Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
0013-4651/2013/160(6)/F487/9/$31.00 © The Electrochemical Society
F487
Effect of Channel Material on Water Droplet Dynamics
in a PEMFC Gas Channel
Preethi Gopalana,∗ and Satish G. Kandlikara,b,∗∗,z
a Microsystems Engineering, Rochester Institute of Technology, Rochester, New York 14623, USA
b Mechanical Engineering, Rochester Institute of Technology, Rochester, New York 14623, USA
Water accumulation in the gas channels affects the gas diffusion to the reaction sites and the overall performance of proton exchange
membrane fuel cells (PEMFCs). Understanding water droplet growth and detachment from the gas diffusion layer (GDL) inside
the channel is critical in developing better water management strategies. It has been reported that the wettability of the channel
affects the droplet-wall interaction. Hence, in this study, the behavior of the liquid water droplet within a channel is experimentally
investigated from a fundamental perspective. Experiments were conducted for different air velocities (0.2–2.4 m/s), GDL types and
channel wall materials which are commercially used in PEMFCs for automotive applications. Experimental results were analyzed
through the Concus-Finn condition using instantaneous dynamic contact angle (IDCA) to determine the corner filling behavior,
which is a precursor to channel flooding. Droplet detachment is influenced by air flow within the channel and the contact angle
hysteresis (CAH) of the wall and the GDL material. Experiments were also performed on the GDLs that have undergone longer test
periods in a fuel cell. Results showed much lower static CA and CAH, and a corresponding change in the droplet dynamics on used
GDL compared to the new GDLs.
© 2013 The Electrochemical Society. [DOI: 10.1149/2.030306jes] All rights reserved.
Manuscript submitted January 8, 2013; revised manuscript received February 25, 2013. Published March 8, 2013. This was Paper
1633 presented at the Honolulu, Hawaii, Meeting of the Society, October 7–12, 2012.
Proton exchange membrane fuel cells (PEMFCs) have shown
tremendous potential in becoming a reliable source for a variety of
power generation applications including automotive purposes.1 Despite many advantages, one of the major outstanding issues in the
commercialization of the PEMFCs in automotive applications is inefficient water management at the startup and shutdown of the system
in the presence of liquid water.2 Water management is of critical importance to enable proper diffusion of gases toward the catalyst layer,
as both the liquid water and the air share the same pathways to initiate
and sustain chemical reactions within the cell. Accumulation of the
liquid water on the cathode side of a PEMFC leads to flooding of the
channels, and thereby hinders the reactant flow toward the reaction
sites, consequently lowering the efficiency of the system. On the other
hand, lack of water in the cell leads to membrane dehydration and
reduction of the proton exchange through the membrane, and thus decreasing the efficiency of the system.3,4 Effective and balanced water
management in PEMFCs is of utmost importance to realize improved
efficiency of the cell prior to commercialization. The primary source
of liquid phase in a PEMFC is the water droplet emerging from the
GDL into the channel. The analysis of the water droplet emergence,
growth, interaction with the channel wall surface, and removal from
the channel as a function of gas flow velocity becomes necessary.
Having a complete understanding of the liquid droplet behavior on
the GDL surface and its interaction with the channel forms the major
motivation for the current work.
There have been several techniques used to visualize the liquid
water emergence and its interaction with the channel walls. Some
of the imaging techniques that have been successfully used are neutron radiography,5–9 X-ray radiography,10 fluorescence microscopy,11
nuclear magnetic resonance (NMR) microscopy,12 and magnetic resonance imaging (MRI).13–15 Drawbacks of using such techniques are
that these are expensive, have limited availability, and require complex system modifications. One of the reliable imaging techniques
that has been used to evaluate the flow regime in the gas channel,
water emergence mechanism, and the droplet interaction with the gas
channel is the visible light imaging. To use the visible light imaging,
the gas channel cover must be transparent or optically clear. Bazylak et al. used an optically clear gas channel to visualize the droplet
breakthrough and its growth in the channel.11 Similarly, Yang et al.
performed experiments using a transparent fuel cell to observe the
droplet emergence pattern on the GDL and its behavior in the gas
∗
∗∗
z
Electrochemical Society Student Member.
Electrochemical Society Active Member.
E-mail: [email protected]
channel.16 It was found that the droplets generally appear at certain
preferential pores and grow to a certain size before detaching from the
GDL. Tuber et al. were among the first few groups to study the effect
of GDL wettability on the water accumulation in the channel using a
transparent fuel cell.17 They found that hydrophilic GDLs result in a
more uniform spread of water on the GDL surface and a decrease in
the fuel cell performance and hence the reduced current densities.
To further understand the mechanism of droplet removal, the surface forces that act within the gas channels on the liquid water and
their influence on the droplet removal have been studied in literature both experimentally (in situ and ex situ) as well as numerically.
