- Email: [email protected]
Synthesis and characterization of a highly active alumina catalyst for methanol dehydration to dimethyl ether
Applied Catalysis A: General 348 (2008) 113–120
Contents lists available at ScienceDirect
Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata
Synthesis and characterization of a highly active alumina catalyst for methanol dehydration to dimethyl ether Seung-Moon Kim, Yun-Jo Lee, Jong Wook Bae, H.S. Potdar, Ki-Won Jun * Alternative Chemicals/Fuel Research Center, Korea Research Institute of Chemical Technology (KRICT), P.O. Box 107, Yuseong, Daejeon 305-600, Republic of Korea
A R T I C L E I N F O
A B S T R A C T
Article history: Received 19 March 2008 Received in revised form 23 June 2008 Accepted 24 June 2008 Available online 1 July 2008
A simple sol–gel method was adopted to synthesize boehmites with high surface area using aluminum iso-propoxide (AIP), acetic acid (AA) and 2-propanol, and the effects of surface area and methanol dehydration on activity were investigated. The hydrolysis conditions of AIP in the presence of AA in 2propanol solvent were systematically varied to observe their effect on phase formation, crystallinity, surface area and pore size distribution of the alumina. The surface area and the number of acidic sites varied considerably with the variation in the molar ratio of AA/AIP. This study revealed that a high surface area boehmite (in the range of 628–717 m2/g) could be obtained by keeping the molar ratio of AA/AIP as 0.5 and that of H2O/AIP at 3. Rod shaped, porous g-Al2O3 powder with a high surface area of 438 m2/g was obtained after calcination of the boehmite at 500 8C for 5 h in air. The temperature programmed desorption of ammonia (NH3-TPD) of the g-Al2O3 samples demonstrated higher concentration of acidic sites when acetic acid was used during preparation than when it was not used. The vapor phase dehydration of methanol (containing 20 mol% H2O) to dimethyl ether (DME) was conducted on the prepared aluminas. With increasing surface area of g-Al2O3, the temperature required to reach 50% conversion of methanol decreased due to the increased number of acidic sites which are favorable for methanol dehydration with low byproduct formation. The catalytic activity for methanol dehydration to DME correlated well with the total number of acidic sites of g-Al2O3, which was controlled by changing the AA/AIP and H2O/AIP molar ratios. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Methanol dehydration Dimethyl ether Sol–gel g-Alumina Acidity
1. Introduction Alumina is a low cost material most widely used as a catalyst and catalyst support [1]. In addition, it is also used as the starting material for the preparation of Al2O3 based ceramics [2]. A wide variety of these applications are possible because of the fact that alumina exists in the corundum and transition alumina forms [3]. The corundum or a-alumina has excellent mechanical, electrical, thermal and optical properties due to hexagonal close packing of oxygen ions. On the other hand, transition aluminas, including gAl2O3, have a cubic close packing of oxygen ions resulting in high surface area, mesoporosity and surface acidity [4]. As a result of these important properties, g-Al2O3 is also extensively used as an adsorbent and a membrane [3]. Solid acid catalysts e.g. g-Al2O3, modified g-Al2O3 with silica, phosphorus or B2O3 based are widely used, excellent catalysts for the dehydration of methanol to DME [5]. However, systematic study on the effect of various preparation
* Corresponding author. Tel.: +82 42 860 7671; fax: +82 42 860 7388. E-mail address: [email protected] (K.-W. Jun). 0926-860X/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2008.06.032
parameters on physical and chemical characteristics of g-Al2O3 affecting DME synthesis is lacking. In the our previous investigations, the efforts to achieve the preparation of g-Al2O3 powders by using various chemical routes through boehmite precursor have been continued to synthesize a thermally stable g-Al2O3 at fixed preparation conditions [6] and to evaluate the catalytic activity of home-made g-Al2O3 prepared from aluminum nitrate precursor via coprecipitation/digestion routes [7]. An attempt is made in the present investigation to synthesize g-Al2O3 catalysts with different surface areas by varying preparation parameters systematically during sol–gel synthesis and to correlate the catalytic activity with the surface area of g-Al2O3. In view of the increasing demand for DME as a raw material for the production of dimethyl sulfate, methyl acetate, light olefins and alternative clean fuels [8–10], the importance of a commercially viable catalyst is further enhanced. In order to get a reliable and reproducible alumina for these applications, a stringent control of composition, surface area, porosity (i.e. pore size and its distribution) and surface acidity are essential. Although various chemical routes [4,11,12] have been tried, the sol–gel route offers an excellent opportunity for controlling the physical, chemical and
114
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
textural properties of the aluminum oxide. Sol–gel derived alumina powders are generally prepared through acid- or basecatalyzed hydrolysis and condensation reactions of aluminum alkoxide precursors such as aluminum iso-propoxide (AIP) or aluminum sec-butoxide [1,2] in organic solvents. In the case of the Yoldas process [13], it is reported that the aging treatment over a period of 24 h at 80 8C yields boehmite [AlO(OH)] whereas aging at room temperature yields bayerite, Al(OH)3. The thermal decomposition of boehmite at 400–500 8C in air produces g-Al2O3. The catalytic activity of the alumina for methanol dehydration is generally dependent on the surface acidity, which could be varied by adding some promoters or controlling the acidic properties of alumina or zeolites [14,15]. Hence, it is important for the synthesis of g-Al2O3 with controllable and reproducible properties to get the stable catalytic activity. It is also necessary to study the effect of various preparation parameters on the physico-chemical properties of the solid–acid catalyst, with particular attention given to the changes in the acidity. The present investigation focuses on elucidating the effects of acetic acid (AA)/AIP and H2O/AIP mole ratios on the properties of boehmite and also on the g-Al2O3 obtained by subsequent heat treatment. Detailed investigations have been carried out to correlate the changes in acidity with those of the textural properties of the g-Al2O3 during its transformation. 2. Experimental 2.1. Syntheses AIP was used as an aluminum precursor, AA as hydrolysis rate controller and 2-propanol as solvent during the synthesis. Initially, AIP was dissolved in 2-propanol under continuous stirring. By controlling the rate of addition of AA and water (H2O) to the above stirred solution, we could make the hydrolysis occur faster and the condensation occur slower so as to get the precipitates in the form of a fine hydroxide gel. The gel was further aged at 80 8C for 20 h. The molar ratio of AA/AIP was varied from 0 to 0.5, whereas that of H2O/AIP was varied from 3 to 25. The product was washed several times with 2-propanol and finally dried at 80 8C in vacuum for 12 h. Finally, the material was calcined in a flow of air at 500 8C for 5 h with a heating rate of 2 8C/min. 2.2. Characterization Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) of the samples were conducted in a TA Instrument (DMA, SDT 2960) in flowing nitrogen atmosphere at a heating rate of 10 8C/min up to 1200 8C using a commercial alumina as the reference material to discover the various decomposition steps occurring in the as-dried precursor as a function of temperature. In order to identify the various phases present and the crystallinity of as-prepared boehmite and calcined g-Al2O3 powder, we carried out powder X-ray diffraction (XRD) studies with a Rigaku diffractometer using Cu-Ka radiation. The Brunauer–Emmett–Teller (BET) surface areas and pore volumes were determined from nitrogen adsorption and desorption isotherm data obtained at 196 8C on a constant-volume adsorption apparatus (Micromeritics, ASAP-2400). The pore volumes were determined at a relative pressure (P/Po) of 0.99. The as-prepared samples were degassed at 150 8C for 3 h before measurements. The pore size distributions in as-prepared samples were determined by a Barett–Joyner–Halenda (BJH) model from the adsorption branch of the nitrogen isotherm [16]. The analysis of temperature programmed desorption of ammonia (NH3-TPD) was performed to determine the total acid sites on the catalyst. About 0.1 g of the sample was initially flushed
with a He flow at 500 8C for 5 h, next cooled to 100 8C and then saturated with NH3. After NH3 exposure, the sample was purged with He until the initial excess of NH3 which is not utilized is removed. Then this sample is heated from 100 8C to 700 8C at a heating rate of 10 8C/min. The BEL-CAT (PCI-3135) instrument was employed to monitor the amount of ammonia in the effluent by a thermal conductivity detector (TCD) and the values were recorded as a function of temperature. The microstructures of both as-prepared and calcined samples were studied by transmission electron microscope (TEM) images obtained on a JEOL JEM 2100F (field emission electron microscope) instrument operated at 200 kV. Fourier transform infrared (FT-IR) spectra of boehmite and g-Al2O3 powder were recorded using a Bio-Rad Digilab FTS-165 FT-IR spectrometer. 2.3. Activity measurement in methanol to DME The performances of the prepared catalysts with different AA/AIP ratios were compared with that of the g-Al2O3 catalyst prepared from catapal-B (SASOL) boehmite. The vapor phase dehydration of methanol containing 20 mol% H2O was carried out in a fixed-bed reactor (inner diameter = 0.8 cm and length = 30 cm). Prior to experiments, the catalyst (volume of 1.5 ml and size of the pellet in the range of 20–40 mesh) was pretreated for 1 h at 300 8C under a N2 flow. The methanol solution was fed into the reactor using a pump. The reaction was performed with N2 as a carrier gas at 10 atm pressure, in the temperature range of 210–400 8C and at a methanol feed rate of 0.25 ml/min (SV = 10 h1). The reaction products were analyzed on a gas chromatograph (GC) equipped with a flame ionization detector connected with a capillary column (Porapack Q). 3. Results and discussion 3.1. Physico-chemical properties of the synthesized boehmites Boehmite is obtained in the present study through hydrolysis of AIP in the presence or the absence of acetic acid, followed by aging at 80 8C for 20 h. In the sol–gel process, two simultaneous reactions, namely hydrolysis and condensation, occur when AIP reacts with water. The amount of water determines the degree of hydrolysis and the type of initial species formed, thus influencing condensation reactions that involve polymerization of hydrolyzed species in alcoholic medium. If the H2O/AIP ratio is kept 3, the AIP would get hydrolyzed completely, leading probably to the nucleation of tiny particles of boehmite after the aging process by the following reactions: AlðORÞ3 þ 3H2 O ! AlðOHÞ3 þ 3ROH
(1)
AlðOHÞ3 ! AlOðOHÞ þ H2 O
(2)
AlðORÞ3 þ 2H2 O ! AlOOH þ 3ROH
(3)
Initially, the amorphous hydroxide gel precipitated by reaction (1) is converted in to boehmite precursor after 20 h of aging at 80 8C by reactions (2) and (3), respectively. To confirm the formation of boehmite by reactions (2) and (3), we undertook DTA/ TGA, XRD, and IR studies on the as-dried boehmite precursor. The DTA/TGA curves of the as-dried boehmite using acetic acid during processing are shown in Fig. 1(a). The as-dried boehmite precursor appeared to have undergone three stages of decomposition reaction with a total weight loss of 35%. The first step corresponds to an endothermic weight loss of 15% which is attributable to the removal of adsorbed water below 200 8C. The second broad exothermic weight loss is due to the decomposition of associated organics including adsorbed acetic acid, followed by a
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
115
Fig. 2. XRD patterns of as-dried boehmite with variation of AA/AIP ratios. (a) AA/ AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.1, (d) AA/AIP = 0.5.
