Hysteretic Phenomena Between Charging and Discharging Process in LiNi0.5Mn1.5O4

Abstract

Introduction LiNi0.5Mn1.5 O4-σ (LNMO) cathode material for lithium ion batteries are promising because of the high-energy density of 686 Wh/kg. However, the large hysteresis between the charging and the discharging profiles induces the low energy efficiency. Andrej et. al. found a hysteresis in the phase transition of LiNi0.5Mn1.5 O4 during battery operation in a range; the solid solution and two phase coexisting reaction for low lithium concentration dominate the charging and discharging, respectively[1]. Meanwhile, the factors determining the extent of the overpotential during the charging/discharging have not been understood. In this study, we measured in-situXAFS and XRD of LNMO and discussed the factors determining the over(under)potential. Experimental Commercial LNMO (Kojundo Chemical Laboratery Co., Ltd. Japan) was used in this study. The cathode was composed of LNMO, carbon black, and polyvinylidenefluoride with a weight ratio of 80:10:10. The cathode and lithium metal anode were separated by a polypropylene membrane separator. LiPF6 in ethylene carbonate/diethyl carbonate (EC/DEC) was used as the electrolyte. The 2032-type coin cell with X-Ray windows was assembled in an argon-filled glove box. We first charged the cell to 4.85 V at a rate of 1.0 C and took 10 minutes for rest when the voltage got to 4.85 V. We subsequently discharged the cell at a rate of 1.0 C to 3.5 V. In-situ Ni K-edge XANES measurement was performed in a transmission mode at beamline 5S1 of Aichi Synchrotron Radiation Center. We took 80 sec. measurement with 40 sec. interval per scan and the patterns were analyzed by using “Athena” program[2]. In-situ XRD measurement was also conducted at beamline 5S1 of Aichi Synchrotron Radiation Center. Results and Discussion The charge/discharge curves show two reaction plateaus: ≃ 4.0 V and 4.7 V regions for LNMO as shown in figure 1. Previous studies reported that these plateaus are ascribed to Mn and Ni redox reactions, respectively[1]. The dashed line indicates the estimated OCV from 1/10C charge-discharge cycle performed separately. We found that both the gap between the CCV (closed circuit voltage) and estimated OCV (open circuit voltage) curve and the gradient of CCV during discharging are larger than that during charging. We anticipate that discharging process requires larger overpotential kinetically, since charging process is a lithium-extraction process through the large lattice spacing requiring little lattice change and discharging process is a lithium-insertion process as expanding the lattice spacing. The gradient is extremely steep for low lithium concentration region( 0.3 ≤ x ≤ 0.5) during discharging, suggesting the requirement for a significant overpotential. Interestingly, Ni-K edge XANES peak top energy during charging plotted in Figure 2 shows the same tendency; the change rate is small before x = 0.5 and increases quickly afterwards. The peak top energy during charging changes in the same way. These results suggest that Ni ion plays more as the redox center in the range of x ≤ 0.5 and that a different redox process in addition to Ni oxidation occurs in the range of x ≥ 0.5. Based on previous studies, it is hardly likely that Mn4+ change to Mn5+, and O2p would attend the charge compensation in the latter region. Taking the above discussion into account, we consider that, during discharging in the range of 0.3 ≤ x ≤ 0.5, a considerable large lattice expansion is required for both lithium-insertion and valence state change from Ni3+ to Ni2+; the ionic radius of Ni3+ and Ni2+ is 0.57 and 0.69 Å, respectively. Further, we deduce that the kinetically preferred state for facilitating the abrupt change requires the significant overpotential. Our in-situ XRD results confirmed the above discussion from the viewpoint of the difference in the crystal structure between charging and discharging. We plan to talk about the details of the crystal structure and Mn-Kedge XAFS result for detailed discussion at PRiME 2016. Conclusion We measured in-situ XAFS and XRD of LNMO and consider that the over(under)potential strongly depends on the kinetic reaction mode, i.e.charging and discharging, and the charge compensator.