Anisotropy of structure and transport properties in sulfonated polyimide membranes

Journal of Membrane Science 214 (2003) 31–42

Anisotropy of structure and transport propertiesin sulfonated polyimide membranes

Jean François Blachota, Olivier Diata, Jean-Luc Putauxb,Anne-Laure Rolleta,1, Laurent Rubatata, Cécile Valloisc,

Martin Müllerd,2, Gérard Gebela,∗a Département de Recherche Fondamentale sur la Matière Condensée/SI3M/Polymères Conducteurs Ioniques,

CEA-Grenoble, 17 rue des martyrs, 38054 Grenoble Cedex 9, Franceb CERMAV-CNRS, BP 53, 38041 Grenoble Cedex 9, France

c Institut Européen des Membranes, CC 047 place Eugène Bataillon, Montpellier 34095, Franced ESRF, 6 rue J. Horowitz, BP 220, 38043 Grenoble Cedex, France

Received 25 April 2002; received in revised form 28 October 2002; accepted 29 October 2002

Abstract

An original analysis of the structure of ionomer films along the three dimensions using micro small-angle X-ray scatter-ing (microSAXS) technique is reported. While the in-plane structure appeared as isotropic, a significant SAXS anisotropywas observed in the transverse direction, revealing a multiscale structural anisotropy from the molecular level to severaltens of nanometers. Using transmission and scanning electron microscopy (TEM and SEM, respectively), a layered struc-ture was identified up to micrometer scales. This structural anisotropy was used to interpret the conductivity data as afunction of the ion content. Moreover, the diffusion coefficients of the counterions, were determined along the planar andtransverse directions using pulse field gradient spin echo-NMR (PFGSE-NMR) in order to correlate structure and transportproperties.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Sulfonated polyimides; Anisotropy; Micro-SAXS; Electron microscopy; Conductivity; Pulse field gradient NMR

1. Introduction

Ion-conducting membranes are increasingly usedin industry as separators for electrochemical applica-

∗ Corresponding author. Tel.:+33-438783046;fax: +33-438785691.E-mail address: [email protected] (G. Gebel).

1 Present address: Center de Recherche sur les Materiaux aHautes Temperatures, CNRS, 1D avenue de la Recherche Scien-tifique, 45071 Orleans Cedex 2, France.

2 Present address: Institut für Exp. und Angwandte Physik, Uni-versität Kiel, Leibnitzstrasse 19, 24118 Kiel, Germany.

tions, such as electrolyse, electrodialysis or fuel cells[1,2]. These materials present a nanophase separationinducing complex structures. They are usually de-scribed as composed on the one hand of a polymericmatrix and on the other hand of connected ionic do-mains where the ionic conduction takes place. Thisphase separation about few nanometers scale wasmainly observed using small-angle X-ray scattering(SAXS) owing to the detection of a scattering max-imum (called the ionomer peak)[1–4]. This peakis usually attributed to the distance between ionicdomains (≈50 Å), corresponding roughly to their

0376-7388/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0376-7388(02)00522-7

32 J.F. Blachot et al. / Journal of Membrane Science 214 (2003) 31–42

diameters since the percolation of these ionic do-mains is necessary to explain the high level of con-ductivity [3]. The ion-conducting membranes alwaysappeared as isotropic materials whatever the scale ofobservation except when deformations are appliedduring membrane formation.

One of the major characteristics of these mem-branes is their ability to swell when soaked in polarsolvents such as water or ionic solutions. The wateruptake (swelling), which depends on the ion contentand varies exponentially with temperature,[5] mod-ifies the membrane transport properties (conductivityand selectivity). It induces also a modification of theSAXS spectra (shift of the ionomer peak to smallerangles and large increase of the scattered intensity),which reveals a structural evolution at the nanometerscale. This hydration leads also to large geomet-ric variations or to internal stress increase for con-strained membranes that can be very troublesome forapplications.

