Mechanism of Mesoporous Silica Formation. A Time-Resolved NMR

It is concluded that the aggregation is caused by a micelle−micelle attraction induced by oligomeric/polymeric silica that adsorbs to the EO palisad...
7 downloads 0 Views 215KB Size
Download PDF
680

Langmuir 2004, 20, 680-688

Mechanism of Mesoporous Silica Formation. A Time-Resolved NMR and TEM Study of Silica-Block Copolymer Aggregation Katarina Flodstro¨m,* Ha˚kan Wennerstro¨m, and Viveka Alfredsson Physical Chemistry 1, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden Received April 23, 2003. In Final Form: September 17, 2003 The dynamics of the synthesis of a mesoporous silica material SBA-15 is followed using time-resolved in situ 1H NMR and transmission electron microscopy (TEM). Block copolymer-silica particles of twodimensional hexagonal symmetry evolve from an initially micellar solution. The synthesis was carried out with the block copolymer Pluronic P123 (EO20-PO70-EO20) at 35 °C and using tetramethyl orthosilicate as the silica precursor. By using TEM, we can image different stages during the evolution of the synthesis. Flocs of spherical micelles held together by the polymerizing silica are observed prior to precipitation. With time, the structure of these flocs evolves and the transition from spherical to cylindrical hexagonally packed micelles can be monitored. The signal from the methyl protons of the PO part was recorded with 1H NMR. One observes a continuous increase in the signal width but with distinct changes in the spectral characteristics occurring in narrow time intervals. The spectral changes can be attributed to structural changes of the self-assembled aggregates. The 1H NMR and TEM studies reveal the same mechanism of formation. It is concluded that the aggregation is caused by a micelle-micelle attraction induced by oligomeric/polymeric silica that adsorbs to the EO palisade layer of the micelles and has the ability to bridge to another micelle. This adsorption also favors the formation of cylindrical aggregates relative to spherical micelles. The sequence of NMR and TEM observations can then be interpreted as the following sequence of events: (i) silicate adsorption on globular micelles possibly accompanied with some aggregate growth, (ii) the association of these globular micelles into flocs, (iii) the precipitation of these flocs, and (iv) micelle-micelle coalescence generating (semi)infinite cylinders that form the two-dimensional hexagonal packing.

Introduction As a result of their high surface areas and uniform pore widths, ordered mesoporous silica materials have a high potential for use in, for example, separation processes and catalysis. These inorganic materials are synthesized using amphiphilic molecules as structure directors. When the surfactant is removed by, for example, calcination, a porous material remains with similar mesostructure as the liquid crystalline of a surfactant-water system. In the first of the syntheses of mesoporous silica using this concept,1,2 cationic surfactants were used as structure-directing agents. By varying the reaction conditions, one could obtain both a two-dimensional hexagonal structure (p6mm), normally referred to as MCM-41, and a cubic structure (Ia3 h d), called MCM-48. Later, it was found also that nonionic surfactants and amphiphilic block copolymers can act as structure-directing agents in the syntheses of analogous mesoporous materials. Both disordered wormhole structures3,4 and ordered materials5-7 have been * Corresponding author. E-mail: [email protected]. Tel.: +46 46 2221536. Fax: +46 46 2224413. (1) Beck, J. S.; Vartuli, J. C.; Roth, W. J.; Leonowicz, M. E.; Kresge, C. T.; Schmitt, K. D.; Chu, C. T. W.; Olson, D. H.; Sheppard, E. W.; McCullen, S. B.; Higgins, J. B.; Schlenker, J. L. J. Am. Chem. Soc. 1992, 114, 10834-10843. (2) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710-712. (3) Bagshaw, S. A.; Prouzet, E.; Pinnavaia, T. J. Science 1995, 269, 1242-1244. (4) Bagshaw, S. A.; Pinnavaia, T. J. Angew. Chem., Int. Ed. Engl. 1996, 35, 1102-1105. (5) Zhao, D.; Huo, Q.; Feng, J.; Chmelka, B. F.; Stucky, G. D. J. Am. Chem. Soc. 1998, 120, 6024-6036. (6) Yang, P.; Zhao, D.; Margolese, D. I.; Chmelka, B. F.; Stucky, G. D. Chem. Mater. 1999, 11, 2813-2826. (7) Yu, C.; Yu, Y.; Zhao, D. Chem. Commun. 2000, 7, 575-576.

