Comparative Study of Selected Wave Function and Density Functional Methods for Noncovalent Interaction Energy Calculations Using the Extended S22 Data Set

In this paper, an extension of the S22 data set of Jurecka et al. ( Jurečka , P. ; Šponer , J. ; Černý , J. ; Hobza , P. Phys. Chem. Chem. Phys. 2006 , 8 , 1985. ), the data set of benchmark CCSD(T)/CBS interaction energies of twenty-two noncovalent complexes in equilibrium geometries, is presented. The S22 data set has been extended by including the stretched (one shortened and three elongated) complex geometries of the S22 data set along the main noncovalent interaction coordinate. The goal of this work is to assess the accuracy of the popular wave function methods (MP2-, MP3- and, CCSD-based) and density functional methods (with and without empirical correction for the dispersion energy) for noncovalent complexes based on a statistical evaluation not only in equilibrium, but also in nonequilibrium geometries. The results obtained in this work provide information on whether an accurate and balanced description of the different interaction types and complex geometry distortions can be expected from the tested methods. This information has an important implication in the calculation of large molecular complexes, where the number of distant interacting molecular fragments, often in far from equilibrium geometries, increases rapidly with the system size. The best performing WFT methods were found to be the SCS-CCSD (spin-component scaled CCSD, according to Takatani , T. ; Hohenstein , E. G. ; Sherrill , C. D. J. Chem. Phys. 2008 , 128 , 124111 ), MP2C (dispersion-corrected MP2, according to Hesselmann , A. J. Chem. Phys. 2008 , 128 , 144112 ), and MP2.5 (scaled MP3, according to Pitoňák , M. ; Neogrády , P. ; Černý , J. ; Grimme , S. ; Hobza , P. ChemPhysChem 2009 , 10 , 282. ). Since none of the DFT methods fulfilled the required statistical criteria proposed in this work, they cannot be generally recommended for large-scale calculations. The DFT methods still have the potential to deliver accurate results for large molecules, but most likely on the basis of an error cancellation.