Volumic electrodes of fuel cells with polymerelectrolyte membranes: electrochemicalperformances and structural analysis by

thermoporometry

S. Escribano,a P. Aldeberta and M. Pinerib*

aDe partement de Recherche Fondamentale sur la MatieÁ re Condense e, SI3M/PCM, CEA/Grenoble,17 rue des Martyrs, 38054 Grenoble Cedex 9, France

bCentre d'Etudes et de Recherche sur les Mate riaux, DEM/SPCM/DIR, CEA/Grenoble, 17 rue desMartyrs, 38054 Grenoble Cedex 9, France

(Received 2 April 1997; in revised form 7 August 1997)

AbstractÐOur studies deal with electrodes and Membrane Electrodes Assemblies (MEAs) for PolymerElectrolyte Membrane Fuel Cells (PEMFCs). In this paper the structure and the electrochemical behaviourof home-made electrodes are correlated with their preparation method. Di�erent aspects of electrode prep-aration are considered such as the composition and the heat treatments of the active layer. First plots ofpotential vs current density are presented to con®rm the important in¯uence of the preparation method onthe electrochemical performance of the fuel cell. Then a particular structural point is studied: the porosityof the active layer always with regard to the preparation method. The paper shows the interest of a speci®ctechnique, thermoporometry, for studying porosity in this type of electrode. # 1998 Elsevier Science Ltd.All rights reserved.

Key words: Polymer Electrolyte Membrane Fuel Cell (PEMFC), electrodes preparation/electrochemical per-formance, porosity of the active layer.

INTRODUCTION

Electric vehicles constitute one way to decrease pol-

lution especially in large cities. Programs to develop

such technology are prompted by state decisions

like the ``zero emission vehicle act'' in di�erent

U.S.A. states. Batteries will certainly be one way to

deliver the necessary energy, especially with the

development of high energy density batteries like

lithium polymer batteries. However, even with

scheduled values of 0100 Wh/kg, performances of

battery-powered electric cars will be always limited

in terms of autonomy and recharging times.

Polymer Electrolyte Membrane Fuel Cells

(PEMFCs) are now considered, among the di�erent

available fuel cell technologies, as the best candi-

dates for electric vehicles.

In fuel cells, electrical energy is produced by the

electrochemical combination of hydrogen and oxy-gen from air which are respectively oxidised and

reduced inside two volumic electrodes separated byan electrolyte membrane. Water is the only by-pro-

duct of such a reaction. PEMFC technology impliesthe use of hydrogen, either stored or produced on

board by reforming, and needs high performancemembranes and Membrane Electrode Assemblies

(MEAs). Demonstration of the feasibility of this

concept has already been made with buses (Ballardcompany) [1] and cars. However, large scale devel-

opment of these PEMFCs implies a reduction ofcost which implies in particular the optimisation of

the MEAs with less expensive membranes and verylow amounts of platinum catalyst.

Advances in reduction of platinum loading beganabout ten years ago when electrodes commonly

contained high loadings (4 mg/cm2) of unsupportedplatinum black. Studies were intended to optimise

Electrochimica Acta, Vol. 43, Nos. 14±15, pp. 2195±2202, 1998# 1998 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0013±4686/98 $19.00+0.00PII: S0013-4686(97)10108-6

*Author to whom correspondence should be addressed.E-mail: pineri@drfmc.ceng.cea.fr

