JARVIS Repository for Automated Materials Discovery Simulations

JARVIS is a repository from the National Institute of Standards and Technology automating materials discovery through thousands of force-field, density functional theory, and machine learning calculations. It contains data on energetics, elastic constants, band-structure, and phonon frequencies for bulk, 2D, 1D, and 0D material systems. The repository was last updated in July 2022.

Use Cases

• Train machine learning models on JARVIS-DFT data to predict material properties like heat of formation and GGA/METAGGA bandgaps.
• Analyze force-field calculation results for properties such as elastic constants and surface energies from the JARVIS-FF section.
• Use DFT-calculated band-structure and density of states data to screen materials for electronic or thermoelectric applications.
• Compare phonon frequencies from gamma-point phonon calculations across different material dimensionalities (3D, 2D, 1D, 0D).

Strengths

• Includes thousands of automated calculations across multiple simulation methodologies (force-field, DFT, ML).
• Covers a wide range of material properties including energetics, elastic constants, band-structure, and phonons.
• Data spans multiple material dimensionalities: 3D-bulk, 2D single layers, 1D nanowires, and 0D molecular systems.

Limitations

• Specific dataset size, row counts, and column structures are not provided in the input.
• Data is provided in HTML format, which is not ideal for direct programmatic analysis and may require parsing.
• The last update was in July 2022, so newer material discoveries or computational methods may not be included.

Provenance

Source

National Institute of Standards and Technology

Collection Method

Automated computational simulations using LAMMPS and VASP software, with machine learning tools trained on the generated data.

Freshness

Last updated on 2022-07-29.