Available online at www.sciencedirect.

com

Applied Catalysis A: General 333 (2007) 245253 www.elsevier.com/locate/apcata

Effect of framework Si/Al ratio and extra-framework aluminum on the catalytic activity of Y zeolite
Bin Xu a, Silvia Bordiga b, Roel Prins a, Jeroen A. van Bokhoven a,*
b a ETH Zurich, Institute for Chemical and Bioengineering, 8093 Zurich, Switzerland ` di Torino, Via P. Giuria 7, I-10125 Torino, Italy Centre of Excellence NIS, Dipartimento di Chimica I.F.M., Universita

Received 14 May 2007; received in revised form 12 September 2007; accepted 17 September 2007 Available online 19 September 2007

Abstract The effect of the Si/Al ratio of the framework on the intrinsic activity of Brnsted acid sites in Y zeolite was investigated by monomolecular cracking of propane and physicochemical characterization. Samples with different Si/Al ratios were obtained by synthesis and by post-synthesis treatments. Two samples with Si/Al ratios of 3.3 and 3.6 exhibited characteristic features similar to those of commercial Y zeolite with a Si/Al ratio of 2.6, but they were more stable. An ultrastable Y (HUSY) was treated with ammonia and ethylenediaminetetraacetic acid (EDTA) to remove extra-framework aluminum (EFAl). The rates per gram increased with Si/Al ratio, but the activation energies did not depend on the Si/Al ratio or on the presence of EFAl. This implies that the intrinsic activity of the Brnsted acid sites that participate in the reaction is identical in all the samples. The enhanced activity of the samples with Si/Al ratios up to 5.0 is attributed to the creation of more isolated Brnsted acid sites, which are active in this demanding reaction. The removal of EFAl did not affect the intrinsic catalytic activity but increased the number of catalytically active framework Brnsted acid sites. # 2007 Elsevier B.V. All rights reserved.
Keywords: Y zeolite; High-silica Y; USY; Monomolecular cracking; Acidity; Steaming; Extra-framework aluminum

1. Introduction Y zeolite (FAU) is the most widely used zeolite in the reneries [1]. Its large pore openings and high surface area make it the catalyst of choice in uid catalytic cracking, one of the most important reactions in oil rening. However, the thermal and hydrothermal stability of Y zeolite is limited. Even at room temperature, the structure of acidic Y zeolite partially collapses when exposed to moisture [27]. It is well known that the higher the Si/Al ratio, the more stable the structure of the zeolite. Therefore, a thermally stable Y, the so-called ultrastable Y zeolite (USY), was developed by dealumination of the framework by the controlled addition of steam [8]. The reduced ion-exchange capacity and a smaller unit cell are indications of the removal of aluminum from the framework [9]. At the same time, extra-framework phases form [912].

* Corresponding author. E-mail address: j.a.vanbokhoven@chem.ethz.ch (J.A. van Bokhoven). 0926-860X/$ see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2007.09.018

Besides its higher thermal stability, HUSY exhibited a much higher catalytic activity than the non-activated samples [1318]. Many explanations have been put forward. First, the next-nearest neighbor effect [1922] suggests that strong Brnsted acidity occurs only on acid sites, of which the aluminum atom has no aluminum atoms as its next-nearest neighbor. Second, extra-framework aluminum has been suggested to modify the catalytic activity because it stabilizes the lattice [23], forms active phases [24], or has a synergistic effect on nearby Brnsted acid sites [23,24]. Sites of enhanced strength have been attributed to a hydroxyl band located at 3600 cm1 in the infra-red spectra of HUSY [13,2527]. Finally, mesopores are generated during steam activation that has been shown to alleviate diffusion limitation in classical cracking [28]. Cracking and isomerization of hydrocarbons are catalyzed by the Brnsted acid sites in the zeolites [29], and the catalytic activity increases with the number of Brnsted acid sites [13,18,25], which is related to the number of framework aluminum atoms. Alkane cracking is a model reaction to probe the nature of Brnsted acid sites in zeolites. Depending on the

246

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

reaction conditions, the two reaction mechanisms are: monomolecular and bimolecular [30,31]. The monomolecular pathway involves the protonation of an alkane by the Brnsted acid site to form a carbocation, which subsequently undergoes cracking or dehydrogenation [32]. The bimolecular pathway involves hydride transfer between an alkane and an adsorbed alkoxide species, followed by isomerization and/or b-scission [32,33]. The hydride transfer does not involve a direct interaction between alkane and the Brnsted acid site. The monomolecular reaction, however, involves proton transfer from the zeolite to the alkane and depends on the nature of the Brnsted acid sites. Monomolecular cracking is a rst-order reaction, in which the protonation of the alkane is the ratelimiting step, when catalyzed by zeolites [34,35]. Therefore, the intrinsic activity of this reaction is a measure of the strength of the Brnsted acid sites in the catalyst. The intrinsic activation energy (Eact) equals the apparent activation energy (Eapp) minus the heat of reactant adsorption (DHads) [36]. The activation energy is indicative of the nature of the active site; the activity alone does not give insight into the intrinsic activity of the catalytic active sites. In earlier studies, the intrinsic reactivity of the zeolitic Brnsted acid sites was determined by the monomolecular cracking of alkanes [34,35,3741,86]. The general conclusion was that the variation in the activity is due mainly to the different sorption of the reactant and that the intrinsic activity of the Brnsted acid sites is identical and does not depend on the zeolite type. A compensation relation between the apparent activation energy and the natural logarithm of the preexponential term was related to the compensation between the heat and entropy of adsorption of alkanes into the pores of the zeolites. In this study, we determined the effect of framework Si/Al ratio on the intrinsic activity of Y zeolite. Samples with different Si/Al ratios of the framework were obtained by synthesis and by post-synthesis treatments. The Brnsted acid sites in these samples were investigated by the monomolecular cracking of propane and various physicochemical characterization techniques. Propane was chosen as a probe molecule, because the cracking and dehydrogenation pathways can be easily distinguished, without consecutive reactions to occur.
Table 1 Sample treatments and nomenclature Sample Y(2.6) HY(2.6) Y(3.3) HY(3.3) Y(3.6) HY(3.6) HUSY H(NH3)USY HUSY-NaEDTA Starting material Y(2.6) Y(3.3) Y(3.6) HUSY HUSY Treatment Calcined at 823 K in air Calcined at 823 K in air Calcined at 823 K in air At 423 K NH3 admitted, cool in NH3 Washed with 0.1 M Na2H2-EDTA at 335 K for 24 h; ion-exchanged with 1 M NH4NO3 at 353 K for 24 h