2195

commercial E-tek electrodes with low catalyst load-ing (R0.5 mg/cm2) made with carbon supported

platinum catalyst: the extension of the three dimen-sional reaction zone by Na®on1 impregnation andthe hot-pressing of the electrodes on to the mem-

brane allowed the same level of performance to beachieved with only one tenth of the noble metalloading [2, 3]. It was then demonstrated that higher

power densities could be attained in PEMFCs bylocalization of platinum near the front surface ofthe electrodes [4, 5]. Since then, several preparation

methods have been developed to increase the utiliz-ation of platinum in order to obtain a high level ofelectrochemical performance with ultra-low Pt load-ing electrodes (0.1 mg/cm2). This aim was reached

by preparing very thin active layers, using new tech-niques such as spraying and rolling the active layeron a di�usion layer [6, 7] or casting thin ®lm active

layers directly on to the electrolyte membrane [8±10].The spraying technique perfected in our

laboratory [11±13] has been used to develop electro-des with thin ultra-low loaded active layers coatedeither on home-made di�usion layers, to form a

classical double layer electrode [6, 14, 15] similar tocommercial E-tek electrodes, or directly on to themembrane. This last original method was perfectedin order to improve the electrode/membrane ionic

interface [11, 12]. These two methods allowed us toprepare electrodes with ultra-low platinum loadings(R0.1 mg/cm2 for the cathode and R0.15 mg/cm2

for the whole MEA) and good electrochemicalbehaviour [13].In the experiments presented in this paper,

Na®on1 117 is used as the polymer electrolytemembrane. The volumic electrodes are made of twolayers:

± the active layer containing the carbon sup-ported platinum catalyst particles on which reac-

tions take place. This layer is also composed ofpoly(tetra¯uoroethylene) (PTFE) which acts both asa binding and a hydrophobic agent and, ®nally,

membrane solution which produces after evapor-ation a thin layer of ion conductive material on thecatalyst particles.

± and the di�usion layer which acts as the currentcollector and allows both the uniform distributionof gas within the active layer and the removing of

water to prevent the ¯ooding of the catalyst. Forthese reasons, this layer is made of a carbon cloth,carbon powder and PTFE. Hot-pressing involvingadditional parameters: temperature, time and press-

ure permits to ®nalise the MEA preparation.

The increase of the fuel cell performances impliesthe optimisation of the MEA and more especiallyof the active layers in which the electrochemical

reactions take place. Such an optimisation is poss-

ible by correlating the di�erent parameters involvedin the fabrication of the MEAs with their structural

and electrochemical properties. One aim of thispaper is to de®ne optimised preparation parametersfor the MEAs.

Among the relevant parameters of the MEAsstructure, porosity is a key one: in one hand, it hasto be high enough to permit gas di�usion up to

each catalyst particle; on the other hand, if toohigh, porosity can lower ionic and electronic con-ductivities by hindering electrolyte polymer and car-

bon particles cohesion. We intended therefore tode®ne the porosity of the active layers and to checkits dependence vs di�erent preparation parametersof our electrodes. Porosity was ®rst evidenced by

scanning electron microscopy (SEM) and thermo-porometry was the technique used to analyse itmore precisely. Another important aim of the paper

is to introduce thermoporometry and show its ade-quacy for such a study.Many experimental methods can be used to

measure the porosity of a porous solid. We canmention mercury porosimetry which measuresthe amount of mercury penetration as a func-

tion of the applied pressure. However, large de-formations of the electrode may occur duringthe measurement because of its poor mechanicalproperties. There is also the BET adsorption

measurements which permit to obtain thespeci®c surface area and the pore size distri-bution from the amount of adsorbed gas

measured at di�erent gas pressures.Thermoporometry was chosen because it is es-

pecially adapted to materials which can not

undergo compression and whose pore radii arewithin the range 1.5±150 nm. The principles of thistechnique have been previously described byQuinson et al. [16, 17]. The calorimetric analysis of

the solidi®cation of a liquid ®lling the pores permitsto obtain both the size of the pores and the overallvoid volume. The freezing temperature of the

trapped liquid depends on the radius of curvatureof the pore. Numerical relationships between thepore radius and the temperature depression have

been proposed, taking into account the interfaceenergy and the solidi®cation enthalpy. These re-lations are valid for cylindrical and spherical pores.

