Use this URL to cite or link to this record in EThOS:
Title: Development of coarse-grained force fields from a molecular based equation of state for thermodynamic and structural properties of complex fluids
Author: Lobanova, Olga
ISNI:       0000 0004 5354 751X
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Access from Institution:
In spite of the vast array of modelling techniques and force fields available, the study of the phase behaviour, structure, microstructure, and dynamics of mixtures remains a challenging task. A systematic coarse-graining (CG) methodology is employed in this thesis involving the parameterisation of force fields using a top-down approach, by effectively describing a large number of target macroscopic thermodynamic states with a rigorous molecular-based equation of state. A recent incarnation of the Statistical Associating Fluid Theory (SAFT-gamma) is used. The underlying force field is based on the Mie intermolecular potential, which is a generalised form of the Lennard-Jones potential with a variable and versatile form of the repulsive and attractive interactions. The coarse-grained force fields developed in this manner are used directly in Molecular Dynamics simulations in order to explore the dynamical, structural, and interfacial properties, which can not be directly accessed by the equation of state (unless a suitable treatment of the inhomogenous properties of the system is made). The goal of any coarse-graining procedure is to derive simple, but accurate, robust, and transferable force fields. By aiming for the simplicity, the coarse-grained models developed in our work are typically based on the three-to-one mapping, i.e., one bead containing approximately three heavy atoms, or one-to-one mapping for the small spherical molecules, with the polar, directional, and long-ranged interactions between the beads treated implicitly using the effective spherically-symmetric Mie potentials. The SAFT-gamma Mie coarse-graining methodology is exemplified for a number of fluid systems of different complexities, including pure component systems, such as: the homologous series of n-alkanes, n-perfluoroalkanes, semifluorinated alkanes, ethers and water; binary and ternary mixtures, comprising the carbon dioxide, n-alkanes, and water; and finally the aqueous mixtures of alkyl polyoxyethylene glycol non-ionic surfactants. An accurate representation of the vapour-liquid properties with both, the equation of state and molecular simulation, is obtained for the molecules of different size and chemical nature. Describing the properties of water is, however, a much more difficult task. The CG model suffers from issues associated with the transferability and representability of the various properties for different thermodynamic conditions, as a consequence of the aggressive averaging of the strongly directional and polar forces into an effective spherically symmetrical potential. It has been shown that an isotropic single-site CG model based on a spherically symmetrical potential cannot capture all of the thermodynamic properties of water simultaneously (the issue of representability). Two different CG models of water are proposed: the first is designed to accurately reproduce the saturation liquid density and vapour pressure, and the second to capture the saturation liquid density and surface tension with high precision. Both models benefit from an accurate parameterisation of temperature dependence following the target properties over the entire temperature range of the fluid. An additional model is developed based on the two-to-one mapping, enabling more efficient large scale simulations in, for example, biomolecular systems. The models of the binary mixtures are developed by using the corresponding pure component models with an additional adjustable parameter to account for the unlike interactions; the latter are obtained by considering appropriate properties of the mixtures such as the fluid-phase equilibria or the thermodynamic properties of mixing. The unlike interactions are shown to be transferable for a quantitative description of the phase behaviour over a wide range of conditions and for the systems of related components. We are able to obtain an accurate prediction of the azeotropic point, critical loci, tree phase line, global density, and the shape of phase envelopes for studied mixtures. The quality of predictions is found comparable to the results from the atomistic models and other equations of state. The aqueous mixtures of alkyl polyoxyethylene glycol non-ionic surfactants are a key final goal of the research presented in this thesis. The CG models of the surfactants are developed within the SAFT-gamma group-contribution framework, where each functional group is derived from an accurate representation of the corresponding chemical moiety. By capturing a delicate interplay of the repulsive and attractive intermolecular interactions and obtaining the right balance between energetic and entropic effects, the various phase morphologies at ambient conditions can be reproduced in agreement with the experimental findings over the entire concentration range. The force fields developed in the current work allow for a prediction of key structural and interfacial properties. The Molecular Dynamics simulations reveal the spontaneous formation of micelles at low surfactant concentrations and a self-assembly into a bilayer at high surfactant concentrations. The aggregation numbers, the critical micelle concentration, area per molecule, the surface excess properties, and bilayer thickness are found in very good agreement with experimental data. This is very encouraging considering that only macroscopic thermophysical properties are used to develop the underlying force fields that describe the fine interactions between the molecules in the system. Despite the simplicity, coarse-grained force fields are shown to be robust and transferable; they can be applied to predict the properties which were not used in the original parameterisation procedure, with an accuracy comparable to the more sophisticated and computationally demanding models.
Supervisor: Jackson, George; Muller, Erich Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral