Big-data analytics for materials science: search for new topological insulators

• Grant number: 16/04496-9
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: June 01, 2016
• End date: May 31, 2017
• Field of knowledge: Physical Sciences and Mathematics - Physics - Condensed Matter Physics
• Principal Investigator: Adalberto Fazzio
• Grantee: Carlos Augusto Mera Acosta
• Supervisor: Matthias Scheffler
• Host Institution: Instituto de Física (IF). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Institution abroad: Max Planck Society, Berlin, Germany
• Associated to the scholarship: 14/12357-3 - Tansistor based on spintronics: topological characterization and ballistic transport in topological insulators, BP.DR

Abstract

A main basic-science objective of spintronics is to understand the mechanisms by which it is possible to achieve efficient control of both spin configurations and spin currents. The generation of spin currents, spin injections and spin conservation is mediated by the Spin-Orbit Coupling (SOC) mainly via the Rashba effect and/or the topological effects. Therefore, the search for systems with these properties is a primary concern for spintronics developments. Study the transport properties of topological insulators and proposing systems whose electronic transport properties come from the metallic topological states is the main objective of the current doctoral work. The usual approaches to predict topological insulators (TI's) start with a trial-and-error learning process to find new materials and then, employing DFT calculations to verify if this material satisfy the required properties. These calculations have a high computational cost, and hence, the trial-and-error learning process is not feasible. Since the amount of calculated data to find systems with a specific property increase exponentially with time, the "big-data of materials science" is a good strategy to overcome the problem of computational cost and to build predictive models. This statistical learning focuses on finding the actuating mechanisms of a certain property or function and describing it in terms of a set of physically meaningful parameters (henceforth termed descriptor). Professor Dr. Matthias Scheffler put forward the requirements for a suitable descriptor and demonstrated how a descriptor can be found systematically. The main purpose of a visit to the Scheffler's group of the Fritz Haber Institute of the Max Planck Society is finding suitable descriptors for TI based on the compressed sensing concepts to construct predictive models for these systems. We will stared to perform DFT calculations for trivial and non-trivial TI's, before we will construct possibles descriptor for TI's and finally we will propose a predictive model for TI's with the suitable transport properties to construct a transistor based on spintronic. We believe this work will open a new area of research in the physics of materials, contributing to the discovery of the atomic variables that determine the topological properties of a system and allow find new materials with nontrivial topological phases that can be integrated with current technology. (AU)