Scientists use AI to identify new materials for carbon capture

Carbon capture is an important technology for reducing greenhouse gas emissions from power plants and other industrial facilities. But a suitable material for efficient carbon capture at low cost has not yet been found. One candidate is metal organic frameworks, or MOFs. This porous material can selectively absorb carbon dioxide.

The molecules of MOFs contain three types of building blocks – inorganic nodes, organic nodes and organic linkers. These can be arranged in various relative positions and configurations. As a result, there are countless possible configurations of MOFs for scientists to design and test.

To accelerate the discovery process, researchers at the US Department of Energy’s (DOE) Argonne National Laboratory are pursuing several avenues. is a generative artificial intelligence (AI) that dreams up previously unknown building block candidates. Another is a form of AI called machine learning. A third route is high-throughput screening of candidate materials. And last theory-based simulations using a method called molecular dynamics.

Among Argonne participants in the project are researchers from the University of Illinois at Urbana-Champaign (UIUC), the University of Illinois at Chicago and the University of Chicago’s Beckman Institute for Advanced Science and Technology.

Designing MOFs with optimal carbon selectivity and capacity is a significant challenge. Until now, MOF design has relied heavily on experimental and computational work. This can be expensive and time consuming.

By exploring the MOF design space with generative AI, the team was able to rapidly assemble 120,000 new MOF candidates within 30 minutes. They ran the calculations on the Polaris supercomputer at the Argonne Leadership Computing Facility (ALCF). ALCF is a DOE Office of Science user facility.

They then turned to the Delta supercomputer at UIUC to perform real-time molecular dynamics simulations using only the most promising candidates. The aim is to screen them for stability, chemical properties and carbon capture capacity. Delta is a joint effort between Illinois and its National Center for Supercomputing Applications.

The team’s approach may ultimately allow scientists to synthesize only the best MOF contenders. “People have been thinking about MOFs for at least two decades,” said Argonne computational scientist Elio Herta, who helped lead the research. “Traditional approaches typically involve experimental synthesis and computational modeling with molecular dynamics simulations. But trying to survey the vast MOF landscape in this way is simply impractical.”

Even more advanced computing will soon be available for the team to employ. With the power of ALCF’s Aurora exascale supercomputer, scientists can simultaneously survey billions of MOF candidates, many of which have never been proposed before.

What’s more, the team is taking chemical inspiration from past work on molecular design to discover new ways in which different MOF building blocks can fit together.

“We wanted to add new flavors to the MOFs we were designing,” Huerta said. “We needed new ingredients to synthesize AI.” The team’s algorithm can improve MOFs for carbon capture by learning chemistry from biophysics, physiology and physical chemistry experimental datasets that have not previously been considered for MOF design.

To Huerta, there’s promise of a transformative MOF material beyond traditional methods — one that can be good at carbon capture, cost-effective and easy to produce.

“We are now combining generative AI, high-throughput screening, molecular dynamics and Monte Carlo simulations into a standalone workflow,” said Huerta. “This workflow incorporates online learning using past experimental and computational research to accelerate and improve the accuracy of AI to create new MOFs.”

An atom-by-atom approach to MOF design enabled by AI will allow scientists to get what Ian Foster, Argonne’s senior scientist and director of the Data Science and Learning Division, calls a “broader lens” on these types of porous structures. ” has said. “The work is being done so that, for new AI-assembled MOFs that are predicted, we incorporate the insights of independent labs to synthesize them experimentally,” Foster said. capacity and carbon sequestration capacity can be verified.” “As the model is fine-tuned, our predictions keep getting better and better.”

A paper based on the study was written by Hyun Park, Xiaoli Yan, Ruijie Zhu, Eliu Huerta, Santanu Choudhuri, Donny Copper, Ian Foster and Emad Tajkhorshid. It was published in the online issue of Nature Communications Chemistry.

“The study shows the great potential of using AI-based methods in the molecular sciences,” said UIUC’s Tajkhorshed. “We hope to expand the scope of the approach to problems such as biomolecular simulations and drug design.”

“This work is a testament to the collaboration between graduate students and early-career scientists from different institutions who came together to work on this science-leading AI,” said Huerta. “The future will be bright as we continue to inspire and motivate talented young scientists.”

This work was supported by DOE’s Office of Science, Office of Advanced Scientific Computing Research, laboratory-directed research and development funds, and the National Science Foundation.