Fig. 1. DTA/TGA curves of as-dried boehmite precursor with and without acetic acid addition during processing.
third process above 500 8C which is attributable to a loss due to the slow continuous dehydroxylation [2]. The total loss in these two steps is 20%. The broad exothermic peak ranging from 400 8C to 900 8C may be ascribed to the crystallization of g-alumina and other metastable forms of alumina [2]. A continuously falling TGA curve suggests that a slow dehydroxylation reaction proceeds as a function of temperature. The exothermic peak at a temperature above 1100 8C, without any weight loss in TGA curve, is indicative of conversion of the metastable alumina to thermodynamically stable a-alumina phase [2]. On the other hand, the DTA/TGA of the as-dried boehmite precursor (Fig. 1(b)), in the absence of acetic acid during preparation, showed a total loss of 20% in the TGA curve in two stages. The DTA/TGA plot of the as-dried boehmite shows 15% more weight loss in the boehmite samples (when acetic acid is used during processing), suggesting that acetic acid might have adsorbed on the boehmite precursor during processing. The XRD patterns of the as-dried precursor are shown in Figs. 2 and 3 respectively with the variation of AA/AIP and H2O/AIP ratios. All the reflection peaks in Figs. 2 and 3 are assigned to the boehmite phase namely AlO(OH) with the orthorhombic symmetry [17]. It is interesting to note from Fig. 2 that all the peaks become broader with the increase of AA/AIP ratio, which is indicating the decrease of crystallinity of boehmite particles. With the increase of H2O/AIP ratio, the peak intensity of all reflection peaks increases. This result suggests an improvement in crystallinity of boehmite phase [17]. It is reported [3] that boehmite with low crystallinity produces gAl2O3 powder with high surface area whereas crystalline boehmite usually gives lower-surface area Al2O3 powder. Thus, it is possible to tune the crystallinity of boehmite precursor by adding required quantities of acetic acid during the preparation step. Figs. 4 and 5 show the XRD patterns of calcined boehmite precursors at 500 8C for 5 h in air. The new peaks appearing in the
Fig. 3. XRD patterns of as-dried boehmite with variation of H2O/AIP ratios. (a) H2O/ AIP = 3, (b) H2O/AIP = 6, (c) H2O/AIP = 12, (d) H2O/AIP = 25.
XRD patterns are assigned to the g-alumina phase formed due to crystallization process, as is described in DTA/TGA studies. The crystallinity of g-alumina decreases with the increase of AA/AIP ratio, as is seen from the broadening of XRD peaks. However, a better crystallinity of g-alumina is obtained at a H2O/AIP ratio 6. A further increase in the H2O/AIP ratio is not effective in improving the crystallinity of g-alumina powder. The diffraction peaks observed in all the XRD patterns were broad and diffuse, because the crystallites were very small. Such a size indicates their nanocrystalline nature in the as-dried boehmite precursor and calcined g-alumina material. The characteristics peaks of g-alumina phase at 2u = 45.908 for (4 0 0) reflection and 2u = 66.98 for (4 4 0) reflection are seen in XRD analyses of the calcined material. Thus, the DTA/TGA and XRD studies on the as-dried boehmite precursor confirm that the as-dried precursor is boehmite and that its thermal decomposition leads to formation of g-alumina phase, the crystallinity of which is found to be dependent on H2O/AIP and AA/ AIP molar ratios. The results of the surface area, total pore volume and average pore diameter for the as-dried precursor of boehmite, are
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
116
Table 2 Physicochemical properties of boehmite with the variation of H2O/AIP ratio Molar ratio of water to aluminum (H2O/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
3 6 12 25
717.8 491.1 494.7 427.7
6.4 8.8 8.7 6.6
1.19 1.08 1.08 0.70
Synthetic conditions: AIP/AA/IPA = 1/0.1/25 and dried at 80 8C in vacuum condition.
range of 6.4–8.8 nm. The nature of N2 adsorption and desorption isotherms represents a typical type IV curve, which suggests the predominance of mesopores for all AA/AIP and H2O/AIP ratios. A change in the hysteresis indicates a change in the pore structure in these materials [2]. 3.2. Preparation and characterization of g-Al2O3 from boehmite Fig. 4. XRD patterns of g-alumina with variation of AA/AIP ratios. (a) AA/AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.1, (d) AA/AIP = 0.5.
A similar trend is also found in the g-Al2O3 powder prepared from the thermal decomposition of boehmite at 500 8C for 5 h in air. The results of the surface area, total pore volume and average pore diameter are summarized in Tables 3 and 4, respectively. It can be seen from the Table 3 that the BET surface area increases from 306.5 m2/g to 437.8 m2/g when the AA/AIP ratio changes from 0 to 0.50. At the same time, the pore diameter decreases from 19.8 nm to 7.7 nm and the pore volume decrease from 1.52 cm3/g to 0.84 cm3/g with the variation of AA/AIP ratio during processing. However, one can see from Table 4 that, as the H2O/AIP ratio increases, the BET surface area and the average pore diameter are stabilized in the ranges of 445.3–359.0 m2/g and 10.5–8.1 nm, respectively. The pore volume is stabilized in the range of 1.16– 0.73 cm3/g. The nature of N2 adsorption and desorption isotherms remains the typical type IV, with changes in the hysteresis indicating the changes in the pore structure in the material [2]. 3.3. The role of acetic acid
Fig. 5. XRD patterns of g-alumina with variation of H2O/AIP ratios. (a) H2O/AIP = 3, (b) H2O/AIP = 6, (c) H2O/AIP = 12, (d) H2O/AIP = 25.
presented in Tables 1 and 2 respectively. In the case of using AA/AIP ratios as a parameter, a definite trend is observed from Table 1. The BET surface area increased from 314 m2/g to 628 m2/g with the increase of AA/AIP ratios. The average pore diameter is reduced from 16.1 nm to 4.5 nm and the pore volume also decreases from 1.25 cm3/g to 0.70 cm3/g with the increase of AA/AIP ratio. It is interesting to see that a maximum surface area of 717.8 m2/g is obtained when the H2O/AIP ratio is kept 3. As can be seen from Table 2 the pore volume decreased from 1.19 cm3/g to 0.70 cm3/g with the increase of H2O/AIP ratio. The pore size remained in the
Table 1 Physicochemical properties of boehmite with the variation of AA/AIP ratio
An analysis of all these results indicates that acetic acid plays a crucial role in controlling the surface area and pore size distribution in the as-dried boehmite and the resulting g-alumina powder. The pore size distribution is not much affected in the galumina powder with the variation of H2O/AIP molar ratio. The pore size distribution changes systematically to mesoporous range when the AA/AIP ratio is increased from 0 to 0.5. For the H2O/AIP Table 3 Physicochemical properties of g-alumina with the variation of AA/AIP ratio Molar ratio of acetic acid to aluminum (AA/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
0.0 0.035 0.1 0.5
306.5 350.5 412.8 437.8
19.8 16.5 12.9 7.7
1.52 1.44 1.33 0.84
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and calcined at 500 8C for 5 h.
