Abstract :
A computational method is described which allows experimental tie line data for two polymer¯one solvent mixtures to be analysed directly in terms of Flory¯Huggins theory. Three Flory¯Huggins parameters and their uncertainties are calculated. These are, one polymer molecular weight and two chivalues (one polymer segment¯solvent chi, i.e. chi12 or chi13, and chi23 the polymer¯polymer segment interaction coefficient). Due to irreducible parameter correlation, the remaining two Flory¯Huggins parameters, i.e. one polymer molecular weight and one polymer¯solvent chi, must be fixed at experimentally determined values. The paper also describes application of the method to experimental (literature) tie line data for some typical ternary systems based on uncharged polymers (synthetic polymers and biopolymers). While the description of phase diagrams was generally good, discrepancies between calculated and experimental polymer molecular weights and second virial coefficients, together with a significant dependence of the chivalues on molecular weight, warn that the Flory¯Huggins approach is unlikely to be an accurate quantitative predictor of phase behaviour for systems of the type discussed. The results were encouraging enough, however, to suggest that semi-quantitative conclusions about the influence of molecular factors on the positions of binodals, tie lines and critical points, and on the slopes of tie lines, were likely to be valid for mixed uncharged ternary systems.
