ShakeNBreak: Symmetry-Breaking and Reconstruction at Defects in Solids

Pre-recording of my talk 'Symmetry-Breaking and Reconstruction at Defects in Solids' at APS March 2023 in Las Vegas, USA – discussing the issue of global optimisation for defects in solids, its importance and our work (ShakeNBreak) on tackling this issue.

Talk recording on YouTube: https://youtu.be/aqXlyLofLSU

ShakeNBreak website: https://shakenbreak.readthedocs.io/en/latest/

Our general defect calculation package doped is available here: https://github.com/SMTG-UCL/doped

See our open-access papers on defect structure-searching here:
https://www.nature.com/articles/s41524-023-00973-1
https://joss.theoj.org/papers/10.21105/joss.04817

For other talks on YouTube, have a look at my channel!
For other research articles see:
https://bit.ly/3pBMxOG

Slides for my talks can be found here:
https://speakerdeck.com/kavanase

Other references:
Matter Preview of Defect Structure Searching: https://www.sciencedirect.com/science/article/pii/S2590238521002733
Metastable defects : https://doi.org/10.1039/D2FD00043A
Recombination at V_Cd in CdTe (case study): https://pubs.acs.org/doi/abs/10.1021/acsenergylett.1c00380

Transcript

1. Standard defect supercell relaxation Seán R. Kavanagh‡ & Irea Mosquera-Lois,‡ Aron Walsh, David O. Scanlon APS March 2023 I. Mosquera-Lois‡ & S. R. Kavanagh‡*, A. Walsh and D. O. Scanlon*, npj Comput Mater, 2023, 9, 1–11 Identifying the Ground State Structures of Point Defects in Solids
2. Standard defect supercell relaxation Identifying the Ground State Structures of Point Defects in Solids Seán R. Kavanagh‡ & Irea Mosquera-Lois,‡ Aron Walsh, David O. Scanlon APS March 2023 I. Mosquera-Lois‡ & S. R. Kavanagh‡*, A. Walsh and D. O. Scanlon*, npj Comput Mater, 2023, 9, 1–11
3. Defect Calculation Workflow 3 Host primitive cell Goyal et al, Comp Mater Sci 2017
4. Defect Calculation Workflow 4 Host primitive cell Goyal et al, Comp Mater Sci 2017
5. Defect Calculation Workflow 5
6. Defect Calculation Workflow 6
7. Defect Calculation Workflow 7
8. Defect Calculation Workflow 8
9. Defect Calculation Workflow 9
10. Defect Calculation Workflow 10
11. Defect Calculation Workflow 11 ➡ Energy ➡ Concentration ➡ Transition Level ➡ Deep/Shallow ➡ Doping ➡ Carrier capture ➡ Diffusion ➡ …
12. VCd in CdTe
13. Metal-metal dimers possible for vacancies in semiconductors: Lany & Zunger Phys Rev Lett 2004 Lany & Zunger Phys Rev B 2005 VCd in CdTe
14. Metal-metal dimers possible for vacancies in semiconductors: Lany & Zunger Phys Rev Lett 2004 Lany & Zunger Phys Rev B 2005 VCd in CdTe
15. Metal-metal dimers possible for vacancies in semiconductors: Lany & Zunger Phys Rev Lett 2004 Lany & Zunger Phys Rev B 2005 VCd in CdTe
16. VCd in CdTe Tei in CdTe Standard Relaxation ShakeNBreak (Our Method)
17. Metal-metal dimers possible for vacancies in semiconductors: Lany & Zunger Phys Rev Lett 2004 Lany & Zunger Phys Rev B 2005 Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 VCd in CdTe
18. Metal-metal dimers possible for vacancies in semiconductors: Lany & Zunger Phys Rev Lett 2004 Lany & Zunger Phys Rev B 2005 Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 VCd in CdTe
19. Potentially the Wrong Defect! Mosquera-Lois & Kavanagh* Matter 2021 Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 Mosquera-Lois‡ & Kavanagh‡*, Walsh and Scanlon* npj Comput Mater 2023 Standard defect supercell relaxation
20. How Prevalent is This? Tested on a diverse range of materials: Si, CdTe, GaAs, Sb2 S3 , Sb2 Se3 , CeO2 , In2 O3 , ZnO, anatase-TiO2 Energy-lowering reconstructions, missed by standard relaxations, found in every material studied Mosquera-Lois‡ & Kavanagh‡*, Walsh and Scanlon* npj Comput Mater 2023
21. Defect Calculation Workflow 21
