Langmuir 1997, 13, 4133-4141
4133
Synthesis of Periodic Mesoporous Silica Thin Films James E. Martin, Mark T. Anderson,* Judy Odinek, and Paula Newcomer Physics and Chemistry of Materials, Sandia National Laboratories, Albuquerque, New Mexico 87185-1421 Received July 19, 1996. In Final Form: May 1, 1997X We describe a synthetic method for the formation of surfactant-templated periodic mesoporous silica thin films. The films, when deposited between 0.22 < t/tgel < 0.67, are crack-free, consist of aggregated submicrometer particles, each of which contains the familiar honeycomb arrangement of 30-40 Å diameter channels, and have specific surface areas greater than 400 m2/g. The synthetic method, which we call gas-catalyzed thin film synthesis (GCTFS), involves diffusing ammonia into a homogeneous micellar coating solution on a nonporous substrate. This method ameliorates the problems associated with coating substrates from inhomogeneous solutions. We have studied and optimized GCTFS with light scattering, 29Si NMR, XRD, optical microscopy, SEM, TEM, and isothermal N2 sorption measurements.
1.0. Introduction Thin films of periodic mesoporous silica have a number of possible applications, which include size-specific coatings for surface acoustic wave (SAW) sensors,1 catalyst supports,2,3 and funtionalized optical coatings. There has been some discussion of the use of hexagonally packed mesoporous silica as a surface coating in the literature.1-10 For example, in the applications patent by Olson, Stucky, and Vartuli,1 a bulk phase was dispersed into a liquid and dip-coated onto a substrate, which resulted in a noncontinuous, nonuniform coating of colloidal particles that had been homogeneously nucleated in solution. Ogawa reported the formation of lamellar4 and hexagonal5 thin films from homogeneous acid-catalyzed alkoxysilane/surfactant solutions that were spin-coated onto nonporous substrates. Yang et al.6 reported a very nice study on the growth of oriented hexagonal thin films of mesoporous silica on a mica substrate suspended horizontally in an acidic solution of tetraethoxysilane (TEOS). They also reported the growth of free-standing, oriented mesoporous silica thin films at air-water interfaces.7 It was proposed that the growth process in these studies involved assembly of silicasurfactant micellar species on interfacially organized structures, such as hemimicellar structures, which have been observed previously on hydrophbic surfaces.8 Ayral et al.9 reported a sol-gel method to prepare surfactanttemplated membranes that exhibit ordered microporosity. Bontha et al.10 reported the synthesis of mesoporous films through liquid-liquid interfacial reactions. A key consideration in all of these methods is the ultimate accessibility of the porosity. For example, in the X
Figure 1. Dependence of the gel formation time on temperature. The dependence is essentially independent of surfactant counterion for bulk gels made from the following TMOS:CTAHS: H2O ratios: 1.0:0.124:129.0:0.289 (circles); 1.0:0.124:97.4:18.1: 0.289 (triangles); 1.0:0.130:97.4:18.1:0.289 (diamonds). The formation time vs temperature can be usefully described by an Arrhenius law (inset).
Abstract published in Advance ACS Abstracts, June 15, 1997.
(1) Olson, D. H.; Stucky, G. D.; Vartuli, J. C. U.S. Patent No. 5 364 797, 1994. (2) Kresge, C. T.; Marler, K. O.; Rav, G. S.; Rose, B. H. U.S. Patent 5 366 945, 1994. (3) Maschmeyer, T.; Rey, F.; Sankar, G.; Thomas, J. M. Nature 1995, 378, 159-162. (4) Ogawa, M. J. Am. Chem. Soc. 1994, 116, 7941-7942. (5) Ogawa, M. Chem. Commun. 1996, 1149-1150. (6) Yang, H.; Kuperman, A.; Coombs, N.; Mamiche-Afara, S.; Ozin, G. A. Nature 1996, 379, 703-705. (7) Yang, H.; Coombs, N.; Sokolov, I.; Ozin, G. A. Nature 1996, 381, 589-592. (8) Aksay, I. A.; Trau, M.; Manne, S.; Honma, I.; Yao, N.; Zhou, L.; Fenter, P.; Eisenberger, P. M.; Gruner, S. M. Science 1996, 273, 892898. (9) Ayral, A.; Balzer, C.; Dabadie, T.; Guizard, C.; Julbe, A. Catal. Today 1995, 25, 219-224. (10) Bontha, J. R.; Kim, A. Y.; Liu, J. Materials Research Society Symposium Proceedings P: Microporous and Macroporous Materials, April 8-12, San Francisco, CA; Materials Research Society: New York, 1996.
