Miscibility studies of engineering thermoplastic polymer blends

Engineering thermoplastic polymer blends constitute a large part of polymer consumption. To optimize the design of such blends, a great deal of attention has been focused on predicting and characterizing the miscibility of these blends. In this thesis, both experimental and molecular dynamics (MD) simulation methods have been used to address miscibility of three kinds of engineering thermoplastic polymer blends. In particular, the heats of mixing and Flory-Huggins interaction parameters of poly(etherimide) (PEI)/polycarbonate (PC), PEI/Poly(butylene terephthalate) (PBT) and PC/PBT blends at different compositions were calculated from MD simulation. For the PEI/PC blend, both differential scanning calorimetry (DSC) and MD simulation showed that they are immiscible. However, the degree of compatibility of a 80:20 PEI/PC blend is higher than that of a PC rich blend. This anomalous thermodynamic behavior was found to be due mainly to intermolecular interactions. Transmission electron microscopy (TEM) results confirmed this observation.; The inherent miscibility of PEI/PBT blends was also successfully predicted from MD simulation. At concentrations used in the simulations, the heats of mixing were found to be negative. The miscibility between PEI and PBT is attributed to favorable van der Waals interaction. The morphology of as-quenched PEI/PBT blend samples and annealed samples was also investigated. MD simulation was also applied to PC/PBT blends to study their inherent miscibility in the melt state, and the results indicate that PC/PBT blends are completely immiscible in the absence of transesterification reactions. The quantitative disagreement between the MD simulated and experimentally determined interaction parameters was also addressed.; A new approach to predict glass transition temperatures (T g) of polymers was developed. In particular, the intensities of peaks of the radial distribution function (RDF) at different temperatures obtained from MD simulation were employed to predict the Tg. Compared with traditionally used specific volume-temperature method, the accuracy of the prediction was significantly improved. This is because within the capacity of current computational power, it is impossible for the bulk specific volume to reach equilibrium below Tg; however, the local specific volume at a shorter length scale related to the intensity of peaks of the RDF can reach equilibrium.