Water Transport in Gas Diffusion Layers of PEMFCs
Shawn Litster, Aimy Bazylak, David Sinton and Ned Djilali
ECS Trans. 2006, Volume 3, Issue 1, Pages 409-414.
doi: 10.1149/1.2356161
Email alerting
service
Receive free email alerts when new articles cite this article - sign up in the
box at the top right corner of the article or click here
To subscribe to ECS Transactions go to:
http://ecst.ecsdl.org/subscriptions
© 2006 ECS - The Electrochemical Society
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
ECS Transactions, 3 (1) 409-414 (2006)
10.1149/1.2356161, copyright The Electrochemical Society
Water Transport in Gas Diffusion Layers of PEMFCs
Shawn Litster, Aimy Bazylak, David Sinton, Ned Djilali
IESVic, University of Victoria, Victoria, BC V8W 3P6, Canada
We present an ex-situ analysis of water transport in the carbon
paper gas diffusion layers for PEM fuel cells. We employ
florescent imaging that allows us to distinguish liquid water from
air through an optical microscope by seeding the water with
fluorescent molecules. We were able to infer the vertical location
of water and create three-dimensional reconstructions of the water
surface by correlating surface height with image intensity. Our
study investigated water phase movement in an uncompressed
GDL as well as the additional effect of the localized compression
that a fuel cell’s bipolar plate introduces. The visualizations
indicate that localized compression promotes regions of increased
wettability. We put forth that the increased wettability results from
fractured carbon fibers and the degradation of the PTFE coating.
Introduction
Proper water management in polymer electrolyte membrane fuel cells (PEMFC) is
required to maintain an adequate level of hydration in the PEM while avoiding
performance decline due to liquid water flooding. Flooding occurs in both the gas
delivery channels and within the pores of the gas diffusion layers (GDLs). There are
numerous strategies for removing water from the gas channels (1, 2), which can be as
simple as running high gas flow rates. However, flooding in gas diffusion layers can be
more difficult to address since the surface tension in the porous media can entrap the
water. The common remedy is to treat the porous carbon material with a hydrophobic
agent to ensure that liquid water exists in only a small fraction of pore network and place
a micro-porous layer between the catalyst layer and macro-porous GDL (3). However,
the concepts of the transport mechanisms in GDL materials have largely been speculative.
In our initial work (4) we performed ex-situ visualizations of liquid flow in a carbon
paper GDL to elucidate the liquid water transport mechanism. Our visualization method
utilized fluorescence microscopy techniques that distinguish the gas and liquid phases by
an optical means. The spatial resolution of images was 5.35 µm pixel-1 and the vertical
resolution was 0.11 µm. Sinha et al. (5) explored the use of microtomography to
determine the saturation of gas diffusion layer materials. An advantage of their method is
that it enables vertical scanning of the GDL layer to generate three-dimensional
reconstructions. However, the vertical resolution of their method was 13.4 µm (5), which
was too large to visualize water interactions with individual fibers in the GDL that are
roughly 8 µm in diameter. Zhang et al. (6) used neutron radiography to measure water
saturation in the gas diffusion layers during fuel cell operation. Their method proved
capable of detecting the variation in saturation at the scale of the fuel cell channel, but not
at the pore scale. In this paper, we present our method for visualizing capillary transport
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
409
ECS Transactions, 3 (1) 409-414 (2006)
of water in GDLs at the pore-level and report on the extension of this method for
studying flow in GDLs under the localized compression caused by a fuel cell’s flow
fields.
