Nondestructive EIS Testing to Estimate a Subset of Physics-based-model Parameter Values for Lithium-ion Cells

Abstract

This paper is the final installment in a series of articles that collectively shows how to estimate parameter values for lumped-parameter physics-based models of lithium-ion cells without requiring cell teardown. In this paper, we leverage electrochemical impedance spectroscopy (EIS) to find estimates of all as-yet-unresolved parameter values. The characterization process regresses the measured cell impedance spectrum to exact analytic closed-form expressions of the frequency response of an extended Doyle–Fuller–Newman model to identify thirteen lumped parameters plus multiple reaction-rate constants. A nonlinear optimization algorithm performs the regression, and so it is important to provide reasonable initial parameter estimates and constraints, which we also discuss. As part of this process, the generalized distribution of realization times technique is used to isolate time constants from the two electrodes as well as to calibrate the laboratory EIS-test data. The overall methodology is studied on a virtual cell and on a laboratory cell (both having graphite//NMC chemistries). Parameter estimates found in the simulation study are highly accurate, leading us to have confidence in the values estimated for the physical cell as well.

Quantifying the Entropy and Enthalpy of Insertion Materials for Battery Applications Via the Multi-Species, Multi-Reaction Model

“…The MSMR model provided quantitative agreement with experimental data for the open circuit voltage of lithiated silicon, 66 graphite, iron phosphate, and spinel manganese oxide 57,58,60 and has been utilized by several groups to probe various battery phenomena. [67][68][69][70] The different species of lithium correspond to different galleries in the host materials and describes the relationship between the Open Circuit Voltage and the mole fraction of each species of intercalated lithium.…”

Modeling Voltage Decay During Calendar-Life Aging

“…10: p , must be known prior to the SEI-model system identification. Their values can be estimated using the methods presented in references, [34][35][36]46,47 for example. Their sensitivity studies are exhibited in Fig.…”

“…Finally, we now apply the self-discharge model and parameteridentification ideas to commercial cells. The cell we chose to model here is one we have used in previous publications: [34][35][36]46,47 a prismatic 25 Ah (≈0.93 mol) graphite//NMC cell manufactured by Table II. OCV and OCP (MSMR-style) models for simulation studies.…”