Bayesian statistics-based analysis of AC impedance spectra

Abstract

AC impedance spectroscopy is an important method for evaluating ionic, electronic, and dielectric properties of materials. In conventional analysis of AC impedance spectra, the selection of an equivalent circuit model and its initial parameters are visually determined from a Nyquist plot; this visual determination can be both inefficient and inaccurate. Thus, analysis based on a rigorous mathematical method is highly desirable. Here, we demonstrate the analysis of AC impedance spectra using Bayesian statistics. We apply the method to artificial AC impedance spectra generated from resistance (R) and capacitance (C) circuits, obtaining a high accuracy ratio (>90%) in model selection when the ratio of the time constants of two RC parallel circuits exceeds 3. Furthermore, this method is applied to an actual electrical circuit comprising a resistance and two RC parallel circuits, yielding highly accurate model selection and parameter estimation. The results demonstrate the effectiveness of the proposed method for AC impedance spectra.