James Kermode

I develop multiscale materials modelling algorithms and the software that implements them, with a particular focus on machine learning and data-driven approaches, and on quantifying the uncertainty in the output of electronic structure and atomistic models. I am also active in applying parameter-free modelling techniques to make quantitative predictions of "chemomechanical" materials failure processes where stress and chemistry are tightly coupled, e.g. near the tip of a propagating crack (left), where local bond-breaking chemistry is driven by long-range stress fields. Prominent examples include: QM-accurate modelling of dislocations. Combining a quantum mechanical description of dislocation cores with an interatomic potential to capture long-range elastic relaxation allowed us to investigate plasticity in nickel-based superalloys and interactions between dislocations in tungsten with plasma components such as hydrogen (upper right, with tungsten atoms red, green and blue, hydrogen impurity purple and dislocation core and glide path red). Machine Learning a General Purpose Interatomic Potential for Silicon. Albert Bartók-Pártay, Noam Bernstein and Gábor Csányi and I created a data-driven potential for silicon using the Gaussian approximation potential (GAP) framework. Our model, published in Physical Review X, is capable of accurately describing its behaviour across a wide range of temperature and pressures, and it comes along with uncertainty estimates that help estimate where it risks straying outside its domain of applicability (right, near a crack tip on the (111) cleavage plane, where uncertainty highest on red atoms). Scattering of cracks by individual atomic-scale impurities. In an article published in Nature Communications, we showed that a single atomic defect (e.g. a boron dopant, coloured orange in movie, above right) can be enough to deflect a crack as it travels through a crystal, leading to macroscopically observable surface features (lower right).

P. Grigorev, A. M. Goryaeva, M.-C. Marinica, J. R. Kermode, and T. D. Swinburne, Calculation of Dislocation Binding to Helium-Vacancy Defects in Tungsten Using Hybrid Ab Initio-Machine Learning Methods, Acta Mater. 247 118734 (2023) [arXiv:2111.11262] L. Zhang, B. Onat, G. Dusson, A. McSloy, G. Anand, R. J. Maurer, C. Ortner, and J. R. Kermode, Equivariant Analytical Mapping of First Principles Hamiltonians to Accurate and Transferable Materials Models, npj Comp. Mater. 8, 158 (2022) [arXiv:2111.13736]