Soft matter, sol–gel process and external magnetic field to design macrocellular silica scaffolds

Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

Soft matter, sol–gel process and external magnetic field to design macrocellular silica scaffolds Florent Carna , Annie Colinb , V´eronique Schmitta , Fernando Leal Calderonc , R´enal Backova,∗ a

Centre de Recherche Paul Pascal, Department of Chemistry, 15 UPR 8641 CNRS, 115 Avenue Albert Schweitzer, 33600 Pessac, France b Laboratoire du Futur, UMR CNRS-Rhodia FRE2771, IECB, 2 Rue Robert Escarpit, 33607 Pessac, France c Laboratoire des Milieux Dispers´ es Alimentaires, ISTAB, Avenue des Facult´es, 33405 Talence, France Received 15 July 2004; accepted 20 December 2004 Available online 21 January 2005

Abstract Rational design toward macrocellular silica scaffolds has been reached. The first technique employed is based on the use of concentrated direct emulsions as a macroscale pattern where the final monolith textures can vary upon the oil volumic fraction, emulsification process, and pH conditions. Also, in order to improve void spaces alignment we can make the use of external magnetic field. In this issue, an hydrophobic Ferro-fluid replaces the oily phase. The second approach is rather focussed on a non-static patterning method that allows a complete control over width, length and curvature of the silica foam Plateau’s borders. Furthermore, those macroscale void space structurations are associated to secondary mesoscale porosity emerging from aqueous media lyotropic mesophase while the SiO4 tetrahedral statistic connectivity is offering a third microscale porosity. This void space texturation in registry provides finally hierarchically organized monolith-type materials. © 2004 Elsevier B.V. All rights reserved. Keywords: Sol–gel; Emulsion; Air–liquid foam

1. Introduction Taking into account the extraordinary catalogue of shapes that exist in nature, researchers are challenging to recover those morphologies, leading to the on-growing field of the so-called “bio-inspired” materials [1–5]. While micro- and mesoscale void space texturation is today a rather well established method [6–11], supermeso- or macroscale morphosynthesis appears more recently (i.e. pore sizes >50 nm) [12–16]. In this issue, monolith-type materials with hierarchical porosity attract considerable attention due to their wild scope of applications [17,18]. To achieve such high organization, patterns or templates at different length scales have to be used for controlling void space sizes, shapes and connectivity [19]. We present two complementary methods that allow ∗

Corresponding author. Tel.: +33 5 56 84 56 30; fax: +33 5 56 84 56 00. E-mail address: [email protected] (R. Backov).

0927-7757/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2004.12.034

preparing porous monoliths hierarchically textured with a strong control over the macroscopic void space sizes and shapes. In a first approach [20], hierarchical inorganic porous monoliths have been prepared using concentrated emulsion direct biliquid foams to tune macroscale void spaces of silica monoliths while providing mesoporous texture with micellar templates. The texture of these monoliths can be tuned by varying either the pH of the continuous aqueous phase, the emulsification process and the oil volumic fraction or by applying an external magnetic field when the dodecane is replaced with an hydrophobic Ferro-fluid. The second route, based on the use of a non-static patterning air–liquid foaming method [21], allows reaching higher macrocellular size, ranging from 50 ␮m to 500 ␮m. This strategy allows independent control of Plateau-border length and width. The Plateau-border length can be controlled by the size of the porous disk used for bubbling whereas the Plateau-border width can be tuned by varying the foam liquid fraction.

342

F. Carn et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

This dynamic approach leads to hierarchically organized materials with a very high specific surface associate to a macroscale morphologies tunable on the demand.

umn (25 cm × 25 cm × 60 cm high). Final foams are frozen over night and lyophilisation is then operated during 10 h. The resulting hybrid organic–inorganic monolith-type materials are finally thermally treated as described above in the text.

