cyclodextrin membranes

cyclodextrin membranes

Separation and Purification Technology 22-23 (2001) 255– 267 www.elsevier.com/locate/seppur Transport properties of Nafion®/cyclodextrin membranes N...

196KB Sizes 8 Downloads 40 Views

Separation and Purification Technology 22-23 (2001) 255– 267 www.elsevier.com/locate/seppur

Transport properties of Nafion®/cyclodextrin membranes N. Grossi, E. Espuche *, M. Escoubes Laboratoire des Mate´riaux Plastiques et des Biomate´riaux, UMR CNRS no. 5627, Uni6ersite´ Claude Bernard, Lyon I, I.S.T.I.L., 43, Boule6ard du 11 No6embre 1918, 69622 Villeurbanne Cedex, France

Abstract The transport properties of Nafion®/cyclodextrin membranes were studied for simple gases and water. Cyclodextrins (CD) in ionic form have been introduced as counter-ions in the ionic phase of Nafion® by solution exchange. The CD and Nafion® transport parameters were at first determined using sorption methods. The cyclodextrin gas and water solubility coefficients are higher than the Nafion® ones but the diffusion coefficients are lower. For the Nafion®/CD composite properties, the gas permeability and diffusion decrease with the CD content showing that the free volume of the cyclodextrin molecules is not accessible to the diffusant molecule. Moreover the detailed analysis of the permeation and sorption data shows that gases do not diffuse through the Nafion® ionic zone. A similar decrease as a function of the cyclodextrin content is obtained on the water diffusion and permeability whereas it is known that water diffuses through the ionic phase. This behaviour has been explained by the fact that cyclodextrins reduce the water swelling capacity of this ionic phase and modify its reference transport properties. A small angle neutron scattering (SANS) microstructural analysis of the composite membrane confirms this phenomenon. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Cyclodextrin cage molecules; Ionic polymer; Gas and water transport properties

1. Introduction The realisation of membranes for gas separation has attracted a great deal of interest for many years. Modification of the physical and chemical structures of polymer have been developed in order to achieve better separation characteristics. More recently, the effects of cage molecules like zeolites or cyclodextrines (CD) on the polymer * Corresponding author. Tel.: +33-4-72431002; fax: + 334-72431249. E-mail address: [email protected] (E. Espuche).

matrix [1–11] have been investigated. These molecules are interesting due to their cavities in the range of the dimension of gas molecules which can be moreover calibrated by functionalisation under neutral (CDNH2) or ionic form (CD N+I− ) in the case of CD [12]. Another interesting characteristic of these molecules is their inclusion capacity towards some diffusants notably ethanol in the case of CD. Concerning the transport mechanism of the filled matrix, different cases can be distinguished in the literature: “ For homogeneous systems with a molecular dispersion of fillers, the direct effect of the

1383-5866/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 3 - 5 8 6 6 ( 0 0 ) 0 0 1 2 6 - X

256

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

cage, by means of either an increase of solubility due to the inclusion effect or an increase of diffusivity, due to the free volume, leads to an increase of the separation factor. For example, the increase of ethanol solubility without a loss of diffusivity has been observed in the case of PDMS/zeolites membranes while no modification of the water transport is observed; the ethanol/water selectivity is thus enhanced [6]. It has also been shown that the water/ethanol selectivity of polyvinyl alcohol increases when CD are introduced in the membrane [10,11]. In that case, a decrease of ethanol diffusivity due to the inclusion effect is noticed whereas an increase of the water diffusion coefficient due to the free volume effect of CD is observed. “ For heterogeneous systems, the interface often plays an important role in the transport mechanism. Indeed in the case of strong polymer – filler interaction, the interphase can control the flux of the diffusant species and favours one of them [8] whereas in the case of weak polymer – filler interactions, the filler may form a void in the interface between the polymer and the filler and in the latter case, a loss of selectivity is often observed [3]. According to the literature results, it seems then interesting to obtain a molecular dispersion of fillers and to achieve a continuous path of cages. In order to attain this goal, we have used as polymer matrix a perfluoronated ionomer polymer Nafion®. Nafion® was chosen because of its exceptional chemical, thermal, mechanical and transport properties and because of its interesting morphology characterized by a microphase separation between hydrophobic fluorocarbon back-

bone and hydrophilic ionic domains [13 –15]. Indeed a CD ionic form can be introduced by ionic exchange as counter ion in the ionic domains and a molecular dispersion in the specific ionic phase can thus be expected. The effect of introduction of neutral or ionic cyclodextrins in Nafion® was thus evaluated on gas and water transport properties and discussed as a function of the CD dispersion mode.

2. Materials

2.1. Cyclodextrins CD are cyclic molecules consisting of six to eight glucose units linked through a1–4 linkages. They form a conic cage with an hydrophobic cavity whose dimensions are precised in Fig. 1 and an hydrophilic border. In this study b cyclodextrins composed by seven glucose units have been used under two forms: a monomer form and an oligomer form corresponding to an average polymerization degree equal to 3. The functionalisation of the monomer consists in replacing the OH group of the carbon 6 (C6) of one of the glucosidic cycles by an ionic group (pyridium group) in order to obtain an ionic CD, or by a neutral group (amine group) in order to obtain a neutral CD. In the case of oligomer CD, the substitution by the pyridium group is made on all the C6 of the CD. Fig. 2 presents, as example, the persubstituted oligomer CD. All these CD are under powder form and they are unfilmable. Their mean diameter deduced from S.E.M. observation and by B.E.T. method

Fig. 1. Shape and dimension of CDs.

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

257

Fig. 2. Persubstituted oligomer b cyclodextrin.

are, respectively, 20 mm for oligomer and 7 mm for monomers. No data on transport properties of CD are available in the literature and one aim of this work is to determine the gases and water diffusion /solubility coefficients.

2.2. Nafion ® Nafion® is a polytetrafluoroethylene with sulfonic pendant groups bound to the perfluorinated backbone by fluoroether side chains. Its chemical structure is:

The Nafion® used in this study is Nafion®117 with n =5 and the counter ion is Li+. The thickness of the film is 170 mm.

The microstructure of Nafion® has been studied extensively because of the potential applications of such membranes: Cl2/NaOH separation, solid polymer electrolyte technology… As shown in different papers [16,17], the phase separation is due to multiplets formation between the acidic dipoles followed by clustering of these multiplets into hydrophilic domains. The latter are  5 nm in size and are connected by short channels of 1 nm diameter. This cluster-channel structure offers good pathways for diffusion of gases or ions. When Nafion® is highly swollen in water (for water uptakes higher than 18% in volume) the clusters percolate forming a continuous diffusive phase that may lead to permselectivity [18]. The changes in structure upon water swelling can be accurately followed by the changes in position and intensity of a peak, called ‘ionomer peak’, which appears in small angle neutron scattering (SANS). A relation can even be made between the water content and the small angle scattering spec-

258

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

trum [19]. This relation has been used for example to determine the water concentration profiles in Nafion® membranes during permeation or pervaporation experiments [20]. The profile results explain some of the discrepancies which have been observed when the percolation of ionic clusters is achieved for water uptakes higher than 18% in volume. Previous studies have shown that oxygen and nitrogen diffusions are not very sensitive to the water content and especially to the percolation phenomenon. At the opposite, a great effect is noticed when liquid crystal is dispersed in Nafion®: for high liquid crystal contents (28 – 35%), the H2 and O2 permeation coefficients jump by some five orders in magnitude at the crystal – nematic transition and retain these high values upon cooling and thermal cycling without any permselectivity [22]. These results are explained by spectroscopic investigations [21] which indicate that liquid crystal appears as phase-separated droplets (0.2 –0.6 nm) and is located in different domains from water. In this study the goal was thus to introduce calibrated cage molecules as counter ions in the ionic domains and to follow the effects on gas transport (dry state) and on water transport (wet state).

2.3. Composite membrane realisation The Nafion® film is desorbed in a SETARAM B70 microbalance under secondary vacuum to obtain a constant weight m1 and it is swollen in N-methyl formamide (NMF) at 80°C. The CD powder is dissolved in the same solvent (NMF) at room temperature. The swollen film is then introduced in this solution for 3 h at 40°C allowing the CD diffusion. At last, the composite membrane is desorbed in the SETARAM microbalance in the same way as the initial film until a constant weight m2 is attained. The CD content is directly deduced from the two dry weights. For neutral CD this amount is exactly the difference m2 – m1. For ionic CD, the calculation is more complex. It can be realised in the case of monosubstituted CD since during exchange no anion I− is introduced in the mem-

brane as verified by the uncolorless of the composite membrane. The number of CD+ introduced, thus the number of the exchanged Li+, can then be deduced from the mass uptake. The initial number of Li+ can be calculated from the equivalent weight of Nafion® (1100 g in the case of Nafion® 117). The ratio of these two quantities determines the ionic exchange rate. The calculation is not possible in the case of ionic oligomer CD because the composite membranes become yellow indicating that exchange is not the only mechanism that takes place. Table 1 presents the range of CD content introduced in the various composite membranes and the rate of ionic exchange when it is calculable. The state of dispersion inside the membrane has not been evidenced from SEM images. The gas transport properties analysis will be used to obtain informations about this morphology as described in the methodology part hereafter.

2.4. Transport properties analysis It consists in the determination of the three transport parameters: the permeability coefficient (Pe expressed in barrer units with one barrer= −1 ), the solubility 10 − 10 ccSTP cm cm − 2 s − 1 cmHg 3 −1 coefficient (S expressed in cmSTP cm − 3 cmHg ) and the diffusion coefficient (D expressed in cm2 s − 1). These parameters being related by the equation: Pe = DS in the case of a Fickian transport mechanism, it is very interesting to determine them independently in order to confirm this fickian behaviour. The solubility and the diffusion coefficients were determined by sorption experiments whereas the permeability and the diffusion coefficients were determined by permeation experiments.

2.5. Sorption analysis The specimen are introduced in a SETARAM B92 microbalance and the studies are made at a constant temperature (20°C) for water partial pressures ranging from 0 to 1 and for gas pressures ranging from 0 to 760 torr. In both cases the weight uptake at the sorption equilibrium allows to obtain the sorption isotherm. A mean solubility

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

coefficient (S) can then be deduced from the mean slope of the isotherm. Furthermore, the evolution of the weight uptake as a function of time gives kinetical informations. The diffusion coefficients D are deduced from Fick’s second law: (c ( c =D 2 (t (x 2



(1)

By introducing the following boundary conditions: the concentration of sorbed molecule is equal to 0 for t=0 at each point of the specimen except at the surface where the equilibrium is reached instantaneously, a simplified solution of the Fick’s equation is obtained: “ In the case of powders Mt 6 1 ( −P2n 2Dt) =1− 2 % 2 exp r2 M P n=0 n

(2)

with r the radius of the powder, Mt the gas or vapour weight sorbed at the time t and M the gas or vapour weight sorbed at the sorption equilibrium.

“

259

In the case of films

1 Mt 8 − P2(2n + 1)2Dt = 1− 2 % exp e2 M P n = 0 (2n + 1)2 (3) with e, the thickness of the film. The value of D can be computed by iterative calculations leading to the best fit of experimental data by the theoretical equation.

