High-Throughput Screening of Symmetry-Breaking Reconstructions at Vacancies in Oxides

When and Where

Nov 27, 2023
2:00pm - 2:15pm

Sheraton, Second Floor, Liberty B/C

Presenter

Seán Kavanagh

Irea Mosquera-Lois

David Scanlon

Aron Walsh

Yu Kumagai

Co-Author(s)

Seán Kavanagh^1,2,Irea Mosquera-Lois¹,David Scanlon²,Aron Walsh¹,Yu Kumagai³

Imperial College London¹,University of Birmingham²,Tohoku University³

Abstract

Seán Kavanagh^1,2,Irea Mosquera-Lois¹,David Scanlon²,Aron Walsh¹,Yu Kumagai³

Imperial College London¹,University of Birmingham²,Tohoku University³

Point defects are a universal feature of crystalline materials, whose identification is often addressed by combining experimental measurements with theoretical models. The standard approach of simulating defects is, however, prone to miss the ground state atomic configurations of defects, due to energy-lowering reconstructions from the idealised crystallographic environment.1–4 Missed ground states severely compromise the accuracy of calculated properties. To address this issue, we report an approach to efficiently navigate the defect configurational landscape using targeted bond distortions and rattling.5 In this study, we apply our defect structure-searching method (implemented in ShakeNBreak6) in a high-throughput methodology to oxygen vacancies in over 200 metal oxides, building on the work of Kumagai et al.7 Our results reveal energy-lowering reconstructions missed by the standard modelling approach in ~50% of cases, demonstrating the widespread prevalence of this phenomenon. Analysis of this large defect dataset allows us to correlate the host material properties with the likelihood of defect symmetry-breaking and the motifs observed. Finally, we extend this analysis by re-training a graph neural network force-field on this large dataset of defect relaxations and applying it to a much wider oxide dataset (~1000 oxides), demonstrating the future potential of machine learning methods to accelerate defect calculation workflows. 1 M. Arrigoni and G. K. H. Madsen, npj Comput Mater, 2021, 7, 1–13. 2 I. Mosquera-Lois and S. R. Kavanagh, Matter, 2021, 4, 2602–2605. 3 S. Lany and A. Zunger, Phys. Rev. Lett., 2004, 93, 156404. 4 S. R. Kavanagh, A. Walsh and D. O. Scanlon, ACS Energy Lett., 2021, 6, 1392–1398. 5 I. Mosquera-Lois, S. R. Kavanagh, A. Walsh and D. O. Scanlon, npj Comput Mater, 2023, 9, 1–11. 6 I. Mosquera-Lois, S. R. Kavanagh, A. Walsh and D. O. Scanlon, Journal of Open Source Software, 2022, 7, 4817. 7 Y. Kumagai, N. Tsunoda, A. Takahashi and F. Oba, Phys. Rev. Materials, 2021, 5, 123803.