Sie sind auf Seite 1von 14

Chem. Mater.

1996, 8, 1451-1464

1451

Synthesis of Pure Alumina Mesoporous Materials


Fre de ric Vaudry, Shervin Khodabandeh, and Mark E. Davis*
Chemical Engineering, California Institute of Technology, Pasadena, California 91125 Received January 18, 1996. Revised Manuscript Received April 29, 1996X

Alumina mesophases have been synthesized by reacting aluminum alkoxides and carboxylic acids with controlled amounts of water in low-molecular-weight alcoholic solvents. Calcination of these materials yields aluminas that are thermally stable to 1073 K and contain randomly ordered pores. Specific surface areas as high as 710 m2/g and narrow pore size distributions centered at 20 that do not contain zeolitic micropores are exhibited by the calcined solids.

Introduction Activated aluminas are attractive catalysts for processes such as petroleum hydrodesulfurization, the Claus reaction, the dehydrogenation of butane to give butenes and the dehydration of alcohols to give alkenes. One of the major problems related to the use of alumina catalysts is the deactivation by coke formation and pore plugging that hinders the diffusion of reactants and products in and out the catalyst particles. It is known that the larger the contribution of micropores to the specific surface area and the wider the pore size distribution, the greater the enhancement in the deactivation rate. Thus, synthesis of aluminas with porosity properties comparable to those exhibited by the new high- and all-silica mesoporous1,2 materials would be of industrial interest. The silica-based mesoporous materials developed by researchers at Mobil,1,2 e.g., M41S, were recently prepared by organizing silica with organic surfactants. These materials can exhibit cubic (MCM48) or hexagonal (MCM-41) symmetries. Thermal decomposition of the surfactant allows for the development of narrow pore size distributions in the range 15100 and BET specific surface areas as high as 1000 m2/g. Mesoporous materials of this type are not restricted to silica, and MCM-41 type materials have been reported recently for oxides of Ti,3 Sb, and Pb.4,5 Our early work on pure-alumina mesoporous materials revealed that aqueous solutions of cationic surfactants do not yield mesophases and, as already reported in the literature,5 that aqueous solutions of anionic surfactants such as alkyl phosphates or sodium dodecylbenzenesulfonate can promote the formation of thermally unstable lamellar phases as the only mesophases. Recently, the synthesis of all-alumina mesoporous materials
* To whom correspondence should be addressed. X Abstract published in Advance ACS Abstracts, June 15, 1996. (1) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710. (2) 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.; Sclenker, J. L. J. Am. Chem. Soc. 1992, 114, 10834. (3) Antonelli, D. M.; Ying, J. Y. Angew. Chem. 1995, 34, 2014. (4) Huo, Q.; Margolese, D. I.; Ciesla, U.; Feng, P.; Gier, T. E.; Sieger, P.; Leon, R.; Petroff, P. M.; Schu th, F.; Stucky, G. D. Nature 1994, 368, 317. (5) Huo, Q.; Margolese, D. I.; Ciesla, U.; Demuth, D. G.; Feng, P.; Gier, T. E.; Sieger, P.; Firouzi, A.; Chmelka, B. F.; Schu th, F.; Stucky, G. D. Chem. Mater. 1994, 6, 1176.

has been mentioned.6 The procedure (not given in detail) involves the use of polyglycols as surfactants. The solids after calcination at 873 K can develop a 400 m2/g BET specific surface area. The objective of our work is to synthesize pure aluminas with high surface areas and narrow pore size distributions that do not contain micropores. Most of the aluminas used for catalytic applications are prepared by precipitation, drying, and calcination of aluminum (oxo)hydroxides. The pseudoboehmite Catapal alumina is among the best catalyst base materials commercially available because it can be obtained essentially free from impurities such as sodium. This pseudoboehmite is synthesized as a byproduct of the manufacture of linear C2-C20 alcohols and the process involves the hydrolysis of the corresponding aluminum alkoxides in organic solvents using controlled amounts of water (Ziegler process).7 Here, we describe a variation of this process where an organic surfactant and an aluminum alkoxide are reacted with controlled amounts of water in organic solvents. Calcination of solids thussynthesized gives activated aluminas with unique porosity properties, e.g., specific surface areas as high as 710 m2/g and narrow pore size distributions centered at around 20 . Experimental Section
Chemicals. Formamide, ethanol, 2-propanol, diethyl ether, ethyl acetate, and pentane were purchased from EM Science, 2-butanol from Fisher; chloroform from Mallinckrodt, and 1-propanol, 1-hexanol, and 1-nonanol from Aldrich. The aluminum-containing reactants and the surfactants were purchased from Aldrich, except for neodecanoic acid (Pfaltz and Bauer), lauryl phosphate (Lancaster), and the chloride form of the aluminum Keggin ion (Reheis Inc.). Syntheses in Formamide. A typical synthesis in formamide yielding a solid denoted as solid F is as follows: 1 g of aluminum sec-butoxide was hydrolyzed in 15.1 g of water, followed by stirring for 1 h. A solution of 0.49 g sodium dodecylbenzenesulfonate in 5.05 g of formamide was then added. After 5 min of aging, the synthesis mixture was heated at 383 K for 2 days. The solid was recovered by filtration, washed with deionized water, and dried at room temperature. Syntheses in Alcohols. Solids denoted as solids JAx were prepared in 1-propanol (alcohol solvent denoted by subscripted
(6) Bagshaw, S. A.; Prouzet, E.; Pinnavaia, T. J. Science 1995, 269, 1242. (7) Misra, C. Industrial Alumina Chemicals; ACS Monograph 184: Washington, DC, 1986.

S0897-4756(96)00033-6 CCC: $12.00

1996 American Chemical Society

1452

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.
for selective polarization transfer to the central (+1/2 T -1/2) transition was established on a sample of kaolin by satisfying the condition 3AlBAl ) HBH. 27Al chemical shifts are reported relative to 1.0 M Al(NO3)3 solution (chemical shift ) 0.00 ppm) and are not corrected for second-order quadrupolar effects. Quadrupolar interactions in nuclei with I > 1/2 such as 27Al (I ) 5/2) have the potential of yielding quantitation of NMR spectra difficult and ambiguous. In particular, for aluminum sites with large quadrupolar coupling constants (QCCs) the line shapes could be broadened to such an extent that would yield those sites invisible by NMR spectroscopy. Thus, one needs to establish if all of aluminum magnetization is observed in the NMR spectrum. This was determined by comparing the signal intensities of known amounts of well-characterized samples to that of a known amount of well-characterized kaolin (for which 100% of aluminum sites are expected to be NMRvisible). 1 H (300.15 MHz) NMR spectra were measured at spinning rates of 11.7 kHz using 4-mm ZrO2 rotors spinning in air or dry N2. 1H NMR chemical shifts were referenced to adamantane (1.74 ppm relative to TMS) and are reported relative to TMS. 13C (75.47 MHz) NMR spectra were measured using cross polarization with a 1H 90 pulse of 7 s and contact times of 5-10 ms at spinning rates of 2.6-3.6 kHz utilizing 7-mm rotors. 13C NMR chemical shifts were referenced to external adamantane (downfield resonance at 38.4 ppm versus TMS). Spectral simulation and integration were performed using Bruker LINESIM and QNMR softwares.