2. Experimental 2.1. Samples Table 1 lists the samples and their nomenclature. The samples were referred to as Y(N) and HY(N), which correspond to the NH4 and the proton form, respectively. N is the Si/Al ratio determined by atomic absorption spectroscopy (AAS). Two high-silica Y zeolites were synthesized by the hydrothermal method with 15-crown-5 ether [4245]. Y(3.6) was prepared from 1.49 g sodium hydroxide (Aldrich, p.a., 99.998%), 0.57 g sodium uoride (Sigma, p.a., 99%), and n, technical grade, 3.03 g sodium aluminate (Riedel-de Hae 54 wt% Al2O3, 41 wt% Na2O), which were dissolved in 27 mL double deionized water [42,44]. After stirring for 5 min, 3.53 g 15-crown-5 (Aldrich, 98%) and 24.11 g silica sol (Ludox AS40, 40 wt% SiO2) were added. After stirring continuously for 2 h, the mixture was transferred to an autoclave reactor to crystallize at 385 K for 7 days. A second sample (Y(3.3)) was made without adding sodium uoride [43]. 1.04 g sodium hydroxide was dissolved in 23 mL double deionized water and 3.02 g sodium aluminate, 2.42 g 15-crown-5, and 23.89 g silica sol were stirred into the mixture. The resulting gel was continuously stirred for 24 h. The crystallization was carried out in an autoclave reactor at 385 K for 7 days. Both assynthesized samples were calcined in air at 823 K for 8 h to remove the template and then converted to the NH4-form by triple ion-exchange with 1 M NH4NO3 solutions at 353 K. The residual Na content determined by AAS was less than 0.05 wt%. Some of both materials was calcined for 6 h at 823 K at a heating rate of 2 K/min and exposed to moisture after cooling to room temperature to form HY(3.6) and HY(3.3). To investigate the effect of dealumination and extra-framework aluminum, a steamed commercial sample, HUSY (LZY-84, obtained from BP), was used. Two treatments were applied to this sample. One was the adsorption of ammonia, which converts part of the octahedrally-coordinated aluminum of HUSY to tetrahedrally-coordinated aluminum [5]. A ow of 10% NH3 in He was adsorbed on HUSY at 423 K and the sample was cooled to room temperature in this ow to form H(NH3)USY. The other treatment was performed with ethylenediaminetetraacetic acid (EDTA), which removes extra-framework aluminum from the sample [46]. This sample, HUSY-NaEDTA, was obtained by reuxing HUSY with 0.1 M Na2H2-EDTA at 335 K for 24 h and then ion-exchanging with 1 M NH4NO3 at 353 K for 24 h [47]. Two high-aluminum samples, Y(2.6) and HY(2.6), are also listed in Table 1. Their structures and activities are described elsewhere [6]. 2.2. Characterization Nitrogen physisorption measurements were performed at liquid-nitrogen temperature with a Micromeritics ASAP 2000 apparatus. Prior to the measurements, all the samples were degassed at 723 K overnight. The surface area was determined by the BET method and the micropore volume was determined from the intercept of the linear part of the t-plot. XRD patterns

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

247

were obtained by spinning capillaries (diameter 0.3 mm), containing the samples, on a Siemens D-5000 powder X-ray diffractometer (Cu Ka radiation) with BraggBrentano geometry. The aluminum content of the framework was determined using the relationship between the unit cell size and the number of aluminum atoms in the unit cell [48,49]. MAS NMR spectra were measured at 104.287 MHz for 27Al and at 79.49 MHz for 29Si on a Bruker Avance AMX-400 spectrometer at a spinning rate of 12 kHz using a 4 mm probe head. Prior to the MAS NMR measurements, the samples were hydrated by placing them in an atmosphere saturated with moisture from a solution of 1 M NH4NO3. In 27Al MAS NMR, the single pulse length was p/6. In 29Si MAS NMR, a highpower decoupling pulse sequence (HPDEC) and a relaxation delay of 6 s were used. The 27Al chemical shifts were referenced to (NH4)Al(SO4)212H2O and the 29Si chemical shifts to octakis-(trimethyl siloxy)silsesquioxane. For the FT infra-red measurements, self-supporting pellets (24 mg) were placed in a laboratory-constructed heated cell equipped with KBr windows and connected to a metal vacuum line capable of attaining a residual vacuum of <103 Pa. The spectra were recorded at room temperature at a resolution of 4 cm1 on a Galaxy 6020 FT infra-red spectrophotometer equipped with an MCT detector. Prior to measurement, the samples were activated for one h under vacuum at 773 K at a rate of 2 K/min. The number of Brnsted acid sites was determined by two methods. Infra-red spectroscopy after pyridine adsorption and thermogravimetric analysis (TGA) with decomposition of ipropylamine. The infra-red measurements were performed in a ow cell with a Bio-Rad Excalibur FTS 3000 infra-red spectrophotometer. After activation at 773 K, the samples were treated for 10 min at 393 K in a ow of 20 mL/min of He, saturated at 273 K with pyridine, at a vapor pressure of 550 Pa; infra-red spectra were recorded after 1 h at 423 K. The number of Brnsted acid sites was determined by means of the integrated molar extinction coefcient of 1.67 cm/mmol for the infra-red absorption band of pyridinium at 1545 cm1 [50]. For the TGA measurements, laboratory-constructed equipment was used to activate the samples. A sample of approximately
Table 2 Characteristics of the samples Sample Y(2.6)e HY(2.6)e Y(3.3) HY(3.3) Y(3.6) HY(3.6) HUSY H(NH3)USY HUSY-NaEDTA Si/Al AAS 2.6 2.6 3.3 3.3 3.6 3.6 2.5 2.5 5.0 Surface area (m2/g) 810 450 670 540 650 600 700 640 730 m-Pore volume (cm3/g) 0.32 0.18 0.27 0.22 0.26 0.25 0.27 0.24 0.27

