Finding new functional materials is a sensitive job. Looking for very specific properties in a small group of known materials is even harder. A team of researchers have overcome this. They built a database of known materials. And then they used machine learning, a sub-discipline of computer science, to construct algorithms that enable learning from data. They used this learning to make better guesses.
With this, they identified chemical compositions that are probable bets for the material they wanted to develop. They investigated over 3000 possible materials with the data science approach, and found over 200 promising materials.
They then applied different meticulous quantum mechanical evaluations. With this, they determined the atomic structures of the potential candidates and examined their stability.
They further investigated whether the filtered candidates have the guessed structure, electric polarization, and favor lab creation.
This narrowed the potential candidates to 19. They immediately recommended these for experimental synthesis.
"Our work has the potential to help save enormous amounts of time and resources. Instead of exploring all possible materials, only those materials that have the potential to be promising will be recommended for experimental investigation" said Prasanna Balachandran, the paper’s co-author.