Abstract: Liquid lead-lithium eutectic remains as a promising candidate for various breeding-blanket designs in future nuclear-fusion technologies. The lack of a generalized theory of interatomic forces in the liquid state is reflected on the wide variety of proposed functional forms to describe interatomic interactions even in simple liquids. Computer simulations facilitate the study of liquid metal properties, due to mathematical and experimental challenges. A classical-MD EAM potential is parametrized using mechanical and non-mechanical (melting-point) properties to minimize the arbitrariness of functional forms, where the employed pair potential stems from the liquid-state theory to avoid the issue of the uniqueness of the potential. Enhanced performance is obtained for liquid density, energy, structure, diffusivity and shear viscosity of Li, and their temperature-dependencies. In a similar manner, reference experimental and ab initio MD data are used to parametrize a functional to describe Pb-Li pairwise interactions in liquid Pb-Li alloy, which is used with the derived EAM of liquid Li and a reference EAM of liquid Pb to investigate properties of liquid Pb-Li alloy. Enhanced transferability characteristics are obtained for low-in-lithium liquid Pb-Li melts, where Coulombic interactions are negligible. In specific, the exhibited behaviour of Li in liquid lead-lithium eutectic is consistent with findings from ab initio MD methods, and drastically different from predictions of previous C-MD studies which suggested a substantial segregation of Li atoms instead of dispersion. It is concluded that the functional form of the pair potential and its uniqueness influence both the pure liquid-metal properties and the validity of the potential transferability in multi-component systems, where a theoretical functional results in enhanced performance in pure and alloyed liquid systems.

Abstract: Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of machine learning IAPs (ML-IAPs) based on four local environment descriptors—atom-centered symmetry functions (ACSF), smooth overlap of atomic positions (SOAP), the spectral neighbor analysis potential (SNAP) bispectrum components, and moment tensors—using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model and, consequently, computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.

Abstract: Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of machine learning IAPs (ML-IAPs) based on four local environment descriptors—atom-centered symmetry functions (ACSF), smooth overlap of atomic positions (SOAP), the spectral neighbor analysis potential (SNAP) bispectrum components, and moment tensors—using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model and, consequently, computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.

Abstract: We demonstrate that the melting points and other thermodynamic quantities of the alkali metals can be calculated based on static crystalline properties. To do this we derive analytic interatomic potentials for the alkali metals fitted precisely to cohesive and vacancy energies, elastic moduli, the lattice parameter, and crystal stability. These potentials are then used to calculate melting points by simulating the equilibration of solid and liquid samples in thermal contact at ambient pressure. With the exception of lithium, remarkably good agreement is found with experimental values. The instability of the bcc structure in Li and Na at low temperatures is also reproduced and, unusually, is not due to a soft T1N phonon mode. No forces or finite-temperature properties are included in the fit, so this demonstrates a surprisingly high level of intrinsic transferability in the simple potentials. Currently, there are few potentials available for the alkali metals, so in addition to demonstrating trends in behavior, we expect that the potentials will be of broad general use.

Abstract: Interatomic potentials for pure Li and the Mg-Li binary system have been developed based on the second nearest-neighbor modified embedded-atom method formalism. The potentials can describe various fundamental physical properties of pure Li (bulk, point defect, planar defect and thermal properties) and alloy behaviors (thermodynamic, structural and elastic properties) in reasonable agreement with experimental data or higher-level calculations. The applicability of the potential to atomistic investigations on the deformation behavior of Mg alloys and the effect of Li is demonstrated.

Abstract: Ionic compounds exhibit a variety of crystal structures that can critically affect their applications. Traditionally, relative sizes of cations and anions have been used to explain coordination of ions within the crystals. Such approaches assume atoms to be hard spheres and they cannot explain the observed structures of some crystals. Here we develop an atomistic method and use it to explore the structure-determining factors beyond the limitations of the hard sphere approach. Our approach is based upon a calibrated interatomic potential database that uses independent intrinsic bond lengths to measure atomic sizes. By carrying out extensive atomistic simulations, striking relationships among intrinsic bond lengths are discovered to determine the B1 (NaCl), B2 (CsCl), and B3 (zinc-blende) structure of binary ionic compounds.

Abstract: Unraveling the liquid structure of multicomponent molten salts is challenging due to the difficulty in conducting and interpreting high-temperature diffraction experiments. Motivated by this challenge, we developed composition-transferable Gaussian approximation potential (GAP) for molten LiCl-KCl. A DFT-SCAN accurate GAP is active-learned from only ~1100 training configurations drawn from 10 unique mixture compositions enriched with metadynamics. The GAP-computed structures show strong agreement across high-energy x-ray diffraction experiments, including for a eutectic not explicitly included in model training, thereby opening the possibility of composition discovery.

Abstract: The in silico modeling of molten salts is critical for emerging "carbon-free" energy applications but is inhibited by the cost of quantum mechanically treating the high polarizabilities of molten salts. Here, we integrate configurational sampling using classical force fields with active learning to automate and accelerate the generation of Gaussian approximation potentials (GAP) for molten salts. This methodology reduces the number of expensive ab initio evaluations required for training set generation to O(100), enabling the facile parametrization of a molten LiCl GAP model that exhibits a 19000-fold speedup relative to AIMD. The developed molten LiCl GAP model is applied to sample extended spatiotemporal scales, permitting new physical insights into molten LiCl's coordination structure as well as experimentally validated predictions of structures, densities, self-diffusion constants, and ionic conductivities. The developed methodology significantly lowers the barrier to the in silico understanding and design of molten salts across the periodic table.

