In situ modification of Nafion ® membranes with phospho-silicate for improved water retention and proton conduction

Page 1

Iw

AI

a

ARRAA

KPHPW

1

Ntmdc5wpparta

mio

0d

Journal of Membrane Science 333 (2009) 50–58

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

n situ modification of Nafion® membranes with phospho-silicate for improvedater retention and proton conduction

ravindaraj G. Kannan, Namita Roy Choudhury ∗, Naba K. Duttaan Wark Research Institute, ARC Special Research Centre for Particle and Material Interfaces, University of South Australia, Mawson Lakes, South Australia 5095, Australia

r t i c l e i n f o

rticle history:eceived 2 October 2008eceived in revised form 24 December 2008ccepted 30 January 2009vailable online 7 February 2009

eywords:roton exchange membraneybridhospho-silicate

a b s t r a c t

Proton-conducting hybrids from Nafion® and phospho-silicate networks were prepared by solvent-directed infiltration and copolymerization of 2-(methacryloyloxy)ethyl phosphate (EGMP) and3-[(methacryloyloxy)propyl]trimethoxysilane (MEMO) at different ratios in Nafion®. Fourier transforminfrared (FTIR) spectroscopy and 29Si and 31P nuclear magnetic resonance (NMR) spectroscopy confirmthe presence of hydrophilic –POH groups and Si–O–P and Si–O–Si bond formation. The hybrid membranesamples show phase-separated morphology. Scanning electron microscopic (SEM) images confirm uni-form distribution of 40–60 nm sized phospho-silicate nanoparticles in the membranes. The use of ethanolas solvent, directed the deposition of nanoparticles to the hydrophilic ionic cluster of Nafion®. The wateruptake of all the hybridized membranes is higher than that of the unmodified Nafion® membrane due to

ater retention the presence of strong hydrogen-bonded water molecules within the phospho-silicate inorganics. In gen-eral, the proton conductivity of hybridized membranes below 100 ◦C is lower than that of the unmodifiedmembrane due to restricted mobility as a result of decrease in free volume and disruption of arrangementof hydrophilic domain thereby disrupting the proton movement path. However, above 100 ◦C and in anhy-drous conditions, the hybrid membranes show increased proton conductivity and thermal stability thanthe blank Nafion®. Among the hybrid membranes, the sample with higher phosphate content exhibits

bette

higher water uptake and

. Introduction

Hydrated perfluorosulfonic acid (PFSA) membranes, such asafion®, are widely used as proton-conducting polymer elec-

rolyte membranes in fuel cells because of their excellent chemical,echanical and thermal stability in addition to their high con-

uctivities [1]. The current polymer electrolyte membrane fuelells (PEMFCs) with PFSA membranes typically operate between0 ◦C and 80 ◦C and exhibit certain disadvantages such as poorater/thermal management and less catalytic tolerance to COoisoning. These problems can be overcome if the operating tem-erature of the fuel cell is increased to intermediate temperaturebove 80 ◦C. Also, an elevated operating temperature enhances theeaction kinetics of the fuel cells. However, at such high tempera-ures, even the most durable perfluorinated membranes lose waternd hence the ionic conductivity reduces.

To enhance the water retention capacity and durability of theembranes at high temperatures, inorganic modification of exist-

ng perfluorosulfonic acid membranes using hydrophilic metalxides has been widely adopted [2–6]. Shao et al. [4,7] studied the

∗ Corresponding author. Tel.: +61 8 8302 3719; fax: +61 8 8302 3755.E-mail address: [email protected] (N.R. Choudhury).

376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2009.01.048

r proton conductivity.© 2009 Elsevier B.V. All rights reserved.

performance of PEMFC with Nafion® composites containing dif-ferent inorganic additives such as silicon oxide (SiO2), titaniumdioxide (TiO2), tungsten oxide (WO3) and SiO2/phosphotungsticacid (PWA). The performance of these modified membranes in thePEMFC, operated at 110 ◦C and 70% RH was found to be betterthan that of Nafion® membrane with an order of Nafion®/SiO2/PWA > Nafion®/SiO2 > Nafion®/WO3 > Nafion®/TiO2. Jiang et al. [1]reported the inorganic modification of Nafion® using tetraethy-lorthosilicate (TEOS) and demonstrated an increase in fuel cellperformance at low loadings (at 5%). However, the performancedecreased with an increase in inorganic content. On the contrary,Malhotra and Datta [8] showed incorporation of inorganic solidacids (phosphotungstic acid) in Nafion® resulted in improved waterretention capacity due to additional acidic sites and hence high fuelcell performance at elevated temperatures. However, higher watersolubility of the incorporated inorganic acids resulted in exces-sive swelling leading to membrane failure. Jalani et al. [9] preparedNafion®-MO2 (M = Si, Ti and Zr) nanocomposite membranes, withhigher water retention capacity than that of Nafion®. However, only

ZrO2-based hybrid membranes exhibited higher proton conduc-tivity than that of Nafion® at elevated temperatures and 40% RHconditions. This was attributed more to the increased acidity ofzirconium-based membranes than to that of silica- and titanium-based membranes. These research investigations demonstrate that

Page 2

A.G. Kannan et al. / Journal of Membrane Science 333 (2009) 50–58 51

brane

tcre

pcwtbStip[sraettgspKmits

N

Scheme 1. In situ modification of Nafion® mem

he incorporation of inorganic oxide particles along with acidicharacteristics has the potential to increase the durability, wateretention capacity and proton conduction of PFSA membranes atlevated conditions.

The phospho-silicate containing particles are particularlyromising candidates due to the presence of thermally and chemi-ally stable hygroscopic silicate along with acidic phosphate groups,hich are known to retain their proton conductivity at higher

emperatures [10–12]. Moreover, hydrogen ions more stronglyind to the non-bridging oxygen in phosphate in comparison toi–OH. This, in turn, has the potential to increase the mobility ofhe protons [13,14]. The strong interaction results in the increasen water retention capacity of the membranes at elevated tem-erature and hence can result in increased proton conductivity15–17]. Aparicio and Klein [16] synthesized Nafion®-phospho-ilicate–zirconia composite membrane using sol–gel technique andeported an improved electrochemical behavior at 130 ◦C. Theuthors [18] also synthesized a series of phospho-silicate glasslectrolyte and reported an increased proton conductivity withhe increase in P2O5/SiO2 ratio. Tung and Hwang [19] reportedhe preparation of phospho-silicate/Nafion® nanocomposite usinground phospho-silicate glasses by solution casting method, whichhowed enhanced thermal and chemical stability along with higherroton conductivity in comparison to phospho-silicate glasses.lein et al. [13] demonstrated the advantages of using infiltrationethod to prepare Nafion®-based hybrid. This method directs the

nfiltrated precursors to specific sites, where in situ polymeriza-ion takes place, which allows better control over the location, size,hape and distribution of the particles within the membrane.

