Rational, Hierarchical Parameterization of Complex Zeolite-guest Molecular Models

A hierarchical approach is presented for rationally parameterizing complex molecular models of guest species in microporous materials. The large number of microscopic processes (e.g. adsorption, desorption, site-to-site migration), and intrinsic complexity of potential energy surfaces characterizing these systems, demands simultaneous optimization to multiple sets of experimental data to ensure model predictive capabilities. The first step of the approach capitalizes on the computational efficiency of mean field models to initially identify sensitive parameters and narrow the window of parameter space. Surface response techniques are subsequently employed in this window for rational optimization of the computationally intensive molecular models. The hierarchical approach is developed in light of both enthalpic and entropic constraints to ensure consistency of multiple microscopic processes. We illustrate this approach using a lattice kinetic Monte-Carlo description of benzene in NaX zeolite. Through application of simulated annealing techniques, the hierarchical framework is capable of optimizing the model parameters simultaneously to disparate sets of experimental data (e.g. adsorption isotherms and PFG-NMR self-diffusivity). Finally, application to other atomistic models is also discussed.