Some of these preliminary studies showed that the surface forces are
the primary causes for water accumulation and channel flooding in
the PEMFCs compared to the viscous, pressure or inertial forces.18
The surface forces are mainly dependent on the advancing and receding contact angles, and the hysteresis on the material surface. In
general, different GDL materials used in PEMFCs possess different
surface characteristics, and thus differ in their droplet accumulation
characteristics within the gas channels.19,20 A few detailed studies have
further attempted to evaluate different channel properties, mainly focusing on the liquid water stagnation. It was found that the hydrophilicity of the channel surface facilitates the removal of water by wicking
it into the channel corners.21 These results were in agreement with the
conclusions derived from the more recent studies performed by Lu
et al., who found that the hydrophilic channels help in uniform water
distribution along the GDL surface.22 They stated that the hydrophilic
channels aid in creating a film flow in the channel, and reduce the
pressure drop in the system compared to the hydrophobic channels.
Similar study was also conducted by Zhu et al.,23 however their results
contradicted the conclusion of Zhang et al.21 They proposed that the
spreading of the liquid water near the hydrophilic walls leads to a film
flow and blockage of the reactant channel pathway. They also stated
that the channel geometry has an important role in the spreading of
the liquid water in the channels.
To address the effect of channel geometry on the water accumulation, Zhu et al. conducted simulations on different channel geometries
and found that the upside down trapezoidal channels had the highest
water coverage ratio compared to any other channel configurations.24
It was also noticed that the triangular channels performed more efficiently compared to the rectangular channels. On a similar note,
Lu et al. investigated the effect of channel geometry and channel
orientation on the water hold up in the cell.22 They concluded that
the sinusoidal geometry for the gas channel helps in obtaining lower
pressure drop in the system compared to the trapezoidal or rectangular
channel. Also, the gravity and the shear forces in the channel affect
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
F488
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
Figure 1. Experimental test setup used to
visualize the channel end to visualize the
droplet dynamics with the channel wall and
the base. A detailed description of the channel
is shown by the image on the right side.
the liquid transport in the gas channel. This implies that the PEM
fuel cells in vertical orientation would result in a better performance
compared to the horizontal orientation.
To tackle the water stagnation issue at a much more fundamental
level, Rath and Kandlikar analyzed the droplet interactions with the
sidewall of the channel at different open angles (angle between the
GDL and the channel sidewall).25 The Concus-Finn condition was
used to predict the filling behavior of the droplet in the corners of the
channel.26,27 They found that 52◦ and lower channel angles would lead
to droplet pinning to the sidewall of the channel and result in no corner
filling. Gopalan and Kandlikar extended this work by incorporating
an air flow into the system to simulate a realistic PEMFC working
condition.28,29 It was established that the corner filling of the channel
is more dynamic in nature and depends upon the instantaneous dynamic contact angle (IDCA) that the droplet makes with the sidewall
and the GDL at a given air velocity. These studies were limited to only
one particular channel wall and GDL material. However, investigation
of the droplet growth and removal pattern on different pairs of materials that are commercially used for the automotive application would
be beneficial for understanding the effect of material properties on
the droplet dynamics. To address this fundamental issue, this work focuses on analyzing the effects of four channel wall materials (stainless
steel (SS-2205), copper (Cu-110), graphite composite, polycarbonate
(Lexan)) and three GDL types (MRC-105, SGL-25BC, TGP-H-060)
on the liquid water droplet dynamics in the presence of airflow within
the gas channel.
Ex-Situ Experimental
Experimental set up.— A schematic of the ex-situ setup used in
the present study is shown in Figure 1. The test setup consists of a
base plate made of polycarbonate, which holds the GDL on top of it.
An inlet for the water droplet was made on the base plate near the
channel end to allow the water to emerge onto the GDL surface and
interact with the sidewall. The channel sidewall plates are made up of
different materials. These plates are studied under two different open
angles of 45◦ and 50◦ . These plates were placed on top of the GDL
layer. The top wall (above the sidewall plates) was attached to this
setup forming a complete 100 mm long channel. In our experiments,
polycarbonate (Lexan), stainless steel (SS-2205), copper (Cu-110)
and graphite composite materials that are commonly used for the
bipolar plates in the PEMFCs were used for the channel sidewalls
as well as for the top plate. SGL-25BC, MRC-105 with 6% PTFE
and TGP-H-060 with 6% PTFE were used as the GDL materials.
