ausblenden:
Schlagwörter:
-
Zusammenfassung:
For the operation of PEM fuel cells the diagnosis of critical states is an important task. Several detrimental conditions can lead to performance loss, degradation or safety problems. Water management for example has a major impact on fuel cell performance. While dehumidification leads to decrease of the proton conductivity, excess water causes flooding of the catalyst, gas diffusion layer or complete channels. Carbon monoxide poisoning of the catalyst is an additional problem, which can arise when CO-containing hydrogen, e.g. reformate gas from renewable sources, is used as a fuel gas.
Therefore, powerful tools for the diagnosis of such states are necessary. Their development is subject of ongoing research. Unfortunately, classical electrochemical methods are either not applicable for in-situ diagnosis (e.g. neutron imaging) or they are not sufficient to distinguish unambiguously between different fuel cell failures (e.g. Electrochemical Impedance Spectroscopy, EIS).
In our contribution, we apply Nonlinear Frequency Response Analysis (NFRA) [1], which might have the potential to overcome these problems. NFRA has originally been developed and applied to electrical circuits and effectively used to analyse mechanical systems and chemical process systems. The theoretical work of Bensmann et al. [2] has shown the applicability of NFRA for model discrimination in the kinetics of direct methanol fuel cells, whereas the work of Kadyk et al. [1] has shown experimentally the applicability to distinguish between flooding, dehydration and CO poisoning of PEM fuel cells.
NFRA is based on the excitation of a nonlinear system with a harmonic signal of high amplitude, in contrast to EIS in which only small amplitudes are used to measure the linear response of the system. Thus, with NFRA the nonlinearities of the system are excited, which are typical for electrochemical systems, e.g. reaction kinetics or transport phenomena. Based on the responding signal so-called higher order frequency response functions (HFRF) are determined using a mathematical framework based on Volterra series evaluation of the responding signal and multidimensional Fourier transform of this series.
Since the principle applicability of NFRA for the diagnosis of PEMFC of technical dimensions has been shown [1], the current work focuses in detail on the most interesting cases of anodic CO poisoning [3] and dehydration. In order to separate the influence of carbon monoxide poisoning and dehydration from other effects in a fuel cell, a H2/H2 cell was used in a “hydrogen pump” operation mode in order to exclude masking effects of the oxygen reduction reaction at the cathode. Additionally, a cell design with a differentially short channel was used in order to exclude along-the-channel effects like fuel depletion and accumulation of CO and water vapour. [3]
The experimental results showed a qualitatively similar behaviour in the linear EIS spectra: in both cases the main performance loss stems from an increase of the polarisation resistance. On the one hand, CO poisoning blocks the catalyst and leads to a decreased active area for the hydrogen oxidation reaction (HOR). On the other hand, dehydration of the proton conducting phase in the catalyst layer leads to a decreased protonic connection of the active sites further away from the membrane, making these sites inactive. Thus, the effective area for the HOR is decreased, too. However, the second order frequency response function gives additional information to the linear EIS and showed features which might be used to distinguish between both cases.
Additionally, the tailored experimental setup allows the usage of a simplified model, which could reproduce the main effects of the experimental results. The results obtained with the theoretical model give further insight into the processes within the catalyst layer and into effects occurring when both CO poisoning and dehydration take place.
[1]Kadyk, T., Hanke-Rauschenbach, R., Sundmacher, K., J. Electroanal. Chem. 630 (2009) 19-27
[2]Bensmann, B., Petkovska M., Vidaković, T., Hanke-Rauschenbach, R., Sundmacher, K., J. Electrochem. Soc. 157(9), 2010, B1279-B1289
[3]Kadyk, T., Hanke-Rauschenbach, R., Sundmacher, K. J. Appl. Electrochem. (2011)
DOI 10.1007/s10800-011-0298-8
