Forecast Li Plating in LiFePO4/Graphite Cells: Proposal and Issues

Lithium-ion battery (LIB) for stationary use is required non-destructive degradation and SOH estimation in operation with sufficient safety and long-term use. We have developed non-destructive and post-mortem analysis for large format batteries. In case of non-destructive analysis, differential voltage analysis (dV/dQ) have been applied for the detection of lithium (Li) plating in operation. In addition, we have introduced Ar-filled glove box specially designed for large format battery disassembly to confirm the capacity fade of cathode / anode materials including Li plating on the anode. We selected 20 conditions as acceleration tests using cylindrical 26650 type LiFePO 4 /Graphite (LFP/Gr) cells with nominal capacity of 3.4 Ah. We applied the following four non-destructive approaches to the cells with under 90% SOH to forecast Li plating; i) slope of open circuit voltage (OCV) during galvanostatic interruption of titration technique (GITT), ii) gap of charge end voltage and OCV, iii) Gr capacity estimation using dV/dQ peak to peak (Gr 4 th and/or 3 rd to 2 nd stage) capacity or Gr 2 nd stage dV/dQ peak position, and iv) peak height and/or width of dV/dQ at Gr 2 nd stage. In case of OCV monitoring, we focused on OCV trend at Gr 2 nd and 1 st stage coexisting region (0.5< x <1 in Li x C 6 ). In the normal reaction, OCV at the region is almost flat due to the two-phase reaction of Gr and also LFP. However, we found the cell with OCV slope at the corresponding region was an indicator to forecast Li plating. In addition, the gap of charge end voltage and OCV during GITT was also an indicator to forecast Li plating. In such case, we could forecast Li plating by using two charge end voltage and subsequent OCV values at D10% SOC at the region of 0.5< x <1 in Li x C On the other hand, Li plating forecast using dV/dQ peak positions (used as an indicator of Gr capacity) could be applied in some cycle conditions but could not be applied other conditions including inhomogeneous capacity fade in operation. Then we focused on peak height and width of dV/dQ at Gr 2 nd stage and found that the trend of its peak shape will be an indicator to forecast Li plating. Furthermore, such forecast can grasp at limited SOC range operation including Gr 2nd stage region. Therefore, it will be a possibility to forecast Li plating in a short time using single dV/dQ profile. We will report merits and issues of the procedure to forecast Li plating including the evidence of Li plating by post mortem analysis.