1032
Vol. 39, No. 8
INDUSTRIAL AND ENGINEERING CHEMISTRY
pignient volunitrs may l>respectec! t o peifonii coupled n-itli stcvl and immersed in thc sea, a (al~ovc,23$ pigment-volume ratios) retain resistance valuv stic of extremely poor conductors nlien uncoupled a n Klien placed in t h e role of a cathode, an initial time c.c;uiied in ivhich the copper particlea u i i t cliange in their relative position t o each othei,, oitbLei breakdo1r-n occurs in tlie resistance of iritervciiiiig t); lators iornietl by envelopes of the matiis. or c o p i ) ~ ~j i solution ant1 redeposits according to a theory (4)p r t ~ p o . > tt>:irlirr. ~l In tht, cast’ of l o x pigmentation the distancc bctn-ccn the iiiqjority of pigment particles is so great that, the insul:iting p i i ) i of thc, organic matris are not disturlird by the intensity applicd pot t,iJtial, and no significant change o c ~ u r sin tlie ~ I < the ~ Q film. a n w ] ) V I ~ I L ~ ~ uf Therefore, it seems obvious t h a t a tl valiir for pigmentation with ic copper pigments somrn-here tJetneen 25 and 30 ent-volume for nixof the type listed here. Abo crititzal value nullificiition of antifouling properties of the paint ni n-lien t h r paint is coupled with steel. -1.3 a ri couple, a corresponding acceleration in the rate of corwsion 01 the steel should follow. T h e d a t a do not indii.:ite that similar behavior niay bc espected of cuprous oyide. The quantitative d a t a obtained from specific paint-steel couples [vas substantiated experimentally by tlie behavior of paints rsposed in the sea t o a n environment of high fouling intensity. Theoretically, films containing a high volume of metallic pigment and performing as a cathode should foul readily. This i ~ a st h e case among t h e samples exposed a t lfiaiiii 13each.
Similarly, i e s highly pigniente(1 foiniulxtioris (vitii metallic copper) [>eI’f~Jrlii equally \ d i \rhcttlicir applied to stwI or wood. Finally, cuprous oside paints are apparently etlually effective whether used over wood or metal, irrespctctive of pigment concentration. t c i i \ ow LE u(; \I E \ -r
,l.;ton anti Scott Ewing of n ~ s,iggestious i in the course
LITE:R.\TLRF: C I T E D
1) LaQ1.ie. l:. L., I n t e r i i a t i o i i a l S i c k e l C o . Kept. on S p e c i m e n s Reinox-ed f r o m K n r e B e a c h . \-. C. (AIay 1941). ( 2 ) Koa?, H o l e Oceanographic I n s t i t u t i o t i ’ s h I o n t h l > ~Rcpt. t o Bur. of S h i p s (Sept. 1, 1914.) ‘3) ITourtg. 12.H., and Schneider, W.K , , ISO.ESG. CHEM.,35, 438 (1943). (4) l-ouiig. G . €I., S e a g r e n , G . IT.,and Z e h n e r , *J. C.. Ihid., 37, 461 (1915). ’
PRESENTED before t h e Di\.i\ion of Paint. Vnrni.sh, a n d Plastics Chemistry at the 110th LIeeting of t h e .-\MERICAN CHEMICAL SOCIETY.Chicago, Ill.
