Ind. Eng. Chem. Res. 1993,32, 397-402
397
KINETICS, CATALYSIS, AND REACTION ENGINEERING Acid-Catalyzed Cracking of Paraffinic Hydrocarbons. 2. Evidence for the Protonated Cyclopropane Mechanism from Catalytic Cracking Experiments S.Tiong Sie KoninklijkelShell-Laboratorium,Amsterdam (Shell Research B. V.),P.O.Box 3003, 1003 A A Amsterdam, The Netherlands
Experimental results obtained in studies on catalytic cracking of normal paraffins as described in the literature are interpreted in terms of the protonated cyclopropane mechanism described in the previous paper of this series. It is shown that this mechanism is capable of explaining the observed sharp increase in reaction rate with increasing carbon number of normal paraffins above C7 in catalytic cracking. It also provides an explanation for many characteristic features of the product spectra, such as the virtual absence of CIand CZas primary cracking products, the high degree of branching in the saturated fractions in contrast to the predominantly linear structures in the light olefin fractions, and the relatively low formation of Ca as compared with C4 and higher homologs under conditions where secondary cracking is minimized.
Introduction In the previous part of this series of papers (Sie, 1992) the generally accepted classical theory of acid-catalyzed cracking (which assumes a ,&scission mechanism for breakage of a C-C bond in a classical type of carbenium ion) was shown to be inadequate for explaining many characteristic features of catalytic cracking and hydrocracking processes as observed in practice. To provide a rationale for these phenomena, we have formulated a new mechanism, in which a nonclassical carbonium ion, viz., a protonated dialkylcyclopropane species, takes the place of the classical carbenium ion (Sie, 1992). Hydride shifts followed by bond breakage result in the formation of a tertiary carbenium ion and an olefin as cracked fragmenta. Hydride transfer from a hydrocarbon molecule to the carbenium ion then gives an isoalkane and a new carbenium ion by which the reaction cycle is closed. In a simplified form the mechanism is depicted in Scheme I for the cracking of a normal paraffin. An important condition to be met for cracking to take place by the reaction mechanism shown in Scheme I is that the values of n and m’ should both be at least 1. If n were 0, the initial classical carbenium ion to be formed from the parent paraffin molecule would be a primary one, which is highly unlikely on energetic grounds. Likewise, if m’were 0, the carbon atom of the alkyl group in &position to the cyclopropane ring would be a primary one, and it is very unlikely on energetic grounds that it would release the hydride ion to be shifted toward the cyclopropanering carbon, which is an essential step in the cracking reaction (cf. part 1 of this series). In the present paper of this series, experimental results of catalytic cracking studies reported in the literature are examined in detail to see whether they can be reconciled with the proposed new mechanism. The formation of cracked products in hydrocracking as well as during the isomerization of paraffins will be a subject of discussion in part 3 of this series of papers.
Scheme I. Simplified Reaction Scheme for Cracking of a Normal Paraffin by the Protonated Cyclopropane Mechanism :HI H-j
H H H
[Hjn H H H n21
!Hi
H
H H :HI
c j-c-c-c- c-c- 1 c I -H
1
n-PARAFFIN
H H ;Hjm*
HYDRiDE ABsTRAcTiowmANsFER
H H H H H !Hi
-1 C j -C-C-C- C-C- j C j -H
CARBENIUMION
[ H j n O H H H H \Hjm,
H
H
HYDRIDE SHIFT, SCISSION
Catalytic Cracking of Pure Normal Paraffine Effect of Chain Length on Crackability. The very strong effect of chain length on the crackability of normal paraffins in the range from CSto ($4 has been observed by Voge (Voge, 19581,who also concluded that this strong effect cannot be accounted for by the classicalmechanism. Assuming that all secondary carbon atoms of the straight paraffin have an equal chance of becoming the charge center of a secondary carbenium ion, one should expect the reactivities to be proportional to the total number of carbon atoms minus 2. Thus, the ratio between the reactivities of n-Cls and n-C,, for instance, should be 1415,
oaaa-~aa~19312632-0397~04.00100 1993 American Chemical Society
398 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 Table I. Effect on Chain Length on Reaction Rate of Paraffins in Catalytic Cracking (Experimental Data from Voge (1958). T = 500 O C ; P = 1 bar; Catalyst, Alumina-Zirconia-Silica) Ist-order rate re1 reactivities on molar basis % feed cracked" const. mol/(L.h) obsd wed 1.0 1.0 3 0.42 n-C7 6.4 6 18 2.7 n-Clz 18 10 n-CI6 42 1.4 28 18 n-CZ4 58 11.8 0 Figures determined by the author from a graph published by Voge.
