Journal of Applied Electrochemistry 31: 275±279, 2001.

275
Ó 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Development and operation of a 150 W air-feed direct methanol fuel cell stack

D. BUTTIN1*, M. DUPONT1, M. STRAUMANN1, R. GILLE1, J-C. DUBOIS1, R. ORNELAS2,

G.P. FLEBA2, E. RAMUNNI3, V. ANTONUCCI4, A.S. ARICOÁ4, P. CRETIÁ4, E. MODICA4,
M. PHAM-THI5 and J-P. GANNE5
1
Sodeteg, 25, rue du Pont des Halles, Chevilly-Larue, 94666 Rungis Cedex, France
2
De Nora Fuel Cells, Via Bistol® 35, 20134 Milan, Italy
3
De Nora, Via Bistol® 35, 20134 Milan, Italy
4
CNR±TAE Institute for Transformation and Storage of Energy, Salita S. Lucia sopra Contesse 39,
98126 S. Lucia, Messina, Italy
5
Thomson±CSF/LCR, Domaine de Corbeville, 91404 Orsay, France
(*author for correspondence)

Received 20 March 2000; accepted in revised form 28 July 2000

Key words: direct methanol fuel cells, fuel cell stack, fuel cell performance, Pt catalysts, Na®on 117Ò

Abstract

A ®ve-cell 150 W air-feed direct methanol fuel cell (DMFC) stack was demonstrated. The DMFC cells employed
Na®on 117Ò as a solid polymer electrolyte membrane and high surface area carbon supported Pt-Ru and Pt
catalysts for methanol electrooxidation and oxygen reduction, respectively. Stainless steel-based stack housing and
bipolar plates were utilized. Electrodes with a 225 cm2 geometrical area were manufactured by a doctor-blade
technique. An average power density of about 140 mW cm)2 was obtained at 110 °C in the presence of 1 M
methanol and 3 atm air feed. A small area graphite single cell (5 cm2) based on the same membrane electrode
assembly (MEA) gave a power density of 180 mW cm)2 under similar operating conditions. This di€erence is
ascribed to the larger internal resistance of the stack and to non-homogeneous reactant distribution. A small loss of
performance was observed at high current densities after one month of discontinuous stack operation.

