Sei sulla pagina 1di 7

Electrochimica Acta 56 (2010) 600–606

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Effect of cathode catalyst layer thickness on methanol cross-over in a DMFC


Saif Matar, Hongtan Liu ∗
Clean Energy Research Institute, College of Engineering, University of Miami, Coral Gables, FL 33146, USA

a r t i c l e i n f o a b s t r a c t

Article history: The effects of cathode catalyst layer (CCL) thickness on the detrimental effect of methanol cross-over in a
Received 21 April 2010 direct methanol fuel cell (DMFC) under various operating conditions are studied. Three membrane elec-
Received in revised form 29 June 2010 trode assemblies (MEAs) with different CCL thicknesses but identical catalyst loading and identical anode
Accepted 2 September 2010
catalyst layer are used. The results show that, when a thicker CCL, approximately twice the thickness of
Available online 21 September 2010
the base case, is used, the fuel cell performance increases significantly. The increase in performance with
a thicker CCL is attributed to the oxidation of the methanol crossed-over in part of the catalyst layer and
Keywords:
leaving the rest of the catalyst layer free from methanol contamination, leading to mitigations of the
DMFC
Methanol cross-over
effects of mixed potentials. The results of electrochemical impedance spectroscopy (EIS) show that the
Cathode charge transfer resistance for the fuel cell with twice the thickness of CCL is 30% lower compared to that
Catalyst layer thickness for the base case, indicating that the active catalyst area available for oxygen reduction reaction (ORR) is
Impedance spectroscopy indeed greater. The results of the electrochemical active surface areas (ECA) show that without methanol
ECA contamination, the ECA of the thicker CCL is actually lower, indicating that the better performance and
the lower charge transfer resistance are not caused by a higher original ECA, but a higher active area for
ORR. A much thicker CCL, about 5 times of that for the base case, is also used and the cell performance
is also higher than that for the base case. The experimental results show that there exists an optimum
cathode catalyst layer thickness and the thickness of cathode catalyst layer has a significant effect on
DMFC performance.
© 2010 Elsevier Ltd. All rights reserved.

1. Introduction ner Nafion® membranes. They also reported that the methanol
cross-over rate decreases with increasing current density due to an
Direct methanol fuel cells (DMFCs) are an attractive potential increase in methanol consumption to sustain higher current densi-
power source for mobile devices, thanks to their simple and com- ties. Besides, the methanol cross-over rate was found to increase
pact designs, quick refueling and low operating temperature. When with the cell temperature. Dohle et al. [2] also found a similar
compared to a H2 PEM fuel cell, a DMFC does not require humid- behavior and noticed that the electro-osmotic drag (EOS) coef-
ification because the liquid methanol in the anode and the water ficient for the methanol/water mixture also increases with tem-
produced in the cathode are sufficient to keep the Nafion® mem- perature. Ren et al. [3] showed the importance of electro-osmotic
brane well hydrated. Despite all these advantages, there are several drag in methanol transport through the Nafion® membrane and
hurdles that prevent DMFCs from being competitive to existing reported that EOS effects are significant even at low methanol con-
power sources, such as lithium-ion batteries. One of these hurdles is centrations. Ramya et al. [4] measured the methanol flux across the
the methanol cross-over from the anode to the cathode, which has Nafion® membrane using cyclic voltammetry and chronoamper-
a serious detrimental effect on cell performances due to the adverse ometry techniques and concluded that the methanol permeability
effect caused by the formation of mixed potentials at the cath- is dependent on methanol concentration and reported that the opti-
ode. Methanol cross-over also reduces the fuel efficiency of the cell mum methanol concentration is between 1 and 2 M. Han and Liu [5]
as the oxidation of the crossed-over methanol does not generate conducted real time measurements of methanol cross-over using
any electrical power. Many efforts have been focused on methanol a CO2 sensor placed at the exit of the cathode side. They observed
cross-over and its effects. Narayanan et al. [1] reported that over peaks in methanol cross-over rates when the cell voltage suddenly
40% of methanol could be wasted due to cross-over through thin- changes and found that the greater the change in cell voltage is,
the greater is the peak in methanol cross-over. This phenomenon
was attributed to methanol permeation hysteresis. Liu and Wang
∗ Corresponding author at: Department of Mechanical and Aerospace Engineer-
[6] developed a three-dimensional 2-phase model to investigate
ing, P.O. Box 248294, Coral Gables, FL 33124, USA. Tel.: +1 305 284 2019;
methanol concentration, current and cross-over current distribu-
fax: +1 305 284 2580. tions. They concluded that optimizing the flow field geometry can
E-mail address: hliu@miami.edu (H. Liu). lead to uniform current density distribution and better methanol

