B. Karasulu1 J.P. Darby1 C.P. Grey2 A.J. Morris3

1, University of Cambridge, Cambridge, , United Kingdom
2, University of Cambridge, Cambridge, , United Kingdom
3, University of Birmingham, Birmingham, , United Kingdom

Conventional rechargeable lithium-ion batteries (LIBs) utilize solid electrodes and liquid electrolytes. Combustible organic electrolytes pose potential safety risks, including volatilization, flammability and even explosion. Solid-state (SS) inorganic electrolytes have the potential to eliminate these safety concerns, while providing high-energy LIBs. Two major challenges, however, have been hindering the practical high-performance all-SS battery applications: (1) the rather low ionic conductivities of SS electrolytes compared to liquid counterparts and (2) high resistance at the electrolyte/electrode interfaces that further curtails the ion migration. Sulfide-based electrolytes are a possible solution for the former, displaying ionic conductivities comparable to the organic counterparts, with potential enhancements by dopants.1 Besides, the interfacial resistances can be lowered by a rational design of the interfaces and the use of buffer layers.2
To tackle these issues, an automated computational procedure is adopted in this work for predicting novel SS electrode/electrolyte interfaces with lower interfacial resistances and high ionic conductivities. The procedure comprises several steps, first of which involves the generation of convex hulls by pre-screening the bulk structures of sulfide-based electrolytes with diverse compositions, phases, vacancies and doping. Subsequently, the promising electrolyte materials are screened this time for their ionic conductivity using molecular dynamics addressing Li-ion conduction. Stable surfaces of the selected bulk structures are then generated through random cuts and interfaced with the known cathode surfaces (e.g. LiCoO2, Li-metal), while minimizing the lattice mismatch. Our research group’s ab initio random structure searching (PyAIRSS) code3 is used for the random search of the initial structures, and for introducing possible dopants and vacancies into the lattices. All electronic structure calculations are performed with the plane-wave density functional theory (DFT) using the CASTEP code.4 Further details of this procedure will be discussed in this contribution along with the most promising interfaces predicted by this approach.
1 C. George, et al. Chem. Mat. 28, 7304-10 (2016); M. Butala et al. Chem. Mat., 29,3070-82 (2017); K. See et al. J. Am. Chem. Soc. 136, 16368.77 (2014)
2 J. Haruyama, et al. Chem. Mater. 26, 4248-4255 (2014)
3 C. J. Pickard, R. J. Needs, PRL 97, 045504 (2006); Medeiros et al., ACS Nano, 11, 6178–6185 (2017).
4 M. D. Segall et al., J. Phys.: Condens. Matter, 14, 2717 (2002).