Do Optimally Tuned Range-Separated Hybrid Functionals Require a Reparametrization of the Dispersion Correction? It...

Area of Science:

• Computational chemistry
• Quantum chemistry
• Theoretical chemistry

Background:

• Range-separated hybrid functionals (RSHs) combined with dispersion corrections are state-of-the-art for molecular studies.
• Optimal tuning of the range-separation parameter (ω) improves excited-state accuracy but may affect dispersion parameterization.

Purpose of the Study:

• To investigate the interdependency between dispersion correction parameters and the range-separation parameter (ω) in RSHs.
• To assess the robustness of dispersion models across various ω values for different RSHs.

Main Methods:

• Refitted the DFT-D4 dispersion model for six established RSHs.
• Explored a wide range of ω values (0.05–0.45 a₀^-1).
• Used a set of noncovalently bound molecular complexes for parameterization and testing.

Main Results:

• PBE-based RSHs and ωB97M-D4 showed weak interdependency and robust performance across ω values.
• B88-based RSHs, particularly LC-BLYP, exhibited strong sensitivity to ω.
• Reduced ω in B88-based RSHs led to systematic overbinding and degraded performance.

Conclusions:

• The choice of RSH significantly impacts the robustness of dispersion corrections with respect to ω tuning.
• Strategies are needed to mitigate sensitivity issues in certain RSHs for accurate molecular simulations.
• Future work should focus on improved D4 parameterization and optimization algorithms.