Table 4 Physicochemical properties of g-alumina with the variation of H2O/AIP ratio
Molar ratio of acetic acid to aluminum (AA/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
Molar ratio of water to aluminum (H2O/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
0.0 0.035 0.1 0.5
314.0 401.3 491.1 628.5
16.1 12.7 8.8 4.5
1.25 1.25 1.08 0.70
3 6 12 25
445.3 412.8 395.8 359.0
10.5 12.9 12.3 8.1
1.16 1.33 1.21 0.73
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and dried at 80 8C in vacuum condition.
Synthetic conditions: AIP/AA/IPA = 1/0.1/25 and calcined at 500 8C for 5 h.
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Fig. 6. FT-IR spectra of boehmite and g-alumina. (a) AA/AIP = 0, (b) AA/AIP = 0.5, (c) AA/AIP = 0.5 calcined at 500 8C.
ratio variation; the effect is less predominant in controlling the pore size and the pore distribution. All of these observations establish the facts that the preparation conditions have tremendous effects on the average pore diameter and pore size distribution during preparation and can be controlled by selecting the proper AA/AIP and H2O/AIP ratios while keeping all other
117
conditions the same. In short, acetic acid is also found to play some crucial role in controlling the microstructure of g-Al2O3 powders probably through its selective adsorption on high-energy faces of randomly oriented boehmite particles nucleated during the aging process [18]. The confirmation for this argument comes from the FT-IR studies on the aged boehmite precursor. The FT-IR spectra presented in Fig. 6 shows that all the absorption bands corresponding to the boehmite phase [19,20] have octahedral Al–O coordination. The IR spectrum shown in Fig. 6(b) clearly, indicate the presence of acetate groups [18] on the surface of boehmite precursor. The bands at 1469 cm1 and 1579 cm1 are assigned to the stretching mode of adsorbed acetate CH3COO groups due to chelating or bridging type acetate coordination with Al(III) [18,19,21]. The presence of a small hump at 1635 cm1 is assigned to the presence of hydroxyl groups in boehmite precursor [19]. The band at 1635 cm1 (assigned to hydroxyl group) still persists in the samples even after calcination in air. The characteristic absorption peaks (Fig. 6(c)) in the region 500– 750 cm1 are assigned to the vibrations of AlO6 and a shoulder observed at 886 cm1 is thought to be due to the vibration of gAlO4 [19]. Thus, the g-Al2O3 powder produced by the calcination of boehmite precursor is found to contain both tetrahedral and octahedral Al–O coordination. The absence of bands at 1276 and 1415 cm1 i.e., the bands characteristic of free acetic acid in the material [21], confirms its absence in the material. TEM studies have been undertaken to understand the effects of acetic acid on the morphology of boehmite and of the calcined gAl2O3 powder as well. Fig. 7(a) gives the TEM image of the as-
Fig. 7. TEM images of boehmite and g-alumina powders (bar = 20 nm). (a) AA/AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.5, (d) AA/AIP = 0.035 calcined at 500 8C.
118
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
prepared boehmite precursor that shows the presence of rod shaped porous particles (consisting of aggregates of crystallites) with different lengths (40–50 nm) having a diameter of 1–2 nm. However, when the AA/AIP ratio is kept at 0.035 during processing, the as-dried boehmite precursor shows material that consists of interconnected networks of thinner porous rods having diameter smaller than that of the powder when the AA/AIP ratio is 0 (Fig. 7(b)). With the further increase of AA/AIP ratio, the boehmite precursor consists of a wormlike microstructure (Fig. 7(c)) with still thinner elongated interconnected networks of nanometer sized smaller rods with variation in length. The gAl2O3 powder, however, shows the material with the presence of rods, as can be seen from Fig. 7(d). The diameter of the particles is found to be in nanometer range (1.5–2.5 nm) with uniform lengths of 20–25 nm. All these observations are consistent with those reported by Hochepied et al. [22] that the alignments of nanoparticles lead to the building of polycrystalline fibers with diameters of 3–8 nm and lengths of about 100 nm. 3.4. The formation of high surface areas in boehmite Careful analyses of FT-IR and TEM together postulate that the morphology of boehmite precursor can be controlled by the adsorption of acetate groups on as-dried boehmite precursor during preparation. One possible mechanism could be that the boehmite particles nucleated by reactions (2) and (3) gradually ripen to form the boehmite nano-crystals during the aging process. Then the reaction-limited aggregation of these particles via specific sites [23] (where no adsorption of acetic acid takes place) can occur to minimize the interfacial energy so as to give a particular morphology to boehmite aggregates. High-energy faces of the boehmite particles where the CH3COO groups are selectively adsorbed [24] are not available for the reaction-limited aggregation process during aging. The interaction between CH3COO groups and those faces possessing the highest energy would be relatively stronger and the growth rates of these facets are decelerated most. In other words, CH3COO ions have a balancing effect on the growth rates of different faces in such a way that the desired rod shaped morphology is obtained in the boehmite precursor. The same morphology is retained in the g-Al2O3 powder by topotactic reactions [25] with the g-Al2O3 powder. Similar observations have been recently been reported by Zhang et al. [26] for the synthesis of ZnO micro-crystals, wherein the growth habit is controlled by its intrinsic structure and external conditions like temperature, pH and capping agent. Such observations are also separately reported by Khollam et al. [27] for the Ba1xSrxTiO3 powder in achieving the star shaped morphology because of selective adsorption of C2O42 groups on the faces of the as-dried crystalline precursor Ba1xSrxTiO(C2O4)2. Therefore, the role of acetic acid during processing in the current investigation is found to be crucial. Further characterization of g-Al2O3 powder was carried out by measuring its PL spectra with excitation at 325 nm at room temperature to confirm the presence of oxygen vacancies in these materials. The spectra are depicted in Fig. 8. The commercial gAl2O3 obtained from catapal-B (Fig. 8(a)) showed similar features in PL spectrum to those of the g-Al2O3 powders (Fig. 8(b)) prepared without using acetic acid during preparation. A strong blue emission at 437.6 nm is observed in both the samples indicating the presence of a large amount of F+ centers [27] in the g-Al2O3 powder. The low intensity broad band centered at 800 nm may be attributed to the presence of defects other than the F+ centers in the samples [28,29]. However, two additional sharp peaks are observed at 585.8 nm and 616.5 nm in the g-Al2O3 powder (Fig. 8(c)) prepared using acetic acid which are absent in the g-
Fig. 8. PL spectra of g-alumina powder obtained from (a) catapal-B, (b) w/o acetic acid, (c) with acetic acid.
Al2O3 powders prepared without using acetic acid during preparation. The doublet appearing at 585.8 nm and 616.5 nm is attributed to the presence of a small amount of u-Al2O3 formed during calcination at 500 8C in air. The doublet is shifted to lower wavelength side as compared to the reported data [30] due to the stress in the material. This observation is consistent with earlier DTA/TGA, XRD and TEM results that the nanocrystalline g-Al2O3 with high surface area may transform to u-Al2O3 phase in small amounts at calcination temperatures close to that of the present work. The phase percentage could be below the detection limit of XRD, as no peaks corresponding to the u-Al2O3 are seen in the XRD pattern. The DTA/TGA also showed a broad exothermic peak due to crystallization of the g-Al2O3 and metastable phases. Therefore, it is possible that the g-Al2O3 powder might contain small amounts of the u-Al2O3 below the detection level of XRD. 3.5. Catalytic activity on the synthesized g-Al2O3 with different surface areas The aluminas synthesized by changing the mole ratio of AA/AIP and by showing different surface areas are further characterized by NH3-TPD and tested for the activities on methanol dehydration to DME. Fig. 9 shows the NH3-TPD profiles of g-alumina powder; quantitative results obtained for the same are summarized in Table 5. It is evident from Table 5 that the amount of ammonia desorbed from the g-alumina powder is higher (1.12 mmol/g) when the AA/AIP ratio is maintained at 0.50 than that of 0.82 mmol/g observed for the AA/AIP ratio 0.0. This amount gradually increases with the increase of AA/AIP ratio due to the increasing surface area. The g-alumina powder derived from the commercial catapal-B boehmite sample, however, gave a lower value for the desorbed ammonia 0.65 mmol/g. The TPD patterns (Fig. 9) of the g-alumina powder showed two intense peaks in the range of 100–800 8C. The first peak (I) corresponds to acid sites which is attributable to the removal of adsorbed NH3 on the catalyst surface. The second peak (II) is due to the slow dehydroxylation process [2] of surface hydroxyl groups on g-Al2O3. The first peak can be assigned to the NH3 desorbed from acid sites. The intensity of the peaks is found to increase significantly by keeping the AA/AIP ratio 0.5 during preparation. From these results, one can conclude that the g-alumina powder reported in the present study possesses higher surface acidic sites as compared to the g-alumina powder obtained from the
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Fig. 9. NH3-profiles of g-alumina obtained from (a) catapal-B, (b) AA/AIP = 0, (c) AA/ AIP = 0.035, (d) AA/AIP = 0.1 (e) AA/AIP = 0.5.
