Cornell researchers are collaborating with the Toyota Research Institute to employ materials science and artificial intelligence in search of eco-friendly batteries that can be produced in the lab.
Prof. Carla Gomes, computer and information science, and John Gregoire Ph.D. ’09, currently a staff scientist at the California Institute of Technology, are combining Artificial Intelligence, in which computers accelerate the process of creating automated materials through algorithms that use physical properties of basic chemicals and compounds to find materials.
“We [are able to do] a few hundred thousand experiments in an hour. There’s just no way for individual scientists anymore to analyze that. There has to be artificial intelligence,” said Prof. Michael Thompson, materials science and engineering.
Gomes and Gregoire contributed to the creation of the Hierarchical Correlation Learning for Multi-property Prediction, a code system that uses physical properties to predict which of the group two elements of the periodic table to use to make materials with desired properties.
Prof R.B. van Dover, materials science and engineering, who works closely with Gomes, said they use X-Ray crystallography, where the material is bombarded with light to discern the composition of materials — like those used in making batteries.
van Dover also explained how the researchers use HCLMP based analysis to interpret the X-Ray data, and then the AI determines what to synthesize next. The synthesis experiments are executed by laser processing thin films of metal oxides — metals such as barium and radium— that are prepared with a deliberate variation in material composition across the film.
According to van Dover, the lasers can change the crystal structure of the material, depending on the composition, peak temperature and processing time. Laser processing changing the crystal structure allows for the creation of materials not found in nature.
Brian Storey, the Director of Accelerated Material Design and Discovery Group at TRI explained how the expansion of renewable energy would greatly benefit the energy efficiency of electric vehicles.
However, electric vehicles still come with certain drawbacks. For example, to manufacture electric vehicle batteries, some resources still need to be extracted from the Earth.
Despite this challenge, Storey said that AMDD has been improving upon making batteries that use less rare resources. AI models such as H-CLMP and other more autonomous approaches contribute to the success of AMDD research, allowing for sustainability to be a top priority in new transportation technologies.