Atomic Design of Superstrong Materials

A new computer model could significantly advance nanoscale engineering.

Researchers have learned how to design the nanoscale features of materials to make them four times stronger without making them brittle. The new insight is the result of a significant improvement to an existing computer model that allowed researchers to, for the first time, simulate the complex mechanical behavior of nanostructures in metals.  

The work, described in the Proceedings of the National Academy of Sciences by researchers at MIT, Ohio State, and the Georgia Institute of Technology, could lead to more-durable materials for gears in microscale machines. It could also lead to coatings that dramatically improve the performance of larger-scale structures, such as metal plating on artificial joints, says Subra Suresh, professor of materials science and engineering at MIT.

The advance is part of a larger ongoing effort to use software to discover new materials that would be impossible or impractical to discover using experiments alone. Computers make it possible to fine-tune and quickly compare various material parameters, such as the ratios and locations of the elements used and their crystal structures. Such techniques have already suggested new combinations of elements for high-performance battery electrodes, for example.

Simulations of the electrical and optical properties, however, have been more successful than models of materials' mechanical behavior. That's because the computations are so involved that even the fastest supercomputers can only simulate the stretching, compressing, and cracking of materials for very short time periods--on the order of a billionth of a second. That's not nearly long enough to test the models against measurements gathered in real-world experiments, which take place over several seconds, minutes, or even hours.

The MIT researchers based their new model, which can work on these longer time scales, on one invented in 2000 in which the occurrence of rare events is used to extrapolate from short time scales to seconds or minutes. All that's needed is a good idea of how often these events take place, as well as how long they take. The method, however, would not work when, as is the case with some nanostructured materials, the "landscape is very complicated," says Efthimios Kaxiras, professor of physics and applied physics at Harvard.

To overcome this problem, the researchers modified the method so that it can be used to model materials in which the duration of these rare events isn't known, or in which the events take place over long time scales. "It's a nice modification that gives you additional freedom if you don't know exactly where you're going to end up," Kaxiras says.

"It gives us the ability to link nanoscale features to the mechanical properties of materials," Suresh says.

The researchers used the model to understand for the first time the way that nanostructures in copper can make the material three to four times stronger without the usual trade-off of making the material more brittle. What they found could help them optimize the properties of copper, Suresh says.