Spin-Scaled Range-Separated Double-Hybrid Density Functional Theory for Excited States

Abstract

Our recently presented range-separated (RS) double-hybrid (DH) time-dependent density functional approach [J. Chem. Theory Comput. 17, 927 (2021)] is combined with spin-scaling techniques. The proposed spin-component-scaled (SCS) and scaled-opposite-spin (SOS) variants are thoroughly tested for almost 500 excitations including the most challenging types. This comprehensive study provides useful information not only about the new approaches but also about the most prominent methods in the DH class. The benchmark calculations confirm the robustness of the RS-DH ansatz, while several tendencies and deficiencies are pointed out for the existing functionals. Our results show that the SCS variant consistently improves the results, while the SOS variant preserves the benefits of the original RS-DH method reducing its computational expenses. It is also demonstrated that, besides our approaches, only the nonempirical functionals provide balanced performance for general applications, while particular methods are only suggested for certain types of excitations.

Assessing recent time-dependent double-hybrid density functionals on doublet-doublet excitations

“…22 By ignoring the same-spin part, SOS-based methods can achieve an improved formal scaling behavior, which makes them particularly valuable for larger systems. 23 The superior performance of some of the fourteen developed SCS/SOSbased TD(A)-DHDFs has been demonstrated for singlet-singlet and singlet-triplet excitations with the best methods outperforming our original 2019 approaches, as well as some alternative DHDF definitions published in 2021, 34,35 which also employ range-separation in the perturbative part. 20 Nearly all benchmark studies exploring TD(A)-DHDFs dealt with closed-shell systems and assessments of open-shell systems have been rare and very limited in scope.…”

Thermal Fluctuations in Conjugation and their Effect on Calculated Excitation Energies: A Case Study on the Astaxanthin Carotenoid

“…Moreover, it should be pointed out that an effect of ~0.05 eV is rather large in relation to the "chemical accuracy" threshold often sought for calculated excitation energies, [90] which by some researchers is set at 1 kcal mol À 1 � 0.04 eV, [5,11] whereas others prefer 0.1 eV. [106] At the same time, the effect is small compared to the average errors that all density functionals, including accurate double hybrids ones, [90,[107][108][109] show for excitation energies. Thus, the effect is not likely to be the main source of error when such calculations are performed in a static fashion.…”

Time-Dependent Long-Range-Corrected Double-Hybrid Density Functionals with Spin-Component and Spin-Opposite Scaling: A Comprehensive Analysis of Singlet-Singlet and Singlet-Triplet Excitation Energies

“…for ΔE CIS and two different versions of SCS-CIS(D) were developed by setting λ=0 and λ=1. Herein, we set this parameter to unity following the same protocol as in Refs 34 and 39. As this final version of our study now also includes the new SCS/SOS-RS-PBE-P86 methods, 85 we would like to point out that theoretical details on those are not clear regarding the application of SCS/SOS to the CIS(D) part. As the authors did not acknowledge Rhee and Head-Gordon's SCS-CIS(D) approach and did not use the phrases "direct" and "indirect" term, it is safe to assume that both terms were subjected to the same scaling parameters.…”

“…84 During the peer-review stage of our work, a related article by Mester and Kállay was published in this very same journal. 85 Therein, the authors apply the SCS and SOS ideas to their RS-PBE-P86 83 functional. Those two new functionals differ from our herein proposed methods in two ways.…”