An Improvement to COSMO-SAC for Predicting Thermodynamic Properties

A modified COSMO-SAC model is presented
to calculate thermodynamic
properties of pure fluids and mixtures using statistical thermodynamics
and the surface charge density of each compound obtained from a quantum
mechanics (QM) calculation. The main differences from the previous
models are that the new model includes a dispersion contribution in
the mixture interaction, and is reparametrized using measured pure
and mixture thermodynamic data simultaneously. With a single set of
universal parameters, the new model provides higher accuracy than
our previous models for predicting mixture thermodynamic properties
while maintaining the same accuracy for pure compound thermodynamic
properties. The overall root-mean-square deviation (RMSD) in the logarithms
of partition coefficients for 992 octanol–water partitioning
systems and 829 other solvent–water partitioning systems with
this new model is reduced by about 10% compared to the results from
previous models. Also, the agreement between the predicted and measured
partition coefficients over a wide range of values is improved as
a result of better activity coefficient predictions at high dilution
by inclusion of the dispersive mixture interaction in the model. The
accuracy in the vapor–liquid equilibrium (VLE) predictions
is comparable to, or better than, the previous model that was developed
for phase equilibria calculations only. The new model also provides
parameters for use with the Amsterdam Density Functional (ADF) in
addition to DMol3.