S0743-7463(96)00714-7 CCC: $14.00
Figure 2. X-ray diffraction data for an as-made mesoporous bulk gel made using a water-formamide mixture (solution A), and the same sample calcined at 550 °C in O2 for 10 h. The as-made gel has a hexagonal lattice constant of ∼40.0 Å compared to ∼36.5 Å for the calcined sample. The inset shows the higher order reflection region for the as-made sample.
case of Yang et al.6 the films are highly oriented, but the 1-d channels run parallel to the substrate surface, which means the pores may not be accessible or may be accessible © 1997 American Chemical Society
4134 Langmuir, Vol. 13, No. 15, 1997
Martin et al.
Table 1. Reactant Types and Molar Ratiosa for Coating Solutions solution A B C D E F G H I J
r
(%)b 90 25 25 25 25 25 50 50 25 25
cosolvent
wt % surfc
surf
formamide methanol methanol methanol methanol methanol formamide ethylene glycol methanol methanol
2 2 4 8 4 2 2 2 4 8
CTAB CTAB CTAHS CTAHS CTAB CTAB CTAB CTAB CTAB CTAB
catalyst NaOH NaOH NaOH NaOH NaOH NH3 NH3 NH3 NH3 NH3
TMOS
surf
water
cosolvent
catalyst
1 1 1 1 1 1 1 1 1 1
0.130 0.130 0.124 0.124 0.130 0.130 0.130 0.130 0.130 0.130
13.6 97.4 47.7 22.9 47.7 96.8 64.5 64.5 47.4 22.7
46.5 18.1 8.89 4.26 8.89 18.1 25.8 18.7 8.89 4.26
0.289 0.289 0.289 0.289 0.289 -e -
a The last five columns give molar ratios. b r is the weight percent cosolvent in the cosolvent/water mixture. c Refers the weight percent surfactant in the water/cosolvent/surfactant mixture. d The molar concentrations of TMOS are 0.40 M for 2 wt % surfactant; 0.76 M for 4 wt % surfactant, and 1.32 M for 8 wt % surfactant. e Not directly monitored; present in a large excess.
Figure 3. Scanning electron micrograph of a thin film made by spin coating premixed reactants in water-formamide (solution A with CTAHS). Solution was mixed at 0 °C and deposited on a room temperature substrate. Coverage is nonuniform and consists only of particle aggregates nucleated in solution.
Figure 4. X-ray diffraction data for an as-made mesoporous bulk gel made using a water:methanol mixture (solution B), and the same sample calcined at 550 °C in O2 for 10 h. The as-made gel has a hexagonal lattice constant of ∼45.3 Å compared to ∼38.4 Å for the calcined sample. The inset shows the higher order reflection region for the as-made sample. The coherent scattering domain size for the as-made sample is ∼140 nm.
only from the edges of the coating. We have devised a method to rapidly form periodic mesoporous thin films that have a controlled, predictable, accessible microstructure. The method builds on our previous work on making mesoporous silica gels from the molecular precursor TMOS.11 The method has the advantage that the solutions can be easily dip- or spin-coated to give films with complete
Figure 5. (a) Optical microscopy (950×) on spin-coated films made from methanol-water mixtures (solution C) showing bizarre tire tread type structures. The lines cannot be dissolved by common organic solvents and may form by the deposition of silica from convective rolls. (b) Optical microscopy (950×) in other areas of the same coated substrate. It is revealed that the lines form branched structures.
coverage and very large accessible pore volumes. We are able to make films of aggregated 150-500 nm periodic mesoporous particles; within each particle exists the familiar hexagonal array of 30-40 Å diameter channels. The film thickness can be varied from a few hundred nanometers to several micrometers. The films are stable to calcination at 500 °C and exhibit intraparticle surface areas in excess of 400 m2/g. The films are being evaluated (11) Anderson, M. T.; Martin, J. E.; Odinek, J. G.; Newcomer, P. P. Chem. Mater. submittted.
