Ind. Eng. Chem. Res. 2004, 43, 8217-8225
8217
Influence of the Binder on the n-Octane Hydroisomerization over Palladium-Containing Zeolite Catalysts Antonio de Lucas, Jose´ Luis Valverde,* Paula Sa´ nchez, Fernando Dorado, and Marı´a Jesu ´ s Ramos Departamento de Ingenierı´a Quı´mica, Facultad de Ciencias Quı´micas, Universidad de CastillasLa Mancha, Avenida Camilo Jose´ Cela s/n, 13071 Ciudad Real, Spain
The influence of the binder on the properties and performance of palladium-containing zeolite catalysts in the n-octane hydroisomerization was studied. Three different framework zeolites were used as catalysts: mordenite, β, and ZSM-5 with or without binder. To characterize the catalysts, surface area measurements, temperature-programmed desorption of ammonia, atomic absorption spectroscopy, inductively coupled plasma emission spectrophotometry, hydrogen chemisorption, and solid-state 27Al NMR were used. Catalytic performance of the zeolites was strongly influenced by the binder because its presence modified both zeolite acidity and porosity. A decrease in the strong acidity of the zeolites could be observed when the samples were agglomerated, due to a solid-state ion exchange between the zeolite protons and bentonite sodium. The neutralization of some zeolite acid sites caused a decrease in the n-octane conversion for mordenite (from 90.5 to 53.8 mol %) and ZSM-5 (from 78.4 to 71.2 mol %) zeolites. The binder modified the porosity of the zeolite providing meso- and macropores, which allow a higher formation of branched isomers for mordenite (from 27.3 to 48.9 mol %) and ZSM-5 (from 14.3 to 35.4 mol %). Activity of β zeolite was modified by the presence of extraframework aluminum species (EFAL) formed during the agglomeration process, improving both n-octane conversion (from 22.9 to 88.7 mol %) and branched isomer selectivity (from 48.7 to 77.5 mol %). 1. Introduction
Scheme 1. Mechanism of n-Octane Isomerization
The current environmental requirements have led the government and the industry to support the production of gasoline with the minimum content of compounds considered harmful to the environment or the public health, such as lead, aromatics, and oxygenated compounds. The reduction in the aromatics content will have a special effect on the units of production of reformed naphtha that use fundamentally heavy naphtha in the range C7-C10, with a paraffin content between 30% and 60% in weight. The technologically more interesting alternative is the utilization of these paraffins for the production of branched paraffins with high octane number. Although the hydroisomerization of C7 and C8 has been widely studied,1-6 the multibranched C8 alkanes are the most useful isomers because of their high octane number. The industrial processes of n-paraffins hydroisomerization need the presence of bifunctional catalysts, which present two kinds of active or functional sites: (i) an acid function that allows the occurrence of carbenium ions involved in isomers formation and (ii) a hydrogenating-dehydrogenating function that originates the precursor of the carbocations by paraffin dehydrogenation and causes the hydrogenation of the unsaturated remains.7 Most industrial hydroisomerization catalysts are based on zeolites, which provide the acid function. As the hydrogenating-dehydrogenating function, several metals have been tested including Pt, Pd, Rh, Ir, Ru, Re, * To whom correspondence should be addressed. Tel.: +34-926-29 54 37. Fax: +34-926-29 53 18. E-mail:
[email protected].
and Ni, mostly associated with mordenite or CaY.8 Platinum was singled out as exhibiting both high activity and good selectivity.9 However, palladium, although rarely studied, was shown to be the most selective in the butane hydroisomerization.10,11 The hydroisomerization mechanism is as follows (Scheme 1): the metallic function converts the linear alkane into the corresponding alkene. The acid function is responsible for the isomerization of the linear alkene into the corresponding isoalkene, via the formation of the secondary carbenium ion first formed upon addition of the linear alkene to the proton of the Bro¨nsted acid site. The secondary carbenium ion is driven by thermodynamics to rearrange into the tertiary carbenium ion, which is likely to restore the Bro¨nsted acid site and to release the isomerized alkene or to crack via β scission to provide a lower alkene and a new carbenium fragment.7 An ideal mechanism is characterized by the control of the acid function; the acid function has to be as strong as possible to convert the intermediate alkene into the corresponding carbenium ion at the lowest possible temperature to favor the isomerization against cracking. Most industrial zeolite-based catalysts require the zeolite to be pelletized with a binder to improve the mechanical properties of the catalyst particles, thus avoiding extremely high-pressure drops in fixed-bed
10.1021/ie040133j CCC: $27.50 © 2004 American Chemical Society Published on Web 11/20/2004
8218
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004
reactors. It has been demonstrated that the binder is not active as a catalyst but could change the acid properties of a zeolite as a result of changes in the proton-exchange efficiency, trapping by the binder of coke precursors, and/or blocking of zeolite channels during the pelletization process.11,12 Information regarding the influence of the binder on the acidity and catalytic performance of zeolites is, therefore, very important for the development of industrial catalysts. Some authors have studied the influence of different binders on the catalytic performance of zeolites,13-16 even though bentonite has not been widely studied. Bentonite is a laminar and expandable clay with wet binding properties and is widely available throughout the world. The scientific interest of bentonite is due to its physical and chemical properties as well as its low price. Consequently, the industrial application of bentonite is an attractive process. The dispersability of clays in aqueous suspensions is the reason for their agglomeration properties: zeolite particles are surrounded by clay laminae, and, when the water is removed, a solid phase is achieved with the zeolite particles agglomerated by the clay. It has been shown that the acid forms of clays do not have binding properties and that their sodium forms exhibit better properties.17 Because zeolites are mainly used as acid catalysts, further transformation of the agglomerated zeolite to the acid form is required.11 The aim of this work was to investigate the influence of the binder on the catalytic activity and selectivity in the hydroisomerization of n-octane over mordenite, β, and ZSM-5 zeolites agglomerated with sodium bentonite using palladium as the hydrogenating-dehydrogenating function. 2. Experimental Methods 2.1. Catalyst Preparation. The parent zeolites mordenite (Si/Al ) 10.4), β (Si/Al ) 13.0), and ZSM-5 (Si/Al ) 15.6) were supplied in the ammonium form by Zeolyst International. The ammonium form was calcined at 550 °C for 15 h to obtain the protonic form. Bentonite (sodium form) was supplied by Aldrich Chemical Co. In the binding process, zeolite (35 wt %) and bentonite (65 wt %) were mixed together and suspended in water at 60 °C for 2 h. The suspension was then dried at 120 °C for 12 h. After being ground and sieved, particles with an average particle size of 0.75 mm were obtained. With this size, no internal diffusion limitations in the catalytic runs were detected. Finally, the agglomerated zeolites were calcined at 550 °C for 15 h, obtaining the samples named as MOR/BentNa, β/BentNa, and ZSM5/BentNa. After the agglomeration process and to reincorporate the acid function in the zeolite, the agglomerated mordenite and ZSM-5 zeolites were ion-exchanged with 0.6 N HCl (35 mL g-1). The ion exchange for the β zeolite was carried out three times with 1 M NH4Cl (30 mL g-1). The ion-exchanged samples were subsequently calcined again at 550 °C for 15 h to obtain the acid form of the zeolites. All of the catalysts (with or without binder) were impregnated with an aqueous Pd(NO3)2 solution. The metal concentration of the impregnating solution was calculated to yield a final Pd content of 1 wt % in the resulting catalysts.
