(294j) Insights into the Mass Transfer through Vapor-Liquid Interfaces from Molecular Dynamics Simulations

In fluid separation processes, both thermodynamic bulk and interfacial properties are important. In distillation and absorption processes, the mass transfer through vapor-liquid interfaces plays a central role. In classical theory, the vapor-liquid interface is a two-dimensional object. In reality it is a region in which properties change smoothly over a few nanometers from its liquid bulk to its gas bulk value. Properties of that nanoscopic interfacial region can presently not be measured experimentally, but can be predicted using methods from molecular thermodynamics. Different theoretical methods consistently predict an enrichment of low-boiling components in many mixtures [1], which is suspected to influence the mass transfer through the interface. This enrichment would be an insurmountable obstacle to mass transfer according to the Fickian theory and, hence, is suspected to influence for example fluid separation processes like absorption or distillation.
To study the influence of the enrichment on the mass transfer, two new simulation methods based non-equilibrium molecular dynamics simulation (NEMD) were developed by our group [2,3]. The first method establishes a stationary mass flux across vapor-liquid interfaces and the second method investigates the transient mass flux of a low-boiling component through an interface into a liquid phase. For the stationary method, averaging is carried out using the classical block averaging method. In the non-stationary method, averaging is carried out using the replica method. Two model systems were used for testing and comparing the new NEMD simulation methods. The non-stationary method is found to be significantly more robust and insightful compared to the stationary method. In particular, the non-stationary method provides a high-resolution insight into the response functions for the component densities, fluxes, and pressure in the interfacial region. The benefits and drawbacks of the two methods will be discussed in detail in this contribution.
To study the influence of the nature of the interface, i.e. enrichment, surface tension, thickness etc., on the mass transfer through the vapor-liquid interface, a comprehensive simulation study was carried out focusing on the non-stationary simulation method. The influence of the temperature, the (bulk phase) composition, and the mixture type on the interfacial mass transfer effects was thereby studied. Five binary Lennard-Jones model systems were considered: An ideal system (A), a high-boiling azeotropic system (B), a low-boiling azeotropic system (C), a heteroazeotropic system (D), and a system exhibiting a high pressure VLLE (E). These systems are good candidates for a systematic study as their phase equilibrium and equilibrium interfacial properties [3,4] as well as their bulk phase transport coefficients [6,7] have been systematically studied in a previous work of our group. In particular, the five systems exhibit important differences in their interfacial properties. For example, at equilibrium states, the enrichment of the low-boiling component at the interface increases going from mixture A to E. The results from the NEMD simulations indicate a significant influence of the enrichment on the mass transfer. Both simulation methods consistently yield a reduced mass flux (about an order of magnitude) in the case of high enrichment (systems D and E) compared to cases with no or low enrichment (systems A to C). It is shown that the differences in transport properties of the bulk phases alone cannot account for the findings of the NEMD simulations. Hence, the differences in the mass flux are attributed to interfacial effects. Furthermore, it was observed, that the component entering the liquid phase through an interface may be repelled from the interface. In some cases, even a net negative mass flux back into the vapor phase is observed due to the repelling of particles from the interface. This repelling of particles significantly depends on the nature of the interface, i.e. its surface tension and interfacial thickness.
References
[1] S. Stephan, H. Hasse, Int. Rev. Phys. Chem. 39, 3 (2020) 319-349.
[2] S. Stephan et al., Mol. Phys. 119, 3 (2021) e1810798.
[3] D. Schäfer, S. Stephan et al., J. Phys. Chem. B (2023) in press.
[4] S. Stephan, H. Hasse, Phys. Chem. Chem. Phys. 22 (2020) 12544.
[5] J. Staubach, S. Stephan, J. Chem. Phys. 157 (2022) 124702.
[6] D. Fertig, H. Hasse, S. Stephan, J. Mol. Liqu. 367 (2022) 120401.
[7] D. Fertig, S. Stephan, Mol. Phys. (2023) e2162993.