Density-functional-based tight-binding parameterization of Mo, C, H, O and Si for studying hydrogenation reactions on molybdenum carbide

Hydrogenation reactions catalyzed by transition-metal-containing nanoparticles represent an important type of reaction in chemical industry. However, the modeling of these reactions in their working conditions requires much longer simulation times than what could usually be achieved with ab initio or first-principle methods. To address this problem, in this work, the density-functional-based tight-binding (DFTB) method was parameterized for hydrogenation reactions on molybdenum carbide catalysts, involving the elements C, H, Mo, O and Si. The overall quality of the DFTB parameters was tested with band structure/molecular orbital energies, molecular/crystal structures, chemisorption bond strengths, hydrogen adsorption energies, hydrogenation reaction energies, molecular vibrational frequencies, energy barriers and the structures of the transition states for systems of interest. The parameterized DFTB method gave errors of <1.45 % for bond distances of hydrocarbons and 4.86 % for non-hydrocarbons. It could reproduce the structure and vibrational frequencies (with errors of about 100 cm⁻¹) of selected hydrocarbon–molybdenum carbide complexes obtained from DFT calculations. Good agreement was reached between DFTB and DFT on the dissociative adsorption of hydrogen on the α-Mo₂C (0001) surface. For most of the hydrogenation reactions examined, DFTB showed errors of ~2 kcal/mol compared to DFT/PBE, with a few exceptions of ~5 kcal/mol. It could also describe the reaction energies, the forward and reverse energy barriers and the transition-state structures for the benzene hydrogenation reaction on a Mo₃₈C₁₉ cluster.