Automated Toolkits and Machine-Learning Acceleration for Defect Simulations
First-principles simulations of atomic and electronic structure in solids offer a powerful route to predict and understand material properties. (1) This is particularly relevant in the case of point defects which dramatically affect material properties yet present many challenges for experimental characterisation.
Recent years have seen significant advances in both computational methodologies (2–4) and associated toolkits (5–9) for modelling defect behaviour. I will discuss our collaborative efforts in this area; including the continued development of the open-source doped defect simulation package5, approaches for exploring defect energy surfaces (including MLIP accelerations) (10,11) and remaining challenges in this area. (12)
1. Freysoldt, C. et al.First-principles calculations for point defects in solids. Rev. Mod. Phys. 86, 253–305 (2014).
2. Arrigoni, M. & Madsen, G. K. H. Evolutionary computing and machine learning for discovering of low-energy defect configurations. npj Comput Mater 7, 1–13 (2021).
3. Alkauskas, A., Yan, Q. & Van de Walle, C. G. First-principles theory of nonradiative carrier capture via multiphonon emission. Phys. Rev. B 90, 075202 (2014).
4. Mosquera-Lois, I., Kavanagh, S. R., Walsh, A. & Scanlon, D. O. Identifying the ground state structures of point defects in solids. npj Comput Mater 9, 1–11 (2023).
5. Kavanagh, S. R. et al. doped: Python toolkit for robust and repeatable charged defect supercell calculations. Journal of Open Source Software 9, 6433 (2024).
6. Mosquera-Lois, I., Kavanagh, S. R., Walsh, A. & Scanlon, D. O. ShakeNBreak: Navigating the defect configurational landscape. Journal of Open Source Software 7, 4817 (2022).
7. Kim, S., Hood, S. N., Gerwen, P. van, Whalley, L. D. & Walsh, A. CarrierCapture.jl: Anharmonic Carrier Capture. Journal of Open Source Software 5, 2102 (2020).
8. Turiansky, M. E. et al. Nonrad: Computing nonradiative capture coefficients from first principles. Computer Physics Communications 267, 108056 (2021).
9. Zhu, B., Kavanagh, S. R. & Scanlon, D. easyunfold: A Python package for unfolding electronic band structures. Journal of Open Source Software 9, 5974 (2024).
10. Kavanagh, S. R. Identifying split vacancy defects with machine-learned foundation models and electrostatics. J. Phys. Energy 7, 045002 (2025).
11. Kasoar, E., Hart, J., Batatia, I., Csányi, G. & Elena, A. M. ML-PEG (Machine Learning Performance and Extrapolation Guide). GitHub Repository (2025).
12. Mannodi-Kanakkithodi, A., Huang, M., Gorai, P. & Kavanagh, S. R. Accelerating point-defect simulations using data-driven and machine learning approaches. MRS Bulletin 51, 600–614 (2026).
Asst Prof in Simulation of Energy Materials
Dr. Seán Kavanagh is an Assistant Professor in Simulation of Energy Materials, at the Department of Chemistry in the University of Cambridge. He earned his undergraduate at Trinity College Dublin in his native Ireland, PhD at University College London and Imperial College London, and a Fellowship at the Harvard University Center for the Environment. Seán was awarded the 2024 IOP and APS Computational Physics thesis prizes, the 2025 RSC Energy Sector Thesis Prize and named a 2025 JPhys Energy Emerging Leader. He serves as an Editorial Board Member for Physical Review Materials, an Assessor for Innovate UK and a Modelling & Simulation Steering Group Member at the Henry Royce Institute.
Seán heads the Simulation of Advanced Materials (SAM) lab (https://sam-lab.net), studying defects and disorder in materials, including methodological/software developments – such as the doped (https://doped.readthedocs.io) and ShakeNBreak (https://shakenbreak.readthedocs.io/) defect modelling toolkits – and machine-learning approaches.