Energy Conversion and Management 45 (2004) 3239–3249

www.elsevier.com/locate/enconman

Performance optimisation of PEM fuel cell

during MEA fabrication
Wan R.W. Daud a, Abu Bakar Mohamad a, Abdul Amir H. Kadhum a,
Rachid Chebbi a, Sunny E. Iyuke b,*
a
Department of Chemical and Process Engineering, Universiti Kebangsaan Malaysia,
43600 UKM Bangi, Selangor DE, Malaysia
b
Department of Chemical and Environmental Engineering, Faculty of Engineering, Universiti Putra Malaysia,
43400 UPM Serdang, Selangor DE, Malaysia
Received 3 July 2003; received in revised form 5 January 2004; accepted 9 January 2004
Available online 18 February 2004

Abstract
We have analysed membrane electrode assemblies (MEAs) involving fabricated and commercially
available electrodes using a scanning electron microscope (SEM) and X-ray photoelectron spectroscopy
(XPS) and developed simple mathematical models to simulate the best performance and design conditions.
The analysis showed that a MEA surface with the catalyst layer consisting of 10 wt% Pt/C and 30 wt%
Teflon
(PTFE, designated E2) loaded with 0.38 mg Pt/cm2 showed good localisation of the platinum
particles. The SEM image of the E2 electrode showed the existence of a diffusion layer, while the cross-
section of electrode E3 (without diffusion layer) showed only the backing layer of the carbon cloth. It was
seen that good adhesion of the catalyst on the membrane was obtained as a result of the hot press used in
fabrication. XPS analysis showed that the electrode surfaces consisted of C, O, F, Si and Pt, whose binding
energies for the PTFE/C layer were C 1s, O 1s, F 1s and Si 2p states and were 285.0, 532.7, 689.5 and 103.0
eV, respectively. While for the catalyst layer, the binding energies for the elements, C 1s, O 1s, F 1s, Si 2p and
Pt 4f states, were 284.3, 532.4, 689.3, 102.9 and 74.1 eV, respectively. Similar observations were made for a
commercial E-TEK electrode. The mathematical and simulation investigations supported the hypothesis
made in an earlier study in terms of optimum PEM fuel cell performance determination and design simu-
lation. The calculated values of the voltage operational limit Vopl cal: agreed quite well with the experimental
data Vopl exp: reported earlier. Other works from the open literature were also correlated using the mathe-
matical model, and it was found that the Vopl values were comparable. Hydrogen usage thus calculated was
best with the E2 electrode compared to E1, E3 and the commercially available E-TEK electrode.
 2004 Elsevier Ltd. All rights reserved.

*
Corresponding author. Tel.: +603-89466294; fax: +603-86567120/099.
E-mail addresses: [email protected], [email protected] (S.E. Iyuke).

0196-8904/$ - see front matter  2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enconman.2004.01.011
3240 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249

Keywords: Fuel cell; MEA; Design and simulation; Hydrogen usage; Optimisation

1. Introduction

The proton exchange membrane (PEM) fuel cell technology promises an environmentally
friendly power source with a variety of applications in several industries including transportation
and portable and stationery power generation. Hence, in such promising technology, manufac-
turing cost reduction would be a great incentive, and such can be realised by optimising the effi-
ciency of the fuel cells through detailed analysis of the complex electrochemical reactions and the
transport phenomena involved in the membrane and electrode assembly (MEA) of the cells. These
processes are major contributors to the fuel cells voltage drop from the usual open circuit potential
(normally about 1.2 V for a PEM fuel cell operating below 100 C) to a much less voltage value
when the cell is run. To get maximum output from the fuel cell, therefore, the MEA should possess
effective porous gas diffusion electrodes, which are typically fabricated from porous conductive
carbon based materials. The requirements for such materials are good gas diffusion properties,
suitable water permeability, low resistance and chemical and physical durability. The paramount
functions of porous gas diffusion electrodes are distributing reactants to the active areas, draining
excess water from the electrodes and providing electrical contact between the electrodes and the
flow field plates. In this article, as we earlier reported [1], the materials used for electrode fabri-
cation were carbon cloth, 10 wt% platinum supported carbon black as electrocatalyst using 60
wt%, polytetrafluoroethylene (PTFE) as a binder and wet proofing agent. A Nafion
solution was
used as a proton conductor between the catalyst layer and the Nafion
117 membrane.
The article is, therefore, aimed at analysing the MEA involving fabricated and commercially
available electrodes using a scanning electron microscope (SEM) and X-ray photoelectron
spectroscopy (XPS) and developing simple mathematical models to simulate the best performance
and design conditions.

