Viscoelasticity of randomly crosslinked EPDM networks

The network structure of plasticized EPDM compounds, crosslinked with resol at different concentrations, was studied by means of rheological methods consisting in oscillatory shear tests, to determine the equilibrium modulus Ge, and long-time relaxation tests in compression followed by strain recovery (a protocol that also yielded values of the compression set of the samples). Ge results were analyzed with respect to the phenomenological model of Langley and Graessley which takes into account the contribution of crosslinks and trapped entanglements to the shear equilibrium modulus. A correction was introduced in order to take into account the presence of plasticizer. The measurement of the soluble polymer fraction in the different samples allowed a more detailed characterization of the networks to be carried out, following a molecular approach by Pearson and Graessley. This method enabled to calculate the crosslink density and trapping factor, but also to compute the probability ψ1 for an un-crosslinked polymer unit to belong to a dangling chain. This probability was shown to increase as resol concentration, and then crosslink density, decreased. The empirical Chasset-Thirion equation was used to model the long-time relaxation data for each sample. Chasset-Thirion parameters were interpreted by Curro and Pincus within a theoretical framework based on the idea that the longest relaxation times are associated with the pendent chains of the network. The relaxation times, obtained from the fitting of experimental relaxation moduli, dramatically increased as the crosslink density decreased. This result corroborates the evolution of ψ1: both tend to demonstrate that in the present compounds, the decrease of crosslink density is accompanied by an increase of the number and length of the dangling chains, leading to increasing relaxation times. The large soluble fraction and long pendent chains of samples showing the lowest crosslink densities were responsible for their poor elastic recovery. The relaxation data were used to model the elastic recovery of the compounds and predict their compression set profiles. Very satisfactory agreement was obtained between experimental data and computations.