reported as products of syntheses with these amphiphiles. Particularly triblock copolymers, consisting of blocks of ethylene oxide and propylene oxide (EOx-POy-EOx), Pluronics, are excellent structure directors for generating ordered mesoporous silica. Different polymers and synthesis conditions can induce lamellar, hexagonal, and cubic structures.5,8-10 The most frequently used structuredirecting polymer, Pluronic P123 (EO20-PO70-EO20), gives rise to a two-dimensional hexagonal (p6mm) product normally known as SBA-15.11 Using block copolymers instead of low-molecular-weight surfactants has the advantage of giving more stable materials, due to thicker silica walls, as well as larger pore diameters. Another interesting property of the SBA-15 material is the coexistence between meso- and micropores.12 The micropores provide connectivity between the larger organized mesopores. A large amount of synthesis work has been carried out within the field of mesoporous materials. However, less effort has been devoted to understanding the mechanism of formation and the kinetics in these processes. Through the years, a number of different models have been suggested, among them the “liquid crystal templating” mechanism proposed by Kresge et al.,2 a “charge density matching” mechanism proposed by Monnier et al.,13 and (8) Flodstro¨m, K.; Alfredsson, V. Microporous Mesoporous Mater. 2003, 59, 167-176. (9) Kipkemboi, P.; Fogden, A.; Alfredsson, V.; Flodstro¨m, K. Langmuir 2001, 17, 5398-5402. (10) Flodstro¨m, K.; Alfredsson, V.; Ka¨llrot, N. J. Am. Chem. Soc. 2003, 125, 4402-4403. (11) Zhao, D.; Feng, J.; Huo, Q.; Melosh, N.; Fredrickson, G. H.; Chmelka, B. F.; Stucky, G. D. Science 1998, 279, 548-552. (12) Ryoo, R.; Ko, C. H.; Kruk, M.; Antochshuk, V.; Jaroniec, M. J. Phys. Chem. B 2000, 104, 11465-11471.

10.1021/la030173c CCC: $27.50 © 2004 American Chemical Society Published on Web 12/25/2003