2. Mathematical models

If it is assumed that the slowness of the reaction taking place on the surface of the electrodes,
which may either be caused by reactants or products diffusivity to and from the gas diffusion
layers, is the only cause of irreversibility, then the PEM fuel cell voltage and irreversibility may be
given as in Eqs. (1) and (2) [1,2], respectively.
V ¼ E
A lnði=i0 Þ ð1Þ
and
Irr ¼ A lnði=i0 Þ ð2Þ
where V is the cell voltage, E the reversible open circuit voltage, Irr the irreversibility, i the current
RT
density and i0 the exchange current density, while A ¼ 2aF in which a is referred to as the charge
transfer coefficient, and it is proportional to the electrical energy applied in changing the rate of
an electrochemical reaction. F is FaradayÕs constant, R the gas constant and T the absolute
 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249 3241

temperature. However, it was reported earlier [1] that at an appropriate exchange current density,
now designated as io;vopl , the irreversibility is equal to the cell voltage at an optimum voltage,
otherwise referred to as the voltage operational limit Vopl , i.e. Irr ¼ Vopl . Therefore, we may write,
pffiffiffiffiffiffiffiffi
i ¼ io;vopl eE=A ð3aÞ
Thus, when io;vopl  1, the equation becomes,
pffiffiffiffiffiffiffiffi
i ¼ eE=A ð3bÞ
Hydrogen fuel usage is defined [2] as,
ni
H2 usage ¼ ; where n is the number of cells ð4Þ
2F
Substituting Eq. (3b) into Eq. (4), considering F ¼ 96 485 coulombs and the molecular mass of
H2 ¼ 2.02 · 10
3 kg/mol for the single cell, Eq. (4) becomes,
pffiffiffiffiffiffiffiffi
H2 usage ¼ 1:05  10
8 eE=A kg=s ð5Þ
Incorporating Eq. (3b) into Eq. (1) with rearrangement will give the expression for the voltage
operational limit as,
pffiffiffiffiffiffiffiffi !
eE=A
Vopl ¼ E
A ln ð6Þ
i0
or, using Eq. (3a) instead of Eq. (3b),
h pffiffiffiffiffiffiffiffii
Vopl ¼ E
A ln ðio;vopl =i0 Þ eE=A

3. Experimental

The major part of the initial experimental works involving electrode preparation and fabri-
cation, treatment of the Nafion
membrane, membrane electrode assembly, experimental setup
for a single cell unit, MEA and mode of operation have been well presented elsewhere [1,3].
Carbon cloth ‘‘A’’ (E-TEK, Inc.) with thickness 350 lm was used as the substrate for casting the
platinum supported carbon, 10 wt% Pt/C (Fluka, Chemie AG), activated carbon black (Ajax
Chemicals, Sydney) and Teflon (PTFE) 60 wt% (Aldrich Chemical, Inc.) mixture. Three different
structures of electrodes were prepared with 70 wt% Pt/C and 30 wt% Teflon. One structure was
with a diffusion layer and 0.18 mg Pt/cm2 loading, designated E1; the second was with a diffusion
layer and 0.38 mg Pt/cm2 loading, designated E2; while the third was without diffusion layer but
with 0.4 mg Pt/cm2 loading, designated E3.
The final electrodes preparation with platinum loadings of 0.18, 0.38 and 0.4 mg/cm2
impregnated with 5% Nafion
solution by weight dissolved in a mixture of lower aliphatic alcohol
and water was done using a brushing technique [4,5]. They were air dried at 80 C and then
weighed. The electrodes and the membrane were coupled into a single cell MEA by the hot
pressing procedure [4] using the press (Scientific Press-Motorized Hydraulic 20 tonnes) at tem-
perature 120 C and pressure of 1000 psi for 1 1/2 min. The MEA was then cooled at room
3242 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249

temperature for 30 min [1]. However, the emphasis in this study only includes the MEA and its
components analysis using a SEM and X-ray photoelectron spectroscopy (XPS) using the XSAM-
HS spectrometer (KRATOS) equipped with magnesium as an X-ray source.