RELATIVE REACTIVITY
50 0 VOGE
20
t
et a t (39581
0
A
GLADROW e t a t
t
45001 A N D W O J C I E C H O W S K I (19S7A)
(19531
m
/
7 K-N-6
IO 0
5 4
i.e., somewhat less than 3. In actual fact, however, the reactivities differed by an order of magnitude. The strong effect of chain length was also observed in the very early days of catalytic cracking in a comparison between n-C16 and n'c8, in which 20% n-Cl6 was converted at 500 OC, against 10% conversion of n-C8 at 570 OC under otherwise identical conditions (Egloff et al., 1939). The new mechanism predicts a much stronger effect of chain length than the classicaltheory. Referring to Scheme I, we may assume that the reactivities should be proportional to the number of different combinations of n and m',if there is no preference for one value of n or m' over another (as long as they are greater than 0). With the latter simplifyingassumption it followsthat the reactivities will be proportional to the total carbon number minus 6. Thus, for n-C16 and n-C7, the predicted reactivity ratio will be 10, which appears to be more in line with the conversions observed by Voge. A more quantitative comparison is given in Table I, which lists the first-order rate constants calculated for the cracking experiments with an alumina-zirconia-silica catalyst as reported by Voge. The first-order kinetics were found to be applicable in the cracking of n-hexadecane (Voge, 1958). It can be inferred that although the agreement is not perfect, at least the order of magnitude of the reactivity differences is correctly predicted. The correspondence is sufficiently good to serve as supporting evidence for the new mechanism. A closer correspondence can hardly be expected because of the simplifyingassumption made, and the possibility that some preisomerization has affected the observed cracking rates supposed to be attributable to cracking of normal paraffins only. In addition, some differences in carbon level on the catalyst as well as adsorption of the heavier hydrocarbons may have played a role. Figure 1 compares the effect of carbon number on cracking reactivity of normal paraffins as predicted from the protonated cyclopropane mechanism (reactivity proportional to N - 6) with the effects predicted if the reactivity were proportional to the number of secondary carbon atoms ( N - 2) or the number of cleavable C-C bonds ( N - 1). The figure clearly shows the much steeper increase of reactivity with carbon numbers above 7,and the precipitous drop of reactivity below that number predicted by the proposed mechanism. In Figure 1 we have also plotted experimentally measured reactivities in terms of first-order rate constants relative to that of n-heptane, as deduced from published data by a number of investigators (Gladrow et al., 1953; Abbot and Wojciechowski, 1987a). The figure shows that the relative reactivities of normal paraffins between C7 and c24,as measured by these investigators, follow much better the trend predicted by the proposed mechanism (k ( N - 6)) than the classical carbenium-ion mechanism (k ( N - 2)). According to the present mechanism Cg
--
3 2
I 0.0
6
S
10
12
14
16
10
6 20 22 24 CARBON NUMBER,N
Figure 1. Comparison of the effects of carbon number on cracking reactivity, expressed as reactivities relative to that of n-heptane. The solid line is the relation predicted by the protonated cyclopropane mechanism (N - 6);the broken lines are based on proportionalities with the total number of secondary carbon atoms ( N - 2) and C-C bonds ( N - 1). Relative reactivitiesderived from experimental results published in the literature are indicated by points.