In this paper, we report the synthesis and characterization ofafion®/phospho-silicate hybrid membrane by solvent-directed

using phospho-silicate by infiltration method.

infiltration and copolymerization of 2-(methacryloyloxy)ethylphosphate (EGMP) and hydrolyzed [3-(methacryloyloxy)propyl]trimethoxysilane (MEMO) into the ionic domains of Nafion® (asshown in Scheme 1). MEMO and EGMP ratios were varied to preparephospho-silicate gel with different P/Si ratio and its effect on thewater retention capacity, thermal stability and proton conductivityare evaluated.

2. Experimental

2.1. Materials

MEMO and EGMP were purchased from Aldrich, Australia andused as received. Nafion® 117 membranes with an ion exchangecapacity of 0.91 mequiv. g−1 were purchased from Aldrich, Aus-tralia. The as-received membranes were cleaned by first immersingin 3% hydrogen peroxide solution at 70 ◦C for 2 h, to remove organicimpurities, followed by rinsing with distilled water. Then, the mem-branes were immersed in 0.5 M sulfuric acid solution at 70 ◦C for2 h and washed several times with distilled water and subsequentlyimmersed in distilled water at 70 ◦C for 1 h to neutralize any trace ofacidity. Finally, the Nafion® membranes were dried in the vacuumoven prior to hybridization.

2.2. Nafion®-phospho-silicate hybrid preparation

The synthesis of the hybrid sol was carried out in a two-step reac-tion process. Initially, MEMO was hydrolyzed with ethanol/watersolution and mixed with varied amount of EGMP to prepare sols ofdifferent compositions. For instance, 10 mmol of MEMO was firsthydrolyzed in 20 mmol/30 mmol of ethanol/water mixture at room

Page 3

5 embr

tvtaio6tthfih1aapcpbc

2

pswpAMcoua

hasa

2

2bwpcb

U

2

wir55mwht

of a shoulder peak at 1696 cm . The peak at 1091 cm correspond-ing to Si–O–C bond in MEMO monomer spectrum disappears and anew broad peak appears at 1108 cm−1, indicating the condensationof silanol groups. Also, the appearance of a new band at 907 cm−1

(Si–OH stretching) in hydrolyzed MEMO spectrum confirms the

2 A.G. Kannan et al. / Journal of M

emperature. The progress of the hydrolysis reaction was monitoredia transmission Fourier transform infrared (FTIR) spectroscopy andhe reaction was deemed complete when the methoxy group peakt 2841 cm−1 disappeared completely. The reaction was completen 6 h. The resulting hydrolyzed solution was mixed with 10 mmolf EGMP (diluted with ethanol to 95%) and stirred vigorously at5 ◦C for 1 h. The prepared sol was diluted with ethanol to 90% prioro soaking the membranes. Detailed description of the synthesis ofhe sol is reported elsewhere [20]. The cleaned membrane was thenybridized using infiltration method by soaking in the diluted solor 2 min. It was followed by rinsing the soaked membrane withsopropanol to remove any physisorbed sol on the surface. Theybridized Nafion® membrane (henceforth referred as Nafion®-:1) was dried using compressed air and stored in a desiccatort room temperature for 2 weeks prior to further characterizationnd performance evaluation. Similar procedure was used to pre-are hybridized Nafion® membranes at two different MEMO:EGMPompositions namely Nafion®-3:7 and Nafion®-7:3 by using solrepared at 3:7 and 7:3 MEMO:EGMP ratios. Also, Nafion® mem-ranes were soaked in hydrolyzed MEMO and EGMP monomers asontrol samples.

.3. Spectroscopic characterization

The spectroscopic characterization of the monomers and therogress of hydrolysis reaction were monitored using transmis-ion mode FTIR spectroscopy. Sodium chloride plates were used asindow material. The synthesized hybrid was characterized using

hotoacoustic Fourier transform infrared (PA-FTIR) spectroscopy.Nicolet MagnaTM IR spectrometer (model 750) equipped with aTEC (model 300) photoacoustic cell was used. The spectra were

ollected in the mid-infrared region with 256 scans at a resolutionf 4 cm−1 and a mirror velocity of 0.158 cm s−1. Carbon black wassed as a reference and the system was purged with helium gas atflow rate of 15 cm3 s−1.

Solid-state nuclear magnetic resonance (NMR) spectra on theybrid samples were recorded on a Bruker 400 spectrometer oper-ting at 79.49 and 161 MHz for 29Si and 31P NMR spectra with atandard pulse sequence and the samples were spun at the magicngle at a frequency of 5 kHz.

.4. Water uptake measurements

Weighed dry membranes were immersed in distilled water for4 h. The membranes were removed from water, gently blottedetween tissue papers to remove surface water and weighed. Thisater uptake measurement was repeated in three different sam-les. The average values are reported. The water uptake (U) wasalculated using the weights of wet membrane (Ww) and dry mem-rane (Wd) using the following equation:

(%) = Ww − Wd

Wd× 100 (1)

.5. Thermal analysis

Thermal stability of the prepared hybrids and blank Nafion®

as investigated using thermogravimetric analyzer (TGA), TAnstruments (model 2950), at a heating rate of 10 ◦C min−1 fromoom temperature to 900 ◦C under a controlled gas flow rate of0 mL min−1. The sample was heated from room temperature to

50 ◦C in nitrogen and between 550 ◦C and 900 ◦C in oxygen. Theass of the sample used was between 10 and 12 mg and the samplesere immersed in distilled water for 24 h to achieve equilibrium

ydration prior to testing. The onset of degradation, weight loss dueo different components and residue remaining were evaluated.

ane Science 333 (2009) 50–58

Dynamic mechanical properties of the samples were measuredusing dynamic mechanical analyzer (DMA), TA Instruments (model2980) in tension mode. Samples were heated from room tempera-ture to 175 ◦C at a frequency of 1 Hz, at 0.08% strain amplitude witha programmed heating rate of 3 ◦C min−1.