2.6. Permeation analysis The permeation cell thermostated at 20°C consists in two compartments separated by the studied membrane (useful area: 3 cm2). A preliminary high vacuum desorption is performed to ensure that the static vacuum pressure changes in the downstream compartment are smaller than the pressure changes due to the gas or water vapour diffusion. The pressures variations in a defined volume allow to determine the gas or vapour quantities that are diffusing through the membrane during

Table 1 Presentation of the composites membranes used in this work CD type

CD weight fraction (%)

CD volume fraction (%)

Ionic monomer CD monosubstituted (CDN+I−)

6.8 7.7 18.2 22.9 23.3 34.6 38 39.4

12.7 14.3 30.8 37.3 37.8 51.5 55 56.6

Ionic oligomer CD persubstituted (CD*N+I−)

9.9 11.3 14.1

18.1 20.3 24.7

Neutral monomer CD monosubstituted (CDNH2)

8.1 9.8 13.5 18.7 22.2 29.1 32.6 36.8

15 17.8 23.8 31.5 36.4 45.1 49.1 53.8

37

54

Ionic exchange (%) 6.8 7.7 20.5 27.5 28.1 49 56.6 60.2

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

260

the steady state. The flux and thus the permeability coefficient can be calculated. Gas studies are performed under a 3.105 Pa and water studies are performed under different pressures in the range 0–15 mmHg, i.e. for partial pressures ranging from 0 to 1. The diffusion coefficient D in both cases is deduced from the time lag (q) provided by the extrapolation of the steady-state line on the time axis. D= e 2/6q

(5)

using the sorption data, the permeability coefficient of CD was calculated.

2.7. Methodology of analysis of transport parameters As we previously precised, the water and gas transport properties of the composite membranes should be very sensitive to the CD dispersion mode and three cases have been taken into account: 1. In the case of a molecular CD dispersion, the permeability of the composite should be higher than that of the pure matrix due to the free volume brought by the cavities but no predictive calculation can be made. 2. In the case of a CD phase formation whose size is in the micrometric range, the transport properties of the dispersed phase may be the same as those of the bulk CD and the evolution of the composites permeability should obey to Maxwell law: Pe =

3PecPedbd +2P 2ecbc +PedPecbc 3Pec + bc(Ped −Pec)

Pe 2bc = Pec 3− bc

(7)

(4)

Permeation measurements were performed on Nafion® and composites membranes. Indeed the experimental determination of CD permeability is not possible due to their unfilmable character. Nevertheless, the application of the law relative to the fickian transport Pe = DS

spheric phase, bc the volume fraction of the continuous phase and bd the volume fraction of the dispersed phase.This law can be simplified in the case of impermeable spheres dispersed in a permeable medium and the following equation is obtained:

(6)

with Pec the permeability of the continuous phase, Ped the permeability of the dispersed

3. At least, in the case of interactions between CD and Nafion® leading either to the creation of an interphase controlling the transport properties or to the modification of the transport properties of one of the components, no predictive values of the permeability coefficient can be proposed. If the first dispersion mode is easily imagined, it seems strange how powder of micrometric range can be incorporated in the membrane, especially for ionic CD in counter ion position. However this last case cannot be excluded and it is the only one which allows a quantitative analysis of the transport data, using the CD reference values measured on the CD powder

2.8. SANS analysis The SANS experiments are detailed in Ref. [20]. They are performed using a special cell with quartz windows that are transparent to neutrons. The membrane separates two compartments that can be both connected either with a vacuum system or with a water-controlled atmosphere. The data acquisition is performed when the water sorption equilibrium is attained. The scattered intensity is recorded over a Q[(4y/u) sin q] range between 0.02 and 0.22 A, − 1 with a wavelength u of 6.36 A, . The scattering curve shows a peak, the ‘ionomer peak’, whose position and intensity depends on the water content. The position of this peak is associated with a distance between scattering objects which correspond to ionic hydrophilic domains.

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

261

Fig. 3. Sorption isotherms obtained for CDNH2 for (a) N2, (b) O2.

