What is the best density functional to describe water clusters: evaluation of widely used density functionals with various basis sets for (H2O) n (n = 1–10)

Abstract

The geometric structures, stabilization energies, dipole moments, and vibrational frequencies of the neutral water clusters (H2O)n, with n = 1–10, were investigated using density functional theory along with a variety of exchange-correlation functionals (LDA with SVWN5 parameterization, GGA with BLYP, PW91, PBE, B3LYP, X3LYP, PBE0, PBE1W, M05-2X, M06-2X and M06-L parameterizations) as well as high-level ab initio MP2 and CCSD(T) methods. Using the MP2 and CCSD(T) results as benchmarks, the effects of exchange-correlation functionals and basis sets were carefully examined. Each functional has its advantage in certain aspects; for example, M05-2X and X3LYP yield better geometries, and the capability of these two functionals to distinguish the relative energies between isomers are more similar to MP2. The size of the split-valence basis set (6-31G or larger), diffuse functions on the oxygen atom, and d(p) polarization on the oxygen (hydrogen) atom are crucial for an accurate description of intermolecular interaction in water clusters. The 6-31+G(2d,p) basis set is thus recommended as a compromise between computational efficiency and accuracy for structural description. We further demonstrated that the numerical basis set, TNP, performs satisfactorily in describing structural parameters of water clusters.