Reproducible Sorbent Materials Foundry for Carbon Capture at Scale

Direct air capture (DAC) is an emerging technology aiming to mitigate climate change by adsorbing CO2 from ambient air for subsequent use or sequestration. There is a dire need to improve the performance of DAC facilities, as measured by metrics including CO2 captured per unit energy. This work presents our vision and current progress towards an autonomous sorbent materials foundry for rapidly evaluating materials for DAC. Our initial work focused on machine learning models predicting CO2 uptake from the sorbent material structure. However, CO2 uptake alone is not enough to evaluate the fitness of a sorbent for DAC. We subsequently develop more holistic performance metrics for DAC sorbents based on the mixed-gas sorption behavior. For example, knowing the mixed-gas sorption behavior and an choosing a refresh cycle it is possible to calculate intrinsic amount of CO2 captured per unit energy for the sorbent material. We show how this work fits into a hierarchical framework of autonomous systems accelerating the development of DAC.