2.6. Microscopic characterization

Scanning electron microscope (SEM) fitted with energy disper-sive X-ray analysis (EDAX) was used to establish the morphologyand elemental composition of the coating. The sample was coatedwith carbon to increase the conductivity prior to scanning and thesamples were examined by Philips XL30 scanning electron micro-scope operated at an accelerating voltage of 20 kV.

2.7. Proton conductivity

Conductivity measurements were performed at differenttemperatures and humidity conditions in an ESPEC SH-240 tem-perature/humidity chamber using a two-electrode (platinum)conductivity cell [21]. The proton conductivity (�) was calculatedfrom Eq. (2).

� = l

Rhw(2)

where l is the distance (cm) between the two Pt electrodes, h andw are the thickness (cm) and width (cm) of the membrane respec-tively, and R (�) is the resistance of the membrane obtained fromthe complex impedance plot measured using Solartron 1260A.

3. Results and discussion

3.1. Spectroscopic studies

Transmission FTIR was used to monitor the completion ofhydrolysis reaction of MEMO. Fig. 1 shows the spectra of MEMOmonomer and the hydrolyzed MEMO. During the hydrolysis ofMEMO, the peak at 2841 cm−1, which corresponds to the methoxygroup (–OCH3) in MEMO, disappears completely, whereas a newbroad peak appears at 3425 cm−1. The appearance of the new peakshows the formation of the silanol groups, which lead to hydrogenbonding with the carboxyl groups, as confirmed by the appearance

−1 −1

Fig. 1. Transmission FTIR spectra of MEMO monomer and hydrolyzed MEMO.

Page 4

A.G. Kannan et al. / Journal of Membr

hrw8crtd1anae

hartbepsttTwgbf

®

Fig. 2. PA-FTIR spectra of blank Nafion® and Nafion®-1:1 hybrid membranes.

ydrolysis of methoxy silyl groups to form silanol groups. Theseesults confirm the complete hydrolysis of MEMO prior to mixingith EGMP and infiltration. The bands at 1635 cm−1, 1405 cm−1 and

17 cm−1, which are present in the unreacted MEMO monomersorresponding to C C, CH2 wag, CH2 twist respectively [22–24],emain unchanged in the hydrolyzed MEMO spectrum. This showshat no significant free radical polymerization reaction takes placeuring the hydrolysis step. The peaks at 1171 cm−1, 1296 cm−1 and320 cm−1 corresponding to the ester groups of MEMO monomerre also present in the hydrolyzed MEMO spectrum. The otherew broad peak at 1023 cm−1 in the hydrolyzed MEMO spectrumppears from the C–OH group of ethanol, which is used as a homog-nizing agent during hydrolysis reaction.

PA-FTIR was used to understand the final structure of theybridized Nafion®. Fig. 2 shows the FTIR spectra of blank Nafion®

nd Nafion®-1:1 hybrid membranes. The broad peak present in theegion of 1080–1370 cm−1 is assigned to the C–F stretching vibra-ions of fluorocarbon chain of Nafion® membrane. The presence ofroad peak in the region of around 3400 cm−1 indicates the pres-nce of bonded –OH groups of sulfonic acid groups. Also, the peakresent at 2346 cm−1 corresponds to the –POH group [25]. Thishows the presence of non-bridging oxygen in phosphate func-ional group. The vibrations at 524 cm−1 and 639 cm−1 are assignedo the stretching bands of C–S and S–O groups respectively [26].

® −1 −1

he Nafion membrane shows peaks at 977 cm and 1058 cmhich are assigned to the symmetric stretching of S–O− and C–O–C

roups respectively; whereas in the case of hybrid membrane, theand in this region is broad indicating the presence of additionalunctional groups. Generally, peaks related to silicates and phos-

Fig. 3. (a) 29Si NMR spectrum of Nafion®-1:1 sample and (b

ane Science 333 (2009) 50–58 53

phates are detected in this region. This observation indicates thatthe hybrid sol has infiltrated the membrane. Since, the peaks cor-responding to S–O− and C–O–C groups overlap with the phosphateand silicate regions, 29Si and 31P CP/MAS NMR spectroscopy havebeen carried out to understand the interaction between phosphateand silicate groups and their environment.

29Si CP/MAS NMR spectrum of the hybrid-1:1 sample is givenin Fig. 3(a) and it shows two characteristic peaks at −58.9 ppmand −68.4 ppm. Usually, T2 and T3 type silicon atoms appear inthe regions of −55 to −60 ppm and −65 to −70 ppm respectively[27,28]. The number in T2 and T3 type silicon indicates the num-ber of bridging oxygen atoms per silicon, which is 2 and 3 inthe case of T2 and T3, respectively. This shows that the Si–OHgroups in hydrolyzed MEMO have either highly condensed to formR–Si–(O–Si/P)3 in case of T3 type or partially reacted to formR–Si–[(O–Si/P)2OH] in T2 type signal. The absence of any peak inthe region of T0 type silicon shows the absence of any residual non-converted MEMO in the membrane. Fig. 3(b) shows the 31P CP/MASNMR spectrum of Nafion®-1:1 hybrid. The spectrum shows twopeaks at 3.7 ppm and −27.6 ppm respectively. The major peak at3.7 ppm can be assigned to O P–(OR)(OH)2 [29] which confirmsthat the major part of phosphate containing group has infiltratedto the membrane and remained unreacted. This also confirms thepresence of –POH groups as determined from FTIR. The other peakat −27.6 ppm indicates the presence of phosphorus atom with twobridging oxygens such as Si–O–P and P–O–P bond [30]. These resultsdemonstrate the formation of phospho-silicate structures and thepresence of –POH, which can form strong hydrogen bonding withwater molecule.

