007) 1248–1255www.elsevier.com/locate/ssi

Solid State Ionics 178 (2

Functionalized zeolite A–nafion composite membranes for directmethanol fuel cells

Xiao Li a, Edward P.L. Roberts a,⁎, Stuart M. Holmes a, Vladimir Zholobenko b

a School of Chemical Engineering and Analytical Science, The University of Manchester, PO Box88, Sackville Street, Manchester M60 1QD, United Kingdomb Department of Chemistry, Keele University, ST5 5BG, Staffordshire, United Kingdom

Received 11 April 2007; received in revised form 13 June 2007; accepted 20 June 2007

Abstract

A series of composite membranes based on zeolite A and Nafion 117 have been fabricated for direct methanol fuel cells. The external surface ofzeolite A has been modified to enhance the interface bonding between inorganic zeolite crystals and Nafion ionomer. The modified zeolite sampleshave been characterised using a combination of thermal analysis and spectroscopic techniques. Methanol permeability of Nafion 117 could bereduced by as much as 86% by incorporating functionalised zeolite NaA crystals into the membrane. The effect of different functionalisation level isdiscussed.© 2007 Elsevier B.V. All rights reserved.

Keywords: Zeolite NaA; Nafion; Composite membrane; Functionalisation; FTIR spectroscopy; Thermal analysis; Proton conductivity; Methanol permeability; Directmethanol fuel cell

1. Introduction

Fuel cell technology has been rapidly advancing over therecent years. In particular, direct methanol fuel cells (DMFC)have demonstrated a good deal of potential for practicalapplications. In terms of performance, it is essential for theDMFC membrane to have a high proton conductivity and a lowpermeability to methanol. Nafion, a membrane material manu-factured by DuPont, is still the most widely used for DMFC.Although Nafion has a high proton conductivity, it suffers fromsevere methanol crossover, which restricts the performance of thefuel cell. Indeed, methanol can easily diffuse through Nafionunder the concentration gradientwhich occurs in an operating fuelcell. In addition, protonated methanol or protons solvated bymethanol can be transported through the membrane by means ofelectro-osmotic drag [1].Methanol crossover leads to a significantdecrease in both the fuel efficiency of the cell and in the opencircuit voltage due to the presence of the “mixed potential” causedby the methanol oxidation at the cathode. In order to overcomethese problems, a wide range of composite membranes based on

⁎ Corresponding author. Tel.: +44 161 306 8849.E-mail address: Edward.Roberts@manchester.ac.uk (E.P.L. Roberts).

0167-2738/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.ssi.2007.06.012

Nafion or other ion-conducting polymers have been prepared andevaluated. Materials which have been combined with the ionconducting polymer include calcium phosphate [2], montmoril-lonite (MMT) [3,4], chabazite and clinoptilolite [5] and TiO2

nano-powder [6]. These materials show a range of usefulproperties, however, few authors have studied the influence ofsurface functionalisation of the particles, or the effect of particlesize.

In this study Nafion–zeolite NaA composite membraneshave been prepared in order to suppress methanol crossover.Zeolite NaA has been selected for this study since its low Si/Alratio provides a hydrophilic (organophobic) character thatallows for preferential adsorption of water, and therefore,preferential proton transfer, whereas methanol is excluded fromthe zeolite channels. Zeolite NaA membranes have been usedfor the separation of alcohols including methanol from water [7]suggesting that this zeolite would be very suitable for thepreparation of composite membranes.

However, zeolite NaA is known to be unstable in strongmineral acids, and it is not known whether the acidic conditionsin the hydrated Nafion will lead to significant breakdown of thezeolite crystals. In addition, poor interfacial compatibilitybetween zeolite crystals and glassy polymer matrix has been

1249X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

reported by a number of authors [8–12] who attributed thiseffect to the presence of pinholes between the zeolite particlesand Nafion. The size of these opening is in the nanometer range[13], which facilitates the methanol transport. In this work, weattempt to overcome this problem by functionalising the surfaceof the zeolite to make it more compatible with the polymermatrix. This has been achieved by grafting organic chains ontothe inorganic surface using silane-based coupling agents such as3-aminopropyltrimethoxysilane (APTS) [14]. The functiona-lised zeolite NaA has been characterised in detail and the effectof functionalizing the zeolite surface on the properties of aNafion–zeolite composite membrane has been evaluated. Inaddition, this study aims to demonstrate that the degree of thezeolite modification can be quantified using a combination ofthermal analysis and infrared spectroscopy. Indeed, it isimportant to establish the most favourable degree of modifica-tion in order to achieve good crosslinking between the polymerand the zeolite in combination with the optimum properties ofthe zeolite surface to minimise the membrane methanolpermeability and maximise its proton conductivity.