Figure 1. TPDA curves for HMOR, PdMOR, and PdMOR/Bent catalysts.
After the impregnation process, the catalysts were calcined at 400 °C for 4 h and reduced in situ under a hydrogen flow of 190 mL min-1 g-1 at 450 °C for 4 h. Agglomerated catalysts were named as “Bent” following the name of the zeolite (MOR, β, and ZSM-5). For example, Pdβ/Bent corresponds to a Pdβ zeolite (in the acid form) agglomerated with bentonite. In this work, zeolites agglomerated without further incorporation of the acid function were considered. Thus, for example, Pdβ/BentNa corresponds to a β zeolite agglomerated without further incorporation of the acid form after the agglomeration process. 2.2. Catalyst Characterization. Pores size distribution and BET surface area were determined by adsorption and desorption data acquired on a Micromeritics ASAP 2010 adsorptive and desorptive apparatus. The samples were evacuated under vacuum of 5 × 10-3 Torr at 350 °C for 15 h. Specific total surface areas were calculated using the BET equation, whereas specific total pore volumes were evaluated from N2 uptake at a relative pressure (P/Po) of N2 equal to 0.99. The Horvath-Kawazoe method was used to determine the microporous surface area and micropore volume.18 The Barret, Johner, and Halenda (BJH) method was used to determine the distribution of the mesopores.19 Surface area measurements have an error of (3%. The concentration of the acid sites was measured by temperature-programmed desorption of ammonia (TPDA) using a Micromeritics TPD/TPR 2900 analyzer. The sample was first heated at 15 °C min-1 under a flow of helium from room temperature to the calcination temperature, holding this temperature for 30 min. After the catalysts were reduced under a hydrogen flow, the system was cooled to 180 °C. Ammonia was then flowed over the sample for 15 min. Later, the sample was purged with helium for 1 h to eliminate physisorbed species. The temperature was ramped at 15 °C min-1 from 180 to 560 °C, and TPDA data were acquired. The total acidity was obtained by integration of the area under the curve. This curve was fitted using two peaks, which were classified as weak and strong acidity depending on the desorption temperature (see Figure 1, where TPDA curves are shown as an example). The use of these peaks was not based on any peak assignment to a specific Bro¨nsted or Lewis acid sites, but it was a convenient way to categorize the acid strength
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004 8219
distribution obtained by this method. The average relative error in the acidity determination was lower than 3%. To quantify the palladium content in the catalysts, atomic absorption spectroscopy (AA) measurements were performed using a SpectrAA 220FS spectrophotometer. The aluminum and silicon content was measured using an inductively coupled plasma emission spectrophotometer LIBERTY RL Sequential ICP-AES. Prior to measurement, the samples were dissolved in hydrofluoric acid and diluted to the interval measurement. The error of these measurements was (1%. The palladium dispersion was determined from chemisorption measurements. The apparatus used was the same as that described for the TPDA. The experiments were carried out using the dynamic pulse technique with an argon (99.9990%) flow of 50 mL min-1 and pulses of hydrogen. To calculate the metal dispersion, an adsorption stoichiometry of Pd/H ) 1 was assumed.20 The chemisorption experiments with hydrogen pulses were carried out at 60 °C to avoid the spillover phenomenon.21 Previously, the sample was pretreated by heating at 15 °C min-1 in flowing helium up to 250 °C and kept constant at this temperature for 20 min. The sample was then reduced in situ. After, the hydrogen was removed by flowing inert gas for 30 min, the temperature being 10 °C higher than the reduction temperature. Finally, the sample was cooled to the experiment temperature in an inert gas flow. The dispersion measurements with H2 pulses had an error of (5%. Solid-state 27Al NMR spectra were collected in a Bruker Avance WB 400 spectrometer. The 27Al NMR spectra were obtained at 12.5 kHz using 15° pulses and a 1 s delay, a total of 5000 pulses being accumulated. 2.3. Reaction Tests. n-Octane hydroisomerization reactions were carried out in an Autoclave Engineers (BTRS-Jr) microreactor. Experimental conditions were as follows: weight of catalyst, 1.5 g; reaction temperature, 250-410 °C; total pressure, 10 bar; WHSV ) 10 gn-C8 h-1 gzeolite-1; and H2/n-C8 molar ratio, 14. All measurements were collected after 1 h on stream. Reaction products were analyzed with a HP 5890 series II gas chromatograph equipped with a flame ionization detector and an automatic valve for continuous analysis. The reactor effluent stream was sent for analysis through a heated line (about 180 °C) to the automatic valve. The gas chromatograph was equipped with a capillary column Supelco Petrocol DH50.2, 0.2 mm i.d. and 50 m length. Results from a reproduced experiment showed that conversion and isomer selectivity had an error of (4%. 3. Results and Discussion 3.1. Characterization of Catalysts. Table 1 shows the acidity data of the raw materials and the nonagglomerated catalysts. The catalysts with or without metal had similar acidity values, which shows the scarce influence that the metal incorporation has on the catalyst acid properties. Bentonite acidity was very low, even not observing strong acidity. Anyhow, the contribution of the bentonite to the acidity of the agglomerated catalysts should not be considered remarkable. Acidity data of the agglomerated catalysts are given in Table 2. It can be observed that the experimental values of weak acidity of these catalysts were always higher than the predicted values, the opposite effect