2040 mg was activated at P < 104 Pa overnight by heating to 723 K at a rate of 2 K/min. The sample was then exposed to i-propylamine at 473 K for 2 h and cooled to room temperature. It was evacuated for 2 h to remove the physically adsorbed i-propylamine. The decomposition experiment was performed at a heating rate of 10 K/min and a ow of 20 mL/min Ar in a Mettler Toledo TGA/SDTA851e instrument, after which a mass spectrometry (QME 200, Balzers) was performed. The amount of decomposed i-propylamine was obtained from the mass changes in the TGA curve between 573 and 650 K. 2.3. Monomolecular cracking of propane The catalytic conditions were similar to those reported elsewhere [6]. The catalytic reaction was performed in a setup with six parallel-ow, tubular quartz reactors, with 10% propane in argon (10 kPa) between 675 and 875 K. Traces of H2O were present in the feed but did not affect the catalytic activity. The Arrhenius plots were determined in this temperature range. During the measurements, the temperature was varied randomly, which did not affect the outcome. The reproducibility was better than 3 kJ/mol. The rates of reaction at 823 K are reported. The ow rates varied from 5 to 50 mL/ min and the weight of the catalyst varied from 50 to 200 mg. Prior to the reaction, the catalysts were activated for 1 h in the reactor in 10% oxygen in argon at 878 K at a heating rate of 2 K/min. Lower temperatures during the pretreatment did not affect the catalytic behavior, thus indicating that the structure and activity of the sample were not altered by the high temperature during pretreatment. The products in the efuent gas stream were analyzed with an on-line Agilent 3000 MicroGC gas chromatograph. The apparent activation energies were determined from the slopes of the Arrhenius plots. The reaction conditions were chosen so that the molar ratios of methane and ethene as well as of H2 and propene were unity. Conversion was kept below 10% and no hydrocarbons with more than three atoms were detected. The conversion to the cracking products methane and ethene is discussed in this paper.

%Td Al 27 Al MAS NMRa 100 93 100 94 100 93 40 89 93

Unit cell, ) a0 (A 24.720 24.643 24.660 24.628 24.580 24.524 24.537 24.545 24.517

FWAl/u.c.b 52 43 45 42 37 31 32 33 30

Brnsted acid sitesc (mmol/g) 1.10 0.20 0.90 0.29 0.71 0.40 0.30 0.36 0.54

Brnsted acid sitesd (mmol/g) 1.95 1.0 1.76 n.d. 1.54 n.d. 1.06 1.13 1.40

n.d.: not determined. a Only the narrow peak at about 60 ppm is considered. b Determined from the unit cell size. c Pyridine infra-red after desorption at 423 K. d Determined by the decomposition of i-propylamine. e Data taken from ref. [6].

248

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

3. Results 3.1. Characterization Table 2 (column 2) gives the Si/Al ratios of all the samples determined by AAS. The high-silica Y zeolites (3.3) and Y(3.6) had Si/Al ratios of 3.3 and 3.6, respectively. The Si/Al ratio of HUSY was 2.5, which did not change with ammonia treatment (H(NH3)USY). The sample treated with Na2H2EDTA had a Si/ Al ratio of 5.0, which indicates that EDTA has removed aluminum from the sample. Table 2 (columns 3 and 4) presents the BET surface areas and the micropore volumes after degassing at 723 K. Zeolites Y(3.3) and Y(3.6) had a very similar surface area (670 and 650 m2/g, respectively) and micropore volume (0.27 and 0.26 cm3/g, respectively). The values were lower than those of Y(2.6) (810 m2/g and 0.31 cm3/ g). The surface area and micropore volume of the calcined samples decreased in the order of increasing aluminum content: HY(3.6) > HY(3.3) > HY(2.6). HUSY and HUSY-NaEDTA had similar surface areas of 700 and 730 m2/g, respectively, and the same micropore volume of 0.27 cm3/g. The ammoniatreated sample H(NH3)USY had a smaller surface area (640 m2/g) and micropore volume (0.24 cm3/g) [51]. Fig. 1 shows the 27Al MAS NMR spectra of the high-silica Y zeolite, HUSY, and of the treated samples. These spectra were recorded after equilibration in a controlled wet environment. The spectra of Y(3.3) and Y(3.6) showed a single peak at 60 ppm, which is assigned to tetrahedrally-coordinated framework aluminum. The spectra of the calcined samples HY(3.3) and HY(3.6) showed two peaks, one at 60 ppm, attributed to tetrahedrally-coordinated aluminum and one at 0 ppm, assigned to octahedrally-coordinated aluminum. There was about 6 to 7% octahedrally-coordinated aluminum (Table 2, column 5) in these samples, similar to that in HY(2.6) [6]. The peak for tetrahedrally-coordinated aluminum was broad, showing disorder in the structure. The spectrum of HUSY consisted of three parts: a broad tetrahedral peak at 60 ppm, a distorted tretrahedral peak from 20 to 40 ppm [10], and an octahedral peak at 0 ppm. The spectrum of H(NH3)USY showed a broad tetrahedral peak at 60 ppm and a weak

Fig. 2. XRD patterns of high silica Y, HUSY, and the treated samples.

Fig. 1.