and lighter paraffins will be hardly cracked under conditions where higher paraffins undergo significant cracking. This is in line with experimental observations by Voge (Voge, 1958)and has implications for the distribution of cracked products, as will be discussed below. Characteristics of the Cracked Product: Carbon Number Distribution. The proposed mechanism provides a logical explanation for the fact that hardly any C1 and C2 hydrocarbons are formed, if at all, in catalytic cracking. With n and m' haviny values equal to or greater than 1,the smallest fragment that can be formed is C3 (see Scheme I). If the above simplifying assumption is valid, primary rupture of the carbon chain of the feed paraffin of carbon number N will have an equal chance of producing any fragment with a carbon number between 4 and N 4 ( N > 7). Each of these fragments may be formed both as a paraffin and as an olefin, and their molar amounts should therefore be larger than that of CBand CN-~,since the latter fragments can only be formed as either an olefin or a paraffin (cf. Scheme I). Under the usual conditions of catalytic cracking secondary cracking of the cracked fragments generally takes place and this will mostly affect the concentration of the heavier fragments, in view of the dependence of crackability on the length of the carbon chain, as discussed above. Since there is a rapid decline of the reactivity below C,, a peak in the carbon number distribution between 3 and 6 is to be expected when substantial secondary cracking occurs. This is indeed observed in practice, as is shown in Figure 2. The occurrence of substantial secondary cracking in this case follows from the fact that 359 mol of cracked fragmenta were found per 100mol of feed paraffin cracked (Voge, 1958). Figure 2 also shows a carbon number distribution obtained by cracking the same feed thermally, with a similar degree of secondary cracking (378mol of cracked product per 100mol of cracked feed (Vogeand Good, 1949). In this case the product distribution shows a peak at CZ, indicating that c4-c6 hydrocarbons are also subject to
Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 399 Table 11. Product Distribution from Short-Contact-Time Catalytic Cracking of n-Hexadecane and a-Dodecane (Experimental Data from Nace (1969). T = 482 O C ; P = 1 bar: Catalyst. R.E./H-X)
CRACKED PROWCT, mol/ 100mol
r O '
~~
feed LHSV conversion, %
---
CATALYTIC THERMAL
c3 c4
c5
1;
CS
\
c7
ca c9
ClO Cn
ClZ mo1/100 mol of C16 cracked 2
4
6
S
10
12
n412 650 9.0
12.1 27.7 23.6 17.5 9.7 4.3 2.8 1.3 0.6 0.4 307
16.9 31.0 22.6 17.6 7.7 3.4 1.0
product distribution, mol 70
so
I'
n-Cl6 1300 15.5
PRODUCT FROM CRACKING. *hmol
14
CARBON NUMBER
Figure 2. Comparison of carbon number distributions in products from catalytic and thermal cracking of n-hexadecane. Catalytic: alumina-zirconia-silica catalyst at 500 OC;359 mol of products/100 mol of cracked feed (Voge, 1958). Thermal: 500 O C ; 378 mol of producta/100 mol of cracked feed (Voge and Good, 1949).
cracking. The fact that the crackability of these hydrocarbons is not so much different from that of the higher ones, as contrasted to observations in acid-catalyzed cracking, can be understood from the simple @-scission mechanism in radical cracking. This constitutes another argument against the concept of 8-scission in acidcatalyzed cracking. The difference in carbon number distributions of Figure 2, which can be attributed to the differences in radical and carbenium-ion cracking, makes it plausible why catalytic cracking is the preferred process for making gasoline + liquefied petroleum gas while thermal cracking is the superior route for ethylene production. The proposed mechanism predicts a relativelylow molar amount of C3 in comparison with the fragments of higher carbon number, which however is barely reflected in the product distribution for acid cracking in Figure 2. This is understandable since secondary cracking of C7 does not produce less C3 than C4, and at the high temperatures possible contributions from other modes of cracking that are responsible for the formation of some methane and C2 found in the product (thermal cracking or cracking via the pentacoordinated carbonium ion mechanism discussed in the previous paper) cannot be ruled out. The lower production of C3 should become more apparent in experiments under milder conditions with less secondary cracking. Table I1 shows the results of experiments performed at very high space velocity, where conversions of the feed paraffin and the degree of secondary cracking were lower than in the experiment shown in Figure 2 (Nace, 1969). It can be inferred that C3 is indeed present in lower amounts than C4, Cg, and CS each. Newer studies on the cracking of n-dodecane and n-hexadecane by Abbot and Wojciechowski show similar carbon number distributions of the product. Initial molar distributions of cracked fragments obtained by cracking n-dodecane a t 400 "C over various catalysts (H-Y, La-Y, H-mordenite, H-ZSM-5 and silica-alumina) show a peak at Cr, with very little C1 and C2. C3 is indeed produced in smaller quantities than C4. However, the higher molecular weight fragmente to be expected from primary cracking of n-Cl2, like C8 and CS,are also found in relatively
401
0
0
2 4 6 8 10 CARBON NUMBER OF PROOUCT MOLECULE
Figure 3. Carbon number distribution in the product of cracking n-decane at very low conversion over a silica-alumina catalyst (Langlois et al., 1966;Langlois and Sullivan, 1969). 5" = 288 O C ; conversion < 2%.
