Development of a Monte Carlo Algorithm for the Prediction of Acid Site Distribution in Zeolites

Zeolites are porous aluminosilicates that afford exceptional benefits as catalysts in the petrochemical industry, or for vehicle emission control, to name a few. The presence of aluminum (Al) in the silicate (SiO4) framework creates a charge defect, which leads to Brønsted acid sites when compensated with a proton. If the defect is compensated by a metal cation, a Lewis acid site is formed. For metal-exchanged zeolites, in particular, the Al distribution and probability of forming paired Al sites in the framework determines the metal speciation and, in turn, the selectivity and activity of the catalyst. This thesis project aims to create an algorithm that can estimate the Al distribution and probability of forming paired Al sites in zeolites based on user-defined parameters: a) Silicon-to-aluminum ratio (SAR), b) simulation temperature, c) number of trials generated for statistical analysis, and d) Al-Al interaction energy. Two different implementations of a Monte Carlo algorithm were tested and evaluated for zeolites with CHA, MFI and MWW framework structures. The traditional (MC1) and modified (MC2) Metropolis Monte Carlo Methods have different means in generating the initial guess for the Al distribution. In MC1, the initial population is random while conforming to Löwenstein’s rule, while in MC2, Löwenstein’s rule is used to generate an initial guess with maximized density of Al site pairs. Even though MC2 has a higher computational cost than MC1, MC2 is better than MC1 because it is less prone to error and has a higher overall efficiency. The final algorithm generates population statistics that are in full agreement with expected results for the provided Al substitution energies and the probability of forming paired Al sites.