(152at) Accelerating Development of Porous Sorbents for Direct Air Capture Using High Throughput Computing and Machine Learning

Development of efficient sorbent-based processes for direct air capture (DAC) of CO₂ requires careful matching of sorbent properties with operating conditions. We have used extensive computational screening to enable detailed assessment of broad range of Metal Organic Framework (MOF) materials for DAC, including the important impacts of coadsorbed water. We have optimized the structures of thousands of experimentally-derived MOFs using dispersion-corrected DFT, and have augmented these calculations with thousands of crystal structures that include physically plausible linker vacancy defects. The binding energies of CO₂ and H₂O as individual molecules and as coadsorbed species have been computed at the DFT level. The resulting data allow us to directly assess this large library of materials for use in DAC and also to test the accuracy of empirical and machine-learning force fields that will enable efficient searches of a comprehensive array of practical operating conditions.