3.2. Water uptake studies

The water uptake of the prepared hybrid membranes plays acritical role, since the amount and nature of water present inthe membrane greatly influence the transport properties of themembrane such as proton conductivity and water diffusion coef-ficient. The water uptake of blank Nafion® membrane and differenthybrid membranes is given in Fig. 4. The water uptakes in terms ofweight percent are 4.5%, 6.3%, 6.8%, 7.9%, 10.5% and 15.8% for blankNafion®, Nafion® hybridized with MEMO, M:E-7:3, 1:1, 3:7 hybridsand EGMP membranes respectively, with the standard deviationof ±0.5%. The water uptake of all the hybridized Nafion® mem-

branes is higher than that of the blank Nafion . This can be ascribedto the presence of strong hydrophilicity of phospho-silicate gels,which holds water by forming hydrogen bonding with them. Asexpected, the water uptake increases with the increase in P:Si ratiowith the maximum uptake registered for Nafion®-EGMP hybrid.

) 31P CP/MAS NMR spectrum of Nafion®-1:1 sample.

Page 5

54 A.G. Kannan et al. / Journal of Membr

F

Sr[b

3

mTgwtptamNdgtocwsmowa7a

Fb

ig. 4. Membrane water uptake measurement for blank Nafion® and the hybrids.

imilar observation has been made with phospho-silicate incorpo-ated sulphonated poly(2,6-dimethyl-1,4-phenylene oxide) (PPO)31,32]. This result demonstrates that the hybridized Nafion® mem-ranes have strong ability to absorb water molecules.

.3. Nature of water present and thermal stability

The nature of water present in the blank Nafion® and hybridembranes and their thermal stability were characterized using

GA and their corresponding thermograms are given in Fig. 5. Ineneral, all the membranes show initial weight loss below 260 ◦C,hich is attributed mainly to the removal of residual water within

he membrane. The unmodified Nafion® membrane loses the majorortion of absorbed water below 105 ◦C and the total water con-ent determined from the weight loss below 260 ◦C is 6.9%. Thisbsorbed water content is slightly higher than the value deter-ined by swelling experiments. The degradation of the blankafion® membrane occurs in two stages, with the initial degra-ation starting at 290 ◦C due to the degradation of sulfonic acidroups and the second stage degradation occurring in the tempera-ure range of 385–560 ◦C, which is attributed to the decompositionf the fluoropolymer backbone of the Nafion® polymer [3]. In thease of hybridized Nafion® samples, the initial weight loss due toater removal is higher than that of the unmodified samples. This

ubstantiates the increase in water uptake in hybridized Nafion®

embranes as observed from the immersion method. The amount

f water loss below 260 ◦C in the hybrid membranes increasesith the increase in EGMP content and the weight loss values

re 11.7%, 9.7%, 9.5% and 7.8% respectively for Nafion®-3:7, 1:1,:3 and Nafion®-MEMO hybrids. The water loss below 100 ◦C islmost the same as that of unmodified membrane; however, the

ig. 5. Thermogravimetric analysis curves of blank Nafion® and the hybrid mem-ranes prepared at three different compositions.

ane Science 333 (2009) 50–58

water loss above 100 ◦C, which is attributed to the strongly boundwater, is much higher than that of the unmodified membrane.This observation demonstrates that the incorporation of phospho-silicate groups in the membrane enables the membrane to bindwater strongly and hence can only be removed at higher tempera-ture.

The thermal degradation of Nafion®-MEMO hybrid is shiftedto higher temperature with the onset of degradation of sulfonicacid groups occurring at 310 ◦C. Subsequent degradation step alsooccurs at higher temperature with the maximum weight loss occur-ring at 447 ◦C. This shows that the in situ sol–gel reaction createsinorganic network, which enhances the thermal stability of thehybrid membrane. In contrast, the hybrid membranes, which con-tain EGMP, show onset of degradation at around 260 ◦C, and thesecond weight loss occurs at higher temperature in comparisonto unmodified Nafion® membrane. The onset of degradation atlower temperatures in EGMP containing hybrids can be attributedto the relative ease of degradation of P–O–C bond in phosphategroup [20,33] and the released volatile material catalyzes furtherdegradation of alkyl group in EGMP and sulfonic acid groups inNafion®. This catalyzed degradation process suggests that a spe-cific molecular contact exists between the sulfonic acid groups andinfiltrated phosphate group. This process leads to form a complexwith carbon compounds, rich in phosphorus [20,34]. This com-plex thermally insulates the hybrid thereby shifting the seconddegradation step of phosphate containing hybrids to higher tem-perature. The residue yield at 800 ◦C ranges from 4% to 6% for thehybrid membranes (with the highest and lowest yields are recordedfor Nafion®-3:7 and Nafion®-MEMO hybrids respectively), whereasthe blank Nafion® membrane is completely degraded at this tem-perature. Hence, the residue can be attributed to the amount ofinorganic content present in the hybrid membranes.

3.4. Thermomechanical investigation

DMA was used to determine the molecular dynamics and glasstransition (Tg) temperature of unmodified Nafion® and the hybrids,and their corresponding tan ı curves vs temperature are shownin Fig. 6(a). For all the membranes, the Tg value was obtainedfrom the peak of tan ı curve. Blank Nafion® membrane exhibitsa transition at ∼107.5 ◦C, which is assigned to the glass transitionof polar ionic clusters [35]. The Tg values of Nafion®-MEMO andNafion®-1:1 hybrid membranes are shifted to higher temperatureat ∼157.8 ◦C and ∼120.9 ◦C respectively. However, the Nafion®-EGMP hybrid shows a slight decrease in Tg in comparison to blankNafion® membrane. This indicates that the thermal transition ofpolar ionic clusters is modified by the establishment of their inter-action with the inorganic filler materials [36]. The increase in Tg

of the polar ionic clusters in Nafion® hybrids containing MEMOcan be attributed to the presence of silicate network. This showsthat the interaction between the silicate and ionic clusters restrictsthe molecular motion of polar clusters thereby increasing the Tg ofthe membranes. As the amount of phosphate group in the hybridincreases, the cluster transition of the hybrid membrane is shiftedto lower temperatures due to the inclusion of flexible phosphateester (P–O–C) group, which is known to act as a plasticizer, withlower Tg [37]. These results are also in agreement with the TGAresults, which show an increase in the degradation temperature forthe hybrid membranes containing MEMO and a slight decrease inthe degradation temperature for the Nafion®-EGMP membrane.

