Research data for "Modelling atomic and nanoscale structure in the silicon-oxygen system through active machine-learning"

This dataset supports the paper "Modelling atomic and nanoscale structure in the silicon-oxygen system through active machine-learning". The paper is online here:

The following files are provided:

• Potential files for the complex ACE potential for Si-O and additionally for the linear and Finnis-Sinclair ACE ("potential")
• Training database in a xyz file format as well as pckl.gzip with weights, which have been used for the fits. Includes also the parameter file for the DFT calculations ("database")
• Parameters files, which have been used for fitting the potentials ("fitting")
• Amorphous matrix embedding example script to extract the cells from the large-scale simulations as well as a script for the LAMMPS simulation to amorphize the boundaries ("amorphous_matrix_embedding")
• Results of various simulations with corresponding input scripts, input structures and analysis scripts ("results")
  
  • High pressure simulations: Compression of amorphous silica up to 175 GPa, energy-volume curves, enthalpies
  • Phase diagram: Input files for the calculation of the free energy and scripts to analyse the data to generate a phase diagram
  • SiO: Contains the amorphous and partially crystalline structure files for the SiO models and the small-scale Si-SiO2 interface models. Moreover, files with structure factor data, crystallinity and interface area can be found here. 
  •  Surfaces: Aerogel structures, which have been shown in the paper and the small-scale amorphous surface models with corresponding DFT data. It also includes the bulk reference data of the small-scale amorphous surface models, which is necessary to calculate the surface energy.
  • Testsets: Include the testsets referenced in Table 1 and Supplementary Table 3.
  • SiO2: Contains SiO2 structural models generated by the hybrid approach and using only the ACE potential.