2572
Langmuir 2001, 17, 2572-2574
Surface Patterned Porous Films by Convection-Assisted Dynamic Self-Assembly of Zeolite Nanoparticles Huanting Wang, Zhengbao Wang, Limin Huang, Anupam Mitra, and Yushan Yan* Department of Chemical and Environmental Engineering, University of California, Riverside, California 92521 Received February 16, 2001 This study demonstrates for the first time that porous zeolite films with surface patterns such as knottedrope web and wrinkled honeycomb can be obtained by convection-assisted dynamic self-assembly of zeolite nanoparticles. The study also shows that an appropriate dispersant and the presence of zeolite nanoparticles with a specific range of particle sizes in the colloidal suspension are critical for pattern formation. The patterned zeolite films have a well-defined bimodal pore size distribution (i.e., 0.55 and 2.6 nm) with a high Brunauer-Emmett-Teller surface area of 680-750 m2/g and are attractive for potential applications in fields such as thin film catalysis, adsorption separation, chemical sensor arrays, and components in microelectronic devices.
Patterned porous films have potential applications as selective chemical sensor arrays and components in microelectronic devices and have been fabricated mostly by micromolding techniques.1-3 At present, all of these studies are limited to surfactant templated mesoporous silica because this class of porous material can be readily synthesized at ambient conditions.4 By contrast, zeolite films are usually synthesized in hydrothermal conditions where a chemically aggressive environment makes direct patterning extremely difficult.5 Recently, we have successfully overcome this difficulty and fabricated patterned zeolite films by combining micromolding with selfassembly of zeolite nanocrystals.6 In this study, we attempt to demonstrate for the first time that surface patterned zeolite films can also be obtained through dynamic selfassembly of zeolite nanoparticles assisted by the wellknown Benard-Marangoni convection. This approach is simple because it does not require a mold. Flow patterns associated with Benard-Marangoni (surface-tensiondriven) convection have been extensively studied both experimentally and theoretically because of their fundamental and practical importance in fluid physics.7 These patterns typically resulted from a vertical temperature gradient in a horizontal thin fluid layer heated from below. Solvent evaporation from a thin film can be another route to generate the vertical temperature gradient and has been recently exploited for pattern formation in solid polymer films.8,9 Inorganic particles have been previously used as labels for visualization of various fluid flows * Corresponding author. E-mail:
[email protected]. Phone: 909-787-2068. Fax: 909-787-2425. (1) Trau, M.; Yao, N.; Kim, E.; Xia, Y.; Whitesides, G. M.; Aksay, I. A. Nature 1997, 390, 674-676. (2) Yang, P. D.; Deng, T.; Zhao, D. Y.; Feng, P. Y.; Pine, D.; Chmelka, B. F.; Whitesides, G. M.; Stucky, G. D. Science 1998, 282, 2244-2246. (3) Fan, H. Y.; Lu, Y. F.; Stump, A.; Reed, S. T.; Baer, T.; Schunk, R.; Perez-Luna, V.; Lopez, G. P.; Brinker, C. J. Nature 2000, 405, 5660. (4) Sellinger, A.; Weiss, P. M.; Nguyen, A.; Lu, Y. F.; Assink, R. A.; Gong, W. L.; Brinker, C. J. Nature 1998, 394, 256-260. (5) Yan, Y.; Davis, M. E.; Gavalas, G. R. Ind. Eng. Chem. Res. 1995, 34, 1652-1661. (6) Huang, L. M.; Wang, Z. B.; Sun, J. Y.; Miao, L.; Li, Q. Z.; Yan, Y.; Zhao, D. Y. J. Am. Chem. Soc. 2000, 122, 3530-3531. (7) Schatz, M. F.; VanHook, S. J.; McCormick, W. D.; Swift, J. B.; Swinney, H. L. Phys. Fluids 1999, 11, 2577-2582. (8) Widawski, G.; Rawiso, M.; Francois, B. Nature 1994, 369, 387389.
