PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
February 25, 2026The Journal of Physical Chemistry Letters0 citations

UTDA-xDH: Accurate Doublet–Doublet Excitation Energies via XYG3-Type Doubly Hybrid Density Functionals within the Unrestricted Tamm–Dancoff Approximation

View Full Paper
QYQ. YangXXXin Xu

Key Points

  • The aim is to develop an accurate method for predicting doublet-doublet excitation energies in small radicals.
  • Developed the UTDA-xDH method combining XYG3-type xDH with UTDA.
  • Assessed three xDH methods: XYG3, XYGJ-OS, and xDH-PBE0.
  • Used a data set of doublet-doublet excitation energies for evaluation.
  • xDH methods achieved mean absolute deviations of 0.07-0.13 eV.
  • Outperformed conventional density functional theory methods.
  • Showed significantly reduced spin contamination compared to non-xDH methods.

Abstract

We develop UTDA-xDH, an excited-state method that combines the XYG3-type doubly hybrid density functional (xDH) with the unrestricted Tamm-Dancoff approximation (UTDA). Three xDH methods, XYG3, XYGJ-OS, and xDH-PBE0, are assessed against a data set of doublet-doublet excitation energies of small radicals. All three xDHs exhibit excellent accuracy and robustness for both valence and Rydberg excitations, with mean absolute deviations of 0.07-0.13 eV. These results not only outperform conventional density functional theory methods but also rival the unrestricted equation-of-motion coupled-cluster singles and doubles (EOM-CCSD) method. Remarkably, the assessed xDHs show substantially reduced spin contamination compared to non-xDH counterparts, enabling highly accurate predictions even for challenging cases such as the 2Σ+ state of CNO. These results support (U)TDA-xDH as an efficient, broadly applicable excited-state approach for open- and closed-shell systems.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Yang et al. (2026) studied this question.

synapsesocial.com/papers/699e927bf5123be5ed050367https://doi.org/10.1021/acs.jpclett.5c04097
Ask AI
Helpful
Bookmark
Share
View Full Paper