1. Introduction [1, 8, 23]. Although a great deal of progress has been

made, Na®onÒ membranes and Pt electrocatalysts still
In the last few decades, direct methanol fuel cells represent the most suitable choice for the production of
(DMFCs) have been actively investigated from both experimental prototypes and for prolonged operation
fundamental and applied points of view [1±26]. Many of [5±10].
the aspects related to the intrinsic mechanism of The DMFC devices have been characterized in terms
CH3OH electrooxidation on Pt-based catalysts have of both performance and eciency [22]. Recent results
been clari®ed and increasing e€orts are currently being have shown interesting prospects for the application in
made on the development of small size DMFC proto- the ®elds of electrotraction and portable power sources
types. At present, the DMFC utilizes a polymer elec- [7, 10, 22]. Yet, the demonstration of DMFC prototypes
trolyte membrane, such as Na®onÒ, on to which of suitable size has only recently been initiated [10]. The
electrocatalyst-containing carbon cloth- or carbon pa- design and operation of these systems require the
per-based electrodes are contacted on either side to form selection of proper components, cell housing and
a membrane±electrode assembly (MEA). A carbon experimental conditions.
supported Pt±Ru alloy is frequently used as an electro- In this communication, the development of a 150 W
catalyst for methanol oxidation whereas oxygen reduc- air-feed direct methanol fuel cell stack within the
tion occurs at suitable rates on Pt catalysts [12±23]. The framework of the NEMECEL project is described.
present Na®onÒ based per¯uoro±sulfonic membranes
are signi®cantly permeable to methanol [14, 15]. The
cross-over of CH3OH molecules through the membrane 2. Experimental details
leads to a parasitic methanol oxidation reaction at the Pt
cathode surface that signi®cantly lowers the cell poten- The DMFC stack was composed of ®ve 225 cm2 cells
tial [14]. Two strategies are currently being pursued connected in series. The stack housing was developed
to overcome this drawback, that is, oxygen-selective and produced by De Nora Fuel Cells and Sodeteg. It
reduction catalysts and methanol resilient membranes was made up of stainless steel (AISI 316) end plates and
276
bipolar plates. No gold or other noble-metal plating was electrodes consisted of carbon cloth, a di€usion layer
used. The materials were surface treated to avoid (carbon cloth and Te¯onÒ) and a reaction layer (catalyst
chemical and electrochemical corrosion. Both fuel and and Na®onÒ ionomer). The Pt content was 2 mg Pt cm)2
air were supplied from one side of the stack and on each electrode. Na®on 117Ò (DuPont) was selected as
distributed over the electrode backings through corro- the membrane. The MEA was manufactured by hot
sion-resistant metallic foams. Reactant inlets/outlets pressing the components at 130 °C for 90 s. The stack
were located in the same end plate. Each cell in the test station was designed and manufactured by Sodeteg.
stack was intercalated with a cooling/heating section in A ¯ow diagram of the DMFC stack plant is shown in
which silicon oil or pressurized water was circulated Figure 1. All variables i.e. temperatures, pressures,
through a thermostat in order to maintain the stack at reactant ¯ows and electrochemical parameters such as
the desired temperature. The cell temperature was stack current and voltage characteristics were computer-
measured through thermocouples located very close to controlled and visualized. Indeed, a speci®c software
each MEA. Both anodic and cathodic catalysts were program has been developed allowing direct piloting
synthesised at CNR±TAE Institute using a colloidal and continuous preview of the actuating gears on the
procedure. A continuous stirred tank reactor (CSTR) interactive screen, recording and ®ling of all the data
with a 12 litre reaction chamber was manufactured to relating to a trial and management of all the safety
produce the required amount of anodic and cathodic thresholds for each critical parameter. Methanol and air
catalysts in one batch in order to have identical catalyst were preheated at the desired temperature (110 °C
properties for each cell assembled in the stack. The methanol and 100 °C air) before being fed to the stack.
anodic catalyst formulation was 85% Pt±Ru (1:1)/ 1 M aqueous methanol solution was supplied to the
Vulcan XC. The methanol oxidation catalyst had an stack at a feed rate of 360 ml min)1; air was humidi®ed
atomic Pt/Ru ratio of 1 and an average particle size of before being fed to the stack at 25 l min)1. Most of the
about 2 nm. The oxygen reduction catalyst was an 85% measurements were collected at 110 °C stack tempera-
Pt/Vulcan with a mean particle size of 3.7 nm. The ture; under such condition, a 1.5 atm backpressure was
electrocatalytic activity of these materials was investi- imposed in the anodic compartment, while the pressure
gated in a 5 cm2 single cell (GlobeTech) equipped with in the cathodic compartment was 3 atm. Galvanostatic
serpentine graphite ¯ow®elds before being used for large steady-state polarization curves were collected through
area MEA production. Electrodes with an area of an electronic load SDCH G4 300 2K 60 backed up with
225 cm2 and MEAs were fabricated at the Thomson± a power supply 7.5-250-3-D-10T-TC. Internal resistance
CSF/LCR Corporate Research Laboratory, using a was measured by the current-interruption method using
procedure already developed at CNR±TAE [8, 12]. The a digital oscilloscope.

Fig. 1. Flow sheet of the DMFC stack plant.