27

Al MAS NMR of high silica Y, HUSY, and the treated samples.

octahedral signal at 0 ppm. The spectrum of HUSY-NaEDTA showed an asymmetric tetrahedral peak and a small broad peak at 2040 ppm. Part of the aluminum that contributes to the tetrahedrally coordinated aluminum could be extra framework. The tetrahedral peaks in the spectra of HUSY, H(NH3)USY, and HUSY-NaEDTA were broad, indicating a large spread in average TOT angles. There was 40% tetrahedral aluminum in HUSY, corresponding to the peak at 60 ppm (Table 2, column 5), which increased to 89% after ammonia treatment. After the EDTA treatment, 93% of the total aluminum was tetrahedral aluminum. Fig. 2 shows the XRD patterns of zeolite Y(2.6), the highsilica Y zeolites, HUSY, and the treated samples. The samples were not pretreated before the measurement. All the samples showed similar patterns and high crystallinity, in agreement with N2 physisorption. The unit cell parameters and the corresponding number of framework aluminum atoms per unit cell are given in Table 2 (columns 6 and 7). The unit cell decreased in size with dealumination, as did the number of framework aluminum atoms per unit cell. Thus, the number of framework aluminum atoms per unit cell, determined from the XRD patterns and AAS, is more or less the same in samples with 100% tetrahedral aluminum. The calcined samples had a signicantly lower number of framework aluminum atoms per unit cell, in agreement with 27Al MAS NMR, which showed octahedrally-coordinated aluminum. The 29Si MAS NMR data are presented in Figs. 3 and 4 and Table 3. The spectra of the high-silica Y zeolites showed that the peak at about 97 ppm was less intense than the signals at about 103 ppm and 107 ppm compared to the corresponding signals of Y(2.6). The spectra are assigned to (Q4(2Al)), (Q4(1Al)), and (Q4(0Al)) species, where Q4 represents a Si atom connected to four T atoms by a bridging oxygen; nAl indicates the number of aluminum atoms in the second coordination shell. Analysis of the spectra of the high-silica Y zeolites gave a Si/Al ratio of 3.2 for Y(3.3) and 4.3 for Y(3.6) (Table 3, column 3), which follow the trend of the bulk values determined from atomic absorption spectroscopy. The Si/Al ratio in HUSY was 4.8 and in H(NH3)USY 5.1; after the EDTA treatment of HUSY, the ratio increased to 5.5. HUSY probably

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

249

Fig. 3.

29

Si MAS NMR of high silica Y zeolite samples.

Fig. 4.

29

Si MAS NMR of HUSY and the treated samples.

contains amorphous silica-alumina phases that interfered with the 29Si MAS NMR and that were washed out during the EDTA treatment. Table 3 (column 3) also shows the number of framework aluminum atoms per unit cell determined from the 29 Si MAS NMR. They follow the same trend as predicted by XRD and by the combination of AAS and 27Al MAS NMR. Fig. 5 shows the infra-red spectra of Y(2.6), Y(3.3), and Y(3.6), measured at room temperature after vacuum activation
Table 3 29 Si MAS NMR data of the samples Sample Y(2.6) HY(2.6)b Y(3.3) HY(3.3) Y(3.6) HY(3.6) HUSY H(NH3)USY HUSY-NaEDTA
a b b

at 773 K. To compare the intensity of the hydroxyl peaks of the samples, all the spectra were normalized in intensity by giving the overtones and the combination modes of the zeolitic framework at about 17002000 cm1 the same intensity. The spectra of Y(2.6) and its calcined sample (HY(2.6)) were taken from reference 6. After the in situ removal of ammonia, the spectrum of Y(2.6) exhibited bands at 3644 and 3545 cm1, assigned to Brnsted acid sites in the super cages (HF, high frequency) and Brnsted acid sites in the sodalite cages (LF, low frequency), respectively [52]. The small band at 3745 cm1 is assigned to the silanol vibration. The HF peak shifted to a lower wavenumber in the spectra of Y(3.3) (3636 cm1) and Y(3.6) (3630 cm1); a higher Si/Al ratio caused a band shift [53,54]. The intensity of the HF and LF bands decreased with increasing Si/Al ratio, which is in line with the number of Brnsted acid sites determined. The spectra after the in-situ temperature treatment of HY(2.6), HY(3.3), and HY(3.6) showed much lower intensity of the HF and LF bands and a higher intensity of the silanol band at 3745 cm1. Compared to HY(2.6), the spectra of HY(3.3) and HY(3.6) showed more intense HF and LF bands and a less intense silanol band. Fig. 6 gives the infra-red spectra of HUSY, H(NH3)USY, and HUSYNaEDTA after an in-situ heat treatment. The spectrum of HUSY is more complex than the spectrum of untreated Y zeolite. In addition to the bands assigned to silanols (3745 cm1), HF (3625 cm1), and LF (3560), two shoulders at 3700 and 3525 cm1 and two distinct bands at 3670 and 3600 cm1 are observed. The shoulder at 3700 cm1 is ascribed to internal silanols and the peak at 3670 cm1 to partial extra-framework aluminum species [55], both of which form during steam activation. The distinct band at 3600 cm1 has been ascribed to Brnsted acid sites in interaction with EFAl and has been suggested to enhance catalytic activity (HF) [56]. After activation at 773 K, the spectra of H(NH3)USY and HUSY-NaEDTA showed an increase in the overall intensity in the hydroxyl region. The HF and LF bands increased and those at 3600 and 3670 cm1 decreased in intensity. The lowfrequency shoulder on the LF peak virtually disappeared and the peak at 3600 cm1 (HF) is barely distinguishable. The number of Brnsted acid sites, determined from infrared spectra after the adsorption and desorption of pyridine at 423 K and the decomposition of i-propylamine (Table 2, columns 8 and 9), showed that the concentration of Brnsted

Si/Ala 2.4 2.6 3.2 3.6 4.3 5.3 5.1 4.8 5.5

FWAl/u.c. 56 53 46 42 36 30 31 33 29

Si(0Al) (%) 4 11 11 20 30 37 51 42 46

Si(1Al) (%) 42 35 58 47 49 50 35 33 38

Si(2Al) (%) 37 42 24 33 20 13 14 25 13

Si(3Al) (%) 17 12 7 0 1 0 0 0 3

From 29Si MAS NMR. From ref. [6].

250

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

Fig. 5. FTIR spectra of high silica Y samples.

