Speaker: Silvana Botti
Topic: Machine-Learning-Assisted Design of Materials for Energy
Date: October 16th, 2024 (Wednesday)
Time: 9.00 a.m.
Venue: Room 428, School of Physics and Electronic Engineering
Sponsors: School of Physics and Electronic Engineering, Institute of Science and Technology
Biography:
Prof. Silvana Botti is a full professor for Theory of Excited States of Integrated Solid State Systems at the Ruhr University Bochum, She did her PhD at the University of Pavia in 2002. After her PhD, she was a Marie-Curie Fellow at the University of Paris-Saclay. She was also appointed CNRS Research Scientist there in 2004. In 2008, she moved to University of Lyon where she habilitated in 2010. In 2014, she became a full professor for physics at the University of Jena. Her research group was a member of the European Theoretical Spectroscopy Facility. Since 2023, she has been a full professor for Theory of Excited States of Integrated Solid State Systems at Ruhr University Bochum. Her research focuses on theoretical spectroscopy and the development of first-principles methods for electronic excitations based on (time-dependent) density functional theory and many-body perturbation theory. She authored more than 220 articles (with a total of more than 10000 citations) and has a Hirsch index of 48 (source: Google Scholar). She edited the book First Principles Approaches to Spectroscopic Properties of Complex Materials. She is an associate editor of npj Computational Materials. Botti was part of an international research collaboration on a silicon-based direct bandgap light emitter, which was announced to be the Breakthrough of the Year by Physics World in 2020.
Abstract:
In this presentation, we explore the utilization of supercomputers to design new functional materials with a focus on energy applications. The combination of high throughput techniques and rapidly advancing supercomputers enables the automatic screening of vast numbers of hypothetical materials, providing solutions to current technological challenges. Moreover, the integration of machine learning methods with density-functional theory [1,2] offers a powerful approach to accelerate materials discovery. We summarize our recent efforts in the discovery, characterization, and understanding of inorganic compounds using these innovative approaches, with a specific emphasis on materials for photovoltaics.
While characterizing the electronic properties of crystalline bulk materials is crucial, it may not be sufficient to design electronic devices. Interfaces, such as those found in transistors, light-emitting diodes, and solar cells, play a pivotal role in exploiting quantum processes involving electrons in tailored multilayers. The ability to shape potential gradients at interfaces opens up opportunities for electron manipulation and the development of new functionalities. However, designing interfaces and gaining a deep understanding of their properties present challenges that exceed the current state of the art. We discuss recent advancements in this direction [3,4].
[1] J. Schmidt et al., Machine-Learning-Assisted Determination of the Global Zero-Temperature Phase Diagram of Materials, Adv. Mater. 35, 2210788 (2023).
[2] J. Schmidt, L. Pettersson, C. Verdozzi, S. Botti, and M.A.L. Marques, Crystal graph attention networks for the prediction of stable materials, Sci. Adv.7, eabi7948 (2021).
[3] L. Sun, M.A.L. Marques, and S.Botti, Direct insight into the structure-property relation of interfaces from constrained crystal structure prediction, Nat. Commun, 12, 811 (2021).
[4] T. Rauch, M.A.L. Marques, and S. Botti, Local modified Becke-Johnson exchange-correlation potential for interfaces, surfaces, and two-dimensional materials, J. Chem. Theor. Comput. 16, 2654-2660 (2020).