LPS Database

Description

We present a Sulfur K-edge x-ray absorption near-edge structure (XANES) database of 18 crystalline and 48 amorphous Lithium-Phosphorous-Sulfur (LPS) compounds. The database contains a total of 2681 XANES spectra of symmetrically inequivalent absorbing S sites. Structures were taken from Materials Cloud entry 2022.17 (DOI: 10.24435/materialscloud:j5-tz) and were originally generated by systematically removing Li, P and S atoms from known crystal structures using an evolutionary algorithm and an artificial neural network based interatomic potential. The details of this procedure can be found in Guo et al. Chem. Mater. 34, 6702–6712 (2022). From this data set, low-energy structures were selected for spectral simulations. The excited electron and core hole method as implemented in VASP 6.2.1 was used to compute the XANES spectra for each symmetrically inequivalent Sulfur atom. The details of the VASP simulations can be found in the associated manuscript.

In summary, the LPS database presented here contains two key types of calculations, with specific files and their contents detailed below.

  • Neutral calculations: */input_SCF contains data in which electronic relaxation is computed without the presence of a corehole and XAS are not computed. These files are usually used for relative spectral alignment.
  • Corehole calculations: */{int}_{int} contains XAS data, i.e., electronic relaxation is performed in the presence of a corehole and XAS is computed. The first integer in the file name represents the site index, and the second is the multiplicity of that site (the number of symmetrically equivalent sites of that type).

Compression steps

Starting from the raw VASP output files, we proceeded with the following steps to generate the database here. Note that these scripts are included for transparency only, and are not required to use the database.

Step 1: run the script: bash process_mu.sh. This finds all sub-directories that do not contain the substring "SCF", indicating that the results correspond to a corehole calculation. From there, a simple grep command is used to find the spectral data, and a hardcoded value of 40000 total spectral grid points (with two extra to account for the header) is used to pipe the output of the entire spectrum to a temporary file called mu.txt. In addition, the Fermi energy is extracted from the OUTCAR and saved to a file called efermi.txt. This data is important for aligning different VASP spectra to each other.

Step 2: run the script: python3 process_mu2.py (requires Python 3, numpy and tqdm by default). All files with name */mu.txt are iterated over and ingested. Each raw text file is processed using numpy to extract the spectral data proper. We keep only the x, y and z component contributions to the spectra (ignoring the off-diagonal components) and keep only the pieces of the spectral range in which the intensity is mostly nonzero. This data is then saved to files mu.dat.

Step 3: run the compression script

find 22-05-13-LPS-no-POTCAR -name "*POSCAR" -o -name "*KPOINTS" -o -name "*INCAR" -o -name "*OSZICAR" -o -name "*mu.dat" -o -name "*efermi.txt" | tar -cvjf 22-05-13-LPS-no-POTCAR.tar.bz2 -T -

This script finds all files of types

  • POSCAR: input structure
  • KPOINTS: input kpoints mesh
  • INCAR: input parameters
  • OSZICAR: output showing the electronic structure convergence
  • mu.dat: see above
  • efermi.txt: see above

and compresses them using bzip2. Thus only these files are included in the compressed data.

Files included

  • 22-05-13-LPS-no-POTCAR.tar.bz2: see above
  • process_mu.sh: see above
  • process_mu2.py: see above
  • README.md: this file
  • structure-index.csv: a main index containing information on the type and origin of each of the 66 materials contained in this database. Specifically, "Type" (glassy or crystalline), Origin (where the material was source/how it was generated), "Notes", "File", "Unique sites" (the number of symmetrically inequivalent sites), "x_P", "x_P2S5" and "Formula".
  • test_plot.py: a simple script for making a quick plot of the spectral data as it is in the files mu.dat. Just used as a sanity check.
  • Spectra_66_compounds.tar.bz2: broadened and unbroadened materials spectra for quick reference.