0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2010.09.001
S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606 601

utilization at the anode and therefore reducing methanol cross- effects of mixed potential are significant; while in the second sub-
over. Wang and Wang [7] employed a 2-phase, multi-component layer, its methanol concentration can be very low, if not zero, and
model to investigate electrochemical kinetics and transport phe- the effect of mixed potential is significantly reduced, thus ORR is
nomena including methanol cross-over in DMFCs. The results dominant. The overall effect could be a significantly higher cathode
showed that the main driving force for methanol cross-over is dif- potential, leading to a much better performance of the fuel cell.
fusion at low and high current densities. However, at higher current
densities there is a significant contribution from electro-osmosis. 2. Experimental system and methodology
Ge and Liu [8] compared the cell performance with and without
methanol cross-over using their three-dimensional MDFC model. 2.1. Fuel cell testing system
The modeling results showed that the effect of methanol cross-
over is significant even at low methanol concentrations. Moreover, The fuel cell testing system used in this study is a fuel cell test
they concluded that methanol cross-over is greatly affected by the station manufactured by Fuel Cell Technologies, Inc. The test station
porosities of both the anode catalyst and diffusion layers. Electro- controls the flow rate, temperature, pressure and humidity of the
chemical impedance spectroscopy (EIS) has proved to be valuable in air flowing in the cathode as well as the cell operating temperature.
diagnosing the effects of methanol cross-over. Mueller and Urban The test station comes equipped with an HP6050 DC electronic load
[9] proposed a method to separately extract anode and cathode that controls the load on the fuel cell and measures the potential
impedance spectra. Piela et al. [10] conducted a comprehensive and current of the cell. Methanol is pumped to the anode side of the
study of EIS diagnostics under various cell parameters and provided fuel cell using a Gilson Minipuls 3 pump. A LabView based software
interpretations for the impedance data. Since methanol cross-over was used in the data acquisition system to control and measure
is recognized as a very serious problem in DMFC, many contri- various parameters of the fuel cell. Moreover, the fuel cell test-
butions have been made to solve or mitigate the problem. One ing system is coupled with a potentiostat (VersaStat 3, Princeton
focus area was developing alternative membranes that have lower Applied Research, Inc) capable of data averaging during acquisition
methanol permeabilities. Li et al. [11] prepared Sulfonated Poly to conduct AC impedance and cyclic voltammetry measurements.
Ether Ether Ketone (SPEEK) membranes that have methanol per- Fig. 1 shows the schematic of the experimental system.
meability of one order of magnitude lower than that of Nafion®
115 and similar proton conductivity to Nafion® 115 at temperatures 2.2. Fuel cell fixture and membrane electrode assembly (MEA)
above 80 ◦ C. Carretta et al. [12] casted Sulfonated Poly-Styrene (SPS)
membranes from sulfonic acid groups in the base polymer. When The fuel cell fixture used was a 5 cm2 cell consists of two stain-
compared to the Nafion® membranes, these membranes exhib- less steel end plates and two Poco graphite plates machined with
ited comparable proton conductivity and 70% lower permeability serpentine flow fields. Two gold plated copper plates were used as
at the highest degree of sulfonation. Carter et al. [13] proposed current collectors. Two Teflon® gaskets were employed to ensure
Polyphosphazene-based membranes for DMFC. These membranes uniform MEA/GDL compression and to prevent damage to the MEA.
showed very low methanol cross-over rate: about 3 times lower Three custom-made MEAs were purchased from BCS Fuel Cells,
than that of Nafion® 117. Another path researchers took was to Inc. The specifications of the MEAs are shown in Table 1. The thick-