A) at a temperature x (in kelvin) using a carboxylic acid J. The composition of the synthesis mixture is as follows:

Al(OCH(CH3)C2H5)3:3.2H2O:0.30CnH2n+1CO2H: 26C2H5CH2OH
J is either caproic acid (n ) 5, J ) C), lauric acid (n ) 11, J ) L) or stearic acid (n ) 17, J ) S). As an example, the synthesis procedure for LA383 (using lauric acid in alcohol at 383 K) is given below: An aluminum hydroxide suspension was obtained by hydrolysis of 43.8 g of aluminum sec-butoxide with 10.3 g of deionized water in 275 g of 1-propanol (99+%). After 60 min of stirring, 10.8 g of lauric acid (99.5+%) was added. The mixture was aged for 24 h at room temperature and heated under static conditions at 383 K in a 1-L glass jar for 2 days. The solid was filtered, washed with ethanol, and dried at room temperature. Syntheses in Other Solvents. The organic surfactant was dissolved in the organic solvent. Water and the aluminumcontaining reagent were subsequently added. A 15-h aging at room temperature preceded the thermal treatment. As an example, the synthesis procedure using lauric acid in pentane at 298 K is given below: A solid denoted as solid LP298 was synthesized in pentane at 298 K using lauric acid from a mixture of the following composition: Al(OCH(CH3)C2H5)3:3.0H2O:0.32C11H23CO2H: 15C5H12. A solid denoted as solid LE298 was synthesized in diethylether at 298 K using lauric acid from a mixture of the following composition: Al(OCH(CH3)C2H5)3:3.0H2O:0.31C11H23CO2H:20(C2H5)2O. Calcination Procedures. The solids were generally calcined for 2 h at temperatures ranging from 673 to 873 K with a temperature ramp of 0.5 K/min from room temperature to the calcination temperature. The calcination atmosphere was either air or nitrogen during the ramp and air at the final temperature. Characterization Methods. XRD patterns were recorded on a Scintag XDS 2000 diffractometer using Cu KR radiation. The diffracted beam was detected by a liquid nitrogen cooled germanium solid-state detector. The samples were analyzed in the 2 range 1-51 in steps of 0.03. Temperatureprogrammed XRD was recorded in a high-temperature diffractometer chamber under flow of air or air saturated with water at flow rates of 500 mL/min. The heating was carried out with a temperature ramp of 10 K/min and the sample was maintained at each set point for 45 min. Scans were taken in the range 1-12.5 2 with a total scan time of 2 h for each temperature. TGA patterns were collected on a Du Pont 951 thermogravimetric analyzer. The samples were heated in air and the temperature ramp was 1 K/min. Nitrogen adsorption/desorption experiments were carried out at 77 K on an Omnisorp Coulter 100CX or a Micromeritics ASAP 2010 analyzer. The pore size distributions were calculated using the Barret-Joyner-Hallender (BJH) method on the desorption branch. n-Heptane (99%, Aldrich) and neopentane (99%, Wiley) adsorption experiments were performed at 298 K using a McBain-Bakr balance. The saturation pressures at 298 K for the two hydrocarbons are 46 and 1306 Torr, respectively. The calcined materials were outgassed at 523 K under vacuum prior to the adsorption experiments. Mid-infrared spectra of as-made solids were recorded on a Nicolet 800 FTIR spectrometer using 1-3 wt % sample in KBr pellets. Solid-state NMR spectroscopy was performed on a Bruker AM 300 spectrometer equipped with a high-power assembly for solids. 27Al spectra were recorded with magic angle spinning (MAS) at a frequency of 78.2 MHz utilizing 4-mm ZrO2 rotors spinning in air or dry N2 at rates of 5-12 KHz. Flip angles of 10.0-15.0 were used for 27Al MAS NMR experiments. 1H-27Al cross-polarization (CP) spectra were recorded at the magic angle with spinning rates of 6.6 kHz and 1H 90 pulses of 4.3 s. The Hartmann-Hahn condition

Results and Discussions The nature of mesophases obtained using the liquidcrystal templating route1,2,4,5 is strongly influenced by electrostatic and steric interactions between solvent molecules, inorganic species, and self-assembled organic surfactants. To date, syntheses of mesoporous materials in polar organic solvents have received less attention than nonaqueous syntheses of zeolites. Although there are many studies on the solubility of surfactants in a variety of solvents, very few reports of liquid crystal phases in non-aqueous solvents have appeared. Auvray et al. have constructed a phase diagram of cetyltrimethylammonium bromide and sodium dodecyl sulfate in formamide that shows phase transitions similar to those observed in water.8,9 Indeed, these authors observed phase transitions from isolated rodlike micelles to a hexagonal phase and then to cubic and lamellar phases upon increasing the surfactant concentration. Moreover, Kalman et al.10 pointed out that formamide and water have similar structural characteristics; every molecule gives rise to four hydrogen bonds in both liquids. Using this information, Kim et al.11 established a nonaqueous liquid-crystal-templating route in formamide for the synthesis of a high-titania (Si, Ti) MCM41 type materials that used a titanium alkoxide reagent (hydrolyzed with controlled amounts of water), and n-alkyltrimethylammonium halogenides as surfactants. Our first syntheses of alumina mesophases were by methods adapted from both the nonaqueous route of Kim et al. and the pseudoboehmite synthesis via the Ziegler process. In our hands, syntheses in formamide using nalkyltrimethylammonium ions as cationic surfactants
(8) Auvray, X.; Petipas, C.; Anthore, R.; Rico, I.; Lattes, A. J. Phys. Chem. 1989, 93, 7458. (9) Auvray, X.; Perche, T.; Anthore, R.; Petipas, C.; Rico, I.; Lattes, A. Langmuir 1991, 7, 2385. (10) Kalman, E.; Serke, I.; Palinkas, G.; Zeidler, M. D.; Wiesmann, F. J.; Bertagnolli, H.; Chieux, P. Z. Naturforsch. 1983, 38a, 231. (11) Kim, A. Y.; Liu, J.; Virden, J. W.; Bunker, B. C. Mater. Res. Soc. Symp. Proc. 1994.