Silica-Block Copolymer Aggregation

a mechanism suggested by Frasch et al.14 where the key step is the formation of silica prepolymers. Patarin et al. have recently reviewed this topic.15 Cryogenic transmission electron microscopy (cryo-TEM)16 and synchrotron X-ray17,18 and electron paramagnetic resonance (EPR)19,20 have been used to gain information about the kinetics in hexagonal mesoporous silica formation. Also, the transitions in the formation of bicontinuous cubic mesoporous silica have been studied with small-angle X-ray scattering (SAXS) by Pevzner and Regev.21 An interesting study based on TEM results (Sadasivan et al.)22 showed that if the synthesis was stopped or, in any case, slowed considerably by dilution and pH reduction, one can follow the rearrangement of the silica polymer composite. Initially, the particles, which have a mean size on the order of 20 nm, consist of unordered micellar subunits, which upon the progress of formation rearrange and form elongated hexagonally arranged micelles (MCM-41). In the case of cationic synthesis but under acidic conditions, Chan et al. described the formation in terms of a phaseseparation model.23 Phase separation occurs when the inorganic oligomers grow and results in droplets of a second liquid rich in oligomers and surfactants. Eventually, microphase separation occurs in the droplets, giving rise to the mesostructure. However, these experiments are all performed for the case where ionic surfactants are used as structure determinants. Not much work is concentrated on the mechanism of formation of material formed in the presence of nonionic amphiphiles, which can give rise to unordered wormhole structures (the MSU-X family) or ordered structures such as SBA-15. Boissie`re et al. used a fluorideassisted two-step synthesis in their work.24,25 Their focus was on nonionic diblock surfactants of the Cx(EO)y type, but experiments were also performed with Pluronic P123. They showed that the association of hybrid units, consisting of Cx(EO)y micelles and small silica oligomers polymerizing in the palisade layer, gives well-structured seeds of the emerging material within 10 min after the addition of fluoride ions. However, no details on the aggregation process of the P123-mediated synthesis were revealed. In a recent report of Ruthstein et al., the formation of SBA15 using EPR spectroscopy is considered.26 It was shown that after 20 min of reaction, the EO chains have a different environment, which is attributed to them being located (13) Monnier, A.; Schu¨th, F. Q.; Huo, Q.; Kumar, D.; Margolese, D.; Maxwell, R. S.; Stucky, G. D.; Krishnamurty, M.; Petroff, P.; Firouzi, A.; Janicke, M.; Chmelka, B. F. Science 1993, 261, 1299-1303. (14) Frasch, J.; Lebeau, B.; Soulard, M.; Patarin, J.; Zana, R. Langmuir 2000, 16, 9049-9057. (15) Patarin, J.; Lebeau, B.; Zana, R. Curr. Opin. Colloid Interface Sci. 2002, 7, 107-115. (16) Regev, O. Langmuir 1996, 12, 4940-4944. (17) Linde´n, M.; Schunk, S. A.; Schu¨th, F. Angew. Chem., Int. Ed. 1998, 37, 821-823. (18) Ågren, P.; Linde´n, M.; Rosenholm, J. B.; Schwarzenbacher, R.; Kriechbaum, M.; Amenitsch, H.; Laggner, P.; Blanchard, J.; Schu¨th, F. J. Phys. Chem. B 1999, 103, 5943-5948. (19) Zhang, J.; Luz, Z.; Goldfarb, D. J. Phys. Chem. B 1997, 101, 7087-7094. (20) Zhang, J.; Carl, P. J.; Zimmermann, H.; Goldfarb, D. J. Phys. Chem. B 2002, 106, 5382-5389. (21) Pevzner, S.; Regev, O. Microporous Mesoporous Mater. 2000, 38, 413-421. (22) Sadasivan, S.; Fowler, C. E.; Khushalani, D.; Mann, S. Angew. Chem., Int. Ed. 2002, 41, 2151-2153. (23) Chan, H. B. S.; Budd, P. M.; de V. Naylor, T. J. Mater. Chem. 2001, 11, 951-957. (24) Boissie`re, C.; Larbot, A.; Van der Lee, A.; Kooyman, P. J.; Prouzet, E. Chem. Mater. 2000, 12, 2902-2913. (25) Boissie`re, C.; Larbot, A.; Bourgaux, C.; Prouzet, E.; Bunton, C. A. Chem. Mater. 2001, 13, 3580-3586. (26) Ruthstein, S.; Frydman, V.; Kababya, S.; Landau, M.; Goldfarb, D. J. Phys. Chem. B 2003, 107, 1739-1748.

