Machine Learning Develops Super Strong Yet Ultra-Light Materials

AI-created nanomaterials combine the strength of carbon steel with the light weight of Styrofoam.

Machine learning has been employed by researchers at the University of Toronto's Faculty of Applied Science & Engineering to create nano-architected materials that are as strong as carbon steel but as lightweight as Styrofoam.

A group led by Professor Tobin Filleter (MIE) explains how they created nanomaterials with a unique blend of remarkable strength, low weight, and customization in a recent work published in Advanced Materials. Numerous businesses, including the automobile and aerospace sectors, might profit from the strategy. 

It would take more than 100 of them arranged in a row to achieve the thickness of a human hair. Nano-architected materials are composed of small building blocks or repeating units that are a few hundred nanometers in size. These building components are stacked in intricate three-dimensional formations known as nanolattices, and in this instance, they are made of carbon.

The Bayesian optimization machine learning algorithm was used by the KAIST team. In order to forecast the optimal geometries for augmenting stress distribution and boosting the strength-to-weight ratio of nano-architected designs, this algorithm learns from simulated geometries. 

Serles and Filleter collaborated with PhD candidate Jinwook Yeo and Professor Seunghwa Ryu at the Korea Advanced Institute of Science & Technology (KAIST) in Daejeon, South Korea, to create their enhanced materials. The International PhD Clusters program at U of T, which promotes PhD training via research collaboration with foreign partners, was the impetus behind this collaboration.