| Details: |
The traditional way of finding the ideal material for a particular application, i.e., having a set of target properties, has been the trial-and-error approach. This is, however, expensive, time consuming and uncertain, leading to an average time scale of two decades between the discovery of a material and its commercial use. Computational methods have been of great help in studying material properties before they are studied in the laboratory, or tested in the factory. Moreover, over and above studying the materials nature has provided us with, computational methods provide us with the means of studying hitherto non-existent materials.
This has opened up two fascinating possibilities—(1) to screen a set of materials for their suitability in a particular application, and (2) to design new materials with desired target properties. Combinatorial design and high throughput computation (HTC) in conjunction with density functional theory (DFT), and Machine Learning (ML) play an important role in this approach to explore the mapping between the vast compositional and structural landscape and material properties.
Equipped with these tools, we address two different problems of immense current interest and practical importance. The first is to find new hard magnetic materials with large magnetization and anisotropy energy, and the other to find inexpensive, earth-abundant materials that are efficient catalysts for hydrogen generation via water splitting. I will discuss some background for each of these problems, and then present our results. Time permitting, I will conclude by indicating some other ways of targeted materials design we are pursuing.
|