Ind. Eng. Chem. Res. 2002, 41, 1973-1985
1973
In Situ Study of Alkane Conversion on Pt-Loaded Acidic Zeolites Niels J. Noordhoek,† Danny Schuring,‡ Frank J. M. M. de Gauw,‡ Bruce G. Anderson,‡ Arthur M. de Jong,*,† Martien J. A. de Voigt,† and Rutger A. van Santen‡ Accelerator Laboratory, Department of Applied Physics, and Department of Inorganic Chemistry and Catalysis, Schuit Institute of Catalysis, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
The hydroisomerization of hexane on platinum-loaded acidic zeolites was studied with positron emission profiling (PEP), a tracer imaging technique based on the same principle as positron emission tomography. The unique character of PEP enables the determination of reaction parameters under in situ reaction conditions that are difficult to obtain via ex situ techniques. A numerical model, including the effects of adsorption, diffusion, and reaction, was used to fit the measured concentration profiles. Preexponential factors and activation energies for some of the elementary proton-activated reaction steps of the bifunctional reaction mechanism have been determined in different transient experiments for a variety of platinum-loaded acidic zeolites. An activation energy for dehydrogenation of 98 kJ/mol was found, in which the heat of adsorption appeared to be included. The heat of adsorption was found to be 61 kJ/mol. An upper value for the deprotonation energy of 100 kJ/mol was determined. This is in agreement with quantum mechanical calculations that have become available only recently. Introduction The hydroisomerization reaction has become an important process in the oil refining industry. It is used to convert linear alkanes to their branched isomers, which are added to fuel mixtures to improve their octane rating.1,2 A higher-octane fuel has better knock resistance, which is needed in modern high compression ratio engines. Knock is the spontaneous and violent ignition of the fuel/air mixture that results in wear and decreased fuel economy. Stringent environmental regulations have increased the interest for the hydroisomerization reaction because alternative additives such as lead compounds [tetraethyl lead (TEL)] and aromatics are toxic. For example, the octane rating of linear n-hexane is 25 (RON), while its branched isomer has an octane rating of 104 (RON). This paper presents the results of an in situ study of the hydroisomerization reaction under typical operating conditions by positron emission profiling (PEP). The aim of our investigation was to image the carbonaceous reaction profile that is generated during the reaction on the catalyst and to probe the reactivity of the carbonaceous residues. As a test reaction, the hydroisomerization of n-hexane on Pt-loaded acidic zeolites was used. The transport parameters of the reactants were determined in an earlier study.3,4 When the mass transport model is simplified3,4 and it is extended with terms for chemical reaction steps in the zeolite, the different types of experiments can be modeled numerically. The results were fitted to the experimental concentration profiles in an attempt to obtain experimental values for the kinetic parameters of the elementary steps of the hydroisomerization process. The unique in situ character of PEP provides * Corresponding author. E-mail:
[email protected]. Fax: +31(0)402438060. † Accelerator Laboratory, Department of Applied Physics. ‡ Department of Inorganic Chemistry and Catalysis.
experimental backup for mechanistic understanding of acidic zeolite catalyzed hydrocarbon conversions. In addition, the kinetic values are of great importance to the design of catalytic processes because they dictate the temperatures, pressures, and geometries of the reactors and processes in practice. Hydroisomerization Reaction Mechanism. Alkanes can be isomerized with high conversions at relatively low temperature by the use of Pt-loaded acidic zeolites. This combination of a transition metal and an acidic oxide zeolite results in a process in which olefinic intermediates are converted to isomers. Mills et al.5 introduced the concept of separate catalytic functions: the (de)hydrogenation activity operative on the Pt sites and the isomerization activity associated with the acid sites. Figure 1 shows the reaction scheme for the hydroisomerization of hexane and Figure 2 the corresponding energy diagram. The platinum serves to convert the hexane into hexene. Hexene is protonated on the acidic site to form a carbenium ion. This happens through a π state, in which the electrons of the double bond are attracted by the proton of the acid site. Subsequently, a σ bond can be formed when the double bond is used to form a covalent bond with the oxygen atom of the acid site and the proton becomes a part of the newly formed alkoxy species. In this protonated form, the hydrocarbon can branch and is desorbed as an isohexene. By hydrogenation on the Pt sites, the end product, isohexane, is obtained.6,7 Although the thermodynamic equilibrium for dehydrogenation is unfavorable, a significant conversion can be obtained through the close presence of the acid sites. The dehydrogenation equilibrium is very fast, and because the transport to the acid sites is fast, high conversion rates can be obtained. These conversions are much higher than can be expected from a reactor with separate dehydrogenation and acid function. The overall reaction rate of this nontrivial polystep reaction is very high, even though the intermediate
10.1021/ie010455c CCC: $22.00 © 2002 American Chemical Society Published on Web 03/14/2002
1974
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002
Figure 1. Schematic representation of Weisz’ bifunctional mechanism. It shows the (de)hydrogenation function of the Pt site and the formation of an alkoxy on the acid site. Chemisorbed on the acid site, the alkoxy can isomerize through a cyclic intermediate form.
Figure 2. Energy diagram of the hydroisomerization reaction mechanism. Some typical values for hexane are given. All values are in kilojoules per mole. These values are only intended to indicate their approximate values. The activation energy for deprotonation has not been determined experimentally before.