3. Results

“

3.1. Gas transport properties “

3.1.1. CD properties The Fig. 3 presents the sorption isotherms obtained for a neutral CD for N2, O2, respectively. The CO2 sorption isotherm is presented in Fig. 4. The weight uptakes increase from N2 to O2 and to CO2. Two different forms are observed:

The sorption isotherm relative to N2 is linear indicating a simple dissolution mode (Henry’s type random mixing with weak interactions). The O2 and CO2 isotherms are curved towards the pressures axis and are thus representative of a dual mode sorption indicating an addition of two contributions: the Langmuir’s type sorption with high interactions on specific sites or in unrelaxed volumes and the Henry’s type mixing. In this dual mode, the diffusion coeffi-

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

262

crete values determined in the range of pressures going from 165 to 600 torr. These values are extremely low. Table 2 gives the permeability coefficient values calculated by the Eq. (5) using the sorption data. The low values obtained are due to the low diffusive terms.

Fig. 4. CO2 sorption isotherms expressed in massic percent of CD, Nafion® and composite membrane containing 39% CD by weight. Confrontation with the theoretical isotherm of the composite assuming the additivity.

cient increases as the pressure increases and as the Henry’s mode is mainly acting. In all cases, a mean solubility value has been calculated from the mean slope of the isotherms (Table 2). The solubility values are in the range of the values commonly found for polymers (0.01 – −1 ). 0.1 cm3STP cm − 3 cmHg The O2 and CO2 diffusion coefficient values have been determined by the best fit of kinetical experimental data using Eq. (2). The values are not sensitive to the pressure showing that the Langmuir mode governs the sorption mechanism in the range of studied pressures. The mean values of 2.510 − 11 cm2 s − 1 for CO2 and 13.510 − 11 cm2 s − 1 for O2 have been calculated from four disTable 2 Transport parameters of the two references CD and Nafion® CD O2

CO2

S (cm3STP cm−3 cm−1 Hg ) D (cm2 s−1) Pe (barrer) S (cm3STP cm−3 cm−1 Hg ) D (cm2 s−1) Pe (barrer)

Nafion®

0.023

Not determined

13.5×10−11 0.03

Not determined Not determined

0.1 2.5×10−11 0.025

0.02 2.9×10−8 5.8

3.1.2. Nafion ® properties The Fig. 4 presents also the CO2 sorption isotherm of Nafion®. It is quite linear indicating low interactions between the gas and the polymer even at low pressures. A mean value of the solubility coefficient has been calculated: 0.02 cm3STP −1 cm − 3 cmHg , a value in the range of that found for CD. The diffusion coefficients calculated for pressures between 400 and 600 torr are almost constant with values of 2.910 − 8 cm2 s − 1, thus higher than the values calculated for CD. As a consequence, the CO2 permeability coefficient deduced from the sorption data is equal to 5.8 barrer, a value 100-fold higher than that calculated for CD. These transport parameters are also presented in Table 2 in order to allow an easy comparison with the CD ones. Nafion® appears as more permeable to CO2 than CD. 3.1.3. Composite properties The composite membrane transport properties have been studied as a function of the CD content by the permeation method. This study was not limited to the gas used in sorption method, the transport of a small and low interactive gas like H2 was also evaluated. For all the gases studied and whatever CD type used, a decrease of the flow is noticed when the CD content increases (Fig. 5). This result is opposite to the result encountered and underlines that the CD cages do not bring additional free volumes in the matrix. Two explanations can be proposed. 1. The CD, even in counter ion position in the ionic zone are not individually dispersed but form small aggregates with poor transport properties equivalent to those of bulky CD powders. In that case, as the permeability of CD is highly lower than that of Nafion®, the

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

263

Fig. 5. Evolution of the permeability coefficient of the membrane as a function of CD content for (a) H2, (b) O2, (c) N2. Confrontation of the H2 theoretical permeability coefficient calculated by the simplified Maxwell law and experimental values.

