Superstructures of Temporarily Stabilized Nanocrystalline CaCO3

The polymers furthermore temporarily stabilize CaCO3 nanocrystals, which are formed by slow CO2 ..... KONA Powder and Particle Journal 2014 31 (0), 15...
0 downloads 0 Views 255KB Size
Download PDF
Langmuir 2004, 20, 991-996

991

Superstructures of Temporarily Stabilized Nanocrystalline CaCO3 Particles: Morphological Control via Water Surface Tension Variation Jan Rudloff and Helmut Co¨lfen* Max-Planck-Institute of Colloids and Interfaces, Colloid Chemistry Research Campus Golm, Am Mu¨ hlenberg, 14424 Potsdam, Germany Received September 30, 2003. In Final Form: November 4, 2003 In this paper, the formation of different complex morphologies of nanocrystalline CaCO3 under the control of double hydrophilic block copolymers (DHBCs) carrying phosphate groups is described. The DHBCs consist of a poly(ethylene glycol) (PEG) block and a pendant poly[2-(2-hydroxy ethyl)ethylene] block with different degrees of phosphorylation up to 40%, some of which show surface activity. The polymers furthermore temporarily stabilize CaCO3 nanocrystals, which are formed by slow CO2 evaporation from a supersaturated Ca(HCO3)2 solution (Kitano method). The polymers are active down to concentrations of 10-4 g/L. In dependence of the nature and concentration of the DHBC, tunable complex shuttlecock flowerlike and other superstructures are formed, which are aggregates of CaCO3 vaterite nanoparticles with an enhanced stability of at least 2 months. It is shown that the aggregation starts around template CO2 gas bubbles at the air/water interface. The size and morphology of the growing aggregates depends on the polymer concentration, phosphorylation degree, and water surface tension. The latter determines when the aggregate sinks to the bottom, interrupting the further growth process. Variation of the water surface tension by addition of the nonionic surfactant Antharox CO880 also allows a variation of the aggregate morphology, thus implying the described method as simple and versatile for the generation of complex CaCO3 morphologies.

Introduction Natural organisms are able to generate crystalline materials with complex morphologies at ambient temperature (e.g., 0-40 °C) in water by the process of biomineralization.1,2 These conditions are in contrast to the formation of materials under hydrothermal conditions in geological or inorganic synthesis processes.3,4 For the controlled crystal morphogenesis in organisms, various mechanisms are applied, including the template-directed aggregation of nanoparticles. Examples for this are the magnetite nanocrystal chains in magnetotactic bacteria5 or calcite single crystal arrangements as gravity sensors in the inner ear of vertebrates.6 To mimic such structure formation, two main requirements have to be fulfilled:7 (1) a template structure is needed that can be as simple as a gas bubble, such as in the present study, and (2) nanosized building units need to be provided that then aggregate in a template-directed fashion to form a complex superstructure. Whereas the first requirement can be easily fulfilled and has already been described, for example, by using microemulsion droplets as templates,8 the second one may be problematic because it requires a delicate balance of temporary colloidal stabilization. Here, microemulsion-based aggregation to complex structures * Author to whom correspondence should be addressed. E-mail: [email protected]. (1) Lowenstam, H. A.; Weiner, S. On biomineralization; Oxford University Press: NewYork, 1989. (2) Mann, S.; Webb, J.; Williams, R. J. P. Biomineralization; VCH: Weinheim, 1989. (3) Sabins, F. F. Ore Geol. Rev. 1999, 14, 157-183. (4) Yu, S. H. J. Ceram. Soc. Jpn. 2001, 109, S65-S75. (5) Blakemore, R. P.; Frankel, R. B. Sci. Am. 1981, 245, 58. (6) Mann, S.; Parker, S. B.; Ross, M. D.; Skarnulis, A. J.; Williams, R. J. P. Proc. R. Soc. London 1983, B218, 415. (7) Mann, S. Biomineralization - Principles and Concepts in Bioinorganic Materials Chemsitry; Oxford University Press: New York, 2001. (8) Walsh, D.; Lebeau, B.; Mann, S. Adv. Mater. 1999, 11, 324-328.