concentration of hexenes is very low. It is essential that the Pt and acid sites are so close that transport between the two sites does not limit the conversion rate. Weisz1 calls this the “intimacy criterion”. An alternative reaction mechanism is presented by Chu et al.8 in which an acid-catalyzed chain reaction involving methyl shifts and hydrogen transfer is responsible for the isomerization mechanism. This was based on at least three observations. First, they observed that pure β-H is only 10% as active as the Pt loaded β-H. However, note that in the Weisz mechanism the Pt dehydrogenation function is essential. Second, they observed that the addition of 1-hexene decreased the conversion and selectivity. This can also be understood from the fact that the catalyst will deactivate rapidly at high alkene concentrations by oligomerization of the alkenes. Besides blocking of the active sites, also pore blocking will result, favoring the more linear isomers. Third, they observed that the equilibrium concentration of hexenes is very small. This, however, agrees well with the statement of Weisz that, if the intimacy requirement is met, “polystep reactions may easily proceed with intermediates at concentrations far below the limit of experimental detection”. More alternative theoretical mechanisms for the hydrocarbon conversion exist. Side Reactions. The Pt site, besides performing (de)hydrogenation, also cracks the alkanes into smaller fragments. Hexane preferentially cracks to C1 + C5 on Pt.9 In addition, coke formation occurs on the Pt sites. By multiple removals of hydrogen atoms, carbonaceous species are formed, which cover the platinum. This can be advantageous for the selectivity of the isomerization process because it suppresses metal-induced cracking
by limiting the number of neighboring platinum sites. Important to note is that the dehydrogenation reaction itself is not structure sensitive (strictly, ensemble-sizeinsensitive),10-12 but further dehydrogenation, leading to the formation of carbonaceous residues, needs larger ensembles. In almost every study, it is found that the formation of carbonaceous species is suppressed more strongly than dehydrogenation when the concentration of an inert element is increased. Tin (Sn), for instance, was found to be an effective modifier for Pt.13,14 The acid site can also crack hexane molecules through β scission. This involves the migration of the two electrons of a β-C-C bond toward a C+ atom to form an olefinic π bond, with the β-carbon atom (with respect to the original C+ atom) being left as the electron-deficient carbon of a smaller carbenium ion. The other product is an olefin that can be protonated by a different acid site or desorbs. The rate of β scission depends on the relative stability of the formed ions. Because the stability order of carbenium ions is tertiary > secondary > primary, the product distribution for hexane cracking has a preference for C3 + C3 products. Because of this difference in product distribution of metal and acid cracking, the ratio of C1 and C3 products yields information on the relative importance of these side reactions. Also fracture into single C atoms, polymerization, and the formation of amorphous carbon or graphite can occur. Because these products block access to the active sites of the catalyst, they have a deactivating effect. Experimental Section Experimental Setup and 11C Labeling. The radionuclide carbon-11, a positron emitter, is produced using the Eindhoven 30 MeV AVF cyclotron by irradiating a target filled with nitrogen gas with 12 MeV protons. This results in the production of 11CO2, which is subsequently converted to 11C-labeled hexane with a two-step alkene homologation reaction. Details about this production procedure have been published previously.3 The PEP detector reconstructs the position of the radio-labeled molecules in tubular reactors.15 When a radio-labeled hexane molecule decays in the reactor, a positron is emitted. Upon an encounter with an electron, the antiparticle of the positron, both particles are annihilated. During the annihilation event, two γ photons are emitted in opposite directions. The PEP detector surrounding the reactor reconstructs the position of the decaying molecule by coincident detection of the two photons. The detector consists of two arrays of
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 1975 Table 1. Overview of the Investigated Zeolites and Their Most Relevant Propertiesa structure code
direction, T atoms, ring and pore size (Å)
H-Beta
*BEA
H-ZSM5
MFI
H-Mordenite
MOR
H-ZSM22
TON
[001], 12, 5.5 × 5.6 [100], 12, 7.6 × 6.4 [010], 10, 5.3 × 5.6 [100], 10, 5.1 × 5.5 [001], 12, 6.5 × 7.0 [010], 8, 2.6 × 5.7 [001], 10, 4.4 × 5.5
zeolite
Si/Al
KH of n-C6 at 573 K
∆Hads of n-C6 (kJ/mol)
crystal size (µm)
wt % Pt
12.5
8 × 10-5
63
1
1.6
28
1 × 10-5
69
3
0.5
10
8 × 10-5
67
5
2
28
3 × 10-6
75
1
0.3
a
given.16
For hexane the Henry constant, heat of adsorption, diffusion constant, and activation energy for diffusion are The zeolites have been ordered from wide pore, two-dimensional to narrow pore, one-dimensional, which is a direct measure for the mobility of the alkanes in the respective type of structure. Though Mordenite, in principle, has a two-dimensional pore structure, effectively it is onedimensional because only the [001] channel is larger that the diameter of hexane (4.5 Å). Table 2. Molecular Dimensions of Hydrocarbon Adsorbates at Room Temperature, Estimated from Courtauld Space-Filling Models17 pore size dependent mobility in zeolite typea molecule
molecule dimensions
Beta
ZSM5
Mordenite
ZSM22
n-hexane 2-methylpentane 2,2-dimethylbutane
3.9 × 4.3 × 9.1 4.6 × 5.8 × 8.6 5.9 × 6.2 × 6.7
3D 1D 1D
3D 0D 0D
1D 1D 1D
1D 0D 0D
a The first two parameters of the dimensions are obviously critical for the transport through the zeolite channels.
nine independent detection elements each. The spatial resolution of the detector is 3 mm. Temporal information is obtained by collecting data during a fixed sampling period. A minimum sampling time of 0.5 s is required to obtain sufficient coincident events for reliable statistics. To reduce errors resulting from the detection of Compton scattered photons, which lead to anomalously positioned annihilation events, energy selection of the photons is also employed. The experiments were carried out on a range of Ptloaded zeolites and Pt/alumina. To reduce the pressure drop across the reactor bed, the crystals were pressed into pellets with diameters ranging from 250 to 500 µm. Some 300 mg of zeolite or alumina was placed in a single-pass flow reactor system with an internal diameter of 4 mm. This reactor was placed inside a tubular furnace, capable of reducing temperature gradients along the reactor to less than 0.1 °C. The experiments were performed at temperatures between 130 and 450 °C and at atmospheric pressure. Zeolites, Pt Loading, and Characterization. The following acidic zeolites have been studied: H-ZSM5, HZSM22, H-Mordenite, and H-Beta. The main differences are pore structure and pore size and their corresponding adsorptive capacity, heat of adsorption, and diffusion properties of alkanes that are applied to them. Table 1 gives an overview of the zeolites used and their properties. Table 2 gives the molecular dimensions of alkanes and the resulting dimensionality of the zeolites. The preparation of the zeolite samples was as follows.18,19 A batch of Na-zeolite is first calcined at 550 °C to remove the template. The H-zeolite was formed by triple exchange with NH4NO3 followed by calcination in dry air at 500 °C. [Pt(NH3)4H-zeolite] was prepared by ion exchange following literature methods.20 A dilute solution of [Pt(NH3)4](OH2) was added as droplets to a stirred H-zeolite slurry. After 24 h of stirring at room temperature, the slurry was filtered and washed twice with doubly deionized water. The stoichiometric composition of the samples was determined using atomic adsorption spectrometry. In this way, the Si/Al ratio and the amount of Pt in the
sample can be determined. A measure for the dispersion of Pt was obtained from hydrogen adsorption. The zeolite crystal size was determined with a scanning electron microscope. Experimental Procedures. All zeolite samples were reduced at 400 °C in hydrogen to remove any alkane species from the zeolite and Pt surfaces. For some experiments, the samples were used as such, i.e., in the reduced form. Other experiments, described as being performed on a preconditioned sample, are conducted on reduced samples on which a steady-state isomerization of C6/H2 at 240 °C has been running before the labeled alkanes are injected. This preconditioning results in a partial coverage of the Pt surface and thus prevents further cracking of the injected hexane, as mentioned in the Side Reactions section. The transient experiments consist of two phases which we call injection and deprotonation. The injection experiment consists of an injection of a pulse of 11Clabeled hexane into a He flow (without a steady-state flow of unlabeled hexane). In principle, all injected hexane molecules either remain irreversibly chemisorbed on the acid sites in a protonated form (the Pt has catalyzed their conversion to an olefin) or leave the reactor without being protonated as they remain in the paraffinic form. This chemisorption is, of course, not completely irreversible. In the absence of hydrogen though, the deprotonated species will not hydrogenate on the Pt sites. The olefin will directly be protonated again at another acid site. Thus, the equilibrium is strongly biased toward protonation. In the absence of hydrogen, the relatively high alkene concentration favors oligomerization and alkylation reactions on the acid sites. The deprotonation experiment is performed after an injection experiment, with the labeled alkanes still adsorbed in the reactor, by replacing the helium by a hydrogen flow. The chemical equilibrium then shifts toward the paraffinic form of C6, which cannot be protonated. Ideally, all alkanes formed will subsequently leave the reactor. These deprotonation reaction experiments have been performed with a constantly rising temperature, referred to as TPR experiments (temperature-programmed reaction), and at a constant temperature, called CTR experiments (constant temperature reaction). Results and Discussion General Shape of the Measured PEP Image. The general shape of a PEP image, after injection of a pulse of a 11C-labeled alkane without reaction, is shown in Figure 3. Because the pulse was injected on a sample without Pt, the dehydrogenation step is absent and no
1976
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 Table 3. Henry Constants on Three Different Zeolites at 300 °C for n-C5 to n-C8 and Some of Their Isomersa K′(mol/kgPa) n-pentane 2-methylbutane n-hexane 2-methylpentane n-heptane 2-methylhexane n-octane
ZSM-5
MOR
ZSM-22
4.7 × 10-6 3.3 × 10-6 9.7 × 10-6 6.7 × 10-6 2.0 × 10-5 1.1 × 10-5 3.9 × 10-5
2.2 × 10-5 1.4 × 10-5 7.7 × 10-5 4.0 × 10-5 2.7 × 10-4 1.4 × 10-4 9.0 × 10-4
1.3 × 10-6 3.2 × 10-7 2.6 × 10-6 5.4 × 10-7 4.7 × 10-6 8.7 × 10-7 8.8 × 10-6
a Together with eq 1, this gives an estimate for the relative retention times of alkanes in the packed bed.
