Ind. Eng. Chem. PrOC8SS Des. Dev. 1981, 20, 307-313 307

Thomas, W. J.; Ogboja, 0. Ind. Eng. Chem. Process Des. Dev. 1978, 17,

Unno, H.; Inoue, I. J. Chem. Eng. Jpn. 1978, 9(2), 92.

McAlllster, R. A. et al. Chem. Eng. Sci. 1958, 9 , 25. Miller, R. H. Ph.D. Thesls, Unhrersity of Mlchlgan, 1959. Nutter, D. E., 88th National Meeting of AIChE, Houston, Paper No. 49C, Feb

429.

1971. Porter, K. E.; Wong, P. F. Y. Inst. Chem. Eng. Symp. Ser. 1080, 32, 2:22. Ram, V. C.; Pavlov, V. P. Khim. from. 1968, &(IO), 776. Rush, F. E., Jr.; Stirba, C. AIChE J. 1957, 3(3), 336. Sterbacek, 2. Brn. Chem. Eng. 1987, 12(10), 1577. Takahashi. T. et al. J. Chem. Ena. J m . 1973, 6(1). 38.

Received for review February 29, 1980 Accepted December 19, 1980

Presented at the AIChE Annual Meeting, San Francisco, Calif., Thomas, W. J.; Campbell, M. Trans. inst. Chem.'€ng. 1987, 45, T53. Nov 1979.

Steam Cracking of High-Molecular-Weight Hydrocarbons

Byouk Blourl, Jean Glraud, Sirous Nourl, and Danielle Herault Laboratoire de G n i e et hformatique Chimiques, €Cole centrale des Arts et Manufactures, 9229O-ChEitenay Malabry, France

The steam cracking of n-tetracosane, 6-methyleicosane, and dodecylbenzene was studied between 550 and 700 OC in a tubular stainless steel reactor whose walls had been passivated by chromaluminization. The kinetic data on the cracking of these heavy hydrocarbons are similar to those previously reported for light hydrocarbons. The order of the reactions is close to 1; the activation energies are 58.2 kcal for tetracosane, 56.3 kcal for 6- methyleicosane, and 45 kcal for dodecylbenzene. The composition of the cracked products of n-tetracosane, in particular for low advance rates, agrees with the prediction of Rice's free-radical theory, while those of 6- methyleicosane and dodecylbenzene are different: there is a slight formation of cracked gas and a larger conversion of the heavy treated hydrocarbons into light fluid ones.

The thermal cracking of hydrocarbons has been the object of much work. In 1931 Rice proposed his classical rules involving the formation of free radicals for the de- composition of these hydrocarbons. For light hydrocarbons (cI-c6), the theory of Rice and Kossiakoff explains the formation of reaction products, in particular when cracking occurs under relatively mild operative conditions: low temperatures (500-650 "C) and low conversion rates.

As mentioned earlier, the cracking of heavier hydro- carbons, such as naphtha or gas-oil, is much more complex (in particular for the preparation of light olefins requiring a higher conversion rate). We studied (Blouri and Giraud, 1977) the steam-cracking of high-molecular-weight hy- drocarbons.

In the present paper are reported the experimental re- sults of a kinetic and chemical study of the cracking of three high-molecular-weight hydrocarbons (with more than 17 carbon atoms): n -tetracosane, 6-methyleicosane, and dodecylbenzene. The composition of the obtained cracked products is compared with the predictions of Kossiakoff- Rice theory (1943). The results are applicable to the in- creasing use of high-molecular-weight hydrocarbons of petroleum and to preparation of raw materials in the pe- trochemical industry. Previous Work

There is much work available on the cracking of mole- cules of higher molecular weight. Kunzru et al. (1972) studied the pyrolysis of n-nonane and Illes et al. (1973) the cracking of n-octane, isooctane, cyclohexane, and (&-branched hydrocarbons.

According to their results, the distribution of the reac- tion products is predicted to within 20% by the theory of Rice and Kossiakoff, but this mechanism was not tested on molecules representative of a fuel oil by their molecular weight. Experimental Section

Apparatus. The dynamic reactor used consisted of three coaxial tubes arranged as shown in Figure 1. In order to better differentiate the vaporization and pre-

heating zones from the reaction zone, water and hydro- carbons were vaporized and preheated in zones I and 11, respectively. Zone 111 was the furnace proper. To improve the thermal profile, the furnace was plunged into a fluid- ized sand bed (particle diameter, 150 pm; 20 L/h of ni- trogen preheated at 500 "C).