Table 5 Summary of surface areas and acidic sites measured by NH3-TPD on g-alumina with the variation of AA/AIP ratio Molar ratio of acetic acid to aluminum (AA/AIP)
BET surface area (m2/g)
Acidic sites (mmol NH3/g)
catapal-B 0.0 0.035 0.1 0.5
220.0 306.5 350.5 412.8 437.8
0.6527 0.8214 0.8598 1.0000 1.1192
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and calcined at 500 8C for 5 h.
commercial catapal-B boehmite and the density of sites is increased with the increase of AA/AIP molar ratio. The presence of acetic acid during the preparation has helped to enhance the concentration of acidic sites in these materials. The use of acetic acid during preparation is not only helpful to improve surface area of boehmite/g-Al2O3 powder (Tables 1 and 3) but also to enhance the amount of surface acidic sites of the g-Al2O3 powders. Thus, it is possible to control easily the amount of surface acidic sites as well as pore sizes, their distribution and the crystallinity of boehmite/g-Al2O3 powder by use of acetic acid during preparation. The g-Al2O3 catalyst obtained by keeping the AA/AIP ratio 0.5 and the H2O/AIP ratio 6.0 during processing is used for the methanol to DME conversion. The performance of this catalysts is compared with that of the g-Al2O3 catalyst obtained from the commercial boehmite i.e. catapal-B. The results on the catalyst indicate 80% conversion and selectivity to DME of 99.9%, almost approaching the equilibrium conversion and the maximum selectivity. The temperature at which the equilibrium conversion of methanol is reached above 320 8C, is for the g-Al2O3 (obtained from catapal-B) and it is also found to be above 320 8C for the catalyst prepared without using acetic acid. However, the catalyst prepared using acetic acid during synthesis needed a temperature above 300 8C to get the equilibrium conversion. Thus, a better performance at lower temperature was obtained on our catalyst prepared using acetic acid than on the g-Al2O3 catalyst derived from the commercially available catapal-B boehmite sample or on the sample prepared without acetic acid. Furthermore, the variations in catalytic activity with surface area and the amount of acidic sites are shown in Fig. 10. In general, the catalytic activity in methanol dehydration strongly depends on the surface acidic properties, such as the total number of acidic sites and their strength. It is apparent that the number of acidic
119
Fig. 10. Correlation of surface area and amount of acidic sites with the catalytic activity of methanol dehydration to DME on the g-alumina obtained at different mole ratios of AA/AIP (filled symbols) and catapal-B (hollow symbols).
sites showing almost the same acidic strength can be precisely controlled by appropriately selecting the synthetic method during alumina preparation (Table 5 and Fig. 9). It is also interesting to see a good correlation between the catalytic activity, the surface area and the amount of acidic sites. Yoo et al. [31] in their recent study on DME synthesis have reported a good correlation between the catalytic activity and the number of weak acidic sites. The present results, reported in Table 5, also agree with this observation, giving the importance of the amount of acidic sites in the dehydration reaction. The acidic sites are also less prone to deactivation during the reaction. The methanol dehydration activity, expressed in terms of the temperature required to obtain 50% conversion, is correlated with the surface area of alumina in Fig. 10. With increasing surface area of g-Al2O3, the required temperature to reach 50% conversion of methanol is found to decrease from 325 8C to 305 8C due to the increased number of acidic sites which are active for methanol dehydration. These results also suggest that a high throughput yield can be obtained at high space velocity on the high surface area alumina without a complicated modification process. 4. Conclusion Nano-sized high surface area boehmite particles can be prepared by adopting the sol–gel method, during which suitable ratios of AA/AIP and H2O/AIP and suitable aging times are required. Addition of acetic acid enhances the textural and structural characteristics of boehmite. Thermal treatment of the boehmite at 550 8C gives a high surface area g-Al2O3 powder having controlled porosity and pore size distribution. The acetic acid adsorbed on the surface plays a crucial role in controlling the morphology of the boehmite precursor as well as that of the gAl2O3 powder. The structure–property relationship observed in the present study is useful to make tailor made g-Al2O3 with high surface acidity. The alumina thus prepared is highly active and selective in the dehydration of methanol, giving an equilibrium conversion and close to 100% selectivity, under the selected reaction conditions. The number of acidic sites on alumina is the crucial factor for catalytic activity of methanol dehydration. A good correlation is obtained between the number of acid sites and the temperature required to reach 50% conversion. With the increase of surface area of g-Al2O3, the temperature required for 50% conversion of methanol decreased due to the increase of available acidic sites.
120
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Acknowledgements This research was supported by a grant from Carbon Dioxide Reduction & Sequestration Center (CDRS), one of 21st Century Frontier Programs funded by the Ministry of Science and Technology, Republic of Korea. Dr. H.S. Potdar is thankful to KOSEF for giving a Brain-Pool fellowship and to the Director, National Chemical Laboratory, Pune 411008, for granting EOL. We are thankful to Dr. K.N. Kim for doing PL measurements of our samples. References [1] T.F. Baumann, A.E. Gash, S.C. Chin, A.M. Sawvel, R.S. Maxwell, J.H. Satcher Jr., Chem. Mater. 17 (2005) 395–401. [2] P. Padmaja, P.K. Pillai, K.G.K. Warrier, M. Padmanabhan, J. Porous Mater. 11 (2004) 147–155. [3] D. Panias, A. Krestou, Powder Technol. 175 (2007) 163–173. [4] E. Elaloui, A.C. Pierre, G.M. Pajonk, J. Catal. 166 (1997) 340–346. [5] F. Yaripour, F. Baghaei, I. Schmidt, J. Perregaard, Catal. Commun. 6 (2005) 147– 152. [6] S.M. Kim, Y.J. Lee, K.W. Jun, J.Y. Park, H.S. Potdar, Mater. Chem. Phys. 104 (2007) 56–61. [7] H.S. Potdar, K.W. Jun, J.W. Bae, S.M. Kim, Y.J. Lee, Appl. Catal. A 321 (2007) 109–116. [8] T.A. Semelsberger, R.L. Borup, H.L. Greene, J. Power Sources 156 (2006) 497–511. [9] J.D. Holladay, Y. Wang, E. Jones, Chem. Rev. 104 (2004) 4767–4790.
[10] S. Lee, A. Sardesai, Top. Catal. 32 (2005) 197–207. [11] K. Okada, A. Tanaka, S. Hayashi, K. Daimon, N. Otsuka, J. Mater. Res. 9 (1994) 1709–1713. [12] T. Fukui, M. Hori, J. Mater. Sci. 31 (1996) 3245–3248. [13] B.E. Yoldas, J. Mater. Sci. 10 (1975) 1856–1860. [14] L. Wang, Y. Qi, Y. Wei, D. Fang, S. Meng, Z. Liu, Catal. Lett. 106 (2006) 61–66. [15] T. Takeguchi, K.I. Yanagisawa, T. Inui, M. Inoue, Appl. Catal. A 192 (2000) 201–209. [16] E.P. Barett, L.G. Joyner, P.P. Halender, J. Am. Chem. Soc. 73 (1951) 373–380. [17] J.A. Wang, X. Bokhimi, A. Morales, O. Novaro, T. Lopez, R. Gomez, J. Phys. Chem. B 103 (1999) 299–303. [18] S.B. Deshpande, H.S. Potdar, Y.B. Khollam, K.R. Patil, R. Pasricha, N.E. Jacob, Mater. Chem. Phys. 97 (2006) 207–212. [19] G.K. Priya, P. Padmaja, K.G.K. Warrier, A.D. Damodaran, G. Aruldhas, J. Mater. Sci. Lett. 16 (1997) 1584–1587. [20] S. Music, D. Dragcevic, S. Popovic, Mater. Lett. 40 (1999) 269–274. ¨ hman, Geochim. Cosmochim. Acta 62 (1998) 3657– [21] P. Persson, M. Karlsson, L.-O. O 3668. [22] J.F. Hochepied, P. Nortier, Powder Technol. 128 (2002) 268–275. [23] Y. Zhou, M. Antonietti, J. Am. Chem. Soc. 125 (2003) 14960–14961. [24] X.C. Jiang, Series of Selected Papers from Chun-Tsung Scholars, Peking University, 2003. [25] P. Alphonse, M. Courty, Thermochim. Acta 425 (2005) 75–89. [26] H. Zhang, D. Yang, D. Li, X. Ma, S. Li, D. Que, Cryst. Growth Des. 5 (2005) 547–550. [27] Y.B. Khollam, H.S. Potdar, S.B. Deshpande, A.B. Gaikwad, Mater. Chem. Phys. 97 (2006) 295–300. [28] S. Liu, L. Zhang, L. An, W. Fei, H. Heinrich, J. Am. Ceram. Soc. 88 (2005) 2559–2563. [29] X. Sun, F. Xu, Z. Li, W. Zhang, J. Lumin. 121 (2006) 588–594. [30] X. Peng, D.R. Clarke, F. Wang, Oxid. Met. 60 (2003) 225–240. [31] K.S. Yoo, J.H. Kim, M.J. Park, S.J. Kim, O.S. Joo, K.J. Jung, Appl. Catal. A 330 (2007) 57–62.
