Empirical D3 Dispersion as a Replacement for ab Initio Dispersion Terms in Density Functional Theory-Based...

In density functional theory-based symmetry-adapted perturbation theory (DFT-SAPT) interaction energy calculations, the most demanding step is the calculation of the London dispersion term. For this bottleneck to be avoided and DFT-SAPT to be made applicable to larger systems, the ab initio dispersion terms can be replaced by one calculated empirically at an almost negligible cost ( J. Phys. Chem. A 2011 ; 115 , 11321 - 11330 ). We present an update of this approach that improves accuracy and makes the method applicable to a wider range of systems. It is based on Grimme's D3 dispersion correction for DFT, where the damping function is changed to one suitable for the calculation of the complete dispersion energy. The best results have been achieved with the Tang-Toennies damping function. It has been parametrized on the S66×8 data set for which we report density fitting DFT-SAPT/aug-cc-pVTZ interaction energy decomposition. The method has been validated on a diverse set of noncovalent systems including difficult cases such as very compact noncovalent complexes of charge-transfer type. The root-mean-square errors in the complete test set are 0.73 and 0.42 kcal mol^-1 when charge-transfer complexes are excluded. The proposed empirical dispersion terms can also be used outside the DFT-SAPT framework, e.g., for the estimation of the amount of dispersion in a calculation where only the total interaction energy is known.