Supplementary Data for Amorphous Alumina Simulations and Analysis

This repository contains the supplementary data, scripts, and pseudopotentials used in the study of defect formation and electronic structure in hydrogenated amorphous alumina.
It is structured into four main directories, each of which originally contained its own README.txt.
Here, all descriptions are merged into a single document for clarity.

Folder Overview

1. simulation_data/

Contains all input and output files from simulations. Subdirectories cover different levels of modeling and validation:

DFT_binding_energy/

Contains input structures for binding energy calculations using Quantum ESPRESSO (QE):
General folder contents:

  • Al_relaxed.xyz
  • amorphous_relaxed.xyz
  • bayerite_relaxed.xyz
  • sapphire_relaxed.xyz
  • README.txt

DFT_charge_density/

This directory contains charge density and Bader analysis results from DFT calculations performed with Quantum ESPRESSO for three materials:

  • Aluminum (Al)
  • Bayerite (Al(OH)₃)
  • Sapphire (α-Al₂O₃)

The calculations were carried out to evaluate charge density differences under two conditions:

  1. With a core hole (tot_charge = 1)
  2. Without a core hole (tot_charge = 0)

Folder structure:

  • Al_tot0_no_core/, Al_tot1_core/
  • Al2O3_tot0_no_core/, Al2O3_tot1_core/
  • bayerite_tot0_no_core/, bayerite_tot1_core/

Contents of each subfolder:

  • charge_density.cube – 3D charge density in cube format
  • ACF.dat, AVF.dat, BCF.dat – Bader charge analysis results
  • aiida.out – QE SCF calculation output

Naming convention:

  • tot0_no_core → no core hole, neutral system
  • tot1_core → with core hole, charged system

These datasets are used to compute and analyze charge density differences between core-excited and ground-state configurations.

MD/

This folder contains the simulation data analyzed in the publication. Each folder corresponds to a specific structure.
Bayerite has two files because its initial structure was too large for VASP analysis; a smaller structure was also used.

General folder contents:

  • structure.lmp – LAMMPS initial structure
  • in.runmd – MD run file for equilibration
  • log.lammps – Thermodynamic data
  • dump.run – Dump file for equilibration
  • dump.equil – Equilibrated trajectory
  • restart – Restart file for continuation
  • in.restart – Restart run file
  • log.restart – Thermodynamic data from restart
  • POSCAR.1000 – VASP structure file from equilibration
  • INCAR – VASP input for charge distribution
  • CHGCAR_structure.vasp – Charge distribution result (Henkelman Bader refined)

Additional pseudominimization files (10K, 100K) exist for AlH₂, AlH, AlH006, Sapphire, and Bayerite (used for low-temperature structural analysis).

Polycondensation/

Contains results and analysis of simulations related to the polycondensation process.

Contents:

  • AlH2/ – simulation data for AlH₂
  • *.ovito files – structural evolution snapshots for OVITO
  • gp1_ap_QD2.pkl – Gaussian process regression model for Auger parameter evolution
  • gp1_water_QD2.pkl – Gaussian process model for water formation dynamics

Potential_validation/

Contains VASP calculation results used to validate the Matlantis machine-learning potential for aluminum–hydrogen and oxide systems.

Folder structure:

  • Aluminum hydrides: AlH006/, AlH013/, AlH1/, AlH2/ and *_gen_500/ versions
  • Oxide/hydroxide phases: bayerite/, sapphire/, water/
  • Thermal simulations: annealing_half/, annealing_full/, heating_510K/, heating_850K/

Contents:

  • OUTCAR – VASP output file with electronic structure, energies, forces, stresses

Purpose:
Benchmark dataset comparing VASP reference data with the Matlantis ML potential across multiple phases and conditions.

2. analysis_scripts/

Contains Python scripts, Jupyter notebooks, and OVITO workflows for analyzing charge densities and the Auger parameter (AP) in aluminum oxide/hydroxide systems.

Python scripts:

  • Analysis_simp_mod.py – Auger parameter per structure (simplified model)
  • Analysis_mod_full_model.py – Auger parameter per atom (full model)
  • compare_charge_densities.py – Comparison of charge densities
  • voxel_grid_center_radial_binningVlad_extract.py – Extracts normalized charge volume from voxel grids
  • electrons_per_sample.py – Computes electrons per voxel grid

Jupyter notebooks:

  • AP_per_nn.ipynb – Auger parameter per nearest neighbor
  • AP_per_ligand_type.ipynb – Auger parameter per ligand type

OVITO scripts:

  • calculate_AP_time_average.ovito – Computes AP time average from trajectory
  • isosurfaces_visualization.ovito – Visualizes charge density isosurfaces
  • compare_Al_valence.ovito – Subtracts valence charge densities between samples

3. structure_generation/

Procedures and scripts for constructing initial amorphous alumina structures from crystalline bayerite.

  • Initial_structure_generation.ovito – OVITO state file with custom modifier
  • OVITO_water_removal_script.py – Removes OH and H in equal proportions to achieve charge-neutral stoichiometry
  • bayerite.lmp – LAMMPS input structure (downloaded from Materials Project, orthogonalized, and replicated)

4. pseudos/

Quantum ESPRESSO pseudopotentials used in the simulations.

Includes standard PAW pseudopotentials and specialized core-hole pseudopotentials distinguished by orbital type and star notation, e.g.:

  • Al.star1s.pbe-n-kjpaw_psl.1.0.0.UPF
  • O.star1s.pbe-n-kjpaw_psl.0.1.UPF

Usage Notes

  • This merged README.md replaces the individual README.txt files for repository upload.
  • Each subsection describes the purpose and usage of its corresponding data.
  • The dataset enables full reproducibility of the simulations, structure generation, and analysis presented in the associated publication.

Citation

If you use these data or scripts, please cite the corresponding publications:

  • https://doi.org/10.1038/s41524-025-01676-5
  • https://doi.org/10.48550/arXiv.2408.08255