Albert Krügel

Deconvolution of electrochemical impedance spectra for the identification of electrode reaction mechanisms in solid oxide fuel cells

The polarization processes occurring at the electrode–electrolyte interfaces of solid oxide fuel cells (SOFC) were investigated by electrochemical impedance spectra measured at single cells under realistic operating conditions. The approach presented is based on distributions of relaxation times which are the basic quantity of interest in electrochemical impedance data analysis. A deconvolution method was developed and implemented that yields these characteristic distribution patterns directly from the impedance spectra. In contrast to nonlinear least squares curve fit of equivalent circuit models, no a priori circuit choice has to be made. Even more importantly, the excellent resolving capacity allows the untangling of the impedance contributions of up to three physically distinct processes within one frequency decade. With the method, processes with the highest polarization losses can be identified and targeted to improve cell performance. Based on the distributions, a general strategy for the identification of the reaction mechanisms is given. The evaluation of the distributions in terms of peak parameters is illustrated by a physical model for oxygen reduction at the SOFC cathode–electrolyte interface. The method is expected to find many applications in electrochemistry beyond the field of solid oxide fuel cell development.

Engineering of SOFC Anode/Electrolyte Interface

Cofiring of anode/electrolyte leads to improved mechanical adherence and lowers polarization losses at the anode. However, caused by different shrinkage of anode and electrolyte the compound is bending or breaking if it is under weight. Sintering experiments revealed that these mechanical stresses can be reduced by using an anode pattern consisting of a large number of small sized individual areas instead of a continuous layer. A simple static model was used to describe the shrinkage due to sintering by effective constant Youngs moduli and pseudo linear thermal expansion coefficients (TEC). By the use of FEM different anode pattern geometries were analyzed with respect to curvature radius and mechanical stress in the electrolyte during cofiring. It turned out that anodes with a hexagonal pattern like honeycombs are preferable as they show small mechanical stress with high coverage at the same time.

Characterization of Multilayer Anodes for SOFC

SOFC anodes have to combine various tasks. In anode supported single cells a thick anode substrate is used for current collecting and gas distribution whereas a thin functional layer adjacent to the electrolyte is the electrochemically active part of the anode. This functional anode layer is cofired together with the thin film electrolyte to obtain an enhanced interface with low polarisation losses. This multilayer structure was transferred to an electrolyte supported single cell. The electrochemical active Ni/8YSZ anode layer was screen printed onto a 8YSZ electrolyte green tape and subsequently cofired at 1350 °. Mechanical stresses during cofiring due to shrinkage mismatch of anode and electrolyte were avoided by changing the geometry of the anode layer from a continuous layer to a large number of small sized individual areas. Simulations by finite element modeling indicated that a hexagonal pattern similar to honeycombs is preferable. The second layer which adjoins to the fuel gas channels and which is responsible for current collecting and gas distribution was later on screen printed on top and sintered together with the cathode. Single cells with a multilayer anode and different functional layers were electrochemically characterised under realistic operation conditions. The performance and reduction/oxidation stability of this type of anode was investigated. The electrochemically active layer showed only small degradation during redox cycling and long term operation at high fuel utilisation. In contradiction to single layer anodes Nickel agglomeration was not observed in the functional layer.