Bipolar Plates

Flow field plates in early fuel cell designs — and still in use in the laboratory — were usually made of graphite into which flow channels were conveniently machined. These plates have high electronic and good thermal conductivity and are stable in the chemical environment inside a fuel cell. Raw bulk graphite is made in a high-temperature sintering process that takes several weeks and leads to shape distortions and the introduction of some porosity in the plates. Hence, making flow field plates is a lengthy and labor-intensive process, involving sawing blocks of raw material into slabs of the required thickness, vacuum-impregnating the blocks or the cut slabs with some resin filler for gas-tightness (Washington et al., 1994), and grinding and polishing to the desired surface finish. Only then can the gas flow fields be machined into the blank plates by a standard milling and engraving process. The material is easily machined but abrasive. Flow field plates made in this way are usually several millimeters (1 mm = 0.04 inch) thick, mainly to give them mechanical strength and allow the engraving of flow channels. This approach allows the greatest possible flexibility with respect to designing and optimizing the flow field.

When building stacks, flow fields can be machined on either side of the flow field plate such that it forms the cathode plate on one and the anode plate on the other side. Therefore, the term bipolar plate in often used in this context. The reactant gases are then passed through sections of the plates and essentially the whole fuel cell stack (see Fig. 4.2).

4.4.1.1 Flow Field Designs and Their Effects on Performance

Let us first consider the main tasks of a flow field plate:

• Current conduction

• Heat conduction

• Product water removal

Figure 4.13 showed that a balance exists between gas supply and current conduction. The best conductor, a solid blank sheet, will not allow any gas access, while an entirely open structure does not allow any current to flow. Therefore, some sort of large-scale "porosity" is needed in flow fields. The rib and channel design shown in Fig. 4.13 is just one way of achieving this. Clearly, the size of the open flow structure (for example channels) depends on the resistivity of the materials used (including that of the gas diffusion layer), the size of the MEA, the operating pressure, and the current range envisaged. The complex task of achieving the right structure can be done by fluid-dynamic modeling in combination with experimental evaluation of a large number of different designs. Modeling will also give sufficiently accurate treatment of heat removal from the power-generating MEA to heat sinks within the stack.

Product water removal (at the cathode) is even more complex as this represents a two-phase flow problem. While some turbulence may help to release water from the open gas diffusion layer structure, opening up room for gas access, turbulent gas flow leads to larger pressure drops between flow field inlet and outlet. Pressure differentials require compression energy to drive the reactants through the flow field structure, and this has an impact on systems efficiency. A possible answer is the construction of many parallel gas channels as indicated in Fig. 4.15(a). Parallel flow through many channels will lower the pressure differential between gas inlet and outlet. Unfortunately, water formed at the cathode accumulates in the channels adjacent to the cathode. Water droplets tend to coalesce and form larger droplets, partially obstructing the channels. Not all channels will be equally blocked, and the main gas flow will be redistributed through the remaining open channels. Ultimately, portions of the MEA will no longer be supplied with reactant gas and will become inactive.

The flow field designs employed by leading stack developers are well-kept secrets. However, the patent literature gives hints as to what principal concepts are used.

The so-called serpentine flow field described by Ballard researchers (Watkins et al., 1992) is sketched in Fig. 4.15(b). It is believed to overcome the blockage of certain channels by product water because the differential pressure between inlet and outlet forces stagnant water out of the channels. When higher flows are needed and widening of channels is not feasible, the number of channels may be increased. The excellent performance of Ballard stacks probably relies on the use of this concept.

General Motors has patented a flow field design sketched in Fig. 4.15(c) that looks like a compromise between straight parallel channels and a single serpentine flow field (a so-called mirrored flow field). The patent (Rock, 2000) is also interesting regarding the materials and the integration of cooling within the bipolar plate, which will be discussed below.

Structures of the type shown in Fig. 4.15(d) have been suggested, where the channels are no longer continuous, but the gas is forced to flow through some part of the gas diffusion layer. The interdigitated structure shown in Fig. 4.15(d) is just one possibility. This structure helps to induce forced water removal from the open structure of the gas diffusion layer but induces higher pressure drops between inlet and outlet than through-flow options do.

Another option is abandoning channels altogether by creating a regular pattern of supporting patches, as shown in Fig. 4.15(e), or dimples, or by using a more or less isotropic gas distribution layer, such as a wire mesh (Wilson and Zawodzinski, 2001) or a foam (Gamburzev et al., 1999), as sketched in Fig. 4.15(f). These two options may work well in a limited current range and single cells or short stacks. When high power is generated in the MEA, gas flow will focus more and more on the central section of the flow field, leaving inactive patches around the edges. What was said about stagnant water in conjunction with the structure shown in Fig. 4.15(a) applies accordingly.

More importantly, it is very difficult to produce irregular structures of the type shown in Fig. 4.15(f) with absolutely identical porosity properties. In a stack, this may lead to differences in flow resistance among different cells, particularly at the high performance end. Certain cells may be shut off because of accumulating water and lack of reactant and, possibly, may be destroyed. The technique of "purging' the stack by venting the outlet at fixed intervals, which is used by some stack manufacturers, may have been applied to fix this problem.