Fig. 6(b) shows the normalized plot of storage modulus as a

function of temperature for the blank Nafion® and the hybridmembranes. The moduli for all the samples decrease with tem-perature as the samples undergo transition from glassy to rubberystate. In general, the hybrid samples show higher stiffness than theblank Nafion®, indicating that the inorganic additives contribute

Page 6

A.G. Kannan et al. / Journal of Membr

Frb

sfiNoiht

ig. 6. (a) Tan delta curves of Nafion® and hybrid membranes in the temperatureange of 25–175 ◦C and (b) normalized storage modulus vs temperature plots for thelank Nafion® and the hybrid membranes.

ignificantly to the materials’ mechanical strength. The unmodi-ed Nafion® shows highest decrease in the modulus, whereas the

afion®-MEMO sample remains more mechanically stable than thether samples. Nafion®-3:7 hybrid displays a steady decrease in

ts modulus values up to 160 ◦C and the Nafion®-1:1 hybrid showsigher strength below 120 ◦C but it decreases rapidly above thisemperature as it passes from the glassy to the rubbery regime.

Fig. 7. (a) SEM image of cleaned blank Nafion® (b) its corresp

ane Science 333 (2009) 50–58 55

Thus, the hybrid membranes containing higher amount of silicatenetwork are more mechanically stable. This result is in correlationwith the TGA results and the change in ionic cluster transition tem-perature. This observation demonstrates that the incorporation ofinorganic particles contributes to a decrease in segmental mobilityof ionic clusters, which results in an increase in mechanical strengthof the composite membranes.

3.5. Morphological investigation

To further understand the interaction between the membraneand the inorganic phase, morphological characterization of theblank Nafion® and the hybrid membranes were carried out. Themorphology of the blank Nafion® as shown in Fig. 7(a) is uni-form and smooth. In contrast, the surface of the Nafion®-1:1hybrid membrane [Fig. 8(a)] shows phase-separated morphologywith near-uniform distribution of oxide particles in one phaseand almost negligible amount in the other phase. The phase-separated morphology arises from the use of polar solvent suchas ethanol, which is selectively compatible with only hydrophilic(ionic domain) part of the copolymer whereas the fluorinatedpolymer backbone in Nafion® is non-compatible with this sol-vent. The uniform distribution of hybrid nanoparticles in the ioniccluster can be attributed to the preferential migration of solvent-directed hydrolyzed MEMO to the nanometer size ionic clusterspresent in Nafion® membrane. The clusters serve as miniature reac-tors in which the condensation of sol–gel reactants takes place.DMA results also indicate a shift in the cluster transition temper-ature indicating that the inclusion of phospho-silicate particles inthose domains, as observed from SEM. Similar results have beenpublished previously [38]. Solvents can play a significant role indirecting the inorganic particles within the membrane and deter-mining the final morphology of the membranes. Since ethanol, apolar solvent is used for membrane preparation; this directs the

inorganic particles to the ionic clusters thereby forming this phase-separated morphology. The particles are approximately 40–60 nmin size and are uniformly distributed. This, in comparison with theliterature [4,7], yields much smaller particles, which is a criticalparameter in composite membranes. The smaller particles along

onding EDAX spectrum and (c) cross-sectional image.

Page 7

56 A.G. Kannan et al. / Journal of Membrane Science 333 (2009) 50–58

; (b) i

wmrtcqhmTssoiNtcSp

3

h(g

TP

DM

NNNNN

Fig. 8. (a) SEM image of Nafion®-1:1 hybrid surface morphology

ith its near-uniform distribution significantly improve the ther-al and mechanical stability of the polymers. Apart from this, the

eduction in particle size enhances the surface to volume ratio ofhe particles thereby facilitating easy proton hopping. This, in effect,an increase the proton conductivity of the membranes [39]. Also,ualitative compositional analysis of the blank Nafion® and Nafion®

ybrids were carried out to determine the elements present in theembrane. EDAX spectrum of blank Nafion® is shown in Fig. 7(b).

he spectrum shows the presence of fluorine, carbon, oxygen andulphur. In hybridized Nafion® membrane [Fig. 8(b)] the inten-ity of oxygen peak increased significantly showing the presencef organic methacrylate and inorganic oxides. The cross-sectional

mage of the hybrid membrane in Fig. 8(c) in comparison to blankafion® [shown in Fig. 7(c)] indicates the presence of cavities due

o the removal of particulate structures from the membrane duringross-sectioning. The EDAX spectrum confirms the presence of P,i, C, O peaks and a lower intensity F peak than Nafion® indicatinghospho-silicate inorganic particles reside in the membrane.

.6. Proton conductivity

The proton conductivities of blank Nafion® and the Nafion®

ybrids at 30 ◦C and 80 ◦C at two different relative humiditiesRHs) are measured and the results are given in Table 1. Theeneral trend in blank Nafion® membrane is that the proton con-

able 1roton conductivity results of unmodified Nafion® and its hybrid membranes.

escription (Nafion®-EMO:EGMP)

30 ◦C, 90% RH(mS cm−1)

80 ◦C, 30% RH(mS cm−1)

80 ◦C, 90% RH(mS cm−1)

afion®-7:3 16 0.38 1.9afion®-1:1 21 0.17 2.1afion®-3:7 35 0.40 9.0afion®-EGMP 28 0.42 7.6afion® 80 8.00 99.0

ts corresponding EDAX spectrum and (c) cross-sectional image.

