mechanics theories developed by Rivlin and Thomas (ref. 8). Their approach to the rubber fracture process is based on an energy balance suggested by Griffith (ref. 9) which gives the following criterion for crack propagation:

(1) - ([differentialU/[differentialA) [greater than or equal to] T where T (tearing energy) is the energy required per unit area of fracture surface, U is the stored energy in the specimen, A is the interfacial area of the crack provided that no external work is done on the system during crack propagation.

Using a pure shear specimen, that is a slab-like specimen having high width/thickness ratio (w/h [greater than or equal to] 10), equation (1) can be written as follows:

(2) T =W[multiplied by]h (2) where W is the strain energy density per unit unstrained volume.

The use of this approach made it possible to describe crack growth resistance in terms of crack growth rate vs. tearing energy plots which, being sample geometry independent, represent a characteristic property of the vulcanizate.

Tests were executed on a MTS 831 system equipped with special clamps to fix the pure shear specimen. A long focal length microscope with X,Y positions indicator was used to measure crack growth during the test. A Sony UP850 hard copy unit linked to the microscope was used to print the path of the growing crack.

The tests have been performed at 50[degrees]C, 20/40 Hz and strain levels in the range of l0-30% were considered.

Dynamic mechanical properties of the compounds were evaluated on cylindrical specimens ([phi] = 17 mm, h = 25 mm) performing strain sweep in compression. Typical tire service conditions were simulated (strain range 1-20%, temperature 80[degrees]C, frequency 110 Hz).

Results and discussion

The main results of our early work (ref.2) concerning fatigue resistance of pure polybutadiene compounds is highlighted in figure l which shows Nd-BR overcoming reference commercial polybutadiene in terms of Weibull mean life (63% of failed specimens). Nd-BR allows an improvement of l5% and 48% respectively in comparison with Co-BR and Ti-BR.

Crack growth experiments performed on the same compounds (ref. 10) showed a slowdown in crack growth rate moving from the lowest 1,4 cis content (Ti-BR) to the highest (Nd-BR). As previously reported (ref.2) this behavior has been associated with strain-induced crystallization.

The use of Nd-BR in sidewall compounds has been suggested to improve their performance.

As the presence of NR in sidewall compounds could obscure the differences mentioned above, the aim of the present work has been to confirm that the benefits conferred by Nd-BR to fatigue and crack growth resistance still persist in NR/high cis BR compounds.

The fatigue life of sidewall compounds is described in figure 2 in terms of the Weibull distribution function and in terms of the Weibull mean life.

It should be noted that owing to NR shadow effect, Ti-BR and Co-BR behave similarly, whereas Nd-BR still shows an advantage of 12% approximately in terms of mean life.

The higher fatigue resistance of sidewall compound containing Nd-BR has been confirmed performing the test in ozone atmosphere.

Crack growth tests on the sidewall compound confirmed the trend obtained with pure polybutadiene even if smaller differences are involved. Crack growth rate decreases with increasing 1,4 cis content of the polybutadiene.

Moreover, the slope of dC/dN vs. T lines is intermediate between those obtained for pure polybutadiene and pure NR compounds (figure 3). This result seems to confirm that crack growth rate depends on the polybutadiene being used in the tearing energy range examined.

According to the energy criterion for fracture propagation suggested by Griffith (ref.9) sufficient energy has to be supplied by the elastic energy stored in the elastomeric network to cause crack growth.

It follows that if a network is capable of dissipating energy when stressed, less elastic energy will be available for crack growth, therefore higher fracture energy will be obtained together with longer fatigue life. Energy dissipation can be achieved by viscoelastic relaxation mechanisms, therefore fracture resistance could be enhanced by increasing the viscous component of the material. This can be evaluated in terms of loss modulus ("E") and loss factor (tan o) by performing dynamic mechanical analysis. The tan o curves (figure 4) show Nd-BR to be less hysteretic than Co-BR and Ti-BR. Therefore, if energy dissipation by viscoelastic relaxation is the dominant mechanism, then Nd-BR should have the lowest fatigue life. Clearly this is not the case.

Energy dissipation can also be increased by structural breakdown of carbon black aggregates under stress and its magnitude can be assessed by the Payne effect (ref. l l), that is by the storage modulus (E,) decrease observed during a strain sweep.

The dynamic spectra of the sidewall compounds show similar dependence of E, on strain (figure 4), therefore any contribution of structural breakdown in increasing fracture energy should be the same.

It can therefore be concluded that the main reinforcing mechanism in delaying fracture both in pure polybutadiene compounds and in BR/NR blends is the strain-induced crystallization which occurs in the highly strained region around the crack's tip and does not affect bulk properties or the hysteresis of the elastomeric network.

As previously mentioned (refs.2 and 12) this mechanism is more efficient in elastomers having high steric purity (high cis content, low vinyl content) and high chain linearity.

These characteristics are typical of neodymium polybutadiene, which represents the latest development in the polybutadiene field and should be recommended to enhance fatigue and crack growth resistance in tire compounds.

Conclusions

Fatigue and crack growth resistance of high 1,4 cis polybutadienes and their blends are strongly dependent on the steric purity of the polybutadiene. Fracture resistance is improved by strain-induced crystallization which acts as a self reinforcing mechanism in delaying crack propagation.

In the case of tire applications where high fracture resistance is requested together with low hysteresis, polybutadienes having very high 1,4 cis content and high chain linearity should be used; neodymium polybutadiene fully meets these specifications so that it is a powerful tool to extend tire life.

References

[1.] L. Gargani, G. Lugli, E. Lauretti and F. Mistrali, IRC 82 proceedings, Paris 1982. [2.] L. Gargani and M. Bruzzone, "Fatigue resistance of polybutadienes and effect of microstructure," in Advances in Elastomers and Rubber Elasticity, J. Lal and J.E. Mark Ed., Plenum Publishing Co. 1986. [3.] M. Cesari, G. Perego, A. Zazzetta and L. Gargani; MakromoL Chem. 181, 1143 (1980). [4.] D. Morero, A. Santambrogio, L. Porri and F. Ciampelli, Chim. Ind. (Milan) 41, 758 (1959). [5.] E.E. Drott., R.A. Mendelson, J. Polym. Sci.-A2, 8, 1361 (1970). [6.] E.C. Harrington, J. Ind. Qual. Control, 21, 494, (1965). [7.] G.C. Derringer, Rubber World, ]68, 49 (1973). [8.] R.S. Rivlin, A.G. Thomas, J. Polym. Sci 10, 291 (1953). [9.] A. Griffith, Philos. Trans. R. Soc. London, A221, 163 (192]). [10.] B. Miani, M. Da Via, unpublished work. [11.] A.R. Payne, Rubb. Plast. Age, Aug. 1961, p. 963. [12.] M. Bruzzone, Makromol. Chem. Suppl. 4, 177(1981).