To evaluate the performance difference due to aging, two MRC-105
GDLs that were run for 40 and 125 hours by Lu et al.22 were compared
with a fresh sample. Furthermore, the superficial air velocities were
varied between 0.2–2.4 m/s, which corresponded to a current density
range of 0.1–1 A/cm2 in a typical fuel cell with an active area of
50 cm2 .6 The Bond number and the Weber number calculated for the
given experimental conditions were in the range of 0.1–0.2 and 1–50
respectively. The water flow rate was kept constant at 0.05 mL/min
(corresponding to the water generation rate for a current density of
3 A/cm2 for an active area of 2.9 cm2 ). More detailed description
of the experimental setup can be found in the paper by Gopalan and
Kandlikar.28,30
For each of the materials considered in this study, the advancing
and receding contact angles were measured using the VCA Optima
Surface Analysis System. The contact angle hysteresis (CAH) was
calculated as θhys = θadv – θrec . The detailed values are listed in
Table I. The Concus-Finn condition was used for the theoretical estimation of the transition angle from non-filling to filling condition for
a given material pair as shown in Table II. The Concus-Finn condition
characterizing the contact angle range for which de-wetting of the
corner occurs is given by
2α < (θ B + θW ) − π
[1]
where θB is the advancing contact angle for the base material (GDL),
θW is the advancing contact angle for the sidewall and 2α is the channel
open angle (angle between the sidewall and GDL). According to the
above condition, the channel corner will not be filled if the above
inequality is satisfied. It was found from the data that all material
combinations had transition angles near 50◦ (48◦ –53◦ ). An open angle
smaller than the transition angle showed no corner filling, and any
open angle above the transition angle showed corner filling behavior.
Since the current study involves several variables, the experimental
test matrix was simplified to an extent by only including 45◦ and 50◦
channel open angles as these angles are closer to the transition angle.
Experimental procedure.— For each pair of materials used in the
experiments, the air flow was set to the desired value and the droplet
was allowed to grow on the GDL surface. High speed videos were
recorded using Keyence VW-6000 camera at 60–125 fps from the time
Table I. Static advancing and receding contact angles of sidewall
materials.
Material
Advancing
Contact
Angle (θadv )
Receding
Contact
Angle (θrec )
Hysteresis
(θhys )
MRC-105 with 6% PTFE
TGP-H-060 with 6% PTFE
SGL-25BC
40 hours MRC-105 GDL
125 hours MRC-105 GDL
Polycarbonate
Copper (Cu-110)
Graphite Composite
Stainless Steel (SS-2205)
148
145
148
144
132
85
80
84
81
138
127
136
136
123
61
53
34
60
10
17
12
8
9
24
27
50
19
Table II. Transition Angles from Non-Filling to Filling for a Given
Material Pair.
Material Pair
Transition Angles
MRC-105 GDL- Polycarbonate
MRC-105 GDL – Copper (Cu-110)
MRC-105 GDL – Graphite
MRC-105 GDL – Stainless Steel (SS-2205)
TGP-H-060 – Polycarbonate
SGL-25BC- Polycarbonate
40 hours MRC-105 GDL- Polycarbonate
125 hours MRC-105 GDL- Polycarbonate
53◦
48◦
52◦
49◦
50◦
51◦
49◦
37◦
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
the droplets appeared on the GDL to the time they were completely
removed from the channel. Typically, in all our experiments, the following droplet growth sequence was observed: A droplet appears on
the GDL and grows to its full size before contacting the sidewall.
Once the droplet touches the sidewall it either fills the channel corner
or remains pinned on the sidewall depending on the channel open
angle until the pressure inside the channel builds up leading to the
removal of the droplet. The high speed videos of the droplet growth
and its interaction with the channel wall were post-processed using
the Keyence motion analyzer. The IDCA the droplet made with the
channel wall and the GDL was measured on each frame. The uncertainty in the measurement of the contact angle was approximately
± 1◦ . Pressure drop across the droplet was measured using a Honeywell FWD differential pressure sensor (0–1 psi range). The uncertainty
in the pressure sensor measurement was ± 0.25% of full scale reading.
Results
Concus-Finn condition.— This condition was used to predict the
liquid droplet behavior at the intersection of the GDL and the sidewall
surfaces. The contact angles the droplet makes with different surfaces
are shown in Figure 2a, and these angles are plotted on a 2D graph as
shown in Figure 2b. It can be inferred from this graph that if the contact
angle points lie in the shaded region, the droplet will fill the channel
corners. Otherwise, the droplet will remain pinned at the channel
sidewall without filling the channel corners.27 In PEMFCs, once the
droplet fills the channel corner, the successive droplets emerging from
the GDL surface near the corner will merge with this water, and may
eventually lead to channel flooding, depending on the airflow rate.
Therefore, it is essential to avoid corner filling in the channel to
prevent such degenerative effects.
In our prior work, it was demonstrated that the material roughness
also influences the contact angle measurements. To avoid discrepan-
F489
Figure 3. Confocal Laser Scanning Microscope images of different GDL and
channel wall materials. (a) MRC-105 (b) SGL-25BC (c) TGP-H-060 (d) Cu
110 (e) Graphite Composite (f) SS 2205MRC-105.
cies due to the local surface roughness variation in our experiments,
the channel walls were polished using an emery cloth. The surface
roughness of the sidewall was measured using the Confocal Laser
Scanning Microscope (CLSM), and the corresponding images of different channel walls and the GDL samples are shown in Figure 3.