Figure 3. Injection of a pulse of 11C6 in hydrogen on H-Beta without Pt at 150 °C. The pulse passes without reacting and with a retention time as given by eq 1, which mainly is a function of adsorption, expressed by the Henry constant KH. The broadening of the peak is determined by axial dispersion and macropore diffusion limitation (eq 2). Details about these equations can be found in ref 2.
reaction occurs. The pulse, injected in a hydrogen flow of 150 mL/min, has a full width half-maximum (fwhm) of 2 s. The pulse is delayed by adsorption, resulting in the retention time µ given by the following equation:
µ)
Lbed [ + (1 - z)y + (1 - z)(1 - y)x(1 + Ka)] vsup z (1)
A typical value for µ is 100 s. The broadening of the pulse is σ2 (eq 2), in which the first term accounts for axial dispersion, the second for film transfer, the third for macropore diffusion, and the fourth for micropore diffusion. Under our experimental conditions, axial dispersion and macropore diffusion are the dominant terms for the large-pore zeolites (>5 Å) with small crystal diameters ( 20 the LDF model works very well. The dimensionless bed length for our system was >30, and the LDF approximation can thus be safely used. Because the PEP detector images all phases within the column, both the gas phase in the bed and the average sorbate concentration contribute to the measured PEP profiles. The measured concentration at point z is therefore given by
〈C(zi,t)〉 ) zCz + (1 - z)q j
(7)
The PEP detector, however, measures the concentration in a finite volume ∆V associated with the position resolution of the detection system. The measured concentration at detector position i then equals
Cmeas(zi,t) )
πRtube2 ∆V
+(1/2)∆z 〈C(z,t)〉 dz ∫zz-(1/2)∆z i
i
(8)
in which Rtube is the radius of the reactor tube and ∆z the width of the detection volume. Because of the discrete nature of the solution, this equation was integrated numerically using the rectangle rule. Reaction Mechanism in the Absence of Hydrogen. The isomerization reaction is normally carried out by feeding a mixture of hydrogen and n-hexane over a platinum-loaded zeolite at temperatures of 240-300 °C. Hydrogen is needed for the hydrogenation of the nhexene and isohexene species formed on the catalyst surface. When the hydrogen is replaced with a helium flow, this last reaction step is prohibited, resulting in the following reaction scheme: Pt
H+
n-C6H14 98 n-C6H12 + H2 798 n-C6H13+ T H+
i-C6H13+ 798 i-C6H12 (9) Once the adsorbed n-hexane is dehydrogenated on a platinum site, it is trapped on the catalyst surface. Chemisorption on the platinum sites, physisorption on the zeolite, and chemisorption as an alkoxy species on the acid sites are possible. The same applies for isohexene, which is formed by isomerization of the n-alkoxy species. Because of the strength of the adsorption of the alkoxy species, the majority will be adsorbed on the acid sites. Although adsorbed species can travel to neighboring sites, these distances are short relative to the resolution of the detector. Effectively, the dehydrogenation step thus determines the activity distribution in the reactor. The catalytic properties of platinum in the dehydrogenation of hydrocarbons have been known for many years. An extensive amount of research has been carried out on this subject. The dehydrogenation reaction can be described with a three-step reaction mechanism, called the Horiuti-Polanyi mechanism.25 First, an adsorbed n-hexane molecule is transported to a nearby platinum site:
ZEO-n-C6H14 + Pt T Pt-n-C6H14 + ZEO (10) The hydrogen atoms are then extracted further one by one:
Pt-n-C6H14 + Pt T Pt-n-C6H13 + Pt-H (11) Pt-n-C6H13 + 2Pt f Pt2-n-C6H12 + Pt-H (12) Associative desorption of the hydrogen then occurs, after which the hydrogen is quickly removed from the reactor. A somewhat simpler reaction mechanism was used by Van de Runstraat19 to model the overall reaction process. In this model, the dehydrogenation was assumed to proceed through a one-step mechanism:
Pt-n-C6H14 + Pt f Pt2-n-C6H12 + H2
(13)
Modeling the Dehydrogenation Experiments. Equation 13 gives a simplified reaction scheme for the dehydrogenation of hexane on the platinum sites. A further simplification can be made when the amount of available platinum sites greatly exceeds the amount of reacting species or the regeneration of these platinum sites is fast. In this case, the number of available sites can be treated as constant, and the reaction reduces to an irreversible reaction from adsorbed n-hexane to n-hexene: Pt
n-C6H14(ads) 98 n-C6H12(ads) + H2
(14)
The n-hexene formed on the platinum sites can then further react to form zeolitic alkoxy species. Hydrogenolysis reactions, leading mainly to methane and pentane, are not taken into account. Because the catalyst has been preconditioned by exposure to unlabeled n-hexane/hydrogen mixtures for at least 30 min prior to injection of the labeled pulse into helium, a sufficient carbonaceous overlayer exists on the catalyst to prevent these cracking reactions.9 To incorporate reaction equation (13) in the model, the LDF equations have to be extended to include an additional immobilized species, whose concentration will be denoted by Cim. The change of the total gas and adsorbed phase concentration in the pellets is now determined by the flux from the fluid phase and by the rate of reaction within the micropores:
Kp
∂Cp ∂Cim ) k(Cz - Cp) - (1 - y)x ∂t ∂t
(15)
The rate of formation of the immobilized species is proportional to the concentration of adsorbed n-hexane and is given by the rate equation:
∂Cim ) kim[C6H14(ads)] ) kimKaCp ∂t
(16)
The reaction constant kim accounts for the transport step from the zeolite adsorption site to the platinum site as the actual dehydrogenation reaction. The transport step is much faster than the dehydrogenation step, and the first step will thus be equilibrated. The rate of formation of n-hexene, catalyzed by Pt, for the complete reaction described by eqs 10 and 13 can be written as
dθPt2C6H12 dt
) kfdehKtransθPt2phexane
(17)
in which kfdeh is the reaction constant for eq 12 (in mol/ f b /ktrans is the equilibrium conm2‚s) and Ktrans ) ktrans stant for transport between the zeolite adsorption site