2. Experimental 2.3. Characterization Tetraethoxyorthosilane (TEOS) and tetradecyltrimethylammonium bromide (TTAB) 98% were purchased from Fluka, HCl 37% and dodecane 99% were purchased from Prolabo, perfluorohexane was purchased from Acrosorganics. Hydrophobic Ferro-fluid was kindly provided by the society ADEMTECH (Pessac, France). 2.1. Concentrated direct emulsion synthesis Procedures are based on the use of both micelles and direct emulsion templates. Typically, 5 g of tetraethoxyorthosilane is added to 16.5 g of tetradecyltrimethylammonium bromide aqueous solution at 35% in weight. The aqueous mixture is then brought to a pH value of 0.035 (5.84 ml of HCl 37%) leading, after the oil emulsification process, to the materials labelled as xSi-HIPE0.035. Detailed procedures can be find elsewhere when varying the pH and the emulsificatin process [20]. In a part of our study, we intend to tune the macroscopic void space texture and connectivity playing either with the emulsification process with the oil volumic fractions (ρo ). In this issue, 35 g, 40 g, 45 g and 60 g of dodecane are then emulsified drop by drop by hand in a mortar within the pH 0.035 aqueous media described above. The material presented in the Fig. 3 has been obtained by using an hydrophobic Ferro-fluid that replaces 35 g of the dodecane. The use of a viscous dispersed phase in high concentration state avoids droplets migration into more mobile regions during the sol–gel process. Contrary to other works [22,23], such characteristic allows to overcome the non-homogeneous macropores structuration hence forming homogeneous polydisperse emulsions. Also, in order to remove the organic supramolecular-type templates, the hybrid organic–inorganic materials were treated at 650 ◦ C during a period of 6 h. 2.2. Air–liquid foam synthesis Typically, sol–gel reactions are first started within a becher for a time period of 20 min. In this issue, TEOS (5 wt.%) is added to a TTAB solution (10 wt.%) where HCl has been previously added in order to reach 2 M HCl concentration. Then the air–liquid bubbling process is started within a Plexiglas column under specific sol debit (Q) [21]. Typically, foaming solutions are prepared by mixing a cationic surfactant (TTAB), tetraethoxyorthosilane, water and HCl. Foam is obtained from bubbling perfluorohexane saturated nitrogen through a capillary (hole diameter: 1 mm, 0.5 mm, 0.2 mm, and 0.1 mm) or a porous glass disk (porosity: 150–200 ␮m, 90–150 ␮m, and 40–90 ␮m) into the foaming solution. The reaction is taking place inside a Plexiglas col-

TEM experiments were performed with a Jeol 2000 FX microscope (accelerating voltage of 200 kV). The samples were prepared as follows: silica scaffolds in a powder state were deposited on a copper grid coated with a formvar/carbon membrane. SEM observations were performed with a Jeol JSM-840A scanning electron microscope operating at 10 kV. The specimens were gold-coated or carbon-coated prior to examination. Surface areas and pore characteristics at the mesoscale were obtained with a Micromeritics ASAP 2010 employing the Brunauer–Emmett–Teller (BET) method. Intrusion/extrusion mercury measurements were performed using a Micromeritics Autopore IV apparatus this to reach the scaffolds macrocellular cells characteristics. X-ray diffraction experiments were carried on a 18 kW rotating anode X-ray source (Rigaku-200) with use of Ge (1 1 1) crystal as monochromator. The scattered radiation was collected on a two dimensional detector (Imaging Plate system from Mar Research, Hamburg). The sample detector distance was 500 mm. 3. Result and discussion Those two strategies allow developing two set of highly organized materials with a complementary range of pore sizes and thus properties. The first strategy using concentrated direct emulsion leads to typical polymerized high internal phase emulsion (poly-HIPE)-type interconnected macroporous textures with polydisperse cell sizes ranging from 4 ␮m to 50 ␮m, while the air–liquid foam patterning approach allows obtaining macrocellular silica scaffolds with higher cell sizes, ranging from 50 ␮m to 500 ␮m, with the additional capability to control the cell morphologies. All materials are associated to typical vermicular-type mesoporosity [24,25]. The main critical factors involved with the macroscopic patterns effects both for the biliquid and air–liquid foams are discussed below. 3.1. Billiquid foam patterning approach: effect of oil volumic fraction Before going through the effect of the oil volumic fraction over the monolith textures it is important to note that two parameters are of great importance, namely pH and emulsification process. Those two parameters are fully discussed elsewhere [20] and will not be a part of the present discussion, we will rather focus on the effect of the oil volumic fraction. To accomplish this study, we did choose the synthesis pH conditions of 0.035 for the high condensation kinetic within

