Prediction of multicomponent mutual diffusion in liquids: Model discrimination using NMR data

Predictions of multicomponent diffusion coefficients in liquids still contain large uncertainties today. Progress is hampered by both a lack of experimental diffusion data and insufficient thermodynamical models. Currently, the thermodynamic factor relating Maxwell-Stefan to Fick diffusivities introduces substantial uncertainty in any diffusion analysis rendering many modeling efforts meaningless in practice. In this work, the influence of the thermodynamic factor is removed by studying the diffusion of one diluted component in a multicomponent mixture. Nuclear magnetic resonance (NMR) intra-diffusion coefficient measurements then directly comply with Fickian mutual diffusivities. A model-based experimental analysis framework is presented that allows the efficient and direct discrimination between suitable prediction models for diffusion coefficients. The general applicability of the method is demonstrated for the test system cyclohexane–n-hexane–toluene. It is shown that the discrimination based on the multicomponent consistency of intra-diffusion and mutual diffusion as developed here provides a stringent test for diffusion models independent of thermodynamic assumptions.