Local-composition models and prediction of binary liquid-liquid binodal curves from heats of mixing

Abstract

The temperature dependence of several local-composition models has been studied in conjunction with the Gibbs-Helmholtz identity. Binary heat-of-mixing data at temperatures near but above the critical solution temperature have been used to fit parameters obtained from excess-free-energy models in conjunction with the Gibbs-Helmholtz equation. These models, if the temperature dependence is adequate, should allow prediction of liquid-liquid equilibria (LLE) from the fitted parameters. The NRTL, UNIQUAC, and modified NRTL and UNIQUAC models (modified by inclusion of temperature-dependent parameters) seldom provide even qualitatively correct LLE predictions. A new local-composition model due to Wang and Chao (1983) yields reasonably good predictions for some systems but incorrect results for others. Reasons for these model inadequacies are discussed in terms of a local-composition model for the excess enthalpy which can be used to predict binodal curves accurately, including reasonably accurate values for the critical solution temperature, if reference excess-free-energy data at higher temperatures are available from VLE maesurements.