The surface roughness for different sidewall materials was confirmed
to be less than 900 nm from the image slices and is summarized in
Table III.
Effect of channel wall material.— For these set of experiments, the
base GDL material MRC-105 was used while the sidewall material
was changed along with the channel open angle.
Stainless steel (SS-2205) sidewall.— According to the theoretical
prediction using the Concus-Finn condition, the transition angle for
SS-2205 and MRC-105 was found to be 49◦ . Experimentally it was
found that for 45◦ channel, the corner filling did not occur for any air
velocity agreeing with the theoretical predictions. A recorded image
sequence of the droplet growth and its interaction with the SS-2205
sidewall for an open angle of 45◦ and air velocity of 1.6 m/s is shown in
Figure 4. The IDCA for the SS-2205 sidewall and the MRC-105 base
measured from each frame of the video sequence was plotted against
the Concus-Finn Limit (CFL) as shown in Figure 5a. According to the
Concus-Finn condition, any point falling below the CFL line would
lead to corner filling condition, and any point above would lead to no
corner filling condition. The observed behavior can also be compared
to the plot of the IDCA points that lie above the CFL line. Hence, the
findings from our experiments regarding droplet filling agreed with
the corresponding theoretical predictions.
However for a 50◦ channel, contradictory results were obtained
between the experimental observation and the theoretical predictions.
For 50◦ channel, corner filling was observed only for the lower air
velocities (0.2–0.6 m/s), and not for the higher air velocities (0.7–2.4
m/s). Image sequences of the droplet interacting with the 50◦ channel
walls for an air velocity of 1.6 m/s are shown in Figure 6. It was also
observed that the IDCA for lower air velocities fell below the CFL line
satisfying the corner filling criterion (Figure 5b). In contrast, the IDCA
for higher air velocities fell above the CFL line leading to non-corner
filling behavior. From the experiments it was observed that the effect
of air velocity was pronounced, and caused noticeable oscillations
Table III. Surface Roughness Measured for the Different Sidewall
Materials Used.
Figure 2. Concus-Finn Condition a) Different contact angles made by the
droplet with the base and the sidewall b) Plot representing the Concus-Finn
condition when the droplet contacts the base and the sidewall.
Material
Surface Roughness (Ra)
Copper (Cu-110)
Graphite Composite
Stainless Steel (SS-2205)
Polycarbonate (Lexan)
554 nm
873 nm
558 nm
666 nm
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
F490
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
Figure 4. Image sequence of the droplet interaction with the Stainless Steel (SS-2205) sidewall and MRC-105 base in the 45◦ open angle channel with 1.6 m/s
air velocity inside the channel. The red circle in the image shows that the droplet does not fill the corner of the channel.
in the droplet at higher velocities. These oscillations affected the
transition between the channel filling and non-filling behaviors at
the transition angles (disagreeing with the traditional Concus-Finn
condition). Due to the oscillations in a droplet, the instantaneous
contact angle the droplet makes with the sidewall may become larger
than the equilibrium contact angle just prior to its contact with the
sidewall. If the conditions are close to the Concus-Finn criterion, the
higher contact angle experienced due to oscillation may cause the
droplet to not fill the corner. Therefore, the use of dynamic contact
angles is emphasized to predict the droplet behavior and thus the
overall water coverage ratio on the GDL more accurately.
Copper (Cu-110) sidewall.— A set of similar experiments were
conducted with the Cu-110 sidewall. According to the theoretical
prediction, the transition angle for Cu-110 and MRC-105 was 48◦ .
The experimental results observed for Cu-110 for both the 45◦ and
50◦ channels were similar to those observed for SS-2205 sidewalls.
Figure 5. Plot of instantaneous dynamic contact angle (IDCA) made by the
droplet with the Stainless Steel (SS-2205) sidewall and the MRC-105 GDL
base for (a) 45◦ open angle channel and (b) 50◦ open angle channel at different
air velocities.
Graphite composite sidewall.— The transition angle for the
graphite sidewall according to the theoretical prediction was 52◦ .
From the experimental results, it was found that for 45◦ and 50◦ open
angle channels, the droplet did not fill the channel corners for any
air velocities (0.2–2.4 m/s) and the results agreed with the IDCA
plot (Figure 7) which showed no corner filling behavior for any air
velocities.