(

—-

-

(

J

±

(

FIGURE 4.15 Various possible flow field designs: (a) Parallel gas channels — this design may lead to pressure imbalances between adjacent channels and gas blockages; (b) serpentine flow field as patented by Ballard Power Systems; (c) mirrored flow field as patented by General Motors; (d) interdigitated flow field with gas conduction through the attached electrode substrate; (e) gas diffusion layer without flow channels (here: ordered support pad structure); (f) metal mesh flow field — the gas passes through the mesh structure below and above the metal wires. (Drawing courtesy of Elke Schnur, Umwelt-Campus Birkenfeld.)

4.4.1.2 Flow Field Design Tools

The best tool available to the flow field designer is probably a powerful computer program (Hontanon et al., 2000) able to solve two-phase flow problems in arbitrary geometries. Yet stack developers have not reported any results of such modeling exercises in the literature.

One technique has been developed for experimental monitoring of the current distribution within a flow field plate. This so-called current mapping makes it possible to study the performance of the combination of flow field and MEA under practical operating conditions, even inside stacks.

FIGURE 4.16 Current distribution of a fuel cell of practical size (304 x 190 mm). The measurement technique was developed by DLR (Germany) (Wieser et al., 2000). The graph shows uneven current distribution due to poor oxidant supply to the cathode as a result of insufficient product water removal using a non-optimized porous isotropic gas distributor (100 mV, 200 A, 80°C, 2 bar abs, humidified air, stoichiometry 1.8). (Graph courtesy of DLR.)

Current Density [mA/cm_]

1850.0 -800.0 -750.0 -700.0 -650.0 -600.0 -550.0 -500.0 -450.0 -400.0 -350.0 -300.0 -250.0 -200.0 -150.0 -100.0 -50.0 -0.0

FIGURE 4.16 Current distribution of a fuel cell of practical size (304 x 190 mm). The measurement technique was developed by DLR (Germany) (Wieser et al., 2000). The graph shows uneven current distribution due to poor oxidant supply to the cathode as a result of insufficient product water removal using a non-optimized porous isotropic gas distributor (100 mV, 200 A, 80°C, 2 bar abs, humidified air, stoichiometry 1.8). (Graph courtesy of DLR.)

Ballard was the first to report current mapping data (Campbell et al., 1997). Figure 4.16 shows the current distribution inside an operating fuel cell. One possible design is a special, segmented plate that fits into the plate arrangement inside a stack. Currents can be measured as voltage drops across each segment. Another approach has been presented by researchers at DLR who incorporated solid state Hall sensors into plate segments (see Fig. 4.16). The Hall sensors are very sensitive to current but also respond to temperature gradients inside the cell (Wieser et al., 2000).

4.4.1.3 Materials and Production Techniques for Bipolar Plates

Some basic materials properties for bipolar plates are listed in Table 4.2. As far as is publicly known, there are currently two competing approaches, the use of graphite-basedflowfieldmaterialsandtheuse of metal. We will discussbothsystemsinturn.

Graphite-Based Materials

The choice of materials for producing bipolar platesin commercialfuel cellstacksisdictatednot only by performance considerations as outlined in Section4.4.1.1 butalsoby cost. Currently, blank graphite plates cost between U.S. $20 and U.S. $50 apiece in smallquantities, i.e., up to U.S. $1000/m2, orperhaps more than U.S. $100/kW, assuming one plate per MEAplus coolingplatesatanMEApowerdensityof 1 Wcm-2 (compare the discussion of membrane costinSection4.3.1.3). Again, automotive costtargets are well beyond reach, even ignoring additional machining and tooling time.

TABLE 4.2 Material Properties Targeted for Bipolar Plates

Material Property

Target Value

Reason

Permeability for gas Electronic conductivity Density

Thermal conductivity

Corrosion resistance Pattern definition

Machining

High <1 mV loss per plate

Low <1 kg/kWel

High

High High

Thermal and pattern stability Medium

Low cost ($15 $/kWel for entire stack)

Separation of anode and cathode compartments High current densities of up to 4 Acm-2 Stack weight

Removal of reaction heat (approx. 1 W per cm2 of MEA)

Proton activity equivalent to 1 M H2SO4 Identical pressure drop across all plates in a fuel cell stack

Gas tightness throughout operating lifetime of 5000 h/50,000 h (automotive/stationary) at 80°C (176°F) Next to MEA most expensive stack component

Mcfc Anode
FIGURE 4.17 Sketch of a serpentine flow field embossed in a graphite foil — Ballard (Wilkinson et al., 1996).

This dilemma has sparked off several alternative approaches. Ballard Power Systems has developed plates based on (laminated) graphite foil, which can be cut (Washington et al., 1994), molded, or carved in relief in order to generate a flow field pattern (Wilkinson et al., 1996) (see Fig. 4.17). This may open up a route to low-cost volume production of bipolar plates. Potential concerns are perhaps the uncertain cost and the availability of the graphite sheet material in large volumes.

