Benchmarking Density Functional Approximations in Nonadiabatic Dynamics: Trans-Cis Isomerization in Retinal Model.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Nonadiabatic dynamics, crucial for photochemistry, are challenging for time-dependent density functional theory (TD-DFT).
• Protonated Schiff base 3 (PSB3) photoisomerization presents a complex benchmark case for evaluating computational methods.
• Accurate modeling of excited-state processes requires robust theoretical frameworks.

Purpose of the Study:

• To conduct an exhaustive benchmark of various density functional approximations (DFAs) for nonadiabatic dynamics.
• To evaluate the performance of different DFA families against a high-level reference method for the PSB3 system.
• To identify suitable DFAs for future studies of nonadiabatic processes.

Main Methods:

• Development of a rigorous protocol for benchmarking DFAs in nonadiabatic dynamics, covering initialization, simulation, and evaluation.
• Comparison of multiple DFA families, including local functionals and those with high Hartree-Fock exchange, against RMS-CASPT2 reference data.
• Static energy scans along relevant torsion coordinates to analyze potential energy surfaces.

Main Results:

• Electronic populations are identified as an unreliable metric for assessing dynamics accuracy.
• Local functionals showed good population decay but followed an incorrect deactivation pathway.
• Functionals with 100% Hartree-Fock exchange predicted correct isomerization but with inaccurate timescales and quantum yields.
• Double hybrid functionals, particularly those balancing nonlocal exchange/correlation and employing range-separation, accurately reproduced the reference energy profiles.

Conclusions:

• Certain double hybrid density functional approximations show significant promise for accurate nonadiabatic dynamics simulations.
• The choice of DFA critically impacts the prediction of reaction pathways, timescales, and quantum yields in photoisomerization.
• Further development and implementation of analytical gradients for these advanced DFAs are crucial for advancing the field of nonadiabatic dynamics.