Contents lists available at ScienceDirect
Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata
Synthesis and characterization of a highly active alumina catalyst for methanol dehydration to dimethyl ether Seung-Moon Kim, Yun-Jo Lee, Jong Wook Bae, H.S. Potdar, Ki-Won Jun * Alternative Chemicals/Fuel Research Center, Korea Research Institute of Chemical Technology (KRICT), P.O. Box 107, Yuseong, Daejeon 305-600, Republic of Korea
A R T I C L E I N F O
A B S T R A C T
Article history: Received 19 March 2008 Received in revised form 23 June 2008 Accepted 24 June 2008 Available online 1 July 2008
A simple sol–gel method was adopted to synthesize boehmites with high surface area using aluminum iso-propoxide (AIP), acetic acid (AA) and 2-propanol, and the effects of surface area and methanol dehydration on activity were investigated. The hydrolysis conditions of AIP in the presence of AA in 2propanol solvent were systematically varied to observe their effect on phase formation, crystallinity, surface area and pore size distribution of the alumina. The surface area and the number of acidic sites varied considerably with the variation in the molar ratio of AA/AIP. This study revealed that a high surface area boehmite (in the range of 628–717 m2/g) could be obtained by keeping the molar ratio of AA/AIP as 0.5 and that of H2O/AIP at 3. Rod shaped, porous g-Al2O3 powder with a high surface area of 438 m2/g was obtained after calcination of the boehmite at 500 8C for 5 h in air. The temperature programmed desorption of ammonia (NH3-TPD) of the g-Al2O3 samples demonstrated higher concentration of acidic sites when acetic acid was used during preparation than when it was not used. The vapor phase dehydration of methanol (containing 20 mol% H2O) to dimethyl ether (DME) was conducted on the prepared aluminas. With increasing surface area of g-Al2O3, the temperature required to reach 50% conversion of methanol decreased due to the increased number of acidic sites which are favorable for methanol dehydration with low byproduct formation. The catalytic activity for methanol dehydration to DME correlated well with the total number of acidic sites of g-Al2O3, which was controlled by changing the AA/AIP and H2O/AIP molar ratios. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Methanol dehydration Dimethyl ether Sol–gel g-Alumina Acidity
1. Introduction Alumina is a low cost material most widely used as a catalyst and catalyst support [1]. In addition, it is also used as the starting material for the preparation of Al2O3 based ceramics [2]. A wide variety of these applications are possible because of the fact that alumina exists in the corundum and transition alumina forms [3]. The corundum or a-alumina has excellent mechanical, electrical, thermal and optical properties due to hexagonal close packing of oxygen ions. On the other hand, transition aluminas, including gAl2O3, have a cubic close packing of oxygen ions resulting in high surface area, mesoporosity and surface acidity [4]. As a result of these important properties, g-Al2O3 is also extensively used as an adsorbent and a membrane [3]. Solid acid catalysts e.g. g-Al2O3, modified g-Al2O3 with silica, phosphorus or B2O3 based are widely used, excellent catalysts for the dehydration of methanol to DME [5]. However, systematic study on the effect of various preparation
* Corresponding author. Tel.: +82 42 860 7671; fax: +82 42 860 7388. E-mail address: [email protected] (K.-W. Jun). 0926-860X/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2008.06.032
parameters on physical and chemical characteristics of g-Al2O3 affecting DME synthesis is lacking. In the our previous investigations, the efforts to achieve the preparation of g-Al2O3 powders by using various chemical routes through boehmite precursor have been continued to synthesize a thermally stable g-Al2O3 at fixed preparation conditions [6] and to evaluate the catalytic activity of home-made g-Al2O3 prepared from aluminum nitrate precursor via coprecipitation/digestion routes [7]. An attempt is made in the present investigation to synthesize g-Al2O3 catalysts with different surface areas by varying preparation parameters systematically during sol–gel synthesis and to correlate the catalytic activity with the surface area of g-Al2O3. In view of the increasing demand for DME as a raw material for the production of dimethyl sulfate, methyl acetate, light olefins and alternative clean fuels [8–10], the importance of a commercially viable catalyst is further enhanced. In order to get a reliable and reproducible alumina for these applications, a stringent control of composition, surface area, porosity (i.e. pore size and its distribution) and surface acidity are essential. Although various chemical routes [4,11,12] have been tried, the sol–gel route offers an excellent opportunity for controlling the physical, chemical and
114
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
textural properties of the aluminum oxide. Sol–gel derived alumina powders are generally prepared through acid- or basecatalyzed hydrolysis and condensation reactions of aluminum alkoxide precursors such as aluminum iso-propoxide (AIP) or aluminum sec-butoxide [1,2] in organic solvents. In the case of the Yoldas process [13], it is reported that the aging treatment over a period of 24 h at 80 8C yields boehmite [AlO(OH)] whereas aging at room temperature yields bayerite, Al(OH)3. The thermal decomposition of boehmite at 400–500 8C in air produces g-Al2O3. The catalytic activity of the alumina for methanol dehydration is generally dependent on the surface acidity, which could be varied by adding some promoters or controlling the acidic properties of alumina or zeolites [14,15]. Hence, it is important for the synthesis of g-Al2O3 with controllable and reproducible properties to get the stable catalytic activity. It is also necessary to study the effect of various preparation parameters on the physico-chemical properties of the solid–acid catalyst, with particular attention given to the changes in the acidity. The present investigation focuses on elucidating the effects of acetic acid (AA)/AIP and H2O/AIP mole ratios on the properties of boehmite and also on the g-Al2O3 obtained by subsequent heat treatment. Detailed investigations have been carried out to correlate the changes in acidity with those of the textural properties of the g-Al2O3 during its transformation. 2. Experimental 2.1. Syntheses AIP was used as an aluminum precursor, AA as hydrolysis rate controller and 2-propanol as solvent during the synthesis. Initially, AIP was dissolved in 2-propanol under continuous stirring. By controlling the rate of addition of AA and water (H2O) to the above stirred solution, we could make the hydrolysis occur faster and the condensation occur slower so as to get the precipitates in the form of a fine hydroxide gel. The gel was further aged at 80 8C for 20 h. The molar ratio of AA/AIP was varied from 0 to 0.5, whereas that of H2O/AIP was varied from 3 to 25. The product was washed several times with 2-propanol and finally dried at 80 8C in vacuum for 12 h. Finally, the material was calcined in a flow of air at 500 8C for 5 h with a heating rate of 2 8C/min. 2.2. Characterization Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) of the samples were conducted in a TA Instrument (DMA, SDT 2960) in flowing nitrogen atmosphere at a heating rate of 10 8C/min up to 1200 8C using a commercial alumina as the reference material to discover the various decomposition steps occurring in the as-dried precursor as a function of temperature. In order to identify the various phases present and the crystallinity of as-prepared boehmite and calcined g-Al2O3 powder, we carried out powder X-ray diffraction (XRD) studies with a Rigaku diffractometer using Cu-Ka radiation. The Brunauer–Emmett–Teller (BET) surface areas and pore volumes were determined from nitrogen adsorption and desorption isotherm data obtained at 196 8C on a constant-volume adsorption apparatus (Micromeritics, ASAP-2400). The pore volumes were determined at a relative pressure (P/Po) of 0.99. The as-prepared samples were degassed at 150 8C for 3 h before measurements. The pore size distributions in as-prepared samples were determined by a Barett–Joyner–Halenda (BJH) model from the adsorption branch of the nitrogen isotherm [16]. The analysis of temperature programmed desorption of ammonia (NH3-TPD) was performed to determine the total acid sites on the catalyst. About 0.1 g of the sample was initially flushed
with a He flow at 500 8C for 5 h, next cooled to 100 8C and then saturated with NH3. After NH3 exposure, the sample was purged with He until the initial excess of NH3 which is not utilized is removed. Then this sample is heated from 100 8C to 700 8C at a heating rate of 10 8C/min. The BEL-CAT (PCI-3135) instrument was employed to monitor the amount of ammonia in the effluent by a thermal conductivity detector (TCD) and the values were recorded as a function of temperature. The microstructures of both as-prepared and calcined samples were studied by transmission electron microscope (TEM) images obtained on a JEOL JEM 2100F (field emission electron microscope) instrument operated at 200 kV. Fourier transform infrared (FT-IR) spectra of boehmite and g-Al2O3 powder were recorded using a Bio-Rad Digilab FTS-165 FT-IR spectrometer. 2.3. Activity measurement in methanol to DME The performances of the prepared catalysts with different AA/AIP ratios were compared with that of the g-Al2O3 catalyst prepared from catapal-B (SASOL) boehmite. The vapor phase dehydration of methanol containing 20 mol% H2O was carried out in a fixed-bed reactor (inner diameter = 0.8 cm and length = 30 cm). Prior to experiments, the catalyst (volume of 1.5 ml and size of the pellet in the range of 20–40 mesh) was pretreated for 1 h at 300 8C under a N2 flow. The methanol solution was fed into the reactor using a pump. The reaction was performed with N2 as a carrier gas at 10 atm pressure, in the temperature range of 210–400 8C and at a methanol feed rate of 0.25 ml/min (SV = 10 h1). The reaction products were analyzed on a gas chromatograph (GC) equipped with a flame ionization detector connected with a capillary column (Porapack Q). 3. Results and discussion 3.1. Physico-chemical properties of the synthesized boehmites Boehmite is obtained in the present study through hydrolysis of AIP in the presence or the absence of acetic acid, followed by aging at 80 8C for 20 h. In the sol–gel process, two simultaneous reactions, namely hydrolysis and condensation, occur when AIP reacts with water. The amount of water determines the degree of hydrolysis and the type of initial species formed, thus influencing condensation reactions that involve polymerization of hydrolyzed species in alcoholic medium. If the H2O/AIP ratio is kept 3, the AIP would get hydrolyzed completely, leading probably to the nucleation of tiny particles of boehmite after the aging process by the following reactions: AlðORÞ3 þ 3H2 O ! AlðOHÞ3 þ 3ROH
(1)
AlðOHÞ3 ! AlOðOHÞ þ H2 O
(2)
AlðORÞ3 þ 2H2 O ! AlOOH þ 3ROH
(3)
Initially, the amorphous hydroxide gel precipitated by reaction (1) is converted in to boehmite precursor after 20 h of aging at 80 8C by reactions (2) and (3), respectively. To confirm the formation of boehmite by reactions (2) and (3), we undertook DTA/ TGA, XRD, and IR studies on the as-dried boehmite precursor. The DTA/TGA curves of the as-dried boehmite using acetic acid during processing are shown in Fig. 1(a). The as-dried boehmite precursor appeared to have undergone three stages of decomposition reaction with a total weight loss of 35%. The first step corresponds to an endothermic weight loss of 15% which is attributable to the removal of adsorbed water below 200 8C. The second broad exothermic weight loss is due to the decomposition of associated organics including adsorbed acetic acid, followed by a
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
115
Fig. 2. XRD patterns of as-dried boehmite with variation of AA/AIP ratios. (a) AA/ AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.1, (d) AA/AIP = 0.5.
