Work function and cleavage energy dataset of paper "Discovery of stable surfaces with extreme work functions by high-throughput density functional theory and machine learning"

Work Function and Cleavage Energy Database

This database contains the work function and cleavage energy calculation results (calculated by high-throughput density functional theory).  

Databases

The content of the two database files is described here. The first one contains results from the work function and cleavage energy calculations of the whole generated surface database for unrelaxed slabs. The second one contains the selected slabs that have work functions below 2 or above 7 eV, which then underwent ionic relaxation. 

The first database JSON file `WF-CE_database_58332.json` contains 58,332 records each containing the following information:

`mpid`: (string) Materials Project ID

`miller_index`: (list) Miller index of surface hkl as a list of length 3

`term`: (int) Termination number indexing (unique terminations are numbered starting at 1)

`surface_elements`: (list) List of atomic numbers present in the topmost atomic layer

`surface_elements_string`: (str) Same information as `surface_elements` but as a concatenated string of elements

`WF`: (float) The DFT-calculated work function in eV of unrelaxed slab of the top surface

`cleavage_energy`: (float) The DFT-calculated cleavage energy in eV/A^2 of unrelaxed slab

`slab`: (str) The unrelaxed surface slab as a Pymatgen dictionary represented as a string (use eval to recover the dictionary which can be read by pymatgen)

`energy`: (float) Total energy of unrelaxed slab in eV

`bulk_energy`: (float) Total energy of the bulk (reoriented in the direction of the Miller index) in eV

`Fermi`: (float) Fermi level in eV -

`convergence`: (list) DFT convergence parameters, a list containing E_cutoff, k_x, k_y, k_z

`nsites`: (int) Number of atomic sites in the slab unit cell

`nterm`: (int) Number of unique terminations for given orientation/material

`thickness_n`: (int) Slab thickness in numbers of bulk unit cells

`thickness_A`: (float) Slab thickness in Angstroms

`area`: (float) Surface area in A^2

`sym`: (bool) Whether or not there exists a mirror or glide plane parallel to the surface, or a 2-fold rotation axis normal to the surface (without adding vacuum). This is equivalent to the surface being non-polar or polar.

`sym_vac`: (bool) Same as `sym` but after adding vacuum in the c-direction.

The second database JSON file `WF-CE_database_relaxed_90.json` contains 90 records each containing the information above (except `WF`, `slab`, `cleavage_energy`, and `energy` have been renamed to `WF_unrelaxed`, `slab_unrelaxed`, `cleavage_energy_unrelaxed`, and `energy_unrelaxed`, respectively) and in addition, the database has the following additional fields:

`WF_relaxed`: (float) The DFT-calculated work function in eV of relaxed slab of the top surface

`cleavage_energy_relaxed`: (float) The DFT-calculated cleavage energy in eV/A^2 of relaxed slab

`slab_relaxed`: (str) The relaxed surface slab as a Pymatgen dictionary represented as a string (use eval to recover the dictionary which can be read by pymatgen)

`energy_relaxed`: (float) Total energy of relaxed slab in eV

`icsd`: (list) List of ICSD numbers of the bulk crystal

`formula`: (str) Chemical formula of the bulk

`spacegroup`: (int) The spacegroup number

How to use

The databases can be loaded with pandas as follows:import pandas as pddata = pd.read_json('./WF-CE_database_58332.json')print(data['WF'].mean())