For most of the applications and especially for fuelcells, perfluorosulfonated ionomer membranes, suchas Nafion® from Du Pont de Nemours are consideredas references in terms of lifetime and performances[6,7]. However, their prohibitive cost, low glass tem-perature transition and low methanol crossover inhibittheir large commercial developments. To overcomethese problems, numerous laboratories work on the de-velopment of alternative membranes generally basedon sulfonated polyaromatic materials (polyetherke-tones, polysulfones, polybenzimidazoles,. . . ) [7]which, however, exhibit lower chemical stability andionic conductivity compared to Nafion®. The con-ductivity of alternative membranes can be improvedincreasing the ion content but at the expense of themechanical properties due to an excessive swellingratio. Among these new membranes, sulfonated poly-imide (sPI) membranes based on naphthalenic struc-tures and rigid polymer chains were shown to be verypromising materials for electrochemical applicationsin terms of ionic conductivity, selectivity, stabilityand cost [8–12]. Their chemical structure (blockcopolymers composed of ionic sequences spaced byhydrophobic sequences), their rigidity and the veryhigh glass transition temperature of the polymer chainconfer some peculiar swelling and transport proper-ties. The water swelling is constant when expressedas the number of water molecules per ionic group

over a large range of ion-exchange capacities whilethis value usually diverges for high ion contents withstandard membranes. Moreover, the geometric expan-sion, mainly observed along thickness, does not cor-respond to the simple addition of water and polymervolumes[11]. As well, no major structural reorgan-isation was observed using SAXS and SANS duringthe swelling of these membranes. These results sug-gested the existence of both porosity and structuralanisotropy[11].

The first objective of the present work was toconfirm structural anisotropy for these rigid poly-mers and to determine the characteristic scales ofthis anisotropy. On a nanometer scale, we studiedthe structure using microSAXS experiments. Theanisotropy was confirmed at a scale of a few tensof nanometers analysing ultrathin transversal slicesusing TEM and at the micrometer scale, thanks tothe observation of the fracture of swollen samples us-ing scanning electron microscopy (SEM) after freezedrying.

The second objective consisted in the study ofthe influence of the structural anisotropy on thetransport properties. The transverse ionic conduc-tivity as a function of the ion content was anal-ysed. Pulse field gradient spin echo-proton NMR(PFGSE-NMR) was performed along the three sam-ple directions in order to study the influence of thestructural anisotropy on the self-diffusion coefficientsat a microscopic scale (ca from ms to s in timescale).

2. Experimental part

2.1. Samples

The synthesis of the polymer solution as well ascasting, acidification and washing procedures used toprepare the membranes, were previously described[13,14]. As a summary, the polymer synthesis is per-formed in two steps using a sulfonated monomer inthe first step and a neutral diamine (the oxydianiline:ODA) in the second step leading to block copolymerswith controlled average lengths.X andY are the aver-age lengths of the sulfonated and neutral sequences,respectively.

J.F. Blachot et al. / Journal of Membrane Science 214 (2003) 31–42 33

Studies were conducted on series characterised byan ion exchange capacity (IEC) of 1.26 meq./g (X/Y =30/70) and 1.98 meq./g (X/Y = 50/50) and with dif-ferentX units, 3, 5, 7 and 9. This type of sPI polymer isqualified as rigid if we compare the relative free move-ment of the molecules around the mean axis of thepolymer backbone with the others sPI which have beenstudied[13]. The water uptake of the different poly-mers was previously reported for either membranesimmersed in water or equilibrated with different vaporpressures[11,15,16]. The average water uptake is 19water molecules per sulfonate ionic group when im-mersed in liquid water and it is independent of both theion content and the length of the polymer sequences.

2.2. Scattering technique

SAXS experiments were performed with using amicrofocus camera (microSAXS) at the EuropeanSynchrotron Radiation Facility (ESRF) on ID13beamline.3 The spectrometer was built to producehighly focused X-ray beam on the sample. In thepresent experiments, the beam size on the sample was10�m× 10�m (width at half maximum). The exper-imental set-up allows to locate the detector (a CCDcamera) at a distance around 50 cm from the sampleand to use a beam stop small enough to get reliabledata at small-angles despite the high beam divergenceafter the optics (qmin = 10−2 Å−1 along vertical di-rection withq defined as 4π sin(θ /2)/λ, whereλ is theincident wavelength andθ the scattering angle). Thesamples which consist of small stripes of polymerfilm (10 mm long and 1 mm wide) are suspended in aspecial support, mounted in a cell, which allows bothrelative humidity from 10 to 98% in relative humidity(using saturated salt solutions) and temperature con-trol (mainly for thermal stability). The membraneswere studied on a Cs+ neutralised form to increase

3 http://www.esrf.fr/expfacilities/ID13/index.html.

