Equation of state of water based on the SCAN meta-GGA density functional.

Based on the newly developed SCAN meta-GGA and the widely used PBE-GGA functionals, ab initio molecular dynamics are performed on water. It is proved that, although the SCAN meta-GGA is not as good as the TIP4P/2005 model potential in describing the equation of state of water, it is much better than the PBE-GGA, the ST2 model potential, and ab initio trained neural network potentials. Moreover, the SCAN meta-GGA predicts a first-order liquid-liquid transition from high- to low-density water at negative pressure, in which the structures are qualitatively consistent with experimental observations, and the spinodal point of high-density water is very close to Speedy's stability limit line.