Phase Separation in Aerospace Polymer Blends

Polymer composites are becoming increasingly attractive to the aerospace industry as a lighter alternative to metal. They allow better fuel efficiency, so reduce environmental impact and operating costs. To create strong and fracture-resistant composites, blends of branched epoxy resin thermosets and linear thermoplastics are used. As the mechanical properties of these materials are largely determined by their morphologies, it is important to understand the phase separation that takes place within the blends. Unfortunately, the usual techniques for studying phase separation, such as the Flory-Huggins model, are not particularly applicable to branched polymers, so predictions of the behaviour of such blends are limited. Even the lattice cluster theory can only be used for polymers with simple, regular architectures, rather than the randomly branched thermosets relevant to the aerospace industry. In this work, a computational approach was developed to calculate the entropy and free energy of branched polymers with arbitrary shapes and sizes. Although these calculations are currently only valid for polymers in infinitely dilute solution, they provide systematic corrections to the Flory-Huggins predictions. Concentration fluctuations, which are the precursor for phase separation, have also been studied directly using Monte Carlo simulations based on the bond fluctuation model. Properties such as total interaction energies, radii of gyration, radial distribution functions and structure factors have been determined for blends containing molecules with various amounts of branching, and various combinations of interaction energies. Finally, a model experimental system based on an industrially-relevant blend has been designed and characterised. This allowed measurements of concentration fluctuations to be carried out using small-angle neutron scattering, and for the competing influences of temperature and cure extent on miscibility behaviour to be studied separately. In the future, it is hoped that the results from these experiments could be compared with structure factors produced using the simulation method mentioned above.