Seán Kavanagh

Have a read if you're interested!

I think this shows exciting potential for MLFFs in defect modelling, but with caveats... they fail dramatically for non-fully-ionised charge states where localisation matters! They work here due to the enormous configuration space but with relatively simple underlying energetics

This allows an efficient tiered screening; scanning 𝘢𝘭𝘭 compounds in the ICSD & Materials Project database for split cation vacancies

Indeed, due to the relatively simple underlying energetics (primarily electrostatics and strain), this problem is well-suited to MLIPs. I find that foundation models (MACE, NequIP, Allegro -- stayed tuned for the latter!) successfully predict split vacancy formation in most cases

We can't just enumerate all potential split vacancy configurations; the search space is enormous (>1000s of candidate geometries per defect). I find instead that electrostatic models can greatly reduce this space, as electrostatics dominate energetics for these 'stoichiometry-conserving' complexes

Unfortunately, they are very challenging to identify with current defect structure-searching methods (e.g. 𝗦𝗵𝗮𝗸𝗲𝗡𝗕𝗿𝗲𝗮𝗸) due to their 'non-local' nature, as most of these methods employ some form of 'local' structure searching techniques

Vacancy defects can sometimes transform to split-vacancies, with dramatic changes in energy & behaviour, e.g. in Ga₂O₃ as discovered by Joel Varley. They have only been witnessed in a handful of cases – are they inherently rare or have we just not had the tools to find them?

Machine learning can be powerful for modelling defects, but currently only in select cases.

MLIPs (& geometric/electrostatic tools in doped) allow screening for challenging 'non-local' defect reconstructions (split vacancies) in all ICSD/MP solids, w/caveats

iopscience.iop.org/article/10.1...

Collaboration with @uclchemistry.bsky.social @imperialmaterials.bsky.social @upc.edu @unibirmingham.bsky.social

Chalcogen Vacancies Rule Charge Recombination in Pnictogen Chalcohalide Solar-Cell Absorbers Pnictogen chalcohalides (MChX) represent an emerging class of nontoxic photovoltaic absorbers, valued for their favorable synthesis conditions and optoelectronic properties. Despite their proposed def...

Starting with a visit to London in 2023, Cibrán began a deep dive on defects in pnictogen chalcohalides (BiChX), finding the chalcogen vacancy to dominate recombination (similar to Sb2Se3!).

He shows that selective anion substitutions can mitigate their effect!
pubs.acs.org/doi/10.1021/...

Chalcogen Vacancies Rule Charge Recombination in Pnictogen Chalcohalide Solar-Cell Absorbers
https://arxiv.org/pdf/2504.18089
Cibrán López, Seán R. Kavanagh, Pol Benítez, Edgardo Saucedo, Aron Walsh, David O. Scanlon, Claudio Cazorla.

Available on the development branches;
- Complex defect multiplicities, symmetries and degeneracies
- N-dimensional chemical potential heatmap plotting using fixed values (to reduce to 3-D)
- Defect "stenciling" to regenerate (relaxed) geometries in arbitrary supercells...

- Tutorial for generating polaron distortions with ShakeNBreak: shakenbreak.readthedocs.io/en/latest/Sh...
- Many efficiency updates.
- Miscellaneous minor bug fixes, improvements and docs updates

github.com/SMTG-Bham/Sh...

Defect Tolerance via External Passivation in the Photocatalyst SrTiO3:Al The efficiency of solar-to-energy conversion in semiconductors is limited by charge carrier recombination, often via defect-induced gap states. Although some materials exhibit an intrinsic defect tole...

- Site-competition handling in defect concentrations; see doi.org/10.26434/che...
- Include 'adsorbate' interstitial sites for structures with significant vacuum volume
- Improved algorithm for defect site clustering (for plotting & concentration analyses)

Release 3.1.0 · SMTG-Bham/doped Update chemical potentials code: Handle recent breaking changes in pymatgen (Apr 2025). Auto check compatibility of INCAR\s and POTCAR\s in competing phases calculations (as already done for super...

doped (3.1.0) and ShakeNBreak (3.4.2) have had new releases!