Periodic Mesoporous Silica Thin Films
Langmuir, Vol. 13, No. 15, 1997 4135
Figure 6. Scanning electron microscopy of the line structures revealing an apparent wound structure that may indicate the interaction of evolving structures with convective rolls.
Figure 7. Optical microscopy (950×) showing that a sample spin-coated from a more concentrated premixed methanolwater solution (solution D) did not give better surface coverage than those with solution C but produced sparse branched structures. Without TMOS we obtained DLA clusters from 8% CTAHS.
Figure 8. (a) Optical microscopy (375×) showing that samples spin-coated using premixed methanol:water solutions (solution D) gave the particle/line-like structures that survived a solvent rinse, which indicates that they are silica based. (b) Optical microscopy (375×) showing that some coatings contain fractallike aggregates that probably formed in bulk solution before spin casting.
as high surface area, size-selective coatings for surface acoustic wave (SAW) sensors. 2.0. Experimental Section 2.1. Coating Solutions. The molar ratios of reactants for solutions A-J are shown in Table 1. To prepare the coating solutions, deionized water, a cosolvent (e.g. methanol, ethylene glycol, or formamide), and cetyltrimethylammonium bromide or cetyltrimethylammonium hydrogensulfate (CTAB or CTAHS) are mixed to form a micellar solution. If sodium hydroxide is the catalyst, it is added to the micellar solution followed by tetramethoxysilane (TMOS). (Warning! TMOS can cause blindness; goggles should be worn.) The solution is immediately deposited onto polished silicon or glass (cover slide) substrates. If ammonia is the catalyst, the TMOS is added to the micellar solution and allowed to age 10-60 min at room temperature before it is coated onto the substrate. The ammonia is diffused into the coating after it is on the substrate. 2.2. X-ray Diffraction. Data were collected with a Scintag PAD V instrument using nickel-filtered Cu KR radiation. Data were collected in continuous scan mode from 1.5 to 10° 2θ witha 0.02° sampling interval and a 1°/min scan rate. Slits widths starting from the source were 1, 2, 1, and 0.3 mm. The tube voltage was 45 kV, and the tube current was 35 mA. Peak positions and full widths at half-height were determined with Scintag analysis software (TC9 package). Coherent scattering domain sizes were determined by application of the Scherrer equation.
Figure 9. X-ray diffraction data from bulk samples made from (a) ethylene glycol (solution H) and (b-d) methanol (solution F) indicating the hexagonal structure. The coherent scattering domain (CSD) size depends on the cosolvent, the agitation, and the temperature of the solution. 2.3. Light Scattering. Dynamic and static light scattering measurements12 were made with a 63 mW NEC He-Ne laser, (12) Berne, B.; Pecora, R. Dynamic Light Scattering; Wiley: New York, 1976.
4136 Langmuir, Vol. 13, No. 15, 1997
Martin et al.
Figure 10. Optical microscopy of a gas-catalyzed thin film made using methanol (solution F) revealing polygonal tessellation that we believe is due to silica deposition from convection cells. The convection cells are driven by the volatility and heat of vaporization of methanol. using a 256 channel Langley-Ford correlator. A Malvern indexmatching temperature-controlled scattering vat, a Malvern detector assembly, using an RCA FW130 photomultiplier tube, and an Aerotech 12 in. stepper-motor-driven goniometer complete the basic light-scattering hardware. The system is automated by a DEC PDP-11/73b computer. Intensity autocorrelation functions were fit to single exponential decays and the micelle radius was determined from the heterodyne decay rate Γ from the standard relation Γ ) Dtq2, where q is the scattering wavevector and Dt is the transitional diffusion constant. The Stokes-Einstein relation Dt ) kBT/6πη was then used to determine the radius. The static intensity data were taken on the same instrument using a statistical procedure to discard anomalously high intensity readings due to dust. Filtration of all micellar samples in a clean bench reduced dust contamination. 2.4. 