Learning to react

A new paper published in the Nature Research family marks a strong result for the department’s work on artificial intelligence and materials science – and highlights the hands-on training of its postgraduate researchers.

The study, published on 31 May in Nature Computational Materials, is part of PhD research by Zheng Jie Liew, funded through the Marie Skłodowska-Curie CINEMA training network. He was supported by Ziad Elkhaiary, who contributed to the project while completing the department’s Advanced Chemical Engineering (ACE) Master’s, under the supervision of Professor Alexei A. Lapkin in the Sustainable Reaction Engineering research group.

Their paper – ‘A multi-model vision assistant for autonomous interpretation of polymer–solvent solvation behaviours’ – presents a new AI system that uses computer vision and language processing to interpret complex polymer–solvent interactions such as swelling, gelation and dispersion from images and videos.

Polymer–solvent systems are notoriously tricky to analyse due to the variety of behaviours involved and the subjective nature of manual assessments. This new approach integrates multiple AI models – including convolutional neural networks to understand static and dynamic visual data and a vision–language module that generates descriptive captions – providing an objective, scalable way to track and describe solvation phenomena.

“Polymers and solvents don’t always behave predictably and human evaluations can vary,” said Liew. “Our AI assistant can see what’s happening in detail and put it into words, making it easier to analyse data quickly and reliably – especially for high-throughput experiments.”

This system promises to accelerate materials discovery by enabling automated, repeatable interpretation of experimental results, removing bottlenecks caused by manual screening.

Elkhaiary’s contribution during his ACE Master’s highlights the research-led teaching ethos of the department.

“It’s rewarding to see our students actively shaping cutting-edge science,” said Lapkin. “Contributing to real projects prepares them for the challenges of sustainable chemical engineering.”

"Our AI assistant can see what’s happening in detail and put it into words, making it easier to analyse data quickly and reliably"

- Zheng Jie Liew -

PhD student at CEB