simplified Maxwell law (Eq. (7)) has to be verified. The theoretical curve is represented in Fig. 5 for H2 (best experimental precision) and a relatively good agreement is noticed between experimental and theoretical values. Nevertheless, in this hypothesis, the additivity of CD and Nafion sorptions must be verified. Fig. 4 underlines that the curve representative of this additivity for a composite with 39% CD by

weight is higher than the experimental one. Thus, CD cannot be considered as a low permeable phase participating to the transport mechanism in the Nafion composite membrane. 2. The CD are excluded from the transport. This implies that the ionic zone of Nafion® does not participate to the gas transport mechanism as the ionic CD are located in this zone. This

264

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

conclusion is in total agreement with the mentioned results concerning the different effects of water and liquid crystal on the gas transport properties as they are, respectively, located in ionic and organic domains [21,22].

3.2. Water transport properties

3.2.1. CD properties The water sorption isotherm of CDN+I− is presented in Fig. 6. It shows that the CD is an hydrophilic component with water uptake of 40% by weight at saturation. This isotherm is of BET type II and it is nearly the same for all the CD studied, showing that the substitutions on one carbon 6 do not modify the water sorption and that the glucosidic cycles of monomer or oligomer CD are the fundamental units for water sorption. Indeed the present isotherm represented in number of water sorbed molecules per glucosidic unit is identical to that obtained for all amorphous celluloses. Furthermore, the sorption mechanism is similar for these materials with two steps: bound water until P/Po =0.8 and swelling water between 0.8 and 1.0 [23] [24].

For the present study, a mean value of the solubility coefficient has been calculated between partial pressures from 0 to 0.8. It is equal to 200 −1 cm3STP cm − 3 cmHg . The diffusion coefficients have been obtained by fitting the kinetic curves. They are constant and equal to 17.10 − 11 cm2 s − 1 in the partial pressures ranging from 0 to 0.8. The calculated permeability coefficient is thus equal to 340 barrer.

3.2.2. Nafion ® properties The water sorption isotherm of Nafion® is also presented in Fig. 6. Nafion® is known as highly hydrophilic ionomer with water uptake of 20% in weight (40% in volume) at saturation. The mean slope of the isotherm for partial pressures between 0 and 0.8 allows to calculate a solubility coeffi−1 cient which is equal to 130 cm3STP cm − 3 cmHg .The diffusion coefficient increases exponentially as a function of the partial pressures to reach a value of 7 – 10×10 − 8 cm2 s − 1 for P/Po ranging between 0.7 and 1 (Fig. 7a). This value is much higher than that obtained for CD and Nafion® appears as a much more diffusive phase than CD phase. The evolution of the Nafion® permeability coefficient is similar to that of the diffusion coeffi-

Fig. 6. Water sorption isotherms expressed in massic percent of CDN+I−, Nafion® and the composite containing 38% by weight of CD. Confrontation with the theoretical isotherm of the composite assuming the additivity.

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

265

Fig. 7. (a) Evolution of the water diffusion coefficient as a function of partial pressure for (*) Nafion® and ( ) the composite membrane containing 23% by weight of CD. (b) Evolution of the water permeability coefficient as a function of partial pressure for (*) Nafion® and ( ) the composite membrane containing 23% by weight of CD.

cient and Pe reaches the very high value of 2.106 barrers at saturation (Fig. 7b); a value 10 000-fold higher than CD one.

3.2.3. Composites properties The diffusion and permeability coefficients of a composite membrane with 23% CD by weight are lower than the Nafion® ones for all partial pressures. The difference can reach two orders of magnitude (Fig. 7a,b).