Catalytic Cracking ot Pure Hydrocarbons T
(:RACKI;NG OF STRUCTURAL ISOMERS
HE catalytic cracking of fiitp-siu pure hydrocarbons a n d a comparison of hydrocarbon classes have been reported in the first five papers of this series (1, b, 3 , 4 6 1 . No detailed study was G. 31. m a d r of the influence of structural isomerism upon the extent of cracking and product composition. However, a large difference was observed in the extent of cracking of normal and isopropylbenzenes and but a small difference in the case of n-dodecane and highly branched isododecane. These data and other observations made previously led us to further experimeuts and study designed to show the effect of structural isomerism on the cracking behavior of certain hydrocarbons. This n-ork esaniines and comp a r e the behavior of (a) the five isomeric hesanes, ( b ) cyclohexane and methylcyclopentane, (c) n-octane and 2,2,4-trimcthylpentane, ( d ) a-propylbenzene and isopropylbenzenc,. ( e ) three isomeric butylbenzenes reported earlier ( d ) , (f)n-dodecyrir and isododecane reported in the first paper (Ij, and ( g Decalin and 2,7-dimethyloctane, a naphthene and paraffin of 10 rarhon atonis each. ~
~~
EXPERISIENTA L PROCEDURE
Ikfinitions and terminology are the same as given i n the first paper ( 1 ) except for the “percentage dccomposcil” o r “estent of cracking” which now includes the hydrogen in thc coke, as \vel1 as carbon, gas, and liquid boiling below the origiuiil, all sumnietf on a no-loss weight basis. EIydrogm and c u b o n \ v e x determined by the hurning of the catalyst deposit iii a n osygcn-nitrogen atmogphere, conversion of carbon monoxidc fornicd to carbon dioxide over copper oside, and absorption anti weighing of tht, i different lot of U.O.P. Rater anti carbon dioxide produced. ; cracking catalyst, type B, of slightly higher activity was used. This catalyst gives results similar to those obtained with the synthetic silica-alumina catalysts currently employed on a la.rgP scale in the petroleum industry.
GOOD, H. H. YOGE, ATD B. S. GREESSFELDER Shell Development C o m p a n y , Emerycille, Calif.
1Iodifications in the original procedure (1) follow: Reactiou products from the vertical catalyst, tube were led directly t o a still kettle cooled by solid carbon dioside. Cncondensed gases passed through a meter to a 255-liter holder containing saturated magnesium sulfate brine. -kt the end of each process period the system was purged directly with 2.8 liters of nitrogen into t h e gas holder. For the experiments with Decalin and 2,7-dimethyloctane, 57 cc. of catalyst \Yere used in a type 302 steel tube with a n inside diameter of 1.58 a n . , a. described previously ( 2 ) . Hexanes and octanes were cracked in a steel tube of 2.66 cm. i.d., with ‘200 and 90 cc. of catalyst, respectively. Xethylcyclopentane and cyclohexane were cracked in a steel tube of -1.08 cni. i.d., with 200 cc. of catalyst. In the larger tubes silira chips
Crachitig of six sets of isomeric hydrocarbons ober a silica-zirconia-alumina catalyst was studied, with e m phasis on the behabior of the hexanes. Some large differences in rate of cracking arid product composition were observed within these sets, which can be correlated with the carbon atom groupings of the isomers. In particular, tertiary carbons enhance the crackability markedly, and quaternary carbons act in the opposite way. Comparison of a paraffin and naphthene of selected C,, structure pro5 ideb added information on cracking behavior.
August 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
1033