3.1. SEM analysis

Four samples of 1 cm2 size each were used for the experiment. The samples consisted of catalyst
layers obtained from a commercial electrode (E-TEK) and the ones prepared as discussed above.
The cross-sections of the MEA with or without diffusion layers were also analysed.

3.2. XPS technique for surface analysis

A 400 lm and 1 cm2 surface area electrode sample was prepared and cleaned to avoid con-
tamination by wrapping it with clean aluminium foil and analysed by XPS (XSAM-HS spec-
trometer, KRATOS). The measurement was made at ultra high vacuum to allow the
photoelectrons to reach the detector without striking a gas atom and to avoid contamination of
the sample surface. The gate valve was slowly opened, and the sample was inserted using a special
arm and the gate closed. The vacuum was left for a few hours until the pressure reached about
10
6 Torr, and the analysis was then performed.

4. Results and discussion

The discussion of the improved MEA design and performance simulation of the single cell
PEM fuel cell consists of the MEA characterisation and surface analysis and the performance
optimisation and simulation or prediction for any specific PEM fuel cell unit.

4.1. SEM image analysis of the electrodes

The SEM was used to identify the electrode surface and the distribution of the catalyst (Pt) and
monomer. Fig. 1 presents the PTFE/C diffusion layer without catalyst deposition, while Fig. 2
presents the surface of the catalyst layer of 10 wt% Pt/C with 30 wt% Teflon
(PTFE) and Pt
loading of 0.38 mg/cm2 , otherwise designated as the E2 electrode. The SEM image of this layer
showed good localisation of the platinum particles on the electrode surface. This was due to the
spraying technique [1,3] used during fabrication of the electrode. Fig. 3 presents the SEM image of
the commercial (E-TEK) electrode, which has fewer grains of platinum (the bright particles) on
the surface of the electrode than were seen on the E2 electrode surface. The distribution of the
platinum particles on the prepared electrode appeared to be more widely spread than on the
E-TEK electrode. The better Pt particles localisation can also be attributed to the spraying
technique used during fabrication. Fig. 4 shows the SEM image of the cross-section of the MEA
of the E2 (0.38 mg Pt/cm2 ) electrode with diffusion layer (PTFE/C), while Fig. 5 presents the cross-
section of the MEA for the E3 electrode but without diffusion layers. In comparison, Fig. 4 (E2
electrode) indicated the presence of the diffusion layer, while the diffusion layer did not exist in
Fig. 5 (E3 electrode). The difference observed in Figs. 4 and 5 may be due to the hot pressing
 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249 3243

Fig. 1. PTFE/C diffusion layer.

Fig. 2. Pt/CPTFE 10 wt% Pt/C catalyst layer with Pt loading of 0.38 mg/cm2 .

employed during the membrane electrode assembly, resulting in good adhesion of the catalyst to
the membrane.

4.2. XPS surface analysis

The XPS surface data for the PTFE/C diffusion layer, Pt/C/PTFE catalyst layer and E-TEK
electrode are presented in Table 1(panels a–c). Table 1(panel a) has four major peaks for C 1s,
O 1s, F 1s and Si 2p having mass concentrations of 47.5%, 8.5%, 36.0% and 7.2%, respectively,
3244 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249

Fig. 3. 10 wt% Pt/C E-TEK electrode with Pt loading of 0.4 mg/cm2 .

