Journal of Membrane Science 279 (2006) 669–674

Preparation and characterization of non-ionic block co-polymertemplated mesoporous silica membranes

Scott Higgins a, Raymond Kennard a, Nathan Hill a, Joseph DiCarlo b, William J. DeSisto a,∗a Department of Chemical and Biological Engineering, Laboratory for Surface Science and Technology,

5737 Jenness Hall, University of Maine, Orono, ME 04469, USAb Yardney Technical Products/Lithion Inc., 82 Mechanic Street, Pawcatuck, CT 06379, USA

Received 9 September 2005; received in revised form 23 December 2005; accepted 5 January 2006Available online 10 February 2006

Abstract

We report on the preparation and characterization of mesoporous silica membranes via micellar templating using the non-ionic ethyleneoxide–propylene oxide–ethylene oxide tri-block copolymer surfactant, Pluronic P123, as a template under various template/silica volume percent-ages (VTS%). The silica membranes were prepared by dip coating silica sols onto porous �-alumina supports followed by calcination to removetsnid(t©

K

1

amatdiisntttb

0d

he template. The membranes were characterized by single gas permeation and permporometry measurements. Using a VTS% of 33 we found thatilica membranes comprised of three dip/fire cycles resulted in highly reproducible continuous membranes with minimal defects. Helium anditrogen permeance values were of the order of 10−6 mol m−2 s−1 Pa−1. The pore size distribution, determined from permporometry measurements,ndicated a dominant peak at 5 nm diameter and approximately 10% microporosity. In addition, there was 8–10% permeance through pores withiameters between 15 and 40 nm. At higher VTS% (47.5 and 65) the silica membranes had considerable bypass flow. The pore size distributionPSD) of the membrane via permporometry indicated a shift to smaller values for the membrane when compared to the PSD of the powder fromhe calcined dipping sol, determined by nitrogen adsorption/desorption.

2006 Elsevier B.V. All rights reserved.

eywords: Mesoporous membrane; Silica; Membrane synthesis; Permporometry

. Introduction

Mesoporous silica membranes and materials are receivingttention due to potential applications for ultrafiltration [1,2],atrices for organic modification [3,4], biological separations

nd drug delivery [5–7], idealized supports for pore size reduc-ion, [8,9] and supports for microporous layers [10]. Since theiscovery of surfactant templated silica [11] recent efforts in sil-ca membranes have shifted from controlling pore size throughnterparticle spacing (resulting from the sintering of particulateols) to controlling pore size through the size of the intercon-ected micelle structures (formed from surfactant aggregates)hat template the silica framework. There are several advan-ages to preparing mesoporous silica membranes using micellaremplating. For example, there is a potential to prepare mem-ranes with a narrow pore size distribution, similar to what has

∗ Corresponding author. Fax: +1 207 581 2323.E-mail address: wdesisto@umche.maine.edu (W.J. DeSisto).

been achieved in powder formation for which pore size controlwithin the 2–10 nm size range has been demonstrated [11,12].Additionally, under specific synthesis conditions, the surfactantassemblies can arrange themselves crystallographically, yield-ing the potential to control pore orientation [13].

In general, the research trend has been to use synthetic strate-gies for powders and apply these to thin film synthesis. Typically,a dilute solution containing surfactant, a molecular source ofsilica (for example tetraethylorthosilicate, TEOS) and solventis prepared at low pH to quench polymerization of the silicasource. Thin film synthesis is then typically achieved via dipor spin-coating, although other routes are possible. Brinker andco-workers established an evaporation induced self-assembly(EISA) technique for preparing mesoporous thin silica filmsusing surfactant assemblies where rapid evaporation of the dilutedipping solution induced the self-assembly of surfactant struc-tures which templated the silica formation [14–16]. A good dealof work has been done on thin films formed on planar supportssuch as single crystal silicon or glass slides [12,13,17,18]. Inaddition to the work done on planar supports, there has been

376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.

oi:10.1016/j.memsci.2006.01.014

670 S. Higgins et al. / Journal of Membrane Science 279 (2006) 669–674

some work on porous supports, required for membrane fabri-cation as reviewed by Guliants et al. [13]. Typically, recipesfor film formation on planar supports are used for porous sup-ports. Important characteristics for membrane films on poroussupports include pore size control, pore connectivity, porosityand the control of thickness and defects. If defects and pore sizecan be controlled then crystallographic arrangement of poresmay be important in the overall permeance characteristics of themembrane.

