Atomic Scale Modeling of Fundamental Characteristics of Metal Anodes and Solid Electrolytes: Overpotentials, Elastic Properties, and Grain Boundary Transport

Abstract

A rechargeable battery with a metallic anode offers the possibility of significant energy density improvement. However, mass commercialization has been hampered by a number of issues, including safety concerns posed by the formation and growth of dendrites during cycling. Use of the solid electrolyte Li7La3Zr2O12 (LLZO) can potentially address these concerns, and is promising in conjunction with Li metal due to its favorable combination of high conductivity and chemical stability. The present work characterizes several important properties of metallic anodes and LLZO at the atomic scale using Density Functional Theory (DFT) and molecular dynamics (MD) simulations. First, thermodynamic deposition/dissolution efficiencies and nucleation rates for seven metals were assessed. Thermodynamic overpotentials were evaluated via DFT at terraces and steps on several low-energy surfaces. In general, overpotentials were observed to be smallest for plating/stripping at steps, and largest at terraces. Differences in coordination numbers between a surface and a bulk atom were found to correlate with overpotential magnitude. Due to their low bulk coordination, body-centered alkali metals were predicted to be thermodynamically efficient for plating/stripping. In contrast, metals with larger bulk coordination, such as Al, Zn, and alkaline earths, exhibited higher thermodynamic overpotentials. The rate of steady-state nucleation during electrodeposition was estimated using a classical nucleation model informed by DFT calculations. Nucleation rates were predicted to be several orders of magnitude larger on alkali metal surfaces than on other metals. This multi-scale model highlighted the sensitivity of nucleation behavior on the morphology and composition of the electrode surface. Next, DFT calculations were employed to assess the elastic properties of eight anode materials. These were predicted as a function of temperature within the quasi-harmonic approximation. Anisotropy was assessed by resolving the moduli as a function of crystallographic direction. The alkali metals were predicted to have the smallest elastic moduli overall, which decreased with increasing atomic number. Al and Mg were predicted to exhibit highly isotropic elastic properties, while the alkali metals were highly anisotropic. In cubic systems, crystallographic directions exhibiting extrema in the elastic properties were diametrically opposed: under axial loading, the stiffest (most compliant) orientation was <111> (<100>), while in shear <100> (<111>) was the stiffest (most compliant). The maximum anisotropic shear modulus of some metals was observed to be more than twice as large as their respective polycrystalline values. Finally, to better understand the impact of grain boundaries (GB's) on LLZO's performance as a solid electrolyte, the structure of a wide range of tilt and twist axis GB's, including amorphous GB's, were predicted via classical MD and Monte Carlo simulations. Their energetics, composition, and Li transport properties were assessed. Little to no change was observed in the concentration of the four constituent elements across GB's, with the exception of amorphous Σ9(221)/[10], which showed a significant decrease in all four. Trajectories indicated disrupted diffusion pathways. Diffusivity showed greater sensitivity to temperature within crystalline GB's than in amorphous GB's. At the GB, Li diffusivity was consistently reduced compared to the bulk, while activation energy (∆Ea) was comparable or higher. Interestingly, diffusivity and ∆Ea both decreased with increasing grain boundary energy. A relationship between the pre-exponential factor (D0) and ∆Ea was observed, which suggested significant variation of D0 within the grain boundaries examined. Diffusivity at the GB's demonstrated anisotropy, with diffusion slower parallel to the GB plane, and faster normal to the GB plane.