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 1979 f b and the platinum site. ktrans and ktrans are the reaction constants for the forward and backward reactions in eq 10. Modeling the Deprotonation Reaction. When hydrogen is added, any hexenes formed by the deprotonation of the carbocation intermediates are rapidly hydrogenated to hexane on the available platinum sites. The deprotonation of the trapped hydrocarbons is, therefore, driven to completion, resulting in the release of 11C-labeled hexane molecules, and the following reaction sequence occurs:
C6H13+(acid) f C6H12(Pt) + H2 f C6H14(ads) f C6H14(g) (18) Once the hexane is formed, it is quickly removed from the reactor in the hydrogen flow. The species formed on the acid sites are very stable, resulting in a high energy barrier for the removal of these species. Therefore, the deprotonation of these carbocations is most likely to be the rate-determining step in the reaction sequence. Assuming that the deprotonation step is rate-determining, the entire reaction can be regarded as a single irreversible unimolecular desorption process. The rate of desorption of the adsorbed species can then be described by a first-order rate equation:
∂Cim ) -kd(T) Cim ∂t
(19)
(20)
in which ν is the preexponential factor, Ea,d is the activation energy for deprotonation, and Rg is the gas constant. The model now has to be modified to account for the desorption of trapped reaction products:
∂Cp ∂Cim ∂q j ) Kp ) k(Cz - Cp) - (1 - y)x (21) ∂t ∂t ∂t At the start of the experiment, only adsorbed carbocations are present in the bed, and Cim is given by
Cim(z,0) ) f(z) in which f(z) is the initial (t ) 0) activity distribution which can be measured using the PEP detector. The initial desorption rate equals
∂Cim (z,0) ) -kdCim(z,0) ∂t
(22)
The change of the pellet concentration at t ) 0 is determined by the initial rate of desorption, and the change of the fluid phase concentration equals zero because no concentration gradients yet exist:
Kp
∂Cp (z,0) ) (1 - y)xkdCim(z,0) ∂t
(23)
(24)
Given the initial activity distribution, f(z), these equations can then be solved using the numerical method of lines. Reaction Profiles. Obviously, to obtain kinetic results from the PEP experiments, it is essential that the shape of the profiles is to a great extent determined by the reaction. An elegant way to analyze this would be to set up an equation for the first and second moments of the measured profiles. Then, simply by comparing the relative contributions from mass transport and reaction kinetics to these moments, the influence of reaction kinetics can be calculated. For the injection experiments, it is obvious from the difference between the images without and with reaction that the reaction has a significant influence on the shape of the profiles. Because the estimates for the deprotonation energy vary over a range from 140 kJ/mol in older publications26,27 to 47 kJ/mol in more recent publications,28 it is useful to develop a tool for estimating the relative importance of mass transport and reaction kinetics. This can be done by comparing the retention time to the reciprocal of the rate of deprotonation. When a firstorder desorption process is assumed for the deprotonation reaction, a measure for the rate would be of the form
r ) ν exp(-Ea,d/kT)
in which kd(T) is the rate constant for deprotonation and Cim is the concentration of the adsorbed carbocations. The temperature dependence of the rate constant is given by the Arrhenius equation:
kd(T) ) ν exp(-Ea,d/RgT)
∂Cz (z,0) ) 0 ∂t
(25)
Figure 7 shows a graph that is an indication for the range of ν and Ea,d, which will result in a reaction that will occur on a time scale of seconds rather than milliseconds or hours. The isotherms connect points for which the combination of the preexponential factor and activation energy results in a practical rate of reaction for PEP at the same temperature. The boundary condition is the minimum binning time of the PEP detector. Currently only one concentration profile can be made per 0.5 s. This value depends on the maximum count rate of the PEP detector; a minimum of counts within the binning time is needed for determining a concentration profile with sufficient accuracy. As a guideline, an experiment with the reaction completing in more than at least 10 binning times is useful for interpretation. The second boundary condition is that the time scale of the transport to the reactor exit should be shorter than the time scale of the reaction process. Results of the Dehydrogenation Experiments. The dehydrogenation model was used to fit the data obtained during the injection experiments. The reaction constant, kdeh, and the adsorption constant were used as fit parameters. The experiments were performed on 0.5 wt % Pt/H-Mordenite, and particular care was taken that the preconditioning was equal for each experiment. Measurements were performed at four different temperatures, and the temperature dependence of the fitted reaction constant was used to determine the activation energy. The resulting Arrhenius plot is shown in Figure 8. As an example, simulated and measured results are shown in Figure 9a-d. The images show the injection of a pulse of 11C-labeled n-hexane in a 75 mL/min helium flow on a 2 wt % Pt/ H-Mordenite sample at 220 and 260 °C, respectively.
1980
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002
Figure 7. From this graph the range of preexponential factors and activation energies for deprotonation can be derived, which will result in a value of the reaction rate that is matched to the time resolution of the PEP detector. Each line represents a set of activation energies and preexponential factors at a fixed temperature that will result in a measurable rate. The numbers next to the lines indicate the temperature, and the plus signs above each line indicate to which line the temperature applies. For example, when experiments are performed at 200 °C and the preexponential factor is estimated to be 1013, the graph shows that the activation energy has to be at least 120 kJ/mol to get a useful PEP experiment.
Figure 9. (a) Measured PEP image of an injection of 11C-labeled n-hexane in a helium flow of 75 mL/min at 220 °C on 2 wt % Pt/ H-Mordenite. (b) Simulated PEP image of an injection of 11Clabeled n-hexane in a helium flow of 75 mL/min at 220 °C on 2 wt % Pt/H-Mordenite. (c) Measured PEP image of an injection of 11Clabeled n-hexane in a helium flow of 75 mL/min at 260 °C on 2 wt % Pt/H-Mordenite. (d) Simulated PEP image of an injection of 11C-labeled n-hexane in a helium flow of 75 mL/min at 260 °C on 2 wt % Pt/H-Mordenite. Figure 8. Arrhenius plot for the reaction constant for dehydrogenation kdeh, for 0.5 wt % Pt H-Mordenite.