Each outer tube was 360 mm in length and 22 and 25 mm in diameter. Each internal tube was 350 mm in length and 14 and 16 mm in diameter. The three sections of the apparatus were heated by electrical resistors (- 20 a) with adjustable pitch. The temperatures were read regulated with the aid of chromel-alumel thermocouples and regu- lators.

The thermal profile of the operating furnace was reg- istered by the mobile thermocouple 4. The thermal iso- lation was provided by a kaolin layer and the fluid pressure was measured at the top (tube 5).

The hydrocarbon was introduced into section I by means of a syringe and of an electrical push-syringe at a flow rate of about 16 g/h. The preheating temperature was 500 "C. In our operating conditions, the treated hydrocarbons have no significant reaction in the preheater. Water was pumped by a peristaltic pump at a flow rate varying from 20 to 150 g/h at the same preheating temperature. The reaction products evacuated through tubing 6 were tem- pered by a refrigerant cooled with tap water and collected in a vessel plunged into melting ice. The noncondensed gaseous products were collected in a Mariotte bottle.

The effluents were collected when the apparatus was in equilibrium, i.e., 15 mn after the beginning of each run. The apparatus had been previously flushed out with cir- culating water for 1 h (E 50 9). After each run the possible carbon deposits where burned out by the passage of air heated at 600 "C.

Wall Effects. The catalytic effect of the walls on the pyrolysis reactions, in particular in the case of stainless steel tubes, has been pointed out in many publications (Albright, 1975). The three sections of our apparatus were made of stainless steel 18/8 protected from oxidation and carburation up to 1200 OC by a 200 pm thick deposit of

0 196-4305/81 I 1 120-0307$01.25/0 @ 1981 American Chemical Society

308 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981

! 1; .. . . . . . , . , ' . . . . ..

'. . _' ..

Figure 1. Experimental device.

chromium and aluminum (chromaluminization). A previous set of runs during which a mixture of n-

paraffins in C,-CB was cracked between 800 and 950 "C in the presence of water showed that a tube of steel 18/8 passivated in this way was stable and led to the same results as a quartz tube, quartz being known to produce small catalytic effects (Pines and Arrizo, 1957).

After 60 runs of about 1 h each alternating with 100 h of regenerations by the passage of air a t 600 L/h (about 1.5 h for each) the results were still perfectly reproducible, contrary to numerous other alloy steels which give more or less rapidly significant amounts of carbon and hydrogen (Partovi et al., 1978).

Analysis. The liquids were analyzed by gas chroma- tography in a chromatograph fitted with a flame ionization detector (the peak surface was determined with an inte- grator) by means of 3 m long, in. diameter s t a i n l a steel column packed with SE 30 on Chromsorb W/AW, 60-80 mesh. The temperature was programmed from 50 to 210 "C at a linear rate of 10 "C/mn.

The gases were analyzed in a chromatograph fitted with a flame ionization detector by means of a stainless steel column which was 4 m in length and in. in diameter and contained Porasil B 80-100 mesh. The hydrogen/ methane ratio was determined by means of a third chro- matograph fitted with a catharometer and of a 2 m lon

in. diameter stainless steel column packed with 5 molecular sieves.

Treated Hydrocarbon. The 6-methyleicosane was prepared in our laboratory by allowing heptanone-2 to react with a magnesium derivative of bromo-1 n-tetrade- cane. By hydrolyzing the resulting complex 6-methyl-6- hydroxyeicosane was obtained, whose hydrogenation under 150 kg/cm2 in the presence of a modified Adkin's catalyst gave 6-methyleicosane with 97.5% purity. Tetracosane

k

and dodecylbenzene with 97% purity are commercial products. The hydrocarbon purity was controlled by gas chromatography. Results

(1) Kinetic Analysis. The experiments were carried out between 550 and 750 "C at a small hydrocarbon partial pressure (e&%), and at the maximal flow rates and tem- peratures, the Reynolds number was Re N 230. The flow was thus laminar. In the following a plug flow was as- sumed.