the electron density contrast between the polymer ma-trix and the ionic domains. Using an internal rotat-ing system, the samples were studied in two differentconfigurations called parallel (the X-ray beam is per-pendicular to the membrane plane but probes the struc-ture parallel to the membrane surface) and transverse(the X-ray beam is parallel to the membrane surfacesbut probes the structure in perpendicular plane). Scat-tering profiles are obtained by averaging the scatteredintensity as a function of the transfer momentumqover a narrow angular sector, centered on the whisheddirection. The background spectrum and more espe-cially the parasitic scattering at very small-angle ofscattering (<10−2 Å−1) fluctuates in time over sev-eral hours. Although thermal drifts and beam positioninstabilities are minimised on the ID13 beamline atthe ESRF, slight fluctuations in position of the highlyfocalised beam create nevertheless some strong slitsscattering through the different pinholes of the opticalset-up. Even if several background files are recordedduring the investigation, the background subtractionat very low scattering angles becomes a critical op-eration below 10−2 Å−1, i.e. when the scattering sig-nal is of the order of the parasitic scattering intensity.This limitation inq is worth in the parallel configura-tion for which the scattering intensity from<100�mthick samples is rather weak compared to the back-ground. In the transverse geometry the sample scat-tering is sufficiently high to resolve the structure until600 Å. On the other hand, corrections at large anglesare applied by subtracting a constant value extractedfrom the background files.

2.3. Electron microscopy

Tractions at a rate of 2 min−1 were applied at 25◦Cto submerged acidic swollen membranes using an In-stron 4301 tester. Fracture surfaces of freeze-fracturedmembranes were observed at room temperature using

http://www.esrf.fr/exp_facilities/ID13/index.html


a JEOL GFC-1100E scanning microscope operated at8 kV, in the secondary electron mode.

sPI membranes in acidic form were embedded inEpon resin. 0.1�m thin cross-sections were obtainedat room temperature using a MT5000 Sorval ultrami-crotome equipped with a 45◦ diamond knife. Differ-ent cross-sections were prepared by varying the knifeorientation with respect to the sPI film. The sectionswere deposited on 200 mesh uncoated TEM coppergrids and observed using a Philips CM200 transmis-sion microscope operated at 80 kV. Due to the highresistance of the material to the electron beam at roomtemperature, micrographs were recorded at a magnifi-cation of 50,000×.

2.4. Transport measurements

The conductivity measurements were performedwith hydrated membranes in acidic form by using animpedance spectroscopy. The membranes were settledin a cell equipped with mercury electrodes enhancingthe electrical contact with the membrane. A Paar 273as potentiostat and a Solartron SI1255 as frequencyanalyser were used for impedance measurements. Thespectra were recorded over a frequency range fromfew hertz to several hundred of kilohertz dependingon the samples. The value of the membrane resis-tance was determined when the imaginary part of theimpedance was is equal to zero.

The self-diffusion coefficients of tetramethyl am-monium cations and water were determined using 1HPFGSE NMR. The experiments were performed on aVarian 500 MHz spectrometer at 298 K using a BPPLED PFG sequence (bipolar-pulse-pairs longitudinal-eddy-current-delay pulsed-field-gradient)[17]. Thegradient strength was ranged from 0 to 30 G/cm witha pulse duration of 4.5 ms. The high resolution ofthe spectrometer permits to differentiate the signalsarising from the ammonium ions, the water insideand the water outside of the membrane. The decreaseof the NMR line intensity versus gradient appears asa mono-exponential attenuation over one decade inintensity, which allows an accurate determination ofdiffusion coefficients and indicates that the diffusionof only one species is observed. Their dependencewere measured as a function the delay between gradi-ent pulses in the NMR sequence,, in order to studythe space explored by cations and to get information

about the existence of a restricted geometry. As ob-served for Nafion® membranes[18], the diffusion ofwater molecules does not depend on in the exploredrange probably because the explored distance is largerthan the characteristic length of the porous structure.So we studied mainly the diffusion of tetramethylam-monium counterions, containing a large number ofprotons that are non-exchangeable with heavy water(used as solvent in the experiments). These ions werechosen as probes because of their large mobility inthe ionic phase—narrow lines allowing the measure-ment of their attenuation over a large range with agood signal to noise ratio—and because their dif-fusion coefficient permits to extract information onthe characteristic lengths of the structure. The pulsedgradients were applied along the NMR tube axis. Theexperiments were performed for two samples orienta-tions: either a piece of membrane (4 cm× 4 cm) wasrolled and introduced in the NMR tube (parallel con-figuration) or more than 300 circular pieces (4 mmdiameter) were stacked in order to get a 3 cm thicksample with the pulsed field gradient perpendicularto the membrane plane (transverse configuration).