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1453

Figure 1. XRD pattern of solid F: alumina synthesized in formamide with sodium dodecyl benzenesulfonate. Inset: TGA curve for solid F.

were unsuccessful. However, nonlamellar mesophases were obtained for syntheses carried out with sodium dodecylbenzenesulfonate as an anionic surfactant (solid F). No well-defined morphology is discernible for these solids using either SEM or TEM. The XRD pattern of solid F is given in Figure 1. In addition to the intense peak at high d spacing, four broad, low-intensity peaks are also apparent and are assignable to pseudoboehmite. XRD patterns with a single intense line at high d spacings are typically observed from mesoporous materials featuring randomly ordered pores.12 This XRD line is much broader than the high-intensity line exhibited by hexagonal, cubic, and lamellar phases where repeating units are very homogeneous in size. A shift to higher d spacings is observed upon calcination, and the line finally disappears for calcination temperatures higher than 783 K. Only octahedrally coordinated aluminum is detected in solid F by 27Al MAS NMR. The surfactant can not be extracted from the solid by contact with HCl/ethanol, HCl/water, or AcOH/ water solutions. The inset of Figure 1 shows the TGA of solid F. The data typically exhibit a 5-10% weight loss below 400 K corresponding to bulk and physisorbed water and a 40-50% weight loss between 500 and 800 K from both the decomposition of the anionic surfactant and the removal of structural water. The surfactant is never totally decomposed at 773 K. A solid F calcined at 783 K for 1 h that still contains organic residues exhibits a BET specific surface area of 500 m2/g. At a calcination temperature above 793 K, the high d spacing XRD line disappears. Elemental analyses show that calcined solid F prepared as above still contains substantial amounts of sulfur that could play a role in the structural collapse of the material. For this reason, we abandoned the use of sodium dodecyl benzenesulfonate as the surfactant. To prepare pure alumina mesophases from anionic surfactants that would be completely removed by calcination, we explored syntheses using carboxylic acids. Syntheses using lower molecular weight carboxylic acids such as caproic acid yielded gels in formamide that after washing with ethanol gave no XRD line at high d
(12) Davis, M. E.; Chen, C.-Y.; Burkett, S. L.; Lobo, R. F. Mater. Res. Soc. Symp. Proc. 1994, 346, 831.

spacings. The solubilities of heavier molecular weight carboxylic acids such as lauric acid in formamide are low (solubilities are also low in formamide/water cosolvents). However, their solubilities in ethanol are quite high. Thus, new syntheses were conducted in ethanol/ formamide/water cosolvents and in ethanol/water cosolvents. Only the latter solvent system yielded aluminas whose XRD patterns revealed a high-intensity peak at high d spacings, indicating that formamide is not necessary for mesophase synthesis. Syntheses performed in ethanol alone yielded solids featuring a single XRD line at high d spacing. We were also able to employ solvents such as ethylacetate, diethylether and pentane for the synthesis of solids with similar characteristics (vide infra). The synthesis route developed here using the aluminum alkoxide/alcohol/carboxylic acid system yields alumina mesophases over a variety of conditions, and some of these are reported in Table 1. The aging and hydrolysis times are found to have very little effect on the formation of alumina mesophases. Branched or normal alcohols ranging from methanol to nonanol can be used as a solvent, while normal or branched carboxylic acids ranging from propionic to stearic acid are suitable as surfactants. The solids are generally formed at room temperature. Nevertheless, the synthesis mixtures were heated because of a general trend observed for MCM-41 silicates that shows that increasing the synthesis temperature improves the thermal stability of the materials (until demicellization of the surfactant occurs).12 Higher temperature syntheses can be carried out in higher boiling point alcohols. However, the aluminum hydroxide suspension obtained at 298 K in 1-hexanol and 1-nonanol coagulates to gels when heated at 423 and 473 K, respectively. A gelation temperature exists for the synthesis mixture and is between 411 and 423 K when 1-propanol is the solvent. Although some gels can yield a high d spacing XRD line, their carboxylic acid content observed by TGA is very low. Consequently, the solids obtained below the gelation temperature are preferred. The most thermally stable solid prepared here is a lauric acid-alumina obtained in 1-propanol that was heated for 2 days at 411 K with rotation of the reaction vessel (solid LA411). This solid maintains the XRD line at high d spacings even after 4 h of heating under air at 1073 K. The XRD pattern of solid LA411 is given in Figure 2. In addition to the high-intensity, high d spacing line in the XRD pattern of LA411, a low-intensity, broad line centered at about 17-20 (d spacings of 5.2-4.4 ) is also present indicating that the inorganic portion of the composite material is amorphous as is the case for the other mesoporous materials.2 For longer synthesis times and for higher synthesis temperatures: (i) a decrease in the intensity of this broad line is observed, (ii) the pseudoboehmite broad lines at d spacings at about 6.5, 3.2, 2.35, and 1.87-1.85 appear concomitantly, and (iii) the TGA curves reveal the same organic content. There are however limits to both the synthesis time and the temperature. Table 2 shows that for syntheses in 2-butanol the high d spacing line is not present for solids recovered after 8 days at 423 K. XRD patterns of alumina gels obtained at temperatures above those used to form the mesophases show, in addition to the broad lines corresponding to pseudoboehmite, a low-intensity

1454

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

Table 1. Syntheses of Aluminas in an Aluminum Alkoxide/Alcohol/Carboxylic Acid System (Composition of All Synthesis Mixtures Falls within the Following Range: Al(OR)3:3.2 ( 0.3 H2O:0.3 ( 0.1 CnH2n+1CO2H:25 ( 5 CmH2m+1OH) carboxylic acid C12 C12 C6 C8 C12 C12 C18 C3 2,2-di-Me C4 C5 2-PrC5 C6 2-MeC6 C8 neo-C10a C12 C12 C12 C12 C12 C18 C12 C12 C12 C6 C12 C18 C18
a

alcohol C1 C1 C2 C2 C2 C2 C2 C3 C3 C3 C3 C3 C3 C3 C3 C3 C3 C3 C3 C3 C3 iC3 sec-C4 sec-C4 C6 C6 C6 C9

aluminum alkoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide isopropoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide isopropoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide

hydrolysis time (h) 0.6 0.6 1.3 1.3 0.6 0.8 15 18 10 18 10 1.0 10 1.2 0.05 1.0 1.2 1.0 0.1 1.0 1.2 5 8 8 3 1.1 3 2.6

aging time (h) 14 0.1 42 42 14 0.2 25 18 10 18 10 24 10 16 6 24 0.3 24 0.1 24 16 5 14 14 17 0.4 17 16

synthesis temp (K) 355 355 355 355 355 355 355 383 383 383 383 383 383 383 383 383 383 411 411 423 411 355 383 423 423 423 423 473

synthesis time (d) 6 6 3 3 5 4 11 6 6 6 6 2 6 4 2 2 5 2 2 2 4 2.5 2 7 3.5 4.5 3.5 4

alumina obtained mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase mesophase gel mesophase mesophase mesophase mesophase gel gel gel gel

XRD d () 22b 22b 28 28 32 31 39 21 23 20 23 24 22 26 26 29 31 30 30 30 36 36 33 21 25 28 34

Mixture of isomers. b Low-intensity, broad line; the solvent-to-aluminum ratio was 70. Table 2. Influence of the Synthesis Temperature and Time on the Position and the Relative Broadness of the XRD Line (Composition of the Synthesis Mixture Is Al(OCH(CH3)C2H5)3:3H2O:0.3C11H23CO2H: 28C2H5CHOHCH3) temp (K) 298 355 355 383 383 411 411 423 423
b

time (days) 0.7a 2 8 2 8 2 8 2 8

d spacing () b 33 33 33 33 34 36 37 b

(d spacing)/ d spacingc 0.48 0.47 0.45 0.48 0.49 0.57 0.64

a The synthesis time here corresponds to the 14-h aging time. Broad, low-intensity XRD line. c Relative broadness of the line estimated at 70% of the maximum intensity.

