For engineers developing new materials or protective coatings, there are billions of different possibilities. Lab tests or detailed computer simulations to determine their exact properties, such as toughness, can take hours, days, or more for each variation. Researchers have developed a new artificial intelligence-based approach that could reduce that to a matter of milliseconds, making it practical to screen vast arrays of candidate materials. The system could be used to develop stronger protective coatings or structural materials; for example, to protect aircraft or spacecraft from impacts.
The focus of this work was on predicting the way a material would break or fracture by analyzing the propagation of cracks through the material’s molecular structure. Molecular dynamics simulations provide a chemically accurate description of how fracturing happens but it is slow because it requires solving equations of motion for every single atom.
Artificial intelligence (AI) systems that include machine learning need a variety of examples to use as a training set to learn about the correlations between the material’s characteristics and its performance. The researchers looked at a variety of composite, layered coatings made of crystalline materials. The variables included the composition of the layers and the relative orientations of their orderly crystal structures as well as the way those materials each responded to fracturing, based on the molecular dynamics simulations.
They generated hundreds of such simulations, with a wide variety of structures, and subjected each one to many different simulated fractures. Then they fed large amounts of data about all these simulations into their AI system to see if it could discover the underlying physical principles and predict the performance of a new material that was not part of the training set— and it did.
For single simulations in molecular dynamics, it has taken several hours to run the simulations but in the artificial intelligence prediction, it only takes 10 milliseconds to go through all the predictions from the patterns and show how a crack forms step-by-step.
Although the method was only applied to one material with different crystal orientations, it could be applied to much more complex materials. And while the team used data from atomistic simulations, the system could also be used to make predictions on the basis of experimental data such as images of a material undergoing fracturing. The system could be applied not just to fracturing but to a wide variety of processes unfolding over time such as diffusion of one material into another or corrosion processes.