PhD Position on Deep Learning and Molecular Modeling for Optimal Catalyst Materials

Updated: about 2 months ago
Deadline: 31 Dec 2021

Are you interested in developing cutting-edge machine learning methods to design new molecules, materials and chemical processes that help to solve the global warming crisis? We are looking for a PhD candidate to investigate how to combine data from quantum chemical calculations, molecular simulations, and experimental catalysis to 'inverse design' optimal catalyst materials for CO2 conversion.

Using CO2 as a feedstock for the industrial production of chemicals is an appealing approach to prevent greenhouse gas emissions. However, conversion of CO2 into useful chemicals and fuels requires efficient catalysts. Design of catalytic materials and processes is challenging due to the large number of design parameters. The aim of this project is to develop a deep probabilistic programming framework to train models from computational and experimental data and to infer optimal catalyst materials and process conditions, given a set of desired properties such as reaction rates, selectivity, stability, and so forth.

This position is on a collaborative project by computational chemist Dr. Bernd Ensing (HIMS/AI4Science Lab), machine learning experts Prof. dr Max Welling and Prof. dr Jan-Willem van de Meent (IvI, AMLAB), experimental chemist/chemical engineer Dr. Shiju Raveendran (HIMS, Catalysis Engineering), and the UvA Data Science Center.

What are you going to do

You will carry out research in the areas of geometric deep learning, probabilistic programming, and molecular modeling, with applications in heterogeneous catalysis. Through this research you will develop new methods for discovering chemical structure-property relationships, learning features for chemical transformations, generating molecular structures, and meta-learning for sampling chemical space given sparse data.

You will:

  • be part of an exciting multidisciplinary community of people interested in developing and applying data-driven solutions for scientific discovery;
  • develop deep probabilistic methods to generate optimal molecular structures given desired properties and specifications;
  • use quantum chemistry calculations and molecular simulations of catalytic materials and chemical reactions for analysis and data generation;
  • become active in the research community and collaborate with other institutes and/or companies that are part of the project;
  • publish and present work regularly at international conferences, workshops, and journals;
  • assist in teaching activities and in supervising Bachelor and Master students.

View or Apply

Similar Positions