3. Results and discussion

As observed in previous SAXS and SANS ex-periments[11], the bidimensional�SAXS patternsof rigid sPI are isotropic in parallel configurationwhatever the polymer composition and the processof casting used to prepare the membrane from thepolymer solution. An example is shown in (Fig. 1afor a membrane withX = 5 and IEC= 1.98 meq./g).The distribution of the scattered intensity is identicalwhatever the azymuthal angle in the detector plane.

Fig. 2apresents the variation of the scattering in-tensity along the parallel directions for different sul-fonated polyimide (sPI) samples varying the sequencelength for a given IEC. As it was shown in[11] on ra-dially averaged scattering curves, these scattering pro-files are characterised first by a broad shoulder at lowq-values. This shoulder is clearly visible forX = 3and much less for largeX due to an intense scatteringupturn at smaller angles, not really accessible usingmicroSAXS technique. This upturn is usually relatedto large scales heterogeneities (about 0.1�m) in ionicpolymers [19]. Concerning the shoulder, it can be


Fig. 1. Two-dimensional�SAXS spectra obtained for a sPI membrane (X = 5 IEC = 1.26 meq./g) equilibrated at 15% of relative humidityusing the a microfocus camera in parallel (A) and transverse (B) configurations.� corresponds to the azymuthal angle in thex–z plan.

assimilated to an ionomer peak, the main scatteringfeature for ionomer films, which originates from con-structive interferences from scattering objects. Indeed,Fig. 3 shows, for a slightly more flexible polymerchain (mAPI as defined in[14]) that the amplitudeof this ionomer shoulder varies as a function of thecounterion contrast without changing in position, afeature which was also studied for other similar sPIsystems[20]. These objects can be considered in avery schematic frame as domains rich in sulfonatedgroups separated by domains containing hydrophobicblocs. For this system, the peak is quite large dueto a weak supramolecular ionic domain organisation.With increasing the number ofX units (keepingX/Yconstant), its position shifts to the lowerq-values. Us-ing neutron or standard X-ray small-angle scatteringtechnique down to 10−3 Å−1, we can determine, witha better accuracy, the characteristic distance definedby this peak, which varies from 200 to 500 Å whenX goes from 3 to 9[11]. These characteristic dis-tances are of the same order than lengths that couldbe estimated summingX andY block chains lengths.Nevertheless, it is important to point out that thisdetermination remains quite dependent on the wayto extract theq-position of the shoulders since theyappear over a strong intensity upturn.

The differences with the usual ionomer spectra[1,2,21,22]concerns first the position of the ionomer

peak which is located at lower angles and indicatesthe existence of three to five times larger ionic clus-ters or larger distances between ionic domains in thepolymeric matrix compared to perfluorosulfonatedsystems[23]. Second, we can observe a second broadand isotropic maximum at larger scattering angles(qmax ≈ 0.42 Å−1), which does not exist in perfluoro-sulfonated polymeric systems and could be attributedto the sequential form of the polymer.

A specificity of this study is the use of microSAXSto scan the membrane along its transverse direc-tion despite its low dimension (<100�m). The2-dimensional microSAXS pattern (Fig. 1b) pointsout the strong anisotropy of these membranes. Thescattering along thez-direction is identical to thatalong thez-direction in parallel configuration sincethis direction is the intersection line of both scatteringplanes (Fig. 2c). However, some major differencesappear along they-direction, perpendicularly to themembrane surfaces.Fig. 2bshows the X-ray scatter-ing intensities of the same samples than inFig. 2a,plotted along the transverse (y−) direction as a func-tion of the transfer momentumq. These spectra arealso characterised by the presence of an ionomerpeak (or shoulder) but located at larger scatteringangles than in the parallel directions and indicating asmaller distance between scattering objects. This dif-ference explains the anisotropy of scattering intensity


Fig. 2. SAXS profiles obtained for slightly hydrated sPI membranes(X = 3, 5 and 9 and IEC= 1.26 meq./g, equilibrated at 15% ofrelative humidity) along the parallel (orz−) (A) and transverse(or y−) directions (B). Fig. 2c presents a comparison betweenboth equivalent scattering curves in thez direction extracted fromspectrum (X = 5 IEC = 1.26 meq./g) obtained in both geometries.The statistic and the signal over background are higher in thetransverse geometry due to a larger scattering volume.

observed and suggests a preferential orientation anddistribution of non centrosymmetric scattering objectsin the membrane. The symmetry in the scatteringpattern is then uniaxial with the axis normal to the

membrane surfaces and was observed for all thedifferent sPI membranes with a short and rigid hy-drophobic diamine in theY block (ODA type); thescattering features depend on the bloc length and theIEC. Analysing the peak position, we still determinean increase of the characteristic transversal distanceas a function of theX bloc length and also with thedecrease of the IEC for a givenX value.