small quantities only (Abbot and Wojciechowski, 1989). This is probably caused by the difficult desorption of these fragments from the catalyst, so that they have to fragment further before escaping to the vapor phase. Hence, even at low conversions and relatively low temperature the product spectrum may not be representative for pure primary cracking. Similar results were obtained by cracking n-hexadecane over H-Y zeolite a t 300 and 400 OC (Abbot and Wojciechowski, 1988). Again a peak in carbon number distribution is observed a t C4, with significant less C3 than C4, and hardly any C1 and C2. Also in this case relatively small amounts of products above C7 are observed. A better picture of the primary fragmentation pattern may be obtained by cracking at low temperature and at low conversion in the liquid instead of the gas phase. Figure 3 shows the carbon number distribution of the product obtained by cracking n-decane over a silicaalumina catalyst at very low temperature (288 "C) and very low conversion (less than 2%). Although cracking was carried out in the presence of hydrogen under pressure, the process was probably not hydrocracking since a metal function was absent on the catalyst (in the presence of a metal function, hydrocracking occurred and resulted in much higher conversions (Langlois et al., 1966; Langlois and Sullivan, 1969)). In this case, cracking is restricted to pure primary cracking, and it is of interest to note that the product apectrum very clearly shows the predicted lower production of CBas well as that of G - 3 , Le., C,.
400 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993 Table 111. Catalytic Cracking of n-Paraffins (Experimental Data from Greensfelder et al. (1949)) feed catalyst T,"C
LHSV
n-Cs UOP-B 550 1.2
n-Clz UOP-B 550 3.2
n-C16 UOP-B 500 4
n-C16 carbon 500 3.9
44 35 11 15 24 13 nd nd
82 28 24 35 21 nd nd nd
85 21 28 40 32 15 13 6
9 4 1 8 1 15 1 17
mo1/100 mol cracked c3=
C3' i-C4= n-C4' i44' n-C4" i-Cjo n-Cjo
Table IV. Catalytic Cracking of n-Hexadecane (Calculated from Experimental Data by Voge (1958). T = 500 "C) catalyst thermodyn A170s-Zr02-Si02 A120s-Si0:! equilib" ~
olefinicity in product C3=/C30 n-C4'ln-C4' i-C4'li-C4" branching in product i-C4=/n-C4' i-C4"/n-C4" i-Cjoln-Cjo
~
2.42 1.84 0.45
2.06 2.02 0.50
0.64 2.62 3.58
0.61 2.45 3.68
0.87 0.59 1.70 0 Calculated from API Project 44 data (1953, 1989). Butene-1, cis-butene-2,and tram-butene-2 are assumed to be in thermodynamic equilibrium.
Characteristics of the Cracked Product: Branching and Olefinicity. According to the protonated cyclopropane mechanism (cf. Scheme I) the CSformed should be unsaturated, the smallest saturated hydrocarbon being C4 in the form of isobutane. As can be deduced from Scheme I, the olefinic fragments formed should have a linear structure, while the saturated fragments should be branched. These predictions are in line with observations made in the cracking of normal paraffins over acidic catalysts, as reported by Greensfelder et al. and by Voge (Greensfelderet al., 1949; Voge, 1958). As can be deduced from Table 111,propene is present in excess over propane. In the unsaturated C4 fraction linear structures are predominant, whereas in the saturated C4and Cg fractions the is0 structures are prominent. These characteristics are in complete agreement with what can be expected on the basis of the proposed mechanism,although at the high prevailing temperatures and relatively long residence times some contributions from other reactions such as thermal cracking or cracking via the pentacoordinated carbonium ion route cannot be completely ruled out. These contributions tend to make the characteristics of the product spectrum resulting from acid cracking less distinct. Table I11 also shows data obtained with active carbon as catalyst. These data may be considered to be representative of thermal cracking under comparable conditions. They are of interest in that they indicate thermal contributions not to be completely absent under the conditions of the catalytic cracking experiments discussed above. Furthermore,they show the predominance of linear structures to be expected from radical cracking. In particular the low degree of branching in the saturated cracked fragments is in striking contrast with the observations made in acid-catalyzed cracking. Table IV lists some data on the olefinicity and branching observed in the products of catalytic cracking of n-hexadecane (Voge, 1958). In agreement with protonated cyclopane mechanism the olefinicity is high in the linear fractions (normal-olefin/normal-paraffinratio > 1) and
Table V. Branching in the C4 Fraction Obtained by Catalytic Cracking of a-Hexedecane (Calculated from Experimental Data from Gladrow et al. (1953). T = 510 "C) conversion,a isobutane/ isobutenei ?i n-butane n-butenes catalyst 2.57 0.69 silica-alumina 13.5 2.13 0.67 20.2 2.85 0.78 30.7 0.72 2.57 39.8 2.22 43.1 0.69 2.70 55.1 0.75 73.4 2.85 0.69 1.70 0.61 silica-magnesia 19.4 1.70 0.56 51.2 2.22 0.61 19.7 0
To products boiling below 221 "C.
