9092
Langmuir 2006, 22, 9092-9095
Biomimetic Growth of Silica Tubes in Confined Media Cle´mentine Gautier,†,‡ Pascal J. Lopez,‡ Miryana Hemadi,† Jacques Livage,† and Thibaud Coradin*,† Chimie de la Matie` re Condense´ e de Paris, UMR CNRS-7574, UniVersite´ Pierre et Marie Curie Paris VI, 4 place Jussieu (T54-E5), 75252 Paris Cedex 05, France, and Diatoms Signalization and Morphogenesis, CNRS-FRE 2910, Ecole Normale Supe´ rieure, 75005 Paris, France ReceiVed June 12, 2006. In Final Form: September 5, 2006 Polymer membranes were used as biomimetic environments to study the effect of confinement on silica formation. Within membrane pores, silica tubes were formed, consisting of a dense silica shell incorporating nanoparticle aggregates. The shell structure does not depend on the membrane pore size, suggesting that its formation proceeds via interfacial interactions with the pore surface. In contrast, the size of primary nanoparticles within core aggregates is influenced by pore dimensions, indicating an effect of confinement on the diffusion-limited growth of silica. A parallel can be drawn with reported roles of confinement in biomineralization processes, providing a basis for future developments in biosilicification mimetic approaches and biofunctional nanomaterials design.
Introduction When trying to design suitable biomimetic systems to study the growth of inorganic materials, chemists often face the complexity of biomineralization processes. Hopefully, most of these processes exhibit some similarities that provide fruitful guides to elucidate the key parameters that control mineral formation in living organisms. More specifically, biomineralization processes involve a nucleation/growth reaction from inorganic precursors within a confined space (vesicle, organic matrix, etc.), in the presence of templating molecules.1,2 For instance, in the case of diatoms and desmosponges, two classes of biosilicifying organisms, silica formation occurs in a specific silica deposition vesicle (SDV) where silica precursors interact with templating macromolecules.3,4 Until now, biomimetic approaches of silicification have mainly focused on the study of interactions arising between silica precursors and macromolecules, either extracted from living organisms or selected/synthesized as suitable models.5-7 However, in contrast to other biominerals, such as calcium carbonate,8 attempts to mimic the confined space of the SDV have only been sparingly studied.9 In this context, we have recently reported the growth of silica nanoparticles within multilamellar phospholipid vesicles, showing that the size of the particles was controlled by the interlamellar distance.10 A few years ago, Meldrum et al. presented an original strategy to mimic calcium carbonate growth in confined media.11 This strategy relies on the use of polymer porous membranes for the * Corresponding author. Tel: +33-1-44275517. Fax: +33-1-44274769. E-mail:
[email protected]. † Universite ´ Pierre et Marie Curie Paris VI. ‡ Ecole Normale Supe ´ rieure. (1) Lowenstam, H. A.; Weiner, S. On Biomineralization; Oxford University Press: New York, 1989. (2) Mann, S. Biomineralization. Principles and Concepts in Bioinorganic Materials Chemistry; Oxford University Press: New York, 2001. (3) Pickett-Heaps, J.; Schmid, A.-M. M.; Edgar, L. A. Prog. Phycol. Res. 1990, 7, 1-168. (4) Weaver, J. C.; Morse, D. E. Microsc. Res. Technol. 2003, 62, 356-367. (5) Sumper, M.; Kroger, N. J. Mater. Chem. 2004, 14, 2059-2065. (6) Lopez, P. J.; Gautier, C.; Livage, J.; Coradin, T. Curr. Nanosci. 2005, 1, 73-83 (7) Patwardhan, S. V.; Clarson, S. J.; Perry, C. C. Chem. Commun. 2005, 1113-1121. (8) Meldrum, F. C. Int. Mater. ReV. 2003, 48, 187-224. (9) Mann, S.; Perry, C. C. AdV. Inorg. Chem. 1991, 36, 137. (10) El Rassy, H.; Belamie, E.; Livage, J.; Coradin, T. Langmuir 2005, 21, 8584-8587. (11) Loste, E.; Meldrum, F. C. Chem. Commun. 2001, 901-902.