involving mesoscopic transformations was already described for vaterite,9 but the temporary stabilization of nanoparticulate building units in an aqueous environment is not an easy task. It can advantageously be fulfilled by tailor-made double hydrophilic block copolymers (DHBCs; for a review, see ref 10), which can be engineered in a way that the stabilizing block has just the length to provide a weak steric colloidal stabilization to allow for subsequent higher order assembly. For example, stabilizing blocks based on poly(ethylene glycol) (PEG) have molar masses in the range of 2000-6000 g/mol. The temporary colloidal stabilization has been demonstrated for spheres and dumbbells of CaCO3 ,11,12 and BaSO4 ,13,14 and otherwise shaped BaSO4 assemblies15 or hollow spherical CaCO3 morphologies.11,16 Especially phosphorylated block copolymers turned out to be very effective for the generation of complex morphologies in a preliminary crystallization study of CaCO3.17 This is not particularly surprising because some of the active biomineralization proteins were found to have a high content of serine units, which can be phosphorylated by casein kinases I and II. In addition, these biomolecules show at the same time high aspartic acid contents and occur in different classes of proteins, namely, dentin sialoprotein, dentin phosphoryn, or special glycoproteins. Specifically, rat dentin phosphoprotein (9) Li, M.; Mann, S. Adv. Funct. Mater. 2002, 12, 773-779. (10) Co¨lfen, H. Macromol. Rapid Commun. 2001, 22, 219-252. (11) Co¨lfen, H.; Antonietti, M. Langmuir 1998, 14, 582-589. (12) Co¨lfen, H.; Qi, L. Chem.sEur. J. 2001, 7, 106-116. (13) Qi, L.; Co¨lfen, H.; Antonietti, M. Angew. Chem., Int. Ed. 2000, 3, 604-607. (14) Qi, L.; Co¨lfen, H.; Antonietti, M. Chem. Mater. 2000, 12, 23922403. (15) Robinson, K. L.; Weaver, J. V. M.; Armes, S. P.; Marti, E. D.; Meldrum, F. C. J. Mater. Chem. 2002, 12, 890-896. (16) Qi, L.; Li, J.; Ma, J. Adv. Mater. 2002, 14, 300-303. (17) Rudloff, J.; Antonietti, M.; Co¨lfen, H.; Pretula, J.; Kaluzynski, K.; Penczek, S. Macromol. Chem. Phys. 2002, 203, 627-635.

10.1021/la0358217 CCC: $27.50 © 2004 American Chemical Society Published on Web 12/23/2003

992

Langmuir, Vol. 20, No. 3, 2004

Table 1. Overview of the Used Polymers, with Different Degrees of Phosphorylationa

a The numbers in parentheses represent the number of repeating units. Abbreviations: PEG ) poly(ethylene glycol), PHEE ) poly[2(2-hydroxy ethyl)ethylene], and PGL ) poly(glycidol).

contains 43% Ser and 31% Asp,18 and examples for dentin phosphoryns are the rat phosphoryn with 56% Ser and 32% Asp19 and the human dentin phosphophoryn with 58% Ser and 26% Asp.20 Finally, a molluscan shell glycoprotein with 32% Ser and 20% Asp21 can be mentioned as an example for special glycoproteins. The previously reported phosphorylated block copolymers17 provided DHBCs with well-defined block lengths and phosphorylation degrees (Table 1). These polymers were applied as models of the just named biomolecules to mimic the process of biomineralization in vitro in the herereported detailed study. The generation of complex CaCO3 morphologies by CO2 gas bubble templates and the role of the soluble polymeric additive is the main focus in this detailed study to shed more light onto the role of the socalled soluble matrix in biomineralization processes. Experimental Section Polymer Syntheses. The set of PEG-based DHBC model polymers was synthesized by anionic polymerization and functionalized by polymer analogue reactions, as described earlier.17,22 These polymers are defined model structures in terms of block lengths, molar mass distributions, and degree of functionalization. A particular advantage is that the phosphorylation degree can be varied without changing the polymer backbone so that the role of the functional groups can be elucidated. Mineralization Process. For the mineralization of CaCO3 in the presence of different DHBCs, the Kitano method was applied.23 The Kitano technique runs on a longer time scale (hours to days) in order to exclude kinetic effects in the precipitation experiment. A supersaturated solution of CaCO3 was prepared (18) Ritchie, H. H.; Wang, L. H. J. Biol. Chem. 1996, 271, 2169521698. (19) Ritchie, H. H.; Wang, L. H.; Knudtson, K. Biochim. Biophys. Acta 2001, 1520, 212-222. (20) Gu, K. N.; Chang, S. R.; Slaven, M. S.; Clarkson, B. H.; Rutherford, R. B.; Ritchie, H. H. Eur. J. Oral Sci. 1998, 106, 10431047. (21) Sarashina, I.; Endo, K. Mar. Biotechnol. 2001, 3, 362-369. (22) Kaluzynski, K.; Pretula, J.; Lapienis, G.; Basko, M.; Bartczak, Z.; Dworak, A.; Penczek, S. J. Polym. Sci., Part A: Polym. Chem. 2001, 39, 955-963. (23) Kitano, Y.; Park, K.; Hood, D. W. J. Geophys. Res. 1962, 67, 4873.