Another difference between these two differenttypes of spectra concerns the appearance larger in-tensity of the a scattering maximum at larger angles(qmax = 0.42 Å−1), which is significantly less visiblein the transverse direction. This second peak cor-responds to a repeat unit distance in the membraneclose to 15 Å and can be attributed to an electronicdensity fluctuation along the rigid ionic sequences dueto the periodic position of the sulfonated groups[24].The anisotropy in intensity, observed at this scale onthe two-dimensional scattering pattern, comes from avariation of the scattering intensity distribution as afunction of the azymuthal angle and suggests eitheran orientation of the polymer chains in the plane ofthe membrane. It will be necessary to perform anadapted wide angle scattering X-ray study in addi-tion to a molecular simulation work to determinewhether the relative positional and orientational orderof the polymer chain is induced from the ionic or thehydrophobic aggregation.

Scattering studies performed in parallel configura-tion with different swelling ratio had shown that thestructure is insensitive to the swelling on a nanometerscale no shift of the ionomer was observed as the func-tion of the water content[11]. The swelling only ledto a large modification of the level of scattered inten-sities owing to a modification of the contrast and wasattributed to the filling by water of free volumes in thefilm. However, the macroscopic swelling has a geo-metrical effect mainly along the membrane thickness.Therefore, using microSAXS technique, we studied aswelling localised along thickness, which is not ob-servable with classical experiments. Namely, we ex-pected to observe a shift of the ionomer peak in thetransverse configuration due to a swelling of the uni-axial structure. But its position in theq-space is to-tally not affected by the swelling along this directionas well (Fig. 4). At the opposite, a strong effect on thecontrast is observed beyond 0.1 Å−1 and in both par-allel and transverse directions, changing the scattered


Fig. 3. SAXS profiles of a dry mAPI naphthalenic sPI sample (IEC= 0.96 meq./g,X = 5, [14]) as a function of different counterions,H+, Na+ and Cs+.

Fig. 4. SAXS profiles obtained for sPI membrane (X = 3 and IEC= 1.98 meq./g) for different water contents in parallel (A) and transverse(B) configurations.


intensity profiles as we increase the water uptake.Moreover we can observe a difference at lowq-valuefor which a very strong increase in the scattered in-tensity scattered is observed only along the transversedirection suggesting the existence of very large andanisotropic water domains.

One way to analyse this anisotropic scattering fea-tures is to consider the scattering objects as largediscoid or sheet-like polymeric aggregates, which arepacked along the thickness. This packing is proba-bly controlled by the film processing. The ionomerpeak (or shoulder) can be considered as the signa-ture of an electronic density heterogeneity in thesenanometric packing, made of an anisotropic alter-nation of ion-rich and ion-poor domains presentingcharacteristic distances two to three times smaller inthe transverse than in the parallel directions. Theselengths are all the larger as the bloc sequences arelong. Then, at few Angstroms length scale, a pref-erential orientation of the chains is observed in theplane of the film, due to the rigidity of the polymericsequences. We can assume that the ionic parts ofdifferent chains are associated forming aggregatesricher in sulfonated groups. These parts can be easilyhydrated over a few tenths of Å that can explain thecontrast change observed at large scattering anglesin the spectra. It is important to remember that thefilm processing is made from ammonium salt form of

Fig. 5. Typical TEM picture of a sPI membrane (X = 5 and IEC= 1.26 meq./g). Thez and y direction are indicated on the picture. Thespace between both arrows at the centre of the picture corresponds to an averaged and observable period.

the polymer, which is then acidified, exchanging theammonium counterions with protons. Since the glasstemperature of these films is high, we can considerthat the removal of the ammonium counterions and ofthe residual solvent leave some free volumes, whichcan be filled by water without a swelling of the ionicstructure. But, as a matter of fact, we observe also ananisotropic swelling with a structural effect at a scalelarger than few thousand of Angstroms, which are notreally accessible, using microSAXS techniques. Thisis not yet explained. Some recent observations un-der optical microscope reveal a kind of macroscopicphase separation made of quite large domains like flatand entangled droplets whose the nature is not clearlydefined but could be at the origin of the macroscopicswelling. Finally, it is important to emphasise that theanisotropy of sPI significantly increases with rigidityof the polymer chains since polymers presenting thesame ionic sequence but with more flexible blocksin the hydrophobic part or in the ionic sequence—for example, a phthalimid group—exhibit isotropicspectra in both parallel and transverse configurations.