Figure 2. XRD pattern of as-made solid LA411.

peak with d spacings of about 8, 9, or 13 when the solvent is 1-propanol, 1-hexanol, or 1-nonanol, respectively. This suggests that solvent molecules are intercalated between the layers along the b axis of the orthorhombic unit cell of the pseudoboehmite. Figure 3a-c gives the TGAs of Catapal pseudoboehmite and of the solids LA298 and LA383, respectively. Three distinct weight losses appear on Figure 3b,c in the temperature ranges of 298-400 K (I), 400-475 K (II), and 475-790 K (III). Weight loss I corresponds to the removal of bulk water, weight loss II to the removal of physisorbed water, and weight loss III to both the decomposition of the surfactant and the dehydration of the inorganic walls (vide infra). The derivative curve reveals a discrete weight loss around 690 K that is mostly due to the loss of structural water as it is the case for the commercial pseudoboehmite sample. Weight loss II is the same (within 9 ( 2%) for any carboxylic

acid-alumina synthesized at room temperature (Figure 3b). The calcination of alumina mesophases for 2 h at temperatures as low as 703 K gives a total decomposition of the surfactant and is appropriate for the preparation of high-surface-area mesoporous aluminas. The field of alumina mesophase formation at 298 K is shown in Figure 4 for the system of aluminum secbutoxide/water/lauric acid/1-propanol as a function of water and lauric acid contents for two levels of concentration. The synthesis procedure involves lauric acid dissolution in 1-propanol and water before aluminum sec-butoxide is added. An elastic transparent gel precipitates for lauric acid-to-aluminum ratios higher than 2, whereas organic-alumina composites (OAC) that do not exhibit an XRD line at high d spacings are synthesized at low lauric acid contents. The total TGA weight loss below 793 K for the former materials exceeds 85% and for the latter falls below 40%. No alumina mesophase (AM) is obtained in the absence of water. For

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1455

Figure 4. Field of formation of carboxylic acid-aluminas at room temperature: (a) Al(OCH(CH3)C2H5)3:yH2O:xC11H23CO2H:15CH3CH2CH2OH; (b) Al(OCH(CH3)C2H5)3:yH2O:xC11H23CO2H:30CH3CH2CH2OH. OAC, organic-alumina composites; AM, alumina mesophase.

Figure 3. TGA curves for (a) Catapal alumina, (b) solid LA298, and (c) solid LA383.

AMs the position of the XRD line is shifted to lower d spacings when the ratio of carboxylic acid to aluminum increases. A concomitant increase in the total weight loss is also observed by TGA. These results suggest that the alumina walls between the organic aggregates decrease in size as the ratio of carboxylic acid to aluminum increases. No differences in the TGA curves are detected between solids synthesized with different water-to-aluminum contents when the carboxylic-toaluminum ratio is held constant. Furthermore, a shift of the XRD line to lower d spacings is evident for solids synthesized with lower water-to-aluminum ratios. Two possible reasons for this behavior are as follows: First, it is possible that the packing of the carboxylic acids in higher water contents results in the formation of aggregates with larger cross-sectional areas, thus shifting the XRD line to higher d spacings. Second, it is reported that rodlike micelles are longer in water than in organic solvents.8,13 Therefore, the aggregates formed in alcohols would be incrementally lengthened by addition of controlled amounts of water, and thus, their number
(13) Perche, T.; Auvray, X.; Petipas, C.; Anthore, R.; Rico, I.; Lattes, A.; Bellissent, M. C. J. Phys. I France 1992, 2, 923-942.

Figure 5. Influence of the alkyl chain length on the d spacing (open figures) and the relative peak broadness of the high d spacing XRD line (solid figures). Circles denote normal and squares denote branched (4 ) 2,2-dimethylbutyric; 5 ) 2propylvaleric; and 6 ) 2-methylhexanoic)carboxylic acids.

would decrease. Since the yield of alumina and organic remains constant (vide infra), the random packing of aggregates would presumably be less close, shifting the high-intensity line to higher d spacings. At any rate, it appears that the self-assembly of the organic molecules in higher water contents results either in aggregates with larger cross sections or aggregates with larger lengths or a combination of both. As expected and already shown for MCM-41, for the alumina mesophases formed here, the position of the XRD line is shifted to higher d spacings with increasing surfactant alkyl chain lengths (Figure 5). Note that the XRD line broadening for solids prepared using short alkyl chain carboxylic acids indicates declining uni-

1456

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

Table 3. Composition of Solids Prepared in 2 days at 373 K from Mixtures of Composition: Al(OCH(CH3)C2H5)3: 3.2H2O:0.30CnH2n+1CO2H:26C2H5CH2OH solid CA383 LA383 SA383 XRD d () 24 29 36 elemental analysis data Al yielda CA%b Al/CAc 0.74 0.80 0.86 30.4 43.5 52.6 3.6 3.4 3.4 I% 10 8 5 TGA datad II% III% W% 4 2 1 35 48 59 8 10 15

a Ratio of the number of aluminum atoms in the solid recovered to the number of aluminum atoms introduced in the synthesis mixture. b Carboxylic acid content based on wt % C. c Molar ratio. d The final temperature of the TGA is 787 K. I% corresponds to bulk water, II% to physisorbed water, III% to both the carboxylic acid decomposition and the removal of structural water from the inorganic walls. W% is the structural water defined by W% ) 100(III% - CA%)/(100 - I% - II% - CA%). W is 17% for the commercial pseudoboehmite sample.

formity, if one assumes that particle size effects are not significant (see adsorption data given in Figures 15a-g at high P/Po for justification of this assumption). When aluminas are synthesized with acetic or formic acid, the XRD patterns show no intense peaks at high spacings. In Table 3 are reported elemental analyses and TGA data for solids synthesized at 373 K in 1-propanol from mixtures featuring the same composition except for the carboxylic acid (solids CA383, LA383, and SA383). The yield of aluminum is higher when the surfactant alkyl chain is longer. Elemental analyses show that all three materials exhibit the same aluminum-to-surfactant ratio as in the synthesis mixture. The difference between the TGA weight loss III and the carboxylic acid content calculated from elemental analysis corresponds to water liberated from dehydration of the walls. Alumina prepared with stearic acid contains almost as much structural water as the commercial pseudoboehmite sample. The structural water amount is shown to decrease when the alkyl chain decreases in size. Also, the shorter the surfactant alkyl chain, the higher the amount of physisorbed water. Investigation of the use of other solvents, surfactants, and aluminum sources was motivated by the search for other alumina mesophases. Alumina mesophases were also synthesized using a wide variety of combinations involving an anionic surfactant, a polar or apolar solvent and an organoaluminum or aluminum salt (Table 4), confirming that nonaqueous solvents are very promising for the synthesis of mesoporous aluminas. At room temperature using lauric acid/aluminum sec-butoxide combinations, solvents such as diethylether or even pentane allow for the formation of alumina mesophases that resemble those obtained in alcohols when investigated by XRD, TGA, and 27Al MAS NMR. The high d spacing XRD line of aluminas synthesized in pentane appears only after 3.5 days, whereas the definitive TGA curve is already obtained after 20 h. These results suggest that the alumina mesophase is obtained from an ill-defined carboxylic acid-alumina via solid-phase transformations. The driving force for the synthesis appears to be the high affinity of the carboxylate ions to complex very small clusters of aluminum oxohydroxides and to organize them into composite materials in which a micellar aggregate of circular cross section is encapsulated by hydrated alumina walls (vide infra). No evidence of layered structures for these composites is found by XRD, SEM, or TEM. The synthesis of mesoporous materials has already been claimed using isopoly- and heteropoly-cations as