Rudloff and Co¨ lfen by bubbling CO2 gas through a slurry of 5 g of CaCO3 in 4 L of pure water for 60 min to shift the CaCO3/Ca(HCO3)2 equilibrium toward the more water-soluble Ca(HCO3)2. The CaCO3 was then subsequently filtered off, and CO2 was bubbled again through the solution for another 30 min to dissolve remaining CaCO3 particles. The solution was left in unsealed vessels (100-mL beaker, diameter 4 cm, height 7.95 cm). Carbon dioxide is formed on a time scale of hours to days, and the equilibrium shifts back to CaCO3, under generation of a CaCO3 supersaturation in the aqueous phase. Precipitation of CaCO3 starts at the air/water interface, where the lowest concentration of CO2 is found. The experiments run on a typical time scale of 1-2 days. Typical rhombohedral crystals of calcite were obtained in the absence of additives, whereas polymeric additives yield vaterite as described earlier.17 For the surface tension variation experiments, the nonionic Antarox CO880 surfactant (GAF Corporation) was applied, which is an oligoethyleneoxide adduct to nonylphenol (MW ) 880 g/mol). The surfactant was added to the readily formed Kitano solution until the surface tension of the solution had the desired value. Sample Preparation for Microscopy. Both light microscopy and scanning electron microscopy (SEM) were applied to all samples. The use of the light microscopy technique is necessary to prove that the SEM micrographs show real structures instead of drying artifacts resulting from the sample preparation. The SEM measurements were performed on a DSM 940A (Carl Zeiss, Jena), and the pictures were taken with a digital camera connected to the SEM. The crystals were directly taken from the solution. Most of the solution was soaked off the sample stub using tissue paper to minimize additional crystallization due to drying on the SEM stub. Light microscopy images were taken in solution at the same time with an Olympus BX50 microscope. Wide-Angle X-ray Scattering (WAXS) Measurement and Sample Preparation. The solution containing the crystals was filtered, and the crystals were collected and dried in a desiccator over phosphorus pentoxide. The WAXS mesasurements on the CaCO3 powders were carried out with a Bruker model D8 Advance at a wavelength of 0.154 nm (Cu KR) in reflective mode. Surface Tension Measurement. The measurements were done with a Lauda (model TD1) at room temperature using the Wilhelmy Plate technique and aqueous polymer solutions with different concentrations. Thermogravimetry Analysis (TGA). TGA was done with a Netzsch (model TG 209). The dried samples in a quantity of 1-5 mg were heated from 25 °C to 900 °C with a rate of 20 K/min.

Results and Discussion The mineralization of CaCO3 using the Kitano technique in the presence of different DHBCs resulted in particles with different crystal structures and particle morphologies. The results of the mineralization processes are systematically influenced by structural parameters of the polymer and experimental parameters. Influence of the Polymer (Block Length and Functionalization). The use of polymers with different degrees of phosphorylation in the Kitano solution with polymer concentrations of 1 g/L resulted in minerals with different particle morphologies. PEG(84)-b-PHEE(13) {PHEE ) poly[2-(2-hydroxy ethyl)ethylene]} polymers with a degree of phosphorylation in the range of 30-40% generate well-defined shuttlecock-like structures, with a narrow particle size distribution. The size of the shuttlecocks is around 50 µm. The particles consist of a core with diameters around 10 µm and an outer sphere of a featherlike structure. Polymers with a lower degree of phosphorylation in the range of 5-10% form minerals with structures such as hollow half spheres with sizes in the range of 5-20 µm. This already implies that the phosphorylation degree of the polymer has a strong influence on the observed crystal morphologies, expressed by a transition of shuttlecock morphologies to hollow halfsphere morphologies upon a decrease of the phosphorylation degree (Figure 1).

