Intermolecular interactions in the condensed phase: Evaluation of semi-empirical quantum mechanical methods

Area of Science:

• Computational Chemistry
• Quantum Mechanics
• Physical Chemistry

Background:

• Approximate quantum mechanical (QM) methods are crucial for modeling condensed phase systems.
• Existing QM methods are often parameterized and validated using gas-phase data, limiting their accuracy in solution.
• Developing accurate QM methods for condensed phase applications requires robust benchmark datasets.

Purpose of the Study:

• To create a benchmark dataset of intermolecular interaction energies for charged dimers in solution.
• To evaluate the performance of various semi-empirical quantum mechanical (SQM) methods using this new dataset.
• To identify SQM methods that perform well in condensed phase calculations and guide future method development.

Main Methods:

• A thermodynamic cycle was used to compute reference interaction energies in solution.
• High-level coupled cluster (CC) calculations were employed for gas-phase dimer binding energies.
• Density functional theory (DFT) was used to calculate solute-solvent interactions.
• Performance of DFTB3-D3, DFTB3/CPE-D3, OM2-D3, PM6-D3, PM6-D3H+, PM7, and HF-3c methods was benchmarked.

Main Results:

• All tested SQM methods generally underestimated gas-phase binding energies (RMSE 2-5 kcal/mol).
• Most SQM methods overestimated solution-phase binding energies (RMSE 3-4 kcal/mol), except for DFTB3/CPE-D3 and OM2-D3.
• HF-3c consistently overestimated binding energies in both gas and solution phases.
• The new dataset revealed systematic deviations in SQM method performance for condensed phase interactions.

Conclusions:

• The developed dataset is vital for calibrating QM methods for condensed phase applications.
• Accurate treatment of polarization and exchange repulsion is critical for condensed phase QM calculations.
• DFTB3/CPE-D3 and OM2-D3 show promising performance for charged systems in solution compared to other SQM methods.