0.6603. n Y 1.3732, bromine number ''3' CATALYTIC CRACKING 2,2-Dimer,hylbutane f r o m the isomerization of methyl(Temperature, 550' C.: process period, 60 minutes: LHSV. 0.5. Liquid product is Ca and above, b u t material pentanes !,.ith alumiIluni &loabove Ca was less t h a n about 170.) ?-Methyl?,3-Dimethyl2,2--Dimethylride catalyst had boiling n-Hexane pentane" 3-Methylpentane butane butane range of 18.5-48.9 C., d:' Feed Product Feed Product Feed Producib' Feed Product' Feed Product 0,6542, n'," 1.3710, bromine number 0.3. Hexane, wi. B n-Hexane 91.2 93.7 ... 0 0.4 1.6 0.2 1 0.0 0.0 n-Octane from Eastman had 2-Methyla boiling range of 124-126.2' pentane 3.4 2.2 90 90-95 13.4 12.8 4.8 8 1.3 2.5 C., dzo 0.7025, ny 1.3078, 3-Methylpentane 2.6 2.7 10 5-10 84.8 83.4 1.3 7 2.0 0.0 bromine number 0.16. 2.3-Dimethyl2,2,4-Trirnethylpentane (isobutane 1.2 1.0 ... 1 1.4 2.2 92.2 83 o.3 0.5(1c) octane) from Shell Oil had a 2,Z-Diinethylbutane 1.6 0.4 ... 0 0.0 0.0 1.5 1 96.4 97.0(93e) boiling range of !)8.6-99.4" c., Bromine number 0.2 4.2 0.3 6.6 0.2 .,. 0.2 9.3 0.3 5.2 d:' 0.6935, ngO 1 . 3 9 2 5 , bromine number 0.04. a Estimated values. Properties and cracking bcb From the run givin- 27.1% conversion. T h e analysis of the product from the run a t 23.7% conversion was also close t o t h a t of t h e feed and bromine number was 1.0. havior of 1%- and isopropylEstimates for t h e r u n k t 9.9% conversion. T h e detailed analysis is from a n unreported similar experiment benzenes a,rld the !,utylbent h a t gave a poor material balance. z e n e s h a v e alreadv been reported (4). Decalin from Eastman, after treatment with silica gel, had d:' 0.8837, n'," 1.4755, and bromine number 0.4. were used to hold the catalyst in place; type 302 steel fillers served this purpose in the 1.58 em. i.d. tube. The electric heater was 1 6 ~ ~ ~ $ $ #% ~ ~ ~ n Of 15'model 1132045 of the Hevi-Duty Electric Company, 62 cm. long Properties and cracking behavior of n-&decane and isododecand 7.6 em. i. d., with four independent 220-volt units totaling arie have been reported (I). Detailed infrared spectrophotometric analyses for four of the hexanes are given in Table I. 5.4 kiloxvatts. These were automatically controlled by a Gelectray (C. J. Tagliabue Manufacturing Company), guided by four iron-constantan thermocouples incased in a stainless steel CRACKING BEHAVIOR sheath which was set in a longitudinal groove in the hfeehanite heat distribution block surrounding the catalyst tube. ill1 The previous papers cited demonst,rated that, for the hgexperiments were made a t atmospheric pressure. drocarbons tested, the extent of cracking of paraffins and naphGltraViolet absorption was used for analyzing c6 to CS arothenes depended primarily upon molecular weight. In conmatics and a n infrared spectrophotometer for the hexanes. Liquid trast, structural isomerism was shown to influence markedly the products Rere distilled in precision columns, usually of about ten extent of cracking of olefins and aromatics, HoFvever, the partheoretical plates, a t 10 to 20 reflux ratio. Cut points were: affinscompared were normal and highly branched structures only, c6,42-74' c . ; c7, 74-99" C.; CS,99-125°C. Olefincontentswere and it was stated ( 1 ) that subsequent data on other paraffins computed from bromine numbers. would reveal a considerable effect of structural isomerism on the Properties and sources of hydrocarbons follow, with compounds cracking of this class of hydrocarbons. Therefore it was decided arranged in order of increasing branching in sets Of increasing t o examine in detail the behavior of the five isomeric hexanes, molecular m i g h t : which offer all the basic carbon-to-carbon groupings of paraffins, reasonable susceptibility to cracking, and much simpler analyCJ'clohexane (Eastman Iiodak Company) was washed wit'h tical and interpretive