29Si NMR. High-resolution spectra were recorded on a Chemagnetics instrument equipped with a 4.7 T magnet. Liquid samples were put into 20 mL quartz tubes. Spectra were recorded at a resonance frequency of 39.7 MHz, with a 50 µs pulse at 90° and a pulse delay of 120 s. Tetramethylsilane (TMS) was used as a standard to define 0 ppm. Data were analyzed and integrated with MacNMR software routines from Tecmag to determine Q ratios and shifts relative to TMS. 2.5. SEM/TEM. A JEOL 1200EX transmission electron microscope (TEM) with ASID (SEM) attachment was used to observe the microstructure of the films. To perform TEM the film was scraped off the substrate or directly formed on a 3 mm copper grid. Bright field TEM or diffraction contrast imaging was done at 120 kV and involved low (20K times) and high (300K times) magnifications of the individual grains and small aggregates. A Hitachi S4500 High-resolution field emission scanning electron microscope was used to obtain micrographs of the calcined films on silicon substrates. In this case the films were ion beam sputter coated with ∼120 Å of Cr to reduce charging. 2.6. Gas Adsorption Measurements. N2 sorption isotherms were collected at 77 K on calcined films that had been deposited on ST-cut quartz surface acoustic wave (SAW) devices.13 Prior to making measurements the system was outgassed at ∼160 °C for 2 h. Mass flow controllers were used to vary the partial pressure of nitrogen (P/Po) in a nonadsorbing helium carrier stream from ∼3% to ∼100% for the adsorption branch and ∼100% back to ∼3% for the desorption branch. The device was oscillated at ∼97 MHz and the change in frequency with changing P/Po was monitored and converted to a change in adsorbed mass. The sensitivity can be as great as ∼80 pg/cm2, where cm2 refers to the active area of the SAW covered by film (in our case ∼0.15 cm2). The BET model is then used to determine surface area, assuming 16.2 Å2 for the area of a nitrogen molecule on the surface. (13) Ricco, A. J.; Frye, G. C.; Martin, S. J. Langmuir 1989, 5, 273276
Figure 11. (a) Optical microscopy (95×) on a gas-catalyzed thin film made using water-formamide mixtures (solution G) shows uniform coverage but with some coarse cracking. (b) Cracking becomes more pronounced and occurs on a finer scale (375×) after a methanol wash. (c) Electron microscopy reveals that the street cracks prior to washing comprise a thin layer above 250 nm aggregated particles.
3.0. Results and Discussion We explored a number of strategies to make thin films before developing a gas-catalyzed synthesis; reviewing these efforts gives insight into the benefits of this approach. In many cases the film morphologies we obtained were unexpected and even bizarre. Before we were able to form
Periodic Mesoporous Silica Thin Films
Figure 12. Optical microscopy (190×) on a gas-catalyzed thin film made using ethylene glycol-water mixtures (solution H) shows a uniform thin film that shrank badly after a methanol wash, producing a marvelous pattern reminiscent of brochi. The change in the characteristic dimension of the pattern is due to varations in the film thickness; thinner films exhibit fracture on a finer scale.
Figure 13. Schematic of the environmental chamber used for dip coating.
thin films, however, we had to address several problems with the coating solutions. 3.1. Solution Homogeneity and Reaction Kinetics. In previous work on the synthesis of bulk periodic mesoporous silica, we found that we could form wellordered products from concentrated, homogeneous precursor solutions by using a combination of tetramethoxysilane, surfactant, catalyst, water, and a cosolvent.11 The water-cosolvent mixture is hydrophobic enough to solubilize the alkoxide in a matter of seconds (where it takes minutes in an purely aqueous system), yet it is hydrophilic enough to allow the surfactant to act as a chemical dipole, which is essential if it is to act as a template.11 We found that a wide variety of solvent systems meet these criteria, including water plus methanol, formamide, or ethylene glycol.11 The reaction kinetics of the TMOS/cosolvent based systems are extremely rapid, Figure 1. The mesoporous silica product forms in 5-7 s at room temperature. The rapid kinetics are advantageous because when TEOS14,15 or nonmolecular sources of silica16-18 are used, a wide variety of species are simultaneously present in the reaction mixture, such as partially hydrolyzed monomers
Langmuir, Vol. 13, No. 15, 1997 4137
Figure 14. Optical microscopy (950×) on a gas-catalyzed thin film made in an environmental chamber using water-methanol mixtures (solution F) producing uniform thin films because the net evaporation of methanol is essentially zero in this case.
