Vast energy density improvements may be achieved in Li-ion rechargeable batteries with lithium metal anodes. However, lithium metal batteries containing liquid electrolytes have been unsuccessful thus far due to safety concerns associated with short circuits that occur when Li dendrites grow through the liquid electrolyte during the charging process. In contrast, thin film batteries using Li metal negative electrodes with an inorganic solid electrolyte (e.g., LiPON) appear to withstand dendrite penetration over extended cycling [1]. Recently, however, multiple research groups have reported and investigated cases where ceramic solid electrolytes paired with a Li metal anode experience a short circuit event [2,3]. The mechanism by which short circuit events occur in all-solid-state Li-ion batteries is unclear.

Thus, in this project, the growth of lithium metal filled cracks through four types of solid electrolytes, glassy Li₂S-P₂S₅ (LPS), β-Li₃PS₄­, polycrystalline and single crystal Li₆La₃ZrTaO₁₂ (LLZTO), was studied using galvanostatic electrodeposition experiments coupled with in-situ and ex-situ microscopy. An electrochemomechanical model for growth of lithium-filled cracks was developed. For current densities up to 5 mA/cm², only the as-fractured surface of glassy LPS exhibits lithium metal deposition at the surface alone, without penetration of the solid electrolyte. All other surfaces of the LPS and LLZTO samples studied had observable defects, the maximum initial sizes of which were characterized. Electrodeposition of lithium at these surfaces is accompanied by the filling, and then propagation, of lithium metal filled cracks. The experiments and model suggest that above a critical current density, the Li plating overpotentials, and corresponding mechanical stresses, reach values sufficiently large to propagate pre-existing surface defects. The results can be summarized, Figure 1, as a plot of the overpotential and corresponding mechanical stress necessary to extend a pre-existing sharp crack as a function of the crack length, for materials parameters suitable to LPS and LLZO. The prevailing failure mechanism in brittle solid electrolytes is Griffith-like, and differs from the amplification of kinetic perturbations at the metal interface that results in dendrite growth in liquid electrolytes. The results show that the shear-modulus criterion proposed by Monroe and Newman [4,5] for prevention of dendrites is not the determining factor for the high modulus, brittle inorganic electrolytes studied here. We suggest that stabilization of inorganic solid electrolyte interfaces against lithium metal penetration will require scrupulous attention to minimize interfacial defects.

Acknowledgments

Tushar Swamy and Lukas Porz contributed equally to this work. The authors gratefully acknowledge support from the US Department of Energy, Office of Basic Energy Science, through award number DE-SC0002633 (J. Vetrano, Program Manager). Brian W. Sheldon acknowledges additional support by the US Department of Energy, Office of Basic Energy Sciences under the award number DE-FG02-10ER46771. The authors also acknowledge use of the MIT Nanomechanical Technology Laboratory (A. Schwartzman, Manager).

References

1. J. Li, C. Ma, M. Chi, C. Liang, and N. J. Dudney, Adv. Energy Mater., 5, 1401408–1401414 (2015)

2. Y. Suzuki et al., Solid State Ionics, 278, 172–176 (2015)

3. R. Sudo et al., Solid State Ionics, 262, 151–154 (2014)

4. C. Monroe and J. Newman, J. Electrochem. Soc., 151, A880–A886 (2004)

5. C. Monroe and J. Newman, J. Electrochem. Soc., 152, A396–A404 (2005)