ENhancing Atomistic ModeLing: Physically Robust Atomistic Machine Learning Models for Predictive Insights into Solvated Chemical Reactions


CALL: 2019

DOMAIN: MS - Materials, Physics and Engineering

FIRST NAME: Alexandre

LAST NAME: Tkatchenko



HOST INSTITUTION: University of Luxembourg

KEYWORDS: Machine Learning, Atomistic Modeling, CO2 reduction, Green Chemical Design

START: 2019-09-01

END: 2020-06-30

WEBSITE: http://www.uni.lu

Submitted Abstract

This proposal will allow Prof. Tkatchenko and his group at the University of Luxembourg to host Prof. Keith of the University of Pittsburgh during a ten-month research visit. Keith, the invited researcher, is among the top scholars in the world in first principles modeling of electrochemical CO2 reduction into fuels and chemicals, and in 2017 he was awarded a highly selective National Science Foundation (NSF) CAREER grant in catalysis and another NSF grant in sustainability engineering. Both projects involve atomistic studies of chemical reaction mechanisms in aqueous solvents to guide the design of new technologies that benefit society. Tkatchenko, the hosting PI, is among the top experts in the world in developing electronic structure and atomistic machine learning models, and he was recently awarded a prestigious ERC Consolidator Grant to develop methods to study quantum fluctuations in complex molecular environments (725291, BeStMo). The quality and impact of Tkatchenko’s and Keith’s respective projects will be dramatically enriched by the two investigators’ synergistic collaboration where key knowledge and technological know-how will be exchanged. Tkatchenko’s robust and accurate methods will be implemented into Keith’s computational models for chemical reactions that are relevant for solar fuels catalysis and green chemical design. Together, Keith and Tkatchenko will introduce new, general, innovative, and transformative computational approaches to physically, predictively, and efficiently model fundamental atomic scale (electro-)chemical reaction mechanisms in solvating environments. This should ultimately guide the design of new technologies for human sustainability such as biomimetic catalyst systems for CO2 conversion and highly effective and biodegradable molecular chelants for environmental sustainability. A secondary objective of this collaboration is to develop a modern textbook that gives a modern perspective of chemical bonding across the periodic table, effective strategies for modeling these bonding interactions with electronic structure and atomistic machine learning methods, and educational examples of how to automate such methods for different applications. Products from both objectives will be disseminated by the two investigators’ groups through publications and research presentations that will reach the superset of international colleagues of the investigators in the physics, chemistry, and engineering communities.

This site uses cookies. By continuing to use this site, you agree to the use of cookies for analytics purposes. Find out more in our Privacy Statement