improve the electro-oxidation process in the catalyst layer as well ness of 8.4 ␮m was the standard/base thickness for anode and
as improve the structure of the catalyst layer and gas diffusion layer cathode catalyst layers. The anode catalyst loading and thickness
(GDL). Abdel Rahim and Hassan [14] studied the electro-oxidation and the membrane used were identical for all three MEAs; only the
of methanol on titanium and platinum modified titanium elec- cathode catalyst layer thickness was varied under the same catalyst
trodes (Pt/Ti). These electrodes manifested less poising effects and loading by varying the wt.% of Pt/C. The platinum catalysts used for
faster methanol electro-oxidation rate when compared to pure Pt. all three catalyst layers are identical and thus have identical particle
Carmo et al. [15] developed an alternative method for preparing sizes and distributions. In addition to the base thickness of 8.4 ␮m,
Pt/C and PtRu/C catalysts. This method involves impregnation of the thicknesses of the two additional cathode catalyst layers used
the support material with the precursors in ethanol and reduc- are 19 ␮m and 44 ␮m, i.e. about double and 5 times the thickness
tion under a hydrogen atmosphere to create active materials for of the base one.
methanol oxidation reaction. Liu et al. [16] compared the difference The break-in procedure for all three MEAs was identical. First,
in performance between coating the membrane or the GDL with to activate the MEA, humidified nitrogen gas at 80 ◦ C and 750 sccm
catalyst. They found that coating the GDL with catalyst instead of was fed to both sides of the fuel cell for 6 h. Second, to activate the
the membrane increases the cell performance and can significantly catalyst layer, humidified hydrogen at 70 ◦ C and 300 sccm was fed
reduce methanol cross-over owning to the increase in thickness. to the anode and oxygen at 70 ◦ C and 700 sccm was fed to the cath-
It is important to note that research efforts have been predom- ode. Then the fuel cell was run under constant voltage of 0.6 V and
inantly focused on anode catalyst layer and the membrane, but the current response was monitored. After a steady state current
little effort has been directed at the cathode side. From experi- was reached, the break-in process was considered to be completed.
mental results, Ge and Liu [17] found that when there is enough In this paper, each polarization curve was obtained by averaging
oxygen supply to the cathode side, the adverse effect of methanol three polarization curves obtained at identical operating condi-
cross-over is reduced. It is believed that when the methanol per- tions. When obtaining the individual polarization curves, there was
meated to the cathode can be oxidized faster, more reaction sites a 240 s delay between each operating point.
will be free for oxygen reduction reaction (ORR). Based on these
results, it is hypothesized here that the performance of a DMFC 2.3. Electrochemical measurements
may be improved if the thickness of the cathode catalyst layer is
increased to a certain value so that the crossed-over methanol can The EIS investigation was carried out according to the method
be oxidized in part of the catalyst layer close to the membrane and described by Muller and Urban [9]. First, the full DMFC impedance
leave the rest part of the catalyst layer close to the GDL relatively spectrum was measured with methanol in the anode and air in
free of methanol. In another word, for a thicker cathode catalyst the cathode. Second, the anode impedance spectrum was mea-
layer, it can be divided into two sub-layers, one in contact with the sured with hydrogen flowing in the cathode side and hence acting
membrane and the other in contact with the GDL. In the first sub- as a dynamic hydrogen electrode. Hydrogen humidification tem-
layer, significant methanol oxidation reaction (MOR) occurs and the perature was set to 90 ◦ C and the backpressure to 152 kPa to
602 S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606

Fig. 1. Schematic of the experimental system.

Table 1
Characteristics of the MEAs used in this study.

