Early detection of lithium plating during fast charging of lithium-ion batteries using nonlinear frequency response analysis

Safe fast charging of lithium-ion batteries is a central development goal of the automotive industry. This requires an early detection of lithium plating during the fast-charging procedure to avoid potentially hazardous dendrite formation. In this paper, Nonlinear Frequency Response Analysis (NFRA) is shown to detect the occurrence of safety-critical lithium plating during the charging process earlier than Electrochemical Impedance Spectroscopy (EIS). NFRA measures higher harmonics, i.e., nonlinear behavior, in the cell's voltage response. This enables, compared to EIS, a more precise determination of the altered reaction kinetics due to plating. For the first time, NFRA is applied during a charging process. When lithium plating occurs, the second harmonic first increases and later decreases with ongoing plating. This feature in the second harmonic is more pronounced and occurs significantly earlier than changes in the first harmonic considered by conventional EIS. A cell model with plating reaction is applied to analyze the sensitivity of the NFR to plated lithium. Plating is modeled as an inhomogeneous layer only occurring on a part of the surface. The model reproduces the characteristic plating feature in the second harmonic as well as the experimentally observed decrease of impedance with state of charge. It thus enables a thorough interpretation of the dynamic cell behavior. Simulations show good agreement with experimental results, confirming that the second harmonic is more sensitive to plating than the impedance. Thus, among all known electrochemical plating detection methods, analyzing the second harmonic can identify the onset of lithium plating earliest during fast charging. • Detection method for lithium plating using Nonlinear Frequency Response Analysis (NFRA). • Significantly earlier detection of plating compared to known EIS-based methods. • Model for inhomogeneous plating in a Newman-type cell model. • Model-assisted understanding of the plating-related features in frequency response.