Passive water management at the cathode of a planar air-breathing proton exchange membrane fuel cell

Abstract

Water management is a significant challenge in portable polymer electrolyte membrane (PEM) fuel cells and particularly in proton exchange membrane (PEM) fuel cells with air-breathing cathodes. Liquid water condensation and accumulation at the cathode surface is unavoidable in a passive design operated over a wide range of ambient and load conditions. Excessive flooding or dry out of the open cathode can lead to a dramatic reduction of fuel cell power. We report a water management design based on a hydrophilic and electrically conductive wick. A prototype air-breathing fuel cell with the proposed water management design successfully operated under severe flooding conditions, ambient temperature 10 °C and relative humidity of 80%, for up to 6 h with no observable cathode flooding or loss of performance.

Introduction

The demand for improved power solutions for portable applications has spurred significant research into miniature fuel cells. Traditional fuel cell systems are complex and bulky; often requiring many auxiliary components such as humidifiers, pumps, compressors, fans and heat sinks. These are in addition to the fuel container and the actual fuel cell. The challenge faced by designers of a miniature fuel cell aimed at replacing batteries in portable power applications is in downscaling and simplifying the existing fuel cell architectures, and at the same time matching the robustness, range of operating conditions and cost of batteries and outperforming their energy density [1].

Water management is a crucial part of any fuel cell system and can be particularly challenging to miniaturize. Most fuel cells for portable applications use polymer electrolyte membranes (PEM), such as Nafion™, as an electrolyte. To obtain high ionic conductivity, these membranes must be fully saturated with water [2], [3], [4]. For efficient operation, therefore, PEM-based fuel cells require water-balanced operation under all operating conditions. Insufficient water production leads to membrane and catalyst layer dry out, causing an increase in resistive and activation losses. On the other hand, excessive water production leads to condensation and flooding, which in turn causes an increase in mass transfer losses [5], [6].

Many approaches have been developed for fuel cell water management: upstream reactant flow humidification [2], [7], cathode water removal by positive cathode to anode gas pressure differential [8], [9], water removal by super-stoichiometric flow [10], [11] or transient purge [12], [13], and water removal by the formation of a pressure gradient between a flow channel and a coolant channel through a porous bipolar plate [14]. For a detailed review see also [15]. Although these water management techniques differ in implementation, they all require bulky components such as humidifiers, blowers, compressors, and are largely unsuitable for miniature planar architectures.

Planar air-breathing polymer electrolyte fuel cells are attractive for the smallest portable power applications due to their simple construction and minimum balance of plant [16], [17], [18], [19], [20], [21], [22], [23]. They are characterized by open-cathode architectures whereby delivery of oxygen from ambient air occurs by diffusion and natural convection (or small amounts of forced convection by local atmospheric or room conditions). As a consequence of the open-cathode construction, water-balanced operation is difficult to achieve over a wide range of operating conditions. None of the aforementioned techniques for water management can be applied in a planar open-cathode design due to the lack of control of the relative humidity and temperature of the ambient air as well as the absence of flow channels with pressure gradients for liquid water advection.

The specific architecture and requirements posed by air-breathing cells demand nontraditional water management approaches. Yao et al. [24] proposed a water management design at the cathode of a miniature air-breathing direct methanol fuel cell (DMFC) based on selective hydrophilic/hydrophobic cathode patterning, variation in cathode through-hole diameter, and gravity driven droplet flow. This architecture enables recirculation of product water but is strongly orientation dependent. In simultaneous work, we presented a demonstration of the use of patterned wicks for water management in air-breathing cathodes [25].

Water management with hydrophilic wicks or water-collecting layers has already been demonstrated in fuel cells. For example, Ge et al. [26], [27] embedded polyvinyl alcohol (PVA) wicks into gas flow channels for water redistribution. However, internal wicks alone cannot prevent flooding since they have finite storage capacity. Furthermore, PVA wicks are non-conductive, require flow channel modifications, and obstruct current paths. As an alternative, Yi et al. [14] implemented a hydrophilic porous graphite plate to serve simultaneously as gas flow manifold and to remove excess water. Water flow from the cathode channels through the porous plate is accomplished by a positive pressure gradient between the cathode flow channels and coolant channels and/or anode flow channels located in the vicinity of the cathode flow channels. This approach simplifies the integration of the wicking structure and can eliminate flooding in a closed stack manifold, but cannot be directly implemented in an air-breathing cathode due to the lack of a closed cathode channel and the requirement for a compressor.

In this paper we report on an approximately scalable, orientation-independent, water management design for an open-air breathing cathode based on a hydrophilic and electrically conductive wick. The hydrophilic conductive wick is in direct contact with cathode gas diffusion layer (GDL) and serves as a water collector and distributor layer which hydraulically links the entire cathode surface. We then discuss the performance of this wicking strategy in the context of a simple cathode water balance model based on our earlier experimental observations [28] and modeling efforts [5], and propose modifications to our model to capture surface flooding effects.