High-Throughput ab-initio computing for materials discovery

Nowadays, High-Throughput (HT) ab-initio calculations allow to compute a large quantity of properties for thousands of materials.

Many public databases such as Materials Project, AFLOWlib, NOMAD share materials' structures and properties computed within this approach. The availability of these large datasets has been proven effective in (a) discovering general trends and correlations between structure, chemistry and physical properties;(b) in accelerating the experimental discovery of new materials. Also, it opens the way to machine-learning methods that can be trained on these datasets to predict new materials.

In this talk I will present how we used HT computing to generate a large dataset of transport properties and to search and design high-performance thermoelectrics, p-type transparent conducting materials, and new electrides.