New methods for the computer simulation of macromolecular liquid crystals

Abstract

Molecular simulation of macromolecular liquid crystal (LC) systems has so far been limited by a number of factors: the large size of the molecules themselves and the fact that mesophase formation takes place on length and timescales that are not reasonable to simulate. The work in this thesis develops three methods that can be used to assist in the computer simulation of macromolecular LC systems. Coarse-graining is a technique where instead of representing every atom within a molecule as a single site, a number of atoms are grouped into interaction centres. This coarse-graining procedure has been applied to a liquid crystal dendrimer to enable the bulk phase simulation of the molecule to be studied. The analysis of the results shows that the behaviour for the coarse-grained model closely matches that of a more detailed atomistic model. Phase behaviour in the bulk matches results from X-ray data. The parallel-tempering method (replica exchange method) uses a series of replicas of the same system at different temperatures to improve the sampling of phase space. This technique was applied to two different systems, a bulk phase simulation of an alkane chain and the gas phase simulation of a silsesquioxane liquid crystal dendrimer. The method was then extended to work with a set of replicas which used different potentials. The Tsallis potential was used to soften potentials and allow replicas to sample a greater area of phase space. A third simulation method was applied which used soft-core potentials. This attempted to address the problem of the long timescales needed to see the formation of mesophases in macromolecular systems. Three different anisotropic single site soft-core models were developed and tested for liquid crystals. The results show that the time needed for mesophase formation is reduced for soft-core models and that these models are able to form multiple liquid crystal phases. In addition, the most promising of these soft-core models has been applied to the simulation of more complex liquid crystal systems, represented by multi-site models.