J.C.S. Favaday I, 1980,76, S4-91Cracking of (5-13C)-n-Nonane with Quartz Wool,Silica-Alumina and Type Y ZeoliteBY THOMAS J. WEEKS, JR,? IRWIN R. LADD AND ANTHONY P. BOLTON"Union Carbide Corporation, Tarrytown Technical Center,Tarrytown, New York 10591, U.S.A.Received 30th January, 1979The cracking mechanism of (5-' 3C)-n-nonane has been studied over quartz wool, silica-aluminaand a type Y zeolite. The products observed at a reaction temperature of 510°C over quartz woolagree reasonably well with the currently accepted mechanism of free radical cracking. Reactionwith silica-alumina at 500°C and zeolite at 230°C results in a 13C labelled product distribution whichagrees with neither a thermal cracking mechanism nor the currently accepted mechanism of /%scissionof carbonium ion intermediates.Rather, the data suggest that the product distribution is a resultof the temperature-dependent random desorption and cracking of a complex polymeric precursor.The cracking of hydrocarbons over zeolite catalysts is of considerable comniercialimportance and has consequently been extensively studied. The results from thecracking of pure components have usually been interpreted in terms of a p-scissionmechanism involving carbonium ion formation, primary cracking and secondary~racking.l-~ However, some recent data on the cracking of (3-1 3C)-n-pentaneshowed that the 13C labelling of the cracked products, particularly propane, ap-proaches that of a totally random di~tribution.~ The complete redistribution of "Cand 13C in the products from the cracking of labelled hydrocarbons over H-mordenite has also been reported by Gault.6 On the basis of these data, it wasconcluded that the cracked products are formed from the breakdown of a sorbedcomplex polymeric material.However, pentanes and hexanes " do not containenough carbon atoms to allow the most probable carbonium ion cracking routes tomanifest themselves by the lowest energy pathways and axe therefore, in a sense,special cases ".' The molecule that is classically selected to illustrate the carboniumion mechanism of catalytic cracking is 2,2,4-trimethylpentane. Indeed, the crackedproducts are, as would be expected, isobutane and the butenes.* However, sincethere is but a limited number of ways that this particular molecule can undergoscission, studies using this molecule might inherently prejudice an evaluation of thecracking mechanism, A more satisfactory evaluation should be offered by studyinga large paraffin that possesses the possibility of cracking in a variety of ways.Thusthe paraffin n-nonane was chosen.As well as studying the cracking of n-nonane over a zeolite catalyst, additionalexperiments were carried out using silica-alumina and quartz wool. Possiblesimilarities or differences between the products from these materials should provideadditional insight into the mechanism of catalytic cracking over a zeolite.EXPERIMENTALMATERIALSThe zeolite was an ammonium-exchanged, hydrothermally stable Linde type Y molecularsieve (LXY-82).The zeolite, a silica-alumina cracking catalyst (Durabead-1 from MobilOil) and quartz wool had the following analysis :847 Present address : Ashland Chemical Company, P.O. Box 2219, Columbus, Ohio, U.S.AT. J . WEEKS, JR, I . R . LADD AND A . P. BOLTON 85quartz wool silica-alumina zeoliteA1203 /wt % 0 11.1 22.8SiOz /wt % 100 89.0 72.2NazO /wt % 0 < 0.1 0.2(NH4)20 /wt % 0 0 4.0surface area/m2 g-1 1.2 309 770~(5-13C)-n-nonane was obtained from Merck, Sharpe and Dohme of Canada and was 95 %enriched. Non-labelled n-nonane was from Matheson, Coleman and Bell and was usedwithout further purification ; chromatographic analysis showed it to contain 99.6 %n-nonane with trace amounts of other C9 isomers.PROCEDUREThe experiments were conducted in a 130cm3 glass reactor.The zeolite and quartzwool were activated in vacuum at 500°C in the reactor for 4 h. The silica-alumina wasactivated in steam at 550°C for 4 h followed by vacuum activation at 500°C for 2.5 h in thereactor. Because of the high cracking activity of the zeolite, less of this catalyst was usedin the experiments than the other two materials. After activation, the appropriate