precipitation of amorphous calcium carbonate via a double diffusion method. A similar process was also applied to polymer replicas of sea urchin skeletal plates, allowing the formation of crystalline calcium carbonate with complex morphologies.12 Because silica formation is obtained from a single molecular precursor, we developed a different approach to mimic silica growth in a confined environment. Polymer membrane multiple impregnations at suitable pH were used instead of the doublediffusion technique. After silica condensation, membrane dissolution allows the recovery of silica micro- and nanotubes partially filled with silica nanoparticle aggregates. Transmission electron microscopy (TEM) analysis of the size of the primary nanoparticles constituting these different phases suggest that the shell formation results from interfacial interactions with the internal pore surface, whereas the inner-tube aggregate growth is influenced by the confinement conditions, through the decrease of particle diffusion coefficients. These results validate the suitability of our model to mimic the growth of silica in confined environments and open new perspectives to design more complex biomimetic systems that may find applications as biofunctional nanomaterials. Experimental Section Silicate solutions were prepared by dilution of a water glass solution (27% SiO2, 10% NaOH from Riedel-de Hae¨n) to reach a 0.5 M concentration and then acidification to pH 5 using HCl (3M). This pH value was selected because it was reported that silica deposition occurs in acidic conditions in diatoms.13 Moreover, at pH 5, the condensation of silica is slow enough to be controlled by biopolymers.14 The silicate concentration was fixed at 0.5 M to obtain a gelation time compatible with the impregnation experiments. A polycarbonate membrane (1.2, 0.4, or 0.2 µm in pore diameter from Millipore) was deposited on a Bu¨chner funnel, and 50 mL of the freshly prepared silicate solution was filtered through the membrane. After the filtration of 40 mL, the vacuum was withdrawn, and the wet membrane was recovered and immersed in 10 mL of the same sodium silicate solution. A silica gel was formed over the whole solution volume after ∼10 min. The membrane was withdrawn and washed thoroughly with water to get rid of most of the silica deposited on the membrane surface. A similar process was repeated up to five times (five impregnation steps). All experiments were performed at room temperature. To observe silica formed within the pores, (12) Park, R. J.; Meldrum, F. C. J. Mater. Chem. 2004, 14, 2291-2296. (13) Vrieling, E. G.; Gieskes, W. W. C.; Beelen, T. P. M. J. Phycol. 1999, 35, 548-559. (14) Coradin, T.; Bah, S.; Livage, J. Colloids Surf., B 2004, 35, 53-58.
10.1021/la061674b CCC: $33.50 © 2006 American Chemical Society Published on Web 09/21/2006
Letters
Langmuir, Vol. 22, No. 22, 2006 9093
Figure 1. SEM images of the silica filling of pores for a 1.2 µm membrane after (a,d) one, (b,e) three, and (c,f) five impregnations as observed from (a-c) top and (d-f) transversal views. The circle in panel c and the cylinder in panel f indicate the initial pore dimensions. (scale bar: 1 µm). mineralized membranes were freeze-dried, and the polycarbonate matrix was dissolved in an excess of CHCl3.11 The resulting powders were studied by scanning electron microscopy (SEM) and TEM, either deposited on a sample holder/grid or after ultrathin sectioning. Primary nanoparticle sizes were determined using the MetaMorph Imaging System software. Statistical analyses were performed from data obtained from at least three different images with 30 particles measured for each image.
Figure 2. SEM images of silica formation in 1.2 µm membrane: (a) full-length tube, (b) tube aperture, and (c) overview showing fractured tubes. As a comparison, an SEM image of silica formed on the membrane surface is shown in panel d.
Results and Discussion First attempts to induce silica formation within membrane pores were performed by soaking the polycarbonate membrane in an acidified silicate solution and waiting for gel formation in the bulk solution. However, no silica could be observed within the porous network. We therefore investigated the filtration of the silicate solution through the membrane to ensure that the precursors penetrated the pores before gel formation. The formation of silica within the 1.2 µm membrane pores was first followed by SEM. While the pore openings did not appear to be significantly modified after the first impregnation, silica particles could be observed on the pore internal surface after three steps, while the pore openings were completely full after five steps (Figure 1a). This evolution was confirmed by observations of pore interior after lateral membrane sectioning. Silica deposition is hardly visible on the pore surface at step 1, and further pore filling by aggregates of particles with diameters in the 50-100 nm range is observed after step 3 and even more so at step 5 (Figure 1b). After dissolution of the polycarbonate membrane in CHCl3, the silica phase could be recovered as microtubes with diameters between 0.8 and 1.1 µm and lengths up to 10 µm, fitting the dimensions of the porous template (Figure 2a). Elemental analyses performed by the energy-dispersive X-ray (EDAX) technique on these tubes yielded a Na/Si ratio below 1:10, indicating that they mainly consist of silica. At the SEM microscale, after five impregnations, the tube surface appears rather smooth, except for unevenly distributed large granules. The ends of the longest tubes appear hollow, with a shell thickness of ∼50 nm (Figure 2b), while the shortest fragmented ones show the presence of internal particles, 50-100 nm in diameter (Figure 2c), suggesting that tubes were uncapped during the membrane dissolution process. The granules are also observed independently of the tubes and correspond to the silica deposit observed on the membrane surface (Figure 2d). Since confinement effects on colloids behavior are drastically enhanced when reactor dimensions get closer to nanoparticle size,15 similar experiments were performed for 0.4 and 0.2 µm (15) Happel, H.; Brenner, H. Low Reynolds Number Hydrodynamics; Prentice Hall: Engelwood Cliffs, NJ, 1965.
