With experimental investigations and current theories, molecular modeling became an inevitable technique to study the perplexing phenomenon of glass transition. Among polymers, small variations in atomic interactions yield different values of the glass transition temperature, $T_g$. To reveal the influence of differences in the atomic functionality on the value of $T_g$, and thus to probe the molecular mechanisms responsible for this transition, atomistic simulations have to be undertaken. However, such simulations are argued not to accurately represent physically the glass transition due to the long relaxation times involved. Here we show the universality of the well-known Williams-Landel-Ferry equation for the experimental thermal dependence of polymer viscosities as demonstrated with atomistic simulations. Consequently, atomic aspects could be explicitly revealed. The contribution of atomistic simulation to the study of glass transition is thus confirmed. However, it has to be used complementarily with experiments and coarse-grained simulation to reveal the atomic aspects of current theories.

• Received 29 April 2006

DOI:https://doi.org/10.1103/PhysRevE.74.061803