학술논문

Model-assisted identification of solid oxide cell elementary processes by electrochemical impedance spectroscopy measurements.
Document Type
Article
Source
Journal of Power Sources. Oct2019, Vol. 436, pN.PAG-N.PAG. 1p.
Subject
*OXYGEN electrodes
*ATTRIBUTION of authorship
*PARTIAL pressure
*CHARGE transfer
*CHEMICAL reactions
*IMPEDANCE spectroscopy
*MATRIX-assisted laser desorption-ionization
*SOLID oxide fuel cells
Language
ISSN
0378-7753
Abstract
Electrochemical Impedance Spectroscopy (EIS) is extensively used to characterize Solid Oxide Cells (SOCs) to extract information on the elementary loss mechanisms. However, these individual mechanisms usually overlap in the frequency domain, requiring dedicated data processing for unambiguous identification. A powerful method for discriminating process contributions is the analysis by the Distribution of Relaxation Times (DRT). The de-convoluted spectrum of SOC generally presents, six peaks from mHz to hundreds of kHz. DRT peak-to-process attribution is often obtained by experimental sensitivity analysis. In the study, six parameters have been systematically varied: temperature, current density, partial pressure of O2 at the oxygen electrode, partial pressure of steam at the fuel electrode, total flow rates and fuel composition. The effect of wires inductance and different cell geometries has also been analyzed. The study provides detailed information about the contribution of the elementary processes to the total losses over a wide range of operation regimes. It further refines peak-to-process attributions by using a dynamic numerical model that includes gas and solid phase transport coupled with charge transfer and chemical reactions. Moreover, the non-univocal literature attribution of processes in the middle frequency range is clarified: strongly overlapping peaks cannot be separated even by DRT. • Operating conditions influence on DRT signals of experimentally recorded data. • Attribution of peaks-to-processes clarified in the middle frequency range: 10–500 Hz. • Peaks-to-processes attribution supported by the DRT of numerically modeled spectra. • Influence of test cell geometry and frequency range on the deconvoluted spectra. [ABSTRACT FROM AUTHOR]