Fictitious phase separation in Li layered oxides driven by electro-autocatalysis (original) (raw)

nderstanding phase diagrams, whether it be for equilibrium (for example, temperature-composition) or kinetics (for example, time-temperature-transformation), is fundamental in materials science. Careful attention to rate and path dependence is crucial for distinguishing equilibrium and kinetic effects in phase behaviour, and battery materials are no exception to this basic prescription. The dynamics associated with the many-particle (ensemble) structure in battery electrodes 1,2 (for example, inter-and intra-particle phase separation) only make such rate and path dependencies ever more critical. In phase-separating LiFePO 4 , for example, it has been recognized that the reaction rate determines both the emergence of a thermodynamically forbidden solid solution 3-5 as well as the transition from particle-by-particle behaviour to concurrent intercalation 1,6. Recently, rate-dependent pathways have also been suggested in Li 4 Ti 5 O 12 (refs. 7,8). All these non-equilibrium phenomena contribute to the excellent rate capability of these materials. Meanwhile, in the so-called solid-solution layered oxides, studies on phase evolution have not been as comprehensive since this material class is deemed a single phase. Included are compounds such as Li(Ni,Mn,Co)O 2 (NMC) and Li(Ni,Co,Al)O 2 (NCA), typically viewed as having extensive single-phase composition ranges down to a lithium fraction of at least 0.5. This standard view is based on monotonic Nernst potential profiles and X-ray diffraction (XRD) data on equilibrated samples 9-14. Contradicting the standard view, phase separation at more than half lithium filling has also been reported in numerous operando XRD studies 15-22. This anomaly has been observed during the first charge, but not during the following discharge. At the rates used in these studies, the effect did not repeat on the second cycle, leading to the prevailing view that the anomaly is a 'first-cycle effect' 17-19. Surface passivation by Li 2 CO 3 has been suggested as one cause 19. More recently, apparent phase separation has also been reported in the second cycle, attributed to sluggish lithium diffusion near fully lithiated compositions 22. Other authors maintain that the observed phases are equilibrium phases 15,16,21 , designating them as H1 and H2 phases analogous to LiNiO 2 (ref. 23). However, rate and path dependencies have not been comprehensively addressed for any NMC or NCA composition, despite their widespread use 24. Partly responsible are the restricted designs of operando experiments based on available instrument time, limiting the range of rates and cycles. Likewise, the lack of particle-resolved composition mapping across an ensemble makes it difficult to assess nanoscale variations that arise from reaction and transport limitations. For these reasons, and despite the success of porous electrode theory 25-27 , a quantitative and predictive model explaining the rate and path dependencies in layered oxides has not been developed. Here, we report that the apparent phase separation persists in later cycles, even in LiNi 1/3 Mn 1/3 Co 1/3 O 2 (NMC111) and