An up to 2 year postdoc position is available to support a collaboration between the University of Toronto and the National Institute of Standards and Technology. The position focuses on building machine learning models to facilitate the design of oxidation resistant high entropy alloys.

Background: Corrosion has broad societal and economic impacts. Despite the US spending ~2.6% of its GDP every year in corrosion remediation and mitigation, incidents such as those that took place at the Surfside condominium tower are far too commonplace.  Recently a new class of metal alloys, so-called high entropy alloys (HEAs) with 5-principal alloying components or more, has been discovered that potentially have superior mechanical and corrosion properties in comparison to traditional alloys.  More recently, studies have shown that their superior oxidation resistance is correlated with the formation of multi-cation oxides not typically associated with passivation behavior. However, the search for HEAs that promote the growth of multi-cation oxides with superior oxidation resistance requires the rational and systematic search of 5+ component alloys systems from a palette of ~70 elements, which would be too time consuming to do using traditional single sample or even combinatorial experiments.  The candidate selected will train machine learning models on calculated properties such as alloy formation energy and defect energies that will be used to screen new potential alloys for their ability to resist rapid oxidation.

Methodology: Machine learning surrogate models will be built to rapidly screen candidate materials with multiple elements, which then can be validated using conventional methods such as density functional theory (DFT), molecular dynamics (MD) and Monte-Carlo. Property-specific models using graph neural networks (GNN) providing state of the art accuracies will be used for pre-screening of multi-component alloy systems. Detailed calculations for defect-energetics and reaction pathways will be carried out with DFT/MD. The postdoc will work with students and researchers at University of Toronto, Canada and National Institute of Standards and Technology, MD, USA with computational facilities from these institutions to expand the workflow for alloy design. The computational workflow will be based on JARVIS infrastructure (https://jarvis-tools.readthedocs.io/, https://jarvis.nist.gov/). The candidate materials thus found will be synthesized and characterized by collaborators at the University of Toronto.

Candidate profile: The preferred candidate is a self-motivated and curious PhD holder in the fields of condensed matter physics, materials science, or metallurgy.  They should have prior experience in performing DFT/MD simulations and a strong interest in learning about and applying machine learning algorithms to the discovery of new materials. Ideally the candidate has experience programming in Python, R, or Julia and experience working in multi-disciplinary teams consisting of theorists and experimentalists.

How to apply:

Please send your application materials (detailed CV, publication list, motivation letter, and names and contact email of at least two references) by email to the supervisors with subject of email as "Postdoc application":

Jason Hattrick Simpers ([email protected])

Kamal Choudhary ([email protected])