Jamie Bennet
Researchers at the Johns Hopkins Applied Physics Laboratory used artificial intelligence to discover a new superconductor made of zirconium, indium and nickel alloy.
The discovery only took three months from data collection to candidate material fabrication, demonstrating the promise of predictive AI models in accelerating targeted discovery in materials science, the Johns Hopkins University-affiliated research center said Wednesday.
The novel superconductor is a product of the laboratory’s Material Invention Through Hypothesis-unbiased, Real-time, Interdisciplinary Learning project. The multidisciplinary team chose superconductors as a test case given the vast amounts of data on their material composition.
Superconductors can generate energy without losing its own, and the new material was found to have around 9 degrees kelvin of superconducting transition temperature.
The study designed the AI model using the available data on superconductors as well as other known materials. The researchers also accounted for the effects of human biases and adjusted the algorithm to compensate for those gaps.
“This approach is being continually improved and will allow us to develop the materials we need, at a drastically faster pace than was previously possible,” APL scientist Leslie Hamilton commented. “And this combination of AI and domain expertise should, in principle, be generalizable to other areas of material and chemical discovery, which is a very exciting possibility.”
