M.Sc Thesis

M.Sc StudentYael Binyamini
SubjectImplementation of Impedance Spectroscopy Genetic Programming
on PEM Fuel Cells
DepartmentDepartment of Energy
Supervisor Professor Tsur Yoed


A fuel cell is an energy conversion device that generates electricity and heat by electrochemically combining a fuel (e.g., hydrogen) and an oxidant (preferably oxygen from air), without a combustion process, through electrodes and across an ion conducting electrolyte. Polymer electrolyte membrane (PEM) fuel cells are constructed using PEMs as a proton conductor and platinum-based materials as a catalyst. PEM fuel cells are developed as a preferred solution for transportation, which accounts for a significant portion (around 30%) of global energy consumption, and for both local and global pollution.

Electrochemical impedance spectroscopy (EIS) is a powerful and non-destructive method for characterizing electrical properties of materials and their interfaces. During the measurement, an electrical stimulus (a known voltage or current) is applied to the electrodes while observing the response (the resulting current or voltage which consists of a variety of microscopic processes). There are several conventional analysis methods to use for IS measurements while characterizing electrochemical systems, e.g. equivalent circuits. The objective of this work is implementing an analysis method based on evolutionary programming techniques for measurements done on PEMFCs. The analysis method is applied through the ISGP program that finds 'the best' distribution function of relaxation times that would generate the measured data. It can lead towards a better analysis of the cells, both for state of health monitoring and to facilitate better informed research and development.

A series of measurements were carried out in order to examine the effect of different conditions on the cell function. All of the obtained models, regardless of the working parameters, contain four peaks:

Ø Rm peak - represents the contribution of the Ohmic resistance of the cell components. This peak should be further studied by experiments consisting change of the humidity in both sides separately.

Ø Right and middle peaks (low frequencies) - represent oxygen diffusion within the CL according to the thin-film-flooded agglomerate model. Additional experiments should be performed in smaller temperature intervals for better understanding of the process.

Ø Left peak (higher frequencies) - The obtained results are not consistent with the explanation provided in the literature (charge transfer resistance). Therefore their origin is still unclear. Finding that particular peak demonstrates the strength of our method (ISGP). This process is not usually identified by the standard equivalent circuits, while other deconvolution methods are very sensitive to the exact parameters of the filter used and hence face difficulties in safely identifying it.