2. Experimental

2.1. Chemical modification method

In order to functionalise the surface of the zeolite with APTSa procedure, based on the method reported by Plueddemann andothers [14,15], was developed. Two parent samples of zeoliteNaAwere utilised in this work, NaA zeolite with a micron-rangeparticle size of 2–3 μm (NaA-M) and zeolite NaA with asubmicron partical size of ∼300 nm (NaA-S). In a typicalpreparation, a mixture of 1.0 g zeolite NaA powder, 1 ml ofAPTS and 30 ml of dichloromethane (DCM) was stirred for24 h at room temperature. After the reaction, the excess reagentwas eliminated by repeated centrifugation (6000 rpm), decant-ing and re-dispersion by ultrasonication in ethanol. Therecovered solid was dried at 100 °C for 24 h (the preparedmaterials were denoted as NaA-F1M or NaA-F1S). In order toinvestigate the effect of the degree of functionalisation, theprocedure was repeated for some samples (these materials weredenoted as NaA-F2M or NaA-F2S). The obtained white APTS-functionalised zeolite powders were used for the compositemembrane fabrication.

2.2. Characterisation

X-ray diffraction (XRD) patterns of modified and unmod-ified zeolite crystals were obtained to monitor possible changesin crystallinity during the functionalisation process. For theinfrared characterisation, self-supporting zeolite discs (∼10 mg,13-mm diameter) were placed into a Pyrex vacuum cell fittedwith BaF2 windows. The cell was installed inside the samplecompartment of an FTIR spectrometer Nicolet Protégé 460. Thecell was attached to a greaseless vacuum system and the celltemperature was externally controlled between 30 and 500 °C.Following the sample disc activation at 30 °C and 10−5 Torr for1 h, the infrared spectrum of the zeolite was collected. The cell

temperature was then steadily increased to the desiredtemperature (the temperature ramp was 10 °C/min) and thestepwise desorption of various species was monitored by IRspectroscopy to the maximum temperature of 500 °C, averaging200 scans at an instrument resolution of 2 cm−1 and a spectralrange of 4000–1300 cm−1. Infrared spectra were analysed(including integration, subtraction and determination of peakpositions) using specialised Nicolet software, Omnic 4.1.

Thermal analysis of the samples (both thermal gravimetricalanalysis, TGA, and differential scanning calorimetry, DSC) wasundertaken in flowing nitrogen or air with a temperature rampof 10 °C/min to the maximum of 800 °C using a STA-1500Rheometric Scientific instrument.

2.3. Membrane preparation

Composite membranes were produced using a solution castingmethod. Nafion was purchased from Aldrich as a 5 wt.% solutionin amixture of lowmolecular weight alcohols andwater.Methanol(99.99%), ethanol (99.7%) and N,N-dimethylformamide (DMF)(+99%) were used without further purification. Deionized waterwas used for the preparation of all aqueous solutions.

Composite mixtures were prepared by combining fivevolumes of the Nafion solution, five volumes of methanol/ethanol mixture (15 vol.% methanol solution in ethanol), threevolumes of DMF [5], and the desired amount of zeolite NaApowder. Two different zeolite powders with particle sizes of 2–3 μm (Valfor® CP100, PQ corporation) and ∼300 nm(Nanoscape, Germany) were evaluated. The mixture wasdispersed by mechanical stirring and ultrasonication for 1–3 h, depending on the concentration of zeolite NaA.

The milky mixture was poured into a glass Petri dish andheated in a vacuum oven at 80 °C for 20 h and 150 °C for 4 h inorder to evaporate the solvents. After cooling, the membrane wasgently separated from the dish. The thickness of the membranewas controlled by the amount of the composite mixture used.

