This repository contains the databases presented in the article "Screening of material defects using universal machine-learning interatomic potentials".

Vacancies.json contains all vacancy formation energy in the JSON format.
For each MPID, the following properties are available :
	- Symbols : List of elements.
	- Vacancy : List of vacancy formation energies.
	- Cell : Lattice constant of the structure.
	- Positions : Atomic coordinates of each atoms.

Vacancies.db also contains the vacancy formation energy, but in SQL format which can be browsed with ASE web-gui using command: ase db Vacancies.db -w
For each entry in the database, the vacancy formation energies can be accessed as entries.data['Ef']['Ef'].


2D_layers.json contains all potential layers from chemical etching in the JSON format.
For each MPID, the following properties are available :
	- Symbols : List of elements.
	- Layer : Formula of the layer.
	- 2D_factor : 2D factor of the layer after relaxation.
	- Formation_energy : Formation energy of the layer.
	- HEE : Highest etched energy (as defined in the article).
	- LNE : Lowest non-etched energy (as defined in the article).
	- LDFE : Lowest vacancy formation energy of the layer. This correspond to the minimal value of the second etching.
	- Cell : Lattice constant of the structure.
	- Positions : Atomic coordinates of each atoms.

2D_layers.db also contains the layers, but in SQL format which can be browsed with ASE web-gui using command: ase db Vacancies.db -w
Additional properties, such as the etched elements, energy above convex hull of the parent phase or the initial dimensionality are also available in this database.