reactionconditions were adjusted to obtain the required conversions and the n-nonane was introducedinto the system. The following reaction conditions were used :quartz wool silica-alumina zeolitewt catalyst/g 1.037 1.006 0.100wt n-nonane/g 0.108 0.082 0.080reaction conditions temp/"C 510 500 230sampling time/min 5, 15 2.5, 7, 15 2.5,7, 15initial pressure/Torr 150 100 10010cm3 samples were taken at the indicated intervals and expanded into 50 or 500cm3vessels.Samples of the products were injected into an integrated gas chromatograph-mass spectrometer. The mass spectrometer was a JEOL JMS-D100 with a Texas Instru-ments 980B computer and Diablo disc. The mass spectrometer was operated at 17eVfor the silica-alumina and quartz experiments and 1OeV for the zeolite experiments. A20 ft, Q in. n-octane Porasil C 80/lOO chromatographic column was used in the silica-alumina and quartz wool experiments and a 20 ft, -$ in. 10 % OV-101 SO/lOO Supelcoportcolumn in the zeolite experiments. The mass spectrometer was repeatedly scanned at9 s intervals at a 20 s scan speed for the full range. The actual scan time was 5-6 s from26 to 170 m/z.Difficulty was experienced insufficiently separating by chromatography the relatively large number of products from thesilica-alumina catalyst for subsequent mass spectral analysis. Consequently, 3C labellingdata could not be obtained for all of these products.Beam conditions were 3 kV and 100 PA.RESULTSProduct distribution obtained by chromatographic analyses is shown in table 1.The experimental apparatus and conditions were not optimized to yield quantitativeconversion data, but rather to obtain precise labelling information. The 3Cdistribution and fragmentation patterns of the starting material, n-nonane, arepresented in table 2. Mass spectrographic analyses of the unreacted n-nonane inthe reaction products showed the labelling to be identical to that in the startingmaterial.The mass spectrographic data for the reaction products from the thre86 CRACKING OF LABELLED n-NONANE WITH VARIOUS CATALYSTSTABLE 1 .-CHROMATOGRAPHIC ANALYSES OF PRODUCTScat a1 yst quartz wool silica-alumina zeolitesampling timelmin 5 15 2.5 15 2.5 15product product distribution/mol %1.84.13.84.00.31.80.20.00.00.01.40.00.00.01.20.90.380.215.411.47.46.70.52.60.00.00.00.02.00.60.00.00.50.00.051.10.02.65.514.13.62.44.22.74.41.90.81.83.20.00.00.00.051 .O0.020.28.13 .O11.92.63.33.614.44.80.11 .o2.50.00.00.00.017.60.00.00.00.02.40.00.00.09.62.80.00.010.40.90.00.00.071.20.00.00.00.06.80.08.00.028.35.50.00.024.81.60.00.00.027.7catalysts me summarized in tables 3, 4 and 5.These data have been corrected fornaturally occurring isotopes and fragmentation observed in the non-labelled expek-ments. Calculations of isotopic labels was carried out by using the Mass SpectralSearch System (EPA) program LABDET. Although analyses were carried out atdifferent reaction times, no differences were observed in product labelling and henceonly those data from the 2.5 and 5 min samples are presented. These data are anaverage of 4 scans per peak from duplicate experiments. The precision of massTABLE MAS MASS SPECTROGRAPHIC ANALYSIS OF (5-1 3C)-n-NONANEion mlz mol %M+ 129128M-29+ 10099M-43+ 8685M-57+ 7271946937919891113C enrichment 10.4 T.J . WEEKS, JR, I . R . LADD AND A . P . BOLTON 87spectral measurement was estimated by calculating the standard deviation fromseven measurements of the M-29f ion of nonane recovered from the 5 min samplefrom the quartz wool reaction. The standard deviation for m/z = 100 and 99 were1.5 and 1.0, respectively. The tables also include the calculated 3C label distributionfrom two possible mechanisms ; one resulting from an intermolecular randomscrambling process in which the distribution is derived from the binomial expansion(a+ b)" where a, b axe the percentages of labelled and non-labelled carbon atoms,respectively, and n is the number of carbon atoms in the molecule ; the other from asimple unimolecular scission mechanism.Attempts to obtain quantitative informa-tion on the methane labelling from the quartz experiment were unsuccessful becauseof the small sample size. Qualitative results showed no labelled methane was presentafter 5 min.DISCUSSIONThe product distributions obtained from the cracking of n-nonane over the threecatalysts are shown in table 1. These data show the quartz wool to promote freeradical cracking which results in the formation of olefins together with quantities ofmethane and ethane.The zeolite yields a completely saturated product, principally isobutane andisopentane, which, incidentally, is very similar to that obtained by reacting an olefinover the zeolite at a comparable temperature. 