Nanocrystalline CaCO3 Particle Morphology Control

Langmuir, Vol. 20, No. 3, 2004 993

Figure 1. Dependence of the obtained CaCO3 morphologies on the PEG(84)-b-PHEE(13) phosphorylation degree at a polymer concentration of 1 g/L: (a) 40, (b) 30, (c) 10, and (d) 5% degree of phosphorylation (scale bars 50, 200, 50, and 20 µm).

Figure 2. Dependence of the obtained CaCO3 morphologies on the PEG(84)-b-PHEE(13)-(30%) concentration: (a) 100, (b) 10-1, (c) 10-2, (d) 10-3, (e) 10-4, and (f) 10-5 g/L (scale bars 200 µm).

On the other hand, the longer PEG(133)-b-PHEE(38) polymer with 10 and 25% phosphorylation degrees did not show such a rich morphological variation and resulted in a “fried egg” vaterite morphology (see Supporting Information, Figure 1). The core in the center of the structure is more distinctive, and the outer sphere is smaller. The size of these structures is again in the range of 20-50 µm. Diffraction analysis (WAXS) of these materials reveals vaterite as a crystal structure. The peak width analysis reveals that these materials are built of nanocrystalline vaterite particles with a primary crystallite size of 20-30 nm. This means that the observed complex morphologies represent defined aggregates of vaterite nanoparticles. Vaterite is a thermodynamically metastable crystal structure of CaCO3 and is usually converted to calcite on a time scale of hours.24 In the here-reported special case, the polymer has a stabilizing effect for the vaterite crystal structure. After 2 months, the crystal structure and primary particle size of the CaCO3 materials was found to be unchanged, checked by repetition of the WAXS measurements. (24) Richter, A.; Petzold, D.; Hofmann, H.; Ullrich, B. Chem. Tech. 1996, 48, 271-275.

Concentration Dependence of the Superstructure Morphologies. The influence of the polymer concentration onto the nanocrystal superstructures was checked in parallel experiments with PEG(84)-b-PHEE(13)-(30%) polymer concentrations ranging from 10-5 g/L to 100 g/L. It is remarkable that the polymer influence is not lost until concentrations down to 10-4 g/L (0.1 ppm additive concentration). Polymer concentrations lower than 10-4 g/L resulted in the well-known rhombohedral calcite morphologies (15 µm; see Figure 2). This means a transition of the particle morphologies can be achieved by variation of the polymer concentration (Figure 2a-f) in addition to the above investigated variation of the phosphorylation degree. At a polymer concentration of 1 g/L, particles with shuttlecock morphologies and sizes around 50 µm were obtained (Figure 2a), in contrast to particles in half-sphere shapes with sizes on the same order of magnitude, which developed at a polymer concentration of 10-1 g/L (Figure 2b). The further lowering of the polymer concentration in the following mineralization experiments to 10-2 g/L resulted in a mixture of particles with fried egg morphologies and irregularly shaped particles (Figure 2c). Pure fried egg particle morphologies (50 µm) were obtained at polymer concentrations of 10-3 g/L (Figure 2d). At a polymer

994

Langmuir, Vol. 20, No. 3, 2004

Figure 3. Surface tension of 1 g/L polymer solutions depending on the phosphorylation degree. The x axis is extended from -5 to +110% for better visualization of the 0 and 100% data points, respectively.

