This dataset contains charge densities for NMC (Ni, Mn and Co) 2x2x1 supercell (12 transition metal atoms and 12 Li/vacancy site) with varying levels of Li content. For each structure we first randomly sample the number of Mn, Ni and Co atoms given that the total number of transition metal atoms is 12 and then randomly assign them to the transition metal positions of the lattice. Similarly the number of vacancies is uniformly sampled between 0 and 12 and vacancies are assigned to the Li site. The generated configurations are then relaxed in two steps: First we relax the atom positions with fixed cell parameters and then we allow both positions and cell parameters to relax. We keep only the electron density (CHGCAR) file after the last cell relaxation step. The atoms are relaxed until forces on each atom are lower than 0.01 eV/Å.

The final relaxation is done with the following VASP settings (DFT level of theory) through the Atomic Simulation Environment:

xc='PBE', gga='PS', istart=1, algo='Normal', icharg=1, nelm=1800, ispin=1, nelmdl=6, isym=0, lcorr=True, potim=0.1, nelmin=5, kpts=[3,3,1], ismear=0, ediff=0.1E-03, ediffg=-0.05, sigma=0.1, nsw=200, isif=3, ibrion=2, ldiag=True, lreal='Auto', lwave=False, lcharg=True, prec='Normal'

The resulting CHGCAR files have been compressed with lz4 compression and packed in non-compressed tar archives with up to 1000 structures in each.

The datasplits json files contain the indices (0-index) of the train, validation and test sets used in the paper "Equivariant Graph neural networks for fast electron density estimation of molecules, liquids, and solids"