Table 1. Acidity Data for Nonagglomerated Samples
sample
total acidity weak acidity strong acidity (mmol of Tda Tda (mmol of (mmol of NH3 gcat-1) NH3 gcat-1) (°C) NH3 gcat-1) (°C)
bentonite HMOR Hβ HZSM-5 PdMOR Pdβ PdZSM-5 a
0.038 0.996 0.626 0.573 0.946 0.580 0.512
0.038 0.164 0.129 0.085 0.122 0.123 0.012
274 312 273 275 315 283 296
0 0.832 0.497 0.488 0.824 0.458 0.500
479 353 390 480 369 386
Desorption temperature.
Table 2. Acidity Data for the Agglomerated Catalysts weak total strong acidity acidity acidity (mmol of Td (mmol of (mmol of Td NH3 gcat-1) NH3 gcat-1) (°C) NH3 gcat-1) (°C)
sample PdMOR/Bent PdMOR/BentNa Pdβ/Bent Pdβ/BentNa PdZSM-5/Bent PdZSM-5/BentNa
0.378b 0.366b 0.373a 0.240b 0.197b 0.244a 0.19b 0.211b 0.225a
0.127b 0.161b 0.082a 0.090b 0.082b 0.070a 0.055b 0.133b 0.054a
291 288 292 291 304 298
0.251b 0.205b 0.291a 0.150b 0.115b 0.174a 0.144b 0.078b 0.171a
422 403 369 402 385 448
a Predicted value calculated from the contribution of the nonagglomerated zeolite and the binder. b Experimental value.
Table 3. Metal Dispersion, Surface Area, and Pore Volume Measurement
catalyst bentonite MOR β ZSM-5 PdMOR Pdβ PdZSM-5 PdMOR/Bent Pdβ/Bent PdZSM-5/Bent a
pore volume deviation (µL gcat-1) surface from the areaa theoretical DH2 meso- and (m2 gcat-1) value (%) (%) micropores macropores 37 560 636 412 559 613 410 228 233 163
3.7 2.5 3.0
31.2 35.4 16.8 24.4 25.3 15.6
4.3
88.5
199.9 116.1 144.4 75.5 47.1 50.6
105.4 840.0 133.3 119.7 350.6 145.7
Experimental value.
being observed for strong acidity. The predicted values were calculated from the contribution of the raw materials: zeolite and bentonite. The acidity values of the agglomerated samples without further incorporation of the acid form after the agglomeration process are included in Table 2. As expected, an increase of the strong acidity when protons were incorporated by ion exchange to the agglomerated samples was verified. First, we considered the possibility that the decrease of the strong acidity could be due to blocking of zeolite channels by the binder. To verify it, the surface area of the different catalysts was measured. This measurement together with the deviation of the theoretical value, calculated from the contribution of the bentonite and the nonagglomerated zeolites, is shown in Table 3. It is clear that there was not blocking of channels because the experimental values of surface area were in agreement with the theoretical values. Values of surface area of samples without metal are included in Table 3. Similar surface area values were obtained for samples with and without metal, suggesting that there is not blocking of the zeolite channels by the metal. These results could suggest that the palladium particles
8220
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004
Figure 2. 27Al NMR spectra of PdMOR, PdMOR/Bent, and PdMOR/BentNa.
are located mainly on the external surface of the zeolite crystals. Palladium dispersion (DH2) values were coherent with this supposition. According to previous works,10,22 the average diameter of the palladium particles would be about 33-75 Å, too big for metal particles to be located inside the main channels of the zeolite. According to some authors,23-25 the decrease of the number of strong acid sites could be due to solid-state ion exchange between the zeolite protons and clay sodium during the calcination that follows the incorporation of the acid function. The Na+ cations are also weak acid sites,26 so they could alter the density of these sites as, according to Table 2, they did in fact. It can be observed in Table 3 that agglomerated catalysts presented a higher meso- and macropore volume than nonagglomerated ones. The binder must provide these meso- and macropores, in accordance with its high meso- and macropore volume. However, how the meso- and macropore volume of the nonagglomerated β zeolite was higher than that obtained in the agglomerated sample could be observed. β zeolite crystallizes with many stacking faults,27 suggesting that most of its mesoporosity is associated with intercrystallite voids.28 When the zeolite is agglomerated, binder fills part of these intercrystallite voids; more important is the mesoporosity lost due to the filling up of the voids that the mesoporosity provides by the binder. Furthermore, β zeolite has a three-dimensional 12-ring pore system, and because of this property, the framework is very flexible. Mordenite zeolite has a one-dimensional 12-ring pore system, and ZSM-5 has a three-dimensional 10-ring pore system. Both structures are less flexible than the β structure. As compared to ZSM-5 or mordenite zeolites, the β framework is less stable and exhibits substantial distortions depending on the zeolite treatment, resulting in an opening of the Al-O framework bonds with formation of a considerable amount of octahedrally coordinated framework aluminum, as well as extraframework aluminum species (EFAL).29,30 EFAL species have been reported to be formed during high-temperatures treatments of protonic β zeolite.29 Figures 2-4 show the 27Al NMR spectra for different samples. The signal at 0 ppm was assigned to octahedral Al (EFAL).28 This signal was significantly higher in agglomerated β-zeolite samples (relative intensity I0ppm/I55ppm in Pdβ/Bent and Pdβ/BentNa was 0.88 and 0.78, respectively) and in PdZSM-5/Bent (I0ppm/I55ppm ) 1.18) and PdMOR/Bent (I0ppm/I55ppm ) 0.48), as compared to their nonagglomerated samples (I0ppm/I55ppm of PdMOR, Pdβ, and
Figure 3.