Figure 15. 29Si NMR data for a neutral solution (solution J with no NH3) showing the evolution of Si-O-Si bridging oxygens (siloxane bond formation). These data show that siloxane bonds do not form until ∼30 min (t/tgel ≈ 0.3). The inset shows that the optimal deposition occurs just prior to the first siloxane bond formation to just after the formation of Q3 species. The hydrolysis of the first methoxy group occurs after about 10 min, which corresponds well with the observed time at which the solutions begins to wet the substrate.
(Si(OCH3)4-x(OH)x), silicate oligomers, colloidal products, amorphous silica, free surfactant, and precipitated product; this is not an optimal mixture from which to form mesoporous thin films. The nearly 2 orders of magnitude enhancement in reaction kinetics with TMOS vs TEOS presumably narrows the distribution. 3.2. Formation of Sodium Hydroxide Catalyzed Films from Chilled Water-Cosolvent Solutions. (14) Huo, Q.; Margolese, D. I.; Ciesla, U.; Feng, P.; Gier, T. E.; Sieger, P.; Leon, R.; Petroff, P. M.; Schuth, F.; Stucky, G. D. Nature 1989, 24, 317-321. (15) Huo, Q.; Margolese, D. I.; Ciesla, U.; Demuth, D. G.; Feng, P.; Gier, T. E.; Sieger, P.; Firouzi, A.; Chmelka, B. F.; Schuth, F.; Stucky, G. D. Chem. Mater. 1994, 6, 1176-1191. (16) Beck, J. S.; Vartuli, J. C.; Roth, W. J.; Leonowicz, M. E.; Kresge, C. T.; Schmitt, K. D.; Chu, C. T-W.; Olson, K. H.; Sheppard, E. W.; McCullen, S. B.; Higgins, J. B.; Schlenker, J. L. J. Am. Chem. Soc. 1992, 114, 10834-10843. (17) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710-712. (18) Chen, C-Y.; Li, H-X.; Davis, M. E. Microporous Mater. 1993, 2, 17-26.
4138 Langmuir, Vol. 13, No. 15, 1997
Martin et al.
Figure 16. Light-scattering intensity as a function of time for a solution made from 8 wt % CTAB, [Si] ) 1.38 M, and r ) 25% water-methanol. This figure indicates an induction time during which hydrolysis occurs but particle formation is negligible.
From initial trials, we found that the kinetics of periodic mesoporous silica formation are too rapid at room temperature to deposit films evenly and reproducibly. As shown in Figure 1, however, the gel time vs temperature shows Arrhenius behavior. We reasoned that by chilling the coating solutions, a homogeneous solution could be deposited onto a substrate and spun into a uniform film before it gelled. The shear forces during spinning provide a liquid layer of essentially constant thickness, except at the periphery of the substrate. The speed at which the substrate is spun thus fixes the reaction volume and the final thickness of the film. 3.2.1. Formamide as a Cosolvent. In initial experiments we mixed the reagents in solution A as if to make a bulk gel and spin-coated these onto a substrate. A bulk powder control was run to ensure that these reagents indeed form periodic mesoporous silica. XRD data (Figure 2) confirm that the hexagonal mesophase forms, with a lattice periodicity of about 40 Å. To make the thin films, the gel point of the coating solution was increased to about 30 s by cooling the alkaline micellar solution and the TMOS to 0 °C before mixing the two liquids. The TMOS was injected into the surfactant solution and spin-coated onto polished silicon or silica (cover slides) substrates before the solution gelled. The resulting films appeared uniform to the unaided eye, but optical and scanning electron microscopies revealed that coverage was incomplete and spotty. We varied the spinning speed from 500 rpm to 9000 rpm and time at which we spun the solution (before, during, and after formation of the product), but always obtained nonuniform coverage, Figure 3. For coating solutions that were chilled we used CTAHS for the surfactant, because quaternary ammonium surfactants with bromide or chloride anions phase separate at low temperature, but CTAHS does not. With CTAHS, we were careful to add enough extra NaOH to titrate the HSO4- ion and maintain a nominal pH of 13 after titration. We examined the effect of leaving out the water from the reactant mixture (r ) 100%). The idea being that after the thin film was spun, moisture from the atmosphere would diffuse into the solution and initiate the reaction. This did not lead to improved results; however, substantially increasing the relative humidity by heating the
Figure 17. (a) Optical microscopy (950×) on a gas-catalyzed film dip coated onto a silicon substrate in the environmental chamber using solution F showing incomplete, spotty coverage. (b) Optical microscopy on a film made using solution J showing continuous, uniform coverage on the same scale (950×).
