Accurate Water Properties from an Efficient ab Initio Method.

Accurate prediction of the water properties from a low-cost ab initio method is still a foremost problem for chemists and physicists. Although density functional approaches starting from semilocal to hybrid exchange-correlation functionals are tested, they are not efficiently performing for all the properties together, especially considering energies, conformal ranking, structures, and dynamics of water. Also, the inclusion of the long-range van der Waals interaction does not improve the ordering stability of isomers. However, relying on the simple revision of the Tao-Mo (revTM) semilocal meta-generalized gradient approximation exchange-correlation functional, we demonstrate that many properties of the water clusters and ice phases can be accurately predicted. A consistent improvement over several popular ab initio methods is achieved, indicating the accuracy of this method for describing hydrogen bonding of water.