27Al
NMR spectra of Pdβ, Pdβ/Bent, and Pdβ/BentNa.
Figure 4. 27Al NMR spectra of PdZSM-5, PdZSM-5/Bent, and PdZSM-5/BentNa.
Figure 5. 27Al NMR spectra of β agglomerated zeolite before the step of calcination that would lead to β/BentNa sample.
PdZSM-5 was 0.25, 0.13, and 0.22, respectively). 27Al NMR analysis of bentonite (figure not included in this paper) showed the presence of a main signal at 0 ppm (I0ppm/I55ppm ) 5.35). Regarding the spectrum of the agglomerated β zeolite before the step of calcination that would lead to the β/BentNa (I0ppm/I55ppm ) 0.75) sample (Figure 5), it is clear that the presence of EFAL species cannot be attributed only to the calcination treatment but also to some specific mechanism occurring during the agglomeration process. According to Jasra et al.,31 the agglomeration process could cause a migration of reasonable mobile cations from the interlayer space of the clay structure to the zeolite. On the other hand, Fougerit et al.32 suggested that the formation of new acid sites in the zeolite could be attributed to the migration of soluble Al species from the alumina binder into the zeolite framework. In turn, mordenite and ZSM-5 zeolite showed an increase in the signal assigned to the EFAL species in the agglomerated samples that incorporated the acid function (I0ppm/I55ppm values for PdMOR/BentNa and PdZSM-5/BentNa of 0.27 and 0.36, respectively). It
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004 8221 Table 4. n-Octane Conversion at 350 °C, Isomer Selectivity, and Yield Obtained to C5+ Products at 50 mol % Conversiona sample
conversion (mol %)b
S mono (mol %)c
Y mono (mol %)d
S multi (mol %)e
Y multi (mol %)f
yield C5+ (mol %)g
PdMOR Pdβ PdZSM-5 PdMOR/BentNa Pdβ/BentNa PdZSM-5/BentNa PdMOR/Bent Pdβ/Bent PdZSM-5/Bent
90.5 22.9 78.4 10.4 28.5 16.5 53.8 88.7 71.2
18.1 29.9 10.4 40.2 53.3 46.7 35.0 55.4 29.4
11.8 13.0 5.9 17.2 30.7 24.0 18.9 35.6 12.7
9.2 18.8 3.9 16.5 19.5 3.6 13.9 22.1 6.0
6.0 8.2 2.2 7.0 11.2 1.8 7.5 14.2 2.6
33.0 25.5 27.8 29.3 45.8 35.5 34.7 53.3 26.6
a Catalyst weight, 1.5 g; H /C ) 14; total pressure, 10 bar; 1 h on stream. b 350 °C. c Monobranched isomers selectivity obtained at 50 2 8 mol % conversion. d Monobranched isomers yield obtained at 50 mol % conversion. e Multibranched isomers selectivity obtained at 50 mol % conversion. f Multibranched isomers yield obtained at 50 mol % conversion. g C5-C8 isomers yield at 50 mol % conversion.
seems clear that the acid treatment should remove the aluminum species from the zeolite framework, remaining in the form of the EFAL species.28 3.2. Influence of the Binder in the Hydroisomerization of n-Octane on Pd Catalysts. 3.2.1. nOctane Conversion. Table 4 shows the main reaction parameters obtained at 50 mol % conversion in the hydroisomerization of n-octane for the agglomerated and nonagglomerated palladium catalysts. It was previously tested that both bentonite and zeolite with no metallic function had not practical catalytic activity. In the case of β-zeolite-based catalysts, the presence of binder improved the conversion, whereas the contrary effect was observed in the case of mordenite and ZSM5. As mentioned above, the agglomeration process introduced Na+ cations in the zeolite framework because of a solid-state ion-exchange process, between H+ from the zeolite and Na+ from the clay,23,24 causing a decrease of the number of strong acid sites responsible of the isomerization reaction (Table 2). This fact was clear for mordenite and ZSM-5 zeolites (Table 4). The n-octane conversion in both agglomerated mordenite and ZSM-5 zeolite based catalysts increased after the incorporation of the acid form, although it was not higher than that corresponding to nonagglomerated samples. The neutralization of the acid sites in ZSM-5 by Na+ from the clay (54.4%) was higher than that in mordenite (29.5%). This could be explained because the solid-state ion exchange between H+ and Na+ in mordenite could be limited by a linear diffusion phenomenon,33 thus partially inhibiting the Na+ transfer from the clay to the zeolite. After the incorporation of the acid form, agglomerated ZSM-5 zeolite based catalyst recovered up to 45.8% of strong acidity of the parent zeolite, whereas mordenite zeolite only recovered up to 18.3% (Table 2). This fact could be explained by attending to the geometry of mordenite channels,33 which could interfere with the incorporation process of the acid function. The strong acid sites recovery in ZSM-5 zeolite based catalysts allowed the PdZSM-5/Bent sample to have a n-octane conversion value similar to that of the PdZSM-5 one. For the Pdβ/BentNa sample, the n-octane conversion was higher than that corresponding to the nonagglomerated sample, despite the neutralization of the acid sites by the binder. As the acid function was incorporated (Pdβ/Bent sample), the n-octane conversion was higher than that reached with the nonagglomerated β zeolite (Table 4). Several authors have demonstrated how the presence of some EFAL species can increase the catalytic activity