Langmuir, Vol. 20, No. 3, 2004 681

either in the micellar corona or in the micropores, respectively. Also, it was shown that the hexagonal structure is present after 2 h of reaction. Because triblock copolymers, such as Pluronics, can assist in forming mesostructured silica with useful properties, more information about the mechanism of formation is desirable. In this work, we have chosen to work with Pluronic P123 because it gives rise to wellordered two-dimensional hexagonal (p6mm) silica, SBA15. Our approach is to use 1H NMR and TEM as tools to follow the time evolution of the aggregation process during the synthesis. In situ NMR has previously been used to follow the mesoporous silica synthesis using 2H, 14N, 29Si, and even 81Br as the nuclei under study.27-31 The advantage with 1H NMR is the high intensity allowing for higher time resolution, but there is a cost of a more demanding analysis of the signals. Because it is not possible to do an in situ study in TEM, we have taken “snapshots” in time of the formation process by taking out samples at certain time intervals within the same time span as in the NMR study. The results from the TEM study agree very well with the information obtained from the NMR investigation, and, thus, we believe that, although it could be argued that the synthesis proceeds slightly after the sample is disposed onto the microscopy grids, the changes we observe is the actual structure forming. This paper starts with an account of the experimental details. Next, we present the results, that is, the in situ NMR data (including an account of 1H NMR of selfassembled aggregates) and the TEM results. Subsequently, the amphiphile phase behavior and its relation to amphiphile aggregation is discussed. Finally, combining the insights from the experimental observations and from generic features of amphiphile aggregation, we present a kinetic model of the events leading from Pluronic micelles in (dilute) solution to precipitated ordered mesoporous particles containing both silica and the block copolymer. Experimental Section The polymers (Pluronics) were obtained as a gift from BASF and used as received without further purification. Tetramethyl orthosilicate, TMOS (98%), was obtained from Aldrich, D2O from Dr. Glaser AG, Basel, and HCl from Merck. Synthesis of SBA-15 was first performed following the literature synthesis procedures, to make sure that the synthesis product would be a well-ordered, hexagonally structured material. A total of 1.92 g of Pluronic P123, EO20-PO70-EO20, was dissolved in 45 g of Millipore water and 30 g of 4 M HCl, resulting in a 2.5 wt % solution. The synthesis temperature was chosen to be 35 °C. When the polymer was fully dissolved, 2.92 g of TMOS was added under vigorous stirring for 1 min, after which the stirring rate was lowered. TMOS was used instead of the more common tetraethyl analogue to minimize interference effects in the 1H NMR spectrum. A precipitate became visible to the unaided eye after 24 min. The mixture was stirred for 24 h, after which it was put in an 80 °C oven for hydrothermal treatment for another 24 h. The product was filtered, washed with water, and thereafter dried in air. The polymer was removed by calcination in air at 500 °C for 6 h. (27) Chen, C.-Y.; Burkett, S. L.; Li, H.-X.; Davis, M. E. Microporous Mater. 1993, 2, 27-34. (28) Firouzi, A.; Kumar, D.; Bull, L. M.; Besier, T.; Sieger, P.; Huo, Q.; Walker, S. A.; Zasadzinski, J. A.; Glinka, C.; Nicol, J.; Margolese, D.; Stucky, G. D.; Chmelka, B. F. Science 1995, 267, 1138-1143. (29) Steel, A.; Carr, S. W.; Anderson, M. W. J. Chem. Soc., Chem. Commun. 1994, 1571-1572. (30) Cheng, C.-F.; Luan, Z.; Klinowski, J. Langmuir 1995, 11, 28152819. (31) Firouzi, A.; Atef, F.; Oertli, A. G.; Stucky, G. D.; Chmelka, B. F. J. Am. Chem. Soc. 1997, 3596-3610.

682

Langmuir, Vol. 20, No. 3, 2004

For the 1H NMR study, the only modification of the synthesis mixture was that the water was exchanged for an equal amount of D2O (on a molar basis), to reduce the intensity of the water peak. The block copolymer was dissolved in a mixture of D2O and concentrated HCl in a Teflon bottle while the solution was heated to our desired temperature, 35 °C. One 1H NMR spectrum was recorded at that stage to show the micellar solution before the addition of the silica precursor TMOS. When the polymer was totally dissolved, TMOS was added under vigorous stirring for 1 min. Subsequently, 1 mL of the solution was added to a 5-mm-diameter NMR tube. The tube was thereafter put into the NMR spectrometer, and the recording of the proton spectra started once a stable temperature (35 °C) was achieved. The first measurement started 7 min after the addition of TMOS. Measurements were thereafter performed every minute during the first hour and thereafter every fifth minute during the following 2 h. The 1H NMR measurements were performed on a Bruker NMR spectrometer operated at 500.20 MHz. The acquisition time was 3 s, the relaxation delay 1 ms, and the pulse width of the 90° pulse 9 µs. Concerning the chemical shifts, no external reference was added. The water peak is located at 4.84 ppm. The evolution of the synthesis was also studied by TEM. Synthesis conditions as mentioned previously (based on H2O) were employed. A total of 1 mL of the synthesis mixture was diluted in 10 mL of Millipore H2O at specific time intervals. These new mixtures were stirred at high speed for 20 s, after which some drops (containing both supernatant and precipitate in the cases where precipitation had occurred) were placed onto holey carbon copper grids. This dilution step has the advantages of quenching/slowing the silica polymerization as well as possible aggregation processes. At the same time, it makes the synthesis mixture less concentrated and more suitable for TEM imaging. The precipitation started 24 min after TMOS addition. Samples were collected for microscopy analysis at the following times: 20, 25, 40, 55, and 100 min. The transmission electron micrographs were recorded with a Philips 120 Biotwin microscope operated at 120 kV, equipped with a Gatan charge-coupled device camera. Also, the final product was characterized with TEM as well as with SAXS. The TEM samples were dispersed in acetone and dripped onto holey carbon copper grids. SAXS experiments were performed on a Kratky compact small-angle system equipped with a position-sensitive wire detector (OED 50M from MBraun, Graz) containing 1024 channels of width 53.6 µm. Cu KR radiation of wavelength 1.542 Å was provided by a Seifert ID 3000 X-ray generator operated at 50 kV and 40 mA. Nitrogen adsorption/desorption isotherms were recorded at 77 K using a Micromeritics ASAP 2400 instrument. The mother liquor was examined by self-diffusion NMR on a Bruker DMX200 NMR spectrometer operating at a 1H resonance frequency of 200 MHz. The surfactant self-diffusion was measured by a stimulated echo pulse sequence. These measurements were performed to get information about what remains in the solution and what is included in the precipitate. The product was filtered through a filter with a 0.1-µm pore size to avoid the particles and only measure on the liquid.