Fig. 7. Arrhenius plots of monomolecular cracking of propane over Y(2.6) (&), Y(3.3) (*), and Y(3.6) (~).

sample, as determined by NMR and XRD. HUSY-NaEDTA had a higher number of Brnsted acid sites per gram, because extraframework aluminum had been removed from the sample. The number of Brnsted acid sites as determined by the decomposition of i-propylamine was consistently higher than when determined by infra-red measurements after adsorption and desorption of pyridine. 3.2. Cracking of propane Fig. 7 shows the Arrhenius plots of the monomolecular cracking of propane of Y(2.6), Y(3.3), and Y(3.6). Table 4 lists the reaction rates at 823 K, the apparent activation energies, and the selectivity to cracking of all the samples. At 823 K, the cracking rates of high-silica Y zeolites are much higher than that of Y(2.6); the rate increased with Si/Al ratio. For Y(3.6), the rate per gram was about 16 times higher than that of Y(2.6) and for Y(3.3) it was about four times higher. For the calcined samples, the rate of HY(2.6) is about six times lower than that of Y(2.6), while the rate of HY(3.3) is about three times lower than that of the parent sample Y(3.3); the rate of HY(3.6) was about 1.5 times lower. The apparent activation energy of Y(2.6),

Fig. 6. FTIR spectra of HUSY and the treated samples.

acid sites decreased with decreasing aluminum content. For HUSY, both the ammonia and the EDTA treatment led to a higher number of Brnsted acid sites. HUSY-NaEDTA had more Brnsted acid sites than H(NH3)USY. The higher number of Brnsted acid sites of H(NH3)USY corresponds to its greater amount of framework aluminum compared to the parent

Table 4 Reaction rates and activation energies of monomolecular cracking of propane of zeolite Y samples Sample Y(2.6)e HY(2.6)e Y(3.3) HY(3.3) Y(3.6) HY(3.6) HUSY H(NH3)USY HUSY-NaEDTA
a b c d e

Rate-crackinga 106 (mol/g s bar) 0.29 0.05 1.3 0.49 5.4 3.3 2.0 2.2 6.8

Rate-crackingb 106 (mol/mol s bar) 0.26 0.25 1.5 1.7 7.6 8.3 6.7 6.1 13.2

Rate-crackingc 106 (mol/mol s bar) 0.15 0.05 0.8 n.d. 3.5 n.d. 1.9 1.9 4.9

Activation energy (kJ/mol) 165 208 166 186 165 176 166 161 162

100 cr/ (cr + de)d 60 38 81 64 79 77 81 69 84

Rate at 823 K/g. Rate per Brnsted acid site determined by pyridine infra-red after desorption at 423 K. Rate per Brnsted acid site determined by the decomposition of i-propylamine. Selectivity to cracking (in %) at 823 K. Data taken from ref. [6].

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

251

Y(3.3), and Y(3.6) was virtually the same at 165166 kJ/mol (Table 4). The apparent activation energy of the calcined samples was higher, 208 for HY(2.6), 186 for HY(3.3), and 176 kJ/mol for HY(3.6). The selectivity to cracking of the highsilica Y zeolite was 79 and 81%, which is higher than that of Y(2.6) (60%). Compared to their parent samples, the calcined samples had a lower selectivity to cracking. Table 4 lists the catalytic results of HUSY and the treated samples. H(NH3)USY had a higher cracking rate than HUSY and both rates were higher than that of Y(2.6) and Y(3.3) but lower than that of Y(3.6). The apparent activation energy of HUSY (166 kJ/mol) and H(NH3)USY (161 kJ/mol) was similar and, within the limits of accuracy, unchanged from those of the Y zeolites. The selectivity to cracking is lower for H(NH3)USY. The reaction rate per gram HUSY-NaEDTA was about three times higher than that of HUSY. Its apparent activation energy was 162 kJ/ mol, which is the same as that of HUSY and H(NH3)USY. The selectivity to cracking of HUSY-NaEDTA was 84%, very similar to HUSY and the high-silica Y samples. 4. Discussion 4.1. Inuence of framework aluminum content from synthesis The XRD patterns of the directly synthesized high-silica Y zeolites (Y(3.3) and Y(3.6)) were the same as that of normal Y zeolite (NH4Y), which indicates that their structure belongs to FAU (Fig. 2). The Si/Al ratios from AAS and MAS NMR are higher than that of normal Y zeolite. All the aluminum atoms in these samples were in tetrahedral coordination (Fig. 1), which implies that all the aluminum atoms are in the framework. Although all the tetrahedral aluminum atoms in Y zeolite do not contribute to the strong Brnsted acid sites, [57,58], the number of sites still correlates to the content of tetrahedral framework aluminum. Compared to normal Y zeolite, a lower number of framework aluminum atoms per unit cell in the high-silica samples led to fewer Brnsted acid sites, which is shown by the intensity in the stretching hydroxyl region, by infra-red measurements after pyridine adsorption, and by decomposition of i-propylamine. The infra-red spectra of the high-silica samples contain three types of hydroxyl bands (silanol, HF, and LF), which are the same as the band of normal Y zeolite except that the HF band of the high-silica samples is at lower frequency, indicative of their lower content of framework aluminum [27]. The calcined high-silica samples, HY(3.3) and HY(3.6), showed a large decrease in the intensity of the stretching hydroxyl bands. This is similar to HY(2.6), which is unstable and shows structural collapse after exposure to moisture and reheating [26]. Because the protonation of the alkane by the accessible Brnsted acid site is the rate-limiting step of monomolecular cracking on zeolites [34], this reaction tests the ability of the zeolite to protonate a weak base. We consider the intrinsic activation energy to be a measure of the acidity of the Brnsted acid sites that participate in the reaction. The high-silica Y zeolites had an almost identical apparent activation energy as