TEM observations were performed in order to visu-alize the anisotropic structure of the sPI films, in thereal space. The micrographs of ultrathin cross-sectionsreveal a lamellar organization with an estimated pe-riodicity of about 500–1000 Å (Fig. 5). This valuecorresponds to a larger distance that measured from


Fig. 6. Typical SEM picture of a sPI membrane (X = 9 and IEC= 1.98 meq./g) stretched to fracture on swollen state and then freeze dried.

the SAXS ionomer peak in transverse configuration.Although the lamellar structure is clearly visible onthe TEM images, the intensity and regularity of thefluctuation of the electron density are weak whichexplains why in the scattering spectra, at lower scale,the ionomer peak is neither sharp nor intense as well.It would be very interesting to perform these experi-ments under suitable humidity conditions—instead ofvacuum—in order to correlate the swelling effect inboth real and Fourier space.

The SEM pictures of samples fractured in theswollen state show also patterns from intercon-nected foliated internal structure (Fig. 6) revealing ananisotropy at the micrometer scale. Such surface rup-ture induced by the combined effect of deformationand swelling are probably not completely represen-tative of the actual membrane structure. However,this foliated structure can be considered as charac-teristic from these films since this organisation wasnot observed for reference samples, such as Nafion®

membranes submitted to the same procedure. Thisanisotropy on a micrometer scale corresponds to alarger scale compared to microSAXS data indicatingthat these materials are really characterised by sev-eral levels of anisotropy that means several order of

packing along the transverse direction. It is importantto notice that most of SEM pictures also suggest theexistence of a thin but dense crust on one side ofthe membrane, which could be very important forthe interpretation of ion transport properties and gaspermeation. Some studies are in progress to probe theappearance of this skinning effect during the castingand drying process.

The study of the conductivity as a function of theion content and along the transverse direction wasanalysed taking into account the anisotropic structuralfeature described previously. Usually, an exponentialincrease of the conductivity over several orders ofmagnitude as ion content increases was observed formost of the proton conducting membranes[25]. Theconcentration increase of ionic mobile species andthe associated water molecules cannot only explainthis observation without using the concept of perco-lation threshold. This behaviour was observed for sPIcontaining flexible polymer chains such as phthalicsPI since the size of the ionic domains for a givensequence length neither depends on water content norion content[20]. Fig. 7 presents the evolution of theconductivity as a function of ion content for naph-thalenic sPI (rigid chains) in comparison with phthalic


Fig. 7. Conductivity data against ion content for sPI membranes(X = 5) with rigid (naphthalenic) and flexible (phthalic) polymerchains.

sPI (flexible chains), all the samples presenting thesame ionic sequence length (X = 5). For large IEC,the evolution of the conductivity is similar for bothchemical structures and corresponds to the progressivedecrease of the tortuosity in a percolated system. Thebehaviour for low IEC is different for rigid polymerssince the conductivity values reach a plateau witha high conductivity level compared to the phthalicsystem for the same IEC. This effect can only be ex-plained assuming the existence of highly anisotropicionic domains with a percolation threshold at a lowerion concentration as experimentally observed.

Then, to characterize the anisotropic charge trans-port through the films, the diffusion coefficients weredetermined using PFGSE-NMR.Fig. 8shows that thediffusion coefficients of the ammonium ions dependon which is characteristic of a diffusion in a re-stricted geometry at the micrometer scale[26]. Twolimiting values can be extrapolated:D0 at very shorttimes, that is the value without structural restrictionon a micrometer scale, andD∞ at long times that in-tegrates all the restricted paths of diffusion.D∞ andD0 are usually correlated through the tortuosity co-efficient,θ (D∞ = θD0). Unfortunately, this relationis probably not valid in the present case because theD0 value extracted fromFig. 8 does not correspondto the diffusion coefficient in the absence of polymersince the membrane present several levels of organ-isation.D∞, about 3× 10−7 cm2/s within the mem-brane, is one hundred times lower than in aqueoussolutions with the same ion concentration. Moreover,

Fig. 8. Diffusion coefficient variation of tetramethylammoniumions as a function of the delay time for a sPI membrane (X = 9and IEC= 1.98 meq./g) in parallel (�) and transverse () con-figurations.

the in-planeD∞ value is four times larger than thetransverse one. This factor indicates clearly the exis-tence of a strong anisotropy in the transport propertiesin agreement with the results obtained from the struc-tural study.

