Benchmark datasets for crystalline organic semiconductors


The database is grouped in folders, each folder corresponding to a single polymorph encoded in the folder name as system_polymorph, where 'system' is the name of the material and 'polymorph' is the polymorph label, e.g. 'cryst-b' is beta-polymorph, 'cryst-aH' is high-symmetry/temperature alpha-polymorph, 'cryst-x' is purely computational polymorph, 'cryst' is the main polymorph and so forth. Each folder contains the following files:

Computational methodology

All calculations and data processing are performed with the same set of computational protocols referenced below:

Selection of materials

Datasets are composed and will be refined in future according to the following principles:


All data are 7-zipped into one archive per method (here is PBE-D3paw600), below are indexes to datasets as text files whose first line contains coma-separated list of entries, and the rest of file provides some technical description of the dataset (there 'na' is number of atoms in unit cell, 'na1' is number of atoms in molecule, 'Ebind' binding energy in eV per molecule, 'freq' are imaginary frequencies, 'missing' lists missing files). Because for isolated molecules much more complicated calculations can be performed in a basis of atomic orbitals, we have added single-molecule calculations (here is CAM-B3LYPp2p).