Researchers at the University of Toronto have achieved an exciting breakthrough by combining machine learning with nanoscale engineering. This new material has the potential to change industries such as aerospace and automotive.
Engineers have long sought materials that are both lightweight and strong. Lighter materials can lead to significant fuel savings, especially in aerospace, where every gram counts. Traditional materials like aluminum and titanium have limitations, while carbon fiber, although innovative, has its downsides.
The research team focused on developing nanoarchitected materials. These materials are designed at the nanoscale to achieve high strength while minimizing weight. They draw inspiration from nature’s designs, such as bones and honeycombs. However, creating these structures is challenging. The key is to design geometries that distribute stress evenly to prevent weak points.
To tackle this problem, the researchers used Bayesian optimization, a type of machine learning that helps find the best design from many possibilities. By inputting data from thousands of simulations, they identified the most effective shapes for their carbon nanolattices.
Peter Serles, the lead author of the study, explains that nano-architected materials leverage high-performance designs. These designs benefit from the principle that smaller structures can be stronger. Standard lattice shapes often have sharp corners, creating stress points that can lead to failure. Recognizing this, the research team saw an opportunity for machine learning to help improve material design.
The process began with the algorithm generating thousands of design options. Each design underwent testing in a virtual environment using finite element analysis to predict material behavior under stress. The algorithm then refined its designs to maximize strength and stiffness while minimizing weight.
After narrowing down the optimized designs, the team turned to two-photon polymerization—a precise 3D printing technique—to create the materials. They produced lattices with beams just 300 to 600 nanometers thick. These lattices consisted of 18.75 million cells and measured 6.3 x 6.3 x 3.8 millimeters. To finalize the process, they used pyrolysis to convert the polymer material into glassy carbon by heating it in a nitrogen-rich environment.
The result? The new nanolattices exhibited more than double the strength of previous designs. They were able to withstand stress of 2.03 megapascals per cubic meter per kilogram of density. This means they are over ten times stronger than many lightweight materials, including aluminum alloys and certain carbon fibers, and about five times stronger than titanium.
Serles noted, “This is the first time machine learning has optimized nano-architected materials, and the improvements were surprising. The algorithm didn’t just replicate existing designs; it learned from what worked and what didn’t, allowing it to create entirely new shapes.
What makes these nanolattices so impressive? It turns out that the unique properties of carbon at the nanoscale are crucial. The team found that reducing the carbon beam diameter to 300 nanometers significantly boosted strength due to a phenomenon called the “size effect.” In this case, the arrangement of carbon atoms creates structures that maximize strength. A remarkable 94% of the carbon in the beams was sp²-bonded, known for its exceptional properties.
The potential applications for this lightweight and strong material are vast. It could lead to components in planes, helicopters, and spacecraft, which would improve fuel efficiency and reduce emissions. For instance, if airplane components made from titanium were replaced with this material, fuel savings could reach 80 liters per year for each kilogram of material replaced.
Looking ahead, the research team aims to scale up their designs. They want to make the material viable for larger, cost-effective components and continue innovating to achieve even lower densities while maintaining strength and stiffness.
The findings were published in the journal Advanced Materials.
Check out this related article: “Unlocking the Secrets of Love: Associate Professor Sandra Langeslag Shares Insights in Pittsburgh Post-Gazette” – UMSL Daily
Source link