F. Carn et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

343

Fig. 1. SEM visualization of the inorganic monolith-type material macrostructure. (a) 1Si-HIPE0.035, (b) 2Si-HIPE0.035, (c) 3Si-HIPE0.035 and (d) 4SiHIPE0.035. The scale bar represents 70 ␮m.

such conditions. The results are depicted within the Fig. 1. We can first observed that whatever the oil volumic fraction conditions the general texture resemble to this “aggregated hollow spheres” appearance. For the xSi-HIPE0.035 series, two kind of cell junctions are observed; the internal windows that connect two adjacent cells (Fig. 1b, black arrow) and an external cell junction (or external porosity) emerging from the hollow spheres aggregation process (see white arrow in Fig. 1b). In this study, we have been played with the oil volumic fractions of the starting emulsions to tune the average macrocellular sizes thus controlling the connecting windows cell sizes. For the 1Si-HIPE0.035 and 2Si-HIPE0.035 porous monoliths, we observe almost the same macrocellular average cell sizes without any dramatic increase of the cell sizes windows (Fig. 1a and b), confirmed with mercury intrusion porosimetry experiment (Fig. 2). When we increase the starting emulsion oil volumic fraction, the macrocellular cell sizes diminish drastically (Fig. 1c and d). In fact, considering the rheology of the emulsion, it is well known that the viscosity of direct emulsions increases dramatically when the oil volumic fractions are reaching values above 0.74 [26,27]. For the starting emulsions of the materials 3Si-HIPE0.035 and 4Si-HIPE0.035, this phenomenon increases the stress applied on the oily droplets that induces smaller macrocellular cells [26] within the solid-state replica. This rheological effect has a strong impact on the cell windows taking into account

Fig. 2. Pore size distribution for the xSi-HIPe0.035 monoliths series as measured by mercury intrusion porosimetry. (a) 1Si-HIPE0.035, (b) 2SiHIPE0.035, (c) 3Si-HIPE0.035 and (d) 4Si-HIPE0.035. The continuous line is related to the internal porosity evolution with the increase of the oil volumic fraction, the dash line is related to the external porosity evolution with the increase of the oil volumic fraction.

344

F. Carn et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

Fig. 3. SEM visualization of the inorganic monolith-type material macrostructure obtained by varying the oily phase toward hydrophobic Ferro-fluid and exposing the sample within a magnetic field during the condensation process.