low in the nonlinear ones (isoolefin/isoparaffin ratio < 1). Expressed in another way, viz., the degree of branching, it follows that the degree of branching is low in the olefin fraction and high among the paraffins. The iso/normal ratio of the butene fraction of about 0.6 is well below the values calculated for the thermodynamic equilibrium, as follows from the data included in Table IV. This means that n-butene is present in concentrations above equilibrium and that at least part of it is a product of the cracking reaction. The iso/normal ratios of the butanes and the pentanes, 2.6 and 3.6, respectively, are clearly above the calculated equilibrium ratios for the temperature of the cracking experiments; see Table IV. This implies that at least part of the isoparaffins is directly derived from cracking and is not produced by consecutive isomerization. Another argument against consecutive isomerization being the only source of branched paraffms can be derived from cracking experiments with n-hexadecaneover silicaalumina and silica-magnesia catalysts at different degrees of conversions (Gladrow et al., 1953). The isobutanelnbutane ratios are remarkably constant over the conversion range of 13-80%, which is not to be expected for a consecutive isomerization; see Table V. The same applies to the isobuteneln-butenes ratio. In agreement with the earlier discussed results, the iso/normal ratio in the butane fraction is considerably higher than the thermodynamic equilibrium ratio, whereas the opposite is true for the butene fraction. The more recent experimentsof Abbot and Wojciechowski on cracking of n-dodecane over various catalysts at 400 OC generally also show a clear predominance of isoparaffins over normal paraffins (Abbot and Wojciechowski, 1989). With H-Y, La-Y, H-mordenite, and amorphoussilica-alumina isolnormal ratios of initially produced paraffins with four to eight carbon atoms are above 1, reaching values as high as 5.5. An exception is H-ZSM-5, which gives rise to iso/normal ratios below 1. This is presumably caused by the shape-selective properties of this zeolite. The presently proposed mechanism will lead to 2methylalkanes as predominant paraffinic fragments (cf. Scheme I). Examinationof detailed product compositions obtained by cracking n-dodecane over various catalysts indeed shows that the 2-methyl isomers are the most abundant isomers among the C6 and C7 paraffins (Abbot and Wojciechomki, 1987b, 1989). The above cracking experiments also show the predominance of propene over propane predicted by the present cracking mechanism. However, in the C4 and Cg olefins branched structures are also present in relatively large proportions. This may be caused by the difficulty which larger fragments experience at the relatively low
Ind. Eng. Chem. Res., Vol. 32, No. 3,1993 401 Table VI. Typical Composition of the C4 Fraction from Two Catalytic Cracking Processes, As Reported by Borrows and Seddon (1953) fluid moving bed components catalytic cracking catalytic cracking isobutane, wt % 48 58 n-butane, wt ?6 10 12 imbutane, wt 9% 10 7.5 n-butenes, wt 7'% 32 22.5 isobutane/ n-butane 4.8 4.8 isobutene/n-butenes 0.31 0.33
temperature in leaving the catalyst (as mentioned in the previous section). If these fragments have a chance of undergoing skeletal isomerization, the cracked olefinic fragments could well be branched olefins. Another factor which may affect the olefinicity and branching in the cracked products is the Occurrence of hydrogen transfer. Hydrogen-transfer reactions leading to the formation of aromatics and coke besides paraffins has been demonstrated to occur, in particular a t low temperatures, with zeolite catalysts (Abbot and Wojciechowski, 1987b, 1988). On the basis of the bulk of the above evidence it can be concluded that linear olefins and isoparaffins are likely to be the primary products in the acid-catalyzed cracking of normal paraffins, which is in line with the proposed new mechanism but in clear conflict with the classical theory (cf. previous paper).