The furnace was rendered nearly isothermal by the use of a fluidized bed. However, to determine more accurately the reaction volume, the equivalent reactor volume concept as developed by Hougen and Watson (1947) was used.

Since the conversion rate in the equivalent reactor is the same as in the real one, it is easy to deduce for a reference temperature T, the expression for the equivalent volume V , as a function of the real volume V

The activation energy E is set equal to 60 kcal. The volume V, calculated with it gives a residence time t leading to a rate constant K and, therefore, to a new value E of the activation energy which replaces the preceding value E. The method had a rapid convergence. The order of reaction was determined using the relation established by Kershembaum and Martin (1967) for constant pressure and temperature profiles

The rate constants were determined by assuming from numerous studies that the order of the hydrocarbon cracking kinetics was close to or equal to 1.

When this assumption is verified, the constant K of the reaction can be determined using Benton's formula (1931)

A ---.+ vB

log (-Af) = n log X + log K

k

hence 1

l - x k t = v ln- - (v - l )X

Since Y is Micult to determine when studying the cracking of high-molecular-weight hydrocarbons, we have estab- lished the kinetics by studying the variation of

1 log - = f(t)

as a function of the residence time t , where a is the hy- drocarbon conversion rate.

(a) Order of the Reactions. Figures 2, 3, and 4 il- lustrate the application of the Kershenbaum-Martin re- lation to n-tetracosane, 6-methyleicosane, and dodecyl- benzene, respectively. The values of n deduced from these graphs are listed in Table I.

The order of the three reactions is close to 1 and not very sensitive to a significant increase in temperature. For n-tetracosane, the order increases with the temperature; for dodecylbenzene, it decreases with increasing temper- ature.

(b) Rate Constants. Figures 5, 6, and 7 show the variation of the logarithm of the 1/1- a ratio as a function of the residence time (calculated after determination of the equivalent volume). The values found for the rate constants k are listed in Table 11.

l - f f

L8

-5

- 4

- 3

- 2

( - A n 600'

6 -5 Log ( A )

- - - 4

Figure 2. Logarithm of rate of transformation vs. logarithm of average mole fraction of n-Czr

1

- 4

- 3

- 2

- 1

I - z € 1

600'

- 4 -5 -6 Log t T )

Figure 3. Logarithm of rate of transformation vs. logarithm of average mole fraction of 6-methyleicosane.

Table I order of reaction, n

temp, 6-methyl- dodecyl- "C n-tetracosane eicosane benzene

600 1.07 1.06 1.18 630 1.10 1.12 650 1.16 660 1.17 1.11 690 1.18 1.11 720 1.13

Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981 309

Logl-:F)

1 1

Log ( F ) - 4 -i

Figure 4. Logarithm of rate of transformation vs. logarithm of average mole fraction of dodecylbenzene.

Residence t m e s

A n124

720'

1 1.0 2.0 3.0 Log-

1-4

Figure 5. Log [l/(l - a)] vs. residence time.

Table I1 rate constants. k. s-I

6-methyl- dodecyl- temp, "C n-tetracosane eicosane benzene

550 0.14 600 0.57 1.19 0.55 630 1.59 1.38 650 3.33 660 4.10 2.82 690 7.32 4.90 920 12.50 "_

310 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981

~ ~ s l d e n c e time s .

f

n-tetracosane 58.2 1.6 x 1014 6-methyleicosane 56.3 1.2 x 10l4 1 0

dodecy lbenzene 45.0 1.2 x 10"

6-Methyleicosane

/ ,530"

Dodecylbenzene

For n-tetracosane, the values of k are independent of the conversion rate a. For dodecylbenzene, they slightly increase and for 6-methyleicosane they sensibly decrease with increasing a.

(c) Activation Energy. From the graphs of the func- tions

we could deduce the values for the global activation en- ergies (E,) and the DreexDonential factors (A) as shown in"Tab1e %I.

Since no data were available on the global activation energies of cracking of such hydrocarbons, it was not posssible to compare these values with other results. The activation energy of n-paraffins is close to 60 kcal. The cracking of ramified hydrocarbons requires a slightly lower energy activation, that of aromatic hydrocarbons requires a slightly lower energy activation, and that of aromatic hydrocarbons requires a much higher energy.

