Pore and cell scale

a) Synchrotron Tomography. The 3D images obtained from synchrotron tomography allow for a resolution below 1 micron per voxel. Compared to the typical fiber diameter of several microns and the typical pore size on the order of 10 microns, synchrotron tomography facilitates the mapping of the real three-dimensional micro-structure of GDL materials.

b) Virtually created materials. The basic parameters determining the structure of the non-woven carbon paper are:

• the porosity
• the radius of the fibers
• the shape of the fiber cross section
• the distribution of the fiber directions (anisotropy parameter)

With these parameters a virtual structure model of the material is generated by the microstructure generator software GeoDict of Fraunhofer ITWM. Once a geometrical model of the GDL is obtained, micro-structure simulation determines the effective material parameters of the GDL. These are:

• capillary pressur - saturation relation
• bubble point
• gas diffusivity tensor
• permeability tensor
• thermal conductivity tensor
  

Except for the bubble point, these parameters all depend strongly on the saturation of the porous gas diffusion layer with liquid water. Simulations of fuel cell stacks on the cell and stack scale need to account for the ocurrence of liquid water in PEMFCs. Hence, saturation dependent parameters are a very important part of cell and stack models. The pore-morphology method is applied for the determination of the parameters to avoid numerically costly two-phase flow simulations. In short, the pore-morphology works as follows: At a given capillary pressure the pore space accessible for the water is determined by the size of the pores. The mean curvature radius between water and air is related to the capillary pressure by the Young-Laplace formula. To simulate an air drainage process, the available pore space is determined by fitting spheres of a certain radius. This part of the pore space is associated with the pore volume that can be occupied by the non-wetting phase. This procedure is repeated with stepwise reduced radii until the entire pore space is filled with water. Relative, i.e. saturation dependent parameters like the permeability are calculated by assuming the phase distributions at a given capillary pressure to be stationary. Then one-phase flow simulations are applied in the corresponding phase space. Upscaling the results of these microscopic flow simulations yields the required effective parameters.

The core of a fuel cell is the membrane electrode assembly (MEA). The MEA consists of two porous electrodes that are separated by an ion conducting membrane. The electrodes are a complex structure of open gas pores, proton conducting ionomer and catalyst on highly porous carbon support that serves as the electron conducting phase.

For excellent performance, high catalyst utilization and, an extensive three-phase boundary is necessary. That means the catalyst particle has to be accessible to protons, electrons and the reaction educts. Therefore, a good dispersion of the ionomer within the highly porous carbon support is needed. Additionally, the ionomer should have a high water content for good protonic conductivity. However, too much accumulated water in the electrode eventually blocks the open gas pores.As a result the educts cannot reach the whole active layer and the electrochemical reaction is limited to only a part of the catalyst particles. This phenomenon is known as flooding of the electrode.

Another critical point is the electrochemical reaction occuring on the cathode side. There is now broad consensus in the literature regarding the reaction mechanism of the oxygen reduction reaction. Since a major part of the fuel cell losses is caused by the oxygen reduction reaction, knowledge of the reaction mechanism and kinetic parameters is essential.

These phenomena will be included in a detailed time-dependent two-phase electrode model which takes into account the electrode microstructure, the spatially resolved educt concentrations, the generation and accumulation of liquid water and the overpotential.
Dynamic simulations of the electrode behavior are compared to dynamic experimental measurement techniques such as:

• cyclic voltammetry
• current interrupt
• impulse and step response
• sine-wave testing
• electorchemical impedance spectroscopy

Relevant operating conditions are, for example:

• relative humidity of the inlet gases
• gas and cell temperature
• flow rates
• gas pressure
• cell potential
• load current

By means of inverse modeling relevant parameters can be extracted from the experimental data. Moreover, homogenization is used to derive a reduced model based on effective parameters and equations for the use in efficient unit cell and stack models.