From the slope of the Arrhenius plot (Figure 8), a value of 98 ( 15 kJ/mol was found for the activation energy of the dehydrogenation reaction, which in essence is the effective activation energy for alkoxy formation. Van de Runstraat29 gives an activation energy of 40-55 kJ/mol for the dehydrogenation reaction given in eq 13, depending on the alkene formed after dehydrogenation. This value is significantly lower than that obtained in this study. The higher value obtained from PEP than used in the literature may be due to the fact that the PEP measurements refer to activation energies with respect to the adsorbed state. In that case this higher value is due to the heat of adsorption that actually has to be subtracted in order to obtain an apparent activation energy. This difference can also be explained from the side reactions that affect the measured value. Coke formation after dehydrogenation is not a problem because this also results in an irreversible adsorption, but hydrogenolysis is likely to affect the measurements and modeling parameters. The short product of hydrogenolysis will have a much smaller retention time than hexane and,
therefore, produces a very different distribution in the zeolite bed. Results of the Deprotonation Experiments Temperature-Programmed Reaction (TPR) Experiments. In the classical TPR experiment, the reactor is heated to create a linear rise in the temperature as a function of time, while the effluent stream is monitored as a function of the reactor temperature. Ideally, with a high carrier gas flow rate, the response is proportional to the rate of desorption, if diffusion and readsorption are not limiting.30 Then the activation energy can be obtained from
2 ln(Tmaximum) - ln(β) )
Ed RTmaximum
( )
+ ln
Ed (26) Rνdes
In this equation Tmaximum is the temperature of maximum desorption, β is the heating rate (in K/s), R is the gas constant, Ed is the activation energy for desorption, and νdes is the preexponential factor for desorption. When the left-hand side of this equation is plotted against 1/Tmaximum for different heating rates, a value
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 1981
for the activation energy can even be obtained without having to assume some preexponential factor. TPR experiments were carried out after injection in helium at 240 °C at different temperature ramp rates. From a plot of the total amount of radioactivity in the bed as a function of temperature, Tmaximum was obtained. For H-Mordenite a value of 61 ( 8 kJ/mol was obtained, and similar results were obtained on H-Y, H-Beta, and H-ZSM5. However, in the case of readsorption, the product may readsorb on the downstream catalyst before it is swept free of the bed by the carrier gas stream, resulting in an incorrect value of the activation energy because readsorption shifts Tmaximum to higher temperatures. Demmin and Gorte31 devised a set of design parameters for “minimizing difficulties” in temperature-programmed desorption (TPD) experiments. These experiments were aimed at obtaining heats of adsorption in packed beds and are, therefore, fundamentally different from our TPR experiment in which desorption from the bed is preceded by a deprotonation step. This deprotonation step has to be rate determining. What can be learned from the design parameters from Demmin and Gorte31 is that our system is far from ideal for performing TPD experiments. Consequently, there are two possibilities: either transport is rate-determining and a value (influenced by readsorption) of the heat of adsorption results or deprotonation is rate-determining and possibly a correct value for the activation energy for deprotonation is found. More clarity on this can be obtained when an estimate is made of the rate of reaction with an activation energy that is close to the value of 61 kJ/ mol. A preexponential factor has to be assumed. A typical value for the deprotonation reaction would be 1013-1015 s-1. This results in a rate of
r ) ν exp(Ea/RT)
Figure 10. Logarithmic plot of the amount of desorbing species from the reduced H-Mordenite in the CTR experiments for three different temperatures.
Figure 11. Arrhenius plot for the CTR experiments on reduced H-Mordenite.
(27)
which has a value of at least 8 × 106 s-1 at 240 °C. This indicates that deprotonation takes place on a much smaller time scale than the transport to the exit of the reactor, which is on the order of 100 s. The obtained value of 61 kJ/mol must, therefore, be related to the heat of adsorption, probably affected by readsorption. Literature4,16,21 is quite unanimous that the real value for the heat of adsorption of n-hexane on Mordenite is 67 kJ/mol and the value of isohexane is only a few kilojoules per mole lower. Constant Temperature Reaction (CTR) Experiments. In the case where the deprotonation step is ratedetermining and the deprotonation is a simple firstorder desorption process, the activation energy can also be obtained from an Arrhenius plot. An injection experiment is performed, after which at a constant-temperature hydrogen is added. When the logarithm of the total detector signal (corresponding to the total amount of adsorbed molecules) is displayed as a function of time, a linear slope proportional to the reaction constant should appear. Figure 10 shows these curves for reduced H-Mordenite at three different temperatures. When the reaction constant is plotted in an Arrhenius plot (Figure 11), the slope of this plot should give the activation energy and the abscissa the preexponential factor. Experiments were conducted between 200 and 340 °C, with intervals of 20 °C. For reduced H-Mordenite, the obtained values were 110 ( 10 kJ/mol and 1 × 108 s-1. For propane a value of 102 ( 10 kJ/mol resulted.
It is not surprising that the activation energies are similar because they do not depend on chain length;26 the low values for the preexponential factor indicate, however, that they do not result from a deprotonation reaction, which would be around 1013. As a check, an experiment on freshly reduced 0.5 wt % Pt/Al2O3 was performed, which does not contain acid sites. This resulted in similar desorption rates, suggesting that species chemisorbed on the Pt sites are desorbing. When this experiment is performed on a preconditioned sample of H-ZSM5, a significant amount of radioactivity remains adsorbed. Even after an increase in the temperature to 400 °C, not all is removed. An Arrhenius plot can still be made when this remaining amount at the end of the experiment is subtracted as a “background”. The results are displayed in Table 4. Fitting the Reactor Model. When the full mass transport/reaction kinetics model is fitted to the measured profiles, readsorption is taken into account. Further, to prevent the situation in which desorption from chemisorbed hydrocarbons from the Pt sites is dominant, the samples were preconditioned with a steady-state reaction of a hexane and H2 mixture. The zeolite samples used were H-ZSM22, H-ZSM5, HMordenite, and H-Beta. After hydrogen is applied, a large difference is observed for these zeolites. The H-ZSM22 sample hardly releases any of the adsorbed labeled molecules: H-Mordenite just a small fraction, H-ZSM5 about 40%, and H-Beta more than 50%. This effect is a direct reflection of the pore structure of the
1982
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002
Table 4. Apparent Activation Energies for Hydrogenation at High Temperatures sample
T range (°C)
alkane
precondition
Eapp (kJ/mol)
v (s-1)
Pt/Al2O3 Pt/H-Mordenite Pt/H-Mordenite Pt/H-Mordenite Pt/H-ZSM5 Pt/H-ZSM5
260-280 240-340 280-340 290-340 140-240 130-170
C6 C6 C4 C3 C6 C4
reduced reduced reduced reduced preconditioned preconditioned
162 110 82 102 52 20
4 × 1012 1 × 108 2 × 105 2 × 105 1 × 104 1
a The value found indicates that carbonaceous species are desorbed from metal (Pt). For comparison, the same experiment was performed on a Pt/Al2O3 sample, giving a similar value. The values for ZSM5 are probably influenced by the long retention times at these much lower temperatures.