(2) Composition of the Cracked Products. (a) n- Tetracosane. Table IV gives the molar distribution of

2 r L O C L I- <

Figure 7. Log [l/(l - a)] vs. residence time.

the cracked products of n-tetracosane obtained from two runs under the following conditions: cracking temperature, 660 "C; n-tetracosane flow rate, 16 g/h; steam flow rate, 50 g/h for 80% conversion, 122 g/h for 48% conversion.

Table IV. Molar Product Distribution in Thermal Cracking of N-Tetracosane conversion weight, %

48 80 48 80 48 80 ~~

gases liquids

mo1/100 mol cracked 26.0 36.1 55.0 80.0 22.0 32.0

13.9 C3H6 53.8 87.8 1 -C,H, 19.0 32.0

C4H6 5.8 14.4

H2 CH4 CZH, C,H, C3H8 4.9

2-C4H8 0.2 1.0

139 180

7.3 6.6 6.6 5.5 6.1 5.8 5.3 3.0 5.3 3.3 4.0 3.4 3.6 1.5 3.3 1.6 1.2 1.5

MO.

1 5 0

IO0

5 0

m o D u c ' 1

0 EX?ERIE!ICE

R-K THEORY a

'3'6 'qhC '4'F I1 C H 4 C z i i 6 C 2 B 4 C3HC

Figure 8. Products in the cracking of n-tetracwane. Experimental data vs. R-K theory (2' = 660 "C).

1-Pentene is not present among these cracked products, since this hydrocarbon is intermediate between the gases and liquids (mp 27 "C) and is lost during the experiment. Comparison of the Results with Rice's Hypothesis

For this first we have used the method developed by Kossiakoff and Rice in 1943 and assumed that all radicals with more than five atoms were isomerized before cracking.

There is first formation of a great number of C, radicals, n being the number of carbon atoms in the molecule. These radicals statistically distributed according to the site number (in reality, the number of hydrogen atoms avail- able) and, moreover, to the exp(E/RT) ratio, E being the difference of the energies binding C-H for each available site. We have taken the following average values: E = 4 kcal between a primary and secondary site and E = 8 kcal between a primary and a tertiary site. For example, for 6-methyleicosane at 600 "C the following repartition in CP1 radicals is possible: primary radicals, 9 X 1 = 9 (7.2%); secondary radicals, 34 X 3.14 = 106.76 (84.7%); tertiary radicals, 1 X 9.88 (7.9%).

Before cracking, each radical is isomerized statistically on the available sites, i.e., from carbon in the 6 position. The statistic ratio is the ratio exp(AE'/RT) with AE'the stabilisation energy after isomerisation: E' = 2000 cal for primary to secondary and E ' = 4000 cal for primary to tertiary. This ratio is a function of available sites.

For the 6-methyleicosane, we add the repartition of the 21 formed radicals, and each radical has a &scission; then the new formed radicals are isomerized and cracked. The molar distribution in thermal cracking of n-tetracosane is represented in Table IV.

The distribution of the gaseous products according to their number of carbon atoms is represented schematically in Figure 8. This graph also reproduces our experimental results obtained at 660 "C (for a 50% conversion rate), expressed as moles per 100 moles of cracked n-tetracosane.

Despite some uncertainties concerning the differences between the activation energies for the abstraction reac- tions of a hydrogen atom between a primary and a sec- ondary carbon atom, as well as between the rearrangement energies of a primary radical, our experimental results agree reasonably well with the results predicted by the R-K theory, in particular for propene and 1-butene, where the difference is of the order of 9%.

The poorest agreement is found between the R-K theory and our results concerning the formation of hydrogen and

Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981 311

fbles P r d u b

100

50

Experience a R-K Theory

H I C", C2H6 C21i4 c p g C3H6 C ' p C C q H s

Figure 9. Products in the cracking of 6-methyleicosane. Experi- mental data vs. R-K theory (T = 600 "C).

ethylene, which are in excess, and methane and ethane, which are in deficit. The following reactions would pre- cisely support the formation of hydrogen, ethylene, and propane at the expense of ethane and methane

CH3CHzCH2 - CH3CH=CHz + H. fi + RH - R + H2

instead of CH3CH2 + RH - CH3CH3 + R (E, = 12 kcal) CH3CH2CH2 - CH2=CH2 + CH3 (E, = 31 kcal)

CH3RH - CHI + R The formation of liquid a-olefins is in agreement with

the R-K theory, but their cracking into gaseous products may explain the deviation of our experimental results obtained for the cracking of n-tetracoaane. In the cracking of n-tetracosane, the absence of aromatics, in particular of benzene, is also remarkable.