Anode loading Cathode loading Membrane Anode catalyst layer Cathode catalyst layer Pt/C wt.%
(mgPtRu cm−2 ) (mgPt cm−2 ) thickness (␮m) thickness (␮m)

MEA 1 4.0 1.0 Nafion 117 8.4 8.4 66


MEA 2 4.0 1.0 Nafion 117 8.4 19 50
MEA 3 4.0 1.0 Nafion 117 8.4 44 40

ensure the reversibility of the cathode electrode. Finally, the cath- 3.1. Comparison of cell performances
ode impedance spectrum was obtained by subtracting the anode
impedance spectrum from that for the full DMFC. The impedance Polarization and power density curves of the three MEAs at
spectra were acquired at frequencies between 10 kHz and 0.1 Hz methanol concentration of 1 M are shown in Fig. 2. These results
with 10 points decade−1 under galvanostatic mode. The amplitude show that the cell performance increased significantly when the
of the input AC current was selected to be 5% of the fuel cell oper- cathode catalyst layer thickness was increased from the base thick-
ating DC current. ness of 8.4 to 19 ␮m and the maximum power density increased
Electrochemical active surface areas (ECA) of the cathode elec- from 44.9 mW cm−2 to 78.4 mW cm−2 . Such a drastic increase in
trode of the three MEAs were measured using cyclic voltammetry performance for 19 ␮m over 8.4 ␮m seems to indicate that the
at sweep rate of 5 mV s−1 with H2 flowing in the anode side as the hypothesis, that a thick cathode catalyst layer can leave part of
reference/counter electrode and N2 flowing in the cathode side as
the working electrode. Both gasses were fully humidified and the
cell temperature was set at 35 ◦ C. ECA (cm2 Pt mg−1 Pt ) values were 0.80 80
calculated using the following equation: 8.4 μm
0.70 19 μm 70
44 μm
Power Density / mW cm-2

QH 0.60 60
ECA = × 0.21 (1)
[Pt]
0.50 50
Voltage / V

where QH is the adsorption charge in mC cm−2 ,


[Pt] is the cata- 0.40 40
lyst loading in mgPt cm−2 and the constant 0.21 (mC cm−2 Pt ) is the
charge required to oxidize a monolayer of H2 on a smooth Pt surface. 0.30 30

0.20 20

3. Results and discussion 0.10 10

0.00 0
In all of the experiments, unless stated otherwise, methanol was 0 100 200 300 400 500
fed to the anode compartment at a rate of 6 ml min−1 at 1 M, 2 M and Current Density / mA cm-2
4 M concentrations, respectively. Air flow rate was set at 600 sccm
Fig. 2. Polarization and power density curves at different cathode catalyst layer
with no humidification and the fuel cell temperature was set at
thicknesses for 1 M methanol concentration. Conditions: 70 ◦ C; air flow rate
70 ◦ C. The operating pressure was atmospheric pressure. 600 sccm; methanol flow rate 6 ml min−1 .
S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606 603

-0.06
20 mA cm-2²
100 mA cm -2²
150 mA cm -2²
-0.04 Model
ZIM / Ω

-0.02

Fig. 4. Equivalent circuit model for DMFC cathode [18].

0
Table 2
Fitting values for 19 ␮m CCL thickness at different current densities and 1 M
methanol concentration.
0.02 Current density C1 (F) R1 () CPE Rct () Ro () Lo (H)
0 0.02 0.04 0.06 0.08 0.1
ZRE / Ω C (F) ˛

20 mA cm−2 0.057 0.0023 0.294 0.78 0.108 0.230 0.028


Fig. 3. Cathode impedance spectra for 19 ␮m CCL thickness at different current
100 mA cm−2 0.057 0.0022 0.294 0.79 0.081 0.204 0.016
densities. Conditions: 70 ◦ C; air flow rate 600 sccm; methanol concentration 1 M;
150 mA cm−2 0.057 0.0021 0.294 0.79 0.058 0.172 0.008
methanol flow rate 6 ml min−1 .

it relatively free from methanol contamination was reasonable.