4. Conclusions

Anisotropy of the structure and of the transportproperties in sPI membranes were pointed out usingmicroSAXS, electron microscopy, conductivity andpulse field gradient NMR experiments. MicroSAXSexperiments allowed the original structural studyalong the three-dimensions of the samples and espe-cially along thickness (or transverse direction). Thestructure along the two planar directions is isotropicwhile a significant SAXS anisotropy is observed alongthickness. This anisotropy was observed from a hun-dred of angstroms (preferential orientation of poly-mer chains) to few micrometers (anisotropic upturnin intensity) and was interpreted as originated fromanisotropic domains close to lamellae or foliated poly-meric aggregates packed along thickness. Althoughthe different techniques do not allow to probe thesame length scales, this anisotropic structure was con-firmed by electron microscopy since both TEM pic-tures of ultrathin membrane slices and SEM picturesof stretched and fractured samples revealed a layered


structure. On a transport properties point of view,ionic diffusion coefficients determined using pulsefield gradient NMR along the planar and transversedirection revealed a significant anisotropy (a factor4) as well, which confirmed the structure–transportrelationships. It would be quite important to elabo-rate a structural model to quantify these correlationsbut at that time the structural information remainsrather complex at the different length scale in orderto present a least a picture which would be not tooschematic. The conductivity data depending on ioncontent can only be interpreted in the frameworkof the model and demonstrate that the use of rigidpolymer sequences in sulfonated polyimide systemsallowed to obtain ionomer membranes with a quitehigh level of conductivity for a lower charge content.Since in these system, an optimum ion content hasto be found to reduce the swelling stress, rigid naph-thalenic sulfonated polyimide polymers appears to bea good candidate for electrochemical applications.

Acknowledgements

The authors would like to thank the EuropeanSynchrotron Radiation Facility (ESRF, SC701), theCentre Grenoblois de Résonance Magnétique and theCERMAV for the access to the micro-SAXS, NMRand mechanical spectrometers, respectively, and tothe French Ministry of Research and Technology forfinancial support through the PREDIT and “réseautechnologique PACo” programs. We are also indebtedto the CNRS/LMOPS and Régis Mercier for polymersynthesis; to the CEA-le Ripault, Philippe Capronand Franck Jousse for the membrane preparation,Armel Guillermo for the help in NMR experimentsand Michel Pinéri for fruitful discussions.

References

[1] Ionomers: Characterization, Theory and Applications, in: S.Schlick (Ed.), CRC Press, Boca Raton, 1996.

[2] Ionomers: Synthesis, Structure, Properties and Applications,in: M.R. Tant, K.A. Mauritz, G.L. Wilkes (Eds.), Chapmanand Hall, London, 1997.

[3] T.D. Gierke, G.E. Munn, F.C. Wilson, The morphology inNafion® perfluorinated membrane products, as determined bywide- and small-angle X-ray studies, J. Polym. Sci. Polym.Phys. Ed. 19 (1981) 1687.

[4] L. Rubatat, A.-L. Rollet, G. Gebel, O. Diat, Evidence ofelongated polymeric aggregates in Nafion®, Macromolecules35 (2002) 40.

[5] G. Gebel, P. Aldebert, M. Pineri, Swelling study of perfluor-osulphonated ionomer membranes, Polymer 34 (1993) 333.

[6] H.P. Dhar, Review on solid polymer fuel cells, J. Electroanal.Chem. 357 (1993) 237.

[7] O. Savadogo, Emerging membranes for electrochemicalsystems. Part I. Solid polymer electrolyte membranes for fuelcell systems, J. New Mater. Electrochem. Syst. 1 (1998) 47.

[8] S. Faure, M. Pineri, P. Aldebert, R. Mercier, B. Sillion,Polyimides sulfonés, membranes préparées avec ceux-ci etdispositif de pile à combustible comprenant ces membranes,PCT Int. Appl. (1997) FR 96-5707.

