The accuracy of quantum chemical methods for large noncovalent complexes

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Molecular Modeling

Background:

• Noncovalent interactions are crucial in molecular systems.
• Accurate computational description of these interactions is essential for understanding molecular behavior.
• Larger, dispersion-stabilized complexes pose a challenge for standard computational methods.

Purpose of the Study:

• To evaluate widely used computational methods for noncovalent interactions.
• To assess performance on larger, dispersion-dominated complexes (L7 dataset).
• To identify methods with optimal accuracy and computational cost.

Main Methods:

• Tested methods include wavefunction (MP2 variants), density functional theory (DFT-D, DFT-D3), and semiempirical approaches.
• Evaluated performance on diverse complexes like dimers and trimers of aromatic and biological molecules.
• Calculated relative root mean square deviation (rRMSD) to quantify accuracy.

Main Results:

• MP2.5 achieved the best performance with 4% rRMSD, serving as a benchmark alternative.
• MP2C showed strong performance (8% rRMSD) among non-DFT methods.
• BLYP-D3 offered the best accuracy/cost ratio within DFT (8% rRMSD).
• Semiempirical methods, despite lower accuracy (>25% rRMSD), presented excellent cost-performance.

Conclusions:

• MP2.5 is a recommended alternative to high-level benchmarks for large systems.
• DFT methods, particularly BLYP-D3, provide a good balance of accuracy and cost.
• Semiempirical methods are computationally efficient with surprisingly good absolute errors.