Fig. 1. DTA/TGA curves of as-dried boehmite precursor with and without acetic acid addition during processing.
third process above 500 8C which is attributable to a loss due to the slow continuous dehydroxylation [2]. The total loss in these two steps is 20%. The broad exothermic peak ranging from 400 8C to 900 8C may be ascribed to the crystallization of g-alumina and other metastable forms of alumina [2]. A continuously falling TGA curve suggests that a slow dehydroxylation reaction proceeds as a function of temperature. The exothermic peak at a temperature above 1100 8C, without any weight loss in TGA curve, is indicative of conversion of the metastable alumina to thermodynamically stable a-alumina phase [2]. On the other hand, the DTA/TGA of the as-dried boehmite precursor (Fig. 1(b)), in the absence of acetic acid during preparation, showed a total loss of 20% in the TGA curve in two stages. The DTA/TGA plot of the as-dried boehmite shows 15% more weight loss in the boehmite samples (when acetic acid is used during processing), suggesting that acetic acid might have adsorbed on the boehmite precursor during processing. The XRD patterns of the as-dried precursor are shown in Figs. 2 and 3 respectively with the variation of AA/AIP and H2O/AIP ratios. All the reflection peaks in Figs. 2 and 3 are assigned to the boehmite phase namely AlO(OH) with the orthorhombic symmetry [17]. It is interesting to note from Fig. 2 that all the peaks become broader with the increase of AA/AIP ratio, which is indicating the decrease of crystallinity of boehmite particles. With the increase of H2O/AIP ratio, the peak intensity of all reflection peaks increases. This result suggests an improvement in crystallinity of boehmite phase [17]. It is reported [3] that boehmite with low crystallinity produces gAl2O3 powder with high surface area whereas crystalline boehmite usually gives lower-surface area Al2O3 powder. Thus, it is possible to tune the crystallinity of boehmite precursor by adding required quantities of acetic acid during the preparation step. Figs. 4 and 5 show the XRD patterns of calcined boehmite precursors at 500 8C for 5 h in air. The new peaks appearing in the
Fig. 3. XRD patterns of as-dried boehmite with variation of H2O/AIP ratios. (a) H2O/ AIP = 3, (b) H2O/AIP = 6, (c) H2O/AIP = 12, (d) H2O/AIP = 25.
XRD patterns are assigned to the g-alumina phase formed due to crystallization process, as is described in DTA/TGA studies. The crystallinity of g-alumina decreases with the increase of AA/AIP ratio, as is seen from the broadening of XRD peaks. However, a better crystallinity of g-alumina is obtained at a H2O/AIP ratio 6. A further increase in the H2O/AIP ratio is not effective in improving the crystallinity of g-alumina powder. The diffraction peaks observed in all the XRD patterns were broad and diffuse, because the crystallites were very small. Such a size indicates their nanocrystalline nature in the as-dried boehmite precursor and calcined g-alumina material. The characteristics peaks of g-alumina phase at 2u = 45.908 for (4 0 0) reflection and 2u = 66.98 for (4 4 0) reflection are seen in XRD analyses of the calcined material. Thus, the DTA/TGA and XRD studies on the as-dried boehmite precursor confirm that the as-dried precursor is boehmite and that its thermal decomposition leads to formation of g-alumina phase, the crystallinity of which is found to be dependent on H2O/AIP and AA/ AIP molar ratios. The results of the surface area, total pore volume and average pore diameter for the as-dried precursor of boehmite, are
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
116
Table 2 Physicochemical properties of boehmite with the variation of H2O/AIP ratio Molar ratio of water to aluminum (H2O/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
3 6 12 25
717.8 491.1 494.7 427.7
6.4 8.8 8.7 6.6
1.19 1.08 1.08 0.70
Synthetic conditions: AIP/AA/IPA = 1/0.1/25 and dried at 80 8C in vacuum condition.
range of 6.4–8.8 nm. The nature of N2 adsorption and desorption isotherms represents a typical type IV curve, which suggests the predominance of mesopores for all AA/AIP and H2O/AIP ratios. A change in the hysteresis indicates a change in the pore structure in these materials [2]. 3.2. Preparation and characterization of g-Al2O3 from boehmite Fig. 4. XRD patterns of g-alumina with variation of AA/AIP ratios. (a) AA/AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.1, (d) AA/AIP = 0.5.
A similar trend is also found in the g-Al2O3 powder prepared from the thermal decomposition of boehmite at 500 8C for 5 h in air. The results of the surface area, total pore volume and average pore diameter are summarized in Tables 3 and 4, respectively. It can be seen from the Table 3 that the BET surface area increases from 306.5 m2/g to 437.8 m2/g when the AA/AIP ratio changes from 0 to 0.50. At the same time, the pore diameter decreases from 19.8 nm to 7.7 nm and the pore volume decrease from 1.52 cm3/g to 0.84 cm3/g with the variation of AA/AIP ratio during processing. However, one can see from Table 4 that, as the H2O/AIP ratio increases, the BET surface area and the average pore diameter are stabilized in the ranges of 445.3–359.0 m2/g and 10.5–8.1 nm, respectively. The pore volume is stabilized in the range of 1.16– 0.73 cm3/g. The nature of N2 adsorption and desorption isotherms remains the typical type IV, with changes in the hysteresis indicating the changes in the pore structure in the material [2]. 3.3. The role of acetic acid
Fig. 5. XRD patterns of g-alumina with variation of H2O/AIP ratios. (a) H2O/AIP = 3, (b) H2O/AIP = 6, (c) H2O/AIP = 12, (d) H2O/AIP = 25.
presented in Tables 1 and 2 respectively. In the case of using AA/AIP ratios as a parameter, a definite trend is observed from Table 1. The BET surface area increased from 314 m2/g to 628 m2/g with the increase of AA/AIP ratios. The average pore diameter is reduced from 16.1 nm to 4.5 nm and the pore volume also decreases from 1.25 cm3/g to 0.70 cm3/g with the increase of AA/AIP ratio. It is interesting to see that a maximum surface area of 717.8 m2/g is obtained when the H2O/AIP ratio is kept 3. As can be seen from Table 2 the pore volume decreased from 1.19 cm3/g to 0.70 cm3/g with the increase of H2O/AIP ratio. The pore size remained in the
Table 1 Physicochemical properties of boehmite with the variation of AA/AIP ratio
An analysis of all these results indicates that acetic acid plays a crucial role in controlling the surface area and pore size distribution in the as-dried boehmite and the resulting g-alumina powder. The pore size distribution is not much affected in the galumina powder with the variation of H2O/AIP molar ratio. The pore size distribution changes systematically to mesoporous range when the AA/AIP ratio is increased from 0 to 0.5. For the H2O/AIP Table 3 Physicochemical properties of g-alumina with the variation of AA/AIP ratio Molar ratio of acetic acid to aluminum (AA/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
0.0 0.035 0.1 0.5
306.5 350.5 412.8 437.8
19.8 16.5 12.9 7.7
1.52 1.44 1.33 0.84
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and calcined at 500 8C for 5 h.
