Developing new materials to tackle pressing global issues is a gap-filled process that can leave potential climate solutions lost in unread academic papers and forgotten dissertations.
A new, end-to-end, machine-learning-guided workflow created in the lab of UChicago Pritzker School of Molecular Engineering and Department of Chemistry Prof. Laura Gagliardi is helping smooth the path from academic idea to manufacture-ready reality in a single discovery process.
Through the Center for Advanced Materials for Environmental Solutions (CAMES) , which Gagliardi co-directs, the lab used this new process to create two new high-performing materials for separating methane from nitrogen. These new zinc-based metal-organic frameworks (MOFs), UCHI-1 and UCHI-2 (named for the University of Chicago but pronounced "you-key" one and two) provide state-of-the-art gas adsorption and separation, but through a smoother, more efficient, less costly path from design to debut.
The work was recently published in the Journal of the American Chemical Society.
The Gagliardi Group collaborated with UChicago Chemistry Department Prof. John Anderson and Anderson Lab postdoctoral scholar Jianheng (Allen) Ling to synthesize the two new materials envisioned through this process. This experimental work is fundamental to closing the loop with computational predictions and, most importantly, to creating the materials that can be tested for potential industrial applications.
"The end-to-end framework we created connects data mining, machine-learning predictions, materials design, synthesis, and experimental validation in a single discovery cycle," Gagliardi said. "This helps overcome a major barrier in computational materials discovery: Many theoretically promising materials are never synthesized, while experimental development traditionally relies on slow and costly trial and error."
Although MOFs are also proving to be powerful tools for fighting airborne carbon dioxide, the team chose to focus on a less-studied greenhouse gas – methane. Methane from agriculture, especially livestock, as well as landfills, coal mining, oil and natural gas operations stays in the atmosphere for about a decade, compared with the thousands of years CO2 can linger. But during that time, it does massive damage.
Over a 20-year time scale, methane has a climate impact 80 times greater than CO2 .
"Methane as a greenhouse gas is more potent than CO2, but it has not been considered as much because it stays in the atmosphere a shorter time than CO2," said Gagliardi Group postdoctoral researcher Andrea Darù , the paper's first author. "Improving how we capture methane will also benefit industry and jobs. Industry leaks methane at a loss of about $10 billion per year through pipes, through compression machines, through anything related to methane gas for distribution."
Filling the gaps
In the traditional process for designing and building new materials, computational groups devise and describe potentially interesting molecular designs. An experimental group must then pick up the work, create the materials and test them. Industry must ultimately incorporate the material into a product – the final step in bringing an academic idea into the real world.
Many promising materials are lost or forgotten along the way.
"In computational groups, predicting and generating structures generally ends in a set of files, which experimental groups might later pick up and take forward to synthesis," Darù said. "Other groups working in MOFs are often experimentalists. They base their research on previous chemical knowledge, improving designs through trial and error while using computation in a lighter way than what we use. Our new end-to-end framework is meant to connect the two sides."
In the best cases, computationalists, experimentalists and industry collaborate to move research from idea to product, he added. In other cases, potentially valuable discoveries may never move beyond the academic research stage.
To create UCHI-1 and UCHI-2, the team trained an AI on datasets from academic literature, honing and iterating the design. They worked in collaboration with experimentalists and industry, rather than handing off the work once done. With an eye toward real-world considerations, the researchers worked in zinc, a material less expensive than the nickel or copper often used for methane-capture MOFs.
"We could obtain slightly methane-nitrogen separation than what is already in the literature, but at a lower cost," Darù said.
The result was a single workflow that created two manufacture-ready MOFs for fighting methane pollution. UCHI-1 and UCHI-2 serve as proof of concept for the process – the team hopes to build new materials that perform better than the current state of the art as they improve their end-to-end workflow.
"Ultimately, we want to have a material that works," Darù said. "We don't only want to always stop with the academic research side of discovery. This starts with research in our labs but connects directly to industry."
This work was made possible by the collaborative framework of the University of Chicago Institute for Climate and Sustainable Growth and by the close partnership among researchers at UChicago, Argonne National Laboratory, and Northwestern University, Gagliardi said.
"By finding a new, replicable route around the bottlenecks and gaps that keep important new materials from mass production, this project exemplifies the central vision of CAMES," said CAMES Co-Director Doug Weinberg. "We want breakthrough science to move beyond the lab and into the world, where it can improve lives."