Abstract: Although large-scale atomistic simulations provide useful insights into various material phenomena, such studies on LiCoO₂, which is the most widely used cathode material for lithium ion batteries (LIBs), have rarely been undertaken due to difficulties in developing adequate interatomic potentials. In this study, an interatomic potential (2NNMEAM + Qeq) for the Li-Co-O ternary system is developed to carry out molecular dynamics (MD) simulation studies on lithium cobalt oxides. Potential parameters are optimized so that the potential can successfully reproduce fundamental materials properties (structural, elastic, thermodynamic and migration properties) of various compounds of sub-binary and lithium cobalt ternary oxide systems. Through MD simulations, we investigate lithium diffusion properties (activation energy for lithium migration and diffusion coefficient) in layered Li_1−xCoO₂ (0 ≤ x ≤ 0.5) of various lithium vacancy concentrations. We find that the lithium vacancy concentration has a significant influence on the activation energy for lithium diffusion and the lithium diffusion coefficient in the Li_1−xCoO₂ cathode. The developed potential can be further utilized for atomistic simulation studies on other materials phenomena (phase transitions, defect formation, lithiation/delithiation, etc.) in LIB cathode materials.

Abstract: Interatomic potentials for pure Li and the Mg-Li binary system have been developed based on the second nearest-neighbor modified embedded-atom method formalism. The potentials can describe various fundamental physical properties of pure Li (bulk, point defect, planar defect and thermal properties) and alloy behaviors (thermodynamic, structural and elastic properties) in reasonable agreement with experimental data or higher-level calculations. The applicability of the potential to atomistic investigations on the deformation behavior of Mg alloys and the effect of Li is demonstrated.

Abstract: An interatomic potential of the Li–Mn–O ternary system has been developed on the basis of the second-nearest-neighbor modified embedded-atom method (2NN MEAM) formalism combined with a charge equilibration (Qeq) concept. The potential reproduces fundamental physical properties (structural, elastic, thermodynamic and migration properties) of various compounds well, including lithium oxides, manganese oxides, and lithium manganese ternary oxides. Through molecular dynamics (MD) simulations using the developed potential, lithium diffusion properties (activation energy for lithium migration and diffusion coefficient) in spinel Li_1–xMn₂O₄ are also reproduced in good agreement with experiments. We have found that the effect of the lithium vacancy concentration is marginal on the activation energy for lithium diffusion in the Li_1–xMn₂O₄ cathode, but it is significant in the lithium diffusion coefficient. The potential can be further utilized for atomistic simulations of various materials phenomena (phase transitions, defect formation, lithiation/delithiation, etc.) in LIB cathode materials.

Abstract: Liquid lead-lithium eutectic remains as a promising candidate for various breeding-blanket designs in future nuclear-fusion technologies. The lack of a generalized theory of interatomic forces in the liquid state is reflected on the wide variety of proposed functional forms to describe interatomic interactions even in simple liquids. Computer simulations facilitate the study of liquid metal properties, due to mathematical and experimental challenges. A classical-MD EAM potential is parametrized using mechanical and non-mechanical (melting-point) properties to minimize the arbitrariness of functional forms, where the employed pair potential stems from the liquid-state theory to avoid the issue of the uniqueness of the potential. Enhanced performance is obtained for liquid density, energy, structure, diffusivity and shear viscosity of Li, and their temperature-dependencies. In a similar manner, reference experimental and ab initio MD data are used to parametrize a functional to describe Pb-Li pairwise interactions in liquid Pb-Li alloy, which is used with the derived EAM of liquid Li and a reference EAM of liquid Pb to investigate properties of liquid Pb-Li alloy. Enhanced transferability characteristics are obtained for low-in-lithium liquid Pb-Li melts, where Coulombic interactions are negligible. In specific, the exhibited behaviour of Li in liquid lead-lithium eutectic is consistent with findings from ab initio MD methods, and drastically different from predictions of previous C-MD studies which suggested a substantial segregation of Li atoms instead of dispersion. It is concluded that the functional form of the pair potential and its uniqueness influence both the pure liquid-metal properties and the validity of the potential transferability in multi-component systems, where a theoretical functional results in enhanced performance in pure and alloyed liquid systems.

Abstract: Sulfur is a very promising cathode material for rechargeable energy storage devices. However, sulfur cathodes undergo a noticeable volume variation upon cycling, which induces mechanical stress. In spite of intensive investigation of the electrochemical behavior of the lithiated sulfur compounds, their mechanical properties are not very well understood. In order to fill this gap, we developed a ReaxFF interatomic potential to describe Li–S interactions and performed molecular dynamics (MD) simulations to study the structural, mechanical, and kinetic behavior of the amorphous lithiated sulfur (a-LixS) compounds. We examined the effect of lithiation on material properties such as ultimate strength, yield strength, and Young's modulus. Our results suggest that with increasing lithium content, the strength of lithiated sulfur compounds improves, although this increment is not linear with lithiation. The diffusion coefficients of both lithium and sulfur were computed for the a-LixS system at various stages of Li-loading. A grand canonical Monte Carlo (GCMC) scheme was used to calculate the open circuit voltage profile during cell discharge. The Li–S binary phase diagram was constructed using genetic algorithm based tools. Overall, these simulation results provide insight into the behavior of sulfur based cathode materials that are needed for developing lithium–sulfur batteries.