The melting temperature of the six site potential model of water.

The calculation by computer simulation of the melting temperature and the phase diagram of different water models has been proven a stringent test of the water force field. The ice Ih melting temperature Tm is systematically underestimated. Only a few models yield more or less satisfactory values for Tm. These are 3 TIP4P/ice 270 K and TIP5P Ref. 5 272 K . The former model also gives a good description of the phase diagram involving solid phases. On the contrary, TIP5P fails completely in this purpose. In fact, for TIP5P, ice Ih is not thermodynamically stable at 1 bar. The actual Tm for TIP5P should then correspond to the melting temperature of ice II which is several degrees higher. Another candidate is TIP4P/2005. This model accurately predicts the stability of the different ice forms and yields excellent predictions for the liquid-vapor coexistence line and critical properties. TIP4P/2005 seems an excellent model for many applications, but its Tm is rather low 251 K . However, some problems ice growth, brine rejection from ice, and nucleation are strongly sensitive to the degree of supercooling and, thus, require a more accurate prediction of the Tm. A first report of the Tm for the six site model of Nada and van der Eerden NvdE was close to the experiment. However, in a recent paper Nada and Furukawa arrived at a melting temperature in the range of 280–285 K. Then, before proceeding to a costly calculation of the phase diagram which is still unknown , it seems interesting to establish more firmly the Tm for this model. This is the goal of this paper. We have used two different methods for the calculation of the Tm for the NvdE potential. The first one is a combination of a Gibbs-Duhem integration and free energy calculations. Let us sketch here the procedure of the Gibbs-Duhem integration. First, we write the intermolecular potential in terms of a reference potential and a parameter ,