Ionic liquids (ILs) are salts with low melting points, often resulting in a liquid state at room temperature. ILs are regarded as highly tailorable designer solvents with many potential applications, such as in organic synthesis, energy storage devices and as solvents for biological molecules. However, for many applications the use of ILs is limited due to their cost and viscosity. One approach used to address this is combining ILs with molecular solvents. However, currently there is insufficient knowledge about the interaction of these IL-molecular solvents with solutes and interfaces, particularly regarding which ions or molecular solvent molecules are involved in solvating various solutes. This thesis expands our understanding and provides insight into the tools available to investigate solvent behaviour of ionic liquids at a molecular level using both experiments and computational simulations.

Machine learning was initially used to understand the trends between chemical structure and physicochemical properties of protic ionic liquids (PILs) in the presence of water. Due to the vast sample space in the field of PILs, it is near impossible to characterise these mixtures experimentally. Machine learning, which allows for the input of experimental data from which extrapolations can be made about new ILs, is a novel technique which has sparked great interest within our field. Machine learning models were created using linear regression and neural network methods using literature experimental viscosity and conductivity data to predict 8605 viscosity values and 8580 conductivity values. The viscosity and ionic conductivity of 10 new PILs of these predicted values were verified experimentally as part of this thesis, which demonstrates that high quality machine learning models can be crafted to complement experimental studies in the future. The machine learning study also demonstrated that physical properties of PILs are subject to drastic changes with minute changes to their chemical structure. This motivated the need to develop a deeper understanding of the role of a PIL in mixtures.

Ethylammonium nitrate (EAN) and ethanolammonium nitrate (EtAN), two PILs which are quite structurally similar with a small change in their chemical structures of replacing a hydrogen with a hydroxyl group, were chosen as the PILs for further investigation. While PILs are widely studied in the literature due to the ease with which they can be synthesised and favourable physical properties such as low viscosity, questions remain regarding how their solvent properties alter in the presence of water. The self-assembly of surfactants Cetyltrimethylammonium bromide (cationic), sodium octyl sulfate (anionic) and Tetraethylene glycol monododecyl ether (non-ionic) in PIL-water mixtures were probed to understand the solvation properties of PILs in the presence of water. The methods used to investigate these properties included surface tensiometry and small angle x-ray scattering (SAXS), both of which were used to understand ternary mixtures of PILs with surfactants and water. Surface tensiometry was able to show that the critical micelle concentration was greatly affected by the concentration of the PILs. The presence of PIL in the mixture led to an initial decrease in the CMC but led to an overall increase in the CMC across all surfactants above 5 mol% of the PIL. To confirm the presence of self-assembled structures in the ternary mixtures, SAXS was used. The SAXS experiments proved to be difficult for EAN due to contrast issues but scattering from micelles were observed in EtAN solutions. No scattering was observed for the EAN rich solvent, whereas for similar concentrations of EtAN x-ray scattering could be observed. To solve the conundrum regarding why such similar PILs led to vastly different results, it was decided computational techniques are necessary.

Molecular dynamics (MD) was explored as a complementary computational technique to probe deeper into the experimental data. As a starting point, a systematic review of 31 existing water models was performed to understand which water model force-field would be suitable for mixing with existing IL force-fields.  OPC3 water model was deemed to be suitable for the purposes of this thesis to combine with the existing OPLS EAN force-field. These force-fields were combined with existing force-fields for the three surfactants to probe the molecular level interactions between the EAN-water mixtures and the surfactants which self-assemble into micelles. The simulations suggested that ethylammonium ions, which are supposed to be in the bulk solvent, were in fact participating in the micelle formation with the surfactants. This provided an explanation regarding the contrast issues which led to inconclusive results from SAXS experiments.

The overall objective of this thesis was to gain a fundamental understanding of how PILs behave in mixtures with other solvents and solutes. To achieve this, a wide variety of experimental and computational techniques had to be explored to understand the mixtures from different perspectives, where that be experimentally or at a molecular level using simulations. The work done during this thesis will form a basis for future work in the space of molecular dynamics and machine learning models for PILs and their mixtures.