2.4. Proton conductivity measurements

The proton conductivity of the membranes was determinedby AC impedance using a two-electrode measuring cell. Thismethod is suitable for measuring the conductivity across themembrane, as appropriate for fuel cell applications [16]. Theproton conductivity of the membranes was measured in a boricacid solution at pH 4.7 at a constant temperature of 35 °C.Platinum electrodes (2mm diameter) were placed on each side ofthe membrane and a pressure of 1.9 N cm−2 was applied toensure a good contact between the electrodes and the membrane.

Prior to the testing, each membrane was boiled in 5% H2O2

solution for 1 h and in boric acid solution (pH=4.7) for 1 h, andthen rinsed in deionizedwater, followed by equilibration for 48 h at35 °C. The AC impedance spectrum of membranes was obtainedusing an Autolab PGSTAT30 Frequency Response Analyzer(FRA) at three different locations on each membrane. Measure-ments were made with a perturbation voltage of 5 mV in thefrequency range from 100 Hz to 1 MHz at 35 °C. The resistancevalue was determined from the high frequency intercept of the

Fig. 1. Schematic diagram of the experimental system for methanol permeability measurements.

1250 X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

impedance with the real axis [2,17,18]. The thickness of eachmembrane was measured using a micrometer.

The proton conductivity can be calculated as follows:

r ¼ dRS

ð1Þ

where σ, R, S and d denote the proton conductivity, theresistance of the membrane, the area of the electrode, and thethickness of the membrane, respectively.

2.5. Methanol permeability measurements

The methanol permeability of each membrane was measuredusing a diffusion cell. The diffusion cell consisted of two

Fig. 2. The comparison of XRD diffraction pattern of zeolite A before or after thefunctionalized zeolite NaA, zeoAF2 — functionalized zeolite NaA where the functi

identical compartments separated by a membrane, with an areaof 13.67 cm2, as shown in Fig. 1. Two litres of a 1 M methanolfeed solution was circulated through one of the compartments at35 °C, while 100 ml of deionised water was circulated throughthe other compartment. Samples of the water were analysed formethanol concentration by using gas chromatography (GC).

The methanol flux across the membrane reached a steadyvalue after an equilibration period and the methanol permeabil-ity was calculated as follows [19]:

b ¼ d½CBðtÞ�dt

� VBLACA

ð2Þ

where CA (1 mol dm−3) and CB are the methanol concentrationin the feed and permeate solutions, respectively; A is the

surface functionalization (zeoA — unfunctionalized zeolite NaA, zeoAF1 —onalization process was carried out twice).

Fig. 3. TGA-DSC analysis of the parent NaA-S, and functionalised NaA-F1S and NaA-F2S zeolites carried out in a flow of nitrogen. All samples were equilibrated at∼40% relative humidity.

Table 1A summary of the TGA data for the parent NaA-S and functionalised NaA-F1Sand NaA-F2S zeolites samples

Weight loss Weight loss Weight loss Totalwater loss

Totalorganics loss20–250 °C 250–650 °C 650–800 °C

% % % % %

NaA 19.5 4.5 0.5 24.5 –NaA-F1S 17 10.5 0.5 21.5 6NaA-F2S 15.5 13.5 0.5 20 9

1251X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

membrane area; L is the thickness of the membrane; VB thevolume of the permeate solution; and t is the permeation time.

3. Results and discussion

3.1. Characterisation of zeolite materials

XRD patterns obtained for the parent and modified zeolitesamples have shown no appreciable differences (Fig. 2),indicating that their structure and crystallinity remain un-changed following the functionalisation procedure.

TGA and DSC data for NaA, NaA-F1 and NaA-F2 zeolitesare presented in Fig. 3. For the parent NaA sample, two regionsof weight loss are identified, a rapid 19.5% weight loss between20 and 250 °C is followed by a steady 5% reduction in thesample mass between 250 and 800 °C (Table 1). The former canbe attributed to the loss of weakly adsorbed water, whereas thelatter should be assigned to the removal of strongly bound waterand, possibly, dehydroxylation of the external zeolite surfaceaccording to the following reaction scheme.

