Machine learning allowed the model to acquire an understanding of the structural properties of different materials. The algorithm uses its knowledge to predict new combinations of atoms to form materials with valuable physical and structural properties.
With the endless possible atomic combinations, materials scientists can be overwhelmed. Computer models can help researchers narrow their focus.
"Understanding which atoms will combine to form new materials from the vast space of possible candidates is one of the grand scientific challenges, and solving it will open up exciting scientific opportunities that could lead to important properties," Matt Rosseinsky, a professor of chemistry, said in a news release.
The model works by identifying only chemically stable combinations that can be reliably synthesized in the lab. Researchers hope the model can help material scientists develop new materials for energy generation and storage.
Rosseinsky and his colleagues detailed their work in the journal Nature.
"The key step in this research was the ability to generate large numbers of truly representative structures that could be used to assess which element combinations were stable," Rosseinsky said, "which greatly narrowed the space that had to be explored experimentally -- like having a map with someone's address, rather than knowing they live in London somewhere."
| Buy Advertising | About Us | Editorial & Other Enquiries | Privacy statement |
| The content herein, unless otherwise known to be public domain, is Copyright 1995-2026 Space Media Network. All websites are published in Australia and are solely subject to Australian law and governed by Fair Use principles for news reporting and research purposes. |
