(178o) Adsorption of Polar Compounds From Gas and Solution Phases Onto Zeolites

Over the past 20 years, molecular simulation studies have been widely utilized to investigate the adsorption of non-polar and weakly polar species from the gas phase onto zeolites. In contrast, investigations of the adsorption of polar and hydrogen-bonding compounds from a solution phase are very sparse because of a lack of transferable force fields and efficient simulation algorithms for these more complex systems. Here, we present methodological developments that overcome these limitations and apply the new methodologies to adsorptive separations of practical interest.

  First, the TraPPE-zeo (transferable potentials for phase equilibria-zeolites) force field is parameterized and shown to be capable of predicting the adsorption isotherms for n-alkanes, carbon dioxide, alcohols, and water with high accuracy. The TraPPE-zeo force field provides Lennard-Jones parameters and partial charges for the framework atoms, and utilizes Lorentz-Berthelot combining rules for all unlike interactions, making it applicable to any sorbate compound as long as suitable force field parameters are available to describe the sorbate-sorbate interactions; a significant advantage over force fields that explicitly tabulate the individual cross interaction parameters.

  Second, a simulation strategy is optimized that combines the configurational-bias Monte Carlo (CBMC) techniques and the Gibbs ensemble (GE). This methodology efficiently samples particle transfers and allows us to investigate adsorption from solution phases over the whole composition range (without reliance on experimental activity models) and any number of sorbate species.

  Third, the TraPPE-zeo force field and CBMC-GE approach are applied to investigate the multi-component adsorption of water, alcohols, and other oxygenated compounds. All-silica zeolites are found to be highly selective for alcohols over water. The ideal adsorbed solution theory is assessed for these systems and found to substantially under predict the amount of sorbed water and leads to very large errors for low-alcohol solution concentrations.

  In addition, results are presented for the sensitivity of the simulation data to minor changes in the framework structure and to different potential truncation methods.