Abstract
This work covers a number of aspects of the thermal properties of empirical and molecular modelling with one chapter of literature survey and four chapters of result and discussion analysis. Chapter One presents a survey of the literature associated with the development of high performance polymers, of which polybenzoxazines are an example. Chapter Two introduces the materials which were supplied by Huntsman Corporation, Basel (i.e. Araldite MT 35600 and Araldite MT 35700). The purity of the samples was analysed in order to get an initial impression of the materials as any impurity in the sample may affect the mechanism of reaction and could influence the thermal properties. Chapter Three explains the thermal characteristics of a telechelic PEK benzoxazine by utilising a molecular modelling approach. The backbone of PEK consisting of five monomers was replicated to have twelve chains in a one cubic cell of Periodic Boundary Conditions. The PEK was then crosslinked and the percentage of crosslinking was subsequently compared with several others of different percentages of crosslinking. Eight atomistic models were generated; 0-0, 58-14, 58-21, 67-31, 83-35, 92-40, 100-45, and 100-54. Chapter Four reports the results of incorporating bisphenol A into Araldite MT 35600 at four different loadings. Empirical data for these blends (i.e. containing 0 wt%, 5 wt%, 15 wt%, and 25 wt%) were generated in-house, whilst a blend containing 35 wt% bisphenol A was incorporated into the molecular modelling study. Spectroscopic and thermal analyses were carried out to obtain the structural features and thermal properties of the materials. A molecular modelling study to predict the thermal properties of the blends was undertaken and the results were compared with the empirical data and corresponded closely. The final Chapter (Five) discusses a new application of the QSPR approach, which relates selected properties to chemical structures. Three case studies were investigated with the best linear equations produced, where the coefficients of determination were close to 1. This approach is still new in the field and therefore refinement work is needed in order to optimise the model.