Nanoporous materials and in particular metal–organic frameworks (MOFs) are considered promising materials for the capture and storage of CO2 due to their exceptional guest-host properties. Since there are numerous MOFs that can be potentially synthesized, the challenge here is to identify those MOFs with the highest CO2 capacities. In this work we discuss and evaluate an iterative self consistent (SC) approach that is based on machine learning (ML) and aims to identify the most important materials from a large pool of candidates using as less information as possible. We have used an existed database of ∼ 325,000 hypothetical MOFs for which the adsorption of the CO2 has been already computed at several thermodynamic conditions. The predictions of the proposed method are compared with those of a traditional ML approach in which the ML algorithm is trained using a randomly selected set of materials. It is seen that the proposed method provides superior predictions for the top-performing materials for the CO2 capture under any scenario. In particular, when computer resources are rather limited, the adoption of our iterative procedure leads to the identification of a significant number of top performing MOFs.

Efficient machine-learning based approach for the identification of top-performing metal-organic frameworks for CO2 storage

Where thousands of chemistry professionals meet to share ideas and advance scientific and technical knowledge. Spring 2023 abstracts, posters, and presentations will launch at the end of March.

ACS Spring 2023

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.