Abstract The cracking of n-heptane proceeds at 623 K and a reaction pressure of 2.

5 MPa preferentially according to the bi olecular cracking route. This !as confir ed for M"#- and "A$-type %eolites !ith &arying density of acid sites. The acti&ity !as found to be linearly dependent on the nu ber of 'r(nsted acid sites for M"#type %eolites. #n the case of "A$-type %eolites it !as obser&ed that the acti&ity per 'r(nsted acid site decreases !ith increasing concentration of acid sites. This !as e)plained by differences in acid strength. All the in&estigated acid %eolite sa ples !ere able to acti&ate hydrogen under the actual reaction conditions. The acti&ated hydrogen !as found to participate in bi olecular hydrogen transfer reactions during n-heptane cracking. The action of hydrogen is discussed on the basis of its influence on the rate-deter ining step during cracking and !as confir ed by a kinetic study of n-heptane cracking on a "A$-type sa ple. *ydrogen transfer fro hydrogen as !ell as fro hydrocarbons !as influenced in the sa e !ay by %eolite structure and acid site density #+T,-.$/T#-+

The cracking of n-alkanes can proceed via a bimolecular or a monomolecular reaction pathway (1). There are several studies dealing with the relation between the properties and performance of zeolites in n-alkane cracking. Haag and essau (1) found that the monomolecular protolytic route is more favored with the mediumpore zeolite !"#-$ (#%&) than with a large-pore ' zeolite (%()). %urthermore* they have shown that bimolecular cracking becomes the prominent pathway with decreasing temperature and increasing partial pressures of the hydrocarbon. +iannetto et al. (,) concluded that classical bimolecular cracking is more pronounced with an aluminiumrich %() sample than with a silicon-rich %() sample. #irodatos and -arthomeuf (.) suggested that protolytic cracking is favored with small-pore zeolites and/or zeolites with a high tortuosity. 0ielers et al. (1) confirmed these results and have shown that with increasing temperature* decreasing aluminium content* and decreasing pore dimensions the relative contribution of the monomolecular protolytic cracking route increases as compared to bimolecular cracking. 0hen performing n-heptane cracking with an H!"#-$ zeolite at a reaction temperature of $1. 2 and a total reaction pressure of ,.1 #3a ($)* it appeared that the reaction follows preferentially the bimolecular pathway* which means that the reaction chain is propagating via bimolecular hydride abstraction from an alkane molecule and its transfer to an intermediate carbenium ion. &n this previous work it was shown that under these reaction conditions the cracking rate was controlled by the rate of desorption of reaction products. Thus* the introduction of hydrogen into this system causes an increase of the initial rate of cracking* which was e4plained by assuming that hydrogen interacts with adsorbed olefins* hydrogenates them* and forces their desorption. The hydrogenation produces saturated molecules in competition with the bimolecular hydride abstraction from hydrocarbons. &n both reactions hydrogen is transferred to adsorbed unsaturated molecules. 0hereas there is much information concerning hydrogen transfer between hydrocarbons* there is a lack of information concerning hydrogen transfer with molecular hydrogen. &n this work we have investigated the role of zeolite structure and acid site density on hydrogen transfer reactions involving n-heptane and molecular hydrogen during n-heptane cracking.