initial film overnight in steam (100 °C) led to periodic mesoporous films that show one peak at ∼30 Å in the XRD spectrum. Unfortunately there is also a significant amount of unreacted surfactant in the film. Higher temperatures and longer reaction times may improve this approach. 3.2.2. Methanol as a Cosolvent. Because of the nonuniform films obtained with the viscous, low vapor pressure water:formamide coatings, we examined the more volatile water:methanol system. Bulk samples made using solution B always gave excellent long-range order (scattering domain sizes up to 140 nm),11 Figure 4. In an attempt to obtain better coverage than with solution A, we doubled the surfactant, TMOS, and catalyst concentrations (solution C). Figure 5a shows the parallel line structures obtained by spin coating 0 °C solutions. In other areas of the substrate coated with the TMOS solution, branched wiggly lines occur, Figure 5b. Neither type of lines can be dissolved by alcohols, amides, or ethers, which led us to believe the lines were silica-based. SEM with EDX confirms that the line structures contain silicon. Electron microscopy of these siliceous structures revealed finer detail, Figure 6, but the actual mechanism of pattern formation remains enigmatic. Given the poor coverage with solution C, we again doubled the concentration of surfactant, TMOS, and catalyst (solution D). Films made from 0 °C solutions have a sparse branched structure, Figure 7, and again have poor surface coverage.
Periodic Mesoporous Silica Thin Films
Finally, we wanted to determine the effect that the surfactant anion had on film structure and coverage. To do this we used solution E with CTAB (bromide vs hydrogen sulfate), cooled on ice to just above 10 °C. Spin coating these solutions led to the particle/linelike structures in Figure 8a, which again were not dissolved by an alcohol rinse. Some coatings contained fractal-like aggregates, Figure 8b, but all still had poor coverage. From these early studies, we realized that the rapid kinetics observed for the TMOS system are advantageous: we avoid a wide distribution of species in solution, and, as we generally use aqueous-based solutions that have a high surface tension, we can form films before the solution dewets. Nonetheless, without strict temperature control and precise timing, the kinetics are problematic: the exact instant of product formation is difficult to predict accurately enough to properly control coating time/speed to get continuous films. Thus, a method is needed to predictably delay the product formation until the coating solution is on the substrate in the appropriate thickness. 3.3. Ammonia-Catalyzed Thin Film Synthesis. In the work thus far described, we attempted to control the rate of reaction simply by reducing the temperature; however, the rates of hydrolysis and condensation of the TMOS are sensitive not only to temperature but also to solution pH. We decided to try to form a stable coating solution at pH ) 7 by leaving out the base and then diffusing the catalyst into the coating in the form of ammonia gas. The idea is that we can deposit the coating solution on a substrate, spin or drain it to the desired thickness, and, by rapid introduction of ammonia (pH switch), form a product that covers the entire substrate uniformly. 3.3.1. Bulk Products. To confirm that we could form periodic mesoporous silica using ammonia gas as the catalyst, we made bulk products by exposing the neutral reaction solution to the gas (solution F). X-ray diffraction clearly shows that this process leads to periodic mesoporous silica, Figure 9. Furthermore, the product forms initially in the top few millimeters of the static liquid within ∼10 s, which ensures that submicrometer thin films will form rapidly. 3.3.2. Methanol (r ) 25%). Our initial attempts at gas-catalyzed synthesis were with dip coating using solution F. Macroscopically, the films appeared to have uniform thickness, but optical microscopy revealed the peculiar pattern of Figure 10. We investigated the formation of this pattern by mixing methanol with an aqueous suspension of 1 µm polystyrene latex spheres that we then squirted onto a microscope slide. Under an optical microscope the latices allowed visualization of complex convection rolls and cells, the formation of which are driven by the volatility and heat of vaporization of methanol. At first, the convection was dominated by rolls, but, as the system became more concentrated, the rolls started to show instabilities along their boundaries. These periodic instabilities grew in amplitude and then pinched the roll into polygonal cells that eventually dissipated as the methanol evaporated away. We believe the pattern in Figure 10, and perhaps the wiggly line pattern (Figures 5 and 6) discussed earlier, is due to deposition of silica from convective cells. If this is indeed the case, there are two reasonable approaches to the problem; switch to a less volatile solvent system, such as water-formamide or water-ethylene glycol, or control the methanol fugacity in the coating apparatus. 3.3.3. Formamide (r ) 50%). Gas catalyzed synthesis using solution G gave the uniform films shown in Figure 11a; the street pattern of cracks is due to film shrinkage during drying. After a methanol wash, the cracking problems became more acute Figure 11b. Electron
Langmuir, Vol. 13, No. 15, 1997 4139
Figure 18. X-ray diffraction data from a calcined film spun at 1600 rpm for 5 s on polished silicon with a methanol-water coating solution (solution F) that was aged for 54 min indicating the hexagonal structure. The inset shows the convolution of the 100 and 200 Bragg reflections, which is similar to that seen in the bulk sample with a 17 nm CSD.
