Resnick Institute


Accelerating Sustainable Inorganic Design with Machine Learning

Chemical space is vast, with best estimates suggesting we have as yet characterized a tiny fraction of all possible compounds. The need for efficient discovery of new materials and catalysts to solve outstanding challenges in sustainable energy and resource utilization mandates that we identify smart ways to map out and explore chemical space.

Over the past twenty years, computational high-throughput screening, typically driven by density functional theory (DFT), has cemented itself as a powerful tool for the discovery of new materials. I will describe our recent efforts to develop new open-source software capable of both leveraging accelerated DFT on novel architectures (i.e., graphical processing units) and in moving beyond DFT to enable exploration of vast chemical space.

I will first describe our efforts to accelerate DFT-driven discovery and chemical space exploration with our divide-and-conquer approach to precise inorganic complex generation from libraries of millions of possible realistic fragments. I will then explain how we have trained machine learning (ML) models to predict inherently quantum mechanical properties in inorganic catalysts and materials such as spin-state ordering, redox potential, or even geometry in seconds instead of hours. Such an approach enables rapid discovery of leads from thousands of candidates in a manner that remains intractable with DFT alone.

I will describe how these developments have advanced design principles for earth-abundant single-site catalysts, enabled sustainable materials synthesis, and revealed new unconventional functional materials for spintronics and sensing. I will close with our outlook on the promise of artificial intelligence to enable discovery of essential materials and catalysts for sustainable energy utilization.

About Heather

Heather J. Kulik is an assistant professor in the department of chemical engineering at MIT. She obtained her BE in chemical engineering from the Cooper Union for the Advancement of Science and Art. She obtained her PhD from the department of materials science and engineering at MIT working with Nicola Marzari on density functional theory method development for transition metal chemistry.

She then completed postdoctoral training at Lawrence Livermore National Lab with Felice Lightstone on biomimetic catalyst design and Stanford University with Todd Martínez on the large-scale electronic structure of biomolecules.

Since beginning her independent career at MIT, she has received several awards including the Office of Naval Research Young Investigator Award, the DARPA Young Faculty Award, the ACS Open Eye Outstanding Junior Faculty Award in Computational Chemistry, the ACS Industrial & Engineering Chemistry Research “2017 Class of Influential Researchers”, and a Burroughs Wellcome Fund Career Award at the Scientific Interface.