Catalytic Cracking of Practical Feedstocks Notwithstanding the fact that the composition of a practical feedstock is quite complex and molecular structures are less well defined than the pure normal-paraffins discussed above, the characteristics of the cracked products predicted by the protonated cyclopropane mechanism are well recognizable in actual catalytic cracking processes. Table VI shows typical compositions of the C4 fractions from fluid catalytic cracking and moving bed catalytic cracking (Borrows and Seddon, 1953). It can be inferred that isobutane is present in excess over n-butane, whereas isobutene is produced in lower amounts than n-butenes. Just as in the products obtained with pure paraffinic feedstocks the isobutaneln-butane ratio is very high and well above the thermodynamic equilibrium values for the temperature levels at which the cracking processes are usually operated. In contrast to this are the low isobutene/ n-butenes ratios, which are well below the equilibrium ratios. The above data pertain to catalysts as used in the prezolite period. In more modem catalytic crackingprocesses riser reactors and zeolitic cracking catalysts are commonly used. Table VI1 shows the distribution of C145 hydrocarbons in the product obtained by cracking of a waxy vacuum gas-oil over a modern zeolite-containing catalyst in a riser reactor (Van Els et al., 1990). It can be seen that the product breakdown has all the characteristicspredicted for the protonated cyclopropane mechanism: very little Cland Cz,a C3 fraction consisting largely of propene, a strong predominance of branched structures among the butanes and pentanes, and a predominance of normal structures among the pentenes. Table VI11 shows the carbon number distributions of the C145 products obtained in the same experiments. It can be seen that C3 is produced in lower amounts than Cq on a molar basis, as predicted by the theory. Although the difference is not large, it is considered significant, since secondary cracking and contributions from other modes of cracking will tend to increase the amount of C3 at the expense of higher hydrocarbons, and formation of higher
Table VII. Yields (in wt %) of C& Hydrocarbons in Riser Catalytic Cracking of a Waxy Vacuum Gas Oil over a Zeolite-Containing Catalyst (Pilot-Plant Experiments by Van Els et al. (1990)) product methane ethane ethene propane propene isobutane n-butane butene-1 + isobutene cis- + trans-butene-2 isopentane n-pentane n-pentenes isopentenes
T = 600.8 O C 0.36 0.21 0.42 0.88 5.85 3.20 0.85 4.95 4.74 4.11 0.65 4.39 2.56
T = 503.9 O C 0.46 0.28 0.60 1.04 7.89 3.95 1.12 6.23 6.12 4.88 0.76 5.45 3.26
Table VIII. Carbon Number Distribution and Branching in the ClCs Products Obtained by Riser Catalytic Cracking of a Waxy Vacuum Gas Oil (Pilot-Plant Experiments by Van Els et al. (1990)) T = 500.8 O C T = 503.9 O C yields in mo1/100 mol of feed cracked 9 11 c1 12 a c2 60 79 c3 92 117 c4 77 63 c5 iso/normal ratio in butanes 3.8 3.5 c1.0 a.0 butenes 6.3 6.4 pentanes 0.58 0.60 pentenes
paraffins from C3 is very unlikely. The table also shows that the iso/normal ratios in the C4 and Cg saturates are very much higher than in the unsaturated fractions, and well above the thermodynamic values listed in Table IV. The expected predominance of the is0 structures in the saturated products of catalytic cracking is not restricted to butanes and pentanes; it can also be found in higher fractions. In catalytic crackingexperiments with different catalysts Magnussen and Pudas reported 4-5 times higher yields of isoheptanes than of n-heptane (Magnussen and Pudas, 1985).