Since the total catalyst loadings are identical on both the cath- fer resistance in the cathode backing. The inductive loop at the
ode and anode and everything else are also identical except for lowest frequency is a result of slow response of CO coverage to
the cathode catalyst layer thickness, the drastic improvement can the changes in electrode potential.
only be attributed to the increased cathode catalyst layer thick- The impedance spectra were fitted with the equivalent circuit
ness as no other mechanisms can explain such a large increase in model proposed by Du et al. [18] shown in Fig. 4. A parallel com-
performance. It is believed that for the base case, most, if not the bination of a capacitor (C1 ) and a resistor (R1 ) was employed to
entire cathode catalyst layer was contaminated with the crossed- fit the high frequency straight line, the faradic branch is fitted by
over methanol and consequently the effects of mixed potentials a series combination of a resistor (Ro ) and an inductor (Lo ) con-
were severe; while with a thicker cathode catalyst layer, methanol nected in parallel to a parallel combination of a resistor (Rct ) and a
could be oxidized within the part of the catalyst layer adjacent to constant phase element (CPE). The constant phase element is used
the membrane and leave the rest of the catalyst layer free from to account for the non-homogeneity in the surface of the electrode
methanol contamination. Therefore, the mixed potential effect is and its impedance is defined by:
significantly reduced. 1
ZCPE = ˛ (2)
Experiments were also conducted with a much thicker cath- C(jω)
ode catalyst layer, i.e. the CCL was further increased from 19 ␮m to
44 ␮m. Even with such an extremely thicker cathode catalyst layer, where C is the magnitude of the capacitance, ω is the angular fre-
the fuel cell performance is still better than that for the base case, quency and ˛ is a unit-less parameter that has a value between
as shown in Fig. 2. The maximum power density is 63.5 mW cm−2 , −1 and 1. The cathode impedance spectrum does not show the
compared with 44.9 mW cm−2 in the base case. As can be seen from membrane resistance because it is canceled out in the subtraction
Fig. 2, with a 44 ␮m CCL, the cell performance lies between the process of anode impedance from the full impedance.
base case and the one with 19 ␮m CCL. Obviously, with a 44 ␮m The fitting values for the impedance spectra for the MEA with a
CCL, even though the crossed-over methanol can be oxidized in CCL thickness of 19 ␮m at different current densities are shown in
part of the CCL and leave the rest relatively free from methanol Table 2. It can be seen that the values of R1 , C1 , C and ˛ are indepen-
contamination, it is so thick that the mass transfer loss for oxygen dent of current density and do not contribute to the electrochemical
transfer and/or the ohmic loss due to the additional ionic and elec- process, as indicated in [18] and reported in [19].
tronic paths in the thicker CCL are significant enough to cause a
decrease in cell performance. Comparing with the other two cases, -0.06
8.4 μm
19 ␮m CCL is thick enough to significantly mitigate the effects of 19 μm
mixed potentials, but it was not too thick to significantly increase 44 μm
other losses due to a thicker CCL. Therefore, there exists an opti- Model
mum cathode catalyst layer thickness, with which the maximum -0.04
fuel cell performance can be attained.
ZIM / Ω

3.2. Electrochemical analyses -0.02

The EIS is a robust diagnostic tool that provides details to pos-


sible losses in a fuel cell. In this work, all reported impedance data
were collected under kinetic conditions to minimize the effect of 0
mass transfer limitations. The cathode impedance spectra for the
one with CCL thickness of 19 ␮m at different current densities are
shown in Fig. 3. The characteristics of a typical cathode Nyquist 0 0.02 0.04 0.06 0.08 0.1 0.12
impedance plot are as follows [10]: the high frequency straight line ZRE / Ω
is due to electrode porosity. The intermediate frequency distorted
Fig. 5. Cathode impedance spectra at different cathode catalyst layers thicknesses.
arc is due to electrode charging capacitance and charge transfer Conditions: 70 ◦ C; current density: 100 mA cm−2 ; air flow rate 600 sccm; methanol
resistance (Rct ). The low frequency portion is related to mass trans- concentration 1 M; methanol flow rate 6 ml min−1 .
604 S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606

Table 3 30
8.4 μm
Fitting values for cathode impedance spectra at different CCL thicknesses at 20 19 μm
100 mA cm−2 and 1 M methanol concentration.
10 44 μm
Cathode C1 (F) R1 () CPE Rct () Ro () Lo (H) 0
-10
C (F) ˛
-20

