ABSTRACT GAWRYS, KEITH LOUIS. How Asphaltenes Aggregate: Role of Chemistry and Solvent.

GAWRYS, KEITH LOUIS. How Asphaltenes Aggregate: Role of Chemistry and Solvent. (Under the direction of Peter K. Kilpatrick) Asphaltenes from three crude oils were separated on a preparatory-scale into 20 30 discrete subfractions by sequential precipitation from mixtures of n-heptane and toluene. The earliest fractions precipitated were characterized by lower than average aromaticity and atomic N/C ratios and contained significant amounts of co-precipitated inorganic solids or metal atoms bound to functional groups intrinsic to the asphaltenes. The subfractions isolated in the regime where asphaltene solubility changed significantly with slight changes in the solvent were characterized by atomic H/C ratios that varied significantly with the yield of precipitated asphaltenes. Atomic N/C ratios decreased and O/C ratios increased systematically with increasing precipitated asphaltene yield in this solvent regime. The higher yield fractions were generally the most aromatic and formed the largest aggregates in solution, suggesting that asphaltene solubility and aggregation behavior is dominated by πbonding interactions between the aromatic moieties. After precipitation of ~ 80 % (w/w) of the asphaltenes, the remaining fractions were more “resin-like” in terms of chemical composition and aggregation behavior. SANS has proven useful for deducing the sizes and morphologies of asphaltenic aggregates in solution. Selection of an appropriate form factor model that closely approximates the structure of asphaltenic aggregates is important for determining the radius of gyration, molar mass, apparent fractal dimension, and second virial coefficients (A2) of the aggregates. SANS measurements were performed on solutions of asphaltenes dispersed in perdeuterated solvents with a broad range of solute and solvent chemical compositions. Comparison of various geometric form factor models to the asphaltene scattering spectra suggested that the shape of asphaltenic aggregates is best described by an oblate cylinder model with radius polydispersity. The polydisperse cylinder model provided typical values of the particle thicknesses from 5 to 32 Å, the average particle radius from 25 to 125 Å, and ~ 30 % radius polydispersity. Subsequent calculation of average aggregate molar masses suggested a range of solvent entrainment from 30 to 50 % (v/v) within the aggregates that was consistent with previous viscosity measurements. Changes in the apparent aggregate molar mass with concentration indicated deviations from ideal solution behavior (i.e., non-interacting aggregates plus solvent), which were quantified through the determination of aggregate A2 values. Aggregate A2 values varied significantly with solvent conditions, concentration, and chemical composition (e.g., atomic H/C ratio) of the solute. These results suggested that interactions of asphaltenes, resins, and solvent are strongly dominated by dispersion and π-bonding interactions. The magnitude of the experimentally measured A2 values generally under-predicted or agreed with the values calculated from an excluded volume model, suggesting that energetic interactions of the solute and entrained solvent are significant. A UV-vis spectroscopic technique was developed to determine the solubility of polynuclear aromatic compounds in binary solvent mixtures with the intention of extending the analyses to more complex multi-component mixtures, such as petroleum asphaltenes. Binary solvent mixtures were selected to probe specific intermolecular interactions between the solvent and solute (i.e., dispersion, polar, and hydrogen bond interactions). The solubility data were fit to a three-dimensional solubility parameter model based on regular solution theory. The model provided accurate solubility predictions for two solutes in several binary solvent mixtures with less than 30 % error, but the predictive capability of the model decreased significantly when the polar and/or hydrogen-bonding contributions to the solvent solubility parameter deviated significantly from that of the solute. Additional experiments measured the solubility behavior of a multi-component solute mixture of ACA and PHD in 2butanol. The results suggested the need to modify the existing model to account for