concentration of 10-4 g/L, the ability of the polymer to influence the mineralization process seems to be almost lost. Therefore, mixtures were found consisting of half spheres with sizes in the range of 30 µm beside calcite rhombohedra, which grow under these conditions without the influence of polymer additives (Figure 2e). In mineralization experiments with lower polymer concentrations, only calcite rhombohedra were found (Figure 2f). For polymers with a lower phosphorylation degree, such as PEG(84)-b-PHEE(13)-(10%), the polymer activity is already lost at higher concentrations than in the previously reported case (see Supporting Information, Figure 2), which implies that the phosphorylation degree plays a crucial role for the morphology development of the crystalline superstructures. Dependence of the Superstructure on the Water Surface Tension. The use of polymers from the PEGb-PGL22 [PGL ) poly(glycidol)] family for the Kitano mineralization experiments means a pronounced change in the polymer structure as compared to the PHEE-based block copolymers.22 The phosphate functionalized block in the PEG-b-PGL polymers has another oxygen atom per repeating unit in the polymer backbone and one carbon atom less in the side chain. Thus, the PEG-b-PGL polymers are more hydrophilic than the PEG-b-PHEE ones and represent a different class of block copolymers with an identical sticking group. The mineral structures obtained in the presence of the PGL block copolymers are closely related to the ones discussed before for the PHEE family, which is an interesting analogy. In the case of the fully phosphorylated PEG-b-PGL, flowerlike structures are obtained. These structures consist also of an inner core and an outer sphere. The size of these structures is around 100 µm. Beside the flowerlike structures, also some irregular structures were obtained.17 The use of partially phosphorylated PEG-b-PGL leads to fried egglike particle morphologies. The content of polymer in the minerals was checked with TGA and is in the range of 1-5 wt %. These results led us to an investigation of the polymer surface activity as expressed in surface tensions of their water solutions. The results are presented in Figure 3. The data in Figure 3 show that PEG(45)-b-PGL(27) is the most hydrophilic of the investigated block copolymers with a surface tension of 66-67 mN/m almost independent of the phosphorylation degree in the investigated range. This value is on the order of magnitude found for very hydrophilic polymers such as, for example, multifunctional hydroxylated poly(acrylamides).25 The PHEE-based block copolymers, on the other hand, show a remarkable influence of the water surface tension on the polymer

Rudloff and Co¨ lfen

phosphorylation degree in the investigated range up to 40%. As expected, the polymers are more hydrophobic compared to the PGL block copolymers but become more hydrophilic with increasing phosphorylation degrees. It is interesting to note that the longer PHEE block appears more hydrophilic than the shorter one at low phosphorylation degrees, but the trend appears to revert at a phosphorylation degree of 25%. This can be attributed to the absolute number of phosphate groups, which is 300% higher for the longer chain compared to the shorter one at the same phosphorylation degree. Thus, the longer chain appears more hydrophilic compared to the shorter one at low phosphorylation degrees until a critical number of phosphate groups is reached, after which the surface tension remains essentially unchanged. At this plateau, it is implied that PHEE is more hydrophobic as compared to PGL with an increasing hydrophobicity with increasing PHEE chain length. This implies that DHBCs are not always purely hydrophilic as the name may suggest but can also be surface-active in aqueous solution. This partial hydrophobicity was previously synthetically realized by attachment of hydrophobic side chains.26,27 If now polymer solutions with different surface tensions are utilized to study the effect of the solution surface tension onto the obtained particle morphology, the results in Figure 4 are obtained. These results (Figure 4) imply the fact that the variation of the water surface tension by the slight variation in the hydrophobicity of the functional DHBC block leads to the variation of the complex vaterite nanocrystal superstructures, although one has to be aware that the phosphorylation degree is simultaneously varied. At a high water surface tension in the presence of PEG(45)-b-PGL(27)-(100%), the big flower morphologies with several spherical cores and extensive rims and a size of about 100 µm are generated. On the other hand, PEG(84)-b-PHEE(13)-(40%) leads to the shuttlecock morphologies with a rim size of 50 µm and a spherical core size of 10 µm. Finally, at the lowest surface tension in the presence of PEG(84)-PHEE(13)-(10%) the hollow halfsphere morphologies with a size of 6 µm are obtained. It is noteworthy that each of the morphologies in Figure 4 exhibits at least partly a spherical moiety, but their sizes differ in a nonsystematic manner if related to the water surface tensions in the presence of the different DHBCs. In particular, the sizes of the spherical morphology parts are in Figure 4: (a) 25, (b) 30, and (c) 6 µm. However, it has to be kept in mind that the variation of the surface tension by a change in the phosphorylation degree also means a change in the functional groups, which have an influence onto the mineral morphology as stated in the previous corresponding subheading. Thus, the influence of only phosphorylation can be observed with the PEGb-PGL block copolymers where the surface tension is essentially unchanged upon variation of the phosphorylation degree in the investigated range. Because the obtained flower- or egglike structures are related to those obtained for the family of PEG-b-PHEE block copolymers at the highest surface tensions, it is implied that not only the phosphorylation degree but also the surface tension determines the mineral morphology. To exclusively check the influence of the surface tension onto the aggregate morphology, a Kitano solution of nonionic surfactant Antarox was prepared. The surfactant (25) Saito, N.; Sugawara, T.; Matsuda, T. Macromolecules 1996, 29, 313-319. (26) Antonietti, M.; Breulmann, M.; Go¨ltner, C.; Co¨lfen, H.; Wong, K. K. W.; Walsh, D.; Mann, S. Chem.sEur. J. 1998, 4, 2493-2500. (27) Sedlak, M.; Co¨lfen, H. Macromol. Chem. Phys. 2001, 202, 587597.