Current / mA
MEA 1 0.015 0.0065 0.294 0.76 0.114 0.352 0.172 -30
MEA 2 0.057 0.0022 0.294 0.79 0.081 0.204 0.016
-40
MEA 3 0.030 0.0030 0.294 0.75 0.104 0.455 0.192
-50
-60
-70
Unlike R1 , C1 , C and ˛, the values of Rct , Ro and Lo decrease as
-80
the current density increases, indicating that methanol is being
-90
consumed in the anode and less permeates thorough the mem- -100
brane. Therefore, it can be concluded that Rct , Ro and Lo define the -110
electrochemical process at the cathode. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
The cathode impedance spectra for the three MEAs at 1 M Voltage/ V
methanol concentration are shown in Fig. 5. The fitting values for
Fig. 7. Cyclic voltammograms for different cathode catalyst layer thicknesses. Con-
8.4 ␮m, 19 ␮m and 44 ␮m CCL thicknesses impedance spectra are ditions: 5 mV s−1 ; 35 ◦ C; H2 and N2 flow rates 200 and 50 sccm, respectively.
shown in Table 3. During the fitting process, it was noticed that
the fitting results were very sensitive to changes in charge trans-
fer resistance, Rct , but not so much to the changes in Ro and Lo, adverse effects of other factors become significant enough to cause
which shows that when the operating parameters are held constant an increase in the charge transfer resistance.
(flow rates, current density, anode thickness, catalyst loadings, etc.) Hypothetical methanol concentration profiles for different
and only cathode catalyst layer thickness is being changed, charge thicknesses of cathode catalyst layer are shown in Fig. 6 to assist
transfer resistance is the main factor that determines the perfor- in the following discussion. For the base case, when the catalyst
mance of the cell. layer is too thin, probably the entire catalyst layer is contaminated
From Fig. 5 and Table 3, the impedance for the cell with a with methanol (Fig. 6a), and hence the area available for oxygen
CCL thickness of 19 ␮m is much lower than that for the base reduction is small, resulting in a higher value of charge transfer
case with a CCL thickness of 8.4 ␮m, and the reduction is mainly resistance. As the thickness increases (Fig. 6b), part of the catalyst
due to the reduction in charge transfer resistance, Rct . When the layer can be free from methanol contamination. Thus the active cat-
cathode catalyst layer thickness increases from the base thick- alyst area available for oxygen reduction is greater and the charge
ness of 8.4 to 19 ␮m, the charge transfer resistance changes from transfer resistance is lower. But when the catalyst layer is too thick
0.114  to 0.080 ; approximately a 30% decrease in charge trans- (Fig. 6c), oxygen transfer through the catalyst layer is hindered and
fer resistance was achieved. Since charge transfer resistance is cannot effectively oxidize the methanol in a thin layer of the cat-
approximately inversely proportional to the active area, it can be alyst layer close to the membrane, resulting in a lower available
deduced that the active area for oxygen reduction is significantly area for oxygen reduction, again leading to a higher value of charge
increased with the thicker cathode catalyst layer of 19 ␮m. Such a transfer resistance. Besides, even the thickness of the catalyst layer
significant increase in the active area of the ORR with identical cat- that is free from methanol contamination in this case is not thinner,
alyst loading clearly shows that the effects of methanol cross-over due to the much lower catalyst loading per unit volume, the active
are significantly reduced by increasing cathode catalyst layer thick- catalyst surface area for oxygen reduction can still be significantly
ness from the base thickness of 8.4 to 19 ␮m. These results further lower. The optimal performance can only be achieved when the
corroborate the original hypothesis that a thicker cathode cata- CCL is thick enough to effectively oxidize the permeated methanol
lyst layer could have the crossed-over methanol oxidized in part of within part of the CCL, but not too thick to have significant adverse
the catalyst layer and leave the rest relatively free from methanol effect on the oxygen transfer and to significantly increase ohmic
contamination, thus reducing the mixed-potential effects. losses.
From Fig. 5 it can be seen that the impedance for the cell with To eliminate the possibility that the thicker CCL has a higher
the cathode catalyst layer thickness of 44 ␮m lies between the original ECA, further experiments were conducted to measure the
other two cases and from Table 3 it can be seen that the charge ECA of the three different CCLs without methanol contaminations.
transfer resistance is 0.104 , which falls between the values for The pristine ECA of the three catalyst layers were measured using
the base case of 8.4 ␮m and the one for 19 ␮m thicknesses. These standard cyclic voltammetry (CV) and the detailed parameters are
results clearly indicate that when the catalyst layer is too thick the provided in Section 2.3. Fig. 7 shows the CV voltammograms and the

