In Brief
Scientists have used high-throughput computational models to predict and build two new magnetic materials. This breakthrough means we can now design and build new magnetic materials at larger scales, lower costs, and unprecedented speeds.

Shortcut to Magnetic Materials

Material scientists from Duke University have used high-throughput computational models to predict and build two new magnetic materials from the atom up. The breakthrough means we can now design and build new magnetic materials at large scales and unprecedented speeds. This is critical because only around five percent of known inorganic compounds show even a hint of magnetism, and few of them are practically useful due to variabilities in their magnetic permanence and effective temperature range.

Supercomputers: To Moore’s Law and Beyond
Click to View Full Infographic

Thanks to this relative scarcity, magnetic materials can be expensive and difficult to obtain. The traditional manner of searching for new options is tedious —researchers churn out new structures hoping to find magnetic properties, with some molecular structures failing while others succeed. High-performance magnets are rare oddities among more predictable chemical and physical trends.

In this new work, the researchers used supercomputers to create a shortcut in the process by modeling the potential magnetism of hundreds of thousands of candidates rapidly, cutting the list of potential configurations from 236,115 to just 14. They know the modeling process works because they’ve already created two completely novel magnetic materials based on their predictions.

“Predicting magnets is a heck of a job and their discovery is very rare,” professor of mechanical engineering and materials science and director of the Duke Center for Materials Genomics Stefano Curtarolo said in a Duke University news release. “Even with our screening process, it took years of work to synthesize our predictions. We hope others will use this approach to create magnets for use in a wide range of applications.”

Super Problem Solvers

The first novel magnetic material the team created was made of cobalt, manganese, and titanium (Co2MnTi). The team predicted the properties of the new magnet with a high degree of accuracy by comparing the measured properties of similarly structured magnets.

Notably, the substance possesses an exceptionally high “Curie temperature” (the temperature at which it loses its magnetism) of 664.85 degrees Celsius (1,228 degrees Fahrenheit) — almost exactly what the researchers predicted. Coupled with its lack of rare earth elements, which are expensive and difficult to acquire, the new substance will likely prove very useful in many commercial applications.

The second material comprised manganese, platinum, and palladium (Mn2PtPd). It is an antiferromagnet, a variety of magnet that is difficult to predict. Its electrons are evenly divided in their alignments, so the material has no internal magnetic moment — its electrons instead respond to external magnetic fields. That property limits its applications to use within hard drives, magnetic field sensing, and Random Access Memory (RAM), but the researchers’ calculations for this material’s properties were accurate as well.

This advancement marks yet another important use for supercomputers, which are already being used to predict monsoons, reveal the origins of cosmic structures, and even extend human life expectancy. As predictive modeling is applied to new situations, supercomputers will solve more and more of humanity’s problems, freeing us up for more creative and innovative pursuits.