One knows that the dehydrocyclization of the alkenyl radical is a way of thermal aromatization of the hydro- carbons. Under our experimental conditions these radicals probably undergo a classical 8-cleavage into a-olefins or a,u-dienes and another radical RCHCH2CH2CH2CH=CH2 -

RCH=CH2 + CH2CH2CH=CH2 without any dehydrocyclisation. (b) 6-Methyleicosane. In Table V are summarized the

molar distributions of the cracked products of 6-methyl- eicosane, determined from two runs under the following operative conditions: conversion in % ,3346; temperature ("C), 600-650; steam flow rate in g/h, 51-67. The 6- methyleicosane flow rate was 16 g/h for the two runs.

Figure 9 shows a comparison of the experimental values, expressed in moles per 100 moles of cracked hydrocarbons (obtained at 600 "C for a 50% conversion rate of 6-

CH3CH2 - CHz=CH2 + H. (E, = 39 kcal) (E, = 30 kcal)

(E, = 8 kcal)

(E, = 8.4 kcal)

312 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981

Table V. Molar Product Distribution in Thermal Cracking of 6-Methyleicosane conversion weight, %

~

33 66 33 66 33 66 gases liquids

mol/100 mol cracked

9.1 8.8 1-C13H26 7.0 5.9 HZ 8.4 15.2 1-C,H1, CH4

C3H8 5.9 2.9 1-C8H16 14.4 13.9 1-CISH30 t 2.2

2-C4H8 t t i-C9H18 3.4 2.6 i-CI6H, 3.0 3.7 C4H6 1.8 5.0 1 -C,oHw 7.9 5.0 1-C17H34 11.5 5.7

i-CloHw 3.8 3.1 i-CI7Hw 5.1 5.4 1-C11Hzz 5.3 5.6 1-C1,H36 1.1 1.2

i-C12H, 3.0 4.4

i-C13Hu 4.7 4.1 31.6 33.6 i-C6H12 18.1 14.7 1-C7H14 8.4 11.5 1-Cl,H, 9.2 10.7

'ZH4 54.2 73.1 i-C H 14 4.6 7.5 i-C14H% 2.9 3.6

C,H6 25.7 33.1 i-C8H16 4.7 4.6 i-C,,H, 2.9 3.6 1-C,H8 7.7 11.2 1-CpH18 5.1 8.2 1-Cl6H32 4.0 3.6

'ZH6

i-C11H22 12.3 5.7 i-Cl8HS 8.2 3.6 1-C1,H, 7.8 7.0 benzene 2.1 23.6

Table VI. Molar Product Distribution in Thermal Cracking of Dodecvlbenzene conversion weight, %

20 43 81 20 43 81 20 43 81 gases liauids

mo1/100 mol cracked H2 5.5 9.6 26.3 1-C6Hlz 5.2 CH, 18.3 16.3 28.5 l-CIHl, 6.1 CzH6 10.1 7.7 8.8 1-C,H16 5.1 CzH, 33.2 39.4 10.1 1-C,H18 3.5 C,H8 0.3 1.4 4.1 l-CloHzo 3.0 C,H6 13.8 15.1 33.0 l-CllHzz 11.6 C,H, 4.6 4.8 11.6 C.H, 1.2 C;H, 0.6 1.0 5.0 C,H;CH3 13.0

C6HS(CH3)Z 3.2 C6H,CH=CHz 28.3

methyleicosane) with the values predicted by the R-K theory.

The agreement is very good for methane, but for the other gaseous constituents there is a lack of cracked products as compared to the theory, which tends to de- crease with proceeding reaction. Such a phenomenon is, however, not observed with n-tetracosane, except a t high temperatures and progression rates a t which one may assume a parallel cracking of the a-olefins formed.

