Langmuir 2004, 20, 957-961
957
Crystallization of Dyes by Directed Aggregation of Colloidal Intermediates: A Model Case Andreas Taden, Katharina Landfester, and Markus Antonietti* Max Planck Institute of Colloids and Interfaces, Research Campus Golm, 14424 Potsdam, Germany Received September 15, 2003. In Final Form: November 11, 2003 The crystallization of two oil-soluble dyes, oil red and oil blue, from the 125-nm-sized nanodroplets of miniemulsions, results in larger well-crystallized species of high quality, which must have been grown by aggregation of colloidal intermediates. This model case not only allows the existence of this nonclassical crystallization process to be proven but also gives evidence for the presence of “super-van der Waals” forces operating between the single nanocrystalline moieties.
Introduction Controlled crystallization is one of the most important chemical processes, for instance, for the production of dyes or drugs where color or biofunction, respectively, depends on the crystal size, morphology, and polymorph. With concepts such as “morphosynthesis”,1,2 crystallization was raised to a synthetic procedure of chemistry on the mesoscale3,4 because it allows the generation of nanohybrids or nanostructured materials with a broad variety of applications not known for classical crystalline materials. This is also known from “biomineralization”;5,6 that is, nature also handles crystallization processes in a by far more complicated fashion than classical physical chemistry. Throughout the sometimes very detailed examinations on morphosynthesis and biomimetic mineralization, doubt was raised whether the classical picture of crystallization is correct in many of those cases. Instead of building a new crystal ion-by-ion (or molecule-by-molecule) and the assorted thermodynamics with the chemical potential of single molecules or effects of Ostwald ripening, it was reasoned that these crystals are constructed in a blockwise fashion, that is, by the addition of nanometer-sized, colloidal intermediates. For instance, both Weiner et al.7,8 and Co¨lfen and Mann9 reviewed independently the role of amorphous nanoparticles in biomimetic mineralization; the latter authors coined the notation “mesoscale transformation” for this process. In primary work, such transformations were observed by Qi et al.10 and Yu et al.,11 resulting either in nanofiber bundles or in complex cones.12 The role of similar * Author to whom correspondence should be addressed. (1) Mann, S. Angew. Chem., Int. Ed. 2000, 39, 3393. (2) Ozin, G. A. Acc. Chem. Res. 1997, 30, 17. (3) Antonietti, M.; Ozin, G. A. Chem.sEur. J. 2003, accepted for publication. (4) Mann, S.; Burkett, S. L. Chem. Mater. 1997, 3, 22300. (5) Addadi, L.; Weiner, S. Angew. Chem., Int. Ed. Engl. 1992, 31, 153. (6) Mann, S. Nature 1993, 365, 499. (7) Beniash, E.; Aizenberg, J.; Addadi, L.; Weiner, S. Proc. R. Soc. London, Ser. B 1997, 264, 461. (8) Addadi, L.; Raz, S.; Weiner, S. Adv. Mater. 2003, 15, 959. (9) Co¨lfen, H.; Mann, S. Angew. Chem., Int. Ed. 2003, 42, 23502365. (10) Qi, L.; Co¨lfen, H.; Antonietti, M.; Li, M.; Hopwood, J. D.; Ashley, A. J.; Mann, S. Chem.sEur. J. 2001, 7, 3526. (11) Yu, S. H.; Antonietti, M.; Co¨lfen, H.; Hartmann, J. Nano Lett. 2003, 3, 379.
