Abstract

The realistic modelling of chemical reactions at finite temperatures is one main goal for computational chemistry, where a primary task lies in accurately calculating free energies. Setting aside the issue of solvation and referring to calculations for the isolated molecule, the latter is essentially a problem of efficient computation of molecular entropy. In this talk, I present an automated composite scheme for the accurate and numerically stable calculation of molecular entropy by efficiently combining DFT and semi-empirical extended tight-binding (xTB) methods.[1,2] Anharmonic effects are included through a modified rigid-rotor-harmonic-oscillator approximation and the Gibbs–Shannon entropy³ for extensive conformational ensembles, which are generated by a specialized meta-dynamics⁴ sampling procedure and extrapolated to completeness. I will demonstrate the approach on an established benchmark set, where unprecedented root mean square deviations, far better than chemical accuracy, are achieved. Furthermore, I will address possible extensions to the workflow and mention effects due to the choice of the underlying potential or the environment, for example, by implicit solvation.[5]

[1] C. Bannwarth, S. Grimme, et al., WIR Es Comput Mol Sci., 2021, 11, e1493.

[2] P. Pracht, S. Grimme, Chem. Sci., 2021, 12, 6551–6568

[3] D. Suárez, N. Dı́az, WIR Es Comput. Mol. Sci., 2015, 5, 1—26.

[4] P. Pracht, F. Bohle, S. Grimme, PCCP , 2020, 22, 7169—7192

[5] J. Gorges, S. Grimme, A. Hansen, P. Pracht, PCCP , 2022, 24, 12249-12259.