In this article we present binding energy results for a scaled MP2.5 method along with several basis sets. The accuracy of the methods tested here is determined by comparison to reference data in the newly developed S66 data set of interaction energies. It is found that scaling of the MP2.5 correction term results in strongly improved binding energies for small basis sets, such as 6-31G* and 6-311++G**. The scaling parameter for larger basis sets, such as aug-cc-pVDZ and aug-cc-pVTZ, approaches a value of 0.50, which corresponds to the original (unparametrized) MP2.5 method. It is found that the MP2.X method yields S66 RMS errors of approximately 0.15 kcal mol(-1) for all basis sets. In addition to generally providing more accurate binding energies than MP2.5, MP2.X generally produces a more well-balanced description of noncovalent interactions, yielding better binding energies for dispersion-bound and mixed dispersion-electrostatic complexes.

Institute of Organic Chemistry and Biochemistry Academy of Sciences of the Czech Republic 166 10 Prague Czech Republic

References provided by Crossref.org

The accuracy of quantum chemical methods for large noncovalent complexes