A Comprehensive Physical‐Based Sensitivity Analysis of the Electrochemical Impedance Response of Lithium‐Ion Batteries

Abstract

The electrochemical impedance spectroscopy (EIS) characterization technique, although widely adopted in electrochemistry for understanding operational issues and degradation, has a less consolidated physical interpretation in lithium‐ion batteries (LIBs), often relying on circuital methods. Herein, the Doyle–Fuller–Newman model is adapted and experimentally validated for the physical simulation of electrochemical impedance; then, it is applied in a comprehensive one‐factor‐at‐time sensitivity analysis on an impedance spectrum from 4 kHz to 0.005 Hz; 28 physical parameters, which represent the kinetic, resistive, diffusive, and geometric characteristics of the battery, are varied within broad literature‐based ranges of values, for each of the 20 analyzed battery states, characterized by different state‐of‐charge and temperature values. The results show a miscellaneous sensitivity of parameters on impedance spectra, which ranges from highly sensitive to negligible, often resulting in a strong dependence on operating conditions and impedance frequency. Such results consolidate the understanding of LIB electrochemical impedance and demonstrate that 40% of the parameters, 12 out of 28, can be considered poorly sensitive or insensitive parameters; therefore, fitting the experimental EIS data, their value can be assumed from the literature without significantly losing accuracy.

Parameter sensitivity analysis of a physico-chemical lithium-ion battery model with combined discharge voltage and electrochemical impedance data

“…Diffusivity factor D e f is a calibration factor introduced in this study to the existing correlation used in COMSOL battery module and this newly introduced factor adapts the model to LGM50 characteristics. The same technique of using a calibration factor with COMSOL P2D models to mimic specific cell characteristics has been seen used by others [17].…”

“…In the literature, such as in [9,17], researchers tend to determine value ranges of the parameters by benchmarking range of values found for the same electrochemistry and selecting the mid value of the range as the nominal value.…”

“…Sensitivity of the parameters was quantified by analysing the dispersion of the states by defining the sensitivity index (SI) as the mean standard deviation of the state x i of the parameter i at the analysed SOC value and C-rate [9,17],…”

“…Most importantly, combined, the two techniques have the potential of one technique being capable of highlighting some parameters with higher sensitivities where the other technique fails to demonstrate competitively higher sensitivities. In [9] authors concluded the electrode kinetic rates with moderate sensitivities for the WLTP real-world dynamic cycling, whereas the same parameter showed the highest sensitivities for EIS spectra in [17]. Even though these studies employed two different models representing different commercial cells while both cells had Nickel-Manganese-Cobalt-Oxide (NMC) positive electrode chemistries, to the best of the authors' knowledge, no study in the literature has tried to analyse the potential of a parallel sensitivity analysis with chargedischarge cell terminal voltage data and EIS spectra.…”

“…However, due to their simplified analytical impedance model formulation of more complex real battery system, the analysis was unable to distinguish some important parameter sensitivities such as positive electrode contributions to negative electrode contributions as the model simplified both negative and positive electrode characteristics using compounded parameters. In a more recent study, Rabissi et al [17] followed a systematic OFAT approach on parameter sensitivity analysis. For the first time, the authors studied influence of both temperatures and state of charge (SOC) using a full-scale physical model for EIS.…”