A group of scientists from Ames Nationwide Laboratory has developed a machine studying mannequin that may predict the Curie temperature of recent materials combos. The Curie temperature is the utmost temperature at which a cloth can keep its magnetism. Through the use of synthetic intelligence, researchers hope to find new everlasting magnet supplies with out the necessity for crucial parts.
Excessive-performance magnets are essential for numerous applied sciences equivalent to wind power, knowledge storage, electrical automobiles, and magnetic refrigeration. Nevertheless, these magnets usually include crucial supplies like cobalt and uncommon earth parts, which have restricted availability. In gentle of this, scientists are motivated to design new magnetic supplies with decreased dependence on these crucial supplies.
Machine studying is a department of synthetic intelligence that makes use of algorithms and knowledge to enhance predictions by way of trial and error. The group at Ames Lab used experimental knowledge and theoretical modeling to coach their machine studying algorithm. The mannequin was educated utilizing details about recognized magnetic supplies, establishing a relationship between digital and atomic construction options and Curie temperature.
To check the mannequin, the group centered on compounds based mostly on cerium, zirconium, and iron. The machine studying mannequin efficiently predicted the Curie temperature of those materials candidates. This success marks a necessary first step in growing a high-throughput technique for designing new everlasting magnets for future technological functions.
The standard strategy to discovering new supplies is commonly time-consuming and costly. Nevertheless, through the use of machine studying, researchers can save time and assets. This new improvement within the area of synthetic intelligence opens up prospects in designing sustainable and high-performance magnetic supplies.
Supply: Ames Nationwide Laboratory, Chemistry of Supplies (DOI: 10.1021/acs.chemmater.3c00892)