 277
3. Results and discussion against current density together with their respective
power density outputs. The open circuit voltages are
Galvanostatic steady-state polarization data for the ®ve- similar, ranging between 0.77 and 0.8 V. There are only
cell air-feed stack at 110 °C are shown in Figure 2. The slight di€erences in the electrochemical behaviour of the
open circuit voltage approaches 4 V, yet a signi®cant various cells in the activation-controlled region at low
voltage loss (about 1 V) is observed at very low current current densities. The di€erence in cell voltages does not
densities indicating strong activation control. Almost exceed 40 mV up to 200 mA cm)2. The last cell (no. 5)
linear behaviour is observed in the stack polarization along the stack section from the inlet, performed better
curve up to 350 mA cm)2. This re¯ects the presence of in the activation-controlled region with respect to the
ohmic control. The curve slope is close to the value of ®rst. The slope of the I/V curves at intermediate current
the stack internal resistance determined by the current- densities is very similar for the various cells, showing that
interruption method (i.e., 5.5 mW). Accordingly, the there are no signi®cant di€erences in terms of internal
average cell resistance is 0.25 W cm2. Above resistance for the various assemblies. This was con®rmed
400 mA cm)2, mass-transfer limitations play a signi®- by measurement of the ohmic drop on each cell using the
cant role and the cell voltage drops signi®cantly. An current-interruption method. The ®rst cell in the stack
output power of 150 W is delivered at 80±90 A. with respect to the reactant inlet does not show a limiting
To understand the behaviour of each cell during stack current at high current densities and its corresponding
operation, the various cell voltages were collected at the power density increases progressively without reaching a
bipolar plate terminals. These are shown in Figure 3 maximum in the current range investigated. The power
output for the ®rst cell is about 170 mW cm)2 at
550 mA cm)2. The other cells show power densities of
between 120 and 140 mW cm)2. They are a€ected by
mass-transport limitations to di€erent extents, the last
cell (cell no. 5) showing the most signi®cant constraints.
In all the experiments carried out, cell voltage at high
current densities decreases almost regularly from the ®rst
(cell no. 1) to the last (cell no. 5). Since there was no
change in temperature pro®le at various cells, such
di€erences are ascribed to a nonhomogeneous distribu-
tion of reactants along the stack section. Although the
reactants enter from one side, they are distributed in
parallel to each cell along the stack section and subjected
to the same pressure imposed on both anodic and
cathodic compartments through backpressure regulators
at the reactant outlets. At this time, the source of such
mass-transfer losses is not individualized. Investigation
Fig. 2. Galvanostatic polarization data for the ®ve-cell air-feed
DMFC stack at 110 °C. and modi®cation in the DMFC stack design are cur-
rently being pursued at De Nora Fuel Cells and Sodeteg
and they should provide an improved design for the
second-term objective which involves the development of
a 1 kW stack. One other aspect related to the same
phenomenon relies on the di€erent behaviour at low
current densities between cell 1 and 5 (Figure 4). The
larger cell voltage losses observed in this region for cell 1
are attributed to a larger methanol concentration gradi-
ent at the anode±membrane interface with a consequent
increase of methanol cross-over through the electrolyte.
If, on the one hand, the fuel mass transport rate is
signi®cantly larger with respect to the reaction rate,
methanol accumulates at the interface and permeates
through the electrolyte under the di€usion driving force.
On the other hand, this accumulation of methanol at the
interface plays a favourable role at high current densities,
diminishing cell discharge due to rapid fuel consumption.
To evaluate the scale-up of the system from a 5 cm2
graphite single cell to a 150 W stainless-steel stack, the
Fig. 3. Galvanostatic polarization data and power densities at 110 °C polarization characteristics of cells 1 and 5 of the stack
for the various cells along the DMFC stack section from the reactant are compared in Figure 4 with those recorded in the
inlet. Key: (j) cell 1, (r) cell 2, (m), cell 3, (d) cell 4 and (¬) cell 5. small prototype under identical conditions. The graphite
278

Fig. 4. Comparison of galvanostatic polarization data and power

densities for cells 1 (j) and 5 ( ) in the stack and the 5 cm2 graphite
single cell (r) at 110 °C.

