In mixtures exposed to temperature gradients, transport of heat and mass is coupled. A precise description of transport in these systems therefore relies on knowledge about the coupling coefficient, the Soret coefficient. Several thermodynamic models have been proposed with the intent of predicting the Soret coefficient in a mixture, requiring only an equation of state, but these have varying degrees of reliability. Reliable prediction of the Soret coefficient can be utilised to achieve optimal design of electrodes, membranes, protective coatings and alloys exposed to large temperature gradients.

A new model is proposed, and its predictions are investigated by comparison with measurements both from experiments and molecular simulations. Additionally, a modification of the model proposed by L.J.T.M. Kempers in 2001 is investigated.