in β zeolites.9,28,34-36 The number of framework aluminum atoms per unit cell is not the only parameter that determines the catalyst activity. Aluminum extraframework species (EFAL) and transient state species can also play a role. As observed in Figure 3, EFAL species are formed during the agglomeration process on Pdβ/Bent and Pdβ/ BentNa samples. Some cationic EFAL species having strong Lewis acidity could interact with the structural Bro¨nsted acid sites, enhancing their acid strength through a synergetic effect28 and, consequently, making them much more active for n-octane hydroisomerization. 3.2.2. Isomer Selectivity. Figure 6 shows octane isomer selectivity versus n-octane conversion for all of the catalysts used in this work. A maximum in the selectivity at low conversions can be observed where the primary reaction products were practically the only formed. They are the monobranched isomers: 2-methylheptane, 3-methylheptane, 4-methylheptane, and 3-ethylhexane. The conversion increased with the temperature, the monobranched isomers yielding the dibranched ones: 2,2-dimethylhexane, 2,3-dimethylhexane, 2,4dimethylhexane, 2,5-dimethylhexane, 3,3-dimethylhexane, 3,4-dimethylhexane, and 3-ethyl-3-methylpentane. The cracking products occurred at moderate conversions, their concentration increasing at high temperatures. It caused an expected decrease of the branched isomers selectivity. The highest selectivity to isomers was found over the β zeolite in both nonagglomerated and agglomerated zeolites. For mordenite and ZSM-5 zeolites, the selectivity to octane isomers was lower than 100% even at low conversions. The product distribution with the different catalysts at approximately 50 mol % conversion is shown in Table 5. The observed selectivities were only a function of the level of conversion, regardless of the reaction temperature.37 Over all catalysts, the same types of products were obtained, which included isooctanes (monobranched and dibranched C8 paraffins) and cracked products up to C8. C9 or higher products were not observed. As expected, β zeolite, because of its relatively large pore size in comparison with mordenite and ZSM-5 zeolites, allowed a higher extent of C8 paraffins branching at the same level of n-octane conversion, because the latter impose transition state-type shape selectivity, thus favoring the formation of the less bulky isomers. Differences in isomerization-cracking selectivity in these zeolites could be explained on the basis of differences in not only pore size within the different framework but also acidity. This way, stronger acidity and narrower pore diameter would make the average lifetime of the
8222
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004
Figure 7. Equilibrium composition of normal, mono-, di-, and tribranched octanes versus temperature obtained by HYSYS (by Aspentech).
Figure 6. Octane isomers selectivity as a function of n-octane conversion: (a) nonagglomerated catalysts; (b) agglomerated catalysts with further incorporation of acid function after the agglomeration process; and (c) agglomerated catalysts without further incorporation of acid function after the agglomeration process.
carbocations on the catalyst surface longer and the diffusion of the branched products slower, respectively. Both factors will favor the cracking of the tertiary carbocations formed during isomerization before they desorb, and the readsorption and cracking of the branched paraffins before they leave the pores and come into the gas stream.38 The selectivity to isoalkanes in the cracked product was much lower over ZSM-5 zeolite than over β and mordenite zeolites. The high selectivity to propane and pentane in the ZSM-5 catalysts is remarkable (Table 5). This fact can be explained because the ZSM-5 framework and its high acidity would favor the cracking of the n-octane rather than the isomerization reaction to linear products. Moreover, the formation of bulky isomers such as 2,2-DMC6 and 3,3-DMC6 is not favored because of its small pore size.9 Figure 7 shows the thermodynamic equilibrium of normal, monobranched, dibranched, and tribranched octanes versus temperature based on the PCP mechanism (via protonated cyclopropane isomerization mech-
anism) obtained by the software package HYSYS (by Aspentech). This mechanism is ideal because the hydrogenating-dehydrogenating function is enough to equilibrate the zeolites acidity, the reaction in the acid site being the rate-limiting state. HYSYS allows one to obtain isomer distributions versus temperature based on the different reactions that could take place at the experimental conditions. PCP mechanism predicts the following relative ratio among the three methylheptane isomers from n-octane hydroisomerization: 2MC7/3MC7/4-MC7 ) 1/2/1.39 From the selectivities shown in Table 5, a deviation from the predicted values could be observed because the monobranched isomers should be further transformed via type A isomerization reactions (methyl shift) and finally approached the thermodynamic equilibrium. The agglomeration process improved the isomer selectivity in all of the zeolites (Table 4). The metal particles are mainly located on the external zeolite surface for the nonagglomerated ZSM-5 sample, inducing a partial blocking of the micropore mouths. When ZSM-5 zeolite was agglomerated with bentonite, the big metal particles were likely located into the meso- and macropores provided by the clay, avoiding the pore partial blockage, with the consequently lower diffusional constraint of the reactants.11 The acidity also plays an important role in the selectivity to octane isomers. Agglomerated ZSM-5 zeolite without incorporation of the acid function (PdZSM-5/BentNa) showed higher isomer selectivity than the PdZSM-5/Bent sample. The acid sites are responsible for the isomerization of linear alkenes into the corresponding isoalkene via the formation of the secondary carbenium ion. The C8 carbenium ions would yield after isomerization the different octane isomers, whereas the C8 carbonium ions, after disproportionation, would yield propane and pentane. The relative concentration of carbenium and carbonium ions would determine the selectivity of the catalysts.11 A higher strong acid site density involves a higher hydrogen-transfer activity, which leads to a shorter lifetime of carbenium ions and a higher concentration of carbonium ions in the zeolite pores. Therefore, a lower isomerization and a higher disproportionation activity would be obtained if the strong acid site density were increased. The octane isomers selectivity, at approximately 50 mol % conversion (Table 4), allowed us to confirm this point.