a metal source.14,15 Here, we were also able to synthesize alumina mesoporous materials using the aluminum Keggin ion. Syntheses in water systematically yield lamellar phases whereas syntheses in formamide using sodium dodecyl benzenesulfonate as the anionic surfactant allow for the formation of alumina mesophases (Table 4). Interestingly, when the mesophases are prepared in short synthesis times, the Keggin structure remains as indicated by the presence of the two typical broad XRD lines at ca. 13 and 5 and as also indicated by the observation that the 27Al MAS NMR spectra of the aluminum source and the mesophase are virtually identical. When the mesophases are treated at 383 K for 10 days, the 27Al MAS NMR spectrum and the XRD pattern reveal that the Keggin ion structure no longer exists. The coordination modes of carboxylates with metals have been extensively studied since metal carboxylate surfaces are of industrial interest for their lubrication and anticorrosion properties.16 Three possible modes of coordination for carboxylates with aluminum ions are as follows:

Although the monodentate and the bridging (bidentate II) modes of coordination have been well documented,17,18 there appears to be no crystallographic evidence for the existence of the chelating mode of coordination of carboxylate to aluminum. In particular, work on 1,3diphenyltriazenide complexes of aluminum in which the triazenide are exclusively chelating has led previous workers to conclude that the chelating coordination of carboxylate with aluminum is not available due to the increased ring strain in such a system.19 Typical midIR spectra of alumina mesophases are given in Figure 6a for solids LA298, LA383, and LA411 and in Figure 6b for CA383, LA383, and SA383. All IR spectra show strong bands at 1550-1580 and 1460-1470 cm-1, indicating a bridging mode of coordination consistent with results reported previously for a number of carboxylatoalumoxanes.17 Solids prepared with lower molecular weight carboxylic acids and those synthesized at lower temperatures show also a shoulder around 1590 cm-1 indicative of the monodentate coordination of carboxylate with aluminum.18 The total decomposition of the organics in solids prepared with higher molecular weight carboxylic acids and solids synthesized at higher temperatures can be performed only at 723 K (rather than 703 K) and for 6 h (rather than 2 h). Thus, it appears that the higher the number of bridging carboxylates relative to unidentate ones, the stronger the interaction between the carboxylate and aluminum.
(14) Ciesla, U.; Demuth, D. G.; Leon, R.; Petroff, P. M.; Stucky, G. D.; Unger, K.; Schu th, F. J. Chem. Soc., Chem. Commun. 1994, 11, 1387. (15) Mu ller, U.; Bayer, B.; Oetter, G.; Gehrer, E.; Ciesla, U.; Schu th, F. J.; Monnier, A.; Unger, K.; Stucky, G. D. Eur. Patent 95102637.6. (16) Mehrotra, R. C.; Rai, A. K. Polyhedron 1991, 10, 1967-1994 and references therein. (17) Landry, C. C.; Rappe , N.; Mason, M. R.; Apblett, A. W.; Tyler, A. N.; MacInnes, A. N.; Barron, A. R. J. Mater. Chem. 1995, 5, 331. (18) Gurian, P. L.; Cheatham, L. K.; Ziller, J. W.; Barron, A. R. J. Chem. Soc. Dalton Trans. 1991, 1449 and references therein. (19) Leman, J. T.; Braddock-Wilking, J.; Coolong, A. J.; Barron, A. R. Inorg. Chem. 1993, 32, 4324.

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1457

Table 4. Syntheses of Alumina Mesophases Using Compositions Other Than Aluminum Alkoxide/Carboxylic Acid/Alcohol surfactant, s lauric acid lauric acid lauric acid lauric acid lauric acid DBSa DBS lauryl phosphate lauryl phosphate
c

solvent, S 2-propanol chloroform ethyl acetate diethyl ether pentane pentane formamide 1-propanol 1-propanol

aluminum source, A acetylacetonate sec-butoxide sec-butoxide sec-butoxide sec-butoxide sec-butoxide chlorhydrolb nitratec sec-butoxide

composition of the synthesis mixtured A:1.00s:3w:30S A:0.24s:3w:6S A:0.30s:3w:17S A:0.31s:3w:20S A:0.32s:3w:15S A:0.17s:3w:18S A:0.14s:14S A:0.28s:27S A:0.20s:3w:32S

synth temp (K) 355 343 355 298 298 298 383 383 355

synth time (days) 8f 3.5 1.5f 2.5f 3.5f 3f 1 4.5 4

XRD d () 30 e 41 39 43 38 35 29 32

a Sodium dodecyl benzenesulfonate. b Chlorhydrol is the aluminum Keggin ion in its chloride form: Al O (OH) (H O) Cl . 13 4 24 2 12 7 Nonahydrate. d A denotes 1 equiv of aluminum, and w stands for water. e A gel is recovered. f A shorter synthesis time does not allow for the formation of alumina mesophases.

When aluminum alkoxides are hydrolyzed in an organic solvent with controlled amounts of water, two reactions can occur:20

Al(OR)3 + 3H2O f Al(OH)3 + 3ROH Al(OR)3 + 2H2O f AlOOH + 3ROH (1)

The initially amorphous precipitate is converted into pseudoboehmite after a few hours of aging. Pseudoboehmite is the end product for a water-to-aluminum ratio of 3.0 ( 0.5. If the temperature does not exceed 350 K, the aluminum trihydroxide bayerite is the final phase to form for water-to-aluminum molar ratios of 20 or more.20 However, the presence of anions greatly influences these processes. Carbonate anions are reported to have the strongest effect with regard to retarding the formation of these phases, as they tend to coordinate with aluminum atoms. Carboxylates also exhibit a similar effect since no bayerite is obtained in 1 day at 298 K from water-rich media (see Figure 4). When carboxylic acid aggregates interact with a freshly precipitated aluminum hydroxide, the following reaction occurs:
Figure 6. Mid-IR spectra of (a) as-made solids CA383, LA383, and SA383 and (b) as-made solids LA298, LA383, and LA411.

>Al-OH + C6H13COOH f >Al-OOCC6H13 + H2O (2)


We believe that the ability for carboxylate ligands to coordinate with one or two aluminum sites on very small clusters of AlOOH allows for the rapid formation of alumina mesophases. Room-temperature syntheses in 1-propanol can yield solids in 15 min whose XRD pattern already exhibits the line at high d spacings. If it is assumed that the d spacing at which the single XRD line appears in the as-synthesized mesophases results from a scattering repeat length that is the sum of the organic aggregate diameters (assuming aggregate shapes of circular cross section) and the average size of the inorganic walls, then the value of the d spacing depends on the conformation and packing of the alkyl chains in the carboxylic acid. Table 5 gives estimates of the aggregate diameters for solids CA383, LA383, and SA383. Estimates for the aggregate diameters were calculated assuming either straight alkyl chains or alkyl chains curved in such a way that methyl groups are in close proximity of polar heads (see Figure 8). Comparison of both sets of values with the observed d spacing suggests that straight chains are not encountered in the
(20) Wefers, K.; Misra, C. Oxides and Hydroxides of Aluminum; Alcoa Laboratories, 1987.