microscopy of the films revealed the formation of clustered 250 nm particles beneath a very thin continuous layer, Figure 11c. The cracking problems arise because the films are too thick to withstand the tensile stresses developed during drying. In an attempt to produce thinner films, the substrates were drawn from the coating solutions more slowly, but because of the high viscosity, the solution drained slowly, dewet, and tended to give spotty coverage. 3.3.4. Ethylene Glycol (r ) 50%). Gas catalyzed synthesis using solution H gave a uniform film that shrank severely upon rinsing with methanol, creating the marvelous fragmentation pattern in Figure 12, reminiscent of the recursive bronchi in Mandelbrot’s The Fractal Geometry of Nature.19 We found that ethanol washes minimize film cracking problems, so we adopted this procedure in further studies. Again the high viscosity of the coating solution gave coverage, cracking, and uniformity problems. 3.3.5. Methanol (r ) 25%) and an Environmental Chamber. Because of the viscosity and coverage problems with formamide and ethylene glycol, we decided to explore the second approach to eliminating the formation of convection cells with methanol, i.e. control the methanol fugacity during coating. We built a simple environmental chamber, Figure 13, for dip coating substrates and catalyzing product formation with ammonia gas. Into this chamber we placed a small dish of the r ) 25% methanol mixed solvent, thus ensuring the fugacity of the methanol in the coating closely matched that in the chamber. Because the net evaporation of methanol is essentially zero, we were able to produce uniform thin films that did not dewet (Figure 14) using solution I. 3.3.6. Chemistry of the pH ) 7 Water-Methanol Coating Solution. After solving the convective problems with the environmental chamber, we focused on the chemistry of the r ) 25% methanol coating solutions. The neutral coating solution with 8% CTAB and [Si] ) 1.38 M remains clear for ∼90 min before a chemical gel forms. Thus, we wanted to determine the optimal time at which (19) Mandelbrot, B. B. The Fractal Geometry of Nature; Freeman: New York, 1977.
4140 Langmuir, Vol. 13, No. 15, 1997
Martin et al.
Figure 20. Lattice image TEM of gas-catalyzed thin film made using a concentrated methanol-water mixture (Solution J) show regions of well-ordered hexagonal silica. Some areas also appear to have uniform pores that are randomly arranged. This microstructure is expected based on the small CSD size measured in XRD.
Figure 19. (a-c) High resolution scanning electron micrographs of calcined periodic mesoporous silica showing a interconnected network of aggregated 150-500 nm particles that bridge the entire substrate. The sample is the same as in Figure 18. (Scale bar on “a” is 50 nm).
to dip the substrates. We examined the evolution of silicate and surfactant species with static light scattering, dynamic light scattering, and 29Si NMR. The NMR measurements, Figure 15, show that in the [Si] ) 1.38 M solution, Q01OH species form after ∼10 min, Q1 species form after ∼30 min, Q2 species begin to form after ∼60 min, and Q3 and Q4 species begin to form after ∼70 min. From light scattering measurements on solution J (no NH3), Figure 16, we see that the species in solutionsmicelles, and partially hydrolyzed alkoxide monomers and oligomerssdo not scatter significantly until after about 1 h of aging, i.e. until Q2 species begin
to form. The chemical gels that form at ∼90 min are amorphous in X-ray diffraction, and therefore they are not a desired product. Thus, we find the optimal time window to deposit the coating solution is between 20 and 60 min (0.22 < t/tgel