Fig. 6. Hypothetical methanol concentration profiles for different cathode catalyst layer thicknesses.
S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606 605

Table 4 (a)
ECA values for cathode catalyst layers with different thicknesses. 0.80
8.4μm
Cathode ECA (cm2 Pt mg−1 Pt ) 0.70 19μm
44μm
MEA 1 282.88
0.60
MEA 2 254.37
MEA 3 243.28
0.50

Voltage / V
0.40
pristine ECA results are provided in Table 4. Each of the ECA values 0.30
listed in Table 4 is an average value of 4 independent CV results. As
can be clearly seen, the pristine ECA for the thicker catalyst layer 0.20
(19 ␮m) is not higher than the base case (it is actually lower). This 0.10
result further supports the original hypothesis.
To further determine if the anode side has any effect on the 0.00
0 100 200 300 400 500
different performances of the cells with different cathode catalyst
Current Density / mA cm-2
layer thickness, anode impedance spectra for the three MEAs at
1 M methanol concentration are measured and shown in Fig. 8. It is (b)
0.80
clear that all three MEAs have the same anode impedance spectra. 8.4 μm
This result suggests that the three MEAs have comparable cell resis- 0.70 19 μm
tances and mass transfer and electro-oxidation parameters in the 44 μm
anode. Therefore, the variation in performance shown in the polar- 0.60
ization curves (Fig. 2) and the variation in the cathode impedance 0.50

Voltage / V
spectra (Fig. 5) are solely due to the difference in cathode cata-
lyst layer thicknesses with no interference from the anode. This 0.40
approach of isolating the effects of cathode side from those from
0.30
the anode can also be used in future studies.
0.20

3.3. Higher methanol concentrations 0.10

Further studies were carried out with higher methanol concen- 0.00
0 100 200 300 400 500
trations. As shown in Fig. 9a and b, when 2 M and 4 M methanol Current Density / mA cm-2
concentrations were used, performance of the cell with the 19 ␮m
CCL thickness is still much better than that of the base case, but Fig. 9. Polarization curves at different cathode catalyst layer thicknesses for (a) 2 M
and (b) 4 M methanol concentrations. Conditions: 70 ◦ C; air flow rate 600 sccm;
the performance of the cell with the very thick CCL of 44 ␮m is
methanol flow rate 6 ml min−1 .
now much better. It can be seen from these two figures, as the
concentration of methanol increases, the performance of the cell
with thicker CCL increases. As the input methanol concentration 4. Conclusion
increases, methanol cross-over becomes more severe and even
the 19 ␮m CCL may not be thick enough to provide a meaning- Earlier experimental results by Ge and Liu [17] found that when
ful amount of CCL that is free from methanol contamination. Only there is enough oxygen supply to the cathode side, the adverse
a much thicker CCL may provide a significant amount of CCL that is effect of methanol cross-over can be reduced. It is believed that
relatively free from methanol contamination, however due to the when the methanol permeated to the cathode can be oxidized
much higher oxygen transfer resistance and higher ohmic losses, faster, more reaction sites will be free for ORR and thus mitigat-
the overall cell performance remains low. ing the effect of methanol cross-over. Based on these results, a
hypothesis was proposed that the performance of a DMFC can be
improved if the thickness of the cathode catalyst layer is increased
-0.1 to a certain value so that the crossed-over methanol can be oxi-
MEA 1, 8.4 μm CCL
dized in part of the catalyst layer close to the membrane and leave
MEA 2, 19 μm CCL
-0.08
MEA 3, 44 μm CCL rest of the catalyst layer free from methanol contamination, thus
Model reduce the detrimental effect of methanol cross-over and increase
-0.06 the active surface area for the ORR. In order to verify this hypothesis,
the performances of DMFCs with three different MEAs each having
-0.04 a different thickness of cathode catalyst layer have been studied.
ZIM / Ω