Y(2.6) in the monomolecular cracking of propane. The sum of the heat of adsorption and intrinsic activation energy remain unaltered. Because the heats of adsorption depend on the structure of the zeolite [60], the intrinsic activation energy remained unchanged. Therefore, sites of equal strength dominate the reactivity in Y(2.6), Y(3.3), and Y(3.6). However, the high-silica samples exhibited higher cracking rates per gram and per acid site than normal Y zeolite, and the rate increased with Si/Al ratio. It is generally accepted that the catalytic activity increases with the number of Brnsted acid sites [13,25,29,60], as shown by the poisoning and the chemical replacement of aluminum by silicon in the framework [25,35,61,66]. The catalytic activity decreases faster than the number of sites that are poisoned [57,61,66,62], showing that not all the Brnsted acid sites contribute to activity. Our results show that the higher activity of the high-silica samples originates from the number of Brnsted acid sites that participate in the reaction. Even though the samples had lower contents of framework aluminum, more catalytically active sites are present. The number of Brnsted acid sites in the high-silica samples, as measured by infra-red with the adsorption of pyridine and the decomposition of i-propylamine were lower than in Y(2.6), which is consistent with the changes in aluminum content of the framework. Normalized by these numbers, the rates per Brnsted acid site of the zeolites decreased in the order Y(3.6) > Y(3.3) > Y(2.6). The Brnsted acid sites in the FAU structure are heterogeneous due to different framework structures and chemical composition [63], and strong bases such as pyridine have the same effect on all the Brnsted acid sites [55]. The monomolecular cracking of propane is a highly demanding reaction and was performed at high temperatures (675 to 875 K); thus, the reaction likely probes only the strongest active Brnsted acid sites [40]. We suggest that the number of Brnsted acid sites, determined either by pyridine or isopropylamine, is higher than the number of sites that participate in the reaction. Beaumont and Barthomeuf [53] studied the acidity of FAUlike zeolites and showed that the effective Brnsted acidity increased with Si/Al ratio and reached a maximum at Si/Al = 6. Assuming that all the aluminum atoms are in the framework, the sample with a Si/Al ratio of six has about 27 framework aluminum atoms per unit cell. A model was developed to interpret the observation that the strength of the Brnsted acid sites depends on the aluminum environment [64], and only aluminum atoms with no next-nearest neighbor aluminum atoms were assumed to exhibit strong acidity [65,66]. Later, theoretical modeling and experimental measurements of the effect of the next-nearest neighbor in zeolites revealed that the catalytic activity increases with the framework aluminum content to a maximum at around 30 framework aluminum atoms per unit cell and then thereafter decreases with increasing framework aluminum content [13,25,6773]. The catalytic results of our samples are in line with these assumptions. 29Si MAS NMR indicated that Si(1Al) increased in the order Y(2.6) (42%) < Y(3.6) (49%) < Y((58%) and Si(0Al) increased with Si/Al ratio from 4 to 30%. The high-silica samples contain a

252

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253

greater number of isolated Brnsted acid sites [65,66,7476]. We attribute the enhanced catalytic activity of high-silica samples to the higher number of isolated framework aluminum atoms. All the characterization methods showed that the calcined samples HY(3.3) and HY(3.6) were partially dealuminated and had smaller unit cell sizes and lower numbers of Brnsted acid sites, which was also observed in HY(2.6) [6]. However, HY(2.6) had only around half the surface area and micropore volume of the parent sample. HY(3.3) and HY(3.6) had a surface area and a pore volume more similar to Y(3.3) and Y(3.6). Moreover, the intensity of the hydroxyl groups of the Brnsted acid sites of these calcined samples was lower than that of their respective parent samples and increased with Si/Al ratio (Fig. 6). These observations were corroborated by the catalytic results. The cracking rate of HY(2.6) was about 17% of that of the parent sample; the cracking rate of HY(3.3) and HY(3.6) was 38 and 61%, respectively. Their activation energy increased in the order HY(3.6) < HY(3.3) < HY(2.6), and the values were all higher than that of the parent samples, implying that structural collapse occurred and that nature of the catalytically active site was different. It has been shown that octahedral aluminum in acidic zeolites forms at room temperature after exposure to moisture [35,7779,87]. The results indicate that the structural collapse is a function of the Si/Al ratio: frameworks with lower aluminum contents are more stable [80]. The spectra of 27Al MAS NMR of HY(2.6), HY(3.3), and HY(3.6) showed a similar percentage of octahedral aluminum. 4.2. HUSY and the treated samples HUSY, a commercial sample with a framework Si/Al ratio of 5.1, was also studied in its protonic form following ammonia adsorption and an EDTA treatment. Steamed zeolites generally show enhanced catalytic activity, as was also shown in this study. The catalytic results of HUSY showed that the activation energy did not depend on these treatments and, within the limits of accuracy, was identical to that of untreated Y zeolite. The cracking rates per gram of HUSY were in the range of those of high-silica Y zeolites. The identical activation energy of all these samples shows that the structures of the catalytically active sites in all these samples are the same. Zeolites with lower contents of framework aluminum can exhibit higher catalytic activity, as shown by Y(2.6), Y(3.3), and Y(3.6), due to a higher number of isolated framework aluminum atoms. The large amount of octahedrally-coordinated aluminum in HUSY can be washed out by an EDTA treatment [47]. This treatment results in a sample (HUSY-NaEDTA) with a higher number of Brnsted acid sites [47] and higher activity. Its activation energy did not change. This indicates that the higher activity of HUSY, compared to the untreated Y zeolite, is not due to the extra-framework aluminum, which was assumed to be the active site or the enhancing site for increasing the catalytic activity of HUSY [23,25,81,82]. Following adsorption of ammonia on HUSY, the 27Al MAS NMR data (Fig. 1 and Table 2) show that most of the

