Understanding Thermal Diffusion in Liquid Mixtures and Macromolecular Solutions via Molecular Dynamics Simulations
Final Report Abstract
In a series of molecular dynamics simulations, thermal diffusion has been studied in several binary molecular mixtures as well as in polymer solutions. The results were quantified by the Soret coefficient ST or the thermal diffusion coefficient DT. The molecular mixtures were simulated with detailed force fields and they were directly compared to thermal diffusion forced Rayleigh scattering (TDFRS) measurements of Simone Wiegand’s group. For efficiency reasons, the polymer solutions were simulated with a generic bead-and-spring model with tunable solvent quality and polymer stiffness. Even with the simple, each of the simulations took months to converge due to the smallness of the Soret effect and the slow convergence of the steady-state concentration gradients. The main findings can be summarized as follows: Modern atomistic force fields are able to describe the Soret effect of molecular mixtures. They reproduce its sign (i.e. which component travels to the hot or the cold side) and they predict its magnitude typically to within 20-50%. In general, trends in the effects of concentration, temperature and branching patterns are correctly estimated. It is therefore now established that molecular dynamics can be used to predict thermal diffusion behavior of molecular mixtures in a useful semi-quantitative way. However, one must keep in mind that simulation times of dozens to hundreds of nanoseconds are required for converged results. The reverse non-equilibrium molecular dynamics method has been established as a robust algorithm for the calculation of Soret coefficients, in addition to thermal conductivities and shear viscosities, which could be calculated already prior to this project. Simple increment rules for estimating the Soret coefficient from properties of the two molecular components (molecular volume, cohesive energy, mass, moment of inertia) have to be used carefully, since they do not always apply. The more phenomenological heat-affinity model of Köhler et al., on the other hand, seems to fit also our data. In polymer solutions, the molecular-weight independence of DT has been confirmed. In addition, the crossover from small-molecule behavior (DT increases with polymer chain length) to polymer behavior (DT is constant) has been localized. In our simple model, it takes place at a chain length of 6-12 monomers. By comparing polymers of different chain stiffness it was found that the cross-over to a constant plateau DT occurred at a chain length of 2.5-3 times the polymer’s persistence length. This might indicate a way of mapping our model results to real polymers by scaling with the persistence length. A sign reversal upon change of the solvent quality, which is experimentally found for some polymersolvent combinations could not be reproduced, since at lower solvent quality demixing occurred before sign-reversal. The reason may be that our generic model is not sophisticated enough to capture this effect and that atomistic simulations are needed. However, they are at present computationally not feasible.
Publications
- “The Soret effect in dilute polymer solutions: Influence of chain length, chain stiffness and solvent quality”. J. Chem. Phys. 125, 124903 (2006)
M. Zhang, F. Müller-Plathe
(See online at https://doi.org/10.1063/1.2356469) - Dissertation (2007). Thermal Diffusion in Liquid Mixtures and Polymer Solutions by Molecular Dynamics Simulations. TU Darmstadt
Meimei Zhang
- “Thermal diffusion measurements and simulations of binary mixtures of spherical molecules”. J. Chem. Phys. 127, 014502 (2007)
P. Polyakov, M. Zhang, F. Müller-Plathe, and S. Wiegand
(See online at https://doi.org/10.1063/1.2746327) - “Reverse nonequilibrium molecular dynamics calculation of the Soret coefficient in liquid heptane/benzene mixtures”. J. Phys. Chem. B 112, 14999–15004 (2008)
P. Polyakov, F. Müller-Plathe, and S. Wiegand
(See online at https://doi.org/10.1021/jp805449j) - “Study of the Soret Effect in Hydrocarbon Chain/Aromatic Compound Mixtures”. J. Phys. Chem. B 113, 13308-13312 (2009)
P. Polyakov, E. Rossinsky and S. Wiegand
(See online at https://doi.org/10.1021/jp904667p) - Dissertation (2010). Molecular Simulation of Transport in Liquids and Polymers. TU Darmstadt
Eduard Rossinsky