Table 4 Physicochemical properties of g-alumina with the variation of H2O/AIP ratio
Molar ratio of acetic acid to aluminum (AA/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
Molar ratio of water to aluminum (H2O/AIP)
BET area (m2/g)
Average pore size (nm)
Pore volume (cm3/g)
0.0 0.035 0.1 0.5
314.0 401.3 491.1 628.5
16.1 12.7 8.8 4.5
1.25 1.25 1.08 0.70
3 6 12 25
445.3 412.8 395.8 359.0
10.5 12.9 12.3 8.1
1.16 1.33 1.21 0.73
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and dried at 80 8C in vacuum condition.
Synthetic conditions: AIP/AA/IPA = 1/0.1/25 and calcined at 500 8C for 5 h.
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Fig. 6. FT-IR spectra of boehmite and g-alumina. (a) AA/AIP = 0, (b) AA/AIP = 0.5, (c) AA/AIP = 0.5 calcined at 500 8C.
ratio variation; the effect is less predominant in controlling the pore size and the pore distribution. All of these observations establish the facts that the preparation conditions have tremendous effects on the average pore diameter and pore size distribution during preparation and can be controlled by selecting the proper AA/AIP and H2O/AIP ratios while keeping all other
117
conditions the same. In short, acetic acid is also found to play some crucial role in controlling the microstructure of g-Al2O3 powders probably through its selective adsorption on high-energy faces of randomly oriented boehmite particles nucleated during the aging process [18]. The confirmation for this argument comes from the FT-IR studies on the aged boehmite precursor. The FT-IR spectra presented in Fig. 6 shows that all the absorption bands corresponding to the boehmite phase [19,20] have octahedral Al–O coordination. The IR spectrum shown in Fig. 6(b) clearly, indicate the presence of acetate groups [18] on the surface of boehmite precursor. The bands at 1469 cm1 and 1579 cm1 are assigned to the stretching mode of adsorbed acetate CH3COO groups due to chelating or bridging type acetate coordination with Al(III) [18,19,21]. The presence of a small hump at 1635 cm1 is assigned to the presence of hydroxyl groups in boehmite precursor [19]. The band at 1635 cm1 (assigned to hydroxyl group) still persists in the samples even after calcination in air. The characteristic absorption peaks (Fig. 6(c)) in the region 500– 750 cm1 are assigned to the vibrations of AlO6 and a shoulder observed at 886 cm1 is thought to be due to the vibration of gAlO4 [19]. Thus, the g-Al2O3 powder produced by the calcination of boehmite precursor is found to contain both tetrahedral and octahedral Al–O coordination. The absence of bands at 1276 and 1415 cm1 i.e., the bands characteristic of free acetic acid in the material [21], confirms its absence in the material. TEM studies have been undertaken to understand the effects of acetic acid on the morphology of boehmite and of the calcined gAl2O3 powder as well. Fig. 7(a) gives the TEM image of the as-
Fig. 7. TEM images of boehmite and g-alumina powders (bar = 20 nm). (a) AA/AIP = 0, (b) AA/AIP = 0.035, (c) AA/AIP = 0.5, (d) AA/AIP = 0.035 calcined at 500 8C.
118
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
prepared boehmite precursor that shows the presence of rod shaped porous particles (consisting of aggregates of crystallites) with different lengths (40–50 nm) having a diameter of 1–2 nm. However, when the AA/AIP ratio is kept at 0.035 during processing, the as-dried boehmite precursor shows material that consists of interconnected networks of thinner porous rods having diameter smaller than that of the powder when the AA/AIP ratio is 0 (Fig. 7(b)). With the further increase of AA/AIP ratio, the boehmite precursor consists of a wormlike microstructure (Fig. 7(c)) with still thinner elongated interconnected networks of nanometer sized smaller rods with variation in length. The gAl2O3 powder, however, shows the material with the presence of rods, as can be seen from Fig. 7(d). The diameter of the particles is found to be in nanometer range (1.5–2.5 nm) with uniform lengths of 20–25 nm. All these observations are consistent with those reported by Hochepied et al. [22] that the alignments of nanoparticles lead to the building of polycrystalline fibers with diameters of 3–8 nm and lengths of about 100 nm. 3.4. The formation of high surface areas in boehmite Careful analyses of FT-IR and TEM together postulate that the morphology of boehmite precursor can be controlled by the adsorption of acetate groups on as-dried boehmite precursor during preparation. One possible mechanism could be that the boehmite particles nucleated by reactions (2) and (3) gradually ripen to form the boehmite nano-crystals during the aging process. Then the reaction-limited aggregation of these particles via specific sites [23] (where no adsorption of acetic acid takes place) can occur to minimize the interfacial energy so as to give a particular morphology to boehmite aggregates. High-energy faces of the boehmite particles where the CH3COO groups are selectively adsorbed [24] are not available for the reaction-limited aggregation process during aging. The interaction between CH3COO groups and those faces possessing the highest energy would be relatively stronger and the growth rates of these facets are decelerated most. In other words, CH3COO ions have a balancing effect on the growth rates of different faces in such a way that the desired rod shaped morphology is obtained in the boehmite precursor. The same morphology is retained in the g-Al2O3 powder by topotactic reactions [25] with the g-Al2O3 powder. Similar observations have been recently been reported by Zhang et al. [26] for the synthesis of ZnO micro-crystals, wherein the growth habit is controlled by its intrinsic structure and external conditions like temperature, pH and capping agent. Such observations are also separately reported by Khollam et al. [27] for the Ba1xSrxTiO3 powder in achieving the star shaped morphology because of selective adsorption of C2O42 groups on the faces of the as-dried crystalline precursor Ba1xSrxTiO(C2O4)2. Therefore, the role of acetic acid during processing in the current investigation is found to be crucial. Further characterization of g-Al2O3 powder was carried out by measuring its PL spectra with excitation at 325 nm at room temperature to confirm the presence of oxygen vacancies in these materials. The spectra are depicted in Fig. 8. The commercial gAl2O3 obtained from catapal-B (Fig. 8(a)) showed similar features in PL spectrum to those of the g-Al2O3 powders (Fig. 8(b)) prepared without using acetic acid during preparation. A strong blue emission at 437.6 nm is observed in both the samples indicating the presence of a large amount of F+ centers [27] in the g-Al2O3 powder. The low intensity broad band centered at 800 nm may be attributed to the presence of defects other than the F+ centers in the samples [28,29]. However, two additional sharp peaks are observed at 585.8 nm and 616.5 nm in the g-Al2O3 powder (Fig. 8(c)) prepared using acetic acid which are absent in the g-
Fig. 8. PL spectra of g-alumina powder obtained from (a) catapal-B, (b) w/o acetic acid, (c) with acetic acid.
Al2O3 powders prepared without using acetic acid during preparation. The doublet appearing at 585.8 nm and 616.5 nm is attributed to the presence of a small amount of u-Al2O3 formed during calcination at 500 8C in air. The doublet is shifted to lower wavelength side as compared to the reported data [30] due to the stress in the material. This observation is consistent with earlier DTA/TGA, XRD and TEM results that the nanocrystalline g-Al2O3 with high surface area may transform to u-Al2O3 phase in small amounts at calcination temperatures close to that of the present work. The phase percentage could be below the detection limit of XRD, as no peaks corresponding to the u-Al2O3 are seen in the XRD pattern. The DTA/TGA also showed a broad exothermic peak due to crystallization of the g-Al2O3 and metastable phases. Therefore, it is possible that the g-Al2O3 powder might contain small amounts of the u-Al2O3 below the detection level of XRD. 3.5. Catalytic activity on the synthesized g-Al2O3 with different surface areas The aluminas synthesized by changing the mole ratio of AA/AIP and by showing different surface areas are further characterized by NH3-TPD and tested for the activities on methanol dehydration to DME. Fig. 9 shows the NH3-TPD profiles of g-alumina powder; quantitative results obtained for the same are summarized in Table 5. It is evident from Table 5 that the amount of ammonia desorbed from the g-alumina powder is higher (1.12 mmol/g) when the AA/AIP ratio is maintained at 0.50 than that of 0.82 mmol/g observed for the AA/AIP ratio 0.0. This amount gradually increases with the increase of AA/AIP ratio due to the increasing surface area. The g-alumina powder derived from the commercial catapal-B boehmite sample, however, gave a lower value for the desorbed ammonia 0.65 mmol/g. The TPD patterns (Fig. 9) of the g-alumina powder showed two intense peaks in the range of 100–800 8C. The first peak (I) corresponds to acid sites which is attributable to the removal of adsorbed NH3 on the catalyst surface. The second peak (II) is due to the slow dehydroxylation process [2] of surface hydroxyl groups on g-Al2O3. The first peak can be assigned to the NH3 desorbed from acid sites. The intensity of the peaks is found to increase significantly by keeping the AA/AIP ratio 0.5 during preparation. From these results, one can conclude that the g-alumina powder reported in the present study possesses higher surface acidic sites as compared to the g-alumina powder obtained from the
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Fig. 9. NH3-profiles of g-alumina obtained from (a) catapal-B, (b) AA/AIP = 0, (c) AA/ AIP = 0.035, (d) AA/AIP = 0.1 (e) AA/AIP = 0.5.
