Quantum mechanical approaches to the prediction of phase equilibria: Solvation thermodynamics and group contribution methods.

摘要

A priori phase equilibria predictions necessary for the development, design, and optimization of chemical processes are one major challenge in chemical engineering. The aim of this dissertation is to develop new computational schemes that allow determination of thermodynamic properties and phase behavior from modern computational chemistry and/or advanced group contribution calculations.; The approach we take involves calculating property changes for molecular solvation, i.e., the transfer of a molecule from an ideal gas to a solution. The corresponding free energy change, referred to as the solvation free energy, can be used to determine the vapor pressure and activity coefficient, from which phase equilibrium predictions are made. A new solvation model is developed to utilize results from high level molecular orbital computations and minimize the error from simplification of the solvent as a continuum. This model is found to be more accurate for infinite dilution activity coefficients and partition coefficients than other methods, such as UNIFAC or modified UNIFAC.; Another issue addressed is the development of new group contribution techniques. Simple group contribution methods do not differentiate between isomers and can lead to severe errors for molecules containing multiple strong functional groups in close proximity. Quantum mechanical calculations reveal that the variation of electron density distribution of a functional group due to the presence of a nearby strongly electronegative or electropositive group changes the group-solvent interactions which lead to the deficiencies in present group contribution methods. A multipole correction method is developed to resolve the isomer and proximity problems. This method results in accurate predictions for octanol/water partition coefficients, Henry's law constants, and several pure fluid properties.; Finally an exact statistical mechanical model is developed to calculate the solvation free energy based on the induced screening charges at the molecular surface during ideal solvation. This model dissects molecules into surface segments, determines the segment activity coefficient from segment-segment interactions, and calculates molecular activity coefficients from summing contributions over all the segments. This approach uses significantly fewer parameters than other methods (2 versus 168 in UNIFAC and 612 in modified UNIFAC) for the prediction of vapor-liquid equilibrium and provides a priori prediction for new compounds.