From your computer to maglev trains, from power tools to MRI scanners, rare Earth permanent magnets are all around us. Modern life without them is difficult so their importance can’t be overstated. However, extracting the rare Earth elements that make them is often laborious and energy-consuming. Scientists have been looking for a better way – and thanks to a machine learning algorithm, they might have found it.

Company Materials Nexus, together with researchers at the Henry Royce Institute and the University of Sheffield, have developed MagNex. This is a permanent magnet that is free of rare earth elements. The MagNex is reported to have been produced with materials that cost one-fifth of regular permanent magnets. The new magnet also saw a reduction of 70 percent in carbon emissions (in terms of kilograms of CO₂ per kilogram of material) compared to rare-Earth permanent magnets.

That in itself is pretty awesome but there’s a wider reason why it is exciting. The permanent magnets that we have were developed from alloys of rare-Earth elements in the 1970s and 80s. Searching for these materials with the right properties was a long undertaking, with a lot of trial and error. MagNex’s development from design to testing was 200 times faster.

“We’re really excited that our first interaction with Materials Nexus has yielded such a hugely positive outcome,” Professor Iain Todd FREng, Professor of Metallurgy and Materials Processing at the University of Sheffield, said in a statement. 

“The combination of Materials Nexus’s approach of using AI [artificial intelligence] for materials discovery and the world-class facilities we have for manufacture of advanced alloys in the Henry Royce Institute here at Sheffield has allowed a novel magnetic material to be developed with breathtaking speed. This achievement showcases the bright future of materials and manufacturing. The next generation of materials, unlocked through the power of AI, is highly promising for research, industry, and our planet.”

The AI system identifies and analyzes the composition of over 100 million potential alloys that would have the right properties to be a permanent magnet, be free of rare-earth elements, and meet the requirements of affordability and sustainability. The potential of this approach for the creation of new materials is enormous.

“This is a ground-breaking discovery using innovative machine learning software, and its development has been enabled by Innovate UK funding,” said Bruce Adderley, Director for Make & Use – Net-Zero at Innovate UK. “This could have a significant future impact on our net-zero ambitions, through renewable energy and low-carbon transport, by removing the need for rare earth elements in high-performance permanent magnets.”

Computers have already helped enormously in allowing us to find new substances. And it seems that thanks to machine learning algorithms, this and similar approaches are being used in labs as well, such as the self-driving lab developed at the University of Toronto.