Examples of mesoporous silica membranes prepared fromsurfactant templates include those made from both ionic sur-factants [19,20] and non-ionic block copolymers [1,21]. Thesestudies have demonstrated the successful formation of meso-porous silica membranes using micellar templating methods.Brinker et al. [22] demonstrated that silica membranes preparedusing polystyrene-block-poly(ethylene oxide)(PS-b-PEO) as atemplating agent resulted primarily in membranes with largermesopores interconnected with micropores. Boissiere et al. [1]prepared silica membranes using a nonionic polyethylene oxide(PEO)-based surfactant resulting in mesopores of 2.5 nm basedon cut-off permeation experiments with PEO polymer solutions.A catalyst was used to localize the silica condensation reactionwithin the pores of the macroporous support.

In our work, we have prepared mesoporous silica mem-branes using an EO–PO–EO tri-block copolymer surfactant as atemplate under various template/silica volume fractions (V =V

aesaptmampVPfs

2

2

lPc4i2Eftw

to 3 h (47.5 and 65 VTS%) in a sealed container to prevent evap-oration of EtOH. An asymmetric alpha alumina support diskwith a 10 �m top layer of 100 nm pore size (HiTK, Germany)was dipped into the solution, using a dipping apparatus, at arate of approximately 2 cm/s. After dip-coating the macrop-orous support, the surfactant was then removed by sintering ata temperature of 400 ◦C for 4 h with a heating/cooling rate of1 ◦C/min. This temperature was found to be sufficient to com-pletely vaporize the surfactant in bulk experiments. The dip/fireprocedure was repeated at least three times in order to ensurecomplete coverage of the support.

2.2. Porosity characterization

The mesoporous silica membranes were characterized usingscanning electron microscopy (SEM), single gas permeationand permporometry. Single gas permeation measurement withhelium, nitrogen and methane were conducted between 25 and150 ◦C at pressure drops between 4.67 × 104 and 1.13 × 105 Pa(350 and 850 Torr). SEM was performed with an AMRAY model1810 scanning electron microscope. Mesoporous silica powdersprepared from calcining the dipping solution were characterizedusing nitrogen porosimetry (ASAP 2010, Micromeritics).

The permporometry measurement apparatus, shown in Fig. 1was constructed based on reference [23]. The apparatus wasconstructed with 1/4 in. stainless steel tubing and fittings. Grade5a(fnt(bait0f

wTtp3tm

K

R

woaAl

TS

template/(Vtemplate + VSiO2 ) corresponding to cubic, hexagonalnd lamellar organized structures, based on the work of Stuckyt al. [12]. The silica membranes were prepared by dip-coatingilica solutions onto porous �-alumina disks. We have char-cterized these membranes using single gas permeation andermporometry measurements. Here, we report: (i) the condi-ions for reproducible synthesis of low defect mesoporous silica

embranes; (ii) the relationship between defect flow and VTS;nd (iii) the comparison of the pore size distribution (PSD) ofesoporous silica membranes and the corresponding powders

repared from the membrane dipping solution. An increase inTS resulted in an increase in defect flow. In comparing theSD’s of the silica membrane and the silica powder preparedrom the corresponding dipping sol, the PSD was shifted tomaller pore diameters for the membrane.

