{"references": ["Molpeceres. G., Zaverkin. V., Watanabe. N., K\u00e4stner. J. Binding energies and sticking coefficients of H2 on crystalline and amorphous CO ice. 2021. A&A 648, A84", "Spicher. S., Grimme. S. Robust Atomistic Modeling of Materials, Organometallic, and Biochemical Systems. Angewandte Chemie International Edition. 2020. Angew. Chem. Int. Ed., 59, 15665", "Becke. A. A new mixing of Hartree\u2013Fock and local density\u2010functional theories. 1993. J. Chem. Phys. 98, 1372", "Weigend.F. and Ahlrichs. R. Balanced basis sets of split valence, triple zeta valence and quadruple zeta valence quality for H to Rn: Design and assessment of accuracy. 2005. Phys. Chem. Chem. Phys., 7, 3297-3305", "Zaverkin. V. and K\u00e4stner. J. Gaussian Moments as Physically Inspired Molecular Descriptors for Accurate and Scalable Machine Learning Potentials. 2020. J. Chem. Theory Comput. 2020, 16, 8, 5410\u20135421"]}

This dataset was employed in the study of the adsorption dynamics of H2 molecules on amorphous and crystalline carbon monoxide ice [1]. It was generated by combining exploratory MD simulations on CO and CO/H2 clusters using the GFN-FF method [2]. Equispaced points in the trajectories were refined using density functional theory, and specifically BHLYP-D4/def2-TZVP [3,4]. The data is stored in python compressed array format (.npz) with the atomization energies in kcal/mol and atomic forces in kcal/mol/Ang. The data set contains five numpy arrays import numpy as np data = np.load('CO_H2_BHLYP.npz') data['R'] # Cartesian coordinates of nuclei (Ang.) data['E'] # Total energy (kcal/mol) data['F'] # Atomic forces (kcal/mol/Ang.) data['N'] # Number of atoms in each structure data['Z'] # Nuclear charges The dataset contains a maximum of 80 atoms per structure. For further details, please check Ref 1 and Ref 5.