9-1 The silica-alumina catalyst, onthe other hand, gives rise to a product more similar to that of the zeolite than thequartz wool but containing a substantial amount of olefins.The difference in productdistribution between the zeolite and silica-alumina is probably a reflection of thedifferent operating temperatures required to achieve reaction and is consistent withprevious studies.Mass spectrographic analyses of the starting n-nonane are shown in table 2.These data show not only the position of the label in this material but also the frag-mentation of a luge n-paraffin in a spectrograph. No rearrangement is observedupon loss of an ethyl, propyl or butyl group. The fragmentation patterns clearlyshow that the starting material is labelled in the 5-position, since the extent of labellingchanges little upon loss of ethyl, propyl and butyl fragments.As in a previous studyusing n-pentane, the mass spectrographic data of the unreacted n-nonane in theproduct was indistinguishable from the starting material.5 Thus the unreactednonane had not participated in any reactions on the zeolite surface.QUARTZ WOOL PRODUCTExamination of the mass spectrographic data from the quartz wool experimentshows that the butene, pentene and hexene are all produced from a simple scissionreaction. A radical chain mechanism for thermal cracking of paraffis was firstproposed by Kossiakoff and Rice.12 The initial propagation step is the abstractionof hydrogen from the paraffin to form a radical, followed by a @cission of the radicalto form an olefin and a smaller radical.A sequence of /?-scissions can occur untilsome small radical is formed for which /?-scission is replaced by hydrogen abstractionto yield a paraflin. The product distribution in this study is consistent with theproposed mechanism and similar to that reported for the thermal cracking of theisomeric hexanes.13 The formation of a C4 hydrocarbon by @-scission from an-nonane radical labelled in the 5-position would require the product to be m-labelled. The data in table 3 show this to be the case. Conversely, the formationof C5 and C6 hydrocarbons by p-scission would require both products to be essentiall88 CRACKING OF LABELLED n-NONANE WITH VARIOUS CATALYSTScompletely labelled.Again, the data in table 3 confirm the scission mechanism.The mass spectrometric fragmentation data for loss of a methyl group from the C5and C6 products show as expected the former to be labelled at a terminal carbon atomand the latter at a non-terminal carbon atom. The sequential formation of ethyleneaccording to the chain mechanism allows for this product to be singly labelled as thedata in table 3 show. If propylene were only formed by the initial scission of anonane radical, it would not be labelled. This was not observed experimentally.TABLE 3 .-MASS SPECTROGRAPHIC ANALYSES OF PRODUCTS FROM QUARTZ WOOLno. labelled isotopic calculated distribution/ %product carbons species/ % random unimolecularC2H4 M+ 21002278119800010021001s8211980001002100227800100325722100010053064001001-CSHlo M' 21008911834580946M-15 mlz 561-C6H12 M+10604021009281136520946M-15 m/z 70 108317However, product distribution data show that propylene is present in the productin a greater quantity than hexene, rather than in an equivalent quantity.This mustindicate that propylene is also derived from further sequential scissions of largerlabelled radicals. This would result in the single labelling of the propylene and beconsistent with the data in table 3. A similar argument apparently holds for thepresence of labelled ethane. The data obtained on the thermal cracking of paraffinshaving more than six carbon atoms are difficult to interpret and are open to somea.mbiguity.14 Thus, the labelling data on propylene and ethane may not be asinconsistent as they might at first appearT .J . WEEKS, JR, I . R . L A D D A N D A . P. BOLTON 89S I LI C A-A LUMI N A PRODUCTNot unexpectedly, both the product distribution and the mass spectrographicanalyses of the products confirm that the cracking mechanism over silica-alumina iscompletely different from that taking place over quartz wool. However, it is readilyapparent from the spectrographic data that, contrary to current opinion, the productscannot be produced by a simple /?