Short- to Intermediate-Range Structure, Transport, and Thermophysical Properties of LiF–NaF–ZrF4 Molten Salts

Abstract

LiF–NaF–ZrF4 multicomponent molten salts are identified as promising candidates for coolant salts in molten salt reactors and advanced high-temperature reactors. This study focused on low-melting point salt compositions of interest: 38LiF–51NaF–11ZrF4, 42LiF–29NaF–29ZrF4, and 26LiF–37NaF–37ZrF4. Ab-initio molecular dynamics (AIMD) calculations were performed and compared with available experimental data to assess the ability of rigid ion models (RIM) to reproduce short- to intermediate-range structure, transport, and thermophysical properties of the LiF–NaF–ZrF4 salt mixtures. It is found that as ZrF4 mol% increases, the average cation–anion coordination number (CN) of monovalent cations (Li+, Na+) obtained from RIM calculations decreases, while multivalent Zr4+ CN varied from 15% to 19% in comparison to corresponding AIMD values. In addition, RIM is found to predict the existence of 7, 8, and 9 coordinated fluorozirconate complexes, while AIMD and the available experimental data showed an occurrence of 6, 7, and 8 coordinated complexes in the melt. The intermediate-range structure analysis revealed that while the RIM parameters are able to reproduce a local structure for lower ZrF4 mol% salts such as in 38LiF–51NaF–11ZrF4, an extensive fluorozirconate network formation is observed in RIM simulations for higher ZrF4 mol% compositions. The network generated by RIM parameters is found to be mainly connected by “corner-sharing” fluorozirconate complexes as opposed to both “edge-sharing” and “corner-sharing” connectively portrayed by AIMD. It is found that a close agreement between AIMD and the RIM salt structure for the 11-mol% ZrF4 salt resulted in good agreement in the calculated Zr diffusivities and the viscosity values. However, due to the inaccurate short- to intermediate-range structure prediction by RIM for higher ZrF4 mol% compositions, thermophysical properties such as densities and heat capacity differ by up to 26% and 27%, respectively, upon comparison with AIMD and experimental values. Also, the network-dominated properties such as diffusion coefficients and viscosities differed by up to two and three orders of magnitude, respectively. This study signifies the importance of accurate salt structure generation for an accurate prediction of transport and thermophysical properties of multicomponent molten salts.

Transferable Deep Learning Potential Reveals Intermediate-Range Ordering Effects in LiF–NaF–ZrF4 Molten Salt

“…9 However, while the local coordination states reported in this Raman study were found to be in good agreement with our previous ab initio molecular dynamics (AIMD) findings, AIMD simulations for salts with 29% and 37% ZrF 4 content revealed fluorozirconate chain formation. 10 Such an intermediate-range structural ordering effect was also previously reported from NMR and EXAFS experimental observations and classical molecular modeling for LiF−ZrF 4 , NaF−ZrF 4 , and KF−ZrF 4 salts where higher ZrF 4 mole content was present. 11,12 In light of the challenges encountered in experimental measurements and their interpretation, AIMD simulations can be used to interpret experimental data and predict temperature-dependent structure, transport, and thermophysical properties of multicomponent molten salts.…”

“…The accuracy of neural network models is limited by the quality of ab initio data used for network training and validation . As such, the data for 42–29–29 and 26–37–37 mol % of LiF–NaF–ZrF 4 was taken from our previous study . In brief, we performed Born–Oppenheimer AIMD simulations using the Vienna Ab Initio Simulation Package (VASP) with the projector augmented wave (PAW) method, a plane wave basis set, and the Perdew–Burke–Ernzerhof (PBE) generalized-gradient-approximation (GGA) exchange correlation functional .…”

“…In order to explore such an intermediate-range structure formation in LiF-NaF-ZrF 4 in training (Figure 3) as well as in testing (Figure 4) regimes, Zr−Zr RDF, Zr−F−Zr angular distribution, and structure visualization (Figures 3 and 4 We have previously found that omitting the polarization contribution results in the chain primarily connected via corner-sharing complexes following the same reasoning provided above. 10 Here, NNMD is able to accurately predict the fluorozirconate chain connected via both cornersharing and edge-sharing complexes for both training and test compositions.…”

“…15 Here, density functional theory (DFT)-accurate as well as nearly CMD-efficient neural network interatomic potentials (NNIPs) can overcome these limitations and can be trained directly on the AIMD data without requiring any significant human intervention in defining and fitting parameters that are extensible to arbitrarily complex systems. Previously, Lam et al 22 and Chahal and Lam 23 have shown the robustness and versatility of NNIP when employed to study multicomponent molten salts containing multivalent cation species. Along this line, Rodriguez et al 24 used the smooth edition of DeePMDkit (DeepPot-SE) 25,26 to develop an NNIP to generate an accurate salt structure for the prediction of transport and thermophysical salt properties.…”

“…As such, the classical molecular dynamics (CMD) simulations have been proven efficient when studying the larger system sizes. However, our previous study showed that the traditional Rigid ion model (RIM) parameters overpredict the zirconium coordination numbers as well as the intermediate-range ordering when ZrF 4 mol % is higher in the melt, which in turn lead to up to 2–3 orders of magnitude differences in the diffusion coefficients and viscosities values . Such issues at higher ZrF 4 mol % content can be essentially fixed by including the polarization effects leading to the development of Polarizable ion model (PIM) .…”