2SiOH→Si–O–Si þ H2O

For the modified samples NaA-F1 and NaA-F2, the results aresomewhat different. Firstly, the amount of weakly bound waterdesorbed at 20–250 °C is reduced for themodifiedmaterials to 17and 15.5%, respectively. Secondly, the steady removal of water at250–800 °C is accompanied by a considerable additional weightloss between 250 and 650 °C (see Table 1). These differences canbe explained by the effects of the surface modification on thezeolite properties. Indeed, functionalised zeolite is less hydro-philic, and therefore, fewer water molecules are adsorbed on theexternal surface of the zeolite— this water is weakly bound and isremoved at relatively low temperatures. The strongly adsorbedwater molecules are bound to the sodium cations within thezeolite cages. The internal pore structure is not affected by the

modifying agent, and consequently, the amount of water removedat higher temperature remains approximately the same for all threesamples. The additional weight loss between 250 and 650 °C forNaA-F1 and NaA-F2 samples should be attributed to thedecomposition of the organic species attached to the surfacefollowing the modification procedure.

These conclusions are confirmed by the DSC resultsobtained under flowing nitrogen, as the weight decrease at250–650 °C coincides with the presence of broad exothermicpeaks, which are observed for NaA-F1 and NaA-F2 samples butnot for the parent NaA zeolite. This is further supported by DSCexperiments carried out under flowing air (Fig. 4). Indeed, theintensity of the exothermic peak between 250 and 650 °Cincreases, compared to the flowing nitrogen conditions, as thedecomposition of the organic surface species is accompanied bytheir oxidation in the presence of oxygen. At the same time,there is no evidence of exothermic reaction taking place on theparent NaA zeolite, for which TGA and DSC traces obtainedunder flowing nitrogen or air are almost identical. In addition,integration of the DSC data yields the values of 4 kJ/g and 6 kJ/g for the heat released between 250 and 650 °C for NaA-F1 andNaA-F2 respectively, which are in good agreement with theweight loss figures for these two materials. Overall, we canconclude that the amount of modifying agent attached to thezeolite surface corresponds to 6% of the sample weight forNaA-F1 and 9% NaA-F2, and that TGA-DSC experiments

Fig. 4. TGA-DSC analysis of NaA-F2S zeolite in nitrogen and in air.

1252 X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

provide quantitative information on the degree of modificationof the zeolite surface.

FTIR spectroscopy has been utilised to gain a betterunderstanding of the surface properties of the APTS functio-nalised zeolite NaA. Fig. 5 shows infrared spectra obtained forthe parent zeolite NaA following its pretreatment at a range oftemperatures under vacuum. Although there is little change inthe appearance of the spectra of the O–H stretching vibrationscollected for NaA between 3800 and 3000 cm−1, the region ofbending O–H vibrations reveals the presence of at least twotypes of water molecules in the sample. Weakly adsorbed water,i.e. physically adsorbed species on the zeolite surfacecharacterised by the absorption band of the bending O–Hvibration around 1660 cm−1, is mostly removed upon sampleactivation between 30 and 200 °C. Water molecules stronglyadsorbed on Na+ cations, probably in two different locations,are responsible for the appearance of two bands at relatively lowfrequencies, 1487 and 1434 cm−1. The intensity of these bandsis steadily decreasing as the activation temperature is increased

Fig. 5. FTIR spectra of zeolite NaA-S obtained following its activation atvarious temperatures: (1) 30 °C, (2) 100 °C, (3) 200 °C, (4) 300 °C, (5) 400 °C,(6) 500 °C. Spectra are offset for clarity.

from 200 to 500 °C indicating slow desorption of the stronglybound H2O molecules at the elevated temperatures.

Infrared spectra of the functionalised zeolite NaA sample(Fig. 6) demonstrate a considerable change in the surfacecharacteristics of the zeolite following its modification withAPTS. In addition to the absorption bands of water molecules, anumber of additional peaks are observed. Absorption peaks at3290–3170 and 1605–1580 cm−1 can be attributed to N–Hstretching and bending vibrations, and bands at 2930–2860 and1445–1410 cm−1 to C–H stretching and bending vibrationsrespectively. Similarly to the zeolite NaA sample, physicallyadsorbed water is removed from the modified zeolite followingits activation up to 200 °C, whereas the intensities of the C–Hand N–H vibrations remain largely unchanged at thesetemperatures. As the pretreatment temperature is increased to300–500 °C, the spectra provide clear evidence of thedecomposition and gradual removal of the organic speciesfrom the zeolite surface. However, even after the sampleevacuation at 500 °C, the intensity of the C–H vibrationsremains significantly high indicating that a considerable amount

Fig. 6. FTIR spectra of functionalised zeolite NaA-F2S obtained following itsactivation at various temperatures: (1) 30 °C, (2) 100 °C, (3) 200 °C, (4) 300 °C,(5) 400 °C, (6) 500 °C. Spectra are offset for clarity.

