The thermal conductivities of a series of molten alkali halides have been evaluated by using molecular dynamics simulation within the framework of Fumi-Tosi potential models. Although the calculated results showed 0%-50% larger values than experimental results depending on system, they are in agreement with each other in showing both negative temperature and ionic mass dependence. In order to clarify the cause of the negative temperature dependence in more detail, the thermal conductivity under constant temperature or constant density was evaluated for all alkali chlorides and all sodium halides. The calculations reveal that the thermal conductivity depends strongly on density but only weakly on temperature. While the integrated value of the autocorrelation function for energy current increases with temperature, this is canceled out by the reciprocal temperature factor in relation to the thermal conductivity. With increasing density the integrated value increases, and this dominates the behavior of the thermal conductivity. By repeating the calculations with different ionic masses, we have concluded that the thermal conductivity is a function of m(-1/2)(N/V)(2/3), where m is the geometric mean of ionic mass between anion and cation and N/V is the number density.

The thermal conductivity of rare gases in liquid and dense fluid states has been evaluated using molecular dynamics simulation with the Lennard-Jones (LJ) potentials and the Green-Kubo (GK) formula. All the calculated thermal conductivities are in very good agreement with experimental results for a wide range of temperature and density. Special attention was paid to temperature and packing-fraction dependence which is nontrivial from dimensional analysis on the LJ potentials and the GK formula. First, the temperature dependence of T(1/4) was determined from the calculations at constant densities. Secondly, in order to obtain the dependence on packing fraction from that on number density separately, a scaling method of particle and/or cell size was introduced. The number density dependence of (N/V)(2/3) which is expected from the dimensional analysis of the GK formulas was confirmed and the packing-fraction dependence of η(3/2) was determined by using the scaling method. It turned out that the summarized functional form of m(-1/2)(N/V)(2/3)η(3/2)T(1/4) can well express both the calculated and experimental thermal conductivities for Ar, Kr, and Xe, where m is the atomic mass. The scaling method has also been applied to molten NaCl and KCl so that it has been found that the thermal conductivity has the packing-fraction dependence of η(2/3) which is much weaker than that of the simple LJ liquids.

No abstract

Dynamic structure factor of charge density was evaluated by MD simulation for molten LiI with artificially varied ionic masses. The dispersion curve was obtained from the peak frequency of collective excitation. The peak frequency at zero wavevector and the modulus of group velocity were estimated from the dispersion curves, as a function of ionic mass. The peak frequency at zero wavevector shows that the optic mode can be thermally excited for all the molten alkali halides. This suggests the ideas that the charge-density modes may dominate the thermal transport in these systems, and that insensitivity of the thermal conductivity of molten alkali halides to details of interionic potentials can be attributed to the behavior. The obtained modulus of group velocity was roughly proportional to m G −1/2 , where m G is the geometric average of anion and cation masses. This suggests that it may partially replace the mass term in the expression of thermal conductivity.