. Experimental

.1. Membrane preparation

Silica membranes have been synthesized using a tetraethy-orthosilicate (TEOS) silica sol templated with Pluronic123 (EO20PO70EO20, MW = 5800, BASF), a non-ionic blockopolymer. Silica membranes were prepared with VTS% of 33,7.5 and 65. For a typical synthesis of a thin mesoporous sil-ca film a mixture of 5.4 mL H2O, 5.4 mL diluted HCl (pH.0), 15.2 mL EtOH, 11.1 mL TEOS (molar ratio of 1 TEOS:5.2tOH:12 H2O:0.015 HCl:0.021 P123) was stirred vigorously

or 20 min. Then a mixture of P123 and EtOH was added tohe solution and mixed until the P123 was dissolved, solutionsere allowed to age at room temperature for 10 min (33 VTS%)

helium was purified (Alltech He purifier) for water, oxygennd particulates and passed through two mass flow controllersMKS Instruments). Mass flow controller 1 (MFC1, 0–50 sccm)ed the He stream into a stainless steel bubbler containing 99%-hexane. The bubbler was allowed to reach equilibrium usinghe bypass valve at the bubbler outlet. Mass flow controller 2MFC2, 0–500 sccm) fed pure He to the membrane. The com-ination of flow through MFC1 and MFC2 allowed for thedjustment of hexane activity (hexane vapor pressure normal-zed to saturation pressure of hexane at experimental condi-ions) over a wide range. A pressure controller (0–1.13 × 105 Pa;–1000 Torr), MKS Instruments was used to provide a constanteed pressure to the membrane.

The hexane bubbler temperature was maintained at 23 ◦Chile the membrane cell temperature was maintained at 25 ◦C.he membrane was sealed in the testing cell with o-rings,

ested for leaks, and heated overnight in flowing He at 110 ◦Crior to measurement. Typically pressures were maintained at.87 × 103 Pa (29 Torr) above atmosphere. Flow rates of Hehrough the membrane were measured using a soap bubble

eter.The activity of hexane was related to the pore size using the

elvin equation

p = −4σVm

RT ln (a)+ t

here σ is the surface tension of hexane, Vm is the molar volumef hexane, R is the gas constant, T is temperature, a is hexanectivity and t is the thickness of the adsorbed monolayer [24].s the hexane vapor pressure is exposed to the membrane capil-

ary condensation occurs in larger pores governed by the Kelvin

S. Higgins et al. / Journal of Membrane Science 279 (2006) 669–674 671

Fig. 1. Permporometry apparatus.

equation. The pore size distribution was calculated based on theanalysis presented by Cao et al. [25] using the expression

f (Rp) = −3l

2R3p

√MRT

8π

dF

dRp

where l is the membrane thickness, M is the molecular weightof the permeate, and F is the permeance of the inert gas (He)based on the expression for Knudsen transport.

3. Results and discussion

The silica membranes prepared by dip-coating �-aluminasupports into surfactant templated silica sols were comprisedof individual, 2–3 �m thick, silica layers. Layer thickness wasestimated by dip-coating on single crystal silicon supports, fol-lowed by examination in a SEM. The individual layers appearedreflective and homogeneous, free of large cracks or defects. Arepresentative SEM photograph of a 33 VTS% prepared mem-brane is shown in Fig. 2. The top layer of the alumina supportis visible (approximately 10 �m). The delineation of the silicalayer on the top layer of the support is not so clear, however thephotograph reveals a continuous, homogeneous layer. After dip-coating and room temperature drying, the support/silica layerwas tested for gas permeance to confirm the lack of defectsprior to calcination and template removal. The inability of thesuictmeb(

Fig. 2. Scanning electron micrograph (SEM) of a cross section of a three-layersilica membrane prepared with 33 VTS% of P123 surfactant in the dipping sol.

