Reticular frameworks are crystalline porous materials that form <i>via</i> the self-assembly of molecular building blocks (<i>i.e.</i>, nodes and linkers) in different topologies. Many of them have high internal surface areas and other desirable properties for gas storage, separation, and other applications. The notable variety of the possible building blocks and the diverse ways they can be assembled endow reticular frameworks with a near-infinite combinatorial design space, making reticular chemistry both promising and challenging for prospective materials design. Here, we propose an automated nanoporous materials discovery platform powered by a supramolecular variational autoencoder (SmVAE) for the generative design of reticular materials with desired functions. We demonstrate the automated design process with a class of metal-organic framework (MOF) structures and the goal of separating CO<sub>2</sub> from natural gas or flue gas. Our model exhibits high fidelity in capturing structural features and reconstructing MOF structures. We show that the autoencoder has a promising optimization capability when jointly trained with multiple top adsorbent candidates identified for superior gas separation. MOFs discovered here are strongly competitive against some of the best-performing MOFs/zeolites ever reported. This platform lays the groundwork for the design of reticular frameworks for desired applications.
Inverse Design of Nanoporous Crystalline Reticular Materials with Deep Generative Models