intermediates in CaCO3 scale formation as well as their experimental identification are also questions of larger industrial relevance and were described in a recommendable review by Horn and Rieger.13 Especially for systems with a low solubility or ion product, nonsoluble nano-intermediates are the only way to explain the sometimes astonishingly fast reaction rates, the occurring additional mesoscopic length scales in the structures, and the resulting structural complexity. Mesoscale transformation of well-defined semiconductor nanocrystals toward nanorods throughout the depletion of stabilizer were first described by Penn and Banfield14 and then observed by Tang et al.,15 Weller et al.,16 and finally, in the most detail with high-resolution transmission electron microscopy (TEM), Liz-Marzan and Giersig.17 In all those cases, the known solubility products of semiconductors are so low that structural transformations via Ostwald ripening (i.e., molecular transport) are not really relevant in the experimental time scales. Dipolar fields were given as a reason for the lineup of nanocrystals by directed aggregation, although the authors were aware that the primary objects possess no dipolar symmetry. For us, this just means that the building block principle toward nanostructures is essentially still not understood. In the present contribution, we want to examine a model case for crystallization by mesoscale transformation, namely, the crystallization from a nanodroplet dispersion. We choose two crystalline but meltable dye structures (oil red and oil blue) with a practical nonsolubility in water (as easily found out by the absence of water-based color under all applied conditions). The dyes were dissolved in chloroform, and the solution of those molecules was then dispersed to small organic nanodroplets in water. For that, a so-called miniemulsification is applied. This new technique allows the generation of nanodroplets with a diameter of 50-500 nm by applying very high shear/ ultrasound to a crude emulsion; the droplets are kept (12) Li, M.; Mann, S. Langmuir 2000, 16, 7088. (13) Horn, D.; Rieger, J. Angew. Chem., Int. Ed. 2001, 40, 4331. (14) Penn, R. L.; Banfield, J. F. Geochim. Cosmochim. Acta 1999, 63, 1549. (15) Tang, Z. Y.; Kotov, N. A.; Giersig, M. Science 2002, 297, 237. (16) Pacholski, C.; Kornowski, A.; Weller, H. Angew. Chem., Int. Ed. 2002, 41, 1188. (17) Liz-Marzan, L.; Giersig, M. Chem. Mater., accepted for publication.
10.1021/la035723l CCC: $27.50 © 2004 American Chemical Society Published on Web 01/10/2004
958
Langmuir, Vol. 20, No. 3, 2004
Taden et al.
Figure 1. Preparation of a dye miniemulsion resulting in 100-nm-sized separated compartments (step 1). The single droplets can undergo crystallization (step 2) and, in case of insufficient colloidal stability of the nanocrystals, mesoscale transformation (step 3).
stable by the simultaneous application of minimal amounts of a surfactant and an osmotic stabilizing compound, which is highly insoluble in the continuous phase.18,19 The surfactant stabilizes the droplets against collisions, and mass exchange (Ostwald ripening) between the droplets is suppressed by the use of the osmotic agent.20 The chloroform is removed by evaporation, and crystallization takes place, as graphically illustrated in Figure 1. The crystallization is confined to the droplets, and if those crystals are sufficiently stable, one ends up with a dispersion of nanocrystals with a maximal size of 100200 nm, similar to the nanodroplets, as recently shown for water,21 n-alkanes,22 and poly(ethylene glycol)s.23 If the surfactant desorbs from the crystals or is not able to handle the interparticular forces that have been altered by crystallization, crystal growth by controlled aggregation can be obtained (step 3). Because we precisely know the size and character of the precursor structures (from the miniemulsion process), structural development can be easily followed. Experimental Section Materials. The dyes oil blue N (97% dye content, melting point 112-114 °C) and oil red O (95% dye content, melting point 120 °C) were purchased from Aldrich. Sodium dodecyl sulfate
(SDS) and hexadecane from Fluka were used as received. Preparation of the Miniemulsions. A total of 400 mg of the dye and 20 mg of hexadecane was dissolved in 4.0 g chloroform and added to a solution of 36 mg of SDS dissolved in 15.0 g of water. After stirring for 1 h, the miniemulsion was prepared by ultrasonicating the emulsion with a Branson sonifier W450 (microtip) at an amplitude of 90% for 2 min. To prevent a temperature rise in the sample, the emulsion was ice-cooled. After sonication, the sample was heated for 3 h at 50 °C for the evaporation of the chloroform. In the case of steric stabilization, additionally 100 mg of a polymeric stabilizer (listed in the following) was added either before or after sonication. (18) Landfester, K.; Bechthold, N.; Tiarks, F.; Antonietti, M. Macromolecules 1999, 32, 5222. (19) Schork, F. J.; Poehlein, G. W.; Wang, S.; Reimers, J.; Rodrigues, J.; Samer, C. Colloids Surf., A 1999, 153, 39. (20) Antonietti, M.; Landfester, K. Prog. Polym. Sci. 2002, 27, 689. (21) Montenegro, R.; Antonietti, M.; Mastai, Y.; Landfester, K. J. Phys. Chem. B 2003, 107, 5088. (22) Montenegro, R.; Landfester, K. Langmuir 2003, 19, 5996. (23) Taden, A.; Landfester, K. Macromolecules 2003, 36, 4037.
