(599a) Active Learning for Efficient Navigation of Multi-Component Gas Adsorption Landscapes in a MOF

Multi-component gases are omnipresent, be it nature or industrial streams, and they are utilized for a range of applications from energy to healthcare. In recent decades, metal-organic frameworks (MOFs), have gained recognition for their potential in gas adsorption applications. Though simulations have been useful in producing the structure-property relationships of MOF-adsorbate systems, they can be computationally expensive and there is a need for faster surrogate models that can predict the adsorption data. In this work, we introduce an active learning protocol that can predict the adsorption for gas mixtures in a MOF for a range of thermodynamic features. We apply this methodology to build a model for three different gas mixtures (CO₂-CH₄, Xe-Kr, and H₂S-CO₂) adsorption in Cu-BTC MOF. Gaussian process regression (GPR) model is used to fit the data as well to leverage its predicted uncertainty to drive the learning. The training data is generated using grand-canonical Monte Carlo (GCMC) simulations as we iteratively add points to the model to minimize the predicted uncertainty. We also introduce an accuracy-based criteria to terminate the active learning process when the accuracy threshold is met. The three systems are tested for a pressure-mole fraction (P-X), and a pressure-mole fraction-temperature (P-X-T) phase space. We demonstrate that using active learning we only need a fraction of the data from simulations to build a reliable surrogate model for predicting mixture gas adsorption. We also show the final GP fit based of AL outperforms IAST predictions.