Materials Informatics Technical Networking

Professor Keith Butler

Associate Professor of Computational Materials Chemistry, UCL

Keith got his PhD in computational chemistry from UCL in 2010. He worked as a PDRA at the University of Sheffield and the University of Bath, using atomisitic simulations design new photovoltaics. In 2018 he moved the the Rutherford Appleton Laboratory where he was involved in setting up the Scientific Machine Learning team. In 2022 Keith moved to Queen May and in 2023 returned to UCL. His group is interested in using machine learning and atomistic simulation to accelerate materials discovery and characterisation.

 

Abstract: Deep learning for sustainable materials design

The discovery and design of new materials is critical for advancing carbon-emission reducing technologies such as renewable energy and electric vehicles. Experimental discovery of new materials is typically slow and costly, quantum mechanics (QM) calculations have brought computational materials design within reach. However, QM calculations are often limited to relatively small sets of materials, as their computational costs are too great for large-scale screening, this is the case for calculating properties required for new energy materials. New methods in machine learning (ML) and deep learning (DL) have emerged as a powerful complementary tool to QM calculations – learning rules from data calculated from QM and applying cheap, efficient models to explore large chemical spaces. However, several challenges still exist for example, learning from small and limited datasets, obtaining measures of confidence in models and understanding the results of DL models. All these challenges must be addressed to fully realise the power of DL for design of new sustainable materials. In this talk I will give examples of recent work in our group to address these issues, including using unsupervised learning to accelerate the characterisation of battery materials without requiring labelled data[1], building models with reliable uncertainty quantification [2], capable of learning on significantly smaller datasets than regular DL models and using DL to match experimental and simulated data [3]. Finally, I will also discuss how the latest exciting developments in large language models could help to solve the challenges of crystal structure prediction [4].
1 Versatile domain mapping of scanning electron nanobeam diffraction datasets utilising variational autoencoders npj Computational Materials 9 (1), 14, 2023
2 Entropy-based active learning of graph neural network surrogate models for materials properties The Journal of Chemical Physics 155 (17), 174116, 2021
3 Using generative adversarial networks to match experimental and simulated inelastic neutron scattering data Digital Discovery 2.578, 2023
4 Crystal Structure Generation with Autoregressive Large Language Modeling arXiv preprint arXiv:2307.04340 2023

Dr Enrique Galindo-Nava FIMMM

Associate professor, UCL

Enrique has over 65 publications, is recipient of numerous awards -including a RAEng Research Fellowship and FIMMM- and has been Keynote speaker at several international conferences. His research covers 3 areas, processing of advanced metallics, computational material design for resource efficiency, and novel materials for the Hydrogen economy -including designing against H embrittlement and material degradation following H combustion. He actively collaborates with industry and has helped partners to deliver sustainable material and manufacturing solutions for energy and transport technologies

 

Abstract:     Novel approaches to predict process-microstructure-property links in heterogeneous alloys

Modern metallic alloys have been developed with significant microstructural heterogeneities demanding new predictive computational tools to understand and prevent processing-related issues, as well as to optimise their in-service performance. This talk will give an overview of recent efforts to study heterogeneous alloys during their manufacturing using novel computational and experimental approaches. We will present two examples where various models for microstructure evolution have been developed and combined with laboratory-scale experiments and Machine Learning to “down-scale” critical industrial manufacturing challenges bringing new insights of complex material behaviour. The first application is on the optimisation of Hydrogen desorption processing -or “bake-out”- in large steel components for nuclear applications. The second example covers predicting microstructure and mechanisms of surface cracking during hot die forging of Ni-based superalloys for reduced waste in aerospace applications. The results deliver cutting-edge strategies to address alloy heterogeneities during manufacturing and develop new materials with improved mechanical performance.

Dr Sergio Gonzalez Sanchez

Senior Lecturer in Mechanical Engineering, Northumbria University

Dr. Gonzalez is Senior Lecturer in Mechanical Engineering. He received his PhD in Materials Physics at Complutense University of Madrid and PhD research work was carried out at the National Centre for Metallurgical Research (CENIM-CSIC) in Spain.

His research is focused on metallurgy, rapid solidification and processing-structure-property relationships, he has received funding from EPSRC, Royal Society, Innovate UK, IOP, Fellowships from Spain (Juan de la Cierva and FPI to pursue his PhD), funding from Japan (internal funding from WPI), Marie Curie travel grant and SRUK Short Term Mobility Fellowship.

He is fellow member of IoM3 (FIMMM), chartered scientist, chartered engineer, full member of EPSRC peer reviewer college, Fellow of the higher education academy (FHEA), British Engineering Excellence Award Finalist and invited editorial board member of numerous journals.