The name “Magneto” emerged amid discussions about using artificial intelligence to discover crystal structures for stronger permanent magnets. In Marvel Comics, Magneto can control all forms of magnetism. In the chemistry lab, the challenges are less dramatic, but no less ambitious.
The federally funded team hopes to find materials that can generate and sustain magnetic fields stronger than those produced by today’s leading permanent magnets. This effort combines machine learning with hands-on synthesis, testing, and materials characterization.
The U.S. Department of Energy’s Advanced Research Projects Agency, known as ARPA-E, donated $2.7 million to a group led by Iowa State University chemistry professor Kirill Kovnir. Researchers aim to identify, manufacture and test new magnetic compounds that can outperform neodymium iron magnets.
These magnets are the strongest permanent magnets available today. They play a key role in electric motors and generators, and are important for transportation, manufacturing, and power production.
Search for unknown compounds
The grant falls under the ARPA-E program called MAGNITO, which stands for “Magnetic Acceleration for New Innovations and Tactical Outcomes.” The program explores “completely new physics, chemistry, and structures for ultra-strong magnets,” according to its synopsis.
MAGNITO is part of a larger $72 million federal effort to support early-stage research aimed at strengthening domestic magnet manufacturing and protecting the supply chain for critical minerals.
Covenir’s team calls their project MAGNUMS, or “Machine Learning-Assisted Generation of New Superstrong Magnets Through Synthesis.” The name reflects the project’s strategy of using computers to narrow down a vast field of candidate materials before chemists begin the time-consuming work of creating them.
Machine learning allows computers to learn from data and identify patterns. Here, the tool screens possible elements, combinations, and structures for signs of useful magnetic behavior.
James Chelikowski, a professor of physics and director of the Center for Computational Materials at the University of Texas at Austin, will help lead the research. Yongxin Yao, an experimental scientist at the Department of Energy’s Ames National Laboratory and an adjunct associate professor at Iowa State University, will also lead the machine learning effort.
“With cutting-edge theoretical and AI-driven tools, this work is truly like embarking on a treasure hunt for new magnetic materials,” Yao said.
Computer narrows search range
Computers can explore many theoretical possibilities, but promising calculations do not automatically generate useful magnets. Researchers still need to create the material, determine whether the predicted structure forms, and test how it behaves.
The research will involve Covenir and Julia Zykina, an associate professor of chemistry at Iowa State. The experimental team also includes Jaroslav Mudryk from Ames National Laboratory and Iowa State’s Department of Materials Science and Engineering, and Michael Shatorak, a professor of chemistry and biochemistry at Florida State University.
“A lot of the current research is about improving known compounds,” Mudryk says. “The goal of the MAGNITO program is to discover new compounds, which is why chemists are involved.”
Chemists try to coax selected elements into structures that have never been created before. You can vary component ratios, synthesis methods, and temperatures and see how these choices affect the final crystal structure and magnetic performance.
The process can be time and material consuming, especially if the proposed combination does not yield any results. The computational group is expected to ease that burden by identifying promising starting points and steering teams away from less useful directions.
“We look forward to working closely with the computational group, which will provide guidance on where to start and where to go, while saving time and resources from exploring ‘dead ends’,” Zykina said.
From prediction to moving magnet
The success of a project depends on how well computational and experimental work can inform each other. While predictions guide synthesis, experimental results can show where the model was accurate, incomplete, or wrong. This interaction can help your team narrow down their search.
Creating new permanent magnets requires more than just finding a compound with attractive theoretical properties. The material is formed under real-world conditions and must remain stable and retain a strong magnetic field after the external magnetic force is removed.
Zykina described the target as a compound that “has the superpower to generate and sustain a high magnetic field.” Achieving that goal means moving from computational possibilities to physical materials that can be synthesized, measured, and compared to neodymium iron magnets.
The team does not claim that such a successor already exists. The task is exploratory, and the program supports early-stage ideas. This work begins with an extensive search for unfamiliar compounds, followed by the difficult process of proving whether a candidate can deliver the required performance.
Artificial intelligence can accelerate exploration, but it cannot eliminate the need for chemical judgment, controlled experimentation, and repeated testing.
Practical implications of the research
Stronger permanent magnets could improve electric motors and generators. According to the project overview, it will improve energy productivity, reduce power generation costs, and make motors for transportation and industrial use smaller and lighter.
The initiative could also support domestic magnet production by expanding the range of materials available to manufacturers. This is important because the federal program aims to strengthen supply chains for critical minerals used in energy and industrial technology.
Even if the team cannot immediately produce magnets that surpass neodymium iron materials, the project could establish a faster method of exploration. Combining machine learning and targeted synthesis could potentially allow researchers to spend less time on unlikely compounds and more time testing reliable candidates.
The bigger question is whether computation and chemistry can turn vast materials exploration into a manageable path to real products. MAGNUMS tests its ideas one compound at a time.