Fig. 8. Proton conductivity (a) and methanol permeability (b) of zeolite NaA/Nafion composite membranes prepared with 5 wt.%, 10 wt.%, and 15 wt.% ofzeolite NaAwith a particle size of∼300 nm (5%S, 10%S and 15%S respectively).

Fig. 7. Proton conductivity (a) and methanol permeability (b) of commercialNafion 117 (CN), recast Nafion (RN) and zeolite NaA–Nafion compositemembranes prepared with 5 wt.%, 10 wt.% and 15 wt.% of zeolite NaAwith aparticle size of 2–3 μm (5%M, 10%M and 15%M respectively).

1253X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

of organic species is trapped on the surface or within the narrowpores of zeolite A under these conditions. Indeed our TGA datademonstrate that ∼2% of the total sample weight loss can beattributed to the removal of organics between 500 and 800 °C,which corresponds to one third of the organic speciesintroduced during the modification procedure.

Overall, the presence of two types of water molecules inzeolite NaA and the changes in the intensities of the absorptionbands in the infrared spectra of both the parent and modifiedsamples are in full agreement with TGA-DSC results describedpreviously.

3.2. Composite membrane performance: Effect of zeoliteloading and crystal size

The methanol permeability of a range of composite mem-branes has been compared to a commercial Nafion 117membraneand to a pure Nafion cast membrane prepared by the sameprocedure as the composite membranes. The data presented inFig. 7a demonstrate that all of the composite membranes show

lower proton conductivity than Nafion, and the proton conduc-tivity decreased with increasing zeolite loading. This is expectedsince zeolite A is a relatively poor proton conductor.

At the same time, the methanol permeability of themembrane is significantly reduced by the addition of only5 wt.% of the parent zeolite A as a 2–3 μm powder (sample 5%NaA-F1M, Fig. 7b). In agreement with the data reported byNunes et al. [20] and Holmberg et al. [21], the methanolpermeability of the composite membrane is found to increasewith the zeolite loading. It is unlikely that this increase is causedby a poorer distribution of zeolite crystals in the polymer athigher loadings, since even if a zeolite layer has formed in themembrane due to settlement, the membrane with a thickerzeolite layer would act as a better methanol barrier. Therefore,one can suggest that the increase in methanol permeability withzeolite loading is caused by poor adhesion between the zeolitecrystals and the polymer matrix. With zeolite loadings over5 wt.%, gaps around the crystals may lead to an enhancedmethanol permeation, while the proton conductivity is stillsteadily decreasing.

Fig. 9. Proton conductivity (a) and methanol permeability (b) of zeolite NaA/Nafion composite membranes prepared with 5 wt.% functionalised zeolite NaAwith a particle size of 2–3 μm (5%FM), 5 wt.% and 10 wt.% functionalisedzeolite NaAwith a particle size of ∼300 μnm (5%FS and 10%FS respectively)and 5 wt.% and 10 wt.% functionalised zeolite NaA with a particle size of∼300 nm where the functionalisation process was carried out twice (5%F2S).

1254 X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

The properties of membranes prepared using smaller zeoliteparticles (∼300 nm in diameter) are shown in Fig. 8. Thecomposite membranes with submicron zeolite crystals (S-membranes) exhibit higher proton conductivity, particularlysamples 5%NaA-F1S and 10%NaA-F1S, than those withmicron-size zeolite particles (M-membranes). This improve-ment can be attributed to a more homogeneous distribution ofzeolite crystals throughout the composite membranes. However,all composite membranes exhibit a considerable methanolcrossover with a slightly higher methanol permeability observedfor S-membranes. This may be explained by the formation ofpinholes which is much more pronounced for composites withfiner crystals. Indeed, if the interfacial adhesion between thezeolite and Nafion is poor, reducing the size of crystals wouldpromote the formation of pinholes and enhance the methanolcrossover, as the smaller crystals have a greater external surfacearea in contact with the Nafion matrix. Provided that thegeneration of pinholes is eliminated or inhibited in composite

membranes containing submicron zeolite particle, which maybe achieved by improving the zeolite–Nafion surface affinity,these S-membranes could demonstrate a combination of highproton conductivity and low methanol permeability.

