Nondestructive Electrochemical Identification of Lithium Plating in High-Energy Automotive Batteries

In this study, commercial 64.6 Ah automotive pouch-cell lithium-ion batteries were used to identify lithium plating (LP) by operando nondestructive electrochemical methods. These methods rely on the analysis of the cell voltage and cell current data, which is usually acquired by a battery management system (BMS). The cells were cycled under specific cycle-rate and temperature boundary conditions (BCs) in order to induce LP. The selected BCs were 0, 10, and 22 °C as test temperatures in combination with 1, 1.5, 2, and 3 C as C-rates. The chosen electrochemical methods are voltage relaxation profile (VRP), differential voltage analysis (DVA), and incremental capacity analysis (ICA). These methods are applied to different sections of charge/discharge cycle voltage and current data to identify the LP. ICA can be applied to the constant current phase of the charge cycle, VRP can be applied during the relaxation time after the charge cycle has finished, and DVA can be applied to the CC-phase of the discharge cycle. These methods are promising, as they do not require additional equipment for data acquisition, proving feasible for online monitoring by the BMS in an electric vehicle. Our investigation shows that in the case of VRP time derivative analysis, with decreasing temperature and increasing C-rate, the plateau/peak position shifts to the right, indicating increased LP during the charge cycle. During the DVA, the characteristic Li-stripping peak shifted to higher capacity values at lower temperatures and higher C-rates. Surprisingly, at a test temperature of 0 °C, the LP is higher at a rate of 2 C than at a rate of 3 C. This effect is attributed to the higher cell polarization at 3 C than at 2 C during the charge phase. In the ICA, at cycling rates of 1.5 C and above, an additional peak was observed at cell voltages above 4 V. This new peak represents an additional reaction taking place at the anode surface, identified as LP. For 0.5 C, no LP was identified in ICA, and LP was only identified at 0 °C at 1 C, whereas LP was always identified at cycling rates of 2 and 3 C at all test temperatures. The described methods and findings open the path toward advanced online LP identification and eventually state-of-health diagnosis of automotive batteries.