Figure 7. 3C CPMAS NMR spectra of solid LA383. The asterisk denotes a spinning sideband.

The 13C CPMAS NMR spectra of LA383 reveal that the organic molecules remain intact in the mesophase with a low-intensity, broad resonance at 180 ppm due to the carboxylic carbon (Figure 7). The broadness of this resonance is likely the result of chemical shift dispersion as well as possible low mobility of the carboxylic group.

1458

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

Table 5. Characterization of the Porosity of Alumina Mesophases Prepared in 1-Propanol (Solids CA383, LA383, and SA383) d spacing () solid CA383 LA383 SA383 as-made 24 29 36 calcined 53 47 43 organic circular cross sectiona () straight 21 31 41 curved 15 19 25 PV1b (cm3/g) 0.338 0.483 0.584 PV2c (cm3/g) 0.312 0.411 0.410 BET SSAd (m2/g) 530 710 700 PDe () 21 19 21 fwhmc () 14 14 14

a The organic circular cross section was estimated assuming straight alkyl chains or alkyl chains curved in such a way that methyl groups reside close to the polar heads. b Pore volume occupied by carboxylic acid assuming a liquid density of 0.90 g/cm3. c Pore volume in calcined materials based on nitrogen adsorption. d BET specific surface area. e Pore diameter (BJH). f Full width at half-maximum on pore size distribution.

Figure 8. Hypothetical straight- and bent-chain conformation of carboxylic acid aggregates coordinated to the alumina phase.

Figure 9. 27Al MAS NMR spectra of as-made alumina mesophases synthesized in different solvents using lauric acid at various temperatures. Asterisks denote spinning sidebands.

solids synthesized at low water-to-1-propanol ratios (note that in solid SA383, the observed d spacing is lower than the estimated aggregate diameter assuming a straight-chain model). In contrast, the curved chain model would be consistent with a wall average width of about 10 for all solids regardless of the length of the alkyl chain (wall width value similar to that of MCM-412,29). Figure 9 shows the 27Al MAS NMR spectra of solids LA411, LA383, LA298, LP298, and LE298. Using the spincounting method described in the Experimental Section, it is established that 100% of the aluminum atoms are NMR-visible for LA411 and LA383, whereas only 80% are in the case of LA298, LP298, and LE298. For any solid

synthesized at room temperature or made in nonpolar solvents, about 15-25% of the total aluminum population remains invisible by NMR spectroscopy. A resonance at ca. 5 ppm (due to 6-fold coordinated aluminum) is the main component in the NMR spectra of all of the solids investigated. In addition, a resonance at ca. 65 ppm, attributable to tetrahedrally coordinated aluminum, comprises a small fraction of the total signal intensity in all of the NMR spectra recorded. A broadening of the resonance at 5 ppm as well as the appearance of a shoulder at ca. 35 ppm are evidenced for materials LA298, LE298, and LP298. The resonance at 35 ppm is attributable to 5-fold coordinated aluminum and has been also observed in a number of aluminumcontaining compounds.21,22 It is speculated that the octahedral sites are aluminum atoms comprising the bulk of the inorganic walls whereas the 4-5-coordinated aluminums are likely due to surface sites. As shown in Figure 9, solids LA411 and LA383 have the narrowest and most symmetric line shape for the main resonance. Furthermore, no 5-fold coordination of aluminum is discernible by NMR for these solid (note that 100% of the aluminum atoms in these samples are detected as octahedral and tetrahedral sites). On the other hand, solids LA298, LE298, and LP298 exhibit a clear shoulder at 35 ppm as well as broader and more asymmetric line shapes, indicating higher numbers of 5-coordinate aluminum atoms and greater chemical shift dispersions consistent with the presence of NMR-invisible aluminum sites. It appears that syntheses performed with nonpolar solvents and those carried out at lower temperatures result in a higher degree of local disorder around the aluminum sites possibly related to the evolution of 5-coordinate aluminum atoms. Figure 10 shows a series of 1H-27Al CP/MAS NMR spectra with contact times of 25 s to 10 ms for as-made solid LA383. A single resonance at 5 ppm is observed signifying that only 6-fold coordinated aluminum atoms are cross-polarized. The maximum signal intensity is obtained at a contact time of 200 s. The signal reaches 20% of its maximum intensity within a contact time of 25 s. Such efficient polarization transfer at such short contact times cannot be attributed to the protons on the lauric acid since the methylene protons are expected to be too far away from the aluminum to account for the fast initial growth rates of the transferred magnetization. Nor can the transfer of polarization be accounted for by the protons of physisorbed water in the as-made material since dehydration at 473 K under vacuum does not appreciably alter the CP intensity for comparable
(21) Alemany, L. B.; Kirker, G. W. J. Am. Chem. Soc. 1986, 108, 6158. (22) Cruickshank, M. C.; Dent Glasser, L. S.; Barri, S. A. I.; Poplett. J. F. J. Chem. Soc., Chem. Commun. 1986, 23.

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1459

Figure 12. 1H NMR spectrum of as-made and dehydrated solid LA383. Inset shows the simulation of the NMR spectrum of the as-made sample [signals at 0.8 ppm (8.6%), 1.2 ppm (53.3%), 2.1 ppm (25%), and 5 ppm (13.3%)].

Figure 10. Variable contact-time 1H-27Al CPMAS NMR spectra of LA383.

Figure 11. 1H-27Al CPMAS NMR intensity vs contact time for as-made solids LA383, LE298, and LP298.

experimental parameters. Thus, the cross-polarization behavior for short contact times (25-500 s) is most likely from Al-OH groups. Figure 11 shows results of variable contact-time CP/ MAS NMR experiments for as-made solids LA383, LP298, and LE298. Only the octahedral aluminum resonance is observed for these solids. Note that all three samples give the same maximum obtainable CP intensity to within 1-2% even though solids LP298 and LE298 contain as much as 20% NMR-invisible aluminum sites. Further, it is evident that the CP behavior for these samples is quite similar indicating comparable local structures for the Al-OH. Given that the total structural water determined from TGA for LP298 and LE298 is the same as that for LA383, it appears that all Al-OH sites are observed by NMR for these solids. Thus, the NMRinvisible sites in LP298 and LE298 are likely due to nonhydroxylated aluminum atoms. Considering the information presented above, it is not unreasonable to presume that a substantial portion of