including Benard-Marangoni convections.10,11 However, these particles are usually large (i.e., micrometer size), and their vitrification into a solid patterned structure has not been reported. In addition, this study shows that nanometer-sized particles are essential for the formation of observed patterns, and thus vitrified patterns are unlikely from micrometer-sized particles. The zeolite silicalite nanoparticle suspensions used in this study were similarly synthesized hydrothermally as reported previously.6,12 One major difference is that extra ethanol was added to the synthesis solution and the assynthesized suspension was used directly without centrifugation separation. The molar composition of the synthesis solution is 1 TPAOH/2.8 TEOS/22.4 EtOH/40 H2O (TPAOH for tetrapropylammonium hydroxide, TEOS for tetraethyl orthosilicate, and EtOH for ethanol). Two batches (noted as A and B) of synthesis solution underwent different hydrothermal histories. Batch A was heated at 70 °C for 9 days, and batch B was heated at 80 °C for 3 days. The resulting milky colloidal suspensions were cooled to room temperature under stirring. The suspensions are stable for months in ambient laboratory conditions, and they contain 25-50 nm diameter zeolite nanocrystals,6,12 1.3 × 4 × 4 nm dimension nanoslabs, and nanoslab aggregates.13,14 The suspension was dropped on a horizontal clean-roomgrade silicon wafer to form a ca. 2 mm thick liquid film with diameter of 2 cm. Surface patterns spontaneously formed throughout the surface after drying at ambient temperature. Typical as-formed patterns include knottedrope web (batch A, Figure 1a,b) and wrinkled honeycomb (batch B, Figure 1c,d). The surface patterns remain unchanged and the films become quite strong mechanically (cannot be scratched off easily) after removal of the organic template by calcination at 500 °C for 15 h. (9) Li, M. Q.; Xu, S. Q.; Kumacheva, E. Langmuir 2000, 16, 72757278. (10) Freymuth, P. Rev. Sci. Instrum. 1993, 64, 1-18. (11) Ball, P. The Self-Made Tapestry: Pattern Formation in Nature; Oxford University Press: New York, 1999. (12) Wang, H. T.; Wang, Z. B.; Yan, Y. Chem. Commun. 2000, 23332334. (13) Kirschhock, C. E. A.; Ravishankar, R.; Jacob, P. A.; Martens, J. A. J. Phys. Chem. B 1999, 103, 11021-11027. (14) Kirschhock, C. E. A.; Ravishankar, R.; Van Looveren, L.; Jacob, P. A.; Martens, J. A. J. Phys. Chem. B 1999, 103, 4972-4978.
10.1021/la0102509 CCC: $20.00 © 2001 American Chemical Society Published on Web 04/03/2001
Letters
Langmuir, Vol. 17, No. 9, 2001 2573
Figure 1. Scanning electron microscope images of surface patterns formed by dynamic self-assembly of zeolite nanocrystals on silicon wafers at ambient conditions (60% relative humidity, 25 °C): (a and b) batch A material with (a) top view and (b) view at a tilt angle of 50° and (c and d) batch B material with (c) top view and (d) view at tilt angle of 50°. Table 1. Characterization of Batches A, B, and C Dried at Room Temperature and Calcined at 500 °C for 15 h
sample batch A batch B batch Ca
BET micropore micropore mesopore mesopore mesopore total pore volume surface area volume surface area surface area diameter pore volume (at P/P0 ) 0.98) crystallinityb d spacingc (cm3/g) (m2/g) (cm3/g) (m2/g) (m2/g) (nm) (cm3/g) (%) (nm) 748 684 723
0.15 0.14 0.14
332 322 233
416 632 490
2.65 2.60 3.10
0.17 0.13 0.31
0.38 0.34 0.45
80 64 75
6.68 8.83
a Batch C is the same as batch B, except for no addition of ethanol (refs 6 and 12). b The crystallinity is estimated from IR spectra (ref 15). c The d spacing is calculated from low-angle XRD patterns.
