Order of stabilities in water nanoclusters: Insight from some recent double‐hybrid functionals

Abstract

In the present work, the applicability of some of the recently proposed and modern double‐hybrid (DH) models and other density functional theory (DFT) approximations has been analyzed for a difficult test, the order of stability in low‐energy isomers of water nanoclusters. In particular, we aim to systematically investigate for these functionals the role played by several factors such as dispersion correction, integrand functions upon which the DHs are based, and different spin scaling for the perturbative term in DH calculations of the relative energies for various isomers of water nanoclusters (H2O)20. From the obtained results, the superior performance of DHs with respect to the functionals from previous rungs is confirmed. It is shown that the dispersion corrected DHs perform better than noncorrected counterparts. Plus, the DH models based on cubic integrand (CI) and quadratic integrand (QI) functions are nearly equivalent in performance. We also find that using only contributions of electron pairs with opposite spin for the perturbative correlation part through scaled opposite spin scheme does not represent a significant improvement on accuracy of the results. Putting all the results together, the dispersion corrected parameterized DHs and parameter‐free DH models involving CI and QI functions outperform other approximations for relative energies of water 20‐mers. Altogether, predicting the correct order of the stability in water nanoclusters may be considered as another Achilles' heel in DFT calculations, although more analyses in this context are still needed. © 2016 Wiley Periodicals, Inc.