Results Product Characterization. In Figure 1, we show a TEM image of the final product synthesized in a NMR tube. The typical pattern of a hexagonal structure aligned perpendicular to the [001] axis can clearly be seen. No differences between material synthesized in the NMR tube and material formed under normal synthesis conditions (in a Teflon bottle) were observed. The products were also examined by SAXS. Figure 2 shows the X-ray diffraction patterns after a 3-h reaction time (at 35 °C) and after a completed synthesis (24 h at 35 °C, 24 h at 80 °C). These diffractograms show clearly the characteristic peaks of the two-dimensional hexagonal structure (p6mm). These results show that the structure can be considered fully evolved at least after 3 h (cf. Ruthstein et al.),26 which is when the NMR experiment was finished. Time-resolved in situ SAXS measurements will provide additional

Flodstro¨ m et al.

Figure 1. Transmission electron micrograph of hexagonal silica (as-synthesized), formed with Pluronic P123. Micrograph recorded perpendicular to the [001] axis.

Figure 2. X-ray diffraction patterns from as-synthesized hexagonal silica formed with Pluronics P123 filtrated after 3 h at 35 °C and 24 h at 35 °C plus 24 h at 80 °C. The (100) peaks are scaled to the same height. Without normalization, the intensity is considerably higher for the 48-h sample than for the 3-h sample, probably because the reaction continues, which leads to a better organized product and a more highly polymerized silica network. The diffractograms are indexed according to p6mm. The unit cell parameter was determined to be a ) 121 Å for the 3-h sample and a ) 124 Å for the 48-h sample.

information on the evolution of the mesostructure. A synchrotron in situ SAXS study has recently been performed and will be published separetly.32 Nitrogen adsorption/desorption measurements resulted in an isotherm typical for SBA-15 materials (results not shown). A BET surface area33 of 780 m2/g was obtained. 1 H NMR of Self-Assembled Aggregates. The 1H NMR spectrum of the solution before the addition of TMOS is shown in Figure 3. Because the peak (at 0.6 ppm) from the protons of the propylene methyl group is located far away from the other peaks, it is suitable to study the time evolution of this signal. To understand the NMR characteristics of systems with amphiphile aggregates, it is essential to recognize that (32) Flodstro¨m, K.; Alfredsson, V.; Linde´n, M.; Teixeira, C.; Amenitsch, H. Manuscript in preparation. (33) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309-319.

Silica-Block Copolymer Aggregation

Langmuir, Vol. 20, No. 3, 2004 683

Figure 3. 1H NMR spectrum of Pluronic P123 (2.5 wt %) in an acidic micellar aqueous solution, prior to the addition of TMOS.

there are molecular motions on several different time scales. Within an aggregate, the amphiphile molecules experience a local motion occurring on a typical liquid time scale. In contrast to the behavior in a normal liquid, the local motion is anisotropic and the tensorial NMR interactions, be it dipolar coupling, quadrupolar coupling, or chemical shift anisotropy, are motionally narrowed, but to a finite value. On a longer time scale, the aggregate might rotate and in an isotropic phase this rotation reduces the tensorial coupling to 0. Additionally, the molecules also have a translational motion along the aggregate surface, and if this is curved, this also results in a motional narrowing. This averaging due to translation occurs both in solution and in liquid crystalline aggregates where the overall rotation of the aggregates is quenched.34 These different motions are illustrated in Figure 4. For the particular case of the 1H signal of the methyl protons of the PO part of the block copolymer, we observe a broad peak with a characteristic shape in the final twodimensional hexagonal particles. This shape, sometimes called a super-Lorentzian, is caused by a superposition of signals from crystallites with random orientation θ relative to the magnetic field. For each orientation, the shape of the signal f(ν) is the same but with a width that scales as 3 cos2 θ - 1. The expression for the band shape is35

