Characterization of a Metal Organic Framework Database

Abstract

Metal organic frameworks (MOFs) are nanoporous materials composed of inorganic and organic structural building units (SBUs). Over the last several decades, interest in MOFs has grown considerably partially due to their promising capabilities for carbon capture and storage (CCS) technologies. This is mostly due to their tunable pore chemistry, high internal surface area and unique structural diversity. This thesis focuses on computational methods that were used to analyze and organize a database of hypothetical structures to facilitate MOF discovery. The work done is detailed in two main parts. In the first part of the thesis, a topologically diverse hypothetical MOF database, containing over 300,000 structures, is screened using simplified molecular-input line-entry system (SMILES) strings to identify SBUs in each structure. The structures in the database are then renamed according to the SBUs identified by the SMILES strings algorithm. The renaming of the structures allows users to have a good idea of the geometrical and topological distribution of the database. Furthermore, a quick and reliable test is developed to identify structures with incorrect bonding patterns/missing hydrogen. In the second part of the thesis, density functional theory (DFT) - derived charges are generated for each structure in the hypothetical MOF database. Using these charges, the CO₂/N₂ selectivity is calculated and compared with the selectivity values obtained from another charge generating method, split-charge equilibration (SQE), and it is determined that there is good agreement, r = 0.96, between the two methods. A machine learning model is then developed to identify relationships between geometrical features and CO₂/N₂ selectivity.