Molecular simulation of the effect of temperature and architecture on polyethylene barrier properties.

We present a multiscale approach for calculating the low-concentration solubility, diffusivity, and selectivity of small molecules through polymer matrixes. The proposed modeling scheme consists of two main stages; first, thoroughly equilibrated and representative poly(ethylene) (PE) atomistic melt configurations were obtained through the application of a Monte Carlo (MC) scheme based on advanced chain-connectivity altering moves (linear architectures) or the combination of localized MC moves followed by molecular dynamics. In the second phase, transition-state theory (TST), as proposed by Gusev and Suter [Gusev, A. A.; Suter, U. W. J. Chem. Phys. 1993, 99, 2228], was invoked in a coarser level of description to calculate the barrier properties of the studied macromolecules to small gas molecules at infinite dilution. The multiscale methodology was successfully applied on PE melts characterized by various molecular weights (MW) (from C78 up to C1000) and polydispersity indices at a wide range of temperature conditions. The effect of molecular architecture on the barrier properties was examined through the comparison between linear and short-chain branched structures bearing the same total number of carbon atoms. Simulation results were found to be in very good agreement with available experimental data. Additionally, the new scheme has been further validated by comparing the qualitative behavior of solubility, diffusivity, and selectivity with previously reported trends in the literature based on both experimental and simulation studies. The present study concludes that density plays a dominant role that determines the behavior of the polymer as a barrier material, especially in terms of diffusivity. Additionally, it is evidenced that short-chain branching has a small effect on the barrier properties of PE when the comparison is performed on purely amorphous samples. The hierarchical method presented here not only is faster when compared against conventional molecular dynamics simulations, but in some cases, like the vicinity of the glass transition temperature or for long polymer chain melts, it opens the way to the calculation of the barrier properties at realistic simulation times.