[9] E. Vallejo, G. Pourcelly, C. Gavach, R. Mercier,M. Pineri, Sulfonated polyimides as proton conductorexchange membranes. Physicochemical properties andseparation H+/Mz+ by electrodialysis comparison with aperfluorosulfonic membrane, J. Memb. Sci. 160 (1999) 127.

[10] Y. Zhang, M. Litt, R.F. Savinell, J.S. Wainrigth, Moleculardesign considerations in the synthesis of high conductivityPEMs for fuel cells, Polym. Preprints 40 (2) (1999) 480.

[11] N. Cornet, O. Diat, G. Gebel, F. Jousse, D. Marsacq, R.Mercier, M. Pineri, Sulfonated polyimide membranes: anew type of ion-conducting membrane for electrochemicalapplications, J. New Mater. Electrochem. Syst. 3 (2000) 33.

[12] T. Watari, J. Fang, K. Tanaka, H. Kita, K.-I. Okamoto, Ionicconductivity and vapor permeation properties of polyimidescontaining sulfonic acid groups, Polym. Mater. Sci. Eng. 85(2001) 334.

[13] S. Faure, Synthèse et caractérisation de nouvelles membranespolyimides sulfoniques pour pile à combustible H2/O2,Université Joseph Fourier, Grenoble, 1996.

[14] C. Genies, R. Mercier, B. Sillion, N. Cornet, G. Gebel,M. Pineri, Soluble sulfonated naphthalenic polyimides asmaterials for proton exchange membranes, Polymer 42 (2001)359.

[15] N. Cornet, G. Beaudoing, G. Gebel, Influence of the structureof sulfonated polyimide membranes on transport properties,Separation Purif. Technol. 22/23 (2001) 681.

[16] V. Detallante, D. Langevin, C. Chappey, M. Métayer, R.Mercier, M. Pinéri, Water vapor sorption in naphthalenicsulfonated polyimide membranes, J. Memb. Sci. 190 (2001)227.

[17] D. Wu, A. Chen, C.S. Johnson Jr, An improved diffusion-ordered spectroscopy experiment incorporating bipolar-gradient pulses, J. Magn. Reson. 115 (1995) 260.

[18] X. Gong, A. Bandis, A. Tao, G. Meresi, Y. Wang, P.T.Inglefeld, A.A. Jones, W.-Y. Wen, Self-diffusion of water,ethanol and decafluoropentane in perfluorosulfonate ionomerby pulse field gradient, NMR Polym. 42 (2001) 6485.

[19] D.Q. Wu, J.C. Phillips, R.D. Lundberg, W.J. MacKnight, B.Chu, Long-range inhomogeneities in sulfonated polystyrenneionomers, Macromolecules 22 (1989) 992.

[20] N. Cornet, Relation entre la structure et les propriétés demembranes en polyimide sulfoné pour pile à combustibleH2/O2, Ph.D. Thesis, University of Joseph Fourier, Grenoble,1999.


[21] L. Jones, P.N. Pintauro, H. Tang, Coion exclusion propertiesof polyphophazene ion-exchange membranes, J. Memb. Sci.162 (1999) 135.

[22] K.D. Kreuer, On the development of proton conductingpolymer membranes for hydrogen and methanol fuel cells, J.Memb. Sci. 185 (2001) 29.

[23] S. Faure, N. Cornet, G. Gebel, R. Mercier, M. Pineri, B.Sillion, Sulfonated Polyimides as Novel Proton ExchangeMembranes For H2/O2 Fuel Cells, in: O. Savadogo, P.R.Roberge (Eds.), Proceedings of the Second InternationalSymposium On New Materials For Fuel Cells and ModernBattery Systems, Montréal, Canada, 1997.

[24] T.J. Shin, B. Lee, H.S. Youn, K.-B. Lee, M. Ree,Time-resolved synchrotron X-ray diffraction and infraredspectroscopic studies of imidization and structural evolutionin a microscaled film of PMDA-3,4′-ODA poly(amic acid),Langmuir 17 (2001) 7842.

[25] R.W. Kopitzke, C.A. Linkous, H.R. Anderson, G.L. Nelson,Conductivity and water uptake of aromatic-based protonexchange membrane electrolyte, J. Electrochem. Soc. 147(2000) 1677.

[26] P.P. Mitra, P.N. Sen, L.M. Schwartz, P. Le Doussal, Diffusionpropagator as a probe of the structure of porous media, Phys.Rev. Lett. 68 (1992) 3555.

Documents

Anisotropy of structure and transport properties in sulfonated polyimide membranes