Fig. 5. Comparison of galvanostatic polarization data and power

cell performs slightly better in the activation-controlled densities for the ®ve-cell DMFC stack at 110 °C during the ®rst stage
region. Since the electrodes employed in the two devices of the experiments (j) and after one month of discontinuous operation
are identical, this aspect could be ascribed to improved (r).
use of the electrode surface area in the graphite cell;
month's operation. The internal resistance did not
further investigations are in progress. The most signif-
change signi®cantly, showing that no passivation pro-
icant di€erence between stack cells and the graphite cell
cess is currently a€ecting the metal components. Chem-
is due to the internal resistance value (0.25 W cm2
ical analysis of outlet aqueous solutions from both
against 0.11 W cm2). Accordingly, the maximum power
anode and cathode compartments revealed no loss of
density for the graphite cell is signi®cantly higher at
metal ions and no apparent change was observed in the
intermediate current densities. At 400 mA cm)2, the
stainless steel plates after one month operation.
power loss in the stack cell vs. the small graphite cell due
The observed increase of mass transport limitations
to the di€erent ohmic behaviour only, is about
may be attributed to some modi®cation occurring in the
25 mW cm)2. Thus, another aspect that needs to be
reactant ¯ow within the stack.
addressed for further stack development is the ohmic
drop. At ®rst glance, the main source of increased ohmic
loss in the stack appears to be due to contact resistance 4. Conclusions
between each electrode backing and metallic foam. The
foam is designed so as to have a sucient degree of A ®ve-cell 150 W air-feed stack was demonstrated
porosity, at the same time achieving a good compromise within the framework of the Nemecel Project. Catalysts,
between electrical contact and mass transport. Further electrodes, stack housing and test bench were developed
developments are currently underway, aiming at by the various partners. An average power density
decreasing all contact resistances in the system. approaching 150 mW cm)2 was achieved at 110 °C in
One signi®cant aspect concerns stack start-up under the presence of 1 M methanol and air feed at 3 atm. The
practical conditions (e.g., in an electric vehicle). Prelim- main limitations observed in the DMFC stack related to
inary experiments carried out on the ®ve-cell stack on the homogeneous distribution of reactants and internal
self-start-up at room temperature and ambient pressure resistance. Accordingly, speci®c advances are needed for
with 1 M methanol impregnated anodes gave an instant the stack components, especially the system housing and
power of 13 W. This is probably not suf®cient, at reactant ¯ow ®elds. Subsequent developments will
present, to drive all the auxiliaries needed to bring the concern the application of high temperature and highly
150 W stack up to suitable conditions in a very short conductive cross-over resilient membranes as an alter-
time. However, all stack characteristics are optimised native to Na®onÒ.
for operation at high temperature and suitable pressure;
more investigation on self start-up is therefore needed.
After one month of discontinuous stack operation Acknowledgements
under various operating conditions (start-up in the
morning and shut-down during the night) no signi®cant Collaboration in this project by Solvay (Belgium),