22. How Important is This? Kavanagh*, Scanlon, Walsh, Freysoldt Faraday Discussions 2022 Inaccurate Structure ➡ Inaccurate Formation Energy ➡ Inaccurate: ➡ Energy ➡ Concentration ➡ Transition Level ➡ Deep/Shallow ➡ Doping ➡ Carrier capture ➡ Diffusion ➡ …
23. How Important is This? Very Standard Relaxation (Metastable) Example: Vacancies in Sb2 Se3 /Sb2 S3 Reveals rare 4-electron negative-U behavior and ultra- strong self-compensation in Sb2 S3 & Sb2 Se3 Difference in predicted VSb concentration = 1021 Our Method (Ground-state) Wang, Kavanagh, Scanlon, Walsh; ‘Four-electron Negative-U Vacancy Defects in Antimony Selenide’ Under Review at Phys Rev Lett (arXiv: 2302.04901)
24. How Important/Prevalent is This? Very Further Examples: • Gallium vacancies, migration and compensation in Ga2 O3 1 • Catalytic activity (divalent metal dopants in CeO2 )2 • CdTe solar cell performance3 • Defect absorption / bandgap lowering (Sn-doped Cs3 Bi2 Br9 )4 • Persistent Photoconductivity in Si, GaAs DX centres1,5 • Oxide polarons (in BiVO4 )6 • Colour centres and deep anion vacancies in II-VI compounds7 1. Varley et al J. Phys.: Condens. Matter 2011 2. Kehoe, Scanlon, Watson, Chem Mater 2011 3. Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 4. Krajewska, Kavanagh et al. Chem Sci 2021 5. Du & Zhang Phys Rev B 2005 6. Osterbacka, Ambrosio, Wiktor J Phys Chem C 2022 7. Lany & Zunger Phys Rev Lett 2004 Inaccurate Structure ➡ Inaccurate Formation Energy ➡ Inaccurate: ➡ Energy ➡ Concentration ➡ Transition Level ➡ Deep/Shallow ➡ Doping ➡ Carrier capture ➡ Diffusion ➡ …
25. Structure Searching Strategies 1. Electron attractor method 2. Random sampling 3. Evolutionary Algorithm ➡ Good for polaronic defects (but only polaronic defects) ➡ Guaranteed to find the ground-state, eventually… ➡ Can be enhanced with ML models Pham & Deskins, J Chem Theory Comp 2021 Huang, M. et al. J. Semicond 2022 Pickard & Needs. Phys Rev Lett 2006 Morris, Pickard, Needs. Phys Rev B 2008 Morris, Pickard, Needs. Phys Rev B 2009 Arrigoni & Madsen npj Comp Mater 2021 ➡ Significant manual effort (setup & hyperparameter tuning) ➡ Inefficient ➡ Require many calculations so typically only possible with cheap inaccurate levels of theory (semi-local DFT) ➡ Infeasible for typical full defect studies Further details in: Mosquera-Lois‡ & Kavanagh‡*, Walsh and Scanlon* npj Comput Mater 2023
26. Our Method: ShakeNBreak Idea: Leverage the localised “molecule-in-a-solid” behaviour of point defects: • Chemically-guided neighbour bond distortions: No. distorted bonds = Δ{Valence Electrons} • Stretch/compress neighbour bonds (±60% range) ➡ Distortion mesh of trial structures • ‘Rattle’: Add small random displacements to break symmetry and aid location of global minimum • Relax Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon*; npj Comp Mater 2023 Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon*; J. Open Source Software 2022
27. ShakeNBreak 11 relaxations with 𝚪-only sampling Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022
28. ShakeNBreak 11 relaxations with 𝚪-only sampling Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022
29. 11 relaxations with 𝚪-only sampling ShakeNBreak Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022
30. 11 relaxations with 𝚪-only sampling ShakeNBreak Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022
31. 11 relaxations with 𝚪-only sampling (Spin-Unpolarised for simplicity) ShakeNBreak :