Fig. 4. Cross-section of E2 electrode with diffusion layer.

with an almost negligible mass concentration of 0.6% for Pt 4f. In the same sequence, Table
1(panel b) (E2) has mass concentrations of 44.96%, 6.85%, 40.43% and 3.94% respectively, with a
greater mass concentration of 3.82% for Pt 4f, while Table 1(panel c) (E-TEK) has mass con-
centrations of 41.82%, 3.0%, 51.3% and 2.6% with the mass concentration for Pt 4f equal to 2.6%,
which is also less than that for the E2 electrode. It implies that five major peaks (C 1s, O 1s, F 1s,
 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249 3245

Fig. 5. Cross-section of MEA of (E3) electrode without diffusion layer.

Table 1
XPS surface analysis for PTFE/C diffusion layer (panel a), Pt/C/PTFE catalyst layer (panel b)and E-TEK electrode
(panel c)
Peak BE (eV) Atomic concentration (%)
Panel a
C 1s 285.0 47.5
O 1s 532.7 8.5
F 1s 689.5 36.0
Si 2p 103.0 7.2
Pt 4f 73.5 0.6
Panel b
C 1s 284.3 44.96
O 1s 532.4 6.85
F 1s 689.3 40.43
Si 2p 102.9 3.94
Pt 4f 74.1 3.82
Panel c
C 1s 284.5 41.82
O 1s 532.9 3.0
F 1s 698.4 51.3
Si 2p 102.8 2.6
Pt 4f 72.4 2.6

Si 2p and Pt 4f) were observed for both the Pt/C/PTFE (E2) catalyst layer and the E-TEK elec-
trode whose binding energies were 284.3, 532.4, 689.3, 102.9 and 74.1 eV, respectively, for
the former, while for the latter (E-TEK), the binding energies were 284.5, 532.9, 698.4, 102.8 and
3246 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249

72.4 eV, respectively. It can also be observed that the elemental compositions of both the prepared
electrode and the E-TEK electrode are somewhat comparable (Table 1(panels b and c)).

4.3. PEM fuel cell performance optimisation and simulation

Table 2 presents the characteristics of both the fabricated and commercially available (E-TEK)
electrodes. Particularly, constant A (column 7) was obtained using the Tafel equation [2,6,7] given
(Eq. (7)) below and other characteristic parameters are presented in Table 2, while i0 was eval-
uated as discussed earlier by Iyuke et al. [1]. The Tafel equation was developed and reported by
Tafel in 1905 [2] from experimental studies that the overvoltage or irreversibility at the surface of
an electrode followed a similar pattern, as in a variety of electrochemical reactions, represented by
Eq. (7).
 
i
V ¼ A ln ð7Þ
i0

Thus, upon using Eq. (6) to calculate the voltage operational limit, which is also known to as the
optimum attainable voltage, the simulated data agreed quite well with the experimental ones
reported earlier [1], as presented in Fig. 6. Fig. 6 also shows the corresponding hydrogen usage
curve, which was determined from Eq. (5). It was also observed that the E2 electrode structure
performed best in terms of power generation and hydrogen usage compared to E1, E3 and the
commercially available one (E-TEK). The results thus support the proposition reported earlier [1]
that the E2 electrode has a greater efficiency than the E1, E3 and E-TEK electrodes. Moreover,
while these electrodes can be operated conveniently, the E2 electrode with 0.38 mg Pt/cm2 loading
possessed the optimum voltage operational limits.
The data in Fig. 6, therefore, confirms the applicability of Eq. (6) in determining the attainable
PEM fuel cell performance with respect to the voltage that can be delivered at an appropriate
exchange current density. Such information would be invaluable in predetermining the design
conditions before specific electrode and MEA fabrication for either a single cell or stack of a PEM
fuel cell unit is embarked upon. Thus, in applying the proven Eq. (6) to previous works in the
open literature [8–10] with different MEA fabrication techniques and platinum loadings and then
comparing with the results obtained in this study provides Table 3. Table 3 shows that the works
of Bender et al. [8], Pozio et al. [9], Kim et al. [10] and this study provide i0 equal to 0.727, 0.627,

