HIU Seminar | Prof. Dr. Francesco Ciucci
Prof. Dr. Francesco Ciucci from University Bayreuth will be guest during the next HIU Seminar taking place on Wednesday, 12th February 2025 at 2pm: The talk will be held in person in the large seminar room (230) at the HIU and streamed under the usual Zoom link: The title of the talk will be “Analyzing Impedance Spectra with the Probabilistic Distribution of Relaxation Times".
Summary
Electrochemical impedance spectroscopy (EIS) is a powerful technique for characterizing electrochemical systems, but its data interpretation is often limited by the non-uniqueness of equivalent circuit models. The distribution of relaxation times (DRT) method offers an exciting alternative, yet extracting DRT from EIS data poses an ill-posed inverse problem. This work addresses this challenge by developing a Bayesian statistical framework for DRT analysis.
The inherent flexibility of the Bayesian approach allows us to incorporate prior knowledge into the DRT estimation. We propose two novel prior models: (1) treating regularization penalty coefficients as multivariate random vectors, leading to timescale-dependent regularization that enhances DRT recovery compared to conventional methods; and (2) modeling residual weights as random vectors, enabling anomaly detection in EIS data and robust DRT recovery against experimental noise. We further extend this Bayesian DRT framework to Gaussian Processes and Neural Networks, demonstrating its ability to uniquely determine hyperparameters and effectively handle multidimensional, experimental-state-dependent EIS data.