Table 5 Summary of surface areas and acidic sites measured by NH3-TPD on g-alumina with the variation of AA/AIP ratio Molar ratio of acetic acid to aluminum (AA/AIP)
BET surface area (m2/g)
Acidic sites (mmol NH3/g)
catapal-B 0.0 0.035 0.1 0.5
220.0 306.5 350.5 412.8 437.8
0.6527 0.8214 0.8598 1.0000 1.1192
Synthetic conditions: AIP/H2O/IPA = 1/6/25 and calcined at 500 8C for 5 h.
commercial catapal-B boehmite and the density of sites is increased with the increase of AA/AIP molar ratio. The presence of acetic acid during the preparation has helped to enhance the concentration of acidic sites in these materials. The use of acetic acid during preparation is not only helpful to improve surface area of boehmite/g-Al2O3 powder (Tables 1 and 3) but also to enhance the amount of surface acidic sites of the g-Al2O3 powders. Thus, it is possible to control easily the amount of surface acidic sites as well as pore sizes, their distribution and the crystallinity of boehmite/g-Al2O3 powder by use of acetic acid during preparation. The g-Al2O3 catalyst obtained by keeping the AA/AIP ratio 0.5 and the H2O/AIP ratio 6.0 during processing is used for the methanol to DME conversion. The performance of this catalysts is compared with that of the g-Al2O3 catalyst obtained from the commercial boehmite i.e. catapal-B. The results on the catalyst indicate 80% conversion and selectivity to DME of 99.9%, almost approaching the equilibrium conversion and the maximum selectivity. The temperature at which the equilibrium conversion of methanol is reached above 320 8C, is for the g-Al2O3 (obtained from catapal-B) and it is also found to be above 320 8C for the catalyst prepared without using acetic acid. However, the catalyst prepared using acetic acid during synthesis needed a temperature above 300 8C to get the equilibrium conversion. Thus, a better performance at lower temperature was obtained on our catalyst prepared using acetic acid than on the g-Al2O3 catalyst derived from the commercially available catapal-B boehmite sample or on the sample prepared without acetic acid. Furthermore, the variations in catalytic activity with surface area and the amount of acidic sites are shown in Fig. 10. In general, the catalytic activity in methanol dehydration strongly depends on the surface acidic properties, such as the total number of acidic sites and their strength. It is apparent that the number of acidic
119
Fig. 10. Correlation of surface area and amount of acidic sites with the catalytic activity of methanol dehydration to DME on the g-alumina obtained at different mole ratios of AA/AIP (filled symbols) and catapal-B (hollow symbols).
sites showing almost the same acidic strength can be precisely controlled by appropriately selecting the synthetic method during alumina preparation (Table 5 and Fig. 9). It is also interesting to see a good correlation between the catalytic activity, the surface area and the amount of acidic sites. Yoo et al. [31] in their recent study on DME synthesis have reported a good correlation between the catalytic activity and the number of weak acidic sites. The present results, reported in Table 5, also agree with this observation, giving the importance of the amount of acidic sites in the dehydration reaction. The acidic sites are also less prone to deactivation during the reaction. The methanol dehydration activity, expressed in terms of the temperature required to obtain 50% conversion, is correlated with the surface area of alumina in Fig. 10. With increasing surface area of g-Al2O3, the required temperature to reach 50% conversion of methanol is found to decrease from 325 8C to 305 8C due to the increased number of acidic sites which are active for methanol dehydration. These results also suggest that a high throughput yield can be obtained at high space velocity on the high surface area alumina without a complicated modification process. 4. Conclusion Nano-sized high surface area boehmite particles can be prepared by adopting the sol–gel method, during which suitable ratios of AA/AIP and H2O/AIP and suitable aging times are required. Addition of acetic acid enhances the textural and structural characteristics of boehmite. Thermal treatment of the boehmite at 550 8C gives a high surface area g-Al2O3 powder having controlled porosity and pore size distribution. The acetic acid adsorbed on the surface plays a crucial role in controlling the morphology of the boehmite precursor as well as that of the gAl2O3 powder. The structure–property relationship observed in the present study is useful to make tailor made g-Al2O3 with high surface acidity. The alumina thus prepared is highly active and selective in the dehydration of methanol, giving an equilibrium conversion and close to 100% selectivity, under the selected reaction conditions. The number of acidic sites on alumina is the crucial factor for catalytic activity of methanol dehydration. A good correlation is obtained between the number of acid sites and the temperature required to reach 50% conversion. With the increase of surface area of g-Al2O3, the temperature required for 50% conversion of methanol decreased due to the increase of available acidic sites.
120
S.-M. Kim et al. / Applied Catalysis A: General 348 (2008) 113–120
Acknowledgements This research was supported by a grant from Carbon Dioxide Reduction & Sequestration Center (CDRS), one of 21st Century Frontier Programs funded by the Ministry of Science and Technology, Republic of Korea. Dr. H.S. Potdar is thankful to KOSEF for giving a Brain-Pool fellowship and to the Director, National Chemical Laboratory, Pune 411008, for granting EOL. We are thankful to Dr. K.N. Kim for doing PL measurements of our samples. References [1] T.F. Baumann, A.E. Gash, S.C. Chin, A.M. Sawvel, R.S. Maxwell, J.H. Satcher Jr., Chem. Mater. 17 (2005) 395–401. [2] P. Padmaja, P.K. Pillai, K.G.K. Warrier, M. Padmanabhan, J. Porous Mater. 11 (2004) 147–155. [3] D. Panias, A. Krestou, Powder Technol. 175 (2007) 163–173. [4] E. Elaloui, A.C. Pierre, G.M. Pajonk, J. Catal. 166 (1997) 340–346. [5] F. Yaripour, F. Baghaei, I. Schmidt, J. Perregaard, Catal. Commun. 6 (2005) 147– 152. [6] S.M. Kim, Y.J. Lee, K.W. Jun, J.Y. Park, H.S. Potdar, Mater. Chem. Phys. 104 (2007) 56–61. [7] H.S. Potdar, K.W. Jun, J.W. Bae, S.M. Kim, Y.J. Lee, Appl. Catal. A 321 (2007) 109–116. [8] T.A. Semelsberger, R.L. Borup, H.L. Greene, J. Power Sources 156 (2006) 497–511. [9] J.D. Holladay, Y. Wang, E. Jones, Chem. Rev. 104 (2004) 4767–4790.
[10] S. Lee, A. Sardesai, Top. Catal. 32 (2005) 197–207. [11] K. Okada, A. Tanaka, S. Hayashi, K. Daimon, N. Otsuka, J. Mater. Res. 9 (1994) 1709–1713. [12] T. Fukui, M. Hori, J. Mater. Sci. 31 (1996) 3245–3248. [13] B.E. Yoldas, J. Mater. Sci. 10 (1975) 1856–1860. [14] L. Wang, Y. Qi, Y. Wei, D. Fang, S. Meng, Z. Liu, Catal. Lett. 106 (2006) 61–66. [15] T. Takeguchi, K.I. Yanagisawa, T. Inui, M. Inoue, Appl. Catal. A 192 (2000) 201–209. [16] E.P. Barett, L.G. Joyner, P.P. Halender, J. Am. Chem. Soc. 73 (1951) 373–380. [17] J.A. Wang, X. Bokhimi, A. Morales, O. Novaro, T. Lopez, R. Gomez, J. Phys. Chem. B 103 (1999) 299–303. [18] S.B. Deshpande, H.S. Potdar, Y.B. Khollam, K.R. Patil, R. Pasricha, N.E. Jacob, Mater. Chem. Phys. 97 (2006) 207–212. [19] G.K. Priya, P. Padmaja, K.G.K. Warrier, A.D. Damodaran, G. Aruldhas, J. Mater. Sci. Lett. 16 (1997) 1584–1587. [20] S. Music, D. Dragcevic, S. Popovic, Mater. Lett. 40 (1999) 269–274. ¨ hman, Geochim. Cosmochim. Acta 62 (1998) 3657– [21] P. Persson, M. Karlsson, L.-O. O 3668. [22] J.F. Hochepied, P. Nortier, Powder Technol. 128 (2002) 268–275. [23] Y. Zhou, M. Antonietti, J. Am. Chem. Soc. 125 (2003) 14960–14961. [24] X.C. Jiang, Series of Selected Papers from Chun-Tsung Scholars, Peking University, 2003. [25] P. Alphonse, M. Courty, Thermochim. Acta 425 (2005) 75–89. [26] H. Zhang, D. Yang, D. Li, X. Ma, S. Li, D. Que, Cryst. Growth Des. 5 (2005) 547–550. [27] Y.B. Khollam, H.S. Potdar, S.B. Deshpande, A.B. Gaikwad, Mater. Chem. Phys. 97 (2006) 295–300. [28] S. Liu, L. Zhang, L. An, W. Fei, H. Heinrich, J. Am. Ceram. Soc. 88 (2005) 2559–2563. [29] X. Sun, F. Xu, Z. Li, W. Zhang, J. Lumin. 121 (2006) 588–594. [30] X. Peng, D.R. Clarke, F. Wang, Oxid. Met. 60 (2003) 225–240. [31] K.S. Yoo, J.H. Kim, M.J. Park, S.J. Kim, O.S. Joo, K.J. Jung, Appl. Catal. A 330 (2007) 57–62.