- Streamlined chemical potential handling
- Auto-compatibility checks w/competing phases calculation settings (as for defects) – common pitfall
- Directly parse spin magnetisation (incl SOC)
...

github.com/SMTG-Bham/do...

We find a high sensitivity of the band edges to lattice parameters (i.e. deformation potentials) in t-Se, which combined with v low elastic constants (vdW-bonded) indicates significant thermal fluctuations and strain effects

We highly encourage sharing these outputs in defect theory papers!
Quick and easy way to ensure reproducibility and queryability.
Of course, this is in addition to making the relevant raw data available (which doped readily sums to lightweight & readable JSON files too)

doped automatically outputs dataframes/tables to aid reproducibility (and reduce manual efforts), showing all contributions to formation energies, estimated charge correction errors, symmetries, degeneracies etc

In particular, we find many energy-lowering reconstructions of defect geometries using ShakeNBreak, which are missed by standard/rattled relaxations (incl the V_Se^0 bipolaron mentioned above).

Full details in the SI for full reproducibility 🤝

The theory portion was primarily performed using doped (doped.readthedocs.io/en/latest/) and ShakeNBreak (shakenbreak.readthedocs.io/en/latest/), which can expedite and expand defect analysis, with many useful tools.
Some directly-output plots in the next reply ⬇️

Extended defects / interfaces are often easier to engineer away than unavoidable 𝘱𝘰𝘪𝘯𝘵 defects (lacking the same entropic driving force), so this is an exciting indication of the potential for t-Se PV! 📈

Overall, our results suggest a strong tolerance to 𝘱𝘰𝘪𝘯𝘵 defects in t-Se, and indicate that GBs/interfaces are the key limiting factors.
This aligns with recent DLTS and DLCP measurements, which suggested extended defects to be dominant in t-Se (refs & details within).

Calculating impurity formation energies (w/doped & SnB), we find strong valence alternation -> amphoteric & charge compensation for H, pnictogens & halogens. Chalcogens are electrically neutral.

F contributes the strongest to hole doping, but still relatively weak (~10¹² cm³)

With ToF-SIMS analysis of high-quality Se films, we show that F, Cl, O and Te are present in the samples, mainly from precursors used.
O and Te are mainly at the surfaces/interfaces.
Br & I are also expected but the Se/Te isotopes prevent direct identification.

Calculating the non-radiative recombination rates of the intrinsic point defects, we find that they 𝘢𝘳𝘦 𝘯𝘰𝘵 fast capture centres and thus do 𝘯𝘰𝘵 limit performance in Selenium PV.
They also do not contribute significantly to hole doping...
So what else could it be?

A bipolaron metastable state is found for the neutral vacancy (using ShakeNBreak), only 27 meV higher energy (depending on its spin configuration).

The V_Se geometries are driven by valence alternation (Kastner et al. PRL 1976; see SI), with Se inter-chain bonds / terminations

We find interstitials to be very low energy due to a 'split-interstitial' geometry (found with doped & ShakeNBreak) and electrically neutral as a result, while vacancies are higher energy and have several charge states / in-gap defect levels

Selenium (t-Se) had several efficiency jumps in the past 5 years, and now shows a high potential for indoor & tandem PV.
The origin of remaining voltage deficits and hole doping are not clear however.
We used a combined theory-expt approach to investigate its defect chemistry 🔎

Intrinsic point defect tolerance in selenium for indoor and tandem photovoltaics Selenium has reemerged as a promising absorber material for tandem and indoor photovoltaic (PV) devices due to its elemental simplicity, unique structural features, and wide band gap. However, despite...

Intrinsic & extrinsic (dopant) defect chemistry of trigonal Selenium for PV, incl metastable states & non-radiative recombination ⬇️

Combined theory & expt analysis, we find an intrinsic tolerance to 𝘱𝘰𝘪𝘯𝘵 defects, with GBs/interfaces the limiting factor for PV 📈
pubs.rsc.org/en/Content/A...