Fig. 4: The Surf13 benchmark for lower-level theories.
From: An accurate and efficient framework for modelling the surface chemistry of ionic materials
The autoSKZCAM framework interaction energy benchmarks are used to assess a selection of functionals along the hierarchy of DFAs, from the GGA to the meta-GGA (mGGA) and hybrid functionals, as well as the RPA with and without the rSE. The deviation from the autoSKZCAM estimate is given as a colour map, with red and blue indicating overbinding and underbinding, respectively. We consider a range of molecules physisorbed on the MgO(001), rutile (r) TiO2(110) and anatase (a) TiO2(101) surfaces. The mean absolute deviations (MADs) across all of the systems (labelled ‘overall’), as well as the subsets involving the MgO(001) and TiO2 surfaces, are given in grey in the right panel. We indicate the typical ‘chemical accuracy’ of 43 meV in yellow on the colour bar. The autoSKZCAM and DFT interaction energies are tabulated in Supplementary Section 2.
