Fluid Structure of Molten LiCl–Li Solutions

Abstract

Molten mixtures of lithium chloride and metallic lithium (LiCl–Li) play an essential role in the electrolytic reduction of various metal oxides. These mixtures possess unique high temperature physical and chemical properties that have been investigated for decades. However, due to their extreme chemical reactivity, no study to date has been capable of definitively proving the basic physical nature of Li dissolution in molten LiCl. In this study, the evolution of the structure of molten LiCl–Li is probed as metallic Li is electrochemically introduced into the melt in situ, using synchrotron radiation experiments based on high energy wide-angle X-ray scattering (WAXS) and small-angle X-ray scattering (SAXS). The time-resolved scattering results indicate the formation of transient Cl– ion cages surrounding low-density voids with a periodicity of ∼8.3 Å, which suggests the formation of metastable Li nanocluster. The structure of the LiCl–Li nanoclusters in the solution is modeled using ab initio molecular dynamics (AIMD) simulations. The simulation results are in agreement with the X-ray diffraction measurement and suggest the nanoclusters are predominantly Li8, along with smaller clusters.

DFT Accurate Interatomic Potential for Molten NaCl from Machine Learning

“…The self-diffusion coefficients of Na + and Cl − in molten NaCl, D α (α = N a + , Cl − ), are obtained from the mean square displacement of particle position according to the Einstein relation described in (7), where r i,α is the position of the i th ion and α is the specific ionic species (i.e., Na + and Cl − ).…”

Automated Development of Molten Salt Machine Learning Potentials: Application to LiCl

“…48,49 The S(Q) obtained via GAP MD simulation is compared in Figure 2a) with that recorded using HEXRD with a transmission geometry at beamline 6-ID-D at the Advance Photon Source (APS), 47 as well as the results generated with ab initio MD simulations. 47 The GAP MD simulation exhibits excellent agreement with both full AIMD (216 atoms, 15 ps total simulation time) simulation and HEXRD over all Q values. GAP also provides improved accuracy over AIMD in resolving the position of the First Sharp Diffraction Peak (FSDP) at Q ≈ 2 Å −1 , as seen in the insert of Figure 2a).…”

“…To validate the performance of the LiCl GAP MLP, we compare the structure of GAP MD-derived configurations to those obtained from high-energy X-ray diffraction (HEXRD) experiments performed at ∼900 K. The partial pair distribution functions (PDF) for Li + –Li + , Li + –Cl –, and Cl – –Cl – atomic pairs are obtained from 900 K GAP MD trajectories of the 1024 atom simulation cell and transformed to the total structure factor, S ( Q ), via Fourier transformation and the use of atomic form factors. , The S ( Q ) obtained via GAP MD simulation is compared in Figure a with that recorded using HEXRD with a transmission geometry at beamline 6-ID-D at the Advance Photon Source (APS) and the results generated with ab initio MD simulations . The GAP MD simulation exhibits excellent agreement with both full AIMD (216 atoms, 15 ps total simulation time) simulation and HEXRD over all Q values.…”

“…To validate the performance of the LiCl GAP MLP, we compare the structure of GAP MD-derived configurations to those obtained from high-energy X-ray diffraction (HEXRD) experiments performed at ∼900 K. The partial pair distribution functions (PDF) for Li + −Li + , Li + −Cl −, and Cl − −Cl − atomic pairs are obtained from 900 K GAP MD trajectories of the 1024 atom simulation cell and transformed to the total structure factor, S(Q), via Fourier transformation and the use of atomic form factors. 50,51 The S(Q) obtained via GAP MD simulation is compared in Figure 2a with that recorded using HEXRD with a transmission geometry at beamline 6-ID-D at the Advance Photon Source (APS) 49 and the results generated with ab initio MD simulations. 49 The GAP MD simulation exhibits excellent agreement with both full AIMD (216 atoms, 15 ps total simulation time) simulation and HEXRD over all Q values.…”

Characterization of the Cathodic Thermal Behavior of Molten CaCl2 and Its Hygroscopic Chloride Mixture During Electrolysis

“…This is probably because the simulated value is for an ideal system; on the other hand, the electrodeposited morphology of Ca in this system exhibits a unique dispersed phase. The excess energy was probably required to create the dispersed metal-and-salt interface and depends on the microscopic fog morphology [39,40]. Unfortunately, no evidence has yet been obtained to support this hypothesis.…”