Dispersion Corrected r2SCAN Based Global Hybrid Functionals: r2SCANh, r2SCAN0, and r2SCAN50

Abstract

The re-regularized semilocal meta generalized gradient approximation (meta-GGA) exchange-correlation functional r2SCAN [J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, J. Phys. Chem. Lett. 11, 8208–8215 (2020)] is used to create three global hybrid functionals with varying admixtures of Hartree–Fock ”exact” exchange (HFX). The resulting functionals r2SCANh (10% HFX), r2SCAN0 (25% HFX), and r2SCAN50 (50% HFX) are combined with the semi-classical D4 London dispersion correction. The new functionals are assessed for the calculation of molecular geometries, main-group, and metalorganic thermochemistry at 26 comprehensive benchmark sets. These include the extensive GMTKN55 database, ROST61, and IONPI19 sets. It is shown that a moderate admixture of HFX leads to relative improvements of the mean absolute deviations (MADs) for thermochemistry of 11% (r2SCANh-D4), 16% (r2SCAN0-D4), and 1% (r2SCAN50-D4) compared to the parental semi-local meta-GGA. For organometallic reaction energies and barriers, r2SCAN0-D4 yields an even larger mean improvement of 35%. The computation of structural parameters (geometry optimization) does not systematically profit from HFX admixture. Overall, the best variant r2SCAN0-D4 performs well for both main-group and organometallic thermochemistry and is better or on par with well-established global hybrid functionals such as PW6B95-D4 or PBE0-D4. Regarding systems prone to self-interaction errors (SIE4x4), r2SCAN0-D4 shows reasonable performance, reaching the quality of the range-separated wB97X-V functional. Accordingly, r2SCAN0-D4 in combination with a sufficiently converged basis set (def2-QZVP(P)) represents a robust and reliable choice for general use in the calculation of thermochemical properties of both, main-group and organometallic chemistry.

Evaluating Noncovalent Interactions in Halogenated Molecules with Double-Hybrid Functionals and a Dedicated Small Basis Set

“…It is also interesting to note that the D4-ATM correction leads to increasing of the MAE on the X40 × 10 set, at least for the PBE0 functional. 66 Indeed, our value for PBE0/def2-TVPP of 0.24 kcal/mol is lower than that obtained at the PBE0-D4-ATM/def2-QZVP level (0.38 kcal/mol). It is difficult to dissect the origin of this discrepancy (D4 potential, ATM correction or basis set), but it was already suggested that the inclusion of three-body effects on top of pairwise potentials could lead to an overestimation of the interaction energies.…”

“…A slightly higher error has been obtained for PBE0-D4-ATM/def2-QZVP (0.38 kcal/ mol). 66 This last functional, PBE0, gives an MAE of 0.18 kcal/ when specific machine learning corrections are applied. 67 In both cases, a significant effort has been done in the definition of the model parameters to obtain a result that, at most, is 0.02 kcal/mol lower than that provided by our PBE-QIDH/DH-SVPD approach.…”

A computational investigation on the adsorption behavior of bromoacetone on B 36 borophene nanosheet

“…Different functionals of the Head-Gordon group [41] were also performed such as B97M, ωB97X, ωB97M [42] in combination with the D4 dispersion correction, and using the modi ed triple-ζ basis sets def2-mTZVP [43]. Further, in order to predict and compare behavior of interaction between BCT and B 36 , we include the recent r2SCAN-3c composite method [44], based on the r2SCAN [45,46] meta-GGA combined with the D4 dispersion correction [47] and geometrical counter poise-correction [48].…”

Working with benchmark datasets in the Cuby framework

“…For example, developments of new density functional theory (DFT) functionals involve benchmarking of tens of computational methods, often implemented in different software packages, applied to tens of datasets, resulting in hundreds of thousands of individual calculations. [3][4][5][6] Automated benchmarking is also useful on a smaller scale, allowing more application-oriented computational chemists to validate their methods on smaller subsets of data covering their topic of interest.…”