different zeolites. If the amount of carbonaceous species formed in all samples is the same, their effect will differ in the resulting amount of pore blocking. The narrow pores of H-ZSM22 are fully blocked, the one-dimensional pore structure of H-Mordenite also suffers from blocking, and the more open three-dimensional structure of H-ZSM5 allows for some desorption. Finally, H-Beta, which has the most open structure, releases the most labeled molecules. The model could be fitted to the experiments for H-ZSM5 and H-Beta, which released enough labeled molecules for a sufficiently accurate fit. The radioactivity that remains at the end of the experiment was subtracted from the measured concentrations as a background term. Sufficiently low temperatures (130190 °C) were used to ensure that desorption of the strongly chemisorbed species during the experiment can be neglected. Experiments were fitted at several temperatures to obtain a preexponential factor for the deprotonation. Such a single fit for H-ZSM5 is shown in Figure 12. For H-ZSM5 the values that are found are 142 ( 5 kJ/mol and (1.0 ( 0.5) × 1015 s1, while the fit of the
H-Beta profiles did not show any sensitivity to a reaction contribution, as evidenced by a symmetrical broadening of the profiles, indicating that the shape of the measured concentration profiles were purely dominated by mass transport. This poses questions on the results for H-ZSM5. The relevant differences between the H-ZSM5 and H-Beta samples are pore structure and size and Pt loading. The more narrow pore structure of ZSM5 may suffer more from blocking. Also in the profiles for ZSM5 further downstream in the reactor, a second pulse seems to form. From calculation of the expected retention time for isomers of C7, it seemed likely that this was iso-C7. The model does not account for this effect, but C7 appears to broaden the desorption, which in fitting the model results in an apparent deprotonation contribution to the peak width (see also Table 4). It must be noted that the value of Ka had to be adjusted to match the retention times of the experiment and model. This adjustment also depended on the temperature at which the desorption experiment was performed. Most likely, this compensates for a reduced retention volume due to coke formation. Cracking and Coke Formation. The amount of labeled material that remains adsorbed after the deprotonation experiment is likely to consist of carbonaceous species or oligomers that were formed during injection. The temperatures at which they can be removed with hydrogen are possibly informative on the kinds of these species. A rise of temperature to 400 °C, while flushing with hydrogen, showed that they are removed at temperatures typically higher than 300 °C. Because the presence of these species deteriorates the accuracy at which a deprotonation rate can be measured, a lot of effort was spent to minimize the formation of these species. By using a smaller wt % of Pt and using a Pt/ Sn alloy instead of pure Pt, the metal-catalyzed cracking could be reduced. Also by execution of a steady-state
Figure 12. Fit of a deprotonation experiment on H-ZSM5 at 160 °C. The model gives a good description of the experiment. For long retention times, however, a second peak seems to appear.
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 1983
reaction of hexane and hydrogen for a prolonged time before the pulse injection, the cracking can be reduced. As mentioned before, a carbonaceous overlayer on the Pt clusters is thought to result from this preconditioning.9 This would reduce the number of neighboring Pt sites and thus the cracking rate without suppressing the dehydrogenation reaction rate itself. Coke deposition in the pores may cause pore blocking, which causes deactivation for isomerization under steady state.32 Therefore, it is interesting to study the relation between pore size/structure and the relative ease with which the protonated fraction is removed in the presence of hydrogen. The removal of the “coke” by an increase of the temperature is a function of the pore size. Beta and Mordenite release their coke at 350 °C. However, ZSM22 hardly releases its coke, even at 400 °C. This may be related to the small one-dimensional micropores of ZSM22 (10 rings) compared to the larger one-dimensional Mordenite micropores (12 rings) and the larger three-dimensional micropores of ZSM5 (12 rings). The product analysis of an injection experiment of 11C in hydrogen on a reduced Pt/H-Mordenite sample 6 illustrates the importance of cracking during injection at the temperatures used (240 °C typical). At 150 °C, the products that leave the reactor mainly consist of C6. When the temperature is subsequently increased in steps to 230 °C, the product distribution shifts to C1 and C2, indicating metal cracking. At high temperatures the rate of coke removal is high enough to determine its activation energy. Also the retention time is small compared to the experiment duration, which means that the desorption rate is not influenced by the retention time in the bed. By assuming a first-order rate of desorption, the apparent activation energy can be determined from the Arrhenius plot of the desorption rate. For H-Mordenite a value of 110 kJ/mol with a preexponential factor of 1 × 108 was found. For comparison, a similar experiment with C6 was performed on Pt-SiO2/Al2O3, which only has the metal function. This gave a value of 162 kJ/mol with a preexponential factor of 4 × 1012. This high activation energy is an indication that hydrogenation from the metal sites is observed instead of a deprotonation from an acid site. Table 4 shows measured rate parameters for several zeolites and alkanes. n-Alkane/Isoalkane Balance. By variation of the time between the injection and the switch to hydrogen and analysis of the product distribution of the labeled alkanes, an attempt was made to measure the elementary rate of the isomerization of hexane. The time difference was varied between 30 s and 5 min. The established isomerization equilibrium on the acid sites seems to be very rapid because the ratio n-hexane/ isohexane that was found in the product stream at these temperatures was constant (equal to the thermodynamic equilibrium). Interestingly, the product distribution shows more shorter products after more experiments have been performed. This indicates that the beneficial effect of the preconditioning treatment of the first experiment is slowly diminished by the hydrogen flow that removes the carbonaceous overlayer from the Pt clusters. Conclusions We have shown that with PEP we are capable of imaging in situ reactions under typical operating condi-
