AI for catalyst discovery

In this project, we will develop innovative AI tools to speed up the process of catalyst discovery, in particular in the domain of renewable energy. Specific catalysts are essential to applications such as the efficient synthesis of solar fuels and fertilizer. However, many known catalysts are suboptimal in their efficacy or require scarce elements. Currently, catalyst development is held back by the computational cost of the quantum chemistry simulations used to guide experimentation, as well as the very high-dimensional search space of potential catalysts. AI offers the opportunity to significantly accelerate the discovery of catalysts, both by performing fast approximate simulations to assess the efficacy of individual catalysts and by intelligently searching for optima in the space of candidate materials. Specifically, we will build AI algorithms that operate on the graphical structure of chemical compounds, incorporating chemistry-related constraints and integrating into larger AI-based materials discovery pipelines.

Faculty Supervisor:

Yoshua Bengio

Student:

Partner:

Inria Saclay - Île-de-France Research Centre

Discipline:

Computer science

Sector:

Artificial Intelligence; Green/Alternative Energy; Achieving Net Zero; Quantum Science

University:

Université de Montréal

Program: