Led by Liam Donnelly (Heriot-Watt University)
This project will harness machine learning and experimental kinetics to repurpose zirconium catalysts beyond their traditional use in olefin polymerisation.
Although more than 600 zirconocene catalysts have been developed over the past five decades, their applications outside olefin polymerisation remain largely unexplored. This ambitious yet focused study will integrate computational featurisation with operando kinetic measurements to build a predictive machine-learning platform capable of identifying high-performing Zr catalysts for a novel nitrogen-insertion reaction. By screening up to 1,000 virtual zirconocene structures extracted from the Cambridge Structural Database and iteratively refining predictions through Bayesian optimisation, the project establishes a closed-loop digital workflow for catalyst discovery that is transferable to other catalyst families and reaction types.
Crucially, the work directly links digital design to sustainable materials innovation. The optimised catalysts will be applied to the backbone editing of poly(olefin) waste to generate recyclable poly(imine) materials — providing a new strategy for polymer upcycling aligned with circular economy goals. Supported by strong partnerships with CCDC and PSDI for data curation and infrastructure, the project exemplifies the integration of Digital Catalysis, Net Zero Transition, and Advanced Characterisation themes. Beyond delivering proof-of-concept sustainable transformations, it provides a powerful ECR-led model for embedding machine learning, open data, and experimental catalysis into the UK’s evolving digital catalysis ecosystem.