-scission mechanism involving carbonium ionformation since the labelling data for the propylene, propane and butylene axe inclose agreement in every instance with that calculated for random distribution.This conclusion is supported by the carbon enrichment data which show that theisotopic content of the products is equivalent to that of the starting material. Thus,as the significant amount of double labelling in the products indicates, the crackedproduct must be derived from a completely random intermolecular mechanism.Incontrast to the singly labelled propylene in the quartz wool product, the propylenein the silica-alumina product is randomly labelled. The latter must result fromsynthesis via a randomly labelled polymeric precursor rather than from a simplescission of a radical or olefin.That the unreacted nonane present in the reactionproducts has the identical isotopic distribution to the starting material argues againstthe possibility of subsequent scrambling of primary paraffinic products. Even ifsome of the products were generated by a secondary reaction with the catalyst, theisotopic labelling data would not be so completely random. It is remarkable thatat similar reaction temperatures the quartz wool and the silica-alumina give rise tosuch different mechanisms. In view of the high surface area of the silica-aluminacatalyst, one might have expected at least some of the olefins to be produced by afree radical mechanism. However, the labelling data are particularly definitive onthis point. Although a random intermolecular mechanism is contrary to thecurrently accepted mechanism of catalytic cracking involving the B-scission of acarbonium ion, the data supporting the proposed mechanism are self-consistent andparticularly explicit and, moreover, not totally unexpected in view of a previousstudy on the cracking of n-~entane.~ The conclusion from the previous study mayalso be invoked to explain the carbon enrichment and labelling data of the silica-alumina products ; that they axe derived from a complex polymeric precursor on thecatalyst which bears no resemblance to the starting material.TABLE 4.-MASS SPECTROGRAPHIC ANALYSES OF PRODUCTS FROM SILICA-ALUMINAno. labelled isotopic 13C calculated distribution/ %product carbons species/ % enrichment/ % random unimo1ecula.rC3Htj M+ 210CjHs M+ 2101-C4Hs M+ 21022969426704326411.0 3257211.3 3257210.0 530640010000100001090 CRACKING OF LABELLED n-NONANE WITH VARIOUS CATALYSTSZEOLITE PRODUCTThe labelling data of the products from the zeolite cracking experiment are moredifficult to explain than those from the quartz or silica-alumina experiments.Theproducts have the same carbon enrichment as the starting material which precludesa simple unimolecular mechanism. However, only the labelling data of isobutaneare consistent with random distribution, those from the other products being neitherunimolecular nor random intermolecular.TABLE 5.-MASS SPECTROGRAPI-IIC ANALYSES OF PRODUCTS FROM ZEOLITEno.labelled isotopic I3C calculated distribution/ %product carbons species/ % enrichment/ % random unimolecularC&j M+ 210i-C4H10 M+ 210n-C4Hlo M+ 210i-C5H12 M+- 210n-C5HI2 M- 2100.53664334630.54257347503465012.3 3257210.0 5306410.8 5306410.6 8345810.4 8345800100001000010009460946The isotopic scrambling observed in the products from both the zeolite and silica-alumina cannot be explained in terms of subsequent interaction of either saturatedor unsaturated products from a direct scission mechanism. The cracked productsfrom a zeolite at M 200°C are characterized by the total