mercury intrusion porosimetry (Fig. 2). First, when the oil volumic fraction is the subject of a small increase from 1Si-HIPE0.035 (ρo = 0.67) to 2Si-HIPE0.035 (ρo = 0.70), both the largest cell junction and smallest cell junction sizes are increased from 1.4 ␮m to 1.6 ␮m and 21 nm to 45 nm, respectively. For those two oil volumic fraction values, the increase of the viscosity is not high enough to overcome the increase of the droplets size and we observe a simple effect of geometry. When the oil volumic fraction is increased-up from 2Si-HIPE0.035 to 3Si-HIPE0.035 (ρo = 0.73) and 4Si-HIPE0.035 (ρo = 0.78) the largest window cell sizes are decreasing from 1.4 ␮m to 0.5 ␮m and 0.25 ␮m, respectively whereas the average smallest cell junction size is weakly increasing from 25 nm to 50 nm for the 1Si-HIPE0.035 and 2Si-HIPE0.035, this value of 50 nm is constant still whatever the increase of the oil volumic fraction (Fig. 2b–d) when going from the materials, 2Si-HIPE0.035, 3Si-HIPE0.035 and 4Si-HIPE0.035, respectively. The decrease of the largest window cell from 1500 nm to 400 nm is certainly related to the decrease of the macrocellular cell diameters. Indeed, decreasing the macrocellular cell sizes reduces the characteristic size of the external porosity while weakly increasing the size of the window cell junctions when going from oil volumic fraction value from 0.67 to 0.70. The small increase of the smallest windows cell, when the oil volumic fraction is increasing from 0.67 to 0.70, might be related to two antagonist effects, namely the small increase of the droplet sizes (Fig. 1a and b), in this range of oil volumic fraction the viscosity is not yet increasing enough to induce a strong shear effect [26,27] and the decrease of the Plateau border thickness. Indeed, when the oil fraction increases above the 0.70 value, the thickness of those Plateau borders decreases, they become thinner and weaker and allowing the amount of macrocellular connecting-cells promoted during the shrinkage process to be enhanced. As a direct consequence when the oil volumic fraction is reaching values above 0.78, we do not obtained monolith-type materials but rather powders as the wall thickness does not offer mechanical strength strong enough to support the silica scaffolds which collapse.

Thermal treatment is employed to access mesoporosity leading to typical values around 800 m2 g−1 for this class of materials [20]. Overall, the process described above, provides materials associated to densities as low as 0.08 g cm−3 [20], which is comparable to values obtained for silica aerogel 0.1 g cm−3 [20]. Also, in order to align the macroscopic void spaces we can substitute dodecane with hydrophobic Ferro-fluid while maintaining the sample within a magnetic field during the condensation mechanism. Preliminary results depicted with Fig. 3 show that the preferential void spaces alignment is parallel to the imposed magnetic field. This process proposes a new insight where for the first time, beyond hierarchical texturation at the microscopic, mesoscopic and macroscopic length scales, external field is apply to assess macroscopic void space preferential orientation. 3.2. Air–liquid foam-patterning strategies: effect of the wetting flux and porous disks sizes in use The major cause of such foams destabilization is the drainage. In order to minimize this phenomenon, we wet the foam from above with the sol solution. A stationary regime is reached and the amount of solution injected at the top of the foam compensates the amount of liquid evacuated at the bottom. This method, allows us to prepare foams with homogeneous liquid fraction from the top to the bottom of the column and to control it. We recall that the liquid fraction (ρ) is the ratio of the volume of liquid present in the foam divided by the total volume of the foam. This technique permits us to tune the liquid fraction by varying the flux of sol at the foam’s top (Q). High fluxes induce high liquid fractions. The foam’s liquid fraction can be checked using conductivity measurements [28,29] and is furthermore well known to have a direct input on the air–liquid foam morphologies [30,31]. More precisely at high Q or at high ρ values, foam’s cells depict circular morphologies whereas at low flux or low fractions they present more polyhedral type morphologies. Hence, gas bubbles are soft materials. They distort at high

F. Carn et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

345

Fig. 4. (a) SEM image of a mineralized foam obtained with Q = 0.038 g s−1 and (b) SEM image of a mineralized foam obtained with Q = 0.192 g s−1 .