3.3. Composite membrane performance: Effect of surfacemodification of zeolite

The data presented in Fig. 9 demonstrate that functionalisationof zeolite crystals leads to a decrease in methanol permeability.As expected, surface modification of zeolite A should improvethe interface bonding between the polymer and the zeolite, andtherefore, suppress the formation of pinholes. Moreover,composite membranes containing 10 wt.% of functionalisedzeolite have a lower methanol permeability than membranes with5 wt.% of the functionalised zeolite, which can be attributed tothe improved interface adhesion between zeolite and Nafion.It is also worth noting that surface functionalization of thefiner rystals has a more significant effect on the membranepermeability.

The second functionalization (sample 5%NaA-F2S) reducedthe methanol permeability to around half that of the zeolitefunctionalized only once (sample 5%NaA-F1S). This suggeststhat the adhesion between the zeolite crystals and the Nafionmatrix is enhanced by further functionalisation. However, thisimprovement comes at the price of a reduction in the protonconductivity to half that of the 5%NaA-F1S compositemembrane. Further work is needed to optimise the level offunctionalisation and zeolite loading to maximise fuel cellperformance.

The incorporation of functionalized zeolite generally leads toa reduction in proton conductivity of the membrane comparedto unmodified zeolite. This can be attributed to the basicity ofNH2-group of the modifying agent attached to the zeolitesurface. The APTS molecules are too large to get into thecavities of NaA, and therefore, only the external surface of thezeolite is modified. The surface alkylamino-groups may interactwith SO3H-groups in the Nafion matrix; hence, some of thesulphonic groups can no longer dissociate and release hydrogenions. In addition, mobile hydrogen ions may be “trapped” byreacting with the basic NH2-groups. Both effects would reducethe rate of proton transfer in the membrane. Our data indicatethat the surface of the inorganic fillers should be neutral oracidic in order to retain or enhance the proton conductivity ofthe composite membranes, which is consistent with the reportfrom Arico et al. [22]. At the same time, the interaction betweenbasic NH2-and acidic SO3H-groups should result in improvedzeolite distribution and mechanical properties of the compositesowing to the enhanced crosslinking between the zeolite crystalsand Nafion polymer. As the functionalization takes place on theexternal surface of zeolite, composite membrane with micron-sized modified zeolite particles show only a marginal decreasein proton conductivity, while the effect is more prominent forthe submicron-sized zeolite which has a much larger externalsurface area, and therefore, a higher proportion of the functionalgroups attached to the surface. The membrane containing moreextensively functionalized zeolites, i.e. 5 wt.%F2S sample,

1255X. Li et al. / Solid State Ionics 178 (2007) 1248–1255

shows the lowest methanol permeability, while its protonconductivity remains comparable to other samples.

In conclusion, the composite membranes tested in this workhave demonstrated good performance under realistic workingconditions despite a number of potential problems which maycompromise the performance, and which may have to beaddressed in future research. For instance, it is known thatzeolite NaA is not stable in concentrated solutions of strongmineral acids, and therefore, its structure may collapse duringthe membrane manufacturing. However, our results indicatethat under the chosen preparation conditions the structuralintegrity of the parent and modified zeolites is sufficiently highto considerably improve the membrane performance, e.g. tosuppress the methanol permeability. It is also known that NaA isa hydrophilic zeolite, and one may expect therefore, that theinteraction between the zeolite crystals and polar groups of theNafion polymer would be reasonably strong eliminating theneed for zeolite modification. Our results, however, suggest thatfunctionalisation of the zeolite surface is necessary to increasethe adhesion between zeolite particles and the polymer. Theaffinity of the external surface of the zeolite towards the Nafionpolymer may be insufficient to produce a pinhole-free com-posite membrane using the unmodified parent zeolite. It is alsopossible that the functionalistion provides some protection tothe surface of the zeolite from acid attack in the presence ofhydrated Nafion. In any case, the performance of the mem-branes prepared using surface-modified materials is indicativeof the much improved adhesion between the polymer and thezeolite, which is reflected by the decreased methanol crossover.As has been mentioned, the proton conductivity of the com-posite membrane is considerably reduced. The latter problemmay be addressed by optimising the surface modification pro-cedure or by utilising less basic functionalising agents.

