(592e) Predictive Models of Azeotropic Adsorption from Molecular Simulations and Macroscopic Theories

The knowledge of multicomponent adsorption equilibria is of crucial relevance in the design of adsorption separation processes¹. Prediction of multicomponent equilibria from single component measurements mainly relies on the assumption that the adsorbed mixture can be treated as an ideal solution². However, some adsorption systems deal with heterogeneity of the solid sorbent and fluid mixtures composed of species different in size and polarity. Azeotropic behaviour can also occur in such cases, and its analysis and macroscopic modelling still remains a major challenge³⁴.

This works has two objectives: to understand behaviour of complex adsorbed mixtures, capable of forming azeotropes, on a molecular level; using insights from a molecular-level picture, develop a predictive macroscopic model of adsorption of these mixtures, suitable for design of adsorption processes.

For this, we employ Grand Canonical Monte Carlo (GCMC) simulations to explore how the heterogeneity of the solid, and mixture of gases of different size and polarity, can lead to azeotropic adsorption behaviour. As a case study we consider binary isobutane-ethylene mixture adsorbed in 13X zeolite⁵. Within molecular simulations adsorptive behaviour of different regions of the porous space can be explored in isolation from the rest of the system⁶. This allows us to define the adsorption sites within the system of interest and obtain specific information about properties of these sites, such as: heat of adsorption, Henry’s law constants, full single and multicomponent isotherms of both isobutene and ethylene on each site. This analysis elucidates a microscopic origins of the azeotropic behaviour of the system.

The results from GCMC simulations and the analysis above provide a foundation for the parametrisation of the thermodynamic “Rigid Adsorbent Lattice Fluid” (RALF) model, recently developed by Brandani⁷. The novel heterogeneous version of RALF is presented here to account for different adsorbing sites. Indeed, RALF model has shown comparable performance for the prediction of single component isotherms with respect to empirical equations, and good a priori predictive capability for complex multicomponent systems. Therefore, using a combination of molecular simulations and advanced RALF theory, we achieve the second objective of this study.

References

1. Ruthven, D. M. Principles of Adsorption and Adsorption Processes. (1984).
2. Myers, A. L. & Prausnitz, J. M., AIChE J. 11, 121–127 (1965).
3. Do, D. D. & Do, H. D., Adsorption 5, 319–329 (1999).
4. Lucena, S., Snurr, R. Q. & Cavalcante, C. L. J., Adsorption 13, 477–484 (2007).
5. Hyun, S. H. & Danner, R. P., J. Chem. Eng. Data 27, 196–200 (1982).
6. Krishna, R. & Paschek, D., Phys. Chem. 1, 453–462 (2001).
7. Brandani, S., AIChE J. 65, 1304–1314 (2019).