Membrane Electrode Assembly (MEA)

Two main production paths exist for the preparation of this core element of a polymer exchange membrane (PEM) fuel cell. In both cases, the final result is basically the same, it consists of five layers already described in the introduction (anode diffusion layer, anode catalyst layer, membrane, cathode catalyst layer, cathode diffusion layer).

One preparation methode is to apply the catalyst layers onto the respective diffusion layers and to hot-press these gas diffusion electrodes onto either side of the membrane. This method is widely used, as pre-fabricated gas diffusion electrodes are commercially available (e.g.

from E-TEK, ElectroChem etc.). The disadvantages are the very high prices for these electrodes and the lack of information one gets from the suppliers with respect to physical properties (porosity, type and manufacturer of the catalyst, thicknesses of the layers of the

Figure 3-13 All parts of the DMFC

58 3 Experimental Setup

structure etc.). Especially the latter is very disadvantageous for the formulation of mathematical models of the physical processes within the fuel cell, as such modeling places the demand for reliable parameters of the structures in focus. Another disadvantage of the pre-fabricated electrodes is the fact, that they are also mainly produced by hand. Each electrode sheet looks different from the other, the thicknesses vary significantly etc.. Therefore, the reproducability of obtained results is in doubt.

For all these reasons it was decided to apply the second widely used method for MEA preparation: The preparation of the catalyst layer on the membrane surface (instead of on the diffusion layer surface). The method applied has been developed by the ZSW Ulm, based on the methods published by W^ILSON and G^OTTESFELD [88]. In the following, the abbreviation MEA always refers to a membrane coated with catalyst layers, but without diffusion layers.

First, the membrane (usually NAFION^TM by DuPont) is pre-treated. This pre-treatment consists of boiling the membrane consecutively in hydrogen peroxide solution (5 mass-%

H2O2), MILLIPORE^TM conductivity water, sulfuric acid (0.5 m H2SO4) and finally again MILLIPORE^TM conductivity water (1 hour in each solution).

In the next step, the wet membrane is fixed in an aluminium frame to prevent it from shrinking. It is dried in an oven at 120°C. Due to the unavoidable shrinking process

Figure 3-14 DMFC connected to miniplant

3 Experimental Setup 59 accompanying the water loss, the membrane is forming an absolutely smooth hard film under high tension stress within the frame. In this state, the frame is placed on a heated table, where the membrane is placed on top of a teflon-coated heated plate with many small holes in, through which a light vacuum is applied. This ensures that the membrane is constantly heated to 120°C, while being fixed on a flat surface.

Now a catalyst ink prepared from catalyst powder (supplied by Alfa Aesar Johnson Matthey Germany), NAFION^TM solution (supplied by C.G.Processing USA) and MILLIPORE^TM conductivity water is applied to the open membrane surface using an airbrush pistol. By weighing of the whole frame, a defined overall catalyst loading can be reached. A main disadvantage of this method is the obvious problem that an even distribution of the catalyst over the total electrode surface can not be guaranteed. A lot of experimental experience and application of many thin layers is necessary to ensure good and reproducible results.

After application of both catalyst layers onto the membrane, the whole MEA is sintered in an oven at 135°C. This results in mechanically stable catalyst layers, which do not break even when the MEA is bent. Then, diffusion layers made from carbon paper (e.g. TORAY^TM TGP-H060) or carbon cloth (e.g. Carbel^TMfrom GORE) can be put on either side of the MEA, and the whole sandwich structure can be mounted between the monopolar plates of the fuel cell. If desired, the diffusion layer can also be hot-pressed onto the MEA to get a compact sandwich, but this was not done in the presented work.

Unsupported platinum black (Alfa Aesar Johnson Matthey HiSPEC^TM 1000) and platinum-ruthenium black (Alfa Aesar Johnson Matthey HiSPEC^TM 6000; Pt:Ru ratio as stated by the supplier: Pt 66 wt%, Ru 34 wt% meaning an atomic ratio of 1:1; according to XRD measurements Pt 45 atomic%, Ru 55 atomic% [89]) were used as catalysts.

For preparation of the catalyst inks, the NAFION^TM solution (DuPont EW1100, 15 wt%

NAFION^TMdissolved in an unknown mixture of alcohols, other solvents and water) is cleared from hydrocarbons by vacuum distillation and further dilution with MILLIPORE^TM conductivity water. The final ready-to-use solution contains 5-8 wt% NAFION^TM in water.

The MEA is prepared based on a NAFION^TMN105 membrane, with catalyst loadings of 5 mg metal per square centimeter on both anode and cathode side. The resulting MEA has a membrane thickness of 100-110 µm and the catalyst layers are each about 35 µm thick. This means the preparation method leads to a decrease of the membrane'sthickness of about 20%

(original thickness of NAFION^TM N105: 125-130 µm).

60 3 Experimental Setup The morphology of the resulting catalyst layers can be studied by SEM (scanning electron microscope) images taken from the surface of some produced MEAs (images taken by W^EINBERG at the Fritz-Haber Institute of the Max Planck Society, Berlin, Germany), as shown in Figure 3-15.

The structure corresponds to that shown schematically in Figure 1-6. One can see that the catalyst particles are forming agglomerates, which are glued together by ionomer threads. The latter can not be seen on SEM images, but have been found by XPS scans over the surface (distinct due to the fluorine content in the used ionomer NAFION^TM, XPS results not presented here as they were not part of this work and need a detailed accompanying analysis and discussion). At the outer surface, the diffusion layer made from carbon paper is ensuring electrical connection, on the other side is the membrane, which transports the protons. As both, electronic and ionic conductivity are necessary to carry out the electrochemical reactions at the catalyst surface, only those catalyst particles are active, which are are in contact with both the diffusion layer material (via other catalyst particles or directly) and a continuous thread of ionomer connected to the membrane. Also it might not be desirable to have the catalyst surface covered totally by the ionomer, as it is not yet understood whether this might place a high mass transfer resistance for the educts and the products, and therefore lead to at

Figure 3-15 SEM image from MEA surface (only metal particles visible)

3 Experimental Setup 61 least a reduced activity of the respective particles. Ideally, therefore, an active catalyst particle has to have also a certain amount of free surface in contact with the fluid phase.

An important physical parameter of the catalyst layer is the porosity . It can be calculated from the catalyst loading, the ink composition and some physical properties of the used materials. The calculation is presented in the appendix (chapter 9.2.2). The obtained values are = 0.81 for the anode catalyst layer, and = 0.86 for the cathode catalyst layer. The porosities are obviously very high, which is of course desired to a certain degree to achieve as many accessible catalyst sites as possible for the fluid phases. Nonetheless also an appropriate amount of ionomer phase is necessary to ensure good protonic contact of the catalyst particles.