Fig. 3. Normalized helium permeance as a function of Kelvin radius for 2, 3and 4 layer silica membranes indicating the progression of support coverage aslayers were added.

upport/silica structure to permeate gases was consistent with aniform, homogeneous pre-calcined layer. Because the supports porous and non-planar, it was expected that complete surfaceoverage would require several silica layers. After the deposi-ion and calcinations of individual silica layers, the now porous

embranes were examined by permporometry to examine thextent of support coverage. This test allowed us to quantifyypass flow which is attributed to viscous flow [26] though largerdia > 100 nm) pores. Fig. 3 shows the normalized helium per-

672 S. Higgins et al. / Journal of Membrane Science 279 (2006) 669–674

Fig. 4. Permeance of nitrogen at various temperatures as a function of pressuredrop indicative of Knudsen diffusion.

meance as a function of Kelvin radius corresponding to hexaneactivity for two-, three- and four-silica layer membranes pre-pared using a dipping sol with VTS% of 33. As can be seen inFig. 3, the coverage after two layers is highly incomplete with85% of the helium flow through pores larger than 15 nm.

These larger pores are most likely the 0.1 �m pores in the toplayer of the �-alumina support due to incomplete support cov-erage with two dipped/fired silica layers. After four silica layersthe flow was nearly identical to that of the three layer membrane.The data demonstrated show that the optimal film is producedwith three layers with no gains in performance with subsequentdepositions. A more detailed analysis of helium permeance ver-sus hexane activity will be presented later.

Further evidence of complete surface coverage after threedip/fire cycles is provided in Figs. 4 and 5, which show per-meance versus pressure drop measurements. Due to the highpermeance (no measurable resistance) of the support, the over-all permeance values reported are representative of the resistanceof the thin silica layer(s). The permeance of nitrogen and heliumas a function of pressure drop across the membrane at varioustemperatures between 25 and 150 ◦C is shown in Figs. 4 and 5,respectively, for a three-layer silica membrane formed from adipping sol containing VTS% of 33. The zero slope of the perme-

Fd

ance versus pressure drop data indicates that Knudsen diffusiondominates the transport of light gases through these membranes.A positive slope is would suggest viscous flow, which is to beexpected when there is a large contribution of pores with diam-eters larger than 100 nm [26]. The inverse relationship betweenpermeance and temperature, as noted in Figs. 4 and 5, is also con-sistent with Knudsen diffusion. The ideal selectivity of nitrogenover helium at 150 ◦C, where surface diffusion is negligible, is2.46, approaching the ideal Knudsen selectivity of 2.65. Perme-ance values on the order of 10−6 mol m−2 s−1 Pa−1 are typicalof mesoporous membranes of this thickness [19,20,27].

In addition to permeance measurements, permporometrymeasurements provided further information regarding the poresize distribution and bypass flow in membranes. In Fig. 6(a)we show normalized helium permeance as a function of hexaneactivity for different three-layer silica membranes formed froma dipping sol with VTS% of 33. Fig. 6(b) shows the number ofpores as a function of the Kelvin pore radius. The reproducibil-ity of the dipping procedure is confirmed in Fig. 6(a) and (b).For the 33 VTS% surfactant silica films, we observed an initialHe permeance at lower hexane activities, corresponding to theblocking of pores with smaller diameters. At the higher hexane

Fig. 6. (a) Normalized helium permeance as a function of Kelvin radius forseveral three-layer silica membranes prepared from 33 VTS% surfactant solutionsshowing reproducibility during synthesis. (b) The pore size distribution (PSD)determined from the data in (a) of several three-layer silica membranes preparedfrom 33 VTS% surfactant solutions showing reproducibility during synthesis.

ig. 5. Permeance of helium at various temperatures as a function of pressurerop indicative of Knudsen diffusion.

S. Higgins et al. / Journal of Membrane Science 279 (2006) 669–674 673

activities (after the initial large drop in permeance) there was stillsome residual helium flow. For example, at hexane activity near0.05 (r < 3.8 nm), there is a gradual filling of pores until higherhexane activities, near 0.15 (r < 5.7 nm), are reached. At evenhigher hexane activities (>0.5), corresponding to Kelvin diam-eters between than 15 and 40 nm, there was still some residualhelium flow (helium flow was completely blocked at hexaneactivities approaching 0.8, pore diameter of 40 nm). Typically,in our membranes we observed helium permeance between 8and 10% in this intermediate region. The pore size distributions(PSD) corresponding to these results are shown in Fig. 6(b). Thedata suggest that the majority of pores have a pore diameter near47 A. The broadening of the PSD at the higher pore diameterssuggest a significant number of pores larger than 6 nm.

