image credit:- Xiaoli Yan/University of Illinois Chicago
Capturing carbon dioxide emissions from industrial facilities could greatly reduce greenhouse gas levels, but finding the right materials to do this efficiently remains challenging. Metal-organic frameworks (MOFs) have potential for selective carbon capture, but with near endless possible configurations, testing each one would be impractical.
To accelerate the discovery process, researchers from the U.S. Department of Energy’s Argonne National Laboratory are utilizing several innovative techniques. These include using generative artificial intelligence (AI) to identify new MOF candidates, high-throughput screening of materials, theory-based molecular dynamics simulations, and harnessing the power of supercomputers.
In one part of the project, the team used generative AI to assemble over 120,000 potential new MOFs in just 30 minutes by intelligently combining molecular building blocks. This AI-based technique enables rapid exploration of the vast MOF design space that would take an impractical amount of time using traditional methods.
The most promising materials identified by the AI were then screened using molecular dynamics simulations on supercomputers like Polaris and Delta at the Argonne Leadership Computing Facility. This approach allows the scientists to synthesize and test only the best candidates from a pool of millions.
“We are now connecting generative AI, high-throughput screening, molecular dynamics, and Monte Carlo simulations into a standalone workflow,” said Argonne computational scientist Eliu Huerta. “This incorporates online learning using past research to improve the AI’s precision in creating new MOFs.”
Looking ahead, the team will leverage Aurora, Argonne’s next-generation exascale supercomputer, to potentially screen billions of AI-generated MOF candidates at once – including many never synthesized before.
Collaborators on the project include UIUC, University of Illinois at Chicago, University of Chicago, and the Beckman Institute. This demonstrates the power of using AI in molecular sciences,said UIUC’s Emad Tajkhorshid. “We hope to expand this approach to problems like biomolecular and drug design.”
The cross-institutional team of young researchers pioneered new ways of accelerating materials discovery for carbon capture using AI. Their innovative approaches will guide future efforts to design optimized frameworks for selectively absorbing carbon dioxide emissions.