ductivity increases with the increase in temperature under humidconditions; whereas at low humidity, the conductivity decreaseswith the increase in temperature. For instance, the unmodifiedNafion® membrane exhibits proton conductivity of 80 mS cm−1 and99 mS cm−1 at room temperature and 80 ◦C respectively at 90% rel-ative humidity. In general, the proton migration in Nafion®-basedmembranes is primarily by the Grotthuss mechanism [7,14,40]. Inthis mechanism, the excess proton hops from the H3O+ donor siteto any neighboring acceptor water molecule. The incorporation ofinorganics to the polymer matrix will have two opposing effects onproton conduction of the hybrid membrane [41]. The first effectis retaining higher amount of water through hydrogen bondingand the second effect is blocking/reducing the proton migrationchannel. Here, we notice a decrease in proton conductivity withinorganic modification, which shows that the disruption of pro-ton movement path is more significant than the increase in wateruptake capacity. The incorporation of inorganics in the membranemodifies the arrangement of hydrophilic phases in the membranethereby hindering the continuous channel for proton movement.Also, non-conducting inorganic inclusions with organic methacry-late group remains embedded in the channels connecting the clus-ters, thereby restricting the vehicular transport of protons from onesite to the other [42]. Further, the excess water molecules are likelyto be involved in hydrating the incorporated phospho-silicatesrather than providing a good connectivity within the hydratedhydrophilic domains [43]. In addition, the interaction of watermolecules with phospho-silicate reduces the water diffusion coeffi-cient. However, the conductivity values at lower and intermediatetemperatures with 90% humidity are comparable to those of ourother phospho-silicate hybrid work [43]. Another significant factorwhich contributes for the lower proton conductivity is the concen-

tration of phosphate groups in the hybrid membrane to achievepercolation threshold. In order to achieve good conductivity, a highconcentration of phosphate group along with no constraint onaggregate formation is necessary to form bulky, hydrogen-bondedaggregates which promote a high-degree of self-dissociation [44].

Page 8

embrane Science 333 (2009) 50–58 57

Rht

dwtahA1ado[N[ticdgttathwchwcTtag

mcNitmtd8tihpnN

Fi

A.G. Kannan et al. / Journal of M

elatively lower concentration of phosphate groups along with theindering of aggregate formation by methacrylate groups also con-ributes for lower proton conductivity of the hybrid membranes.

In correlation with the water uptake measurements, the con-uctivity of the hybrid membranes at room temperature increasesith the increase in EGMP content (up to 70% EGMP con-

ent in the reaction mixture). The highest proton conductivitymong the hybridized membranes is observed for Nafion®-3:7ybrid samples (35 mS cm−1) at 30 ◦C and 90% RH conditions.t the same conditions, the proton conductivities of Nafion®-:1, 7:3 and Nafion®-EGMP samples are 21 mS cm−1, 16 mS cm−1

nd 28 mS cm−1 respectively. The proton conductivity at this con-ition is significantly higher than the phosphonic acid graftedrganic–inorganic hybrid membrane as given in the literature45]; however, this is lower than the conductivity values ofafion®/phospho-silicate hybrids reported by Tung and Hwang

19]. The significant factor which influences the proton conduc-ion is the availability of mobile protons and the ease with whicht forms. The generation of mobile protons depends on the asso-iation/dissociation of the hydrogen bond, which results in moreelocalized protons [32,40,46]. The availability of more –POHroups with the increase in EGMP content along with its tendencyo form strong hydrogen bonding with absorbed water enhanceshe number of mobile protons available for proton transfer. Also,t 80 ◦C and 30% RH, Nafion®-3:7 shows higher proton conductionhan the other hybrid membranes. The higher conductivity at lowumidity and high temperature may be attributed to the fact thatater forms strong hydrogen bonding with P–OH group than sili-

ate network. This hydrogen bonding reduces the loss of water atigher temperatures, thereby retaining more water in comparisonith other hybrids at low humidity. Hence, this hybrid shows higher

onductivity at lower humidity conditions than other membranes.his observation demonstrates that phospho-silicate network inhe hybrid membrane improves the thermal stability of the polymernd the phosphate groups enhance the water retention capacity andeneration of mobile protons.

Further conductivity measurements are carried out at inter-ediate temperature range from 100 ◦C to 150 ◦C and anhydrous

onditions and the proton conductivities of blank Nafion® andafion®-3:7 hybrid membranes are given in Fig. 9. The conductiv-

ty of unmodified Nafion® decreases linearly with the increase inemperature from 100 ◦C to 150 ◦C, whereas the Nafion®-3:7 hybrid

embrane shows sharp increase in conductivity up to 140 ◦C andhen decreases slightly at 150 ◦C. The drastic decrease in the con-uctivity of blank Nafion® membrane at anhydrous condition above0 ◦C can be attributed to the loss of water. In comparison to this,he Nafion®-3:7 hybrid membrane shows higher conductivity atntermediate temperatures (as high as 0.27 mS cm−1 at 140 ◦C and

as slightly lower Tg than Nafion®). This can be attributed to theresence of more hydrogen bound water in hybrid membrane, asoticed from TGA. The slight decrease in the conductivity of theafion®-3:7 hybrid membrane at 150 ◦C shows that the hybrid

ig. 9. Proton conductivity of blank Nafion® and Nafion®-3:7 hybrid membranes atntermediate temperatures and anhydrous condition.

Fig. 10. Photographic images of (a) unmodified Nafion® and (b) Nafion®-3:7 hybridmembranes taken after conductivity measurements up to 140 ◦C.

membrane starts losing water above 140 ◦C. However, the hybridmembrane remains transparent and colorless even after severalhours of testing at 150 ◦C and anhydrous conditions; whereas dis-coloration (brown color) of the unmodified Nafion® samples isobserved at these conditions (Fig. 10). This observation demon-strates that the Nafion®-3:7 hybrid membranes are more thermallystable than the blank Nafion® membranes and can be used for inter-mediate temperature membranes. Thus, it is clear that water ismore tightly held in the hybrid film than blank Nafion® membrane.Also, the pore structure and free volume of these membranes haveimportant roles in determining proton conductivity of these mate-rials, as the structures of pore water (average amount, extent ofhydrogen bond and their connectivity) affect the mobility of proton.

4. Conclusions

In summary, we have demonstrated the in situ modifica-tion of Nafion® membranes via sol–gel reaction and infiltrationmethod, with phospho-silicate gels directed to ionic cluster regionsthrough acid–base interactions. The hybrid membranes show bet-ter thermomechanical properties than the unmodified Nafion®

membranes and increased water uptake and water retention atintermediate temperatures than the blank Nafion® membrane,due to the strong interaction with the functional inorganic addi-tives. The water uptake increases with the increase in P/Si ratio inthe hybrid membranes. The lower proton conductivity of hybridmembranes below 80 ◦C is attributed to the disruption of pro-ton conduction path by disrupting the arrangement of hydrophilicdomains. At intermediate temperatures, hybrid membrane showshigher proton conductivity and thermal stability than blankNafion® membrane. Among the hybrid membranes, Nafion®-3:7membranes show promise as proton-conducting membranes atintermediate temperatures.