octahedrally-coordinated aluminum had converted to tetrahedral coordination [5]. Thus, the framework aluminum content increased (Tables 2 and 3). The infra-red data (Fig. 6) for HUSY and H(NH3)USY after activation at 773 K indicate that the number of Brnsted acid sites represented by the HF band at 3625 cm1 increased at the cost of the 3600 cm1 peak (HF). We ascribe the enhanced catalytic activity of H(NH3)USY to a higher number of active Brnsted acid sites. It has been suggested that the band at 3600 cm1 is related to the enhanced catalytic activity [56,83]. However, a decrease in its intensity is paralleled by a higher activity. This suggests that it may play a minor role in catalytic activity. The similar activation energy of H(NH3)USY indicates that the treatment did not change the properties of the active sites but created more active sites by bringing extra-framework aluminum back into the framework. This suggests that the catalytic potential of a steamed zeolite is not fully utilized and that structural collapse is reversible [6]. The EDTA treatment was used to obtain ultrastable zeolites from Y zeolites by dealumination [46,84]. The resulting sample showed a higher catalytic activity and it was suggested that new strong Brnsted acid sites were created by the EDTA treatment [47,85]. Our HUSY sample after EDTA treatment had the same structure as the parent sample as shown by nitrogen physisorption and XRD [6]. The infra-red spectrum of HUSY-NaEDTA (Fig. 6) showed decreased intensity of the band at 3600 cm1 compared to the spectrum of H-USY, which conrms that this band may not be important for the catalytic activity. The number of Brnsted acid sites and the catalytic activity were higher, but the activation energy did not change. This implies that the sites that participated in the reaction do differ from the sites in the untreated Y zeolite. 5. Conclusion The strong Brnsted acid sites in Y zeolite with different Si/ Al ratios and with different contents of extra-framework aluminum are identical and show the same catalytic activity. Their intrinsic activity does not depend on the Si/Al ratio or on extra-framework aluminum. Only isolated sites that have no aluminum next-nearest neighbors are strong Brnsted acid sites. Their number is higher in Y zeolites with higher Si/Al ratio up to a ratio of 5. Acknowledgement We thank Dr. Anuji Abraham for her help with the MAS NMR measurements. References
[1] R.P. Silvy, Oil Gas J. 100 (2002) 48. [2] G.T. Kerr, J. Catal. 15 (1969) 2000. [3] E. Bourgeat-Lami, P. Massiani, F. Di Renzo, P. Espiau, F. Fajula, Appl. Catal. 72 (1991) 139. [4] J.A. van Bokhoven, A.M.J. van der Eerden, D.C. Koningsberger, Stud. Surf. Sci. Catal. 142 (2002) 1885. [5] A. Omegna, J.A. van Bokhoven, R. Prins, J. Phys. Chem. B 107 (2003) 8854.

B. Xu et al. / Applied Catalysis A: General 333 (2007) 245253 [6] B. Xu, F. Rodunno, S. Bordiga, R. Prins, J.A. van Bokhoven, J. Catal. 241 (2006) 66. [7] J. Jiao, W. Wang, B. Sulikowski, J. Weitkamp, M. Hunger, Micropor. Mesopor. Mater. 90 (2006) 246. [8] C.V. McDaniel, P.K. Maher, US Patents 3,292,192 (1966); 3,449,070 (1969). [9] D.P. Siantar, W.S. Millman, J.J. Fripiat, Zeolites 15 (1995) 556. [10] J.A. van Bokhoven, A.L. Roest, D.C. Koningsberger, J.T. Miller, G.H. Nachtegaal, A.P.M. Kentgens, J. Phys. Chem. B 104 (2000) 6743. [11] A.L. Blumenfeld, J.J. Fripiat, Top. Catal. 4 (1997) 119. [12] M.J. Remy, D. Stanica, G. Poncelet, E.J.P. Feijen, P.J. Grobet, J.A. Martens, P.A. Jacobs, J. Phys. Chem. 100 (1996) 12440. [13] S.J. DeCanio, J.R. Sohn, P.O. Fritz, J.H. Lunsford, J. Catal. 101 (1986) 132. [14] R.A. Beyerlein, G.B. McVicker, L.N. Yacullo, J.J. Ziemiak, J. Phys. Chem. 92 (1988) 1967. [15] F. Lonyi, J.H. Lunsford, J. Catal. 136 (1992) 566. [16] P.V. Shertukde, W.K. Hall, J.-M. Dereppe, G. Marcelin, J. Catal. 139 (1993) 468. [17] Y. Hong, V. Gruver, J.J. Fripiat, J. Catal. 150 (1994) 421. [18] R.A. Beyerlein, C. Choi-Feng, J.B. Hall, B.J. Huggins, G.J. Ray, Topics Catal. 4 (1997) 27. [19] D. Barthomeuf, J. Phys. Chem. 83 (1979) 249. [20] B. Beagley, J. Dwyer, F.R. Fitch, R. Mann, J. Walters, J. Phys. Chem. 88 (1984) 1744. [21] U. Lohse, B. Parlitz, V. Patzelova, J. Phys. Chem. 93 (1989) 3677. [22] B. Hunger, M. Heuchel, L.A. Clark, R.Q. Snurr, J. Phys. Chem. B 106 (2002) 3882. [23] R. Carvajal, P.J. Chu, J.H. Lunsford, J. Catal. 125 (1990) 123. [24] A.I. Biaglow, D.J. Parrillo, G.T. Kokotailo, R.J. Gorte, J. Catal. 148 (1994) 213. [25] P.O. Fritz, J.H. Lunsford, J. Catal. 118 (1989) 85. [26] G. Garralon, A. Corma, V. Fornes, Zeolites 9 (1989) 84. [27] U. Lohse, E. Lofer, M. Hunger, J. Stockner, V. Patzelova, Zeolites 7 (1987) 11. [28] M. Kuehne, H.H. Kung, J.T. Miller, J. Catal. 171 (1997) 293. [29] W.O. Haag, R.M. Lago, P.B. Weisz, Nature 309 (1984) 589. [30] W.O. Haag, R.M. S Dessau, in: Proceedings 8th International Congress on Catalysis Berlin, vol. 2, Berlin, 1984, Dechema Frankfurt-am-Main, (1984), p. 305. [31] K.A. Cumming, B.W. Wojciechowski, Catal. Rev.-Sci. Eng. 38 (1996) 101. [32] W.O. Haag, R.M. Dessau, R.M. Lago, Stud. Surf. Sci. Catal. 60 (1991) 255. [33] A. Corma, J. Planelles, J. Sanchez-Marin, F. Thomas, J. Catal. 93 (1985) 30. [34] T.F. Narbeshuber, H. Vinek, J.A. Lercher, J. Catal. 157 (1995) 388. [35] W.O. Haag, Stud. Surf. Sci. Catal. 84 (1994) 1375. [36] M. Temkin, The Kinetics of Some Industrial Heterogeneous Catalytic Reactions in Advances in Catalysis, vol. 28, Academic Press, New York, 1979. [37] J. Wei, Chem. Eng. Sci. 51 (1996) 2995. [38] S.M. Babitz, B.A. Williams, J.T. Miller, R.Q. Snurr, W.O. Haag, H.H. Kung, Appl. Catal. A 179 (1999) 71. [39] J.A. van Bokhoven, B.A. Williams, W. Ji, D.C. Koningsberger, H.H. Kung, J.T. Miller, J. Catal. 224 (2004) 50. [40] B. Xu, C. Sievers, S.B. Hong, R. Prins, J.A. van Bokhoven, J. Catal. 244 (2006) 163. [41] C.E. Ramachandran, B.A. Williams, J.A. van Bokhoven, J.T. Miller, J. Catal. 233 (2005) 100. [42] F. Delprato, L. Delmotte, J.L. Guth, L. Huve, Zeolites 10 (1990) 546.

