Accountability of Modern Adiabatic Connection-Based Double-Hybrids Constructed from Cubic and Quadratic Integrand Functions for Dipole Polarizabilities of Water Nanoclusters

In this work, we dissect the performance of two modern Perdew-Burke-Ernzerhof (PBE)-based double-hybrid (DH) density functionals to predict the isotropic and anisotropic polarizabilities of water nanoclusters (H2O)n [n = 2-6]. The considered models include the cubic integrand (CI) and quadratic integrand (QI) functions as well as the spin-opposite-scaled (SOS) scheme for perturbative correlation term. It is shown that all the tested CIDHs and QIDHs underestimate the isotropic polarizabilities, while in the case of anisotropic polarizabilities there is also overestimation of data for a few of nanoclusters when employing the PBE-QIDH-OS model. Putting all the results together, the recommended DH functionals for predicting the dipole polarizabilities of water nanoclusters turned out to be PBE-CIDH, PBE-CIDH-OS, PBE-QIDH, and PBE0-DH with deviations smaller than those provided from Møller-Plesset perturbation calculations.