Recent advances in computational screening of metal-organic frameworks for carbon capture

Adsorption is regarded as a potential solution for carbon dioxide (CO₂) capture due to its high gas storage capacities, selectivity, and recovery of CO₂. The design and development of the carbon capture unit are significantly influenced by the choice of adsorbent. Due to their large surface area, adaptable pore architectures, design versatility, and CO₂ selectivity, metal-organic frameworks (MOFs) have drawn a lot of interest in this sector. Due to their high tunability and customizable structures, ∼one million MOFs are experimentally and computationally synthesized and reported in databases such as Computational Ready (CoRE), topologically based crystal constructor (ToBaCCo), hypothetical MOFs (hMOFs), in silico, and Zr MOFs, etc. However, testing MOFs experimentally from the millions of structures for the identification of top-performing MOFs for CO₂ capture is infeasible. To overcome this challenge, a high-throughput screening (HTS) technique is applied to segregate datasets based on their adsorption performance characteristics, such as selectivity of target adsorbate, working capacity, regenerability, adsorbent performance score, etc, measured computationally. Although high-throughput screening alleviates the experimental effort, computational techniques consisting of various simulation tools and density functional theory are expensive computationally. Rapid growth in computational power and advancement in data-driven modeling techniques, such as machine learning, could mitigate the HTS time and labor enormously. This data-driven screening technique requires the physical, chemical, and adsorption characteristics to develop an accurate model to predict the carbon capture performance. This technique enables predictive modeling, optimizes the MOF design, and provides interpretability towards the affecting parameters.