As for gases, the free volume effect of a molecular dispersion of CD in the polymer matrix is not observed. The decrease of the permeability could thus be due to the low permeability of CD aggregates with bulky properties. In that case, the additivity of the sorption of each phase should be verified since they are both in the water diffusive zones. Fig. 6 underlines no satisfactory agreement between theoretical and experimental curves. A possible explanation is that the matrix behaviour and especially the ionic phase behaviour

266

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

are greatly affected by the presence of CD and as a consequence the reference isotherms of Fig. 6 are no more valid. The SANS analysis of the composite membranes confirms this hypothesis. The ‘ionomer peak’ at small angles is observed in the composite as in pure matrix confirming a microphase separation phenomenon in both cases with a distance between scattering ionic domains in the same range. Nevertheless the variations of the position (Fig. 8) and of the height of the peak as a function of water volumic fraction are respectively less regular and less important for the composites than for the matrix. The CD content decreases the water swelling capacity of the ionic zone impeding any use of pure matrix properties as references.

4. Conclusion The introduction of controlled cavities in a polymer matrix was envisaged in order to improve the permeabilities and selectivities of small molecules like simple gases and water. The initial idea was to carry out a molecular dispersion able to bring free volumes of angstroem size, moreover

if possible, percolated volumes (pearl necklace) through the membrane thickness. This idea was preferable especially since a preliminary study has shown that the micronic cyclodextrin aggregates in unfilmable powders present very poor transport properties for gas and water due to poor diffusive factors. The CD functionalised into CDN+I− were thus introduced as counter ion in the ionic zones of Nafion® ionomer membrane. The known percolation of these zones might favour the percolation of the cage molecules. The present study does not confirm these ideas. We never observe a free volume effect. On the contrary, in all cases we observe a decrease of transport properties as a function of CD content: (i) For gases, the data analysis shows that the CD located in the ionic clusters are excluded of the transport mechanism as the gas diffuse preferentially in the organic phase or at the interface between the organic PTFE and the ionic clusters. (ii) For water, the data analysis shows that the CD introduction decreases the swelling capacity of the clusters and thus the water transport properties of the matrix. Unfortunately, due to these observations, no information has been obtained about the dispersion mode of the CD in the Nafion® matrix.

Acknowledgements The authors would thank M. Pineri (C.E.A Grenoble) for fruitful discussions, A. Gadelle (C.E.A. Grenoble) for giving us the functionalised CD and G. Gebbel (C.E.A Grenoble) for performing SANS experiments.

References

Fig. 8. Evolution of the position of the ionomer peak as a function of the Nafion® volume content for, respectively, Nafion® at different hydratation states and the composite membranes at different hydratation states.

[1] A. Yamasaki, T. Iwatsubo, T. Masuoka, K. Misoguchi, Pervaporations of ethanol/water through polyvinylalcohol/cyclodextrin (PVA/CD) composite, in: R. Bakish, Bakish Materials Corp., Englewood, NJ (Eds.), Proceedings of the 6th International Conference on Pervaporation Processes in the Chemical Industry, Ottawa, 27 – 30 September, 1992, p. 252– 261. [2] A. Yamasaki, K. Misoguchi, Preparation of PVA membranes containing b-cyclodextrin oligomes (PVA/CD

N. Grossi et al. / Separation/Purification Technology 22-23 (2001) 255–267

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10] [11] [12]

