STRUCTURE-PROPERTY CORRELATIONS OF ATOMISTIC AND COARSE-GRAINED MODELS OF POLYMER MELTS

Polymer reference interaction site model (PRISM) integral equation theory of polymer melts is applied to investigate the role of local chemical structure on interchain packing and thermodynamic properties such as the isothermal compressibility, cohesive energy, and solubility parameters. Chemically realistic rotational isomeric state model (RIS) level calculations are carried out for polyethylene and isotactic and syndiotactic polypropylene. A strong correlation between the thermodynamic properties and the local depletion regime of interchain packing is found. A series of coarse-grained models of decreasing local realism are then investigated, and various schemes are constructed to estimate the coarse-grained model parameters in terms of polymer structure. For the discrete semiflexible chain and the Gaussian thread models the effects of backbone characteristic ratio and chain branching are incorporated by employing an effective stiffness, or aspect ratio. A strategy for carrying out the mapping from the atomistic to coarse-grained level. based on single-chain conformational and interchain packing considerations is studied. For polyethylene and polypropylene, by enforcing an equality of the zero-angle scattering of the atomistic and coarse-grained models, good agreement is found between the local radial distribution functions and reduced solubility parameters predicted by the various models. Numerical predictions of the semiflexible chain model and analytical results for the Gaussian thread liquid are obtained and compared against RIS-level calculations and experimental estimates of the melt solubility parameters. At least within a homologous series of materials, it appears that appropriately calibrated coarse-grained descriptions can reproduce various structural and thermodynamic properties. This suggests that polymer alloy miscibility can be studied within a homologous series of materials by computationally convenient, coarse-grained models.