All the three MEAs have identical anode catalyst layer and identical
-0.02 cathode catalyst loading. The only difference is in the thickness
of the cathode catalyst layers. The technique of electrochemi-
0 cal impedance spectroscopy (EIS) has also been used to further
study the causes for the differences in fuel cell performances.
0.02 From the experimental results, the following conclusions can be
made:
0.04
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24
ZRE / Ω • The thickness of the cathode catalyst layer has a significant effect
on DMFC performances. Comparing with the base case, when the
Fig. 8. Anode impedance spectra at different cathode catalyst layers thicknesses.
thickness of the cathode catalyst layer is doubled, fuel cell current
Conditions: 70 ◦ C; current density: 100 mA cm−2 ; methanol concentration 1 M;
methanol flow rate 6 ml min−1 .
densities increased drastically.
606 S. Matar, H. Liu / Electrochimica Acta 56 (2010) 600–606

• Such significant increase in cell performance supports the [2] H. Dohle, J. Divisek, J. Mergel, H.F. Oetjen, C. Zingler, D. Stolten, Journal of Power
hypothesis that a thicker cathode catalyst layer allows the oxi- Sources 105 (2) (2002) 274.
[3] X. Ren, T.E. Springer, T.A. Zawodzinski, S. Gottesfeld, Journal of the Electro-
dization of methanol in part of the catalyst layer and leave the chemical Society 147 (2) (2000) 466.
remaining part free of methanol contamination, thus leading to [4] K. Ramya, K.S. Dhathathreyan, Journal of Electroanalytical Chemistry 542
a significant reduction in the mixed-potential effect. (2003) 109.
[5] J. Han, H. Liu, Journal of Power Sources 164 (1) (2007) 166.
• Further study by the EIS technique shows that the charge transfer
[6] W. Liu, C.-Y. Wang, Journal of Electrochemical Society 154 (2007) B352.
resistance for the fuel cell with thicker cathode catalyst layer is [7] Z.H. Wang, C.Y. Wang, Journal of Electrochemical Society 150 (2003) A508.
significantly lower, indicating that the active catalyst area avail- [8] J. Ge, H. Liu, Journal of Power Sources 160 (1) (2006) 413.
[9] J.T. Mueller, P.M. Urban, Journal of Power Sources 75 (1) (1998) 139.
able for oxygen reduction reaction is indeed greater, thus further [10] P. Piela, R. Fields, P. Zelenay, Electrochemical Society 153 (10) (2006) A1902.
supporting the original hypothesis. [11] L. Li, J. Zhang, Y. Wang, Journal of Membrane Science 226 (2003) 159.
• The experimental results also indicate that there exits an [12] N. Carretta, V. Tricoli, F. Picchioni, Journal of Membrane Science 166 (2000)
189.
optimum cathode catalyst layer thickness and the optimal
[13] R. Carter, R. Wycisk, H. Yoo, P.N. Pintauro, Electrochemical and Solid-State
CCL thickness depends on operating conditions, especially the Letters 5 (9) (2002) A195.
methanol concentration. [14] M.A. Abdel Rahim, H.B. Hassan, Thin Solid Films 517 (11) (2009) 3362.
[15] M. Carmo, V.A. Paganin, J.M. Rosolen, E.R. Gonzalez, Journal of Power Sources
142 (2005) 169.
References [16] F. Liu, G. Lu, C.-Y. Wang, Journal of the Electrochemical Society 153 (2006) A543.
[17] J. Ge, H. Liu, Journal of Power Sources 142 (2005) 56.
[1] S.R. Narayanan, A. Kindler, B. Jeffries-Nakamura, W. Chun, H. Frank, M. Smart, [18] C.Y. Du, T.S. Zhao, W.W. Yang, Electrochimica Acta 52 (2007) 5266.
T.I. Valdez, S. Surampudi, G. Halpert, Battery Conference on Applications and [19] C.Y. Du, T.S. Zhao, C. Xu, Journal of Power Sources 167 (2) (2007) 265.
Advances (1996) 113.

Potrebbero piacerti anche