Implications of the Mechanical, Electronic and Ionic Properties of Bulk and Surface Li₆PS₅Cl Argyrodite Structures on Lithium Filament Resistance

Li-filament growth has been experimentally observed in numerous promising solid<br/>electrolytes (SEs), such as garnet Li₇La₃Zr₂O₁₂ (LLZO), NaSICON-type Li_1+xAl_xTi_2-x(PO₄)₃(LAPT), Li₂PO₂N (LiPON), and argyrodite Li₆PS₅Cl (LPSC), therefore preventing the use of Li-metal anodes in all-solid-state-batteries (ASSBs). Herein, we probed the mechanical, electronic, and ionic properties of LPSC, with density functional theory (DFT) calculations, and compared<br/>with other SEs to determine the relevant factors for predicting Li-filament resistance in SEs.<br/><br/>LPSC has a complicated structure which can incorporate anion inversion between S^2-/Cl^-; and has Li⁺distributed among two Wyckoff sites (24g and 48h). In order to probe the bulk and surface properties of LPSC, we first determined a representative bulk structure that incorporates S^2-/Cl^- inversion, and Li⁺ distributed among 24g and 48h sites, via random structure sampling. The lowest energy bulk structures contained no S^2-/Cl^-inversion and had ~80% of Li⁺ in 48h sites<br/>after relaxation, which agrees with previous experimental studies.<br/><br/>The shear modulus and fracture energy of the bulk LPSC were also calculated to<br/>understand the mechanical resistance to filament growth. Surfaces were cleaved in the (100), (110), and (111) directions, with the (100) direction having the lowest fracture energy. For the (100) direction of LPSC, four additional surfaces were cleaved with the following stoichiometries: Li₂S-rich, Li₂S-deficient, LiCl-rich, and LiCl-deficient. Of these surfaces, the Li₂S-deficient slab had the lowest fracture energy (0.20 J/m²), which is far lower than the fracture energy of LLZO (1.76 J/m²). Due its low fracture energy, and small shear modulus (10 GPa),<br/>especially when compared to LLZO (56 GPa), LPSC is likely to experience mechanical cracking.<br/><br/>Surface bandgaps were found and compared to the bulk for all four (100) surfaces. It has been proposed that a reduced surface bandgap and excess electrons trapped inside internal defects, such as pore surfaces or crack surfaces, can directly reduce Li⁺to Li⁰, leading to filament growth ahead of the crack tip in SEs. Bandgap and electron distribution calculations indicated no significant excess electron trapping in the LPSC surfaces. Additionally, due to the low fracture energy of all the surfaces tested, we predict that LPSC will form Li-filaments through dry-cracks, with a mechanical crack opening up first, then Li⁺ flowing in to fill the crack. This is consistent with experimental observations and opposite to the wet-cracks found in LLZO.