L(ν) )

∫01|3 cos2 θ - 1|-1f[ν/|3 cos2 θ - 1|] d cos θ

(1)

For the methyl protons, the dominant dipolar coupling arises from the interaction with protons on the same carbon. Taken alone, this would result in a quartet, but the coupling to other protons in the chain provides a broadening of the same order of magnitude as the splitting in the quartet. Thus, one expects the band shape to be a broad featureless signal where the width approximately reflects the magnitude of the averaged dipolar coupling within the methyl group. Using a Gaussian band shape, the observed signal in the hexagonal phase can be simulated using eq 1 if the width at half-height of the Gaussian is 1.5 kHz. This corresponds to an order parameter S ) 1/2〈3 cos2 θHH - 1〉 ≈ 0.02 for the protonproton vector, which is substantially lower than what is typically found in lipid or surfactant aggregates. (34) Lindman, B.; So¨derman, O.; Wennerstro¨m, H. In Surfactant solutions. New methods of investigation; Zana, R., Ed.; Marcel Dekker: New York, 1987. (35) Wennerstro¨m, H. Chem. Phys. Lett. 1973, 18, 41-44.

Figure 4. Illustration of the different motions modulating the dipolar coupling between two neighboring protons. (a) The quantity 1/2〈3 cos2 θHH - 1〉 is substantially reduced by local motions such as rotations around C-C and C-O bonds. However, this does not reduce the average value to 0. (b) Diffusion along the curved surface of an aggregate results in a further reduction of 1/2〈3 cos2 θHH - 1〉. As the axial ratio a/b of a prolate ellipsoid, or the length/diameter ratio of a cylinder with spherical caps, goes to unity, the average value of 1/2〈3 cos2 θHH - 1〉 goes to 0 and the 1H NMR signal becomes sharper.

At the start of the reaction, one has essentially spherical micelles in the solution and in addition to the local motion there is micelle rotation with a tumbling rate given by

τ)

4πr3η 3kT

(2)

which for a Pluronic P123 micelle of radius 10 nm in water36 gives τ ≈ 1 × 10-6 s. With such a correlation time, one is clearly outside the extreme narrowing limit for protons at 500 MHz because 2π‚5 × 108‚1 × 10-6 . 1. On the other hand, using the estimate of 2 × 1.5 kHz for the residual dipolar coupling, one is certainly in the motionally narrowed regime because 2π‚3 × 103‚1 × 10-6 , 1. In the motional regime between extreme and strong narrowing, (36) Jansson, J.; Schille´n, K.; Olofsson, G.; Cardoso da Silva, R.; Loh, W. J. Phys. Chem. B, in press.

684

Langmuir, Vol. 20, No. 3, 2004

Flodstro¨ m et al.

Figure 5. 1H NMR following the time evolution of the synthesis. The peak from the protons on the PO methyl group is followed, first every minute during the first hour and thereafter every fifth minute. Each consecutive spectrum is displaced 100 Hz to the right.

the signal should be a superposition of Lorentzians as

T (ν)

∫-∞+∞f(ν)1 + 4π2(ν 2- ν )2T 2(ν) dν

(3b)

Figure 6. Transmission electron micrographs following the time evolution of the synthesis. (a) 25 min after TMOS addition, just after precipitation. The micelles (still globular) have aggregated and are surrounded by polymerizing silica. (b) 40 min after TMOS addition. The micelles have started to become elongated. (c) 55 min after TMOS addition. The wormlike micelles have started to pack hexagonally. (d) 100 minutes after TMOS addition. The hexagonal structure is apparent.

