Understanding Density-Driven Errors for Reaction Barrier Heights | JoVE Visualize

Area of Science:

Computational Chemistry
Quantum Chemistry
Materials Science

Background:

Density functional approximations (DFAs) are computationally efficient but suffer from delocalization errors like charge-transfer and self-interaction errors.
Evaluating a semilocal DFA non-self-consistently on the Hartree-Fock (HF) density (HF-DFT) is a common strategy to mitigate these errors, particularly for meta-GGAs like SCAN.
HF-DFT often improves accuracy, presumed due to the HF density being more reliable than the self-consistent DFA density.

Purpose of the Study:

To investigate the underlying reasons for the success of HF-DFT in improving barrier height calculations.
To analyze the nature of errors, specifically charge-transfer errors, in HF-DFT using density-corrected density functional theory (DFT) metrics.
To explore the relationship between density-driven and functional-driven errors in DFT calculations.

Main Methods:

Applied density-corrected DFT metrics to analyze HF-DFT calculations of barrier heights.
Quantitatively analyzed charge-transfer errors in selected transition states.
Employed three nonlocal proxy functionals (SCAN 50% global hybrid, LC-ωPBE, SCAN-FLOSIC) with their self-consistent densities as proxies for the exact functional and density.

Main Results:

HF-DFT improves barrier heights by introducing a localizing charge-transfer error, effectively canceling density- and functional-driven errors.
Analysis of proxy functionals showed that density-driven errors in self-consistent DFT are typically second-order in density error.
Large density-driven errors originate from electron transfers over atomic length scales.

Conclusions:

HF-DFT offers a computationally inexpensive yet effective remedy for delocalization errors in DFAs, particularly for barrier height calculations.
The improvement stems from a beneficial cancellation of errors, where a localizing charge-transfer error compensates for functional-driven inaccuracies.
Understanding density-driven errors is crucial for developing more accurate DFT functionals.