253

[43] F. Dougnier, J. Patarin, J.L. Guth, D. Anglerot, Zeolites 13 (1993) 122. [44] K. Karim, J. Zhao, D. Rawlence, J. Dwyer, Micropor. Mater. 3 (1995) 695. [45] C. Berger, R. Glaser, R.A. Rakoczy, J. Weitkamp, Micropor. Mesopor. Mater. 83 (2005) 333. [46] G.T. Kerr, J. Phys. Chem. 72 (1968) 2594. [47] N. Katada, Y. Kageyama, K. Takahara, T. Kanai, H.A. Begum, M. Niwa, J. Mol. Catal. A 211 (2004) 119. [48] D.W. Breck, F.M. Flanigen, Molecular Sieves, Society of Chemical Industry, London, UK, 1968, p. 47. [49] J.R. Sohn, S.J. Decanio, J.H. Lunsford, D.J. ODonnell, Zeolites 6 (1986) 225. [50] C.A. Emeis, J. Catal. 141 (1993) 347. [51] M.A. Makarova, J. Dwyer, J. Phys. Chem. 97 (1993) 6337. [52] F. Eder, M. Stockenhuber, J.A. Lercher, J. Phys. Chem. B 101 (1997) 5414. [53] R. Beaumont, D. Barthomeuf, J. Catal. 26 (1972) 218. [54] J.R. Sohn, S.J. DeCanio, P.O. Fritz, J.H. Lunsford, J. Phys. Chem. 90 (1986) 4847. zinger, Langmuir 17 (2001) 6233. [55] W. Daniell, N.-Y. Topse, H. Kno [56] J.H. Lunsford, ACS Symp. Ser. 452 (1991) 1. [57] S. Kotrel, M.P. Rosynek, J.H. Lunsford, J. Catal. 182 (1999) 278. [58] A.I. Biaglow, D.J. Parrillo, R.J. Gorte, J. Catal. 144 (1993) 193. [60] D.H. Olson, W.O. Haag, R.M. Lago, J. Catal. 61 (1980) 390. [61] R.A. Beyerlein, G.B. McVicker, L.N. Yacullo, J.J. Ziemiak, Preprint ACS Div. Petro. Chem. 31 (1986) 190. [62] E.A. Lombardo, G.A. Sill, W.K. Hall, J. Catal. 119 (1989) 426. [63] M. Sierka, U. Eichler, J. Datka, J. Sauer, J. Phys. Chem. B 102 (1998) 6397. [64] E. Dempsey, J. Catal. 33 (1974) 497. [65] R.J. Mikovsky, J.F. Marshall, J. Catal. 44 (1976) 170. [66] L.A. Pine, P.J. Maher, W.A. Wachter, J. Catal. 85 (1984) 466. [67] D. Barthomeuf, Mater. Chem. Phys. 17 (1987) 49. [68] V.B. Kasansky, Stud. Surf. Sci. Catal. 18 (1984) 61. [69] W.J. Mortier, J. Catal. 55 (1978) 138. [70] P.A. Jacobs, Catal. Rev.-Sci. Eng. 24 (1982) 415. [71] J. Dwyer, F.R. Fitch, E.E. Nkang, J. Phys. Chem. 87 (1983) 5402. [72] H. Stach, J. Janchen, U. Lohse, Catal. Lett. 13 (1992) 389. [73] J. Datka, B. Gil, J. Catal. 145 (1994) 372. [74] G.M. Zhidomirov, V.B. Kazansky, Adv. Catal. 34 (1986) 131. [75] B. Gil, E. Broclawik, J. Datka, J. Klinowski, J. Phys. Chem. 98 (1994) 930. [76] M. Occelli, U. Voigt, H. Eckert, App. Catal. A 259 (2004) 245251. [77] B.H. Wouters, T.H. Chen, P.J. Grobet, J. Am. Chem. Soc. 120 (1998) 11419. [78] J.A. van Bokhoven, A.M.J. van der Eerden, D.C. Koningsberger, J. Am. Chem. Soc. 125 (2003) 7435. [79] S. Altwasser, J. Jiao, S. Steuernagel, J. Weitkamp, M. Hunger, Stud. Surf. Sci. Catal. 154 (2004) 3098. [80] A. Abraham, S.H. Lee, C.H. Shin, S.B. Hong, R. Prins, J.A. van Bokhoven, Phys. Chem. Chem. Phys. 6 (2004) 3031. [81] E. Brunner, H. Ernst, D. Freude, M. Hunger, C.B. Krause, D. Prager, Zeolites 9 (1989) 282. [82] Y. Sendoda, Y. Ono, Zeolites 8 (1988) 101. [83] E. Guillon, A.A. Quoineaud, T. Armaroli, S. Lacombe, S. Gautier, Stud. Surf. Sci. Catal. 154 (2004) 1539. [84] G.T. Kerr, J. Phys. Chem. 73 (1969) 2780. [85] M. Niwa, K. Suzuki, K. Isamoto, N. Katada, J. Phys. Chem. B 110 (2006) 264. [86] T.L.M. Maesen, E. Beerdsen, S. Calero, D. Dubbeldam, B. Smit, J. Catal. 237 (2006) 278290. [87] J.A. van Bokhoven, P.J. Kunkeler, H. van Bekkum, D.C. Koningsberger, J. Catal. 211 (2002) 540547.