Investigation of Exchange Energy Density Functional Accuracy for Interacting Molecules

Area of Science:

• Computational chemistry
• Quantum chemistry
• Theoretical chemistry

Background:

• Generalized gradient approximation (GGA) functionals are widely used for approximating exchange energies in electronic structure calculations.
• Hartree-Fock (HF) calculations provide a baseline for evaluating the accuracy of approximate methods.
• Accurate calculation of interaction energies is crucial for understanding molecular behavior and chemical reactions.

Purpose of the Study:

• To compare exchange-only interaction energies from various GGA exchange functionals against Hartree-Fock results.
• To identify GGA functionals that best reproduce HF interaction energies for interacting molecules.
• To investigate the impact of the density gradient dependence on the accuracy of exchange functionals.

Main Methods:

• Calculated exchange-only interaction energies for interacting molecules using several standard GGA exchange functionals.
• Compared the obtained energies with reference Hartree-Fock calculations.
• Analyzed the performance of functionals based on their enhancement factor and dependence on the density gradient.

Main Results:

• Functionals incorporating a 2/5 power dependence on the density gradient, particularly for large gradients, demonstrated superior agreement with HF results.
• These functionals consistently outperformed alternative GGA functionals in reproducing HF exchange-only interaction energies.
• Recalculation of the best-performing functional with its exact large gradient dependence further refined the agreement.

Conclusions:

• The 2/5 power dependence on the density gradient is a key feature for accurate GGA exchange functionals in describing molecular interactions.
• Specific GGA functionals with appropriate gradient dependence can effectively approximate Hartree-Fock exchange-only interaction energies.
• Further refinement of exchange functionals should consider precise modeling of large density gradients.