Deep learning-driven molecular dynamics simulations of molten carbonates: 1. Local structure and transport properties of molten Li2CO3-Na2CO3 system

Computational methods to simulate molten salt thermophysical properties

“…Their strategy is always to run a short AIMD simulation of <100 atoms, use DeepMD-Kit to fit a deep potential, and then follow up with a longer MD simulation using the trained potential. They report consistently good agreement with experimental values for similar studies investigating LiCl-KCl mixture 125 , alkali chlorides 126 , lanthanum chlorides 127 , KCl-CaCl 2 128 , Li 2 CO 3 -Na 2 CO 3 129 , and SrCl2 130 .…”

Study of Speciation and Transport Properties for Different Compositions of Carbonates in Li2CO3–Na2CO3 and Li2CO3–K2CO3 Binary Systems at High Temperature in Molten State

“…The RDFs shown here are equivalent to the one presented in the literature of molten carbonate simulations for the same pressure and a similar temperature range. 28,29,31,32 In particular, previous studies have also noted the invariance of the first peak positions for the RDF O cations with respect to the composition. 31,32 Only a large increase of pressure is able to decrease the O-cation distance in the first solvation shell.…”

“…28,29,31,32 In particular, previous studies have also noted the invariance of the first peak positions for the RDF O cations with respect to the composition. 31,32 Only a large increase of pressure is able to decrease the O-cation distance in the first solvation shell. 17,28 To get more insights into the microscopic solvation structure around the carbonate anion, we also extracted the cation 3D densities.…”

“…We focus here on the intermolecular structure and dynamicswithout reactivityand we opted thus for a classical molecular dynamics simulation where energies and forces are calculated at each time step using a force field function, i.e., without electronic structure calculation. Among the existing force fields in the literature, ,, we chose the one recently developed by Desmaele et al It shows a good comparison with experiments for thermodynamic, structural, and dynamical properties, and it is easily transferable for all pure, binary, and ternary mixtures of alkali–carbonate . This force field consists of intermolecular pair interactions between alkali atoms and alkali and oxygen atoms, with a van der Waals and electrostatic terms, and intramolecular interactions within the carbonate anion.…”

“…We thus chose the force field recently developed by Desmaele et al that allows to explore the microscopic properties of molten carbonate in a wide range of thermodynamical conditions (900−2000 K, 0−15 GPa). 29 We finally noted a recent force field developed using machine learning with a dataset generated from first-principle molecular dynamics; 32 this interesting strategy will deserve considerations in future works.…”