Experimental
We optically distinguish water from air in the GDL by seeding the water with
fluorescent molecules. The fluorescent molecules absorb photons at the excitation beam
wavelength (blue light, λ = 513 nm); the molecules then dissipate a portion of the photon
energy and emit photons at a longer wavelength (green light, λ = 490 nm). In our
experiments, we visualized the capillary transport of water in Toray TGP-H-60 gas
diffusion media (10% wt. PTFE treatment). We seeded the water with a 0.5mM
concentration of fluorescein dye (332 MW, Molecular Probes) and pumped that solution
at a constant flow rate through a GDL, which was compressed above a large fluid
reservoir (4). A mask featuring a 2 × 2 mm opening was placed on the reservoir side of
the GDL to control the location of water flux. A Leica Microsystems DM LM optical
microscope with an 8-bit cooled CCD camera captures the transient images. To quantify
the liquid height in each pore, we exploit the depth of field property of the microscope
objective (5X, NA = 0.15), which provides a vertical range of 30µm when inferring
fluorescein solution height from image intensity (4). This vertical range allows us to
resolve water transport within four fiber cross-sections of the GDL.
Figure 1. Schematic of the fluorescence microscopy apparatus for visualizing liquid
transport in GDLs.
Results
Uncompressed GDL
The left hand image in Figure 2 presents the water rise in the GDL where the
fluorescein/water solution produces the dark areas. The right hand image is the threedimensional rendering of the liquid height that we inferred from the image intensity data.
Figure 3 presents a plot of surface height versus time for five specific pores in the images
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
410
ECS Transactions, 3 (1) 409-414 (2006)
of Figure 2. Path 1 rose quickly, but abruptly stopped when it intersected a small fiber
cross-section, and at that time paths 2 and 3 began to rise more rapidly to maintain the
flow rate. At approximately 200 s, the coupling between flow in adjacent pores is evident
when the flow in path 2 rapidly rose due to a local pore expansion, and the height in path
3 dropped due to the local pressure drop induced by the lowered capillary pressure in
path 2.
Figure 2. Image of dye transport in a TGP-H-060 GDL with 10% wt. PTFE treatment
with the contrast inverted (left) and 3D rendering of the liquid height (right) at 145 sec.
Figure 3. Identification of individual water pathways in the GDL (left). Water height in
the pathways of the GDL (right). Key characteristics include the rapid rise of path 2 that
correlates the drop in water height in Path 3 at 200 sec., which indicates pore
breakthrough phenomena.
From our experimental observations, we propose a revised hypothesis for the water
transport mechanism in GDLs. Prior works (7, 8) depict capillary transport in GDLs as a
network of pore flows that converge at the GDL surface, which resemble an “upsidedown tree.” However, our analysis indicates that a fingering and channeling hypothesis
is more correct. These results show that the water will travel from the catalyst layer to the
gas channel via a path constructed from the series of interconnected pores with the
greatest pore radii. When the liquid path intersects a small fiber cross-section, the flow
stops at that location and redirects itself through the largest available empty pore. Our
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
411
ECS Transactions, 3 (1) 409-414 (2006)
hypothesis has implications for the modeling of two-phase flow in PEMFCs as it implies
that the liquid saturation in GDLs is highly dependent on the heterogeneity and
anisotropy of the porous structure.
Effect of Channel Structure and Compression
PEMFCs are typically compressed to ensure proper sealing and minimize contact
resistance. Various studies have examined the effect of compression: Lee et al. (9)
presented polarization curves as a function of the compression pressure; Escribano et al.
(10) showed that higher compression induced lower performance due to increased mass
transfer losses; Ge et al. (11) found that generally performance decreases with increasing
assembly pressure. In a working fuel cell, a bipolar plate imposes the entire compression
force onto the GDL over the land area. The reduced area amplifies the mechanical
stresses on the GDL and may cause the GDL to fracture. It is previously unknown to
what degree this fracturing affects capillary flow in GDLs.
We have modified the previous set-up to investigate the effect of land area interfaces
and compression on liquid water transport in GDLs. Figure 5 provides a schematic of the
new apparatus for holding the GDL. The channel land area is composed of borosilicate
glass with a uniform thickness of 150µm.
Figure 4. Schematic of Plexiglas GDL holding apparatus.