The three channel wall materials commonly considered for use
in the automobile applications (stainless steel, copper, graphite) have
been found to have varying effects on the droplet behavior depending
upon the CAH value. The graphite composite, which has the largest
CAH, behaved differently compared to the materials that have lower
CAH values near the transition angle. It can be concluded that the CAH
affects the droplet behavior in the gas channels. Materials that have
larger CAH can accommodate a larger range of IDCA that the droplet
makes with the wall and the GDL, facilitating later corner filling than
that given by the Concus-Finn condition. Due to the oscillations in
a droplet, the instantaneous contact angle the droplet makes with the
sidewall may become larger than the equilibrium contact angle just
prior to its contact with the sidewall. If the conditions are close to
the Concus-Finn criterion, the higher contact angle experienced due
to oscillation may cause the droplet to not fill the corner. On the other
hand, materials having smaller CAH can only accommodate smaller
oscillations in the droplets produced by the air flow without leading
to corner filling of the channel.
Figure 6. Image sequence of the droplet interaction with the Stainless Steel (SS-2205) sidewall and MRC-105 base in the 50◦ open angle channel with 1.6 m/s
air velocity inside the channel. The red circle in the image shows that the droplet filled the corner of the channel.
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
F491
Figure 7. Plot of instantaneous dynamic contact angle made by the droplet
with the graphite sidewall and the MRC-105 GDL base for (a) 45◦ open angle
channel and (b) 50◦ open angle channel at different air velocities.
Figure 8. Plot of instantaneous dynamic contact angle made by the droplet
with the polycarbonate sidewall and the TGP-H-060 base for (a) 45◦ open
angle channel and (b) 50◦ open angle channel at different air velocities.
When using MRC-105 GDL with any of the tested channel materials, the droplet filling of the channel corners did not take place
for open angles below 45◦ . The 50◦ open angle channel behaved as a
transition between filling and non-filling conditions. The Concus-Finn
condition (using static contact angles) could not accurately predict the
corner filling behavior of the droplet near the transition angle. Hence,
using IDCA information would be helpful in predicting the droplet
behavior in the gas channel of PEMFC instead of static contact angle
information.
with the theoretical predictions for all air velocities. The IDCA plots
for the 45◦ and the 50◦ open angle channels at 0.2 and 1.6 m/s air
velocities are shown in Figure 8.
Effect of GDL material.— There are different types of GDL materials that are used in the PEMFCs. The most commonly used are
Toray, SGL and the MRC-105. To evaluate the effect of a GDL material on the droplet growth and removal, experiments were performed
under similar conditions as mentioned earlier. In these experiments
the channel wall material was kept constant (polycarbonate) while
the GDL materials were changed along with the channel open angle.
The GDL materials used for these experiments were TGP-H-060 and
SGL-25BC with 6% PTFE.
SGL-25BC and polycarbonate channel wall.— The transition angle for SGL-25BC and a polycarbonate sidewall was found to be
51◦ . Similar experimental results were observed for SGL-25BC in
which the droplets did not fill the channel corners for the 45◦ and
50◦ channels for any air velocities. The image sequence of the droplet
growth in the 50◦ open angle channel at 1.6 m/s air velocity is shown
in Figure 9a. The plot comparing the IDCA for both SGL-25BC—
polycarbonate and MRC-105-polycarbonate material pairs, observed
in our previous study,28 at air velocities 0.2 and 1.6 m/s is shown in
Figure 9b. It is observed from the plot that the IDCA point for the
MRC-105-polycarbonate material pair covers larger range of contact
angles compared to the SGL-25BC-polycarbonate material pair. This
may also be one of the reasons why at lower air velocities, the MRC105 showed corner filling behavior. However the SGL-25BC did not
show corner filling behavior in the 50◦ open angle channel.
TGP-H-060 and polycarbonate channel wall.— For TGP-H-060
with 6% PTFE and a polycarbonate sidewall, the transition angle
was found to be 50◦ according to the theoretical prediction using the
Concus-Finn condition. The experimental results showed no corner
filling for both 45◦ and 50◦ open angle channels. These results agreed
Effect of used GDL material.— To understand the effect of GDL
deterioration on the droplet dynamics inside the gas channel of PEMFCs, MRC-105 GDLs, which were previously run for 40 and 125 hours
in an experiment performed by Lu et al.,22 were employed. Experiments were performed using these GDL samples with polycarbonate
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
F492
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
Figure 9. (a) Image sequence of the droplet
interaction with the polycarbonate sidewall
and SGL-25BC base in the 50◦ open angle
channel with 1.6 m/s air velocity. (b) Plot of
instantaneous dynamic contact angle made by
the droplet with the polycarbonate sidewall
and the base MRC-105 and SGL-25BC GDL
for 50◦ open angle channel at different air velocities.