Figure 3. SEM images of silica formation in 0.4 µm (a) and 0.2 µm (b) tubes. Inset shows fractured nanotubes (scale bar: 500 nm).
pore size membranes. Nanotubes are obtained whose diameters closely fit the membrane dimensions (0.3-0.4 and 0.1-0.2 µm, respectively) (Figure 3). However, in contrast to the 1.2 µm membrane, long tubes could only be observed after three impregnations, possibly because of the decrease in stability of the silica network with diameter. For both pore sizes, the silica shell thickness was ∼50 nm, while the internal aggregate size was ∼50 nm in diameter, in close similarity with tubes obtained at 1.2 µm. At the TEM nanoscale, all tubes present zones of different contrast, suggesting variations in filling content and/or the presence of granular particles on the surface (Figure 4a,b). However, images of the tubes are far less contrasted at the smallest pore diameters because of the smaller thickness of the samples and possibly because of the better homogeneity of the structure. For the three samples, the size of the aggregated silica nanoparticle dp could be evaluated using image-analysis software on both tube external surface from TEM images at higher resolution (Figure 4c) and within the tubes on ultrathin sections of the materials (Figure 4d), with the corresponding values being gathered in Table 1. It appears that nanoparticles forming the tube shell are similar in size to those found on the membrane surface (∼9 nm), whereas they appear larger in the tube interior, especially for the largest pores (∼15 nm). The formation of silica from sodium silicate solution is welldocumented.16-18 Water glass solutions consist of monomers Si(OH)4, oligomers, and subcolloidal species that condense to form primary nanoparticles ∼1-2 nm in size. In moderately (16) Iler, R. K. The Chemistry of Silica; John Wiley & Sons: New York, 1979. (17) Wijnen, P. W. J. G.; Beelen, T. P. M.; Rummens, K. P. J.; Saeijs, H. C. P. L.; Van Santen, R. A. J. Appl. Crystallogr. 1991, 24, 759-764. (18) Gerber, Th.; Himmel, B.; Hu¨bert, C. J. Non-Cryst. Solids 1994, 175, 160-168.
9094 Langmuir, Vol. 22, No. 22, 2006
Letters
of larger particles to the detriment of the smallest ones.24 The latter argument indicates that, as membrane pore size decreases, gel formation (i.e., larger particle aggregation) should be favored compared to primary nanoparticle growth (i.e., smaller nanoparticles aggregation). In other words, the size of the primary nanoparticles that form the gel is expected to decrease with confinement. This is observed for 0.2 µm where a primary particle size of ∼10 nm is observed, compared to ∼15 nm for 1.2 and 0.4 µm. A contribution of diffusion coefficient decrease upon confinement is also possible since, taking into account the 50 nm shell of the tubes, a dw/dp ) 10 value is reached for dp ) 10, 20, and 110 nm, for 0.2, 0.4, and 1.2 µm pore membranes, respectively.
Conclusion
Figure 4. TEM micrographs of (a) 1.2 µm and (b) 0.2 µm tubes (scale bar: 500 nm). Higher magnification images showing external (shell) primary nanoparticles (c) (scale bar: 50 nm) and ultrathin sections, and images showing internal (core) primary nanoparticles (d) (scale bar: 100 nm) are also shown. Table 1. Diameters dp of Silica Primary Nanoparticles Constituting Tube Shell and Core for Different Membrane Pore Sizes and for the Surface-Deposited Gela pore size/µm dp (shell) /nm dp (core)/nm a
1.2
0.4
0.2
gel
8.4 [0.9] 14.2 [1.5]
9.2 [1.0] 16.8 [2.8]
9.8 [1.9] 10.8 [1.0]
8.8 [1.6]
Corresponding standard deviations are given in the brackets.