Acknowledgments

The authors gratefully acknowledge the financial support of the

Australian Research Council’s Special Research Centre and the Dis-covery project for carrying out this work, Dr. Z. Shi of National FuelCell Centre, Canada for carrying out some of the proton conductiv-ity tests and Prof. S. Holdcroft, Simon Fraser University, Canada, forhelpful discussion of the project and Dr. Pramoda Pallathadka, Insti-

Page 9

5 embr

to

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

8 A.G. Kannan et al. / Journal of M

ute of Materials Research and Engineering, Singapore, for carryingut the NMR work.

eferences

[1] R. Jiang, H.R. Kunz, J.M. Fenton, Composite silica/Nafion® membranes pre-pared by tetraethylorthosilicate sol–gel reaction and solution casting for directmethanol fuel cells, J. Membr. Sci. 272 (2006) 116.

[2] K.T. Adjemian, R. Dominey, L. Krishnan, H. Ota, P. Majsztrik, T. Zhang, J. Mann,B. Kirby, L. Gatto, M. Velo-Simpson, J. Leahy, S. Srinivasan, J.B. Benziger, A.B.Bocarsly, Function and characterization of metal oxide-nafion composite mem-branes for elevated-temperature H2/O2 PEM fuel cells, Chem. Mater. 18 (2006)2238.

[3] M. Lavorgna, L. Mascia, G. Mensitieri, M. Gilbert, G. Scherillo, B. Palomba,Hybridization of nafion membranes by the infusion of functionalized siloxaneprecursors, J. Membr. Sci. 294 (2007) 159.

[4] Z.-G. Shao, H. Xu, M. Li, I.M. Hsing, Hybrid nafion-inorganic oxides membranedoped with heteropolyacids for high temperature operation of proton exchangemembrane fuel cell, Solid State Ionics 177 (2006) 779.

[5] H. Tang, Z. Wan, M. Pan, S.P. Jiang, Self-assembled nafion-silica nanoparticlesfor elevated-high temperature polymer electrolyte membrane fuel cells, Elec-trochem. Commun. 9 (2007) 2003.

[6] V. Baglio, A.S. Arico, A.D. Blasi, V. Antonucci, P.L. Antonucci, S. Licoccia, E.Traversa, F.S. Fiory, Nafion-TiO2 composite dmfc membranes: physico-chemicalproperties of the filler versus electrochemical performance, Electrochim. Acta50 (2005) 1241.

[7] Z.-G. Shao, P. Joghee, I.M. Hsing, Preparation and characterization of hybridnafion-silica membrane doped with phosphotungstic acid for high temperatureoperation of proton exchange membrane fuel cells, J. Membr. Sci. 229 (2004)43.

[8] S. Malhotra, R. Datta, Membrane-supported nonvolatile acidic electrolytesallow higher temperature operation of proton-exchange membrane fuel cells,J. Electrochem. Soc. 144 (1997) L23.

[9] N.H. Jalani, K. Dunn, R. Datta, Synthesis and characterization of Nafion®-MO2

(M = Zr, Si, Ti) nanocomposite membranes for higher temperature PEM fuelcells, Electrochim. Acta 51 (2005) 553.

10] S. Subianto, N.R. Choudhury, N.K. Dutta, Palladium-catalysed phosphonationof SEBS block copolymer, J. Polym. Sci., Part A: Polym. Chem. 46 (2008)5431.

11] M. Schuster, T. Rager, A. Noda, K.D. Kreuer, J. Maier, About the choice of theprotogenic group in PEM separator materials for intermediate temperature, lowhumidity operation: a critical comparison of sulfonic acid, phosphonic acid andimidazole functionalized model compounds, Fuel Cells 5 (2005) 355.

12] A. Matsuda, T. Kanzaki, K. Tadanaga, M. Tatsumisago, T. Minami, Medium tem-perature range characterization as a proton conductor for phosphosilicate drygels containing large amounts of phosphorus, Electrochim. Acta 47 (2001)939.

13] L.C. Klein, Y. Daiko, M. Aparicio, F. Damay, Methods for modifying protonexchange membranes using the sol–gel process, Polymer 46 (2005) 4504.

14] K.D. Kreuer, Proton conductivity: materials and applications, Chem. Mater. 8(1996) 610.

15] A. Matsuda, T. Kanzaki, K. Tadanaga, M. Tatsumisago, T. Minami, Proton conduc-tivities of sol–gel derived phosphosilicate gels in medium temperature rangewith low humidity, Solid State Ionics 154–155 (2002) 687.

16] M. Aparicio, L.C. Klein, Synthesis and characterization of nafion/60SiO2–30P2O5–10ZrO2 sol–gel composite membranes for PEMFCs, J. Electrochem. Soc.152 (2005) A493.

17] M. Nogami, K. Miyamura, Y. Abe, Fast protonic conductors of water-containingP2O5–ZrO2–SiO2 glasses, J. Electrochem. Soc. 144 (1997) 2175.

18] M. Aparicio, J. Mosa, A. Duran, Hybrid organic–inorganic nanostructured mem-branes for high temperature proton exchange membranes fuel cells (PEMFC), J.Sol–Gel Sci. Technol. 40 (2006) 309.

19] S.-P. Tung, B.J. Hwang, Preparation and characterization of Nafion®/hydratedphosphor-silicate hybrid membranes for the direct methanol fuel cell, Fuel Cells7 (2007) 32.

20] A.G. Kannan, N.R. Choudhury, N.K. Dutta, Synthesis and characterization ofmethacrylate phospho-silicate hybrid for thin film applications, Polymer 48(2007) 7078.

21] X. Zhong, S. Chaojie, A. Bernhard, N. Titichai, S. Zhiqing, Z. Jiujun, H. Steven,Discrepancies in the measurement of ionic conductivity of PEMs using two- andfour-probe ac impedance spectroscopy, J. Electrochem. Soc. 153 (2006) E173.