channel walls under similar test conditions. It was observed from these
experiments that after 40 hours run time, the used GDL performed
similarly to a fresh MRC-105 GDL. For 45◦ open angle channel, it did
not show corner filling behavior for any air velocities introduced in the
system. For 50◦ open angle channel, it showed corner filling behavior
at lower air velocities (0.2–0.6 m/s), and no corner filling at higher air
velocities (0.7–2.4 m/s). However, the GDL with 125 hours run time
showed a different behavior. It was observed that the droplet filled the
channel corners for both 45◦ and 50◦ open angle channels for all the
air velocities. To further understand this behavior, the advancing and
receding contact angles on the used GDL samples were measured and
are shown in Table I. It was observed that 125 hours run time GDL had
much lower advancing and receding contact angles compared to those
for the fresh samples. However, samples with 40 hours run time did
not show much difference from the fresh samples in terms of advancing and receding contact angles. This shows that the wettability of the
GDL samples reduced with longer run time. The transition contact
angle between corner filling and non-filling for both 40 and 125 hours
run time samples with polycarbonate were measured. It was seen
that the transition angles were 49◦ and 37◦ respectively as shown in
Table II. This explains the discrepancy observed in the droplet dynamics for the 125 hours run time sample. To further validate these
results, the IDCA for 50◦ open angle channel were plotted against the
CFL for both 40 and 125 hour run time samples with polycarbonate sidewall, and are shown in Figure 10. Figure 10a shows that for
the 40 hours run time sample, the droplet fills the channel corners
at lower air velocities (0.2–0.6 m/s), and does not fill at higher
air velocities (0.7–2.4 m/s). For the 125 hours run time sample,
Figure 10b shows that for all air velocities, the IDCA point fell below
the CFL line. This indicates that the droplet should fill the channel
corners for all air velocities. This work indicates that the GDL material deteriorates or loses its wettability after longer run time which
perhaps affects the corner filling behavior of the droplet and lead to
larger accumulation of water inside the gas channel. Further study
with a larger number of samples and different run times is warranted
to explore this effect further.
Pressure drop across the droplet.— Pressure drop across the
droplet is one of the important parameters in understanding the dynamics of water removal. In PEMFCs, lower pressure drop in the system
corresponds to a lesser amount of water hold up in the channel and
thus, lower gas transport resistance. Therefore, lower pressure drop in
the system would result in higher gas transport and thus increasing the
efficiency of the PEMFC system. To measure the pressure drop across
the droplet, a pressure sensor was connected to the channel using a
pressure tap just before the droplet inlet location in the upstream of
the channel. The peak pressure drop across the droplet in a channel
for different channel wall materials and air velocities are plotted in
Figure 11. The uncertainty in the peak pressure drop value was about
0.25%. For the 45◦ channel, Figure 11a showed that for different channel wall materials, the peak pressure drop across the droplet followed
a similar trend. The pressure drop in the system would increase with
an increase in the air velocity. This is mainly due to the blockage
created by the droplet in the channel. Once the pressure inside the
system reaches its maximum value, the droplet would be removed
from the system due to the force exerted by the air flow on the droplet.
However, at very high air velocities (1.8–2.4 m/s) the peak pressure
drop value starts to decrease.
At higher air velocity, the droplet starts to transition toward film
flow regime instead of slug flow, as observed for lower air velocities
in the system. Therefore, the force required to remove the droplet is
lower for higher air velocities as compared to that required at lower air
velocities. The different flow patterns observed in the 50◦ open angle
channel is shown in Figure 12. It is seen from the figure that, at lower
air velocities, the droplet grows and completely fills the channel cross
section to form the slug flow. Once the pressure inside the channel
overcomes the surface tension force with which the liquid is held in the
channel, the liquid is removed from the channel. However, at higher
air velocities, the force exerted by the air flow in the channel is much
higher than the surface tension force exerted on the liquid. Hence,
before the liquid completely blocks the channel and the pressure drop
increases in the channel, the liquid is removed from the channel as a
film flow.
It was also observed that different channel materials showed significant differences in the peak pressure drop value in the 50◦ open
angle channel. From Figure 11b, it was observed that Cu-110 and SS2205 produced a much higher pressure drop compared to the graphite
composite channel wall at lower air velocities (0.2–1 m/s). This was
mainly due to the channel corner filling observed for both Cu-110
and SS-2205 at lower air velocities. On the other hand, graphite
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
F493
Figure 10. Plot of instantaneous dynamic contact angle made by the droplet
in a 50◦ open angle channel at different air velocities with the polycarbonate
sidewall; (a) 40 hours run MRC-105 GDL; and (b) 125 hours run MRC-105
GDL.
composite channel wall showed no corner filling at lower air velocities,
resulting in a much lower pressure drop in the system. At medium air
velocities (1.1–1.7 m/s) the channel wall materials showed a similar
peak pressure drop in the system, which increased with an increase in
the air velocity. This corresponded to the non-corner filling condition
observed in the experiments. At higher air velocities (1.8–2.4 m/s), the
peak pressure drop in the system starts to gradually decrease, since
the droplet transitions to the film flow and produces a much lower
blockage in the channel cross section compared to that at lower and
medium velocities.