acidic media (pH 5-6), these particles tend to grow via particleparticle aggregation with possible intraparticle condensation leading to larger primary nanoparticles.17 These particles can further aggregate until gel formation. These processes were shown to occur mainly via diffusion-limited aggregation.17,18 After a first impregnation step, a dense silica coating, ∼50 nm in thickness, is observed on the pore internal surface. As a comparison, the silica gel deposited on the membrane surface is composed of large globular particles, ∼200 nm in size. However, both materials are formed from similar primary particles, ∼9 nm in diameter. This suggests that silica formation mainly results from interactions with the polycarbonate surface. Indeed, the ability of neutral polymers to interact with silica particles via H-bond formation has already been reported.16,19 The formation of a dense monolayer-like silica coating on the pore internal surface may be attributed to the reported effect of surface curvature on particle growth and packing, showing that negative curvatures favor dense-packed arrays of monodisperse particles.20,21 After the silica coating deposition, tubes are progressively filled from the initial shell to the core with silica nanoparticle aggregates. These particles are loosely packed within the tube interior, in contrast to the dense coating deposited on the pore internal surface, in agreement with the absence of direct contact with polycarbonate surface.The effect of confinement on colloid motion and aggregation has been studied both theoretically and experimentally.15,22-24 It was shown that the diffusion coefficient of particles of diameter dp confined between two walls separated from a distance dw significantly decreases for dw/dp values below ∼10.23 Moreover, confinement appears to favor the aggregation (19) Shchipunov, Y. A.: Karpenko, T. Y. Langmuir 2004, 20, 3882-3886. (20) Peczak, P.; Grest, G. S.; Levine, D. Phys. ReV. E 1993, 48, 4470-4481. (21) Rubinstein, M.; Nelson, D. R. Phys. ReV. B 1983, 28, 6377-6386.
In biomineralization processes, confinement has many functions, including spatial delineation to control the size and shape of the growing mineral phase, mineral nucleation inducement by providing specific surface sites, and diffusion-limited ion flow to control solution compositions and, hence, the nucleation/growth process.2 The first two aspects are illustrated in this work as tube shells are formed via interfacial interactions between confined space boundaries and silica-growing particles, more specifically via curvature effects. The last effect is present here, although in a different manner, as the control of silica formation is not achieved via ion diffusion through a permeable barrier but via an intrinsic influence of confinement on the diffusion-limited aggregation of primary nanoparticles. In this context, the striking transition observed between 0.4 and 0.2 µm membranes nicely illustrates the strong sensitivity of these confinement effects on the dimensions of the reactor.23 As mentioned earlier, another key aspect of biomineralization is the presence, within the confined space, of soluble macromolecules that are also involved in the control of the growth of the mineral phase. In this context, we are currently studying the influence of the preimmobilization of polyamines, which mimic some diatom extracts and are well-known to interact with silica precursors,5-7 in the membrane pores on the formation of silica. Finally, it is worth noting that inorganic nanoporous membranes have already been used as templates for the formation of silica nanotubes, but none of these approaches involved conditions resembling those observed in living organisms.25 For instance, Stucky et al.26 showed that nanoconfinement could influence the meso-organization of silica-copolymer composites with a sharp transition from cylindrical to spherical morphologies at a defined pore diameter. However, this effect was analyzed in terms of polymer self-assembly induced by solvent evaporation, leading to one-dimensional nucleation and growth mechanisms. In contrast, the here-described approach occurs in a threedimensional reactor at a constant water volume, and confinement effects are observed for purely inorganic systems. Finally, the fact that our process is closer to biocompatible conditions, together with the possibility to fill interior of the tubes with additional silica grains, opens the route to the design of novel biofunctional nanomaterials.27 (22) Lancon, P.; Batrouni, G.; Lobry, L.; Ostrowsky, N. Europhys. Lett. 2001, 54, 28-34. (23) Benesch, T.; Yiacoumi, S.; Tsouris, C. Phys. ReV. E 2003, 68, 021401. (24) Bentz, J. L.; Kozak, J. J. Phys. ReV. E 2006, 73, 011414. (25) (a) Martin, C. R. Science 1994, 266, 1961-1966. (b) Zhang, M.; Bando, Y.; Wada, K. J. Mater. Res. 2000, 15, 387-392. (c) Kovtyukhova, N. I.; Mallouk, T. E.; Mayer, T. S. AdV. Mater. 2003, 15, 780-785. (26) Wu, Y.; Cheng, G.; Katsov, K.; Sides, S. W.; Wang, J.; Tang, J.; Fredrickson, G. H.; Moskovits, M.; Stucky, G. D. Nat. Mater. 2004, 3, 816. (27) Mitchell, D. T.; Lee, S. B.; Trofin, L.; Li, N.; Nevanen, T. K.; So¨derlund, H.; Martin, C. R. J. Am. Chem. Soc. 2002, 124, 11864-11865.
Letters
Supporting Information Available: SEM images of tubes recovered after one impregnation in a 1.2 µm membrane; SEM and TEM images of silica tubes recovered after membrane dissolution for
Langmuir, Vol. 22, No. 22, 2006 9095 different membrane pore sizes. This material is available free of charge via the Internet at http://pubs.acs.org. LA061674B
