Superhard materials are of the incredible interest in different handy applications, and an increasing number of research efforts are centered around their turn development.
A superhard material has two crucial features, hardness, and fracture toughness, representing its resistance to deformation and cracks propagation, respectively.
Materials with such properties that would suit specific industry necessities can be discovered computationally using progressed techniques for computational materials science supported by an excellent hypothetical model to figure the desired properties for superhard materials.
A group of Skoltech scientists used machine learning (ML) methods to predict superhard materials based on their crystal structure.
Scientists developed a machine learning model using neural networks on graphs and a recently developed physical model of hardness and fracture toughness. The model is trained using available elastic data from the Materials Project database and has good accuracy for predictions.
Using the model, scientists screened all crystal structures in the database and systematized all the promising hard or superhard materials.
Efim Mazhnik, a Ph.D. student at the Skoltech Center for Energy Science and Technology (Computational Materials Discovery Laboratory), said, “Faced with a lack of experimental data on hardness and fracture toughness to train the models properly, we turned to more abundant data on elastic moduli and predicted their values to obtain the sought-for properties using the physical model we had created earlier.”
Skoltech and MIPT Professor Artem R. Oganov said, “In this study, we applied ML methods to calculate hardness and fracture toughness for over 120,000 crystal structures, both known and hypothetical, most of which have never been explored in terms of these properties. While our model confirms that diamond is the hardest known material, it suggests the existence of several dozen other potentially very hard or superhard materials.”
- Efim Mazhnik et al. Application of machine learning methods for predicting new superhard materials. DOI: 10.1063/5.0012055