the NMR-invisible sites in solids LA298, LE298, and LP298 are asymmetric 5-coordinate aluminum atoms. Furthermore, these materials exhibit a greater IR absorption around 1590 cm-1, indicating a higher number of unidentate carboxylates. Thus, it appears that the evolution of 5-coordinate aluminum atoms and NMRinvisible sites is related to the presence of higher amounts of monodentate carboxylate groups on aluminum. Figure 12 compares the 1H NMR spectra of solid LA383 in the as-made and dehydrated (under vacuum at 473 K) forms. Spectral simulation of the 1H NMR spectrum of the as-made LA383 is also provided. The two sharp resonances at 1.2 and 0.8 ppm (relative ratio 6.2:1) are assigned to the CH2 and CH3 protons of lauric acid, respectively. The broad peak at around 5 ppm for the as-made material disappears upon dehydration at 473 K in vacuum and is assigned to physisorbed water. A second broad peak at ca. 2.0 ppm is also observed in the 1H NMR spectra of both as-made and dehydrated materials. The intensity of this peak relative to the sum of the peak intensities from the organic protons remains the same upon dehydration but decreases upon contact of the as-made sample with D2O. Thus, the peak at 2.0 ppm is assigned to hydroxyl protons. Furthermore, the absence of any peaks in the region > 8 ppm indicates that there are no acidic protons associated with the carboxylate groups. It is not possible to remove the organic from the alumina mesophase by ion exchange or extraction of the surfactant without destruction of the material. Calcination, on the other hand, results in the development of mesoporosity. As mentioned previously, the organics are decomposed completely upon calcination in air for 2 h at 703 K. The temperature ramp of 0.5 K/min must be carried out under nitrogen in the case of aluminas prepared with octanoic acid or lower molecular weight carboxylic acids in order to avoid the destruction of the alumina mesoporous structure and to retain the XRD line at high d spacings. It is also interesting to note that the full decomposition of the organics is obtained at lower calcination temperatures when the alumina mesophases are synthesized at lower temperatures. For example lauric acid is fully decomposed in 2 h at 673 K

1460

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

Figure 14. 27Al MAS NMR spectra of: (a) as-made, dehydrated and calcined solid LA298 and (b) as-made, dehydrated, and calcined solid LA383. Figure 13. Variable-temperature XRD patterns of solid LA383 collected under air/water (a) and air (b) atmospheres.

for solid LA298, at 703 K for solid LA383, and at 723 K for solid LA411. Figure 13 shows a series of XRD patterns obtained from aluminas prepared with lauric acid (solid LA383) and calcined under air and air/water atmospheres at different temperatures. It is evident that the solid is stable up to 1073 K, even under the air/water atmosphere. An increase in the intensity of the XRD line is observed upon removal of the organic (673-773 K) under both air and air/water atmospheres. This is expected since the difference in the electron density between the inorganic walls and the pores become progressively larger as the organic is decomposed. A shift of the XRD line to higher d spacings with increasing temperatures is observed for the solid under the air/ water atmosphere. Water readsorption at lower temperatures on the calcined alumina further shifts the XRD line to higher d spacings and the XRD line disappears completely after 20-30 min of rehydration at 50 C as shown in Figure 13a (top XRD pattern). Recalcination does not recover the XRD line. In the absence of water, the position of the XRD line does not change significantly with increasing temperature (Figure 13b). This indicates that the shift in the position of the XRD line is related to the presence of water in the system. Water chemisorption always occurs to a large extent in transition aluminas.7 It is known that the reaction of water with transition aluminas at elevated temperatures results in a decrease in the surface area due to sintering of the pores.23 Silicon grafting of the aluminas increases their stability toward
(23) Harris, D. J.; Young, D. J.; Trimm, D. L. Proceedings, 10th Australian Chemical Engineering Conference, Sydney, 1982.

sintering upon exposure to high-temperature steam since minimal hydroxylation of the surface takes place on silicon-modified alumina.24 Therefore, the observed shift of the XRD line to higher d spacings in Figure 13a is likely due to the coalescing of the pores upon hydroxylation of the surface. This local sintering due to the increase in the number of Al-OH groups effectively results in larger pores and concomitantly smaller surface areas. This is consistent with results of nitrogen isotherms recorded on rehydrated-recalcined solids (not shown). These results indicate that the pores become larger in size, while specific surface areas decrease with increasing number of rehydration-calcination cycles. Figures 14a,b gives the 27Al MAS NMR spectra for solids LA298 and LA383, as-made, dehydrated (at 473 K under vacuum), and calcined at 703 K. For dehydrated solid LA383 a slight increase in the line width of the main resonance as well as an enhancement of the resonance at 65 ppm are observed. Furthermore, a small but discernible peak at ca. 30 ppm appears for solid LA383. Note that the appearance of resonances at 30 and 65 ppm upon dehydration is more pronounced for solid LA298. Thus, dehydration results in an increase in the amount of 4- and 5-fold coordinated aluminum at the expense of 6-coordinated aluminum. Calcination enhances this effect further. Three broad resonances around 5, 32, and 65 ppm are observed in the calcined mesoporous aluminas, and it is apparent that the intensities for the peaks at 32 and 65 ppm have increased relative to the dehydrated samples. About 30% of the total aluminum population is NMR-invisible for the calcined solids. It is known for aluminas that increasing surface area is concomitant with a decrease
(24) Beguin, B.; Garbowski, E.; Primet, M. J. Catal. 1991, 127, 595.

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1461


Table 6. Low-Pressure Hydrocarbon Adsorption Data n-heptane adsorption neopentane adsorption solid CA383 LA383 SA383 Catapal alumina P/P 0.07 0.03 0.07 0.05 vol (cm3/g) 0.138 0.089 0.167 0.090 P/P 0.05 0.04 0.09 0.05 vol (cm3/g) 0.142 0.104 0.154 0.039

in the intensity of the 27Al NMR signal.25 This effect has been attributed to a distorted environment for the surface aluminum sites as well as to the mobility of protons that terminate the surface which produces an efficient quadrupolar relaxation pathway resulting in broadened line shapes and loss in signal intensity.26 The greater the surface area, the greater the number of distorted and hydroxylated aluminum atoms and the greater the loss in signal intensity. Thus, the loss in signal intensity upon calcination is consistent with the large internal surface area of the calcined samples. The 1H NMR spectra of the calcined materials (not shown) reveal two broad resonances at 2.0 and 7.0 ppm. The resonance at 2.0 ppm is assigned to nonacidic AlOH groups. The signal at 7.0 ppm suggests that strongly bound water or clusters of hydrogen-bonded hydroxyls or a combination of both are present. The broadness of these signals can be attributed to the quadrupolar effects of 27Al on the MAS averaging of 1H27Al dipolar interactions as well as chemical shift dispersion due to the presence of a variety of proton environments. No 1H-27Al cross polarization is observed from any of the calcined samples. On the basis of thermogravimetric and elemental analysis data, it is established that the ratio of H to Al is about 1/3 in calcined solids; thus, the fact that no cross polarization is observed is not attributable to lack of an available 1H source. The following are known: (i) about 1/3 of the aluminum in calcined LA383 is invisible by 27Al NMR; (ii) the ratio of aluminum to proton in this sample is about 3; (iii) a substantial portion of the aluminum in the sample is at surface sites; (iv) these surface sites in the calcined samples are hydroxylated; (v) these surface aluminum atoms are NMR-invisible due to either the increased disorder in their local structure, or the dynamic processes involving the surface protons, or both. On the basis of the aforementioned observations, it is reasonable to assume that a substantial portion of aluminum sites that do have protons available for cross polarization are surface sites which are in turn expected to be NMR-invisible. This accounts for the lack of any observed 1H-27Al polarization transfer. Furthermore, adsorption of organic molecules such as cyclohexane and triisopropylbenzene in the calcined materials does not result in any observable 1H-27Al polarization transfer. Therefore, the addition of such organic adsorbates has no appreciable effect on the local structure of the surface aluminum sites and does not reduce large geometrical distortions expected for these atoms. In fact, no difference was observed in the 27Al MAS NMR spectra for calcined, cyclohexane-loaded, and triisopropylbenzeneloaded samples. The nitrogen adsorption-desorption isotherms for alumina mesophases CA383, LA383, SA383, and Catapal alumina are given in Figure 15a-d (see Table 3 for the characterization of as-made solids CA383, LA383, SA383). The corresponding pore size distributions are given in Figure 16a-d. Solid CA383 was calcined at 693 K for 2 h under air after a ramp to 693 K under nitrogen. Solid LA383, Catapal alumina, and solid SA383 were calcined at 703 K under air, for 1.5, 2, and 6 h, respectively. The shape of these isotherms (except Catapal alumina) is
(25) O Reilly, D. E. Adv. Catal. 1960, 12, 31. (26) Huggins, B. A.; Ellis, P. D. J. Am. Chem. Soc. 1992, 114, 2098.