The cracked liquids are mostly composed of normal or branched (Cs-Czo) a-olefins. There are also aromatic hy- drocarbons, among which benzene was identified (3% of the cracked liquids), but no heavier aromatics could be detected, in particular after the mixture analysis by gas chromatography.

The olefins composition determined by experiment is a little higher than that predicted by the R-K theory, and it is more important as the members are lighter.

Moreover, this theory does not predict the formation of the heavier aromatics and a-olefins (C15, C17, C18) that we obtained in small amounts. For these heavy a-olefins, the amount of 1-hendecene and, in particular, of 1-tetradecene is found to increase significantly with the temperature. With a 91% reaction rate, 20 mol of 1-tetradecene are obtained from 100 mol of cracked 6-methyleicosane.

Such a preferential formation may be imputed to preferential cleavages on C6 of 6-methyleicosane

CH3

C H j ( C H;! .$ H t C H,(C H p ) &H3

In reality, owing to the different binding energy of a

1.0 6.1 1.3 5.6 5.1 5.0 4.8 4.0 4.7 3.2

16.6 6.7 t 4.3 7.7 10.6 4.2 7.8

19.7 16.4

Table VI1

3-phenylpropene 5.0 8.2 8.0 4-phenyl-2-butene 3.5 7.9 5-phenyl-1-pentene 7.7 5.7 6-phenyl-1-hexene 8.7 4.1 7-phenyl-1-heptene 8.0 3.3 8-phenyl-1-octene 4.3 2.8 1,2-diphenylethane 5.1 5.9

10-phenyl-1-decene 9.6 3.9 9-phenyl-1 -nonene 6.9 2.3

11-phenyl-1-hendecene 3.4 1.3

conversion 20 43 81 temperature, "C 600 630 690 steam flow rate, g/h 75 104 106

tertiary C-H (of the order of 2 kcal), it is easier to obtain a tertiary radical, which is cleaved into a molecule of a- olefin, and another secondary radical which will be dehy- drocyclized into benzene. In fact, together with an im- portant formation of 1-tetradecene, there are also 12 mol of benzene for 100 mol of cracked hydrocarbon.

(c ) Dodecylbenzene. In Table VI are summarized the molar distributions of the cracked products of dodecyl- benzene determined from three runs under the conditions listed in Table VII. The dodecylbenzene flow rate was 18.2 g/h for the three runs.

Figure 10 shows the results obtained for gaseous mix- tures resulting from the cracking of dodecylbenzene and the values calculated with free-radical Rice-Kossiakoff model. Except for hydrogen, propane, and butadiene, there is a great deviation (deficit) for all the other gaseous constituents. Dodecylbenzene is clearly very sparingly cracked into gas.

For the composition of the cracked liquids, some agreement is found for the a-olefins and a-olefins with a phenyl end group, but the R-K theory does not take ac- count of the addition of a phenyl radical on the a-olefins, which explains the significant deviation of the experi- mental results from the theory.

For cracking occurring below 700 OC, as was the case in the present work, dodecylbenzene is mostly converted into a cracked liquid. For low conversion rates, i.e., 20-30%

I d . Eng. Chem. Process Des. Dev., Vol. 20, No. 2, 1981 313

formation of styrene by abstraction and subsequent B- cleavage is more probable. The amount of styrene is ac- tually very high for small progression rates of reaction.

The a-olefins and a-olefins with a phenyl group are formed according to the free radical mechanism and to the rule of @-cleavage of the Rice-Kossiakoff hypothesis. Conclusions

The experimental study of the cracking of three high- molecular-weight hydrocarbons belonging to three types of hydrocarbons-normal type (n-tetracosane), branched type (6-methyleicosane), and aromatic type (dodecyl- benzene)-has revealed some common characteristics of the cracking reactions: nearly first order kinetics; acti- vation energies close to those reported in the literature for lighter hydrocarbons (nonane, hexadecane); and funda- mental difference in the cracking mechanism.

For a normal paraffin (n-tetracosane C24H50) the cracking mechanism of the radical type and the compo- sition of the cracked products formed exclusively of gases and a-olefins agree fairly well with the predictions of Rice-Kossiakoff s theory.

For an hparaffii (6-methyleicosane C21H44) the reaction mostly develops according to a radical mechanism, but there also occurs an intramolecular cleavage at the branching which favors the liquid formation at the expense of the cracked gases and the composition of the products obtained differs from the Rice-Kossiakoff predictions.