Fig. 5. (a) SEM image of a silica foam obtained with a bubbling apparatus pore sizes of 5 ␮m, (b) SEM image of a silica foam obtained with a bubbling apparatus pore sizes of 40 ␮m and (c) SEM image of a silica foam obtained with a bubbling apparatus pore sizes of 250 ␮m.

liquid fraction to adopt shape determined by surface tension and volume-packing constraints [32]. We wet the foam until the condensation of the silica begins noticeably. Using the drainage properties, we first mineralized foam’s cells at different ρ values while maintaining the same porous glass dick used to perform the bubbling process. We can note with

the Fig. 4 that the air–liquid foam’s water volume fraction appears as a very nice tool to tune silica macroscale void space morphologies. As a direct template of the metastable thermodynamic foam’s form, high liquid volume fraction results in a circular-type silica foam cell morphologies (Fig. 4b) whereas at lower liquid volume fraction they are modified to

Scheme 1. Scope over the macroscopic open-cell sizes and shapes obtained by varying the macroscopic thermodynamic metastable systems from concentrated direct emulsions to air–liquid foams.

346

F. Carn et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 263 (2005) 341–346

a more polyhedral-type shape (Fig. 4a). Also, as observed for the air–liquid foams, the silica Plateau-borders widths increase with an increase of the water volume fraction (Fig. 4 a and b). In order to improve this idea of controlled macroscale morphologies, we used different categories of glassy porous disks for the same liquid fraction volume, this time to tune the Plateau’s border lengths (Fig. 5). Whatever the silica scaffold morphologies discussed above, the associated skeletons do not collapse when exposed in air for a time period of six months. To remove the organic supramolecular-type templates, the hybrid organic–inorganic materials were calcined. This thermal treatment, beyond the mesoscale surfactant removal, induces a reduction of the Plateau’s border width by more or less 45% and 30% for the lengths [21], this phenomenon is in relation with the inorganic core sintering effect [33]. Final multi-scale porous materials depict typical mesoporosity around 900 m2 g−1 [21]. Taking into account, the two main strategies proposed here a wide range of well define macroscopic cell sizes [4–500 ␮m] can be reached (Scheme 1). The versatility of this combinatory strategy between soft-matter (lyotropic mesophases, emulsions, air–liquid foams), sol–gel process that might be associated to divers inorganic polymers (SiO2 [21], TiO2 [34], V2 O5 [35]) and application of an external field should lead to the emergence of complex architectures dedicated to specific applications as for instance, thermal and/or acoustic insulators, biologic cell promoters, photo-catalysis, filtrations, cathode for lithium batteries and so forth.

References [1] [2] [3] [4] [5] [6] [7]

[8] [9] [10]

[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

4. Conclusion In a first static method based on direct concentrated emulsion patterns, monolith-type materials associated to hierarchical porosity are reached. For this approach, future work will be dedicated to employ monodisperse concentrated direct emulsion toward a better understanding of both the internal and external cell junction morphosynthesis. New insight toward the capability of obtaining macroscopic void spaces alignment within monolith-type materials by combining hydrophobic Ferro-fluid, as an oily phase, with an external magnetic field has been demonstrated for the first time. The second non-static air–liquid foam templating route allows independent control over the silica foam Plateau-borders length, width and curvature. The Plateau-border length can be controlled by the size of the porous disk employed for the bubbling process, whereas varying the foam liquid fraction induces a control over the Plateau-border width. This dynamic approach leads to hierarchically organized materials with a very high specific surface associated to macroscale morphologies tunable “on demand”.

[22] [23] [24]

[25]

[26] [27] [28] [29] [30] [31] [32]

[33] [34] [35]

