Materials Informatics for
Materials and Device R&D
Using AI to Create Materials
That the World Has Never Seen Before!
March 4, 2020
Standing in front of the world's most powerful class of atomic resolution electron microscope
Completing a 10-year materials research project in one week. That's the kind of miraculous potential inherent in the field of materials informatics (MI), where Shigetaka Tomiya, R&D Center, Sony Corporation, is conducting his research. This technology allows us to search for new materials by analyzing material development data with artificial intelligence (AI). “In the past, researchers would search for new materials by coming up with hypotheses and verifying them,” explains Tomiya. “In other words, they were predicting results based on factors. However, MI is an approach whereby an AI learns from existing experimental and simulation data to derive new materials with the desired physical properties, or in other words, estimating factors based on data. MI is attracting attention worldwide for greatly reducing development time compared to conventional methods, which tend to rely on experience and instinct.*”
*The four paradigm shifts of materials informatics
In 2009, Jim Gray (an American computer scientist) defined data-driven problem-solving (materials informatics) as the fourth paradigm in his book “The Fourth Paradigm: Data-Intensive Scientific Discovery” (published by Microsoft research).
The materials informatics approach
Tomiya and his team are currently using MI to develop materials for next-generation imaging and sensing devices. With MI being increasingly employed around the world, where do Sony's strengths lie? “Since MI derives results from data, the better the data, the more accurate the results,” explains Raku Shirasawa, R&D Center, Sony Corporation. “By using the proprietary data that Sony has accumulated from previous material development projects, we can perform searches that cannot be replicated elsewhere.”
Yuya Kanitani, R&D Center, Sony Corporation, is currently working on acquiring new experimental data, and here too, Sony's strengths come into play. “A defect on even an atomic level can be fatal for our imaging and sensing devices. To prevent this, we use an electron microscope with the world's highest level of resolution to gather data at the atomic level. Excellent hardware is also one of our strengths.”
As Tomiya-san explains, having high-quality data isn't the only requirement for MI. “To search for new materials, it's important to use the AI correctly, based on a correct understanding of physics and chemistry. We're working with experts in the fields of AI, simulation, and materials analysis for physics and chemistry, and in fact, we discovered a new search method just the other day. Our ability to combine specialized knowledge is another great strength. Moving forward, we will continue using MI to develop new materials for next-generation imaging and sensing devices.”