Making strong steel for 3D printing is usually a slow, expensive process. To get a metal that is both tough and flexible, engineers typically have to add expensive components like cobalt or molybdenum. Even then, the parts often have to sit in industrial furnaces for days to reach their full strength, and they still tend to rust easily.

A research team from the University of South China and Purdue University found a way to bypass the trial-and-error phase. Instead of guessing which metals to mix, they used a smart machine-learning model to find the perfect recipe. The result is a new type of steel that is cheaper to make, resists rust, and only takes a few hours to process.

Using AI to Map the Metal

metal
The novel AI-designed ultra-high strength metal; Photo: Yating Luo, Tao Zhu, Cunliang Pan, Xu Ben, Xudong An, Xiaoming Wang and Hongmei Zhu

The team fed the algorithm 81 different physical and chemical traits of elements, things like how big their atoms are or how fast sound moves through them.

Advertisement

The computer suggested a specific blend of iron and chromium mixed with small amounts of budget-friendly elements like silicon, copper, and aluminum. After 3D printing this mix, they baked it for just six hours. The results were exactly what the computer predicted: the steel became 30% stronger and twice as flexible as it was before the heat treatment.

This happened because the short stint in the furnace created tiny “roadblocks” inside the metal. These microscopic particles stop cracks from spreading, while softer pockets of the metal act like shock absorbers to keep it from snapping under pressure.

Fixing the Rust Problem

One of the biggest accomplishments was solving the corrosion issue. Usually, when you make steel this strong, it loses its ability to fight off rust. However, in this new recipe, the copper particles actually help keep the rust-fighting chromium spread out evenly. In saltwater tests, this new alloy held up much better than the standard stainless steel used in many industries today.

This specific AI model works best for the exact type of 3D printing the researchers used. If they want to use it for different manufacturing methods, they’ll have to adjust the data.