4. Conclusion

This study demonstrates that silane-based coupling agentscan be used to improve the interface compatibility betweenzeolite crystals and Nafion. Our data reveal that the incorpora-tion of zeolite NaA into DMFC membranes results in a decreasein their methanol permeability. This effect is particularlysignificant for composites containing functionalised zeoliteparticles, which is attributed to a better adhesion between theorganic matrix and the inorganic filler and the inhibitedformation of pinholes. Disappointingly, this is accompaniedwith a noticeable reduction in the membrane proton conductiv-ity. The best functionalised membranes tested in this work showmethanol permeability which is an order of magnitude lower

than that of Nafion, whereas their proton conductivity is reducedby a factor of four. It appears that the next step in improvement inthe performance of DMFC membranes should come from amodification procedure that would allow retention of the highproton conductivity of Nafion. This may be accomplished byutilising different functionalising agents, probably with lesspronounced basic properties, which would not restrict the protontransfer to a significant extent. In addition, the optimum degreeof modification should be established to ensure good cross-linking between the polymer and the zeolite in combination witha minimal concentration of basic groups on the zeolite surfacewhich may trap mobile protons. This should be achieved viaprecise quantitative characterisation of the modified zeolites andcomposite membranes utilising a combination of spectroscopictechniques, thermal analysis and electron microscopy. The suc-cessful application of the former two techniques has beendemonstrated in this work.

References

[1] C.H. Rhee, H.K. Kim, H. Chang, J.S. Lee, Chem. Mater. 17 (2005) 1691.[2] Y. Park, Y. Yamazaki, Solid State Ion. 176 (2005) 1079.[3] M.K. Song, S.B. Park, Y.T. Kim, K.H. Kim, S.K. Min, H.W. Rhee,

Electrochim. Acta 50 (2004) 639.[4] R.F. Silva, S. Passerini, A. Pozio, Electrochim. Acta 50 (2005) 2639.[5] V. Tricoli, F. Nannetti, Electrochim. Acta 48 (2003) 2625.[6] V. Baglio, A.S. Arico, A. Di Blasi, P.L. Antonucci, F. Nannetti, V. Tricoli,

V. Antonucci, J. Appl. Electrochem. 35 (2005) 207.[7] D. Shah, K. Kissick, A. Ghorpade, A.R. Hanna, D. Bhattacharyya,

J. Membr. Sci. 179 (2000) 185.[8] Y. Li, H. Guan, T. Chung, S. Kulprathipanja, J. Membr. Sci. 275 (2006) 17.[9] J.M. Duval, J.B. Kemperman, M.H.V. Mulder, G. Desgrandchamps, C.A.

Smolders, J. Appl. Polym. Sci. 54 (1994) 409.[10] D. Metin, F. Tihminlioglu, D. Balkose, S. Ulku, Composites: Part A

35 (2004) 23.[11] G. Alberti, M. Casciola, Annu. Rev.Mater. Sci. 33 (2003) 129.[12] J.P. Boom, I.G.M. Punt, H. Zwijnenberg, R. de Boer, D. Bargeman, C.A.

Smolders, H. Strathmann, J. Membr. Sci. 138 (1998) 237.[13] R. Mahajan, W.J. Koros, Polym. Eng. Sci. 42 (2002) 1420.[14] E.P. Plueddemann, Silane Coupling Agents, Plenum Press, New York,

1982.[15] Y. Li, H.M. Guan, T.S. Chung, S. Kulprathipanja, J. Membr. Sci. 275 (2006)

17.[16] B.S. Pivovar, Y. Wang, E.L. Cussler, J. Membr. Sci. 154 (1999) 155.[17] H.Y. Chang, C.W. Lin, J. Membr. Sci. 218 (2003) 295.[18] G. Zhang, Z. Zhou, J. Membr. Sci. 261 (2005) 107.[19] Y. Woo, S.Y. Oh, Y.S. Kang, B. Jung, J. Membr. Sci. 220 (2003) 31.[20] S.P. Nunes, B. Ruffmann, E. Rikowski, S. Vetter, K. Richau, J. Membr.

Sci. 203 (2002) 215.[21] B.A. Holmberg, H. Wang, J.M. Norbeck, Y. Yan, Polym. Preprints

24 (2004) 45.[22] A.S. Arico, V. Baglio, A. Di Blasi, E. Modica, P.L. Antonucci, V.

Antonnucci, J. Power Sources 128 (2004) 113.