A comparison of permporometry measurements of mem-branes prepared with different surfactant volume percent isshown in Fig. 7. The different membranes were prepared usingVTS% of 33, 47.5 and 65 P123 surfactant in the dipping solu-tion. These volume ratios corresponded to conditions used toprepare cubic, hexagonal and lamellar pore ordered silica layers[12]. The dip-coating of porous alumina supports resulted in sil-ica films with no discernable X-ray diffraction pattern, possiblydue to the restriction of domain growth within the smaller poresof the support. The results of permporometry indicate that for thenumber of dip cycles (3 or 4) and the experimental parameters ofour study, only the 33 V membrane was without significantd45aa

ttVbtd

Flcc

Fig. 8. The pore size distribution (PSD) of the calcined dipping sols and themembrane for a 33 VTS% surfactant membrane as determined from nitrogenadsorption–desorption data and permporometry data, respectively.

The majority of pore volume for the powder was from poreswith a diameter near 75 A, whereas the majority of pores in themembrane had a pore diameter of 47 A. The membrane had alarger contribution of pores near or greater than 10 nm than thepowder material. The powder material also showed only 2%micropore volume whereas the membrane microporosity wasestimated at 15% based on the permeance through pores lessthan 2 nm.

In comparing the permporometry data for a silica membraneto BJH nitrogen adsorption data for the dried/sintered dippingsol, we observed a shift to a larger pore size for the dipping sol(Fig. 8). It must be noted that permporometry will only mea-sure connected pores, whereas nitrogen adsorption/desorptionwill measure dead-end pores in addition to connected pores.The lack of data at smaller pore sizes in the permporometrymeasurement increased the error for the pore size distributionparticularly at smaller pore sizes, when comparing to the nitro-gen adsorption data. Since both these measurements were takenin the adsorption mode there is a layer thickness of condensatethat must be estimated (0.4 nm was used in our analysis) in bothmeasurements. Therefore, the difference between measurementtechniques may explain some of the observed difference in poresize distribution between the membrane and powder. We wouldalso expect different pore size distributions because the pro-cesses for preparing the thin membrane on the porous supportare quite different from preparing bulk powder. The essentialdtEtbccmTmaT

TS%efects. For the 47.5 and 65 VTS% membranes, bypass flow of0 and 80%, respectively, was measured for pores greater than0 nm, indicating either significant defects or poor surface cover-ge. This could be consistent with the large volume of surfactantnd the expected shrinkage during surfactant removal.

The powders prepared from the drying and calcining ofhe dipping solution were analyzed with nitrogen adsorp-ion/desorption for pore size distribution. Since the 47.5 and 65TS% dipping solutions resulted in membranes with poor mem-rane properties, we compared the permporometry results andhe nitrogen adsorption results for the membrane and calcinedipping solution of the 33 VTS% synthesis, shown in Fig. 7.

ig. 7. Normalized helium permeance as a function of Kelvin radius for three-ayer silica membranes prepared from 33, 47.5 and 65 VTS% surfactant solutions,orresponding to synthesis conditions to prepare cubic, hexagonal and lamellarrystallographically ordered mesoporous thin silica films.

ifference is seen in the rapid drying during membrane forma-ion, because of the thin layer. This phenomenon occurs duringISA when the PEO chains are rapidly forming into micelles and

he drying process occurs faster than the surfactant can assem-le, leaving coiled chains extending out of the micelle. Thesehains lead to interconnected microporous channels betweenontinuous mesopores. Under the appropriate conditions theseicroporous channels may connect noncontinuous mesopores.hese microporous channels will lead to a percentage of activeicropores by contributing to the bulk flow and pore necking,

s has been observed and commented on by Brinker et al. [22].hese pores can be quantified in a permeation experiment as

674 S. Higgins et al. / Journal of Membrane Science 279 (2006) 669–674

shown in Fig. 3. Due to the significant drop in relative perme-ance in the presence of hexane activity less than 0.005 (Kelvinradius <2.3 nm) suggests that the microporosity makes a signif-icant contribution to the total flow.