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004 8223 Table 5. Product Selectivities (mol %) over Pd-Zeolite Catalysts with and without Binder at Aproximately 50 mol % Conversiona products
PdMORb
Pdβc
PdZSM-5d
PdMOR/ Bente
Pdβ/ Bentf
PdZSM-5/ Bentg
C1 C2 C3 i-C4 n-C4 i-C5 n-C5 2,3-DMC4 2-MC5 3-MC5 n-C6 DMC5 2-MC6 3-MC6 n-C7 2-MC7 3-MC7 4-MC7 3-EC6 2,2-DMC6 2,3-DMC6 2,4-DMC6 2,5-DMC6 3,3-DMC6 3,4-DMC6 3-E,2-MC5
0.1 0.1 13.8 27.3 7.9 14.1 4.6 0.6 0.6 0.6 0.8 0.5 0.3 0.5 0.6 5.0 7.9 3.4 1.6 1.7 1.7 2.4 1.7 0.5 0.9 0.3
0.0 0.0 8.6 25.3 11.9 10.2 4.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.4 11.3 4.0 1.5 2.0 2.1 2.6 2.1 1.0 1.8 1.5
0.1 0.3 20.7 14.8 15.0 9.8 21.6 0.3 0.6 0.5 1.4 0.0 0.0 0.0 0.5 3.6 4.5 1.7 0.7 0.5 0.6 1.0 0.9 0.3 0.5 0.0
0.5 0.6 8.9 17.0 8.5 9.5 3.7 0.0 0.4 0.5 0.7 0.0 0.0 0.0 0.8 12.2 14.8 6.1 1.9 1.4 2.4 3.7 2.9 1.0 1.6 0.8
0.0 0.0 3.0 9.7 4.3 4.8 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 21.6 22.8 8.4 2.7 2.5 3.3 6.3 5.9 1.4 1.8 0.7
0.1 0.3 20.7 14.8 15.0 9.8 21.6 0.3 0.6 0.5 1.4 0.0 0.0 0.0 0.5 3.6 4.5 1.7 0.7 0.5 0.6 1.0 0.9 0.3 0.5 0.0
a Catalyst weight based on zeolite, 0.53 g; H /C ) 14; total pressure, 10 bar; 1 h on stream. b At 330 °C. c At 370 °C. 2 8 350 °C. f At 330 °C. g At 350 °C.
When the acid function was incorporated into the agglomerated sample (PdZSM-5/Bent), the isomer selectivity decreased because the C8 carbonium ions which yield disproportionation products were favored. This way, propane and pentane selectivity increased from 11.2 and 5.6 mol % to 15.0 and 9.4 mol %, respectively. Concerning the mordenite zeolite, side pockets, corresponding to the crossing of the 12MR and 8MR channels, with a highly constrained space, are the preferred sites for the reaction to occur; in this case, even readsorption of the branched products could be highly favored slowing down the diffusion of products. The high acidity of mordenite along with the diffusional constraints of the framework causes the cracking of the intermediates. If mordenite is pelletized with a binder, the meso- and macropores provided by the later would allow a better accommodation of the metal, likely located into the meso- and macropores, allowing a lower diffusional constraint. The selectivity toward branched isomers was increased (Table 4) because of a decrease in the length of the effective diffusional pathway and to the neutralization of acid sites by the binder, avoiding further cracking of these branched products. Because the neutralization of the acid sites in mordenite zeolite was partially inhibited by its pore geometry, the increase in the isomer selectivity was not as high as ZSM-5 zeolite. Comparing agglomerated mordenite zeolite samples (PdMOR/BentNa and PdMOR/Bent), the acid function incorporation produced a selectivity decrease (Table 4), because of the high concentration of the C8 carbonium ions, which yield propane and pentane. The selectivity to these products increased from 7.2 and 2.1 mol % to 8.9 and 3.7 mol %, respectively. The increase in the propane and pentane selectivity was lower than that observed in ZSM-5 zeolite based catalysts. This fact was because the recovery in the acid function was higher in the latter because of fewer diffusional constraints.