The peak pressure drop measured for different GDL materials
also followed a similar pattern, as observed for different channel wall
materials and is plotted in Figure 13a. From the experiments it was
observed that for both SGL-25BC and TGP-H-060, the droplet did
not fill the channel corners in 45◦ , as well as 50◦ open angle channels
for any air velocities. The peak pressure drop values measured for a
given channel angle with different GDL materials also showed similar
results. For both SGL-25BC and TGP-H-060, the peak pressure drop
in a channel increased as the air velocity increased, for air velocities
lower than 1.7 m/s. However, for higher air velocities (1.8–2.4 m/s)
the peak pressure drop started to decrease. It was also interesting to
note that the pressure drop values observed for different GDL types
with polycarbonate sidewall were similar to those observed for MRC105 GDL and polycarbonate channels in our previous study.28 Hence,
Figure 11. Plot of peak differential pressure drop across the droplet as a
function of air velocity, for different channel wall material, (a) 45◦ open angle
channel, and (b) 50◦ open angle channel. The uncertainty in the pressure drop
was around 0.25%.
this shows that different types of GDLs do not have any significant
differences in the droplet-wall interactions.
Peak pressure drop for used GDL materials differed significantly
compared to those of the fresh GDL samples. Figure 13b shows the
peak pressure drop measured for a 50◦ open angle polycarbonate
channel with used GDL samples at different air velocities. From
the plot it is observed that the peak pressure drops for the 40 hour
run time sample at lower air velocities were higher compared to those
at higher air velocities. This showed that the droplet fills the channel
corners, forms a slug flow at lower air velocities, and results in a
higher pressure drop. At higher air velocities, the droplet transition
to film flow and does not fill the channel corners, leading to a lower
pressure drop in the system. For the 125 hours run time sample, the
droplets resulted in a slug flow for all air velocities. The peak pressure
drop values for all air velocities were much higher compared to those
for the 40 hour run time samples. This showed that after a longer run,
the hydrophobicity of the GDL was reduced and lead to a slug flow,
causing increased liquid stagnation in the channels and reduced area
available on the GDL surface for gas transport.
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
F494
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
Figure 12. Different flow patterns observed in the 50◦ open angle channel (a)
Slug flow at lower air velocities (b) Film flow at higher air velocities.
Conclusions
In this work, ex-situ experiments were conducted to study the water droplet dynamics in a PEMFC gas channel with different channel
sidewall angles, channel wall materials, and GDL materials under
controlled air flow velocities between 0.2–2.4 m/s. High speed video
images showing the droplet interaction with the channel walls were
recorded. The pressure drop across the droplet was also recorded simultaneously. It was revealed through the recorded high-speed videos
that the droplet dynamics at the channel corners are significantly affected by the channel open angles, air flow velocities and the channel
materials.
r
Sidewall material: Characteristic behaviors of different channel
materials differ considerably based on their material property. The
CAH played a major role in determining the corner filling behavior
for each type of channel. Materials such as graphite, which have larger
CAH, showed no corner filling behavior for any air velocities at the
transition angle. These materials can accommodate larger oscillations
before the droplet starts to move on the wall surface to fill the channel
corner. However, for materials having smaller CAH (copper, stainless
steel and polycarbonate), corner filling occurred at lower air velocities.
r Air flow rate: Air flow in the channel produced oscillations in
the water droplet which determined the instantaneous contact angles
the droplet made with the sidewall and the base GDL. In this case,
the IDCA of the droplet determines the corner filling behavior. Also,
at lower air velocities the droplet completely blocks the channel and
forms a slug flow, whereas at higher air velocities the droplet transforms into a film flow.
r GDL materials: For channel open angles near the transition
point, SGL-25BC and TGP-H-060 showed no corner filling. However,
Figure 13. Plot of peak differential pressure drop values across the droplet (a)
For different GDL materials in 45◦ and 50◦ open angle channels as a function
of air flow velocity (b) For used GDL materials in 50◦ open angle channels as
a function of air flow velocity. The uncertainty in the pressure drop was around
0.25%.
MRC-105 showed corner filling at lower air velocities. The CAH and
oscillations in the water droplet determined the corner filling behavior
for a given GDL and wall material.
r Used GDL materials: Hydrophobicity of the GDL material reduces with a longer run time, and leads to larger accumulation of
liquid water in the channel compared to that with fresh GDL samples.
This would reduce the available area for gas transport on the GDL
surface and increased gas transport resistance with longer run times.
Further work in this area is warranted with a larger number of GDL
samples.
r Concus-Finn condition: The Concus-Finn condition using static
contact angle predicted the corner filling behavior inaccurately in a
dynamic condition. Instead, using IDCA in the Concus-Finn condition detected the corner filling more accurately, especially near the
transition angle.
Acknowledgments
This work was conducted in the Thermal Analysis, Microfluidics,
and Fuel Cell Laboratory in the Mechanical Engineering Department
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use
Journal of The Electrochemical Society, 160 (6) F487-F495 (2013)
at the Rochester Institute of Technology and was supported by the US
Department of Energy under contract No. DE-EE0000470.
