Researchers at the University of Toronto’s Faculty of Applied Science & Engineering have created an innovative class of nano-architected materials that are as strong as carbon steel yet as light as Styrofoam. Published in Advanced Materials, the research, led by Professor Tobin Filleter, highlights how machine learning was used to design nanomaterials with remarkable properties—high strength, low weight, and the ability to be customized for various applications. This breakthrough could revolutionize industries such as automotive and aerospace, where materials must balance strength and lightness.

Nano-architected materials, composed of tiny repeating units just a few hundred nanometers in size, are structured into complex 3D shapes known as nanolattices. These materials take advantage of the “smaller is stronger” principle, where nanoscale designs achieve superior strength-to-weight and stiffness-to-weight ratios compared to conventional materials. However, traditional lattice shapes often have sharp intersections and corners, creating stress concentrations that lead to premature failure.

Peter Serles, the first author of the paper, recognized this challenge and realized that machine learning could offer a solution. “As I thought about this challenge, I realized that it is a perfect problem for machine learning to tackle,” says Serles.

In collaboration with Professor Seunghwa Ryu and Ph.D. student Jinwook Yeo from the Korea Advanced Institute of Science & Technology (KAIST), the team employed a cutting-edge machine learning algorithm known as multi-objective Bayesian optimization. This algorithm was trained on simulated geometries, allowing it to predict the best lattice designs that would optimize stress distribution and maximize the strength-to-weight ratio.

“Machine learning is normally very data intensive, and it’s difficult to generate a lot of data when using high-quality data from finite element analysis,” Serles explains. “But the multi-objective Bayesian optimization algorithm only needed 400 data points, whereas other algorithms might need 20,000 or more. This allowed us to work with a much smaller but extremely high-quality dataset.”

Once the optimal lattice designs were identified, Serles and his team used a two-photon polymerization 3D printer at the University of Toronto’s Center for Research and Application in Fluidic Technologies (CRAFT) to create the prototypes. This advanced additive manufacturing technology enables 3D printing at the micro and nanoscale, allowing for the creation of highly optimized carbon nanolattices.

The results were groundbreaking. The new nanolattices more than doubled the strength of existing designs, withstanding a stress of 2.03 megapascals for every cubic meter per kilogram of its density—about five times the strength of titanium. This level of strength, combined with its lightweight nature, makes the material ideal for applications where both durability and low weight are crucial.

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

The team envisions a wide range of applications for these materials, particularly in aerospace. “We hope these new material designs will lead to ultra-lightweight components in aerospace applications, such as planes, helicopters, and spacecraft,” says Filleter. “This could reduce fuel demands during flight while maintaining safety and performance, ultimately helping to reduce the high carbon footprint of flying.”

For example, replacing titanium components with this new material in a plane could save up to 80 liters of fuel per year for every kilogram of material replaced.

This work brought together experts in material science, machine learning, chemistry, and mechanics, and included contributions from professors Yu Zou, Chandra Veer Singh, Jane Howe, and Charles Jia at the University of Toronto, as well as international collaborators from the Karlsruhe Institute of Technology (KIT) in Germany, MIT, and Rice University in the United States.

“We’re looking ahead to scaling these material designs to create cost-effective macroscale components,” Filleter notes. “In addition, we will continue to explore new designs that push the material architectures to even lower density while maintaining high strength and stiffness.”

With further development, these machine learning-optimized nano-architected materials could transform industries that rely on lightweight yet strong materials, offering a more sustainable and efficient path forward in fields like aerospace, automotive manufacturing, and beyond.

By Impact Lab