Analytical Methods. The particle sizes were measured using a Nicomp particle sizer (model 370, PSS Santa Barbara, U.S.A.) at a fixed scattering angle of 90°. Scanning electron microscopy (SEM) was performed with a Zeiss DSM 940 electron microscope. The samples were prepared by drying a droplet of the diluted dispersion on a glass support at room temperature. Afterward, the samples were sputter-coated with gold. Polarized light microscopy was carried out using a Microscope Olympus (model BX50). Images were taken on a digital camera directly connected to the microscope.
Results The miniemulsification of the dyes was optimized with respect to the type and amount of surfactant. For the stabilization of the miniemulsions, a variety of different surfactants were applied. The use of ionic surfactants such as SDS or cetyltrimethylammoniumchloride allows the production of colloidally stable primary miniemulsions (after evaporation of the chloroform) which can, for instance, be diluted and characterized by light scattering. The droplet diameter for the recipe described is typically at 125 nm, with a narrow droplet size distribution. However, the samples turned out not to be colloidally longterm stable, as crystallization and gelation of the sample took place within 1 day. By adding block copolymer surfactants (up to 25 wt % compared to the polymer phase) such as poly(butylene-co-ethylene)-b-poly(ethylene oxide), it was possible to suppress the secondary processes and maintain a colloidally stable situation for up to 3 days after miniemulsification. It, however, turned out that not a single one of those samples was long-term stable, indicating that even polymeric surfactants are not sufficient to efficiently shield the long-range interactions discussed in the following. In the case of the oil blue, the crystallization and gelation of the dispersion is nicely paralleled by a pronounced change of the color from deep blue to a pleochroic bronze tone, indicating that the underlying process is by far more complicated than statistical aggregation. The resulting color changes after 12 h at different storage temperatures are shown in Figure 2. SEM (Figure 3) reveals the reason for these changes: the dispersion completely transfers into an arrangement of crystalline nanorods with a diameter of about the primary particle size. In many cases, one of the tips is sligthly curved, and the primary nucleus can be vagely identified. Indeed, a mesoscale transformation has taken place: the dye nanoparticles have aggregated in a controlled fashion and transformed to a larger crystalline object. We tried to identify the primary crystallite with TEM microscopy and differential scanning calorimetry but were not successful: as long as the dye dispersion is amorphous, the dispersion is stable; however, at the moment where crystals are first formed, these crystals nucleate the
Directed Aggregation of Colloidal Intermediates
Langmuir, Vol. 20, No. 3, 2004 959
Figure 2. (a) Freshly prepared oil blue sample; (b) stored 12 h at 20 °C; (c) stored 12 h at 40 °C, and (d) stored 12 h at 60 °C.
Figure 3. SEM picture of the oil red sample prepared at room temperature.
aggregation of further, presumably amorphous, dye particles and propagate via their transformation. From the length of the crystals of about 5 µm, one can estimate that, in the case of oil red, about only 1 of 40 nanoparticles is nucleating; all the others propagate the growth. It is very interesting how the primary dye molecules and crystals are arranged in the nanofiber or final crystals. Dye molecules and dye crystals have per se a highly dipolar structure; if the final crystal is pleochroic, the main polar axis of the dye is arranged parallel to all the other dyes in the crystal. This is true for both examined species. Pleochroism gives the easy possibility into our hands to identify the mechanism of mutual attraction. The maximum dipole momentum, if not compensated by the intermolecular arrangement in the crystal unit cell, is obviously along the dye main axis, whereas the maximal in-plane polarizability is found perpendicular to this axis (electrons can be easily moved along the backbone and in this specific crystal plane). This results in a tensorial force field (x, y, and z directions are different) around the single nanocrystals, where the maximum of the dipole-dipole interactions and fluctuating dipole-induced mirror dipole
interactions (van der Waals forces) are perpendicular to each other. We also tried to base our findings on singlecrystal structures but did not find these data in the accessible data banks. In general, there are only a few publications24-26 available on the crystal structure of standard azobenzene dyes because of their reluctance to form crystals of a sufficient size. In this sense, crystallization from miniemulsions represents also a promising way to obtain unusually large and well-defined crystals. Ripening the primary crystals shown in Figure 3 for some time promotes further growth on the scale of months. Figure 4 shows a series of polarization tilts of a selected large crystal of oil blue. It is obvious that the dye molecules are arranged in the crystal exactly perpendicular to the preferred growth direction; that is, the growth only weakly proceeds in the direction of the estimated dipole moment, but the very strong van der Waals attraction of a crystalline material drives the controlled aggregation. This is illustrated in Figure 5. Because the van der Waals force of a polar crystal is much stronger than the one of a corresponding amorphous material (because all elementary polarizabilities add up in a coherent fashion), we might call this a “super-van der Waals” force, quantified by a “super-Hamaker” constant. It is obvious that colloidal stability is much harder to obtain under these conditions because the attractive part easily exceeds the repulsive part (brought in by the surfactants) optimized for amorphous materials. This is especially true for miniemulsions, which are “critically stabilized”;19 that is, they show the minimal particle size for a given amount of stabilizer. The spontaneous change from a colloidally stable amorphous precursor to a nonstable nanocrystalline species with spontaneously growth by directed aggregation is, therefore, probably due to the altered polarizability of the nanocrystalline starting unit. It is interesting to repeat that such a “destabilization” has tensorial properties; that is, the nonstability and coupled aggregation preferentially proceeds along the (24) Anderson, S.; Clegg, W.; Anderson, H. L. Chem. Commun. 1998, 2379-2380. (25) Olja, W. H.; Lu, L. K.; Albers, K. E.; Gleason, W. B.; Richardson, T. I.; Lovrien, R. E.; Sudbeck, E. A. Acta Crystallogr., Sect. B 1994, 50, 684. (26) Olja, W. H.; Gleason, W. B.; Richardson, T. I.; Lovrien, R. E. Acta Crystallogr. 1994, 50, 1615.