32. Successfully reproduces all previously-reported cases (so far!) (Benchmarks: Si, CdTe, GaAs, CeO2 , ZnO…) Energy-lowering reconstructions identified in a diverse range of materials & defects (Sb2 S3 /Sb2 Se3 , In2 O3 , TiO2 , Si, CdTe, GaAs, CeO2 , ZnO) Can locate low-energy metastable structures ➡ Important for diffusion (transition states) and carrier recombination.3-5 Efficient (<10% computational cost of full defect study) Automated & super user-friendly (Python API or CLI), trivially parallel… ShakeNBreak 1. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon*; npj Comp Mater 2023 2. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon*; J. Open Source Software 2022 3. Alkauskas et al. Phys. Rev. B, 2016 4. Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 5. Kavanagh*, Scanlon, Walsh, Freysoldt; Faraday Discussions 2022 :
33. Sneak Preview: Slide Credit: Aron Walsh (Defects Seminar/Invited Talks 2023) Example: VO +1 in Na2 TiSiO5 Collaboration w/Prof. Yu Kumagai @ Tohoku University, Japan High-throughput search for symmetry- breaking at oxygen vacancies >200 oxides calculated so far Empirically, symmetry-breaking particularly likely with: - Multinary composition - Reduced crystal symmetry - Mixed ionic/covalent bonding (i.e. more complex potential energy surface, as expected)
34. Acknowledgements Profs David Scanlon & Aron Walsh Irea Mosquera-Lois Studies using ShakeNBreak (and finding lower energy defect structures): • X. Wang, S. R. Kavanagh, D. O. Scanlon, A. Walsh; Under Review at Phys Rev Lett (arXiv: 2302.04901) • C. Krajewska, S. R. Kavanagh et al. Chem Sci • S. R. Kavanagh*, D. O. Scanlon, A. Walsh, C. Freysoldt Faraday Discussions • J. Cen et al; Under Review at J. Mater. Chem. A, (ChemRxiv: 10.26434/chemrxiv-2023-nk8lr) • J. Willis, Q. Zhou et al. In preparation. • Y. T. Huang & S. R. Kavanagh et al. Nature Communications • A. Nicolson et al; In preparation – ChemRxiv next week! • Y. Kumagai et al; In submission • A. Samli et al;. In preparation
35. Key Takeaways • Obtaining the correct defect structure is important! • Our current procedure for defect calculations is incomplete • Energy-lowering reconstructions prevalent in a wide & diverse range of materials/defects. • ShakeNBreak = easily-implemented method to combat this and ensure the accuracy of defect calculations @Kavanagh_Sean_ sean.kavanagh.19@ucl.ac.uk 1. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 2. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022 3. Mosquera-Lois & Kavanagh*, Matter 2021 4. Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 5. Kavanagh*, Scanlon, Walsh, Freysoldt; Faraday Discussions 2022 Catch me at Tuesday’s poster session for more discussion!
36. Key Takeaways • Obtaining the correct defect structure is important! • Our current procedure for defect calculations is incomplete • Energy-lowering reconstructions prevalent in a wide & diverse range of materials/defects. • ShakeNBreak = easily-implemented method to combat this and ensure the accuracy of defect calculations @Kavanagh_Sean_ sean.kavanagh.19@ucl.ac.uk 1. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* npj Comp Mater 2023 2. Mosquera-Lois‡ & Kavanagh‡*, Walsh, Scanlon* J. Open Source Software 2022 3. Mosquera-Lois & Kavanagh*, Matter 2021 4. Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 5. Kavanagh*, Scanlon, Walsh, Freysoldt; Faraday Discussions 2022
37. How Important is This? Very VCd -1 VCd 0 h+ e- Our Method (Ground-state) Standard Relaxation (Metastable) h+ capture e– capture h+ capture e– capture Kavanagh, Walsh, Scanlon ACS Energy Lett 2021 Inaccurate Structure ➡ Inaccurate Formation Energy ➡ Inaccurate: ➡ Energy ➡ Concentration ➡ Transition Level ➡ Deep/Shallow ➡ Doping ➡ Carrier capture ➡ Diffusion ➡ …
38. Why isn’t this an issue for bulk structure prediction? Good initial guesses from experimental databases, starting us close to the global minimum For unknown crystal structure prediction, this is a huge avenue of research Ø PES exploration But defects are unknown structures! No database of known defect structures Ø Efficient structure-searching techniques required
39. Parameter Testing Tip of the iceberg…