d
At 330 °C. e At
In contrast with ZSM-5 and mordenite, the diffusional limitations in the nonagglomerated β zeolite, with a big pore size, are not important. The presence of the binder should not have a positive effect. When β zeolite was agglomerated and the acid function was not incorporated (Pdβ/BentNa), the isomer selectivity increased opposite from that expected (Table 4). As in mordenite and ZSM-5 zeolite, this fact could be due to a higher concentration in C8 carbenium ions, which leads to a higher isomers selectivity. In this case, the incorporation of the acid function should lower the selectivity to branched isomers. However, it can be observed that this selectivity was very similar when Pdβ/Bent and Pdβ/ BentNa were used as catalysts, this selectivity being higher than the nonagglomerated sample (Table 4). It is clear that the high selectivity of β agglomerated zeolite not only depends on the sample acidity. One possible explanation for the high isomerization selectivity of β agglomerated zeolite could be the presence of a large amount of aluminum extraframework species (EFAL).40 Le Van Mao and Saberi41 found that Al3+ incorporated into Pt/HY significantly increased the yield of branched alkanes. They ascribed this behavior to the formation of new desorption-transfer promoting sites, which can rapidly remove the carbocationic intermediates, thus decreasing their residence time on the acidic sites. For this reason, only the monobranched products are clearly favored (Table 4). Both samples, Pdβ/Bent and Pdβ/BentNa, presented EFAL species (Figure 3). Also, a large amount of Lewis acid sites would be due to EFAL species42 and can abstract a hybride from a paraffin, generating a carbenium ion which would yield, after isomerization, the different octane isomers. Because the aim of this work is to obtain branched isomers to increase the octane number of gasolines, it was interesting to observe the selectivity to C5+ products. Isomer products from C5 to C8 contribute and
8224
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004
increase the degree in the octane number of the gasoline. It is clear that Pdβ/Bent and Pdβ/BentNa samples led to the highest C5+ yield. 4. Conclusions The influence of the binder on the acid properties and performance of catalysts based on mordenite, β, and ZSM-5 zeolites in the hydroisomerization of n-octane has been studied. The presence of the binder decreased the zeolite strong acidity because of the solid ion exchange between zeolite protons and clay sodium. The catalysts that presented the biggest selectivity to branched products were those based on β zeolite due fundamentally to its structural characteristics. The selectivity improved considerably when the zeolites were agglomerated because of the presence of meso- and macropores provided by the binder, because the branched products can diffuse with no steric hindrances. The catalytic performance of the β agglomerated zeolite was influenced by EFAL species, which were responsible for the high n-octane conversion and branched isomers selectivity observed. The catalytic activity of mordenite and ZSM-5 zeolites was enhanced by the presence of the binder, which provided meso- and macropores and favored suitable products diffusion. The decrease in the length of the effective diffusional pathway allowed the branched products to migrate from the pores without suffering further cracking reactions. Acknowledgment Financial support by the Ministerio de Ciencia y Tecnologı´a (CICYT) of Spain (Project PPQ2001-1195C02-01) is gratefully acknowledged. Literature Cited (1) Blomsma, E.; Martens, J. A.; Jacobs, P. A. Mechanism of heptane isomerization on bifunctional Pd/H-beta zeolites. J. Catal. 1996, 159, 323. (2) Blomsma, E.; Martens, J. A.; Jacobs, P. A. Reaction mechanism of isomerization and cracking of heptane on Pd/H-Beta zeolite. J. Catal. 1995, 155, 141. (3) Patrigeon, A.; Benazzi, E.; Travers, Ch.; Bernhard, J. Y. Influence of the zeolite structure and acidity on the hydroisomerization of n-heptane. Catal. Today 2001, 65, 149. (4) Blomsma, E.; Martens, J. A.; Jacobs, P. A. Isomerization and hydrocracking of heptane over bimetallic bifunctional PtPd/ H-Beta and PtPd/USY zeolite catalysts. J. Catal. 1997, 165, 241. (5) Me´riaudeau, P.; Tuan, V. A.; Nghiem, V. T.; Lai, S. Y.; Hung, L. N.; Naccache C. SAPO-11, SAPO-31 and SAPO-41 molecular sieves: synthesis, characterization and catalytic properties in n-octane hydroisomerization. J. Catal. 1997, 169, 55. (6) Me´riaudeau, P.; Tuan, V. A.; Nghiem, V. T.; Sapaly, V.; Naccache, C. Comparative evaluation of the catalytic properties of SAPO-31 and ZSM-48 for the hydroisomerization of n-octane: effect of the acidity. J. Catal. 1999, 185, 435. (7) Belloum, M.; Travers, Ch.; Bournonville, J. P. Isome´risation des paraffines de C4 a C7 sur catalyseurs zeolithiques. Rev. Inst. Fr. Pet. 1991, 46, 89. (8) Braum, G.; Fetting, F.; Shoenberger, H. In Molecular Sives II; Katzer, J. R., Ed.; ACS Symposium Series 40; American Chemical Society: Washington, DC, 1977; p 504. (9) Zhang, W.; Smirniotis, P. G. Effect of zeolite structure and acidity on the product selectivity and reaction mechanism for n-octane hydroisomerization and hydrocracking. J. Catal. 1999, 182, 400. (10) Can˜izares, P.; De Lucas, A.; Valverde, J. L.; Dorado, F. n-Butane hydroisomerization over Pd/HZSM-5 catalysts. Palladium loaded by impregnation. Ind. Eng. Chem. Res. 1998, 37, 2592.
