(220a) Computationally-Efficient High-Throughput Screening of Nanoporous Materials for Hydrogen Storage

Metal-organic frameworks (MOFs) are nanoporous materials consisting of metal nodes connected with organic linkers. These materials can have high surface area and porosity, which make them interesting candidates for applications in gas separation and storage. Since MOFs can be made from many different nodes, linkers, and functional groups, there are a huge number of possible MOF structures. High-throughput computational screening can efficiently characterize large numbers of materials to identify the best candidates for applications such as gas separation and storage. [1,2]

We have used high-throughput screening techniques paired with grand canonical Monte Carlo simulations (GCMC) to calculate the hydrogen storage capacity for over 130,000 MOFs. We will discuss commonalities in the structural and textual properties among the top-performing candidates from this screening and explore how the information gleaned from high-throughput screening can inform the design of new materials that are tailored for efficient hydrogen storage applications.

High-throughput screening generates a large amount of data, which makes it amenable to using machine learning and data mining to analyze the data for thousands of structures. We introduce an efficient regression model that can estimate the storage capacity for thousands of structures in a small fraction of the time required for full GCMC simulations. This allows us to quickly identify the most promising candidates for more detailed study, and this technique can be applied to other gases, such as methane, as well as a wide range of temperatures and pressures.

We demonstrate the utility of this screening method by searching a database of 55,000 known MOF structures [3] to identify the top candidates for hydrogen storage. After finding the best structures, we worked with experimental collaborators to synthesize one of the top structures and test its hydrogen storage capacity. This paradigm of using molecular simulation and machine learning techniques coupled with experiments can greatly accelerate the discovery and synthesis of advanced materials.

1. J. Colon, R.Q. Snurr, Chem. Soc. Rev.43, 5735-5749, 2014
2. S. Bobbitt, J. Chen, R. Q. Snurr, J. Phys. Chem. C,120, 27328-27341, 2016
3. Z. Moghadam, A. Li, S. B. Wiggin, A. Tao, A. G. P. Maloney, P. A. Wood, S. C. Ward, D. Fairen-Jimenez, Chem. Mater., 29, 2618-2625, 2017