Uni team aims to extract instant materials ‘recipes’ from research papers

The goal is to create a database that contains key details extracted from millions of sources.

Scientists and engineers will be able to target material and any criteria such as precursor materials, reaction conditions, and fabrication processes to arrive at suggested recipes.

The team from MIT, the University of Massachusetts at Amherst, and the University of California at Berkeley, is led by Elsa Olivetti, Atlantic Richfield assistant professor of energy studies in MIT’s department of materials science and engineering (DMSE).

Olivetti and her colleagues have developed a machine-learning system that can analyse a research paper, deduce which of its paragraphs contain materials recipes, and classify the words in those paragraphs according to their roles within the recipes:

In a paper in the journal Chemistry of Materials, the team demonstrates that a machine-learning system can infer general characteristics of classes of materials — such as the different temperature ranges required — or particular characteristics of individual materials, such as different physical forms taken when fabrication conditions vary.

“Computational materials scientists have made a lot of progress in the ‘what’ to make: what material to design based on desired properties,” said Olivetti.

“But because of that success, the bottleneck has shifted to, ‘Okay, now how do I make it?’”