960
Langmuir, Vol. 20, No. 3, 2004
Taden et al.
Figure 4. Series of polarization tilts of a selected large crystal of oil blue (the bar represents 20 µm).
direction of maximal polarizability, the radial symmetry of the DLVO potential is broken, and straight, in the beginning, partly bent nanofibers with the thickness of the assembed blocks are obtained (see Figure 3 again). It is an open question if the primary crystals can not only aggregate with other nanocrystals (making nucleation in all nanodroplets a prerequisite) but also attract amorphous intermediate particles, which afterward undergo mesoscale transformation. Directly after emulsification, the droplets are indeed amorphous, as proven by wide-angle X-ray measurements of the miniemulsions. High-resolution SEM at the beginning of gelation show the coexistence of rodlike crystals and spherical, apparently amorphous nanoparticles. All trials to isolate and depict further intermediate growth structures were not successful. The aggregation of amorphous nanoparticles onto the growing crystalline tips would go well with the observations made at fiber-
forming inorganic systems, which is what the present miniemulsion system was chosen to be a model system for. Conclusion The crystallization of the miniemulsions of two dyes (oil blue and oil red) with primary droplet diameters of 120 nm results in single-crystalline nanofibers of high quality, uniformity, and chromatic definition. Because these dyes are absolutely insoluble in the continuous phase, the observed growth of the crystals must have proceeded via the controlled aggregation and mesoscale transformation of colloidal intermediates. This is regarded as a model case for this nonclassical crystallization process, which has been postulated to explain a whole bunch of morphosynthesis experiments of inorganic species by a variety of groups but not elaborated in the clarity possible for the present system.
Directed Aggregation of Colloidal Intermediates
Langmuir, Vol. 20, No. 3, 2004 961
Figure 5. Scheme to illustrate the directed mesoscale aggegation. The formed primary nanocrystals, because of the symmetry of the unit cell and the outer shape, show different polarizabilities along different planes. As is typical for van der Waals forces, similar polarizabilities attract each other optimally (“attraction of the same”) here, preferentially the green faces with the highest polarizability giving the strongest van der Waals force. As a results, aggregation and mesoscale formation occurs only in this direction.
The fact that the crystals of those dyes are pleochromic reveals additional information about the aggregation process. Dye absorption and, therefore, the maximum dipole moment are oriented perpendicular to the growth direction, which gives strong indications that the controlled aggregation is not mediated by dipole fields (as usually speculated) but by polarization forces. As these, for polar crystals, add up coherently in a highly anisotropic fashion, van der Waals attraction in a certain direction can obviously become stronger than the ionic and steric stabilizers that have kept the original miniemulsion stable. We called this a “super-van der Waals” force, which is directional and breaks the radial symmetry of the DLVO
potential. As a result, highly selective and spatially controlled aggregation takes place, which is the prerequisite of the morphosynthetical control of the crystal habitus. In addition, the existence of such forces also explain why industrially optimized procedures to make dye or drug nanocrystals12 succeed or fail from system to system in a way nonpredictable from the properties of the molecule alone. In this picture, nanocrystals can only be stabilized from crystal structures that do not exhibit coherent addition of molecular polarizabilities. LA035723L