(11) Dorado, F.; Romero, R.; Can˜izares, P. Influence of clay binders on the performance of Pd/HZSM-5 catalysts for the hydroisomerization of n-butane. Ind. Eng. Chem. Res. 2001, 40, 3428. (12) Dorado, F.; Romero, R.; Can˜izares, P. Hydroisomerization of n-butane over Pd/ HZSM-5 and Pd/Hβ with and without binder. Appl. Catal. 2002, 236, 235. (13) Rough, S. L.; Wilson, D. I.; Bayly, A.; York, D. Tapping characterization of high shear mixer agglomerates made with ultrahigh viscosity binders. Powder Technol. 2003, 132, 249. (14) Devadas, P.; Kinage, A. K.; Choudhary, V. R. Effect of silica binder on acidity, catalytic activity and deactivation die to coking in propane aromatisation over H-gallosilicate (MFI). Stud. Surf. Sci. Catal. 1998, 113, 425. (15) Wu, X.; Alkhawaldeh, A.; Anthony, R. G. Investigation on acidity of zeolites bound with silica and alumina. Stud. Surf. Sci. Catal. 2002, 143, 217. (16) Chen, W. H.; Huang, S. J.; Lai, C. S.; Tsai, T. C.; Lee, H. K.; Liu, S. B. Effects of binder, coking and regeneration on acid properties of H-mordenite during TPD reaction. Res. Chem. Intermed. 2003, 29, 761. (17) Uguina, M. A.; Sotelo, J. L.; Serrano, D P. Toluene disproportionation over ZSM-5 zeolite: effects of crystal size, silicon-to-aluminium ratio, activation method and pelletization. Appl. Catal. 1991, 76, 183. (18) Horvath, R.; Kawazoe, K. J. Method for the calculation of effective pore size distribution in molecular sieve carbon. Chem. Eng. Jpn. 1983, 16, 470. (19) Barret, R. M.; Joyner, L. G.; Halenda, P. P. The determination of pore volume and area distribution in porous substances. J. Am. Chem. Soc. 1951, 73, 373. (20) Rodrı´guez, F.; Rodrı´guez, I.; Moreno, C.; Guerrero, A.; Lo´pez, J. D. Pt catalysts supported on activated carbons (I). Preparation and characterization. J. Catal. 1986, 99, 171. (21) Bond, G. C.; Mallat, T. Studies of Hydrogen Spill over. J. Chem. Soc., Faraday Trans. 1981, 77, 1743. (22) Can˜izares, P.; De Lucas, A.; Dorado, F.; Dura´n, A.; Asencio, I. Characterization of Ni and Pd supported on H-mordenite catalysts: Influence of the metal loading method. Appl. Catal. 1998, 169, 137. (23) Choudhary, V. R.; Devadas, P.; Kinage, A. K.; Guisnet, M. Influence of binder on the acidity ad performance of H-gallasilicate (MFI) zeolite in propane aromatisation. Appl. Catal. 1997, 162, 223. (24) Romero, M. D.; Calles, J. A.; Rodrı´guez, A.; De Lucas, A. Acidity modification during the agglomeration of ZSM-5 with montmorillonite. Microporous Mesoporous Mater. 1997, 9, 221. (25) Yoon, K. B.; Park, Y. S. Visual observation of salt migration between zeolite and clay in the solid state. J. Chem. Soc., Chem. Commun. 1993, 522. (26) Choudhary, V. R.; Nayak, V. S. Effect of degree of proton exchange on the acidity distribution of Hna-ZSM-5. Zeolites 1985, 5, 15. (27) Mu¨ller, M.; Harvey, G.; Prins, R. Comparison of dealumination of zeolites beta, mordenite, ZSM-5 and ferrierite by thermal treatment, leaching with oxalic acid and treatment with SiCl4 by 1H, 29Si and 27Al MAS NMR. Microporous Mesoporous Mater. 2000, 34, 135. (28) Fernandes, L. D.; Monteiro, J. L. F.; Sousa-Aguiar, E. F.; Martinez, S. A.; Corma, A. Ethylbenzene hydroisomerization over bifunctional zeolite based catalysts: the influence of framework and extraframework composition and zeolite structure. J. Catal. 1998, 177, 363. (29) Bourgeat-Lami, E.; Massiani, P.; Di Renzo, F.; Espiau, P.; Fajula, F. Study of the state of aluminium in zeolite-β. Appl. Catal. 1991, 72, 139. (30) Kiricsi, I.; Flego, C.; Pazzuconi, G.; Parker, W. O.; Millini, R.; Perego, C.; Bellussi, G. Progress toward understanding zeolite beta acidity: An IR and 27Al NMR spectroscopic study. J. Phys. Chem. 1994, 98, 4627. (31) Jasra, R. V.; Tyagi, B.; Badheka, Y. M.; Choudary, V. N.; Bhat, T. S. G. Effect of clay binder on sorption and catalytic properties of zeolite pellets. Ind. Eng. Chem. Res. 2003, 42, 3263. (32) Fougerit, J. M.; Gnep, N. S.; Guisnet, M.; Amigues, P. Effect of the binder on the properties of a mordenite catalyst for the selective conversion of methanol into light olefins. Stud. Surf. Sci. Catal. 1994, 84, 1723.
Ind. Eng. Chem. Res., Vol. 43, No. 26, 2004 8225 (33) Ka¨rger, J.; Petzold, M.; Pfeifer, H.; Ernst, S.; Weikamp, J. Single-File Diffusion and Reaction in Zeolites. J. Catal. 1992, 136, 283. (34) Kunkeler, P. J.; Zuurdeeg, B. J.; van der Waal, J. C.; van Bokhoven, J. A. Zeolite Beta: the relationship between calcination procedure, aluminium configuration and Lewis acidity. J. Catal. 1998, 180, 234. (35) Berreghis, A.; Ayrault, P.; Fromentin, E.; Guisnet, M. Acetylation of 2-methoxynaphthalene with acetic anhydride over a series of dealuminated HBEA zeolites. Catal. Lett. 2000, 68, 121. (36) Heinichen, H. K.; Ho¨lderich, W. F. Acylation of 2-Methoxynaphthalene in the presence of modified zeolite HBEA. J. Catal. 1999, 185, 408. (37) Chica, A.; Corma, A. Hydroisomerization of Pentane, hexane, and Heptane for improving the octane number of gasoline. J. Catal. 1999, 187, 167. (38) Chica, A.; Corma, A.; Miguel, P. J. Isomerization of C5C7 n-alkanes on unidirectional large pore zeolites: activity, selectivity and adsorption features. Catal. Today 2001, 65, 101.
(39) Martens, J. A.; Jacobs, P. A. In Theoretical Aspects of Heterogeneous Catalysis; Moffat, J. B., Ed.; Van Nostrand Reinhold: New York, 1990; p 52. (40) Kinger, G.; Vinek, H. n-Nonane hydroconversion on Ni and Pt containing HMFI, HMOR and HBEA. Appl. Catal. 2001, 218, 139. (41) Le Van Mao, R.; Saberi, M. A. Catalysts for the hydroisomerization of n-heptane, prepared according to the concept of ‘triangular’ site configuration (acid/metal/desorption-transfer promoting sites). Appl. Catal. 2000, 199, 99. (42) Corma, A.; Forne´s, V.; Navarro, M. T.; Pe´rez-Pariente, J. Acidity and stability of MCM-41 crystalline aluminosilicates. J. Catal. 1994, 148, 569.
Received for review April 29, 2004 Revised manuscript received October 7, 2004 Accepted October 7, 2004 IE040133J
