A Study of Channel Dimension Effect On Performance of Interdigitated Flow Field PEM Fuel Cells

            Novel water management and reactant distribution strategies are important to the development of next generation polymer electrolyte membrane fuel cell systems (PEMFCs). Over-hydration (flooding) or under-hydration (membrane drying) of channels and critical membrane electrode assembly components can have detrimental effects on cell performance and lifetimes. Improving these strategies in PEMFCs leads to higher power density and reduced stack size for vehicle applications, which in turn reduces weight and improves the price competitiveness of fuel cell systems.

            Parallel and interdigitated flow fields, Figure 1, are two common types of PEMFC designs that have benefits and drawbacks depending upon operating conditions and flow-field geometry. Parallel flow fields suffer from longer diffusion lengths which inhibit delivery of reactants and removal of byproduct water. Interdigitated flow fields induce convective transport, known as cross flow, through the porous GDL between adjacent channels and therefore are superior at water removal beneath land areas, which can lead to higher cell performance. Unfortunately, forcing flow through the GDL results in higher pumping losses as the inlet pressure for interdigitated flow fields can be up to an order of magnitude greater than that for a parallel flow field. Additionally, the pressure gradient between inlet and outlet channels in an interdigitated flow field may not be evenly distributed along the channel length. This may cause cross flow maldistribution in the same way Z-manifolds may unevenly distribute gasses across the system. Such maldistribution of reactant gasses along the active area of a cell may lead to areas of low cross flow, and areas of excessive cross flow. This, in turn, can cause areas of low oxygen concentration, water build up, which can lead to higher pressure losses and uneven membrane hydration all of which reduce overall cell performance. This research seeks to examine the affect varying channel dimension sizes have on the performance of a fuel cell under interdigitated

Figure 1: Interdigitated and parallel diagrams. Inlet channels: PH (high pressure), Outlet channels: PL (low pressure)

conditions. The features of interest are channel length, width, and depth. The width of lands for the bipolar plates mirrors the width of channels on that plate. Changing feature size affects the distribution of liquid water, as well as the distribution of gaseous reactant in the cell. This work will help develop a better understanding of cross flow for interdigitated flow fields.

            For the length study portion of this body of work, a PEMFC was designed that could be configured to run under varying interdigitated channel length configurations through the use of separate manifold exit ports. The lengths studied were 5 cm, 15 cm and 25 cm each with a width of 2 cm. The flow field is comprised of 10 channels with width/depth/land dimensions all of 1 mm. The cell was controlled using stoichiometry to account for the change in active area at the different channel lengths. Tests were performed at stoichiometric conditions of 1.5 on the anode and 2.0 on the cathode (normal), and 2.0 on the anode and 4.0 on the cathode (high). Humidity, temperature and cell compression were held constant for all trials. Pressure was measured at the inlet of the cell while the outlets were exhausted to atmosphere. The cathode was run in interdigitated configuration, while the anode was run in parallel to isolate the effects of the cathode. Polarization and inlet pressure curves were produced for each length case at the two different stoichiometries. A total of three trials for each case were conducted in random order so as to verify repeatability. Figure 2 displays the average polarization, inlet pressure and net system power data, taking into account pumping losses, for each different length. Reduced cell performance with increasing channel length is a trend that was observed.

            This experimental work was complemented by a modeling study of the reactant flow through the GDL at these various lengths and flow rates using the multiphysics package COMSOL. Using the dimensions from the experimental cell, interdigitated flow fields at varying lengths and conditions were simulated in three dimensions. The model focuses on the cathode flow field and GDL layer and does not take into account the electrochemical reaction. As this study is interested in how channel length affects pressure and cross flow distribution, modeling the entire electrochemical process is not necessary since the electrochemical reaction has little effect on bulk cross flow; additionally, the experimental portion of this work characterizes cell performance. Reaction by-product water is a factor;

Figure 2: Polarization, inlet pressure (cathode) and net system power for each length test case. The top graphs are for the high stoichiometry condition (2.0/4.0) and the lower are for the normal condition (1.5/2.0).

however, no models can currently account for this completely. Figure 3 displays some of the key results from the modeling work. The pressure distribution in inlet and outlet channels was shown to be more variable in the 25 cm cell than in the 5 cm cell. As a result, the cross flow between adjacent channels in the long cell is maldistributed along its length compared with the short cell, which shows a very consistent cross flow rate. This effect may help explain why the shorter channel length configuration produced the best performance. The center region of the 25 cm cell, where the cross flow rate is lower, may be subject to reduced oxygen concentration and increased water buildup. Regions toward the entrance and exit, while receiving higher cross flow rates, may be oversupplied and contributing to increased pressure drop due to higher velocities. 

The even cross flow rate in the short cell appears to be more advantageous when these results are compared to the experimental data and suggests that interdigitated designs are sensitive to maldistribution along the channel length. The results of this work were compared to a similar earlier study on parallel flow field channel length; the overall trends are summarized in Figure 4. These findings suggest that for large aspect ratio fuel cells (e.g. long channel lengths and small channel widths) if interdigitated designs are attempted, attention must be paid to the distribution of reactant gas cross flow. Compared to parallel flow fields, which follow a reverse trend due to increased slug removal with longer channel length, interdigitated design flow patterns should be oriented so channel length is minimized. Neutron radiography is being conducted to study the water removal characteristics of interdigitated

Figure 3: (Top) Pressure distribution in long and short interdigitated flow fields. (Bottom) Distribution of cross flow rate along the length of the cell.

cells of varying length. Through-plane images of in-situ cell operation should allow for comparison of inlet and outlet water content as well as comparison of GDL water distribution.

            The affect of width and depth of channels are currently being examined. Channel dimensions of interest are (width x depth, in millimeters) 1x1, 0.5x0.5, 0.25x0.25, 0.5x1, and 0.25x1. The effect of these features is being explored through the use of interchangeable bipolar plates. The bipolar plates are 20 cm long and 1.5 cm wide, for a total active area of 30 cm². The bipolar plates are machined out of aluminum and have a nonreactive gold coating on the active areas to prevent corrosion and eliminate unaccounted for reactions on the surface of the plate. The use of gold further reduces the interfacial contact resistance with the GDL, decreasing the voltage loss due to ohmic resistance.

            These bipolar plates can fit into a cell superstructure that has the ability to change from parallel configuration to interdigitated configuration through the use of valves and a novel manifold – channel connection design. The valves close the inflow of gas to every other channel, and close the outflow of gas to the remaining channels. This can be done without taking apart the cell, allowing for direct in situ comparison of parallel and interdigitated flow fields.

            The work of this study has already been started, with preliminary testing of a nickel plated version of the 1 x 1 mm plates already completed. These results can serve as a baseline for comparison of the gold plated set of bipolar plates. This preliminary testing has further given a better understanding of how the system behaves during operation and improvements which may be made in the new plates.

            Further research will be required in the area of modeling fuel cells with electrochemical reactions to

Figure 4: Summary of parallel and interdigitated max power density trends with channel length.

determine the effect of these channel dimensions on cell performance. Neutron radiography will need to further be performed on the plates with varying channel widths and depths to isolate the water removal characteristics of these designs. Coupling this data with performance trends should give researchers and cell industry a better understanding of design consequences.