Table 2
Characteristics of the fabricated and commercial electrodes for a single cell PEM fuel cell (25 cm2 ) operated at
30 C 6 temperature 6 80 C and atmospheric 6 pressure 6 2 bar
Electrode Pt loading Open volt- Current, i Power (mW/ Resistance Exch. cur- A (Eq. (1))
structure (mg Pt/cm2 ) age (V) (mW) cm2 ) (X/cm2 ) rent density, (V)
i0 (mA/cm2 )
E1 0.18 0.937 14.8 4.44 0.05 0.4 0.031
E2 0.38 1.077 41.2 21.57 0.01 0.87 0.032
E-TEK 0.4 0.989 48.4 20.32 0.012 1.2 0.03
E3 0.4 0.934 12.8 2.94 0.064 0.3 0.034
 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249 3247

0.6 0.25
E2
E- TE K

Voltage Operational Limit, Vopl (V)

E1
0.5 E3
0.2

Hydrogen Usage (kg/s)

0.4
0.15

0.3
0.1
0.2 Vopl.exp
Vopl.cal
0.05
0.1 Hydrogen usage

0 0
0 0 .25 0.5 0.75 1 1.25 1. 5
2
Exchange Current Density (mA/cm )

Fig. 6. Effect of exchange current density on voltage operational limit and hydrogen fuel usage in determining PEM
fuel cell performance and design.

Table 3
Comparison of the estimated voltage operational limit (Vopl ) in the study with literature
Authors Methods Platinum i0 A Vopl (V)
loading (mA/cm2 )
Bender Water ink machine coated on kapton 0.20 mg 0.727 )0.052 0.516
et al. [8] for 5 cm2 cell at 80 C Pt/cm2
[average]
Pozio Previously prepared MEA by spray 0.68 mg 0.627 )0.059 0.508
et al. [9] technique with Pt/C catalyst powders on carbon Pt/cm2
black, was sealed in a moulded gasket 106 cm2 cell [designated
operated at 75 C MEGA 1]
Kim Hot pressing of perfluorosulfonyl floride copolymer 0.2 mg 0.716 )0.072 0.504
et al. [10] powder and Pt/C catalyst on sheet at 230 C. 25 cm2 Pt/cm2
active area cell operated at 80 C
This study Carbon black, Teflon
and PTFE ink solution was 0.38 mg 0.823 )0.072 0.509
sprayed on backing layer, and MEA hot pressed at Pt/cm2
120 C. [designated
25 cm2 cell operated at 80 C E2]

0.716 and 0.823, respectively; the A values were )0.052, )0.059, )0.072 and )0.072, respectively;
while the Vopl were 0.516, 0.508, 0.504 and 0.509, respectively. Interestingly, it is also very
important to observe that if the exchange current density is taken at 1.0 mA/cm2 in Fig. 6, the
voltage operational limit and hydrogen usage fall just to the right of the maxima of the curves at
approximately the same numerical values as those for electrode E2. It is expedient to suggest that
this point be explored further in order to maximize the voltage operational limit. From Tables 2
3248 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249

and 3 and Fig. 6, it seems that the optimum performance might be approximated better if the
exchange current density at which V ¼ Irr were taken as 0.9 mA/cm2 rather than 1.0 mA/cm2 .
On the whole, the results imply that particularly the voltage operational limits for all processes
considered with this study were comparable, thereby supporting the applicability of the model
(Eq. (6)), which is the first of its kind.