N2 adsorption measurements show that pattern A and B materials have high Brunauer-Emmett-Teller (BET) surface areas (748 m2/g for A and 684 m2/g for B, Table 1). Detailed analysis of adsorption-desorption isotherms (see Figure 1 in the Supporting Information or S-Figure 1) clearly shows that patterned materials have a welldefined bimodal pore size distribution (i.e., narrow pore size distributions in both the micropore and mesopore regions). For example, pattern A material has a micropore diameter of 0.55 nm, a micropore volume of 0.15 cm3/g, and a micropore surface area of 332 m2/g and a mesopore diameter of 2.65 nm, a mesopore volume of 0.17 cm3/g, and a mesopore area of 416 m2/g (Table 1). Wide-angle X-ray diffraction (XRD) patterns (S-Figure 2) of both pattern A and B materials match typical silicalite structures. Low-angle XRD (S-Figure 2) shows a peak at 2θ ≈ 1.00-1.15 (d ) 8.83-6.68 nm) confirming presence of mesopores owing to close packing of nanoparticles in the patterned materials. The estimated mesopore wall thickness from nitrogen adsorption and low-angle XRD is 4.1-6.3 nm. Transmission electron microscopy (TEM) (S-Figure 3) reveals disordered mesopores consistent with N2 adsorption and XRD results. Fourier transform infrared (FT-IR) spectra (S-Figure 4) show typical MFI framework bands including the characteristic double ring vibration at ca. 550 cm-1 and a shoulder at 980 cm-1 attributable to Q3
Si-OH groups. The crystallinity for pattern A and B materials is estimated to be 80% and 64%, respectively15 (Table 1). Thermogravimetric (TG) analyses (S-Figure 5) show weight loss of 18% for A and 23% for B at T > 170 °C because of decomposition of occluded TPA. These two values are greater than a theoretical value of 11.7 wt % for pure silicalite (four TPA+ cations per unit cell13), and this implies that more TPA was trapped in the patterned materials. It is possible that the extra TPAOH molecules reside in mesopores. Differential thermogravimetric (DTG) curves (S-Figure 5) show that the decomposition peak is at 200 °C for A and 170 °C for B, values which are lower than the decomposition temperature for pure silicalite nanocrystals. This again suggests that some TPAOH molecules are located in mesopores and decompose more readily. Figure 2 illustrates a possible pattern formation mechanism via evaporation-driven convection. After a liquid film is formed on a horizontal flat surface, EtOH preferentially evaporates, removing heat from the film surface, and leads to a vertical temperature gradient; this gradient in turn may lead to Benard-Marangoni convection across the fluid layer (Figure 2a). The suspension wells up in the center of the convection cell and flows back down along (15) Coudurier, G.; Naccache, C.; Vedrine, J. C. J. Chem. Soc., Chem. Commun. 1982, 1413-1415.
2574
Langmuir, Vol. 17, No. 9, 2001
Figure 2. Schematic illustration of the proposed spontaneous pattern formation during evaporation: (a-d) cross-sectional view of the proposed Benard-Marangoni convection flow pattern during evaporation (fluid layer becomes thinner as evaporation proceeds) and (e) plot of weight change vs evaporation time (weight measured by a microbalance).