Pore necking could also explain the micropore formationobserved in our membranes. Although our permeance valuesare similar to those observed by Brinker [22] we did not observethe same extent of micropore formation. It should be noted thatthe membranes prepared in [22] were made with a different sur-factant under different synthesis conditions.

4. Conclusion

Mesoporous silica films with structure directing tri-block copolymers were synthesized by dip-coating/evaporationinduced self assembly (EISA) in various ratios of silica:tri-block to produce a variety of mesostructures. The films werecharacterized by permporometry, gas permeation, N2 sorption,SEM, TEM, XRD. Characterization of numerous three cyclemembranes demonstrated that a highly reproducible continuoussilica membrane with minimal defects was produced. Compari-son between permporometry and N2 sorption techniques showsa significant increase of micropores in the EISA membranes incomparison with the bulk powder.

A

#RdC

R

[8] B.A. McCool, W.J. Desisto, Self-limited pore size reduction of meso-porous silica membranes via pyridine-catalyzed silicon dioxide ALD,Chem. Vapor Deposit. 10 (2004) 190–194.

[9] B.A. McCool, W.J. DeSisto, Synthesis and characterization of silicamembranes prepared by pyridine-catalyzed atomic layer deposition, Ind.Eng. Chem. Res. 43 (2004) 2478–2484.

[10] K. Kuraoka, N. Kubo, T. Yazawa, Microporous silica xerogel membranewith high selectivity and high permeance for carbon dioxide separa-tion, in: Proceedings of the 10th International Workshop on Glass andCeramics, Hybrids and Nanocomposites from Gels, 20–24 September1999, vol. 19, J. Sol–Gel Sci. Technol. (2000) 515–518.

[11] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck,Ordered mesoporous molecular sieves synthesized by a liquid–crystaltemplate mechanism, Nature 359 (1992) 710–712.

[12] C.A. Peter, K.L. Alberius, R.C. Frindell, E.J. Hayward, G.D. Kramer,B.F. Stucky, B.F. Chmelka, General predictive syntheses of cubic, hexag-onal, and lamellar silica and titania mesostructured thin films, Chem.Mater. 14 (2002) 3284–3294.

[13] V.V. Guliants, M.A. Carreon, Y.S. Lin, Ordered mesoporous and macrop-orous inorganic films and membranes, J. Membr. Sci. 235 (2004) 53–72.

[14] K. Yu, B. Smarsly, C.J. Brinker, Self-assembly and characterization ofmesostructured silica films with a 3D arrangement of isolated sphericalmesopores, Adv. Funct. Mater. 13 (2003) 47–52.

[15] K. Yu, B. Smarsly, C.J. Brinker, Self-assembly and characterization ofmesostructured silica films with a 3D arrangement of isolated sphericalmesopores, Adv. Funct. Mater. 13 (2003) 47–52.

[16] A. Gibaud, D. Grosso, B. Smarsly, A. Baptiste, J.F. Bardeau, F. Babon-neau, D.A. Doshi, Z. Chen, C. Jeffrey Brinker, C. Sanchez, Evaporation-controlled self-assembly of silica surfactant mesophases, J. Phys. Chem.B 107 (2003) 6114–6118.

[17] G.J. De, A.A. Soler-Illia, E.L. Crepaldi, D. Grosso, C. Sanchez, Block

[

[

[

[

[

[

[

[

[[

cknowledgements

The authors would like to acknowledge U.S. Army grantDASG60-03-C-0073. We also would like to thank Prof. Douguthven, Dr. Harry Deckman and Dr. Ben McCool for usefuliscussion and comments. Additional thanks are given to Prof.arl Tripp for the use of the ASAP 2010.

eferences

[1] C. Boissiere, M.U. Martines, A. Larbot, E. Prouzet, On the specificfiltration mechanism of a mesoporous silica membrane, prepared withnon-connecting parallel pores, J. Membr. Sci. 251 (2005) 17–28.