identical with the shape of the MCM-41 or MCM-48 isotherms.12,25 The large uptake at low partial pressures is due to the monolayer coverage of the mesopores and the increase in the adsorbed volume up to pressures of about 0.4 is accounted for by the filling of the mesopores.27 No further adsorption is then observed up to a reduced pressure of 0.981 (at this pressure, pores 500 in radius would be filled). No hysteresis loop is observed in the isotherms of the alumina mesophases LA383 calcined at 703 K. The absence of micropores in our materials is confirmed by the data on hydrocarbon uptakes at low partial pressures (presented in Table 6). Similar n-heptane and neopentane uptakes are evidenced for pure alumina mesoporous materials, whereas the neopentane uptake is much lower than the nheptane uptake on Catapal alumina. Thus, pores in the zeolitic region are not present in the aluminas prepared with carboxylic acids. The absence of microporosity is an important feature if one considers that the contribution of micropores to the total surface area in conventional aluminas is generally significant at calcination temperatures as low as 703 K. The pore size distributions also show that no pores larger than 35-40 in diameter are present in the solids. The pore size distributions from the aluminas prepared here are narrower than those of conventional aluminas and in particular narrower than the pore size distribution observed for Catapal alumina (Figure 16d). The BET specific surface areas are high, namely, 700 m2/g in the case of aluminas prepared with lauric or stearic acid. The specific surface area is lower for aluminas synthesized with lower molecular weight carboxylic acids. In contrast to silica-based mesoporous materials, average pore sizes cannot be tailored directly by synthesis via the choice of the surfactant and are equal to 20 , independent of the carboxylic acid used (Table 5). A comparison between the volume occupied by the carboxylic and the pore volume determined by nitrogen adsorption on the calcined solids shows that the N2 pore volumes are always less than the volume occupied in the composite material by the carboxylic acid. Since the N2 pore volume is affected by the calcination temperature, one must be cautious in interpreting these results. The isotherms and pore size distribution of solid LA298 are given in Figures 15e and 16e, while those for solid LA411 in Figures 15f and 16f. A hysteresis loop is present in the isotherm of LA298, and for this solid, the pore size distribution is shifted slightly to lower pore sizes; a contribution by micropores to the specific surface area cannot be rejected. Solid LA411 must be calcined at higher temperature (723 K) in order to completely remove the organics. Lower specific surface areas as well as higher pore sizes are present in this solid and are consistent with the usual correlation observed for aluminas between calcination temperature and porosity
(27) Chen, C.-Y.; Li, H.-X.; Davis, M. E. Micropor. Mater.1993, 2, 17-26.

1462

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

Figure 15. N2 adsorption isotherms of solids: (a) CA383, (b) LA383, (c) SA383, (d) Catapal alumina, (e) LA298, (f) LA411, and (g) LP298.

features.7 Note however that the pore size distribution is very narrow for this solid. In a study of the influence of preparation methods on the pore structure of aluminas, Kotanigawa et al.28 showed that aluminas prepared by neutralization of sodium aluminate with CO2 and calcination at 823 K have pores of similar size (note however that theses materials also contain much larger
(28) Kotanigawa, T.; Yamamoto, M.; Utiyama, M.; Hattori, H.; Tanabe, K. Appl. Catal. 1981, 1, 185-200.

pores in addition to the 20 pores). Similarities in carbonate and carboxylate interactions with aluminum ions are probably responsible for similarities in pore structure of the 20 pores. Figures 15g and 16g give the isotherms and the pore size distribution for the solid LP298 prepared in pentane. The solid was calcined in the same way as LA383 and exhibits comparable specific surface areas. However, the pore size distribution is much broader and confirms the previous suggestion that carboxylic acid-aluminas

Synthesis of Pure Alumina Mesoporous Materials

Chem. Mater., Vol. 8, No. 7, 1996 1463

Figure 16. Pore size distributions of solids: (a) CA383, (b) LA383, (c) SA383, (d) Catapal alumina, (e) LA298, (f) LA411, and (g) LP298.

prepared in apolar solvents contain organic aggregates that are very heterogeneous in size. The processes that occur upon heating the alumina mesophases are not likely to parallel those of the silicabased mesoporous materials. As opposed to silica-based materials,29 alumina mesophases calcined in humid air undergo an increase in the size of the repeating units upon heating (shift of the XRD line to higher d spacings). Whereas calcination of silica-based mesoporous materials leaves the coordination of the silicon atoms

unaffected (silicon remains 4-coordinated), calcination of alumina mesophases involves a lowering of the local order and symmetry of the aluminum as exhibited by the coexistence of 4-, 5-, and 6-fold coordinated aluminum atoms, broad 27Al NMR line shapes, and substantial amounts of NMR-invisible aluminum sites. This lowering of the short-range order upon heating is related
(29) Chen, C.-Y.; Burkett, S. L.; Li, H.-X.; Davis, M. E. Micropor. Mater. 1993, 2, 27-34.

1464

Chem. Mater., Vol. 8, No. 7, 1996

Vaudry et al.

to the high structural water-to-alumina molar ratios that are in the range 0.5-1.0. The thermal history of dehydration strongly governs the properties of pure alumina mesoporous materials and prevents the tailoring of their pore size directly by the choice of the surfactant size. Summary The driving force for the synthesis of carboxylic acidalumina mesophases is a strong interaction between carboxylic acids and small clusters of aluminum hydroxide. The state of aluminum in the as-made and calcined solids depends strongly on the synthesis temperature and the type of solvent used. The loss of a substantial amount of structural water upon calcination reduces the local order of the resulting alumina, al-

though the solids can be thermally stable to 1073 K. The extent of dehydration of the inorganic walls strongly influences the porosity properties of the alumina. Calcination of the mesophases at temperatures as low as 703 K yields pure alumina mesoporous materials with BET specific surface areas as high as 710 m2/g and narrow pore size distributions (centered at 20 ) that do not contain zeolitic micropores. Acknowledgment. The authors thank Mr. Paul Wagner for carrying out variable-temperature XRD measurements and Dr. Larry Beck for helpful discussions concerning the interpretation of the NMR spectra. Akzo America is gratefully acknowledged for financial support.
CM9600337

Das könnte Ihnen auch gefallen