For benzene:

r �nm� � ÿ132=DT � 0:54 �1�

The volume of liquid freezing at a temperature T

is determined by taking into account that theapparent solidi®cation energy per volume unit ofliquid depends on the solidi®cation temperature de-

pression DT, according to the following equationfor benzene:

W � ÿ8:87 � 10ÿ3DT 2 ÿ 1:76DT �2�

S. Escribano et al.2196

EXPERIMENTAL

Preparation of the MEAs [12, 13]

Na®on1 117 membranes were treated with H2O2

and HNO3, as described in the literature [4, 8], toremove respectively organic and mineral impuritiesbefore being rinsed with water and dried under

pressure to keep them ¯at.Active layers were made of E-tek carbon sup-

ported platinum (20% Pt on Vulcan XC72R), ®ne

PTFE particles (obtained from a 50% w/w suspen-sion of PTFE in water, type GP2, ICI), andNa®on1 powder home-made using Grot and

Chadds process [18].These components were dispersed in mixtures of

50% ethanol and 50% distilled water. These disper-sions were stirred and ultrasonicated, before being

mixed, to avoid agglomerates of polymers and car-bon particles. The mixture was sprayed on twotypes of support: in the ®rst case, it was on a di�u-

sion layer, made of a carbon cloth coated with amixture of carbon powder and PTFE; in the secondcase, it was sprayed directly on to the Na®on1

membrane. In both cases, the mixture was sprayedon one side of the support while the other side washeated to evaporate the solvents.

The Membrane Electrodes Assembly was thenobtained by hot-pressing at 1408C, for a few min-utes and under 60 atm pressure, the membrane withthe active layers and additional di�usion layers.

Elaboration of the samples for thermoporometry

measurements

Since the active layer porosity is one of the keyfactor which governs the gas di�usion properties

inside the MEA, we focused our porosity measure-ments on the active layers. Samples consisted ofactive layers deposited on the electrolyte membrane

and removed from the support after a simple dryingor after a hot-pressing between two pieces of Te¯onribbon.

Thermoporometry measurements

A few tens of milligrams of the cathodic activelayer sample were used for each experiment. Twosolvents are classically used for thermoporometrymeasurements: water and benzene. Benzene was

chosen because this solvent does not cause swellingof Na®on1. Moreover, the hydrophobic productadded to the active layer prevented it from using

water. The sample and the condensate are sealed inthe sample holder and placed in the di�erentialcalorimeter. In order to keep the three phases near

equilibrium, a linear and slow decrease of tempera-ture is programmed. The cooling rate depends onthe device characteristics (inertia) and on the pore

size. In the case of our samples the cooling rate was5 or 30 K/h.The pore size distribution (dV/dr) vs pore radius

(r) plot is deduced from the thermogram by using

equations (1) and (2), and the cumulative porevolume (V) is then obtained by integrating the

curves dV/dr.

RESULTS AND DISCUSSION

In¯uence of the MEA preparation conditions on the

electrochemical results

The fuel cell performances strongly depend on

the preparation conditions of the MEAs as shownin Figs 1±3 where potential±current density (U, i)plots are given for di�erent types of electrodes.These plots were obtained with 100 cm2 MEAs

using a home-made cell operating with internal hu-midi®cation of gases [13, 19].In¯uence of the electrodes preparation method.

(In¯uence of heat-treating PTFE at 3508C). Wenoticed that even using the same support, the elec-trodes behaviour and performances could be very

di�erent depending on the preparation method. Forinstance we compared two active layers depositedon a di�usion layer: the ®rst one was made in onestep, by spraying a mixture containing all the com-

ponents, and the second one was made in twosteps. In this last case, a layer containing only thecatalyst and some PTFE was made in the ®rst step

and heat treated at 3508C, the active layer was thencompleted by spraying solubilised electrolyte in thesecond step. When considering the ®rst part of the

U vs i plots (Fig. 1), one can see that activationoverpotentials are always higher for the ®rst type ofelectrodes and it becomes even clearer when com-

paring the stabilised results. This experimentshowed that in the second case, when Na®on1 wasadded after the formation of a Pt/C-PTFE layer,more platinum was engaged in the reaction and

hence that the reconstituted Na®on1 ®lm was moree�cient in this case. It also con®rmed the bene®ciale�ect of the heat treatment at 3508C on the per-

formances stability. This treatment was in fact apriori performed to sinter the hydrophobic polymerand to remove poisoning solvents contained in the

original PTFE suspension.In¯uence of the hot-pressing at 1408C. The e�ect

of hot-pressing was studied in the case of MEAswith active layers sprayed on the membrane (Fig. 2).