[2] C. Boissiere, U. Martines, Ultrafiltration membrane made with meso-porous MSU-X Silica, Chem. Mater. 15 (2003) 460–463.

[3] R.P. Singh, J.D. Way, S.F. Dec, Silane modified inorganic membranes:effects of silane surface structure, J. Membr. Sci. 259 (2005) 34–46.

[4] D.M. Ford, E.E. Simanek, D.F. Shantz, Engineering nanospaces: orderedmesoporous silicas as model substrates for building complex hybridmaterials, Nanotechnology 16 (2005) 458–475.

[5] A. Katiyar, L. Ji, P. Smirniotis, N.G. Pinto, Mesoporous molecular sievesfor size selective separation of biomolecules, in: F.T. Akin, Y.S. Lin(Eds.), Proceedings of Eighth International Conference on InorganicMembranes, 2004, pp. 63–66.

[6] J. Andersson, J. Rosenholm, S. Areva, M. Linden, Influences of mate-rial characteristics on ibuprofen drug loading and release profiles fromordered micro- and meso-porous silica matrices, Chem. Mater. 16 (2004)4160–4167.

[7] A.L. Doadrio, E.M.B. Sousa, J.C. Doadrio, J. Perez Pariente, I.Izquierdo-Barba, M. Vallet-Regi, Mesoporous SBA-15 HPLC evalua-tion for controlled gentamicin drug delivery, J. Control. Rel. 97 (2004)125–132.

copolymer-templated mesoporous oxides, Curr. Opin. Coll. Interf. Sci.8 (2003) 109–126.

18] D. Zhao, P. Yang, K.I. Margolese, B.F. Chmeklka, G.D. Stucky, Syn-thesis of continuous mesoporous silica thin films with three-dimensionalaccessible pore structures, Chem. Commun. (1998) 2499.

19] M. Klotz, A. Ayral, C. Guizard, L. Cot, Synthesis and characterizationof silica membranes exhibiting an ordered mesoporosity. Control of theporous texture and effect on the membrane permeability, Separ. Purif.Technol. 25 (2001) 71–78.

20] B.A. McCool, N. Hill, J. DiCarlo, W.J. DeSisto, Synthesis and character-ization of mesoporous silica membranes via dip-coating and hydrother-mal deposition techniques, J. Membr. Sci. 218 (2003) 55–67.

21] G.J.A.A. Soler-Illia, E.L. Crepaldi, D. Grosso, D. Durand, C. Sanchez,Structural control in self-standing mesostructured silica oriented mem-branes and xerogels, Chem. Commun. (2002) 2298–2299.

22] B. Smarsly, G. Xomeritakis, K. Yu, N. Liu, H. Fan, R.A. Assink,C.A. Drewien, W. Ruland, C.J. Brinker, Microstructural characteriza-tion of polystyrene-block-poly(ethylene oxide)-templated silica filmswith cubic-ordered spherical mesopores, Langmuir 19 (2003) 7295–7301.

23] T.E. Clark, H.W. Deckman, D.M. Cox, R.R. Chance, In situ determina-tion of the adsorption characteristics of a zeolite membrane, J. Membr.Sci. 230 (2004) 91–98.

24] F.P. Cuperus, D. Bargeman, C.A. Smolders, Permporometry, The deter-mination of the size distribution of active pores in UF membranes, J.Membr. Sci. 71 (1992) 57–67.

25] G.Z. Cao, J. Meijerink, H.W. Brinkman, A.J. Burggraaf, Permporom-etry study on the size distribution of active pores in porous ceramicmembranes, J. Membr. Sci. 83 (1993) 221–235.

26] A.J. Burggraaf, Fundam. Inorg. Membr. Sci. Technol. (1996) 331–433.27] G. Xomeritakis, S. Naik, C.M. Braunbarth, C.J. Cornelius, R. Pardey,

C.J. Brinker, Organic-templated silica membranes. I. Gas and vaportransport properties, J. Membr. Sci. 215 (2003) 225–233.