Nanocrystalline CaCO3 Particle Morphology Control

Langmuir, Vol. 20, No. 3, 2004 995

Figure 4. CaCO3 crystallization using the Kitano method with different polymer additives at 1 g/L observed after 80 h. (a) PEG(45)-b-PGL(27)-(100%) (γ ) 66.02 mN/m, SEM image), (b) PEG(84)-b-PHEE(13)-(40%) (γ ) 63.52 mN, SEM image), (c) PEG(84)-b-PHEE(13)-(10%) (γ ) 46.11 mN, SEM image), and (d) Antarox nonionic surfactant additive (γ ) 63.00 mN, light microscopy image).

concentration was adjusted in a way that the surface tension of the pure Kitano solution was lowered to 63 mN/m, which equals that of PEG(84)-b-PHEE(13)-(40%). If only the surface tension would be responsible for the morphogenesis, a shuttlecock morphology would be expected. Instead, spherical vaterite particles with sizes between 2 and 15 µm are obtained (Figure 4d). This shows that the surface tension certainly has an influence on the particle morphology but is not the only control factor in the morphogenesis process. On the other hand, the influence of the surface tension is clearly visible because the default calcite rhombohedra are not formed. The particle size is also strongly influenced by the surface tension because the particles obtained with the surfactant additive are much smaller than the big flowerlike structures, which are obtained without surfactant. An interaction of the surfactant with the crystals cannot be definitely ruled out but can be considered as unlikely because the surfactant consists of oligoethyleneoxide and nonylphenol, both having a low affinity to ionic crystal surfaces. In addition, the surfactant concentration was very low. Mechanism The previously described results suggest an important role of the water surface tension onto the vaterite nanocrystal superstructures. Thus, it is implied that a common growth/aggregation mechanism is responsible for the formation of the observed unusual complex crystalline superstructures. In the Kitano method, the initiation of the mineral growth process takes place in the upper layers of the Kitano solution, caused by the low carbon dioxide concentration in this region due to the diffusion of carbon dioxide to the surface of the solution. A low carbon dioxide concentration shifts the carbonate/hydrogen carbonate equilibrium to the carbonate side, resulting in CaCO3 crystallization according to

CaCO3(s) + CO2(g) + H2O(l) h Ca2+(aq) + 2HCO3-(aq) The solubility of calcium carbonate is much less than the solubility of the hydrogen carbonate, hence the solution becomes supersaturated with carbonate and the mineralization process starts with the nucleation near the air/ solution phase boundary and the carbon dioxide bubbles, which are temporarily stabilized by the, at least weakly, surface-active DHBCs. The nanoparticles formed in solution are then immediately stabilized by surrounding

Figure 5. Mineral growth observed by light microscopy. (a) Early stage of growth, ring morphology of CaCO3 particles with polymer PEG(45)-b-PGL(27)-(34%) after 20 h of growth. (b) Dendritic growth of the outer mineral sphere with polymer PEG(45)-b-PGL(27)-(100%) after 70 h of mineral growth.

