Assessments of DFT-based energy decomposition analysis methods for intermolecular interactions

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Energy Decomposition Analysis (EDA) is crucial for understanding intermolecular interactions.
• Comparing different EDA methods is essential for accurate theoretical predictions.
• Existing methods like GKS-EDA, ETS-EDA, and DFT-SAPT have distinct theoretical underpinnings.

Purpose of the Study:

• To extensively assess and compare the results of three EDA methods: GKS-EDA, ETS-EDA, and DFT-SAPT.
• To analyze the performance of these methods across various intermolecular interaction types.
• To validate the consistency of EDA term categorizations (electrostatics, exchange-repulsion, polarization, dispersion/correlation).

Main Methods:

• Utilized three distinct EDA methods: Generalized Kohn-Sham (GKS) EDA, Extended Transition State (ETS) EDA, and Density Functional Theory Symmetry-Adapted Perturbation Theory (DFT-SAPT).
• Applied these methods to a large dataset of 1092 non-covalent interaction complexes from standard benchmark sets (S66, PNICO23, HAL59, IL16, S66 × 8, X40 × 10).
• Grouped EDA terms into four physically meaningful categories: electrostatics, exchange-repulsion, polarization/induction, and correlation/dispersion (CD).

Main Results:

• The results from the three EDA methods were found to be comparable across different basis sets and computational platforms.
• Excellent agreement was observed for electrostatics, exchange-repulsion, and polarization terms among the methods.
• The correlation/dispersion (CD) term in GKS-EDA showed good comparability with the dispersion term in DFT-SAPT.

Conclusions:

• GKS-EDA, ETS-EDA, and DFT-SAPT provide consistent and comparable insights into intermolecular interactions.
• GKS-EDA serves as a valuable alternative for investigating electronic correlations within DFT frameworks.
• The study validates the physical interpretations and categorization of EDA terms across different computational approaches.