The comparison of hot-pressed and non-hot-pressedMEAs clearly evidenced that hot-pressing wasnecessary to have high and stable electrochemical

performances. Without this treatment a dissolutionof the electrolyte could occur and cause thedecrease of the active layer e�ciency. In fact, hot-

pressing was necessary both to obtain good inter-face conductive properties between the membraneand the active layer and also to improve the mech-

anical properties of the cast Na®on1 ®lm byrecrystallisation [20]. It can be noticed that, in thecase of active layers sprayed on di�usion layers, thedissolution of Na®on1 could be hindered by heat

Volumic electrodes of fuel cells 2197

treating the electrodes in an oven before making the

assembly; hot-pressing was then exclusively used to

improve ionic contact between the electrodes and

the membrane.In¯uence of the composition of the active layer.

Finally, experiments were performed to observeessentially the in¯uence of the ratio of polymer elec-

trolyte contained in the cathodic active layers.

We ®rst present its in¯uence on the di�usion ofgases to the catalyst particles (Fig. 3). A larger dif-

fusion overpotential was observed for the cathodes

containing 33% of Na®on1 despite the fact that

the oxygen ¯ow used in this case was ®ve times

higher. It can be noticed that with a smaller oxygen¯ow the potential drop occurred at lower current

densities because of pores ¯ooding.

Meanwhile a systematic study was done to de®ne

the in¯uence of the composition of the active layeron the platinum e�ciency. This study consisted in

measuring by cyclic voltammetry the electrochemi-cally active surface area of the electrodes and in

testing the electrochemical performances (electrode

Fig. 1. Potential (U)±current density (i) plots for two 100 cm2 MEAs with active layers sprayed on di�usion layers (cath-

odes: 0.1 mg Pt/cm2, anodes: 0.05 mg Pt/cm2). O2/H2 pressures = 6/4 bar, T = 808C. . Active layer made in two steps:

with a heat treatment at 3508C between the deposit of mixed Pt/C and PTFE and the deposit of solubilized Na®on1. wActive layer made in one step: deposit of mixed Pt/C, PTFE and Na®on1 without any heat treatment. The performances

decreased with time before stabilising at a low level

Fig. 2. Potential (U)±current density (i) plots for two 100 cm2 MEAs with active layers sprayed on a Na®on1 117 mem-

brane (cathodes: 0.1 mg Pt/cm2, anodes: (0.05 mg Pt/cm2). O2/H2 pressures = 6/4 bar, T = 808C. . MEA hot-pressed. wMEA non-hot-pressed. The four plots show the rapid decrease in performances caused by the electrodes damaging

S. Escribano et al.2198

potential vs current density behaviour). The rough-ness factor (de®ned as the electrochemically active

surface area/geometrical surface area ratio) was

used to compare di�erent active layers. Roughness

factor vs Na®on1 ratio by weight (Fig. 4) showed

that maximum catalyst e�ciency was reached for

25% of Na®on1. To con®rm this result, weobserved the electrochemical performances of our

cathodes at high potentials (>0.7 V) because, in

this potential range, catalytic activation could be

considered as the only limiting factor (mass trans-

port limitations occurred at higher current den-sities). Figure 5 shows indeed that the current

density supplied by the cathodic electrode increased

when the amount of Na®on1 increased and reached

a maximum for a Na®on1 ratio higher than 25%.