EXPERIMENTAL
Preparation and Characterization of Catalysts

H!"#-$ samples with different "i/(l ratios were synthesized template-free and were activated by ammonium e4change and calcination (5). The "i/(l ratio of the modified zeolites was determined by chemical analysis. "ample %()1 with a unit cell size of ,.116 nm was obtained by steam calcination at 67. 2 and 1889 steam of a partially ammonium-e4changed :a' zeolite (;inde "2 18)* followed by ammonium e4change. "ample %(), was prepared by steam calcination of %()1 at <,. 2* followed by ammonium e4change. 3ortions of these samples were treated with a 8.1M solution of (:H1),"i%5 at .5. 2 for . h. This treatment was done with (:H 1),"i%5-to-zeolite ratios of 8.,5 g g-1 (%()1 -= %()1%1* %(), -= %(),%) and 8.17 g g-1 (%()1-= %()1%,). The unit cell size of the final)"'zeolites was determined by >-ray diffraction using CuKa radiation and following ("T# procedure -.<1,-68. The estimated standard deviation was 8.881 nm. The crystallinity of these samples was calculated by comparing the peak height of the ($* .* .) reflection and considering the reference :a' "2-18 as presenting 1889 crystallinity. The "i/(l (>? ) ratio for the samples was calculated on the basis of the %ichtner-"chmittler e@uation (7). The differences between these values and those obtained by chemical analysis ("i/(l (chem)) allowed us to calculate the amount of (l in e4traframework positions (A%(;). The &? spectroscopic measurements were carried out in a conventional greaseless &? cell. 0afers of 18 mg cm-, were pretreated overnight at 57. 2 and 18-. #3a dynamic vacuum. The spectra in the 1888B1.88 cm -1 region were recorded at room temperature in a 3erkinBAlmer $68 - spectrometer e@uipped with a data station. %or pyridine adsorption e4periments 5.55418, 3a of pyridine were introduced into the cell at room temperature. (fter e@uilibration* the samples were outgassed at $,.* 5,.* and 57. 2 under vacuum and the spectra were recorded at room temperature. &ntensities of the bands at 1$18 cm -1 (-rCnsted sites) and 11$8 cm-1 (;ewis sites) and apparent e4tinction coefficients reported by Ameis (6) were used to determine the concentration of acid centers. %urthermore* the -AT surface area and the micropore volume of the %()-type samples were determined by adsorption of nitrogen at 77 2 in an ("(3 ,888 instrument. Catalytic Experiments The reactions have been carried out in a stainless steel tubular reactor whose inner surface was passivated by graphitation. The details are described elsewhere ($). 3erforming blank e4periments with n-heptane (pn-heptane* 8.1 #3a) and 1-butene ( p1-butene* 8.8, #3a)it was verified that there was no conversion of the alkane and not more than.9of the olefin was converted. The reaction was studied at 5,. 2 and a total pressure of reaction of ,.1 #3a. The n-heptane flow was chosen in the range between 8..$ and 8..< ml min-1. &f not indicated otherwise* the ratio of n-heptane to the diluting gas was kept constant at 8.16 mol mol-1. :itrogen or hydrogen or mi4tures of both gases were taken as diluting gases. The zeolites have been used as pressed powders with a mesh size between 8.,$ and 8.1, mm. The amount of catalyst was varied between 8.1$ and 8.1$ g. "ample #%&1 was used to check that under these conditions the reaction is not controlled by e4ternal diffusion. To e4amine the relevance of internal mass transfer* e4periments with different mesh sizes (#%&1) were carried out and no significant influence on reaction rate was observed in the chosen range of mesh sizes. The reaction was always followed at different times-onstream (TD") starting at 8.$ min TD". &t was supposed that at this short TD" the influence of deactivation was negligible. The activity of the different zeolites for cracking n-heptane under these

reaction conditions was e4pressed by the conversion. The reaction kinetics were determined at levels of conversion below 189 and then the cracking rate was calculated from the rates of formation of hydrocarbons with one to si4 carbon atoms* i.e.* cracking rate E 1/7.(rF1 ,rF, .rF. 1rF1 $rF$ 5rF5 )G where the values of rFi represent the rate of formation of a hydrocarbon with i carbon atoms* the dimensions of the rate being mol g-1 min-1.
,01$2T1 A+. .#1/$11#-+

Physico-Chemical Properties of Used Zeolites &n Tables 1 and , the physico-chemical properties of the zeolite samples derived from the methods applied are given. &n the case of #%&-type zeolites the synthesis products were checked by >? and all samples were identified as pure and highly crystalline !"#-$ zeolite. ( characteristic &? spectrum of sample #%&1 after pyridine adsorption is given in %ig. 1 and the low intensity of the band at 11$8 cm H1 was taken as evidence that the number of e4traframework aluminium species (A%(;) was negligible. Therefore* it was assumed that nearly all of the aluminium was located on framework positions (%(;). Taking this into account* the values obtained from the chemical analysis were used to calculate the (l/((l F "i) ratios given in Table 1. The concentrations of -rCnsted acid sites were calculated from &? spectra recorded after outgassing the pyridine-loaded samples at $,. 2 (Table 1). %rom the (l/((l F "i) ratios of the #%& samples* it is possible to assume that all the framework (l is isolated and the associated acid sites should be e@uivalent. Therefore* differences in the catalytic activity should be related to differences in the number of sites (<) rather than to their acid strength. &n Table , the total amount of %(; and A%(; and other physico-chemical properties of the %()-type zeolite samples are summarized. The pretreatment of sample %()1 and sample %(), with (:H1),"i%5 causes the removal of A%(;(18) while the framework composition remains practically the same. This is supported by the low value of e4traframework species in the samples %()1%1 and %(),% calculated from chemical analysis and the unit cell parameters* as well as by the small number of ;ewis acid sites derived from the &? spectroscopy. -y increasing the concentration of the (:H1),"i%5 it was possible to remove not only most of the A%(; but also framework (l* producing a further dealuminated almost A%(;-free ' zeolite (%()1%,). The hydrothermal pretreatment* and especially the A%(; e4traction* produces a decrease in zeolite crystallinity. Fomparing the composition of sample %()1 with %()1%1 and %(), with %(),% one can see that the removal of A%(; results in an increase in the amount of -rCnsted acid sites measured by pyridine adsorption/desorption.This can be e4plained by assuming that cationic A%(; species partly compensate the negative charge of the anionic zeolite framework and/or that these species are deposited in the channels of the zeolite* blocking the access of pyridine to acid hydro4yl groups related to the framework aluminium (18).The decrease in the micropore volume of sample %()%, in comparison to %(), should be attributed to the loss in crystallinity (Table ,). These results indicate that)"'zeolites are characterized by more comple4 acidic properties. ( direct correlation between catalytic activity and framework (l content should not necessarily be e4pected. %urthermore* the samples containing more than .$B ,$ (l per unit cell may have -rCnsted acid sites with lower acid strength than samples with lower (l contents (11). Hydrogen ransfer during n-Heptane Crac!ing in "itrogen #tmosphere The product distribution and selectivities obtained during n-heptane cracking on #%& zeolites are given in Table .. %rom these results it becomes clear that despite possible differences due to the different levels of conversion the product distribution with respect to the F-number of cracking products reflects an e4cess in the