membrane) and their pervaporation characteristics for ethanol/water mixtures, J. Appl. Polym. Sci. 51 (1994) 2057– 2062. J.M. Duval, B. Folkers, M.H.V. Mulder, G. Desgrandchamps, C.A. Smolders, Adsorbent filled membranes for gas separation. Part 1. Improvement of the gas separation properties of polymeric membranes by incorporation of microporous adsorbents, J. Membrane Sci. 80 (1995) 189– 198. A. Yamasaki, T. Iwatsubo, T. Masuoka, K. Misoguchi, Pervaporation of ethanol/water through a poly(vinyl alcohol)/cyclodextrin (PVA/CD) membrane, J. Membr. Sci. 89 (1994) 111– 117. A. Yamasaki, K. Ogasawara, K. Misoguchi, Pervaporation of water/alcohol mixtures through the PVA membranes containing cyclodextrin oligomer, J. Appl. Polym. Sci. 54 (1994) 867– 872. M.D. Jia, K.V. Peinemann, R.D. Behling, Preparation and characterization of thin-film-zeolite-PDNS composite membranes, J. Membr. Sci. 73 (1992) 119–128. C.H. Lee, Synthetic membranes containing Schardinger cyclodextrin additives, J. Appl. Polym. Sci. 26 (1981) 489– 497. M.G. Su¨er, N. bac¸, L. Yilmaz, Gas permeation characteristics of polymer-zeolite mixed matrix membranes, J. Membr. Sci. 91 (1994) 77–86. M. Goldmann, D. Fraenkel, G. Levin, A zeolite/polymer membrane for separation of ethanol-water azeotrope, J. Appl. Polym. Sci. 37 (1989) 1791. M.L. Bender, M. Komiyama, Cyclodextrin Chemistry, Springer Verlag, Berlin, 1978. J. Szejtli, S. Jorszef, Cyclodextrins and their Inclusion Complexes, Budapest Akademiai, Budapest, 1982, p. 296. J. Boyer, R.J. Corcoran, J.M. Lehn, Cyclodextrin chemistry: selective modification of all primary hydroxyl groups of a and b cyclodextrins, Helv. Chem. Acta 61 (1978) 2190.

.

267

[13] T.D. Gierre, G.E. Munn, F.C. Wilson, The morphology in Nafion perfluorinated membrane products, J. Polym. Sci. Phys. Ed. 19 (1981) 1687– 1704. [14] H.L. Yeager, Transport properties of perfluorosulfonate polymer membranes, ACS Symp. Ser. 180 (1982) 41 – 63. [15] M. Escoubes, M. Pineri, Water absorption studies of carboxilic and sulfonic ionomers: correlations with their microstructure, NATO ASI Ser. 198 (1987) 341– 359. [16] B. Dreyfus, Model for the clustering of multiplets in ionomers, Macromolecules 18 (1985) 284– 293. [17] W. Forsman, Statistical mechanics of ion-pair association in ionomers, NATO ASI Ser. 198 (1987) 39 – 50. [18] A. Eisenberg, H.L. Yeager (Eds.), Perfluoronated Ionomer Membranes, ACS Symp. Ser., 1982, pp. 180. [19] M. Pineri, R. Duplessix, F. Volino, Neutron studies of perfluorosulfonated polymer structures, ACS Symp. Ser. 180 (1982) 249– 282. [20] M. Thomas, M. Escoubes, P. Esnault, M. Pineri, Determination of the permeant concentration profile in a membrane during gas permeation, J. Membr. Sci. 46 (1989) 57 – 65. [21] M. Thomas, Transport de l’eau a` travers une membrane perfluorosulfonique Nafion®. Relations avec la microstructure. De´veloppement de membranes composites a` haute perme´abilite´ aux gaz, Thesis, Lyon, 1989. [22] J.A. Ratto, S. Ristori, F. Volino, M. Pineri, M. Thomas, M. Escoubes, Investigation of a liquid crystal dispersed in an ionic polymeric membrane, Chem. Mater. 5 (1993) 1570– 1576. [23] C. Fringant, J. Desbrie`res, M. Milas, M. Rinaudo, C. Joly, M. Escoubes, Characterisation of sorbed water molecules on neutral and ionic polysaccharides, Int. J. Biol. Macromol. 18 (1996) 281– 286. [24] N. Grossi, Re´alisation de membranes composite a´ cairte´s contro´lee´s pour la perme´ation gozeuse. Introduction de cyclodextrines dans un polyme´re ionique et dans un polyalacol. Universite´ Claude Bernard, Lyon I, Thesis, 1998.