A DFT accurate machine learning description of molten ZnCl2 and its mixtures: 2. Potential development and properties prediction of ZnCl2-NaCl-KCl ternary salt for CSP

Abstract

ZnCl2-NaCl-KCl ternary salts are promising thermal storage and heat transfer fluid materials with a freezing point below 250 °C, thermal stability up to 800 °C, and other favorable properties that fit the use in the next generation concentrated solar thermal power. This work for the first time developed a machine learning-based interatomic potential for ZnCl2-NaCl-KCl ternary salt (0.6:0.2:0.2 in mole fraction) on the basis of energies and forces estimated by ab initio molecular dynamics calculations. The proposed machine learning potential was validated with the obtained partial radial distribution functions and the coordination numbers with the AIMD. The structural and thermophysical evolutions with temperature over the entire operating temperature range were documented. Adding Na+ and K+ ions deteriorated the network by corner-sharing and edge-sharing ZnCl4 tetrahedra, and apparently affected self-diffusion coefficient, thermal conductivity, and viscosity of the melt. The calculated thermophysical properties agreed with experimental data. A negative temperature dependence of thermal conductivity was noted and discussed. Based on the experimental data, viscosity data by Li et al. and those of this work, yielded reliable experimental values in the Vogel-Tamman-Fulcher form.