and L(ν) is the same quantity as in eq 1. The transition from the superposition of Lorentzians as in eq 3 to the super-Lorentzian of eq 1 occurs when the inverse correlation time is of the same magnitude as the residual dipolar coupling, which means correlation times on the order of 10-4 s. Inserted into eq 2, such rotational correlation times are obtained for spherical particles with a radius on the order of 100 nm. Figure 5 shows how the peak from the protons on the methyl group on the PO part evolves during the synthesis. We observe the transition from motionally narrowed to the static regime after a time of approximately 20 min. After a few more minutes, we also observe the formation of such large particles that the light scattering becomes clearly visible. After this simultaneous emergence of superLorentzian line shape and visible particles, the NMR signal changes slowly but continuously for another 150 min. The transition from motional narrowing to a static spectrum occurs over a relatively narrow range of correlation times on the order of a factor of 10. The time dependence of the NMR signal in this time window could then be attributed to a very slow increase in particle size. However, direct observation using TEM shows that the particles have obtained sizes substantially larger than compatible with the moderately fast tumbling motion required to produce the observed effects and the shape of the signal is not compatible with this interpretation. The more likely alternative is that within the precipitated particles there is an ongoing slow reorganization that leads to a continuous increase in the residual dipolar coupling as, for example, would occur if aggregated micelles slowly grow in size toward a more elongated shape typical for the normal hexagonal phase. For times in excess of about 25 min, we also observe a separate sharp peak superimposed on the broader main signal (see Figure 5). This extra peak has less than 10% of the total signal intensity, and it can be attributed to a

soluble fraction of the amphiphile. This was further demonstrated by separating the supernatant in which the NMR signal at 0.6 ppm consisted solely of this peak. The signal of the PO methyl protons is in this sample narrow enough for the scalar coupling to the methine proton to be resolved, and NMR self-diffusion measurements gave a diffusion coefficient of 1.6 × 10-10 m2/s, indicating a soluble polymer in monomeric form. Apparently, the batch of P123 that was used contained a short and/or EO-rich polymer fraction that appears to be incorporated into the spherical micelles but not in the more tightly packed cylindrical aggregates. TEM Observations. TEM is a useful tool to study the evolution of the synthesis. The synthesis was repeated under as equivalent conditions as possible as those in the NMR study, and samples were collected at different times. The first sample investigated was collected 20 min after the addition of the silica source, just prior to when the precipitate becomes visible to the unaided eye. At this stage, one can clearly image the micelles, which are encapsulated and held together by the polymerizing silica. The micelle-silica composites form large flocs with a size of several micrometers. The second sample was collected after 25 min, which is directly after precipitation. Figure 6a shows that the sample still contains spherical micelles, and no major differences can be detected between this sample and the previous one. However, the fact that precipitation has occurred is a sign that the aggregates have become larger and/or denser. After 40 min, the micelles have started to grow and become elongated, which is the first sign of a structural change. The sample then consists of a mixture of spherical and elongated micelles; see Figure 6b. Fifty-five minutes after TMOS addition, the sample is mainly consisting of elongated wormlike micelles, which have started to get more closely packed (Figure 6c). Only a small fraction of the micelles remain spherical. At t ) 100 min, the two-dimensional hexagonal

L(ν) )

0

(3a)

2

where

1 4π2 (ν - ν02) ) 5 T2(ν)

Silica-Block Copolymer Aggregation

structure is fully developed (Figure 6d) and the particle morphology is similar to that of the final product. We observed that the diffuse flocs created at an early stage of the synthesis gradually rearrange and turn into more well-defined particles with time. More TEM images are available as Supporting Information. The TEM images are taken on samples that after dilution have been dried primarily by liquid flow through holes of size 1 µm in the microscope grid and subsequently by evaporation in the film formed on the grid. It is, thus, pertinent to ask the question: to what extent is the aggregate size affected by the preparation procedure? First, we note that aggregates/micelles of a size smaller than the holes in the grid will primarily follow the liquid flow through the grid and they will not appear in the images. However, on the basis of both direct visual observation and NMR measurements, it is clear that for all the samples studied by TEM the source sample contains large aggregates. During the drying process, it is conceivable that these aggregates could associate to even larger systems. Two observations strongly indicate that this does not occur in the experiments. The final dry grid contains deposited flocs separated by empty areas. Thus, the flocs cannot be the result of the simple crowding of aggregates during the drying. Second, in the images of the flocs there is no indication of internal inhomogenities that one expects from a floc that has been produced by the fast association of two or more aggregates on the grid. The drying process is fast (