ABSTRACT
Atomistic modeling techniques have been employed for the investigation of mechanical properties of amorphous polymeric solids for years [1-5]. Theodorou and Suter [1]
successfully calculated the elastic constants of glassy atactic polypropylene (aPP) by molecular mechanics techniques, surmising that the entropic contributions to the elastic response are negligible for such glasses. Mott et al. [2] strained polypropylene glasses beyond the macroscopic elastic range and found evidence for the occurrence of diffuse plastic unit events and estimated, by comparison with experimental results, that the size
of the plastically transforming domains in elementary shear-transformation events extends over roughly 100Å (well beyond the system size that can conveniently be
simulated). Plastic relaxation occurs through cooperative rearrangements with rather small local transformation strains . These findings were assumed to be true for all polymeric glasses, in contrast to the long-held belief that the elementary process of
plastic transformation is a well-localized conformational transition. Hutnik et al. [3] used the same technique on glassy bisphenol-A-polycarbonate (BPA-PC) and confirmed the
results on aPP. All these calculations suffered from small system size and the fact that the atomistic box shape was used as a simulation variable, prescribed and controlled during the athermal simulations. While this is perfectly appropriate for the elucidation of elastic
properties, it is not optimal when the “natural” system path is unknown, as in plastic deformation. It would be preferable to be able to simulate the behavior of atomistic
subsystems embedded in a medium (of identical properties) and free to follow the driving forces that the gradient of (free) energy suggests. The overall system would have to be considerably larger, however, than the ones employed for these atomistic simulations.