References
1. M. Rikukawa and K. Sanui, Progress in Polymer Science, 25, 1463 (2000).
2. J. T. Gostick, M. W. Fowler, M. A. Ioannidis, M. D. Pritzker, Y. M. Volfkovich, and
A. Sakars, Journal of Power Sources, 156, 375 (2006).
3. J. J. Kowal, A. Turhan, K. Heller, J. Brenizer, and M. M. Mench, Journal of Electrochemical Society, 153, A1971 (2006).
4. P. K. Das, A. Grippin, A. Kwong, and A. Z. Weber, Journal of The Electrochemical
Society, 159, B489 (2012).
5. J. P. Owejan, T. A. Trabold, D. L. Jacobson, D. R. Baker, D. S. Hussey, and M. Arif,
International Journal of Heat and Mass Transfer, 49, 4721 (2006).
6. J. P. Owejan, J. J. Gagliardo, J. M. Sergi, S. G. Kandlikar, and T. A. Trabold, International Journal of Hydrogen Energy, 34, 3436 (2009).
7. R. Satija, D. L. Jacobson, M. Arif, and S. A. Werner, Journal of Power Sources, 129,
238 (2004).
8. R. J. Bellows, M. Y. Lin, M. Arif, A. K. Thompson, and D. L. Jacobson, Journal of
The Electrochemical Society, 146, 1099 (1999).
9. P. A. Chuang, A. Turhan, A. K. Heller, J. S. Brenizer, T. A. Trabold, and
M. M. Mench, ASME Conference Proceedings, 2005, 31 (2005).
10. I. Manke, C. Hartnig, M. Grunerbel, W. Lehnert, N. Kardjilov, A. Haibel, A. Hilger,
J. Banhart, and H. Riesemeier, Applied Physics Letters, 90, 174105 (2007).
11. A. Bazylak, D. Sinton, and N. Djilali, Journal of Power Sources, 176, 240
(2008).
12. K. W. Feindel, S. H. Bergens, and R. E. Wasylishen, Physical Chemistry Chemical
Physics, 9, 1850 (2007).
F495
13. S. Tsushima, K. Teranishi, and S. Hirai, Electrochemical and Solid-State Letters, 7,
A269 (2004).
14. S. Tsushima, T. Nanjo, K. Nishida, and S. Hirai, ECS Transactions, 1, 199 (2006).
15. Z. Dunbar and R. I. Masel, Journal of Power Sources, 171, 678 (2007).
16. X. G. Yang, F. Y. Zhang, A. L. Lubawy, and C. Y. Wang, Electrochemical and
Solid-State Letters, 7, A408 (2004).
17. K. Tuber, D. Pocza, and C. Hebling, Journal of Power Sources, 124, 403 (2003).
18. S. W. Cha, R. O’Hayre, and F. B. Prinz, Solid State Ionics, 175, (2004).
19. J. P. Owejan, T. A. Trabold, D. L. Jacobson, M. Arif, and S. G. Kandlikar, International Journal of Hydrogen Energy, 32, 4489 (2007).
20. D. Spernjak, S. Advani, and A. K. Prasad, Experimental Investigation of Liquid Water
Formation and Transport in a Transparent Single-Serpentine PEM Fuel Cell, in ASME
Conference Proceedings, p. 453, ASME (2006).
21. F. Y. Zhang, X. G. Yang, and C. Y. Wang, Journal of The Electrochemical Society,
153, A225 (2006).
22. Z. Lu, C. Rath, G. Zhang, and S. G. Kandlikar, International Journal of Hydrogen
Energy, 36, 9864 (2011).
23. X. Zhu, P. C. Sui, and N. Djilali, Journal of Power Sources, 181, 101 (2008).
24. X. Zhu, Q. Liao, P. C. Sui, and N. Djilali, Journal of Power Sources, 195, 801 (2010).
25. C. D. Rath and S. G. Kandlikar, Colloids and Surfaces A: Physicochemical and
Engineering Aspects, 384, 653 (2011).
26. P. Concus and R. Finn, Microgravity Science and Technology, 7, 152 (1994).
27. P. Concus and R. Finn, SIAM Journal on Mathematical Analysis, 27, 56 (January
1996).
28. P. Gopalan and S. G. Kandlikar, Journal of Electrochemical Society, 159, F468
(2012).
29. P. Gopalan and S. G. Kandlikar, ECS Transactions, 41, 479 (2011).
30. P. Gopalan and S. G. Kandlikar, in ECS 222 Prime Meeting, Honolulu, Hawaii
(2012).
Downloaded on 2013-05-09 to IP 129.21.225.12 address. Redistribution subject to ECS license or copyright; see ecsdl.org/site/terms_use