Type of presentation: Oral

MS-14-O-1515 Path-dependence modelling of lithium iron phosphate cathode studied by STEM-EELS spectrum imaging

Honda Y.1, Muto S.2, Tatsumi K.2, Kondo H.3, Horibuchi K.3, Kobayashi T.3
1Graduate school of engineering, Nagoya University, Japan, 2Eco Topia Science Institute, Nagoya University, Japan, 3TOYOTA Central R&D labs., Japan
honda.yoshitake@a.mbox.nagoya-u.ac.jp

LiFePO4 is used as a practical active material for positive electrodes of lithium-ion secondary batteries. It shows several advantages such as low cost, excellent cycle life and safety compared to other candidate materials. Its lithium insertion/extraction proceeds via a two phase mechanism: LiFePO4 ⇔ FePO4 + Li+ + e-. Several models for the charge-discharge mechanism of this material have been proposed, though the Domino-cascade model [1] is now recognized as the most plausible to explain the observed experimental results. However, the path-dependence, the characteristic polarization behavior depending on the preceding charge/discharge history [2] cannot be explained by this model. In this study we re-examine the microstructure of electrochemically half charged Li0.5FePO4 electrodes using scanning transmission electron microscopy (STEM) and electron energy loss spectroscopy (EELS) to clarify the standing problem above.
The lithium content of Li1-xFePO4 electrode was controlled electrochemically using a two-electrode cell. Thin specimens for STEM were prepared by focused ion beam. STEM-EELS spectrum imaging (SI) applied using a Jeol JEM ARM200F equipped with a GIF Quantum EELS, operated at 200 kV. O-K and Fe-L2, 3 spectra were simultaneously measured to distinguish between LiFePO4 (LFP) and FePO4 (FP) phase. The spectral data were collected with a scan step of 3 nm for many particles, from the datacubes of which were well separated into the spectral profiles characteristic of the LFP and FP phases (Fig. 1) and their respective spatial distributions using a multivariate spectral decomposition technique [3].
Particles prepared by lithium extraction from the fully discharged state generally exhibited the structure of LFP shell/FP core, whereas particles by lithium insertion from the fully charged state the FP shell/LFP core structure, contrary to the well accepted Domino-cascade model, as shown in Fig. 2(a) and (b) respectively.
The size dependence of the core/total volume ratios were plotted in Fig. 3 for the many particles. We assumed that the lithium insertion/extraction reactions occurred on the surface of the particles and LFP/FP interface proceeded inward. If the rate of phase transition was proportional to the surface area of each particle, the volume ratios predicted by the model in plotted with the broken line in Figure 3. Assuming the shell layer to act as resistance for ion diffusion, the path-dependence can be roughly explained by the present model.

References:
[1] C. Delmas, et al., Nature Materials 7, 665 (2008).
[2] V. Srinivasan, Electrochem. Solid-State Lett., 9, A110 (2006).
[3] S. Muto, et al., Mater. Trans. 50, 964 (2009).

A part of this work was supported by a Grant-in-Aid on Innovative Areas "Nano Informatics" (Grant number 25106004) from the Japan Society of the Promotion of Science.

Fig. 1: Component spectra extracted with MCR, each corresponding to LFP and FP respectively

Fig. 2: ADF images and spatial phase distributions of (a) Li0.5FePO4 particles prepared by Li extraction and (b) Li0.5FePO4 particles prepared by Li insertion

Fig. 3: Experimental results (symbols) and theoretical prediction (broken line) of size dependence of relative core/total volume ratio, based on a model where lithium insertion/extraction reactions occurred on the surface of the particles with LFP/FP interface proceeding inward, which rate is proportional to the surface area of each particle