Another cost-effective volume production technique is injection or compression molding. Difficulties with molded plates lie in finding the right composition of the material, which is usually a composite of graphite powder in a polymer matrix. Although good electronic conductivity requires a high graphite fill, this hampers the flow and hence the moldability of the composite. Thermal stability and resistance toward chemical attack of the polymers limit the choice of materials. Energy Partners (Barbir et al., 1997) and Los Alamos National Laboratory (LANL) both claim to have found suitable composites with the LANL composite consisting of 68 wt% of graphite powder in a vinyl ester matrix. More recently, patents by LANL (Wilson and Busick, 2001) and Premix (Ohio) (Butler, 2001) were published almost simultaneously. Particularly in the Premix patent, a wide range of recipes for molded plates is described: resins, rheological modifiers, initiators, inhibitors, fibers, and mold releases. The fillers are a combination of a graphite powder of one or more narrow size ranges and a carbon black. Bulk conductivities range from 40 to 96 Scm-1, which is at the lower end of what is currently achieved with machined graphite.

Plates based on a spacer layer of porous carbon have also been presented (Gamburzev et al., 1999). But what was said in Section 4.4.1.1 on performance should be borne in mind.

A slightly different approach is taken by researchers at the Dow Chemical Company. A recent patent (Hinton, et al., 2000) describes a bipolar plate made of two layers of a porous electronically conductive material (for example, carbon fiber paper) with a gas-tight solid layer of some conducting polymeric material in between. The bipolar plates can then be molded with the polymeric material sandwiched between the two porous layers.

Plug Power (Carlstrom, 2000) has patented flow field plates that consist of conducting parts framed by non-conducting material that may form part of the flow field. DuPont has patented the concept of a molded polymer plate (with a metal core) that is made conducting only at the surface by coating with a metal, metal nitride, or metal carbide. Surface conduction will not suffice for high power density stacks but may work with portable systems.

Metallic Bipolar Plates

Metals are very good electronic and thermal conductors and exhibit excellent mechanical properties. Undesired properties are their limited corrosion resistance and the difficulty and cost of machining.

The metals contained in the plates bear the risk of leaching in the harsh electrochemical environment inside a fuel cell stack; leached metal may form damaging deposits on the electrocatalyst layers or could be ion-exchanged into the membrane or the ionomer, thereby decreasing the conductivity (Ma et al., 2000). Corrosion is believed to be more serious at the anode (Makkus et al., 2000), probably due to weakening of the protective oxide layer in the hydrogen atmosphere.

Several grades of stainless steel (310, 316, 904L) have been reported to survive the highly corroding environment inside a fuel cell stack for 3000 h without significant degradation (Davies et al., 2000) by forming a protective passivation layer.

Clearly, the formation of oxide layers reduces the conductivity of the materials employed. Therefore, coatings have been applied in some cases. In the simplest case, this may be a thin layer of gold or titanium (Hodgson et al., 2001). Titanium nitride layers are another possibility and have been applied to lightweight plates made of aluminum or titanium cores with corrosion resistant spacer layers (Yang Li et al., 2001). Whether these approaches are commercially viable depends on the balance between materials and processing cost.

Directing the gas flow is sometimes done by shaping thin metal plates to form dimples or other protrusions. Alternatively, meshes (Wilson and Zawodzinski, 2001) and metal foams (Faita and Mante-gazza, 1996) have been used as spacers. What was said in Section 4.4.1.1 applies to all these approaches.

Meanwhile, mechanical machining of flow fields into solid stainless steel plates is difficult. A number of companies such as Microponents (Birmingham, U.K.) and PEM (Germany) attempt to achieve volume production of flow field plates by employing chemical etching techniques. Yet etching is a slow process and generates slurries containing heavy metals, and it is hence of limited use for mass production.

Another solution to the problem of creating a (serpentine) flow field is using well-known metal bashing techniques. To date, no data on flow fields successfully produced in this fashion have been published, although the GM patent (Rock, 2000) clearly refers to a layered metal flow field plate. Figure 4.18 shows a picture taken from this patent. The bipolar plate is made up of three parts: two thin metal sheets with flow field structures and a metal spacer that allows cooling water flow in between adjacent MEAs. The three layers are for example brazed together.

Working with industrial partners, Umwelt-Campus Birkenfeld has developed a technique for producing well-defined channeled flow fields in large volumes at viable cost (see Fig. 4.19).

Guide to Alternative Fuels

Guide to Alternative Fuels

Your Alternative Fuel Solution for Saving Money, Reducing Oil Dependency, and Helping the Planet. Ethanol is an alternative to gasoline. The use of ethanol has been demonstrated to reduce greenhouse emissions slightly as compared to gasoline. Through this ebook, you are going to learn what you will need to know why choosing an alternative fuel may benefit you and your future.

Get My Free Ebook


Responses

  • clare
    Why are bi polar plates made of graphite?
    8 years ago

Post a comment