High-performance, nano-architected materials

Researchers at the University of Toronto’s Faculty of Applied Science & Engineering have used machine learning (ML) to design nano-architected materials with the strength of carbon steel but the lightness of Styrofoam.

In a paper published in Advanced Materials, a team led by Professor Tobin Filleter, MIE, describes how they made nanomaterials combining exceptional strength, light weight, and customizability.

“Nano-architected materials combine high performance shapes, like making a bridge out of triangles, to achieve some of the highest strength-to-weight and stiffness-to-weight ratios of any material,” says Peter Serles, Ph.D., the paper’s first author.

“However, the standard lattice shapes and geometries used tend to have sharp intersections and corners, which leads to the problem of stress concentrations. This results in early local failure and breakage of the materials, limiting their overall potential.”

Nano-architected materials are made of tiny repeating units measuring a few hundred nanometers – it would take more than 100 of them in a row to reach the thickness of a human hair. The carbon building blocks are arranged in 3D structures called nanolattices.

Serles and Filleter worked with Prof. Seunghwa Ryu and Ph.D. student Jinwook Yeo at the Korea Advanced Institute of Science & Technology (KAIST) in Daejeon, South Korea to design the material.

The KAIST team used the multi-objective Bayesian optimization ML algorithm to predict the best geometries for enhancing stress distribution and improving the strength-to-weight ratio of nano-architected designs. Serles then used a two-photon polymerization 3D printer to create prototypes. The optimized nanolattices created more than doubled the strength of existing designs, withstanding a stress of 2.03MPa for every cubic meter per kilogram of its density, about 5x higher than titanium.

“This is the first time machine learning has been applied to optimize nano-architected materials, and we were shocked by the improvements,” Serles says. “It learned from what changes to the shapes worked and what didn’t, enabling it to predict entirely new lattice geometries.”

ML is normally very data intensive, but the multi-objective Bayesian optimization algorithm only needed 400 data points instead of 20,000 or more. “We were able to work with a much smaller but an extremely high-quality data set,” says Serles, who is now a Schmidt Science Fellow at the California Institute of Technology (Caltech).

“We hope these new material designs will lead to ultra-lightweight components in aerospace applications that can reduce fuel demands while maintaining safety and performance,” Filleter says.

“If you were to replace components made of titanium on a plane with this material, you’d be looking at fuel savings of 80L per year for every kilogram of material you replace,” Serles adds.

Other contributors to the project included collaborators from Karlsruhe Institute of Technology (KIT) in Germany, Massachusetts Institute of Technology (MIT), and Rice University.

University of Toronto Faculty of Applied Science & Engineering
https://www.engineering.utoronto.ca/