(Invited) Linear and Nonlinear Electrochemical Impedance Spectroscopy: A Data Science Perspective

Abstract

Electrochemical impedance spectroscopy (EIS) is widely used in the study of lithium ion batteries to determine important kinetic and transport parameters1 as well as to estimate internal state metrics and health prognostics.2,3 Typical analysis of the frequency dependent response involves fitting an equivalent circuit model to extract a few characteristic values for further study. This simple process is easy to implement and computationally efficient, but can artificially limit the information content of these techniques. Several rigorous physics-based models have laid the groundwork for a more sophisticated analysis of linear impedance for lithium ion batteries.4,5 These physics-based approaches offer improved confidence in parameter estimates, better phenomenological insight, and easier adaptation to changing chemistries. Previous work has been limited to analysis of the linear impedance response and variations in single physicochemical parameters. Nonlinear electrochemical impedance spectroscopy (NLEIS) provides greater information content than linear impedance.6 Here we extend the previous physics-based models to compute the linear impedance, Z1,1(ω), as well as the nonlinear impedance from the 2nd harmonic, Z2,2(ω), and 3rd harmonic, Z3,3(ω), response of an LiCO2|C battery subject to moderate amplitude sinusoidal current modulations. To show the sensitivity to particular parameters (and interactions between parameters) variance-based sensitivity analysis is presented. The results enable the partitioning of the response into measures of sensitivity that are used to generate insight into the techniques and inform design of experiments to optimally extract specific information related to electrochemical processes or internal states. The increased informational content in nonlinear impedance measurements are also evaluated viz-a-viz standard linear impedance.