The analysis of the cross section of the MEAs by

scanning electronic microscopy (SEM) evidenced

that the thickness of active layers, with the same

distribution of components, was proportional to

their weight. A study was intended to de®ne the in-

¯uence of thickness on platinum e�ciency. 10 mmwas found to be the maximum thickness allowing

good cell performances, in terms of ohmic drop and

di�usion, associated with an e�cient use of plati-

num.

By comparing the results of these two studies

concerning the in¯uence of the active layer (on gas

di�usion and on platinum e�ciency) we concluded

that in our operating conditions the best level of

cathodic performance was attained with 10 mmthick active layers containing: 54% Pt/C, 23%

PTFE, and 23% Na®on1 by weight and a total

amount of platinum of 0.1 mg/cm2.

Fig. 3. Potential (U)±current density (i) plots for two 100 cm2 MEAs with electrodes containing di�erent ratios of poly-

mer (cathodes: 0.1 mg Pt/cm2, anodes: (0.05 mg Pt/cm2). O2/H2 pressures = 6/4 bar, T = 1008C. . 23% of Na®on1, oxy-

gen ¯ow = 2� stoichiometric ¯ow. w 33% of Na®on1, oxygen ¯ow = 10� stoichiometric ¯ow

Fig. 4. Roughness factor vs Na®on1 ratio to the total

weight of the active layer. Results obtained by cyclic vol-

tammetry at 20 mV/s for 5 cm2 electrodes with active

layers sprayed on di�usion layers (0.1 mg Pt/cm2). The

Roughness factor is the electrochemically active surface

area/geometric surface area ratio

Fig. 5. Current density at ®xed cathodic potentials [0.7 V

(Q) and 0.8 V (.)] vs Na®on1 ratio to the total weight of

the active layer. 5 cm2 electrodes with active layers sprayed

on di�usion layers (0.1 mg Pt/cm2). O2/H2 pressures = 5/

3 bar, T = 508C

Volumic electrodes of fuel cells 2199

The previous observations evidenced the major

in¯uence of the preparation method on the electro-des behaviour. We will now discuss in detail the in-¯uence of this parameter on the porosity.

In¯uence of the MEA preparation conditions on the

porosity

By comparing the thickness measured on the pic-tures obtained by SEM with the thickness calcu-

lated using the density of the active layercomponents, we noticed an important void volumein this layer. The analysis done on many samples

showed that the total void volume was proportionalto the distribution by volume of the active layercomponents. Thermoporometry was, as previously

explained, the technique used to allow a quantitat-ive analysis of this porosity.Porosity of an optimised active layer. In our oper-

ating conditions, the optimised active layer (i.e. giv-

ing the best electrochemical results) was found to

be as follows: sprayed on to the membrane, hot-

pressed at 1408C, 10 mm thick and composed of

54% Pt/C, 23% PTFE, and 23% Na®on1 by

weight with 0.1 mg Pt/cm2.

The distribution by volume of the constituents of

an optimised active layer is given in Table 1. The

total void volume was an experimental result and

the volumic void ratio was deduced from the com-

parison with the total full volume. These ®rst results

of porosimetry showed that an important part of

the active layer (26%) had to be porous to achieve

the highest level of performance.

In Fig. 6, the pore size distribution inside this

active layer was calculated from thermoporometry

measurements. Pore radii were mostly within the

range 10±35 nm with an average radius of 15 nm.

No micropore (r< 2.5 nm) was detected.

In¯uence of the hot-pressing on the porosity. Scan-

ning electronic microscopy measurements evidenced

an important e�ect of the hot-pressing on the active

layer aspect with a noticeable change of its thick-

ness associated with a change of the overall poros-

ity. Moreover, thermoporometry experiments

evidenced an important decrease of the pore size

after such a treatment (Fig. 7).In¯uence of the composition of the active layer on

the porosity. The composition of the active layers

studied by thermoporometry are given in Table 2.

They were all loaded with 0.1 mg Pt/cm2.