the cell boundary. The convection during initial solvent evaporation is verified by in situ optical microscope observation. As evaporation proceeds, the colloidal suspension becomes more concentrated, and its viscosity increases significantly, changing the characteristics of Benard-Marangoni convection (e.g., the Marangoni number is changing). Some nanoparticles may agglomerate and settle to the silicon wafer surface to form a static layer due to favorable hydrogen bonding interaction among the nanoparticles and between the particles and the silicon surface.6 Because convection-cell size is scaled with fluid layer thickness, it is expected that the convection-cell size decreases as convection depth shrinks7 (Figure 2b,c). Eventually, the suspension dries up and the zeolite nanoparticles are locked into the observed solid patterns (Figure 2d). The weight change of the suspension layer was measured by a high-sensitivity balance and is reported in Figure 2e. The evaporation rate remains almost constant until around a weight loss of 50% because of preferential evaporation of ethanol (total ca. 55 wt % ethanol in the nanoparticle suspension). This constant evaporation is important to provide the uniform driving force needed for the pattern formation. Some secondary (underdeveloped) patterns are observed underneath the primary patterns (Figure 1a,b), and this can be easily explained if the convection cells indeed shrink as the fluid layer becomes thinner as we have assumed. Well-defined hexagonal, pentagonal, or square patterns are expected from an ideal time-independent Benard-Marangoni convection,7 whereas the cells of our patterns are slightly irregular and their edges are highly wrinkled. This may have been the result of time-dependent convection flow and the final drying step. It should be noted that the colloidal system used here is much more complex than the single homogeneous fluid used in an ideal Benard-Marangoni convection. For example, ethanol may have been preferentially depleted from regions close to the liquid-air interface, leading to
Letters
coupled thermosolutal Benard-Marangoni convection.16 In addition, patterned deposition on a silicon surface may provide different boundary conditions and thus affect the convection flow pattern, which in turn impacts the final vitrified surface pattern.17 Although not a strong amphiphile, TPAOH may still act as a weak surfactant that influences surface tension and thus the surface-tensiondriven Benard-Marangoni convection. Incorporation of ethanol into the synthesis solution before hydrothermal treatment appears essential for the surface pattern formation because the zeolite nanoparticle suspension prepared without additional ethanol did not produce any patterns (batch C). Extra ethanol is known to slow the silicalite crystallization process to produce small and uniform zeolite nanocrystals that self-assemble better than large zeolite crystals. Also, a nanosuspension with a range of particle sizes appears to be important for pattern formation. No pattern was observed with a suspension that contains monodisperse 50 nm diameter zeolite nanocrystals (monodisperse zeolite nanocrystals are collected by separation with centrifugation). As mentioned earlier, the suspension we used contained zeolite nanocrystals (25-50 nm diameter), nanoslabs, and nanoslab aggregates.13,14 It is believed that the small nanoslabs and nanoslab aggregates serve as glue holding nanoparticles in place during the pattern formation process. In addition, concentrations of ethanol, TPAOH, and water in the final suspension are also important (SFigure 6). For example, addition (postsynthesis) of ethanol, TPAOH, and water produces wavelike patterns, patterns that are intermediate between knotted web and wavelike patterns, and shallow knotted web patterns, respectively. In addition, when ethanol was replaced with 2-propanol or butanol in the colloidal suspensions, only phaseseparated patterns were observed (S-Figure 7). We have demonstrated that surface patterns can be generated via a simple evaporation-convection process. This approach could be a general strategy to fabricate surface patterns from other colloidal systems. This study also provides a new route to the hierarchical highly crystalline porous materials with well-defined microporosity and mesoporosity that are attractive for catalysis and separation because of facilitated mass transport through the mesopores.2 In particular, patterned zeolite films with mesoporosity may find potential applications in fields such as thin film catalysis, adsorption separation, chemical sensor arrays, and components in microelectronic devices.18 Acknowledgment. This work was supported by Honeywell, UC-SMART, UC-TSR&TP, UC-EI, and CECERT. We thank Professor Harry W. Green of U. C. Riverside for allowing us to use his polarized optical microscope. Supporting Information Available: S-Figures 1-7. This material is available free of charge via the Internet at http://pubs.acs.org. LA0102509 (16) Coriell, S. R.; Cordes, M. R.; Boettinger, W. J. J. Cryst. Growth 1980, 49, 13-28. (17) Sullivan, T. S.; Liu, Y. M.; Ecke, R. E. Phys. Rev. E 1996, 54, 486-495. (18) Wang, Z. B.; Wang, H. T.; Mitra, A.; Huang, L. M.; Yan, Y. Adv. Mater., in press.