For a long chain aromatic hydrocarbon (6-methyl- eicosane) simultaneous cleavages occur according to radical and intramolecular mechanisms with an important for- mation of cracked liquids a t the expense of the cracked gases and the composition of the resulting products is sensibly different from the theoretical one.

From a practical point of view, the cracking of normal paraffins gives the best results in the preparation of gases rich in ethylene, which is the basic product in the petro- chemical industry.

By selective cleavages at the ramification of high-mo- lecular-weight isoparaffins one can obtain more cracked liquids of low molecular weight which can be regarded as light fuels.

The presence of an aromatic ring on a long chain highly modifies the cracking mechanism of the hydrocarbon treated. So, by pyrolyzing dodecylbenzene, one can obtain very little gases and a large amount of liquids rich in styrene, a-olefins and a-olefins with an end phenyl group, which are very precious intermediates in the petrochemical industry. Literature Cited Albright, L. F. Chem. Eng. News 1976, 53, 29, 8-12. Benton, A. F. J . Am. Chem. Soc. 1931, 53, 2984. Blouri, B.; Giraud, J. Inf. Chim. 1977, 171, 229. Hougen, 0. A.; Watson, K. M. "Chemical Process Prlncipales". Vol. 111, WC

Illes, V.; Welther, K.; Fleskats, I. Acta Chlm. (Budapest) 1973, 78 (4), 357. Kershembaum, L. S.; Martin, J. J. AIChE J. 1987, 13(1). 148. Kosslakoff, A.; Rice, F. 0. J. Am. Chem. Soc. 1943, 85, 520. Kunzru, D.; Shah, Y. T.; Stuart, E. B. Ind. Eng. Chem. Process Des. Dev.

Miller. D. B. Ind. Eng. Chem. Process Des. Dev . 1983, 2. 220. Partovi, A. R.; Giraud, J.; Blouri, B. Inf. Chlm. 1978, 184, Ill. Pines, H.; Arrlzo. S. T. J. Am. Chem. Soc. 1957. 78, 4958. Rice, F. 0. J . Am. Chem. Soc. 1931, 53, 1959.

ley: New York, 1947; p 884.

1972, 11, 605.

~ l o l e s p r o d u c t

4 Ex,,erience n F-K l'i:ecry

h2 c ' q C2i6 C 2 c 3 b 6 C 3 b 6 C 4 1 8 f r ~ ' 6

Figure 10. Product in the cracking of dodecylbenzene. Experi- mental data vs. R-K theory (2' = 660 "C).

dodecylbenzene converted, the cracked liquids represent 98% of the formed products, while for high conversion rates, i.e., 78% initial hydrocarbon converted, they rep- resent 87% of the liquids formed. These liquids comprise the whole range of a-olefins, up to l-decene, with a-olefins with an end phenyl group from styrene to ll-phenyl-l- hendecene.

Four other aromatic hydrocarbons are also formed: benzene, toluene, xylene, and 1,2-diphenylethane. For the 20-30 % conversion rates of dodecylbenzene, l-hendecene, toluene, and styrene are found to form the main hydro- carbons of the cracked liquids. The cracking mechanism of dodecylbenzene may be explained as follows.

Toluene is formed together with l-hendecene in sig- nificant amounts. These two hydrocarbons may result from a nonradical cleavage of dodecylbenzene involving a four-member ring similar to a six-member ring in Miller's hypothesis (1963) for the cracking of the a-olefins 6 3 > C H ( C H 2 ) 7 C H 3 - C H 2 C H 2

6"' + C H 2 = C H ( C H 2 1 7 C H 3

This is more probable as the binding energy of the C, - C, carbons is very small as compared to the other bonds (67 kcal instead of 93 or 94 kcal)

The weakness of the C,-C, bond and its homolytical cleavage explains the great number of benzyl radicals 4-CH2, The high stability of these radicals due to the conjugation of an odd electron with the ring electrons still favors their concentration. The duplication of the benzyl radicals leads to 1,Zdiphenylethane.

Since the hydrogen and carbon a-bondings are weaker than the other C-H bonds (95 kcal instead of 94 kcal), the

Received for review April 29, 1980 Accepted October 15, 1980