tions. When the hydrogen feed is replaced with helium and a labeled pulse injected in the reactor, dehydrogenation could be studied separately. When a reactor model that includes mass transport and reaction to the profiles, measured at different temperatures, is fitted, a value for the activation energy for dehydrogenation of 98 ( 15 kJ/mol was found. This value deviates from the values typically found in the literature. The fact that this activation energy is determined with respect to the adsorbed state, and thus the heat of adsorption is included in this value, explains the difference. Another explanation might be that the model does not account for the complex side reactions on the Pt and acid sites. These affect the distribution of the labeled molecules because the products have different mass transport and reaction properties. In addition, it proved difficult to reproduce the effects of the preconditioning treatment. Ideally, the products of the injection experiment are only protonated hexane isomers. When helium is replaced by hydrogen again, the reaction pathway toward hydrogenation is opened and the equilibrium of the chemisorbed species shifts to deprotonation. From the PEP images and (radio) gas chromatographic analysis of the products at 240 °C, it was clear that a large part of the products consisted of hexane. This is a strong indication that the theoretical reaction mechanism of a stable protonated species on the surface exists. A value of 61 kJ/mol was found for the heat of adsorption of hexane. A large difference between the studied zeolites in the part of the total radioactivity that could be removed by applying hydrogen was observed. A strong correlation with the pore structure of the zeolites was found. The more limited the dimensionality and the narrower the pores, the less labeled material was released, indicating that pore blocking plays an important role. Because standard techniques such as TPR are not valid, when mass transport affects the rate of reaction, a full mass transfer and reaction model was fitted to the measured concentration profiles. TPR though does reveal at what temperatures the carbonaceous residues can be removed. The high temperature (>350 °C) at which they were released indicates that they consist of highly dehydrogenated carbonaceous species. Apparently, the degree of deactivation is not a direct measure for the degree of coke formation. Of major importance is the structure of the zeolite, which together with the amount of coke determines the effective deactivation. Constant-temperature reaction experiments on reduced catalysts resulted in an activation energy of 110 ( 10 kJ/mol. This value appears to relate to desorption of chemisorpted species on the Pt sites. For experiments conducted on the preconditiond catalysts, the deprotonation model could only be fitted well to H-ZSM5 and H-Beta experiments because only those samples released a significant amount of labeled molecules. In this case the irreversibly adsorbed carbonaceous species were treated as a background term in the concentration profiles. The H-Beta experiments showed to be completely dominated by mass transport, and no activation energies for reaction-related phenomena could be obtained. The H-ZSM5 result of 142 kJ/ mol might also be affected by the side products in the reactor effluent, which would also broaden the concentration profiles, like a slow deprotonation process. However, by assuming that mass transport is slower than the reaction rate, at least an estimate of the upper
1984
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002
boundary for the deprotonation energy can be derived. For the expected preexponential factor of 1013 and the fact that the reaction seems to proceed in less than 100 s (retention time), an upper boundary of about 100 kJ/ mol results. This is in agreement with theoretical calculations as depicted below. Until recently, quantum-chemical calculation on proton activation by zeolites was only possible using the cluster approximation.33,34 The activation energy in reactions with hydrocarbons can be computed with an accuracy of 5-30 kJ/mol, using programs based on density functional theory.35-37 However, a major zeolite property that cannot be studied with small clusters is the effect of steric matching of the zeolite micropore and the size of the transition state complex. Now the periodical structure of the zeolite can be taken into account in calculations.38-40 These calculations resulted in significantly lower values for the activation energies. The structure of the transition state complex is highly polar. The high polarizability of the oxygen atoms cavity (zeolite) stabilizes this complex. The expected stabilization is, of course, larger for the small-pore zeolites than for the large-pore zeolites. Stabilization energies as high as 50-70 kJ/mol have been reported.41,42 As an example, we give the values for the protonation of propylene. Because the energetics hardly depends on chain length, these values are also representative for hexene protonation. While the cluster approach gives a value of 105 kJ/mol, the periodical calculations result in a significantly lower value of 47 kJ/mol.40 In that case, the PEP profiles will certainly not be determined by the rate of deprotonation. Nomenclature x ) microparticle porosity, m3 of micropores/m3 of microparticle y ) macroparticle porosity, m3 of macropores/m3 of macroparticle z ) packed bed porosity, 1 - m3 of macroparticles/m3 of column Cim ) concentration of the adsorbed phase (mol/m3) Cp ) average gas-phase concentration in the pellets (mol/ m3) Cz ) fluid-phase gas concentration (mol/m3) Dx ) microparticle gas-phase diffusion coefficient of nhexane (m2/s) Dy ) macroparticle diffusion coefficient of n-hexane (m2/s) Dz ) fluid-phase diffusion coefficient of n-hexane (m2/s) k ) effective mass transfer coefficient (s-1) kd ) rate constant for desorption (deprotonation) (s-1) Ka ) dimensionless adsorption equilibrium constant Kp ) adsorption constant on a pellet volume basis Lbed ) column length (m) q j ) average sorbate concentration in porous particles (mol/ m3) R ) gas constant (J/mol‚K) Rx ) microparticle radius (m) Ry ) macroparticle radius (m) kf ) film mass-transfer coefficient (m/s) vsup ) superficial carrier gas velocity (m/s) z ) fluid-phase axial coordinate (m)
Literature Cited (1) Weisz, P. B. Polyfunctional Heterogeneous Catalysis. Adv. Catal. 1962, 13, 137. (2) Maxwell, I. E.; Stork, W. H. J. Hydrocarbon Processing with Zeolites. Stud. Surf. Sci. Catal. 1991, 58, 571.
