We present the GMTKN55 benchmark database for general main group thermochemistry, kinetics and noncovalent interactions. Compared to its popular predecessor GMTKN30 [Goerigk and Grimme J. Chem. Theory Comput., 2011, 7, 291], it allows assessment across a larger variety of chemical problems-with 13 new benchmark sets being presented for the first time-and it also provides reference values of significantly higher quality for most sets. GMTKN55 comprises 1505 relative energies based on 2462 single-point calculations and it is accessible to the user community via a dedicated website. Herein, we demonstrate the importance of better reference values, and we re-emphasise the need for London-dispersion corrections in density functional theory (DFT) treatments of thermochemical problems, including Minnesota methods. We assessed 217 variations of dispersion-corrected and -uncorrected density functional approximations, and carried out a detailed analysis of 83 of them to identify robust and reliable approaches. Double-hybrid functionals are the most reliable approaches for thermochemistry and noncovalent interactions, and they should be used whenever technically feasible. These are, in particular, DSD-BLYP-D3(BJ), DSD-PBEP86-D3(BJ), and B2GPPLYP-D3(BJ). The best hybrids are ωB97X-V, M052X-D3(0), and ωB97X-D3, but we also recommend PW6B95-D3(BJ) as the best conventional global hybrid. At the meta-generalised-gradient (meta-GGA) level, the SCAN-D3(BJ) method can be recommended. Other meta-GGAs are outperformed by the GGA functionals revPBE-D3(BJ), B97-D3(BJ), and OLYP-D3(BJ). We note that many popular methods, such as B3LYP, are not part of our recommendations. In fact, with our results we hope to inspire a change in the user community's perception of common DFT methods. We also encourage method developers to use GMTKN55 for cross-validation studies of new methodologies.

The development of van der Waals density functional approximations (vdW-DFAs) has gained considerable interest over the past decade. While in a strictest sense, energy calculations with vdW-DFAs should be carried out fully self-consistently, we demonstrate conclusively for a total of 11 methods that such a strategy only increases the computational time effort without having any significant effect on energetic properties, electron densities, or orbital-energy differences. The strategy to apply a nonlocal vdW-DFA kernel as an additive correction to a fully converged conventional DFA result is therefore justified and more efficient. As part of our study, we utilize the extensive GMTKN55 database for general main-group thermochemistry, kinetics, and noncovalent interactions [Phys. Chem. Chem. Phys. 2017, 19, 32184], which allows us to analyze the very promising B97M-V [J. Chem. Phys. 2015, 142, 074111] and ωB97M-V [J. Chem. Phys. 2016, 144, 214110] DFAs. We also present new DFT-D3­(BJ) based counterparts of these two methods and of ωB97X-V [J. Chem. Theory Comput 2013, 9, 263], which are faster variants with similar accuracy. Our study concludes with updated recommendations for the general method user, based on our current overview of 325 dispersion-corrected and -uncorrected DFA variants analyzed for GMTKN55. vdW-DFAs are the best representatives of the three highest rungs of Jacob’s Ladder, namely, B97M-V, ωB97M-V, and DSD-PBEP86-NL.

Previously, we introduced DFT-D3(BJ) variants of the B97M-V, ωB97X-V and ωB97M-V functionals and assessed them for the GMTKN55 database [Najibi and Goerigk, J Chem. Theory Comput. 2018, 14, 5725]. In this study, we present DFT-D4 damping parameters to build the DFT-D4 counterparts of these functionals and assess these in comparison. We extend our analysis beyond GMTKN55 and especially turn our attention to enzymatically catalyzed and metal-organic reactions. We find that B97M-D4 is now the secondbest performing meta-generalized-gradient approximation functional for the GMTKN55 database and it can provide noticeably better organometallic reaction energies compared to B97M-D3(BJ). Moreover, the aforementioned DFT-D3(BJ)-based functionals have not been thoroughly assessed for geometries and herein we close this gap by analyzing geometries of noncovalently bound dimers and trimers, peptide conformers, water hexamers and transition-metal complexes. We find that several of the B97(M)-based methods-particularly the DFT-D4 versions-surpass the accuracy of previously studied methods for peptide conformer, water hexamer, and transition-metal complex geometries, making them safe-to-use, cost-efficient alternatives to the original methods. The DFT-D4 variants can be easily used with ORCA4.1 and above.

We investigate the effects of range separation of the exchange energy on electronic ground-state properties for recently published double-hybrid density functionals (DHDFs) with the extensive GMTKN55 database for general main-group thermochemistry, kinetics and noncovalent interactions. We include the semi-empirical range-separated DHDFs ωB2PLYP and ωB2GP-PLYP developed by our group for excitation energies, together with their ground-state-parametrized variants, which we denote herein as ωB2PLYP18 and ωB2GP-PLYP18. We also include the non-empirical range-separated DHDFs RSX-0DH and RSX-QIDH. For all six DHDFs, damping parameters for the DFT-D3 dispersion correction (and for its DFT-D4 variant) are presented. We comment on when the range-separated functionals can be more beneficial than their global counterparts, and conclude that range separation alone is no guarantee for overall improved results. We observe that the BLYP-based functionals generally outperform the PBE-based functionals. We finally note that the best-performing double-hybrid density functionals for GMTKN55 are still the semi-empirical range-separated double hybrids ωDSD3-PBEP86-D4 and ωDSD72-PBEP86-D4, the former of which includes a third-order perturbative correlation term in addition to the more conventional second-order perturbation that DHDFs are based upon. File list (2)download file view on ChemRxiv rsdhdfs_GMTKN55.pdf (816.36 KiB) download file view on ChemRxiv rsdhdfs_GMTKN55_SI.pdf (191.84 KiB)

Orbital optimization (OO) has been suggested as a way to solve some shortcomings of second-order Møller-Plesset (MP2) variants and double-hybrid density functionals (DHDFs). A closer inspection of the literature, however, shows that the only two studies on OO-DHDFs were limited to three nonempirical PBE-based functionals, which are known to be of only mediocre accuracy. Herein, we provide a more in-depth analysis of OO-DHDFs with the main focus being on main-group thermochemistry, kinetics, and noncovalent interactions. We reanalyze two PBE-based OO-DHDFs and present four new OO-DHDF variants, two of which make use of the spin-component-scaling idea in their nonlocal correlation part. We also provide a more thorough analysis of three OO-MP2 variants. After assessing more than 621 reference points, we come to the conclusion that the benefits of OO are not as straightforward as previously thought. Results heavily depend on the underlying parent method. While OO-SCS/SOS-MP2 usually provide improved results-including for noncovalently bound systems-the opposite is true for OO-MP2. OO-DHDFs, like their nonoptimized counterparts, still require London-dispersion corrections. Among the DHDFs, the largest effect of OO on thermochemical properties is seen for PBE0-2 and the smallest for PBE0-DH. However, results can both worsen and improve with OO. If the latter is the case, the resulting OO-DHDF is still outperformed by the currently most accurate conventional DHDFs, namely DSD-BLYP and DSD-PBEP86. We therefore recommend the OO technique only to be used in specialized cases. For the general method user we re-emphasize using conventional dispersion-corrected DHDFs for robust, reliable results. Our findings also indicate that entirely different strategies seem to be required in order to obtain a substantial improvement over the currently best DHDFs.