The Chemistry Seminar Program brings visitors from across Canada and beyond to present seminars in their area of expertise.  The seminars are open to everyone, including undergrad students.

Thursday October 13
11:30 am
David Turpin Building A110

Oleksandr Voznyy
University of Toronto, Scarborough
http://www.mapageweb.umontreal.ca/waldronk/Files/Home.html

“Applications of machine learning for studying phase stability in energy materials”

There is a growing demand for design and discovery of new materials that can replace their costly counterparts without the loss of performance. Focusing on clean energy applications, our experimental efforts show that in catalysis, batteries and photovoltaics, even minor changes to the material composition (using alloying or doping) may bring substantial benefits to materials’ performance.  However, it is becoming increasingly evident that traditional trial and error methods are simply too inefficient, given the vastness of the chemical combinatorial space, where human intuition is often inadequate to capture the trends in materials properties.  We find that having one tool that could predict formation energies and bandgaps for arbitrary materials, provided the accuracy is adequate, could help address a host of problems in a vast range of clean energy applications.  In this talk I will present our efforts in developing such a tool and the advances we’ve made in improving the accuracy of such predictions (in particular, formation energies). I will also demonstrate how this simple tool can be used for predicting battery cathode voltages, solid electrolytes ionic conductivity, solar cells phase stability, and catalysts’ stability in acid.