Double-hybrid density functionals with long-range dispersion corrections: higher accuracy and extended applicability.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Materials Science

Background:

• Double-Hybrid Density Functionals (DHDF) improve accuracy by incorporating non-local correlation effects.
• A known limitation of DHDF is the underestimation of long-range dispersion (van der Waals) interactions.
• Empirical dispersion corrections (DFT-D) can address this deficiency.

Purpose of the Study:

• To systematically improve the accuracy of DHDF by adding empirical dispersion corrections.
• To evaluate the performance of the corrected DHDF for various chemical properties.
• To propose a benchmark set for testing new density functionals.

Main Methods:

• Augmenting existing DHDF with a previously developed empirical dispersion term (DFT-D).
• Testing the composite approach on the S22 set for non-covalent interactions.
• Evaluating performance on the G3/99 set for heats of formation and peptide conformational energies.

Main Results:

• The corrected DHDF significantly increases accuracy for non-covalent interactions, achieving low Mean Absolute Deviations (MAD) on the S22 set.
• Unprecedented accuracy was obtained for challenging peptide conformation energies.
• Improvements were observed for the G3/99 set, highlighting the importance of intramolecular dispersion in large molecules.

Conclusions:

• The proposed composite approach using DHDF with dispersion corrections offers accuracy comparable to high-level coupled-cluster methods.
• Dispersion corrections are crucial for accurate DFT treatments, especially for larger molecules and non-covalent interactions.
• The corrected B2PLYP DHDF is recommended for general use due to its high accuracy.