Computer applications to the study of the thermodynamics of coal liquid interactions and to the study of the protonation of weak organic bases

Viscosity phenomena are surveyed for both pure nonideal liquids and for liquid mixtures. It is concluded that previous attempts to characterize the viscosity building interactions in various coal-derived liquid fractions have suffered from the inexact methods used to separate the coal liquid into fractions. Chromatographic methods for fractionating coal liquids are to be preferred over the more conventional solvent extraction techniques.

The Bolles--Drago calorimetric method for determining the free energy and enthalpy of an equilibrium reaction is shown to be inapplicable to mixtures such as coal liquids.

Methods of data analysis in acidity function studies of weak organic bases are surveyed, and it is pointed out that characteristic vector analysis is inappropriate for eliminating medium effects from families of uv spectra. The isobestic points which are recovered by reconstituting the absorption curves using only the first characteristic vector are an artifact and have no chemical significance.

Statistical problems with the conventional treatment of acidity function data are pointed out and a general nonlinear regression model is developed which takes into account medium effects. A computer program for carrying out the regression analysis is available on magnetic tape from the author for use on IBM S/370 compatible computers.

Application of the regression model to nmr data for the protonation of ketones indicates that, contrary to previous reports, ketones may be found to follow the same acidity function if proper account is taken for medium effects on the chemical shifts. The present data is inadequate for a complete statistical analysis. The need for at least 30 independent, distinct data points which are equally spaced along the pH--H_O scale and which clearly define the arms of the sigmoidal titration curve is pointed out, as well as the need for maintaining an adequate number of significant figures and for several replicate data points placed in each arm and near the point of inflection of the sigmoidal titration curve.