[

[

ane Science 333 (2009) 50–58

22] S.K. Medda, D. Kundu, G. De, Inorganic–organic hybrid coatings on poly-carbonate. Spectroscopic studies on the simultaneous polymerizations ofmethacrylate and silica networks, J. Non-Cryst. Solids 318 (2003) 149.

23] J. Ding, W. Shi, Thermal degradation and flame retardancy of hexaacry-lated/hexaethoxyl cyclophosphazene and their blends with epoxy acrylate,Polym. Degrad. Stab. 84 (2004) 159.

24] H. Liang, W. Shi, Thermal behaviour and degradation mechanism of phosphatedi/triacrylate used for uv curable flame-retardant coatings, Polym. Degrad. Stab.84 (2004) 525.

25] A. Bozkurt, W.H. Meyer, J. Gutmann, G. Wegner, Proton conducting copolymerson the basis of vinylphosphonic acid and 4-vinylimidazole, Solid State Ionics164 (2003) 169.

26] C. Heitner-Wirguin, Infra-red spectra of perfluorinated cation-exchanged mem-branes, Polymer 20 (1979) 371.

27] Y.-Y. Yu, C.-Y. Chen, W.-C. Chen, Synthesis and characterization oforganic–inorganic hybrid thin films from poly(acrylic) and monodispersedcolloidal silica, Polymer 44 (2003) 593.

28] R. Joseph, S. Zhang, W.T. Ford, Structure and dynamics of a colloidal silica-poly(methyl methacrylate) composite by 13C and 29Si MAS NMR spectroscopy,Macromolecules 29 (1996) 1305.

29] A. Matsuda, T. Kanzaki, M. Tatsumisago, T. Minami, Comparison of structureand proton conductivity of phosphosilicate gels derived from several kinds ofphosphorus-containing compounds, Solid State Ionics 145 (2001) 161.

30] N.J. Clayden, A. Aronne, S. Esposito, P. Pernice, Solid state NMR study of phos-phosilicate gels, J. Non-Cryst. Solids 345–346 (2004) 601.

31] W. Lu, D. Lu, J. Liu, C. Li, R. Guan, Preparation and characterization of sul-fonated poly(phenylene oxide) proton exchange composite membrane dopedwith phosphosilicate gels, Polym. Adv. Technol. 18 (2007) 200.

32] D. Lu, W. Lu, C. Li, J. Liu, J. Xu, Proton-conducting composite membranes derivedfrom poly(2,6-dimethyl-1, 4-phenylene oxide) doped with phosphosilicate gels,Solid State Ionics 177 (2006) 1111.

33] S. Zhu, W. Shi, Thermal degradation of a new flame retardant phosphatemethacrylate polymer, Polym. Degrad. Stab. 80 (2003) 217.

34] K. Mequanint, R. Sanderson, H. Pasch, Thermogravimetric study of phosphatedpolyurethane ionomers, Polym. Degrad. Stab. 77 (2002) 121.

35] I.D. Stefanithis, K.A. Mauritz, Microstructural evolution of a silicon oxide phasein a perfluorosulfonic acid ionomer by an in situ sol–gel reaction. 3. Thermalanalysis studies, Macromolecules 23 (1990) 2397.

36] B. Mecheri, A. D’Epifanio, E. Traversa, S. Licoccia, Effect of an ORMOSIL-basedfiller on the physico-chemical and electrochemical properties of nafion mem-branes, J. Power Sources 169 (2007) 247.

37] R.-J. Jeng, J.-R. Wang, J.-J. Lin, Y.-L. Liu, Y.-S. Chiu, W.-C. Su, Flame retardant epoxypolymers using phosphorus-containing polyalkylene amines as curing agents,J. Appl. Polym. Sci. 82 (2001) 3526.

38] Q. Deng, Y. Hu, R.B. Moore, C.L. McCormick, K.A. Mauritz, Nafion/ormosil hybridsvia in situ sol–gel reactions. 3. Pyrene fluorescence probe investigations ofnanoscale environment, Chem. Mater. 9 (1997) 36.

39] V. Ramani, H.R. Kunz, J.M. Fenton, Effect of particle size reduction on the con-ductivity of Nafion®/phosphotungstic acid composite membranes, J. Membr.Sci. 266 (2005) 110.

40] M. Eikerling, A.A. Kornyshev, A.M. Kuznetsov, J. Ulstrup, S. Walbran, Mechanismsof proton conductance in polymer electrolyte membranes, J. Phys. Chem. B 105(2001) 3646.

41] G. Ye, C.A. Hayden, G.R. Goward, Proton dynamics of nafion and nafion/SiO2

composites by solid state NMR and pulse field gradient NMR, Macromolecules40 (2007) 1529.

42] C. Li, G. Sun, S. Ren, J. Liu, Q. Wang, Z. Wu, H. Sun, W. Jin, Casting nafion-sulfonated organosilica nano-composite membranes used in direct methanolfuel cells, J. Membr. Sci. 272 (2006) 50.

43] M.K. Mistry, N.R. Choudhury, N.K. Dutta, R. Knott, Z. Shi, S. Holdcroft, Novelorganic–inorganic hybrids with increased water retention for elevated temper-ature proton exchange membrane application, Chem. Mater. 20 (2008) 6857.

44] H. Steininger, M. Schuster, K.D. Kreuer, A. Kaltbeitzel, B. Bingol, W.H. Meyer, S.Schauff, G. Brunklaus, J. Maier, H.W. Spiess, Intermediate temperature protonconductors for PEM fuel cells based on phosphonic acid as protogenic group: aprogress report, Phys. Chem. Chem. Phys. 9 (2007) 1764.

45] S. Li, Z. Zhou, H. Abernathy, M. Liu, W. Li, J. Ukai, K. Hase, M. Nakanishi, Synthesisand properties of phosphonic acid-grafted hybrid inorganic-organic polymermembranes, J. Mater. Chem. 16 (2006) 858.

46] E. Spohr, P. Commer, A.A. Kornyshev, Enhancing proton mobility in polymerelectrolyte membranes: lessons from molecular dynamics simulations, J. Phys.Chem. B 106 (2002) 10560.