Distribution of Relaxation Times for Analysis of Solid Oxide Fuel Cell Stacks

Electrochemical Impedance Spectroscopy (EIS) is a frequently used measurement technique toinvestigate the electrical and structural properties of electrochemical energy converters such as fuelcells and electrolyzers. Recently the Distribution of Relaxation Times (DRT) analysis became apromising method to increase the resolution of electrochemical processes on the relaxation timescale and to support the Equivalent Circuit Modelling (ECM) approach by an a priori estimation ofthe number of processes contributing to the total polarization loss. Among the possibilities tocalculate the DRT function, the Tikhonov approach of regularized regression is a promising way todetermine the DRT numerically. However, a main drawback of this method is the fact that a suitableregularization parameter has to be chosen that has a big impact on the shape of the DRT.The aim of this thesis is to investigate the influence of constant phase element (CPE) behavior,inductive effects and EIS data structure on the accuracy and repeatability of the DRT with DRTtools.CPE behavior is observed in EIS measurements of fuel cells and leads to depressed semicircular arcsin the complex plane and a broadening of the relaxation time distribution. Inductive effects andmeasurement errors can originate from parasitic inductances in the test rig or the cables and leadto a disturbance of the impedance measurement especially in the high-frequency regime. For thesimulation study conducted in this thesis, a theoretical impedance is calculated with an equivalentcircuit model consisting of an ohmic resistor, an inductor and three parallel connections of an idealresistor and a CPE (called RQ elements) in series to simulate three electrochemical processes in anSOFC which exhibit frequency dispersion behavior in different extents depending on the magnitudeof the CPE exponent n. Additionally, simulation sets with different error structures and data pointdensities are simulated in varying frequency ranges to investigate their impact on the DRTcalculated with DRTtools, as well as co-effects of the above mentioned parameters such ascombined effects of parasitic inductances and high degrees of frequency dispersion. Subsequently,the results of the simulation study are verified on EIS measurements performed on a two-layer SOFCstack in F10 design of Forschungszentrum Jülich, highlighting the practical relevance of thesimulation results. Furthermore, the numerical origin of the observed calculation artefacts isinvestigated and a method to remove the artefacts is proposed