Cryogenic Focused Ion Beam Milling of Intact Na Metal Battery Stacks Reveals Separator Infiltration and Delamination Failure Mechanism

Metal anodes are the next frontier of high-energy-density rechargeable batteries. Sodium (Na) metal batteries are particularly promising due to the widespread abundance and low cost of Na precursors. The structure, composition, and function of the solid-electrolyte-interphase (SEI) as well as the metal growth mechanisms must be understood to improve and commercialize Na metal batteries. We developed a cryogenic workflow utilizing a cryogenic sample transfer system to prepare intact electrode:separator:electrode stacks, allowing us to probe the intrinsic structure and composition of undisturbed battery interfaces. By milling with a gallium ion beam at cryogenic temperatures, we produced clean milled surfaces of the electrodes and separator with minimal beam-induced artifacts allowing for high resolution imaging, spectroscopic mapping, and 3D reconstructions generated from serial milling and imaging. Using these techniques, we studied how the Na electrode, SEI, and separator evolve during cycling in both symmetric and asymmetric cells with a tri-layer porous polymer separator and either a carbonate- or ether-based electrolyte.<br/>Though often discarded before characterization, we find that the separator is a crucial piece of the puzzle when investigating battery performance and failure. We found overwhelmingly that cells begin with a conformal interface between the Na metal and the electrolyte-filled polymer separator instead of being separated by a layer of electrolyte. After extended cycling, Na metal infiltrates the separator pores, growing towards the opposing electrode. At the interfaces between the tri-layers, the Na can grow laterally, delaminating the separator layers. High resolution imaging and electron dispersive X-ray spectroscopy (EDS) mapping capture the Na metal in-between and within the polymer layers. 3D reconstructions of the electrodeposited Na and SEI suggest that the robustness of the SEI and the heterogeneity of its distribution (or lack thereof) relates to the ability of the Na to grow into the porous network of the separator. We combine these results to suggest a new failure model that relies on short-circuiting via the formation of a conductive pathway of Na templated by the porosity of the separator, rather than a penetration of the separator by discrete dendrites.