Benchmarking Quantum Mechanical Levels of Theory for Valence Parametrization in Force Fields

Area of Science:

• Computational chemistry
• Quantum chemistry
• Molecular modeling

Background:

• Accurate molecular electronic structure calculations are vital for developing efficient force fields.
• Quantum mechanical (QM) methods are computationally expensive for condensed-phase simulations.
• Parametrizing general force fields requires large QM datasets, necessitating cost-effective QM methods.

Purpose of the Study:

• To evaluate the accuracy of various density functional theory (DFT) methods and basis sets.
• To determine the suitability of different QM approaches for generating molecular data.
• To identify QM methods that balance computational cost and accuracy for force field development.

Main Methods:

• Assessed 59 diverse small molecules.
• Compared approximately 25 combinations of DFT functionals and basis sets.
• Validated against coupled cluster (CC) calculations at the complete basis set (CBS) limit.

Main Results:

• Identified specific DFT functional/basis set combinations that accurately predict conformer and torsion energies.
• Quantified the accuracy of various DFT methods relative to high-level CC/CBS calculations.
• Provided insights into the performance of DFT for generating QM data.

Conclusions:

• Certain DFT methods offer a good balance of accuracy and computational efficiency for generating molecular data.
• The findings guide the selection of QM methods for large-scale force field parametrization.
• This work aids in the development of more accurate and efficient molecular simulations.