Use of Improved Orbitals for CCSD(T) Calculations for Predicting Heats of Formation of Group IV and Group VI Metal Oxide Monomers and Dimers and UCl₆

Abstract

The prediction of the heats of formation of group IV and group VI metal oxide monomers and dimers with the coupled cluster CCSD(T) method has been improved by using Kohn-Sham density functional theory (DFT) and Brueckner orbitals for the initial wave function. The valence and core-valence contributions to the total atomization energies for the CrO3 monomer and dimer are predicted to be significantly larger than when using the Hartree-Fock (HF) orbitals. The predicted heat of formation of CrO3 with CCSD(T)/PW91 is consistent with previous calculations including high-order corrections beyond CCSD(T) and agrees well with the experiment. The improved heats of formation with the DFT and Brueckner orbitals are due to these orbitals being closer to the actual orbitals. Pure DFT functionals perform slightly better than the hybrid B3LYP functional due to the presence of exact exchange in the hybrid functional. Comparable heats of formation for TiO2 and the second- and the third-row group IV and group VI metal oxides are predicted well using either the DFT PW91 orbitals, Brueckner orbitals, or HF orbitals. The normalized clustering energies for the dimers are consistent with our previous work except for a larger value predicted for Cr2O6. The prediction of the reaction energy for UF6 + 3Cl2 → UCl6 + 3F2 was significantly improved with the use of DFT or Brueckner orbitals as compared to HF orbitals.

In silico prediction of annihilators for triplet–triplet annihilation upconversion via auxiliary-field quantum Monte Carlo

“…53 A similar protocol has been shown to improve the accuracy of CC methods. 81 Fig . 3 presents adiabatic triplet energies obtained from KS-DFT, TD-DFT with three different representative functionals, DLPNO-CCSD(T), and AFQMC/U for the functionalized anthracenes shown in Fig.…”

In-Silico Prediction of Annihilators for Triplet Fusion Upconversion via Auxiliary-Field Quantum Monte Carlo

“…54 A similar protocol has been shown to improve the accuracy of CC methods. 79 Table 1: AFQMC results from various trial wavefunctions for the adiabatic triplet energy of benzonitrile, in eV. Parentheses denote statistical error of AFQMC, i.e.…”

Sensitivity of Coupled Cluster Electronic Properties on the Reference Determinant: Can Kohn-Sham Orbitals Be More Beneficial than Hartree-Fock Orbitals?

“…This Slater determinant is constructed from KS-DFT derived auxiliary Kohn-Sham one-electron orbitals 5 whose physical interpretation is considered somewhat vague 6 . Nevertheless, during the last two decades, a considerable number of KS-CC studies appeared in the literature [7][8][9][10][11][12][13][14][15][16] which repeatedly argued that the use of KS reference determinants is not only permitted by the CC equations, but it might even lead to improved results due to the decrease of the multi-reference character in some aspects.…”

“…Unfortunately, experimental data cannot be used for the assessment of accuracy in this case, as they are only available for metal oxide surfaces, rather than for separate dioxide molecules. In their subsequent works 13,14 , novel computational protocols were developed for the prediction of the heat of formation of transition metal hydrides, oxides, sulfides and chlorides, which treats the electronic energy at CCSD(T) level of theory. Again, it was observed that the use of PW91 or B3LYP orbitals instead of the usual HF orbitals decreases the T 1 diagnostic and also alters the energy (or the derived enthalpy) by up to several kcal/mol.…”

“…In this case, although experimental dissociation energies and enthalpies are abundantly available in the literature, most of these come with error bars exceeding the orbital choice related deviation. The Dixon group still managed to find one example where the B3LYP-KS reference appears to be unequivocally superior to HF, i.e., the enthalpy of the reaction of UCl 6 + 3F 2 → UF 6 + 3Cl 2 , experimentally measured to be 278.0±1.2 kcal/mol at room temperature, was predicted to be 285.9 kcal/mol and 277.8 kcal/mol by HF-CCSD(T) and KS-CCSD(T) based protocols, respectively 13 . Even though KS-CCSD(T) clearly produces better numerical agreement, assessing the quality of the computational models for the single benchmark is still challenging.…”