A Critical Review of Thermodiffusion Models: Role and Significance of the Heat of Transport and the Activation Energy of Viscous Flow

In this paper thermodiffusion models developed to estimate the thermal diffusion factor in nonideal liquid mixtures are reviewed; the merits and shortcomings of each model are discussed in detail. Most of these models are multicomponent in principle; however our focus here is on binary mixtures. Two rather different groups of models are identified: models needing a matching parameter to be obtained usually from the outside of thermodynamics, and the self-contained or independent models. Derivation of the matching parameter models using linear non-equilibrium thermodynamics and the details of how to find the matching parameters are investigated. The physical meaning of parameters such as the net heat of transport and the activation energy of viscous flow is elucidated, as the literature is overwhelmed with confusing and misleading information. The so-called dynamic and static models and their relations to the matching and non-matching parameter models are also discussed. We conclude that modeling the net heat of transport by the activation energy of self-diffusion may provide better results than approximating it by the activation energy of viscous flow. Nonetheless, the matching parameter models, which use the activation energy of viscous flow, are more dynamic and predict the thermal diffusion factor better than the non-matching parameter or static models, such as those of Kempers and Haase.