When Na®on1 ratio increased, both average

pore size and size range decreased (Fig. 8). The

ratio of polymer electrolyte in the active layer

appeared to be a key parameter for the size of the

pores and the overall porosity (Fig. 9).

For the sample 5 containing 54% of Na®on1, no

pore could be detected with a usual cooling rate

(comprised between 5 and 30 K/h), a rate of 90 K/h

had to be used. As this value was too high for a

Fig. 6. Pore size distribution [dV/dr (R)] and cumulative

pore volume [V (r)] for an optimised active layer (i.e.

sprayed on to the membrane, hot-pressed at 1408C, 10 mmthick and composed of 54% Pt/C, 23% PTFE, and 23%

Na®on1 by weight with 0.1 mg Pt/cm2)

Table 1.

Determination of the volumic void ratio in an optimised cathodic active layer

Constituent Pt C PTFE Na®on1 Porosity

Ratio by weight (%) 11 43 23 23

Partial volume (mm3/g) 5 215 115 115 158

Ratio by volume (%) 1 35 19 19 26

Table 2.

Composition of the cathodic active layers studied by thermoporometry

Sample Carbon (vol.%) PTFE (vol.%) Na®on1 (vol.%)

Amount of Na®on1

(mg/cm2)

1* 48 26 26 0.2

2 55 30 15 0.1

3 52 28 20 0.15

4 48 26 26 0.2

5 30 16 54 0.4

Vol.% are expressed according to the total full volume.

*Sample 1 corresponds to an active layer non-hot-pressed.

S. Escribano et al.2200

good thermoporometry measurement, the pore size

distribution obtained had no real meaning but this

experiment allowed to evidence a small residual

porosity.

This last porosimetry result is in agreement with

the electrochemical results which showed an

increase in gas di�usion limitation when increasing

the amount of Na®on1 in the active layer (Fig. 3).

However, it can be noticed that the upper limit

determined by thermoporometry (54% of Na®on1)

was much higher than the limit observed during

fuel cell operation (33%). During operation the

water contained in the electrodes caused the swel-

ling of Na®on1 probably inducing a reduction in

pore size. As thermoporometry can be used with

materials swelling in liquid media, it should there-

fore be very interesting to perform thermoporome-

try measurements with water instead of benzene

and observe the e�ect of the electrolyte swelling on

the pore size distribution (taking into account the

water desorption which occurs during cooling [21]).

CONCLUSION

This work has allowed the optimisation ofPolymer Electrolyte Membrane fuel Cells electrodes(including supports, heat treatments, and compo-

sition of the active layer).Thermoporometry has been evidenced to be an

adequate technique for characterising the porosity

in this type of carbon/polymers composite electro-des. Thermoporometry measurements have shownthat the porosity of the active layers was in¯uenced

both by their hot-pressing and by their compo-sition. It was also evidenced that the porous struc-ture created by the platinised carbon powder could

be ®lled by the polymers added in the active layer.On the cathodic side which has been essentially

studied, best performances have been obtained withhot-pressed electrodes containing about 0.1 mg Pt/

cm2 and ratios of Na®on1 and PTFE in the activelayer of 26% by volume. This optimum compo-sition of the active layer corresponded to an im-

portant porosity (26%), large enough pores topermit the gas di�usion (average radius: ~15 nm)and thick enough Na®on1 ®lms to allow a good

ionic conductivity within the active layer.Knowing the interest of thermoporometry for

this type of study, one could now use it to study

the in¯uence of the composition on a wider range,and, as outlined before, to study the e�ect of wateron the porosity of the active layer.

ACKNOWLEDGEMENTS

We are grateful to J. F. Quinson (Laboratoire de

Chimie Applique e et de Ge nie Chimique del'Universite de Lyon I) for doing the thermoporo-metry measurements. One of us (S.E.) has received

a grant from the ``Agence de l'environnement et dela ma|Ãtrise de l'e nergie'' (Ademe) and the work wassupported by the ``Ve hicule Propre et Econome/PileaÁ Combustible'' French fuel cell program.