formation of F$ and F5 hydrocarbons in comparison to F1 and F, hydrocarbons. This can be rationalized by assuming that besides the direct cracking events* which give F./F1* F,/F$ and F1/F5 in e@ual amounts* further reactions have taken place. &n a previous work (1,) a stoichiometric network including dimerization steps was derived in order to e4plain the unbalance in light cracking products during nheptane cracking at $1. 2. &n the present work the high hydrocarbon partial pressure and the relatively low temperature permit the occurrence of dimerizationcracking reactions in the medium pore size #%&zeolite.Thus* it is not surprising that under these conditions a large contribution to the bimolecular cracking route may e4ist in the case of #%& zeolites. &ndeed* when the relative contribution of the monomolecular to the bimolecular cracking mechanism has been calculated from the (F1FF,) to isobutane ratio (cracking mechanism ratio* F#?) (1)* it can be seen that the ratio is much lower than one. (s e4pected* the F#? value decreases when the n-heptane partial pressure increases (%ig. ,). %urthermore* the paraffin to olefin ratio (P$%) can be taken as a yardstick for hydrogen transfer reactions (1.) and can be interpreted as the average chain length (repetition of hydrogen transfer steps). &t can be e4pected that this value behaves sensitively to variations in the conversion* zeolite composition* and zeolitic geometry. The molar ratio of paraffins to olefins ( P$%) was e4amined for different levels of conversion for sample #%&1. The results given in %ig. . clearly indicate an increase of the P$% ratio with increasing conversion. Therefore* the slightly higher values of the P$% ratio observed with increasing number of %(; in #%& type zeolites should rather be related to the higher total conversion than to changes in the properties of the zeolite samples. &n general* the observed values of the P$% ratio at 8.$ TD" underline the importance of the bimolecular hydride transfer reactions during cracking under the given reaction conditions with all the investigated #%& zeolites. (nalyzing the product distribution for n-heptane cracking on %()-type zeolites* one can conclude that the process is characterized by bimolecular cracking events and includes steps of dimerization Iust as was shown to occur for the medium-pore zeolites (Table 1). %urthermore* the values for the cracking mechanism ratio and the ratio of isobutene to n-butane also support this statement (Table 1). ?egardless of the zeolite type used* the cracking of n-heptane seems to be characterized preferentially by bimolecular cracking under the given e4perimental conditions. However* comparing the results fromTables . and 1* it becomes clear that an increase in the pore dimensions of the zeolite increases the relative contribution of the bimolecular versus the monomolecular reaction pathway. This can be deduced from the lower values of the F#? as well as from the higher P$% ratio found with )"' than with #%& zeolites. Hydrogen ransfer &n'ol'ing Molecular Hydrogen during n-Heptane Crac!ing &n previous papers ($* 1,) it has been shown that hydrogen can become activated on acid H!"#-$ zeolite and intervenes in acid-catalyzed cracking reactions. To e4plain the e4perimental results the following reactions were proposed to occur (1,). (n olefin JF nH,nK interacts with a -rCnsted site to form an adsorbed carbenium ion JF nH,nHF(ads)K. "urface reaction (1) proceeds between this ion and an educt paraffin molecule JFmH,mF,K such that via hydride abstraction a paraffinic product molecule JF nH,nF,K and a long-chain carbenium ion JFmH,mHF(ads)K are formedL FD:F;)"&D: )nder reaction conditions such as a temperature of 5,.2 and a total reaction pressure of ,.1 #3a* cracking of n-heptane proceeds preferentially according a bimolecular chain reaction mechanism on #%&- and %()-type zeolites. ?egardless of

the zeolite used the bimolecular hydrogen transfer between adsorbed species and educt molecules is catalyzed and dominates the observed processes. The activity for this type of hydrocarbon conversion can be attributed to the number of -rCnsted acid sites. The cracking activity increases linearly with concentration of these sites in cases where no difference in the intrinsic acidity of sites is evident. This was confirmed for #%&-type zeolites. -rCnsted sites of lower acid strength cause lower specific activity in n-heptane cracking. Hence* %()-type zeolites have shown lower activities for n-heptane cracking than #%&-type zeolites under similar reaction conditions. %urthermore. %()-type zeolites rich in aluminium (concentration of DH higher than 68 Mmol gH1) e4hibit lower specific activities* which becomes reasonable by assuming a lower acid strength of -rCnsted acid sites for these zeolites. Hydrogen transfer involving molecular hydrogen has taken place on all zeolite samples investigated. This study points out that transfer of hydrogen from molecular hydrogen and from hydrocarbons occurs symbiotically. The influence of molecular hydrogen on activity and selectivity during n-heptane cracking allows the establishment of a link between isomerization-cracking reactions carried out on monofunctional and on bifunctional (acid plus metal) catalysts.