Design of active sites for amine-functionalized direct air capture materials using integrated high-throughput calculations and machine learning

[Geoengineering News]

‎https://www.nature.com/articles/s42004-025-01819-1

‎

Authors: Megan C. Davis, Wilton J. M. Kort-Kamp, Ivana Matanovic, Piotr Zelenay & Edward F. Holby

08 December 2025

‎

‎Abstract

Direct air capture (DAC) of carbon dioxide is a critical technology for mitigating climate change, but current materials face limitations in efficiency and scalability. We discover novel active sites for DAC materials using a combined machine learning (ML) and high-throughput atomistic modeling approach. Our ML model accurately predicts high-quality, density functional theory-computed CO2 binding enthalpies for a wide range of nitrogen-bearing moieties. Leveraging this model, we rapidly screen over 1.6 million binding sites from a comprehensive database of theoretically feasible molecules to identify binding sites with superior CO2 binding properties. Additionally, we assess the feasibility of experimentally synthesizing these structures using established ML metrics, discovering nearly 2,500 novel binding sites. This set of binding sites may be used for targeted design of functionalized amine sorbents for multi-objective optimization strategies. Altogether, our high-fidelity database and ML framework represent a significant advancement in the rational development of scalable, cost-effective carbon dioxide capture technologies, offering a promising pathway to meet key targets in the global initiative to combat climate change.

Source: Communications Chemistry