absence of olefins and tosatisfy hydrogen stoichiometry it could be assumed that olefins are irreversiblyadsorbed on the zeolite and undergo further reaction.Thus, one may argue thatthe saturated products from the zeolite are derived from both the cracking reactionand a subsequent olefin condensation-polymerization reaction. This would yielda labelling distribution intermediate to that of random scrambling and simple uni-molecular scission. However, the carbon enrichment data of the products do notsupport this explanation. For example, if a portion of the isobutane were derivedfrom a simple scission, this material would contain no 13C and, therefore, wouldhave zero enrichment. The extent to which the total isobutane incorporates someisobutane from simple scission would reduce the overall enrichment of the productisobutane to considerably less than that of the starting material, i.e., 10.4 %.Infact, this is not the case since a value of 10.0 % was observed. Thus the primarT. J . WEEKS, JR, I . R. LADD AND A. P. BOLTON 91products from catalytic cracking must be formed directly from a sorbed polymericintermediate and not as a result of secondary reactions.It may be significant that the redistribution of 12C and 13C in the products fromthe zeolite is not as complete as that taking place over silica-alumina at the higherreaction temperature. This observation might indicate that both the productdistribution and the extent of 12C and 13C redistribution are more a function ofreaction temperature than the particular catalyst system. It has been suggestedthat the principal difference in products between amorphous and crystalline catalystsis due to the different temperature ranges over which these catalysts 0perate.l Theseinvestigators showed that at a reaction temperature where both catalysts could bedirectly compared the cracked products are remarkably similar.The suggestionthat cracked product distribution is primarily dictated by reaction temperatureimplies that a prerequisite for cracking activity is the possession of a pore system toact as a " pot " in which the reactants axe converted into a complex polymericprecursor and from which they randomly desorb as products characteristic of thattemperature.CONCLUSIONWe conclude from the labelling data that the results from the thermal crackingof n-nonane over quartz wool are in reasonable agreement with those from previousstudies.However, the labelling data from the silica-alumina experiments indicatethat the products are derived from an intermediate polymeric precursor and cannotbe explained in terms of p-scission of carbonium ion intermediates. The labeldistribution results from the zeolite study are less explicit, but the 13C enrichmentdata from the zeolite experiments together with both the labelling and I3C enrichmentdata from the silica-alumina study raise considerable doubts as to the validity of thecurrently accepted mechanism of catalytic cracking.We thank Dr. M. L. Poutsma for fruitful discussions on reaction mechanismsand Mr. P. W. Shafer for many stimulating talks.J. N. Miale, N. Y. Chen and P. H. Weisz, J. Catalysis, 1966, 6, 279.H. A. Benesi, J. Catalysis, 1967, 8, 368.P. B. Venuto and E. T. Habib, catalysis Rev. Sci. Eng., 1978, 18, 1.A. P. Bolton, I. R. Ladd and T. J. Weeks, Proc. VI Int. Congr. Catalysis, London, 1976, p. 316.F. G. Gault, Proc. V.. Int. Congr. Catalysis, London, 1976, p. 321. ' M. L. Poutsma, Zeolite Chemistry and Catalysis, ed. by J. A. Rabo, A.C.S. Monograph 171,(Amer. Chem. SOC., 1976), vol. 8, p. 506. * D. Barthomeuf and R. Beaumont, J. Catalysis, 1973, 30,228.T. J. Weeks and A. P. Bolton, J.C.S. Faraday I, 1975, 70, 1676.lo T. J. Weeks, C. L. Angell, I. R. Ladd and A. P. Bolton, J. Catalysis, 1974, 33, 256.l1 P. B. Venuto, L. A. Hamilton and P. S. Landis, J. Catalysis, 1966,44, 589.l2 A. Kossiakoff and F. 0. Rice, J. Amer. Chem. Soc., 1943,65,590.l3 J. A. Rabo and M. L. Poutsma, Molecular Sieve Zeolites-II, Adu. Chem. Series No. I02 (Amer.l4 D. A. Leathard and J, H. Purvell, Ann. Rev. Phys. Chem., 1970,21, 197.* S. E. Tung and E. J. Mininch, J. Catalysis, 1968, 10,166, 175.Chem. SOC., 1971, p. 284).(PAPER 9/148