Assessment of liquid‐liquid equilibrium data by solving the Gibbs‐Duhem equation

A methodology, based on the Gibbs-Duhem equation, is developed to study the thermodynamic consistency of liquid-liquid equilibrium (LLE) data. The method proposes evaluating the experimental data in two ways: (a) integral-form, verifying dependences among variables in the same phase, both for temperatures: Tj=kappa T epsilon T,j0-Tj,exp-Tj,cal, and compositions: x1,jJ=kappa x epsilon x1J,j0-x1,j,expJ-x1,j,calJ, and (b) differential-form, which validates relationships between the compositions of each phase by zeta j=kappa zeta epsilon zeta,j0-zeta j,exp-zeta j,cal. A coverage factor kappa is introduced in both approaches to define the degree of confidence of the evaluation using a database (T, x1I,x1II) of 50 binary systems to define the limits of inconsistency. With kappa = 3, more than 60% of the set is accepted (kappa > 3 falls to 38%) providing a guarantee of quality for LLE data. The results show that the consistency test can detect errors in the set {data + model} under study, although the absence of a reference state for the LLE gives the model a limited sensitivity to systematic errors.