High-throughput Simulations of Metal Organic Frameworks (MOFs) for CO2 Capture

Metal-organic frameworks (MOFs), composed of metallic nodes that are linked together by organic linkers, have gained considerable attention as a type of nano-porous materials in the past three decades. MOFs are highly appealing for their potential in various applications, such as gas storage and separations owing to their high porosity and adjustable surface properties. Nevertheless, finding the most suitable MOFs for industrially relevant gas adsorption and separation processes is challenging due to the vast number of possible combinations of inorganic and organic building units, which result in hundreds of thousands of different MOFs. With this large number of structures, it is critical to develop computational protocols to rapidly and efficiently screen databases of MOFs for energy applications such as CO2 capture or hydrogen storage.
Firstly, to validate the computational methods and models used, this thesis presents computational geometric properties and CO2 uptake characterisation of Zirconium-based MOFs. We perform systematic periodic density functional theory (DFT) calculations comparing 25 different combinations of basis sets and functionals to calculate framework partial atomic charges. We then compare simulation results with published experimental data for CO2 adsorption isotherms, and demonstrate the good agreement between simulated and experimental data. We then use 102 structures containing Zr-oxide secondary building units (SBUs) extracted from the Cambridge Structural Database (CSD) MOF subset to perform high-throughput adsorption simulations and identify top candidates for post-combustion CO2 capture.
The second focus of this thesis is on the use of simulations to explain the remarkable CALF-20 properties to capture CO2 from flue gas, while maintaining stability and resisting water. Here, we employ atomistic-level simulations and experiments to investigate the adsorptive characteristics of CALF-20 and gain insights into its flexible crystal structure. We analyse and compare CO2 and water adsorption isotherms, and elucidate the significance of water-framework interactions and hydrogen bonding networks in CALF-20's hydrophobic properties. In addition, we conduct molecular dynamics simulations using both density-functional theory (DFT) and machine learning potentials (MLPs) trained to DFT energies and forces. The simulations reveal the impact of adsorption-induced flexibility in CALF-20.
Lastly, through a collaborative work with our industrial collaborator, a multi-scale computational strategy that includes DFT and grand canonical Monte Carlo (GCMC) simulations is established. This strategy develops a systematic high-throughput computational screening to explore The Quantum MOF (QMOF) database and the Computation-Ready, Experimental (CoRE) MOF database for selective adsorption of CO2 from a wet flue gas mixture. The screening protocol effectively reveal several interesting MOFs that exhibit selectivity towards CO2 in the presence of water vapor. By establishing a tight feedback loop between simulations, and experiments performed by our industrial collaborator, we are able to effectively synthesise and evaluate several MOFs with relatively high stability in the laboratory setting. This process confirms the validity of our technique in selecting exceptional MOFs for the purpose of CO2 capture.