Summary
Researchers at Cardiff University are combining high-performance computing, artificial intelligence and advanced microscopy to improve our understanding of catalysts — materials that are fundamental to modern chemical processes and have an important role to play in addressing challenges including decarbonisation, carbon dioxide utilisation and the circular economy.
The work brings together experimental data, detailed molecular simulation and emerging machine-learning techniques to help researchers understand how catalysts behave and support the development of more efficient future materials and processes.
The research challenge
Catalysts reduce the energy required and waste produced by chemical processes. More than 90% of chemical processes involve at least one catalyst during their lifecycle, making improved catalyst design important both scientifically and economically.
Researchers at the Cardiff Catalysis Institute (CCI) study catalysts at very small scales, including through high-resolution electron microscopy. Modern microscopy can generate terabytes of experimental data, creating a substantial computational challenge alongside the simulations needed to understand atomic structures and reaction mechanisms.
Understanding these materials under realistic reaction conditions increasingly requires experimental science, advanced simulation and data-driven techniques to work together.
How advanced computing helps
High-performance computing enables researchers to process large experimental datasets and carry out computationally demanding simulations of atomic structures and chemical reactions.
GPU acceleration and machine learning offer further opportunities. AI-based techniques can help analyse electron microscopy images rapidly, potentially providing useful feedback while experiments are being carried out. Machine-learning approaches can also complement traditional molecular simulation, helping researchers explore more efficiently.
This combination of simulation, data analysis and AI illustrates the increasingly close relationship between traditional high-performance computing and emerging AI research.
Results, impact and scientific excellence
CCI Researchers have already demonstrated the potential of these approaches across several areas.
Published research includes the development of a neural-network pipeline for automated analysis of morphologically diverse catalyst systems, alongside recent work using computational modelling and machine-learning techniques to understand materials and catalytic reactions.
The wider research activity has contributed to publications in journals including Nature, ACS Catalysis, Digital Discovery and the Journal of Chemical Theory and Computation.
By enabling researchers to analyse increasingly complex experimental data and model chemical processes in greater detail, advanced research computing can help accelerate the search for new catalysts and support research addressing major challenges in energy use, sustainability and the circular economy.
Research area
Computational Chemistry · Artificial Intelligence · Materials Science · Net Zero
Technical summary
Research computing: High-performance CPU and GPU computing
Techniques: Molecular simulation, machine learning, high-resolution electron microscopy data analysis
Data scale: Electron microscopy workflows can generate terabytes of data
Research team
The work involves researchers within Cardiff University’s School of Chemistry and the Cardiff Catalysis Institute, bringing together expertise in computational chemistry, catalysis, materials simulation and electron microscopy.
- Prof. Sir Richard Catlow
Focuses on computational techniques for complex materials and catalysts, with a 50-year career in simulation. - Dr. Alberto Roldan Martinez
Reader in Computational Chemistry and Catalysis, working on surface science and materials simulation. - Prof. David Willock
Expert in computer simulations of catalysts and related materials. - Dr. Andrew Logsdail
Specializes in developing computational models for functional materials and catalyst synthesis. - Dr. Stefano Leoni
Focuses on theoretical and computational chemistry, including solid-state materials. - Dr Thomas Slater
An electron microscopist with a focus on the characterisation of nanomaterials and heterogeneous catalyst.
Further information
- https://www.cardiff.ac.uk/campus-developments/projects/translational-research-hub
- Treder, K.P., Huang, C., Bell, C.G. et al. “nNPipe: a neural network pipeline for automated analysis of morphologically diverse catalyst systems”. npj Comput Mater 9 (18) (2023). https://doi.org/10.1038/s41524-022-00949-7
- Xie, J., Fu, C., Quesne, M.G. et al., “Methane oxidation to ethanol by a molecular junction photocatalyst”. Nature 639 (2025), 368–374. https://doi.org/10.1038/s41586-025-08630-x
- Xu Li, Guodong Qi, Richard J. Lewis, Mark J. Howard, Barry A. Murrer, Brian Harrison, David J. Morgan, Christopher J. Kiely, Qian He, Thomas E. Davies, David J. Willock, Michael Bowker, Jingxian Cao, Feng Deng, Jun Xu, and Graham J. Hutchings, “Partial Oxidation of Methane to Acetic Acid with Oxygen Using AuPd/ZSM-5”. ACS Catalysis 15 (21) (2025), 18663-18674. DOI: 10.1021/acscatal.5c03918
- Matthew Lindley, Pavel Stishenko, James W. M. Crawley, Fred Tinkamanyire, Matthew Smith, James Paterson, Mark Peacock, Zhuoran Xu, Christopher Hardacre, Alex S. Walton, Andrew J. Logsdail, and Sarah J. Haigh, “Tuning the Size of TiO2-Supported Co Nanoparticle Fischer–Tropsch Catalysts Using Mn Additions”. ACS Catalysis 14 (14) (2024), 10648-10657 DOI: 10.1021/acscatal.4c02721
- Amit Chaudhari, Kushagra Agrawal and Andrew J. Logsdail, “Machine learning generalised DFT+U projectors in a numerical atom-centred orbital framework”, Digital Discovery, 4 (2025), 3701-3727, https://dx.doi.org/10.1039/d5dd00292c
- Alexandre Boucher, Cameron Beevers, Bertrand Gauthier, and Alberto Roldan, “Machine Learning Force Field for Optimization of Isolated and Supported Transition Metal Particles”, Journal of Chemical Theory and Computation 21 (5) (2025), 2626-2637 DOI: 10.1021/acs.jctc.4c01606