Figure 5 shows the fluorescein distribution in a GDL, placed beneath a horizontal gas
flow channel located midway through the image. The second image capture is 300 ms
after the first. The gas flow channel floods immediately following the liquid
breakthrough under the land area. In our observations, it was evident that the water
flooding the channel was emerging from under the land area. Unfortunately, the frame
rate of the image capturing system was not fast enough to capture this phenomenon. The
results indicate that preferential pathways for liquid transport exist under the land area
due to the relative hydrophilic surface of the land area combined with locations of
relatively hydrophilic pathways created by the local compression.
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
412
ECS Transactions, 3 (1) 409-414 (2006)
Figure 5. Image of dye transport in a TGP-H-060 GDL with 10% wt. PTFE treatment
below a gas flow channel (left) and a 3D rendering of the liquid height (right).
The compression of gas diffusion layers under the land area introduces morphological
changes in the material that can affect the capillary transport of water. SEM images made
before and after compression, shown in
Figure 6, illustrate these changes. The images show that compression fractured a
large portion of the fibers and degraded the hydrophobic PTFE coating. The degraded
PTFE coating exposed areas of bare carbon surface, which are significantly more
hydrophilic. The wettability of the locally compressed GDL visibly increased due to
larger mean pore diameters and lower contact angles generated by fiber breakage and
PTFE degradation. Our observations of a preferential water pathway under the land area
support the suggestion that increased wettability exists under the land area. The
consequence of the increased wettability is problematic for PEM fuel cell operation since
the advective action of the gas flow for removing water is limited under the land area.
Figure 6. SEM image of an uncompressed Toray TGP-H-060 GDL (left), and a SEM
image of the same material after being compressed for five minutes at 0.68 kPa (right).
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
413
ECS Transactions, 3 (1) 409-414 (2006)
Conclusion
We visualized water transport through the gas diffusion layer of a PEM fuel cell in
two scenarios. The first scenario is a uniform GDL that is not under compression but has
a mask with a small opening to control the location of the water flux. In a second scenario,
we imposed a clamping pressure on a GDL using the geometry of a fuel cell channel. The
analysis of water transport in individual pores in the first scenario supports a “channeling
and fingering” model of water transport. In this model, the pore size distribution dictates
the direction of the water transport. With the introduction of localized compression, we
reveal the emergence of preferential water pathways in the compressed regions of the
GDL. The preferential water pathway results from the fracture of the carbon fibers and
the degradation of the PTFE coating.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
C. Buie, J. D. Posner, T. Fabian, S.-K. Cha, J. Eaton, F. Prinz and J. G. Santiago,
J. Power Sources, In Press (2005).
J. S. Yi, J. D. L. Yang and C. King, AIChE J., 50, 2594 (2004).
G. G. Park, Y. J. Sohn, T. H. Yang, Y. G. Yoon, W. Y. Lee and C. S. Kim, J.
Power Sources, 131, 182 (2004).
S. Litster, D. Sinton and N. Djilali, J. Power Sources, 154, 95 (2006).
P. K. Sinha, P. Halleck and C.-Y. Wang, Electrochem. and Solid-State Letters, 9,
A344 (2006).
J. Zhang, D. Kramer, R. Shimoi, Y. Ono, E. Lehmann, A. Wokaun, K. Shinohara
and G. G. Scherer, Electrochimica Acta, 51, 2715 (2006).
J. H. Nam and M. Kaviany, Int. J. Heat Mass Transfer, 46, 4595 (2003).
U. Pasaogullari and C. Y. Wang, J. Electrochem. Soc., 151, A399 (2004).
W.-k. Lee, C.-H. Ho, J. W. Van Zee and M. Murthy, J. Power Sources, 84, 45
(1999).
S. Escribano, J.-F. Blachot, J. Etheve, A. Morin and R. Mosdale, J. Power
Sources, 156, 8 (2006).
J. Ge, A. Higier and H. Liu, J. Power Sources, In Press (2006).
Downloaded on 2012-09-19 to IP 128.100.48.214 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
414