change in stack performance was observed at low and University of Poitiers (France), CNRS±LMOPS (Ly-
intermediate current densities up to 250 mA cm)2, on-France) and PSA (France) is greatly appreciated.
whereas, a slight decrease in limiting current was The Nemecel Project is being developed with the
registered. Figure 5 shows the polarization curves re- ®nancial support of the European Commission, Direc-
corded in the ®rst stage of stack operation and after one torate General XII, Programme Joule III.
 279
References 15. S. Cleghorn, X. Ren, S. Thomas and S. Gottesfeld, `Book of
Extended Abstracts', 1997 ISE-ECS Joint Symposium, Paris, Sept.
1. K. Kordesch and G. Simader, `Fuel Cells and their Applications' (1997), Abstract 182, pp. 218±219.
(VCH, Weinheim, 1996). 16. A.S. AricoÁ, P. Creti, H. Kim, R. Mantegna, N. Giordano and
2. A. Hamnett, Phil. Trans. R. Soc. Lond. A354 (1996) 1653. V. Antonucci, J. Electrochem. Soc. 143 (1996) 3950.
3. M.P. Hogarth and G.A. Hards, Platinum Metal Rev. 40 (1996) 150. 17. S. C. Thomas, X. Ren and S. Gottesfeld, in S. Gottesfeld and T.F.
4. C. Lamy and J-M. LeÂger, in O. Savadogo and P.R. Roberge (Eds), Fuller (Eds), `Proton Conducting Fuel Cells in Proton Conducting
Proc. 2nd International Symp. on `New Materials for Fuel Cells Membrane Fuel Cells' (Second International Symposium), Proc.
and Modern Battery Systems', MontreÂal, Canada (1997), pp. 98±27, (Electrochemical Society, Pennington, NJ, 1999), pp. 267±
477±487. 272.
5. A.K. Shukla, P.A. Christensen, A. Hamnett and M.P. Hogarth, 18. T.J. Schmidt, H.A. Gasteiger and R.J. Behm, Electrochem.
J. Power Sources 55 (1995) 87. Commun. 1 (1998) 1.
6. L. Liu, C. Pu, R. Viswanathan, Q. Fan, R. Liu and E.S. Smotkin, 19. M. Gotz and H. Wendt, Electrochim. Acta 43 (1998) 3637.
Electrochim. Acta 43 (1998) 3657. 20. K. Lasch, L. Jorissen and J. Garche, in C. Lamy and H. Wendt
7. X. Ren, M.S. Wilson and S. Gottesfeld, J. Electrochem. Soc. (Eds), Proceedings of the Workshop `Electrocatalysis in Indirect
143 (1996) L12. and Direct Methanol PEM Fuel Cells', 12±14 Sept. (1999),
8. A.S. AricoÁ, P. Creti, P.L. Antonucci and V. Antonucci, Electro- Portoroz, Slovenia, p. 104.
chem. Solid State Lett. 1 (1998) 4. 21. H. Kim and A. Wieckowski, in O. Savadogo (Ed), Third
9. S.R. Narayanan, W.Chun, T.I. Valdez, B. Je€ries-Nakamura, International Symposium. on `New Materials for Fuel Cells and
H. Frank, S. Surampudi, G. Halpert, J. Kosek, C. Cropley, A.B. Modern Battery Systems', MontreÁal, Canada (1999), pp. 125±126.
LaConti, M. Smart, Q. Wang, G. Surya Prakash and G.A. Olah, 22. R.M. Moore, S. Gottesfeld and P. Zelenay, in S. Gottesfeld and
Program and Abstracts, Fuel Cell Seminar (1996), pp. 525±8. T.F. Fulles (Eds), Proceedings, ref. [17], 98±27, (1999), pp. 365±
10. M. Baldauf and W. Preidel, J. Power Sources 84 (1999) 161. 379.
11. S. Wasmus and A. Kuver, J. Electroanal. Chem. 461 (1999) 14. 23. B.D. McNicol, D.A.J. Rand and K.R. Williams, J. Power Sources
12. A.S. AricoÁ, A.K. Shukla, K.M. el-Khatib, P. CretõÁ and V. 83, (1999) 15.
Antonucci, J. Appl. Electrochem. 29 (1999) 671. 24. M. Watanabe and S. Motoo, J. Electroanal. Chem. 60 (1975) 275.
13. A. Hamnett and B.J. Kennedy, Electrochim. Acta 33 (1988) 1613. 25. P.S. Kaurenan and E. Skou, J. Electroanal. Chem. 408 (1996) 189.
14. K. Scott, W.M. Taama, P. Argyropoulos and K. Sundmacher, 26. P.L. Antonucci, A.S. AricoÁ, P. CretõÁ , E. Ramunni and V.
J. Power Sources 83 (1999) 204. Antonucci, Solid State Ionics 125 (1999) 431.