Conclusions The proposed new mechanism for acid-catalyzed cracking of paraffinichydrocarbons,which assumes a protonated dialkylcyclopropane as a reaction intermediate, is in good agreement with many characteristic features of catalytic cracking as observed experimentally. The mechanism accounts for the generally low production of CIand CZ, the strong effect of increasing chain length of paraffins above C7, the refractoriness of Ce and lower paraffins toward cracking, and the resulting tendency for a carbon number distribution to peak between CS and C7, if secondary cracking occurs. The mechanism predicts that CS should be largely propene, and that the saturated hydrocarbons should be characterized by a high degree of branching, in contrast to the olefins, which should be predominantly linear. All these features, which are borne out in actual practice, cannot be explained by the classical carbenium ion/@-scissiontheory, which fails to provide a satisfactory explanation for the dependence of crackability on chain length, and which would predict the opposite trends with respect to branching and olefinicity.
402 Ind. Eng. Chem. Res., Vol. 32, No. 3, 1993
Literature Cited (1) Abbot,J.; Wojciechowski,B. W. Kinetics of Catalyticcracking of n-Paraffins on HY Zeolite. J. Catal. 19878,104,80-85. (2) Abbot,J.; Wojciechowski,B. W. HydrogenTransfer Reactions in the Catalytic Cracking of Paraffins. J. Catal. 1987b,107,451462. (3)Abbot, J.; Wojciechowski, B. W. The Effect of Temperature on the Product Distribution and Kinetics of Reactions of n-Hexadecane on HY Zeolite. J. Catal. 1988,109,274-283. (4)Abbot, J.; Wojciechowski, B. W. Catalytic Reactions of n-Dodecane on Aluminosilicates. J. Catal. 1989,115,521-531. ( 5 ) API Project 44. Selected Values of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds; Rossini, F. D.; Pitzer, K. S.;Arnett, R. L.; Braun, R. M.; Pimentel, G. C. Carnegie Press: Pittaburgh, PA, 1953. (6)API Project 44. TRC Thermodynamic Tables-Hydrocarbons. Thermodynamic Research Center, Texas A & M University: College Station, TX, 1989. (7) Borrows, E.T.; Seddon, W. L. Chem. Znd. 1953,Aug, 57-74. (8)Egloff, G.; Morrell, J. C.; Thomas, C. L.; Bloch, H. S. The Catalytic Cracking of Aliphatic Hydrocarbons. J. Am. Chem. SOC. 1939,61,3571-3580. (9)Gladrow, E.M.; Krebs, R.W.; Kimberlin, C. N., Jr. Reactions of Hydrocarbons over Cracking Catalysta. Znd. Eng. Chem. 1953, 45,142-147.
(10)Greensfelder, B. S.;Voge, H. H.; Good, G. M. Catalytic and Thermal Cracking of Pure Hydrocarbons. Znd. Eng. Chem. 1949, 41,2573-2584. (11) Langlois, G. E.; Sullivan, R.F. Chemistry of Hydrocracking. Symposium on Refining Petroleum for Chemicals;National Meeting of the American Chemical Society, New York, Sept 7-12, 1969; Prepr.-Am. Chem. SOC.,Diu. Pet. Chem. 1969,14 (3),D 18-39. (12)Langlois, G.E.;Sullivan, R. F.; Egan, C. J. The Effect of Sulfiding a Nickel on Silica-AluminaCatalyst. J.Phys. Chem. 1966, 70,3666-3671. (13) Magnussen, J.; Pudas, R.Activity and Product Distribution Characteristics of the Currently used FCC Catalyst Systems. Paper presented at the Katalistiks’ 6th Annual Fluid Catalytic Cracking Symposium, Mtinich, May 22-23,1985. (14) Nace, D. M. Catalytic Cracking over Crystalline Aluminosilicates. Znd. Eng. Chem. Prod. Res. Dev. 1969,8, 24-33. (15) Sie, S.T. Acid-Catalyzed Cracking of Paraffinic Hydrocarbons. 1. Discussion of Existing Mechanisms and Proposal of a New Mechanism. Znd. Eng. Chem. Res. 1992,31,1881-1889. (16)Van Els, A. C. C.; Breed, P.; Kompier, G. Koninklijke/ShellLaboratorium, Amsterdam; unpublished work, 1990. (17) Voge, H. H.Catalytic Cracking. In Catalysis; Emmett, P. H. Ed.; Reinhold: New York, 1958;Vol. VI, Chapter 5,pp 407-492. (18) Voge, H. H.; Good, G. M. Thermal Cracking of Higher Paraffins. J. Am. Chem. SOC.1949,71,593-597.
Received for review November 16, 1992 Accepted November 28, 1992