Interpretation and Modelling of the Electrochemical Impedance of LiFePO4/Li4Ti5O12 Batteries

Abstract

Significant effort has been devoted to the research and development of safe lithium ion batteries in the last three decades. The commercial 26650 LiFePO4/Li4Ti5O12 (LFP/LTO) battery has been specifically found in many applications in which safety and cycle-life are of prime importance. The cell consists of olivine LiFePO4 as cathode, which is one of the most thermally stable, non-toxic, and long-cycle-life cathode materials, and spinel Li4Ti5O12 as anode, which features zero-strain Li intercalation, flat and relatively high operating voltage, and high thermal stability. In order to assess and improve the power capability and stability of these safe and long-life batteries, it is important to investigate and understand performance characteristics of individual cell components and interfaces, in order to elucidate the rate determining processes at each electrode. In this work, electrochemical impedance spectroscopy (EIS) analysis facilitated by distribution of relaxation times (DRT) is used to study the LFP/LTO cell at different states of charge (SOC) and SOC history, and an equivalent circuit model (ECM) is proposed in order to better understand the main processes at the anode and cathode and quantify their contribution to the full cell impedance. The modeling and interpretation of the full cell EIS is complicated because the LFP and LTO responses appear over similar frequency ranges. In order to unambiguously separate the anode and cathode contributions, EIS spectra from LTO/Li and LFP/Li half cells were also measured at different SOC and SOC history and the fit results were used to develop the full-cell ECM. The LFP/LTO cell impedance changes slightly only at low SOCs during charge, whereas it changes significantly during discharge. This SOC history dependence, i.e., the EIS response of a particular SOC depending on the previous SOC (Figure 1a), was mainly attributed to the change in the LFP electrode charge transfer process frequency and resistance (Figure 1b). This variation was found to be minimized at high SOCs and maximized at low SOCs (Figure 1), e.g., the LFP electrode charge transfer resistance in the full cell at 90% SOC was 11.3 Ωcm2 in charge vs 11.4 Ωcm2 in discharge (~1% difference) and at 10% SOC was 9.7 Ωcm2 in charge vs 13.3 Ωcm2 in discharge (~37% difference). As seen in the DRT results (Figure 1b), the frequency of the charge transfer process at the LFP electrode showed the same variation, i.e., it happens at 2.15 Hz at 90% SOC in charge vs 1.73 Hz in discharge (~20% difference) and at 1.31 Hz at 10% SOC in charge vs 0.55 Hz in discharge (~42% difference). The LTO electrode showed relatively little change in the charge transfer resistance and frequency with SOC and SOC history (Figure 1b).Figure 1. EIS response from the full 26650 cell at different SOC during charge (red lines) and discharge (back lines): a) shows the Nyquist plots and b) shows the DRT results. The spectra are shown after the truncation of the negative values of -Zimag and are normalized to the surface area of the electrodes. Figure 1