S. Mann, Nature 365 (1993) 499. D. Douglas, S. Mann, Nature 364 (1993) 430. S. Mann, G.A. Ozin, Nature 382 (1996) 313. I. Soten, G.A. Ozin, Curr. Opin. Colloid Interface Sci. 4 (1999) 325. S.A. Davis, S.L. Burkett, N.H. Mendelson, S. Mann, Nature 385 (1997) 420. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Nature 359 (1992) 710. J.S. Beck, J.C. Vartuli, J.W. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Scmitt, C.T.W. Chu, D.H. Olson, E.W. Sheppard, S.B. Mc Cullen, et al., J. Am. Chem. Soc. 114 (1992) 10834. S.A. Bagshaw, E. Prouzet, T. Pinnavaia, Science 269 (1995) 1242. T. Czuryszkiewicz, F. Kleitz, F. Sch¨uth, M. Lind´en, Chem. Mater. 15 (2003) 3704. S. Cabrera, J. El Haskouri, C. Guillem, J. Lattore, A. Beltr´an-Porter, D. Beltr´an-Porter, M.D. Marco, P. Amr´os, Solid State Sci. 2 (2000) 405. E.L. Crepaldi, G.J.D.A.A. Soler-Illia, D. Grosso, F. Cagnol, F. Ribot, C. Sanchez, J. Am. Chem. Soc. 125 (2003) 9770. B.T. Holland, C.F. Blanford, A. Stein, Science 281 (1998) 538. D. Zhao, P. Yang, B.F. Chmelka, G.A. Stucky, Chem. Mater. 11 (1999) 1174. P. Schmidt-Winkel Jr., W.W. Lukens, P. Yang, D.I. Margolese, J.S. Lettow, J.Y. Ying, G.A. Stucky, Chem. Mater. 12 (2000) 686. G.T. Chandrappa, N. Steunou, J. Livage, Nature 416 (2002) 702. H. Zhand, G.C. Hardy, M.J. Rosseinsky, A.J. Cooper, Adv. Mater. 15 (2003) 78. N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga, N. Tanaka, J. Chromatogr. A 797 (1998) 133. A.-M. Siouffi, J. Chromatogr. A 1000 (2003) 801. S. Mann, S.L. Burkett, S.A. Davis, C.E. Fowler, N.H. Mendelson, S.D. Simms, D. Walsh, N.T. Wilton, Chem. Mater. 9 (1997) 230. F. Carn, A. Colin, M.-F. Achard, E. Sellier, M. Birot, R. Backov, J. Mater. Chem. (2004) 1370. F. Carn, A. Colin, M.-F. Achard, H. Deleuze, Z. Saadi, R. Backov, Adv. Mater. 16 (2004) 140. A. Imhof, D.J. Pine, Nature 389 (1997) 948. G.-R. Yi, S.M. Yang, Chem. Mater. 11 (1999) 2322. S.A. Bagshaw, E. Prouzet, T.J. Pinnavaia, Science 269 (1995) 865; E. Prouzet, T.J. Pinnavaia, Angew. Chem., Int. Ed. 36 (1997) 516; C. Boissi`ere, A. Larbot, E. Prouzet, Chem. Mater. 12 (2000) 1937. P.A. Webbs, C. Orr, Surface area and pores structure by gas adsorption, in: Analytical Methods in Fine Particles Technology, Micromeritics, Norcross, GA, 1997. M. Mooney, J. Colloid Interface Sci. 6 (1951) 162. T.G. Mason, J. Bibette, D.A. Weitz, J. Colloid Interface Sci. 179 (1996) 439. R.A. Lemlich, J. Colloid Interface Sci. 64 (1978) 107. R. Phelan, D. Weaire, E.A.J.F. Peters, G. Verbist, J. Phys.: Condens. Matter. 8 (1996) 475. D. Weaire, N. Pittet, S. Hutzler, Phys. Rev. Lett. 71 (1993) 2670. V. Carrier, S. Destouesse, A. Colin, Phys. Rev. E 65 (2000), 0614041. D. Exerowa, P.M. Kruglyakov, in: D. M¨obius, R. Miller (Eds.), Foams and Foam Films Theory, Experiment, Application, Vol. 5, Elsevier, Amsterdam, 1998. C.J. Brinker, G.W. Scherer, Sol–Gel Science: the Physics and Chemistry of Sol–Gel Processing, Academic Press, San Diego, 1990. F. Carn, A. Colin, M.-F. Achard, H. Deleuze, R. Backov, Adv. Mat. 17 (2005) 62. F. Carn, N. Steunou, A. Colin, J. Livage, R. Backov, Chem. Mater., in press.