(403a) High-Throughput Approach to Studying Adsorption on Zeolite Surfaces

Zeolites are nowadays abundantly used as adsorbents and catalysts for many industrial applications. Computational studies have recently complemented the widely available experimental data and observations on zeolites by shedding light on reaction mechanisms and the active sites. The computational studies of zeolites, however, are challenging due to the numerous possible active sites. A zeolite structure can have many unique tetrahedral sites (T). Substitution of a trivalent aluminum atom to a tetravalent silicon atom in the zeolite framework leads to a net negative charge that can be compensated by a proton or a cation. However, there are numerous possibilities for the placement of Al and the compensating proton or cation. The number of possibilities grow exponentially for higher Al/Si ratios in the framework. Due to this complexity, most theoretical studies have focused on a limited number of T sites based on physical intuitions. For example, many studies of the MFI zeolite, which has 12 T-sites, have focused only on one or two T sites based on the accessibility of guest molecule to be the active site. These choices are justifiable given the high computational cost of accessing the full descriptor space, however, they probably miss promising active sites. In this work, we attempt to thoroughly explore the possible active sites using a high-throughput approach. We apply our approach to the problem of NO adsorption on metal-exchanged zeolites. It is known that Pd-exchange CHA, for example, is an effective adsorbent to limit NOx emissions. Full exploration of this problem requires enumerating the possible Al-Al pairs on the starting zeolites, identifying the favorable sites for the compensating protons and cations and searching for a global minimum in each case, which means thousands of calculations even for simple zeolites. The problem is further compounded when the cation is a metal that might have multiple oxidation states. We show how we systematically explore the possible sites, starting with a lower level of theory on the initial calculations, and then narrowing the potential candidates and using a higher level of theory afterwards. The automation of this calculation reduces the amount of manual labor and highly improve the efficiency of the workflow. It presents a new approach to tackle the complicated problems related to zeolite catalysis.