Fig. 7. Pore size distribution (dV/dr) for an active layer

non hot-pressed (r) and an active layer hot-pressed at

1408C (R)

Fig. 8. Pore size distribution (dV/dr) for the hot-pressed

active layers, composition of which are given in Table 2.

W 15% Na®on1 (sample 2). Q 20% Na®on1 (sample 3).

R 26% Na®on1 (sample 4).�54% Na®on1 (sample 5)

Fig. 9. Dependence of the average pore radius (R) and

void volume ratio (.) with the Na®on1 ratio to the total

full volume of the active layer

Volumic electrodes of fuel cells 2201

REFERENCES

1. K. B. Prater, J. Power Sources 37, 181 (1992).2. I. D. Raistrick, US patent 4 876 115, 1989.3. S. Srinivasan, E. A. Ticianelli, C. R. Derouin and A.

Redondo, J. Power Sources 22, 359 (1988).4. E. A. Ticianelli, C. R Derouin and S. Srinivasan, J.

Electroanal. Chem. 251, 275 (1988).5. E. A. Ticianelli, J. G. Beery and S. Srinivasan, J.

Appl. Electrochem. 21, 597 (1991).6. S. Srinivasan, A. C. Ferreira, R. Mosdale, S.

Murkejee, J. Kim, S. Hirano, S.-M. Lee, F. N. Buchi,A. J. Appleby and J. McBreen, in Proceedings of theSan Diego Fuel Cell Seminar, ed. A. Pittman and S.Conrad. Courtesy Associates, Inc., Washington DC,1994, p. 424.

7. R. Mosdale, M. Wakizoe and S. Srinivasan, inProceedings of the ECS Meeting, ed. S. Srinivasan, D.D. Mac Donald and A. C. Khandkar. San Francisco,1994.

8. M. S. Wilson and S. Gottesfeld, J. Appl. Electrochem.22, 1 (1992).

9. M. S. Wilson and S. Gottesfeld, J. Electrochem. Soc.139(2), L28 (1992).

10. M. S. Wilson, J. A. Valerio and S. Gottesfeld,Electrochim. Acta 40, 355 (1995).

11. R. Mosdale and P. Stevens, Solid State Ionics 61,251 (1993).

12. S. Escribano and P. Aldebert, Solid State Ionics 77,318 (1995).

13. S. Escribano, S. Miachon and P. Aldebert,Proceedings of the First International Symposium onNew Materials for Fuel Cell Systems, ed. O. Savadogo,P. R. Roberge and T. N. Veziroglu. Editions del'Ecole Polytechnique de Montre al, Montreal, 1995, p.135.

14. G. S. Kumar, M. Raja and S. Parthasarathy,Electrochim. Acta 40(3), 285±290 (1995).

15. S. Besse, G. BronoeÈ l, N. Tassin, Y. Naimi and A.Tounsi, in Proceedings of the First InternationalSymposium on New Materials for Fuel Cell Systems,ed. O. Savadogo, P. R. Roberge and T. N. Veziroglu.Editions de l'Ecole Polytechnique de Montre al,Montreal, 1995, p. 144.

16. C. Eyraud, J. F. Quinson and M. Brun, inCharacterization of Porous Solids, ed. K. K. Unger.Elsevier, Amsterdam, 1988, p. 295.

17. J. F. Quinson and M. Brun, in Characterization ofPorous Solids, ed. K. K. Unger. Elsevier, Amsterdam,1988, p. 307.

18. W. Grot and F. Chadds, European Patent 0 066 369,1982.

19. S. Miachon and P. Aldebert, J. Power Sources 56,31 (1995).

20. G. Gebel, P. Aldebert and M. Pineri, Macromolecules20, 1425 (1987).

21. M. Pineri, F. Volino and M. Escoubes, J. Polym. Sci.Polym. Phys. 23, 2000 (1985).

S. Escribano et al.2202