DHBC molecules as the formation of the inorganic surface triggers their amphiphilicity.10 As the stabilizing poly(ethylene oxide) block is rather short (2000-6000 g/mol), the nanocrystals are only temporarily sterically stabilized. They start to agglomerate on the surface of the solution, forming the observed bigger superstructures on the surface, presumably starting on the rim of the CO2 gas bubbles. Thus, the carbon dioxide bubbles act as templates and ring morphologies are formed on the air/water surface in the early stage as indirectly observed by light microscopy (Figure 5a). These structures indicate an average CO2 bubble size of about 7-8 µm. Any earlier growth stages could not be observed because they were beyond the resolution of the light microscope. Light microscopy, however, is the only technique for the visualization of the species in solution in contrast to electron microscopy or scanning force microscopy, which require a drying step. Also, the CO2 bubbles could not be observed directly because they were destroyed upon preparation of the slides for light microscopy. Summarizing the previously reported results, the schematic formation mechanism shown in Figure 6 can be suggested. First, vaterite nanoparticles are formed presumably near the air/solution interface and are temporarily stabilized by the DHBCs (Figure 6a). CO2 bubbles do not completely leave the solution because they are stabilized by the slightly surface-active DHBCs at the air/water interface so that they can act as templates for the particle aggregation (Figures 6b and 5a). In the next step, the whole water/carbon dioxide bubble interface is subsequently covered by aggregating vaterite nanoparticles, and the ring morphology grows to hollow half spheres (Figure 6c). As a result of the increasing weight of the formed superstructure, it sinks somewhat deeper and the solution/air surface develops a curvature (Figures 6d and 5b). Now it is a matter of the water surface tension how long the mineral superstructure with its increasing mass due

996

Langmuir, Vol. 20, No. 3, 2004

Rudloff and Co¨ lfen

Conclusions

Figure 6. Schematic growth mechanism of CO2 bubble templated CaCO3 nanoparticle superstructure formation (see text).

to further nanoparticle aggregation can be kept at the water surface. If the surface tension is high, the vaterite nanonoparticle aggregates can develop a plain shape, such as in the case of the flower morphology PEG(45)-b-PGL(27)-(100%) (Figure 4a) or egg morphology with polymer PEG(45)-b-PGL(27)-(34%). The precondition for the growth of the outer sphere is the development of a curvature of the air/water surface. This can only be achieved with a high surface tension of the Kitano solution, otherwise the growing material would break away from the surface and sink down to the bottom of the vessel, where the growth would stop. Morphologies with outer spheres were only observed in the case of the higher phosphorylated polymers and, thus, high water surface tensions. If the water surface tension is lowered, the initial hollow sphere partly sinks and further aggregation can take place at the newly formed interface above the hollow sphere (Figure 6d) so that the shuttlecock morphologies can develop, such as in the case of polymer PEG(84)-b-PHEE(13)-(40%) (Figure 4b). If the structures become too heavy, they sink down to the bottom of the vessel where their further growth by aggregation is quenched. This happens at an early stage in the case of the hollow spheres in the presence of PEG(84)-b-PHEE(13)-(10%), which already significantly lowers the water surface tension (Figures 4c and 6e,f). Therefore, small (light) hollow spheres develop the shape of nearly full spheres and bigger (heavy) spheres develop only the shape of half hollow spheres (Figure 2, Supporting Information).

In this work, we could demonstrate that very simple templates such as gas bubbles can already be used to generate complex crystal aggregate morphologies in a crystallization reaction occurring at the air/solution interface. Here, the solution surface tension was found to play a significant role beside the already identified parameters such as the polymer functional groups and the polymer concentration. Although the particle size can crudely be controlled via the solution surface tension, the morphology and size of the nanocrystal aggregates is determined in a synergistic way. This was shown by a control experiment with nonionic surfactant addition, where the obtained morphology differed from the expected one, which was obtained at the same surface tension but with a block copolymer additive. The block copolymers fulfill the role of a temporary nanoparticle stabilizer and are active down to very low concentrations of 10-4 g/L. The temporarily stabilized nanoparticles are the building units for the larger structures with complex morphologies, which are formed by a simple aggregation mechanism. These results underline the importance of mesoscale selfassembly for biomineralization and biomimetic mineralization processes, as recently reviewed,28 and suggest a very simple way for the generation and tuning of complex crystalline nanoparticle superstructures. Acknowledgment. Financial support of the MaxPlanck-Society and DFG(SFB 448) is gratefully acknowledged. We also thank Erich C. for the fruitful discussions. Supporting Information Available: SEM micrographs showing the dependence of the obtained CaCO3 morphologies on the PEG(133)-b-PHEE(38) phosphorylation degree at a polymer concentration of 1 g/L and on the PEG(84)-b-PHEE(13)-(10%) concentration. This material is available free of charge via the Internet at http://pubs.acs.org. LA0358217

(28) Co¨lfen, H.; Mann, S. Angew. Chem., Int. Ed. 2003, 42, 23502365.