(3) Anderson, B. G.; Noordhoek, N. J.; Van IJzendoorn, L. J.; De Gauw, F. J. M. M.; Van Santen, R. A.; De Voigt, M. J. A. Mass Transfer of Alkanes in Zeolite Packed-Bed Reactors Studied with Positron Emission Profiling (PEP). 1. Experiments. Ind. Eng. Chem. Res. 1998, 37 (3), 815. (4) Noordhoek, N. J.; Anderson, B. G.; Van IJzendoorn, L. J.; De Gauw, F. J. M. M.; Van Santen, R. A.; De Voigt, M. J. A. Mass Transfer of Alkanes in Zeolite Packed-Bed Reactors Studied with Positron Emission Profiling (PEP). 2. Modeling. Ind. Eng. Chem. Res. 1998, 37 (3), 825. (5) Mills, G. A.; Heinemann, H.; Milliken, T. H.; Oblad, A. G. Paper presented at the 121th American Chemical Society Meeting, Milwaukee, WI, March 1952. Ind. Eng. Chem. 1953, 45, 134. (6) Van de Runstraat, A.; Van Grondelle, J.; Van Santen, R. A. On the Temperature Dependence of the Arrhenius Activation Energy for Hydroisomerisation Catalyzed by Pt/Mordenite. J. Catal. 1997, 167, 460. (7) Van de Runstraat, A.; Kamp, J. A.; Stobbelaar, P. J.; Van Grondelle, J.; Krijnen, S.; Van Santen, R. A. Kinetics of Hydroisomerisation of n-Hexane over Platinum Containing Zeolites. J. Catal. 1997, 171, 77. (8) Chu, H. Y.; Rosynek, M. P.; Lundsford, J. H. Skeletal Isomerization of Hexane over Pt/H-Beta Zeolites: Is the Classical Mechanism Corrrect? J. Catal. 1998, 178, 352. (9) Van Santen, R. A.; Anderson, B. G.; Cunningham, R. H.; Mangnus, A. V. G.; Van IJzendoorn, L. J.; De Voigt, M. J. A. In Situ Observation of Transient Reaction Phenomena Occurring on Zeolite Catalysts with the Aid of Positron Emission Profiling. Angew. Chem., Int. Ed. Engl. 1996, 35 (23/24), 2785. (10) Sinfelt, J. H.; Carter, J. L.; Yates, D. J. C. Catalytic Hydrogenolysis and Dehydrogenation over Copper-Nickel Alloys. J. Catal. 1972, 24, 283. (11) Menon, P. G. Chem. Rev. 1950, 94, 118. (12) Biswas, J.; Bickle, G. M.; Gray, P. G.; Do, D. D.; Barbier, J. The Role of Deposited Poisons and Crystalite Surface Structure in the Activity and Selectivity of Reforming Catalysts. Catal. Rev.Sci. Eng. 1988, 30, 161. (13) Aranda, D. A.; Noronha, F. B.; Schmal, M.; Passos, F. B. Characterization of Pt-Sn Bimetallic Catalysts Supported on Alumina and Niobia. Appl. Catal. A 1993, 100, 77. (14) Del Angel, G.; Tzompantzi, F.; Gomez, R.; Baronetti, G.; De Miguel, S.; Scelza, O.; Castro, A. React. Kinet. Catal. Lett. 1960, 42, 67. (15) Mangnus, A. V. G.; Van IJzendoorn, L. J.; De Goeij, J. J. M.; Cunningham, R. H.; Van Santen, R. A.; De Voigt, M. J. A. A BGO Detector for Positron Emission Profiling in Catalysts. Nucl. Instrum. Methods 1995, B99, 649. (16) Denayer, J. F.; Souverijns, W.; Jacobs, P. A.; Martens, J. A.; Baron, G. V. High-Temperature Low-Pressure Adsorption of Branched C5-C8 Alkanes on Zeolite Beta, ZSM-5, ZSM-22, Zeolite Y and Mordenite. J. Phys. Chem. B 1998, 102, 4588. (17) Wu, E. L.; Landholt, G. R.; Chester, A. W. In New Developments in Zeolite Science and Technology; Murakame, Y., Iijama, A., Ward, J. W. Eds.; Studies in Zeolite Science and Catalysis 28; Elsevier: Amsterdam, The Netherlands, 1986; p 547. (18) Van den Broek, A. C. M.; Van Grondelle, J.; Van Santen, R. A. Preparation of Highly Dispersed Platinum Particles in HZSM-5 Zeolite: A Study of the Preatreatment Process of [Pt(NH3)4]2+. J. Catal. 1997, 167, 417. (19) Van de Runstraat, A. Adsorption Effects in Acid Catalysis by Zeolites. Ph.D. Thesis, Eindhoven University of Technology, Eindhoven, The Netherlands, 1997; ISBN 90-386-0169-7. (20) Homeyer, S. T.; Sachtler, W. M. H. Elementary Steps in the Formation of Highly Dispersed Palladium in NaY. J. Catal. 1989, 117, 91. (21) Eder, F. Thermodynamics and Siting of Alkane Sorption in Molecular Sieves. Ph.D. Thesis, Universiteit Twente, Enschede, The Netherlands, 1996; ISBN 90-3650861-4. (22) Raghavan, N. S.; Ruthven, D. M. Simulation of Chromotographic Response in Column with Bidispersed Structured Particles. Chem. Eng. Sci. 1985, 40, 699. (23) Schiesser, W. E. The Numerical Method of Lines; Academic Press Inc.: New York, 1991, ISBN 0-12-624130-9. (24) Ka¨rger, J.; Ruthven, D. M. Diffusion in Zeolites and Other Microporous Solids; John Wiley & Sons: New York, 1992.
Ind. Eng. Chem. Res., Vol. 41, No. 8, 2002 1985 (25) Polanyi, M.; Horiuti, M. Exchange Reactions of Hydrogen on Metallic Catalysts. Trans. Faraday Soc. 1934, 30, 1164. (26) Kazansky, V. B.; Senchenya, I. N. On the Real Nature of Aliphatic Carbenium Ions as Active Intermediates of Homogeneous and Heterogeneous Acid Catalysis. J. Mol. Catal. 1992, 74, 257. (27) Blaszkowski, S. R.; Nascimento, M. A. C.; Van Santen, R. A. Activation of C-H and C-C Bonds by an Acidic Zeolite: A Density Functional Study. J. Phys. Chem. B 1996, 100, 3463. (28) Rozanska, X.; Van Santen, R. A.; Hutschka, F.; Hafner, J. A Periodic DFT Study of Intramolecular Isomerization Reactions of Toluene and Xylenes Catalyzed by Acidic Mordenite. J. Am. Chem. Soc. 2001, 123 (31), 7655. (29) Van de Runstraat, A.; Van Grondelle, J.; Van Santen, R. A. Microkinetics Modeling of the Hydroisomerization of n-Hexane. Ind. Eng. Chem. Res. 1997, 36 (8), 3116. (30) Falconer, J. L.; Schwartz, J. A. Temperature-Programmed Desorption and Reaction: Applications to Supported Catalysts. Catal. Rev.-Sci. Eng. 1983, 25 (2), 141. (31) Demmin, R. A.; Gorte, R. J. Design Parameters for Temperature-Programmed Desorption from a Packed Bed. J. Catal. 1984, 90, 32. (32) Ponec, V.; Bond, G. C. Catalysis by Metals and Alloys; Studies in surface science and catalysis; Elsevier: New York, 1995. (33) Blaszkowski, S. R.; Van Santen, R. A. In Transition State Modelling for Catalysis; Truhlar, D. G., Morokuma, K., Eds.; ACS Symposium Series 721; American Chemical Society: Washington, DC, 1999; Chapter 24. (34) Kazansky, V. B. Adsorbed Carbocations as Transition States in Heterogeneous Acid-Catalyzed Transformations of Hydrocarbons. Catal. Today 1999, 51, 419. (35) Parr, R. G.; Yang, W. Density Functional Theory of Atoms and Molecules; Oxford University: New York, 1989.
(36) Ziegler, T. Approximate Density unctional Theory as a Practical Tool in Molecular Energetics and Dynamics. Chem. Rev. 1991, 91, 651. (37) Labonowski, J. K., Andzelm, J. W., Eds. Density Functional Methods in Chemistry; Springer: Berlin, 1991. (38) Haase, F.; Sauer, J. Ab Initio Molecular Dynamics Simulation of Methanol Interacting with Acidic Zeolites of Different Framework Structure. Microporous Mesoporous Mater. 2000, 35, 379. (39) Boronat, M.; Zicovich-Wilson, C. M.; Corma, A.; Viruela, P. Cluster and Periodic ab initio Study of the Ethane-ethene Hydride Transfer Reaction Catalyzed by Acid Chabazite. Is the Cluster Model Able to Describe Accurately the Host-guest Interactions? Phys. Chem. Chem. Phys. 1999, 1, 537. (40) Rozanska, X.; Van Santen, R. A.; Hutscka, F.; Hafner, J. A Periodic DFT Study of Intramolecular Isomerization Reactions of Toluene and Xylenes Catalyzed by Acidic Mordenite. J. Am. Chem. Soc. 2001, 123 (31), 7655. (41) Rigby, A. M.; Kramer, G. J.; Van Santen, R. A. Mechanisms of Hydrocarbon Conversion in Zeolites: A Quantum Mechanical Study. J. Catal. 1997, 170, 1. (42) Sinclair, P. E.; De Vries, A.; Sherwood, P.; Catlow, C. R. A.; Van Santen, R. A. Quantum-Chemical Studies of Alkene Chemisorption in Chabazite: A Comparison of Cluster and Embedded-Cluster Models. J. Chem. Soc., Faraday Trans. 1998, 94, 3401.
Received for review May 21, 2001 Revised manuscript received January 22, 2002 Accepted January 23, 2002 IE010455C
