A low-cost composite variant of the r2SCAN density functional approximation

The recently proposed second revision of the SCAN meta density functional approximation (DFA) [Furness et al., J. Phys. Chem. Lett. 2020, 11, 8208−8215, termed r2SCAN] is used to construct an efficient composite electronic-structure method termed r2SCAN-3c, in analogy to the well established PBEh-3c and B97-3c approaches. It combines the unmodified r2SCAN functional with a specifically adapted def2-TZVP Gaussian AO-basis, the D4 dispersion model and a re-adjusted gCP correction for inter- and intramolecular BSSE. Due to the inherently higher accuracy of r2SCAN over many other (m)GGAs, further empirical adjustments for covalent bond lengths as in B97-3c are not required. The performance of r2SCAN-3c is evaluated comprehensively on several thousand energy evaluations, i.e., the GMTKN55 thermochemical database, extended benchmarks for non-covalent interactions (e.g., S30L, HB300SPXx10), organometallic reaction energies (MOR41), lattice energies of organic molecules (DMC8, X23) and ices (ICE10), as well as for the adsorption of small molecules on apolar coinage metals and polar surfaces. These tests reveal spectacularly accurate and consistent reaction energies and non-covalent interactions in molecular and periodic systems, as well as particularly outstanding conformational energies. For a suprisingly wide range of chemistry, r2SCAN-3c competes with hybrid/QZ-based methods at a small fraction (factor of about 100) of computational cost, surpassing the accuracy of its already efficient predecessor B97-3c at only twice the cost. As other (m)GGAs, r2SCAN-3c suffers from effects of the SIE but this is notably reduced as indicated by smaller artificial charge-transfer in various complexes. We thus chose this numerically robust method as our new group default in standard DFT applications for medium to large systems or large scale screening projects.