5. Conclusion

Different structures of electrodes, namely electrodes with or without diffusion layer, and
different compositions of electrocatalysts were used to study their effects on the performance of
the electrodes and the MEAs. The performances of the electrodes were analysed by the elec-
trochemical response of the electrodes in a test cell. The morphological characteristics and
chemical state of the electrodes were also analysed by a scanning electron microscope (SEM)
and X-ray photoelectron spectroscopy (XPS), respectively. It was observed that the surface with
the catalyst layer with 10 wt% Pt/C and 30 wt% Teflon
(PTFE loaded with 0.38 mg Pt/cm2
showed good localisation of the platinum particles. The SEM image of the E2 electrode showed
the existence of the diffusion layer, while the cross-section of electrode E3 showed only the
backing layer of carbon cloth. This was attributed to the absence of a diffusion layer and
the hot pressing used in the MEA fabrication, showing that good adhesion of the catalyst on
the membrane was achieved. The XPS analysis showed that the electrode surfaces consisted of
C, O, F, Si and Pt. The binding energies for the PTFE/C layer were the C 1s, O 1s, F 1s, and
Si 2p states with 285.0, 532.7, 689.5 and 103.0 eV, respectively, while for the catalyst layer, the
binding energies for the elements states, C 1s, O 1s, F 1s, Si 2p anf Pt 4f were 284.3, 532.4, 689.3,
102.9 and 74.1 eV, respectively. These results were found to be in agreement with ones obtained
from the commercial E-TEK electrode where similar elemental components of C, O, F, Si and
Pt were observed.
The mathematical and simulation studies made supported the hypothesis made in an earlier
study in terms of optimum PEM fuel cell performance determination and design simulation.
The calculated values of the voltage operational limit Vopl cal: agreed quite well with the
experimental data for Vopl exp: reported earlier. The model obtained for such calculation was
given as,
pffiffiffiffiffiffiffiffi !
eE=A
Vopl ¼ E
A ln
i0

which also successfully simulated other works in the open literature. Convincingly, it was also
observed that if the exchange current density was taken at 1.0 mA/cm2 in Fig. 6, the voltage
operational limit and hydrogen usage fall just to the right of the maxima of the curves at
approximately the same numerical values as those for electrode E2. It is expedient, therefore, to
suggest that this point be explored further in order to maximize the voltage operational limit.
From Tables 2 and 3 and Fig. 6, it seems that the optimum performance might be approximated
better if the exchange current density at which V ¼ Irr were taken as 0.9 mA/cm2 rather than
 W.R.W. Daud et al. / Energy Conversion and Management 45 (2004) 3239–3249 3249

1.0 mA/cm2 . Results also worth mentioning are the hydrogen usage, which supported the E2
electrode as the best performing one compared to the others.

Acknowledgements

Financial support from the Malaysia Ministry of Science, Technology and Environment
through the project IRPA 02-02-02-0002 is gratefully acknowledged. The authors also want to
emphasis their gratitude to the staff and the Department of Chemical and Process Engineering,
UKM where the experimental works were conducted.

References

[1] Iyuke SE, Mohamad AB, Kadhum AAH, Daud WRW, Chebbi R. Improved membrane and electrode assemblies
for proton exchange membrane fuel cells. J Power Sources 2003;114(2):195–202.
[2] Larminie J, Dicks A. Fuel Cell Systems Explained. Chichester: John Wiley & Sons, Ltd; 2001. p. 17.
[3] Chebbi R. Fabrication of low platinum loading electrode for proton exchange membrane fuel cells system
(PEMFCs), Masters Thesis, Universiti Kebangaan Malaysia, 2001, p. 43.
[4] Ticianelli EA, Derouin CR, Redondo A, Srinivasan S. Methods to advance technology of proton exchange
membrane fuel cell. J Electrochem Soc Electrochem Sci Technol 1988;135:2209–14.
[5] Felix NB, Srinivasan S. Operating proton exchange membrane fuel cells without external humidification of the
reactant gases. J Electrochem Soc 1997;144:2767–72.
[6] Rangarajan KS, Yegnanaranan V. Current losses in a bipolar cell––an analysis of the Tafel regime. Electrochim
Acta 1997;42(1):153–65.
[7] Lasia A. Hydrogen evolution/oxidation reactions on porous electrodes. J Electrochem Chem 1998;454:115–21.
[8] Bender G, Zawodzinski AT, Saab PA. Fabrication of high precision PEFC membrane electrode assemblies.
J Power Sources 2003;124:114–7.
[9] Pozio A, Giorgi L, De Francesco M, Silva FR, Lo Presti R, Danzi A. Membrane electrode gasket assembly
(MEGA) technology for polymer electrolyte fuel cells. J Power Sources 2002;112:491–6.
[10] Kim CS, Chun GY, Peck HD, Shin RD. A novel process to fabricate membrane electrode assemblies for proton
exchange membrane fuel cells. Int J Hydrogen Energy 1998;23(11):1045–8.