problems than do systems of higher molecconcentrated sulfuric acid and distilled t o a boiling range of 80ular weight. The eleven naphthenes a1read.r studied (.S) rep81" C., dZ,' 0.7826,' % : 1.4262, bromine number 0.5. resent a \Tide variety of st,ructures and demonstrate the preMethylcyclopentane from distillation of a California petroleum fraction had a boiling range of 71.8-71.9' C., d:' 0.7477, n 2 1.4095, bromine number 0.2. TABLE 11. CATALYTIC CRACKIXG O F HEXAXES n-Hexane (Shell Oil Com2-3Iethyl2,3-Dimethyl2,2-Dimethylpany, Inc.) had a boiling range Hydrocarbon n-Hexane pentane 3-Methylpentane butane butane of 68.0-69.0' C., di' 0.6613, n Y 1.3766, bromine number 0.2. Flow rate, moles/l./hr. 2-1\letliylpentane from hyProcess period, min. 60 60 60 drogenation of 2-methylpenGaseous product tenes had a boiling range of 0.249 Moles/mole charge 0.296 0.492 0.620 0.510 0.638 59.6-66.0' C., d:' 0.6603, n"," Volume % 7.7 10.1 1.3730, bromine number 0.3. 9.4 9.1 8.8 7.8 Hz 12.0 36.7 12.5 6.7 16.0 9.6 CHI A n a l y s i s s h o w e d lOYc 36.5 12.0 7.3 16.3 12.7 CiHi 7.9 methylpentanc (Table I). 4.6 0.8 3.8 7.5 CzHa 5.6 6.8 3-Methylpentane was a dis41.2 18.8 31.6 32.2 40.7 C~HE 40.3 3.9 1 6 . 6 1 2 . 9 1 9 . 4 CaHa 1 5 . 1 1 6 . 9 tillation fraction of hydro0.0 0.0 0.0 C4Hs 0.4 0.0 0.0 genated, catalytically cracked 2.3 1.4 ISO-CIHB 1.3 1.2 0.7 i.7 2.6 gasoline from Standard Oil 2.0 1.0 2.2 2.5 TI-CiHa 2.1 3.2 7.0 6.6 3.7 1.90-CaHio 6.5 3.1 C o m p a n y of N e w Jersey 1.0 0.3 1.2 1.5 n-CaH:o 0.7 1.5 (Louisiana Division). .ifter 0.6 0.6 0.8 0.5 0.4 0.5 CsHio treatment with silica gel it 4.9 2.9 1.0 C .~ H..T . 0.4 1.4 0.5 had d:' 0.6619, n'," 1.3765, Material balance, wt. % of charge bromine number 0.2, and anal25.6 24.2 30.8 9.3 Gas (including Cs) 13.0 24.3 ysis (Table I) indicated the 67.4 87.6 79.5 69.6 Remaining product 82.1 73.8 presence of 13.47, 2-methyl0.5 0.3 0.5 Coke 0.3 0.1 0.2 1.3 2.8 4.5 -4.2 Loss 4.8 1.7 pentane. 2 7 . 1 2 3 . 7 3 1 . 7 9 .9 % decomposed, no-loss basis 24.9a 1 3 . 8 2,3-Dimethylbutane f r o m the alkylation of isobutane a T h e shorter process period implies somewhat severer cracking, but not enough difference t o affect the relative results materially, with ethylene had a boiling range of 56.5-57.3" C., d;' O F HEX.4NE FEED STOCKS .4ND TABLE I. ISFR.4RED AXALYSES
OF
LIQUIDPRODUCTS
FROM
THEIR
~~.~o~~~~~i,~~~~~~I;~
~
stock. Assuming only 2,3dimethylbutane cracked, thtt concentration of original 2and 3-met~hylpentanes in the 67.4% product, would be 9% at most; the remaining 6% would require isomerization of 4 % of the feedstock, a minimum figure. On the basis ot’ equal cracking of all components, 6% (by weight of charge) isomerization is indicated. It is concluded that direct isomerization of hexanes is of small importance. T h e amounts of olefins in the liquid products were low, ranging from 2 to 5 weight per cent. Traces of benzene and boluene viere found in some of the runs.
~~~
TABLE 111. CATALYTIC CRACKING O F N.4PHTHEUES A N D Hydrocarbon Experimental conditions Temperature, C. LHSF’ Flow rate, moles/l./hr. Process period, min. Gaseous roduct Molesymole charge Volume % Hz CHI CzHa CzHs C3H6 C~HR C4H6 ISO-C~H~ n-CaHs Iso-CPHIn n-CIILo Material balance, w t . % of charge Gas Liquid below ired range Feed range Liquid above feed range Hydrogen in coke Carbon in coke
IL
Cyclohexane
Methylcyclopentane
Octane
Isooctane
550 0.50 4.6 60
550 0.50 4.5 60
550 0.61 3 7 90
550 0 60 3.6 30
0.350
Boiling range of product considered as feed, ’ C. % decomposed, no-loss basis b
Vol. 39, No. 8
INDUSTRIAL AND ENGINEERING CHEMISTRY
1034
0.582
n-
1.216
0.967
48.7 10.9 6.8 2.5 10.5 13.5 0.0 0.2 1.1 4.5 1.3
30.0 16.2 7.5 2.5 16.6 14.1 0.0 1.0 1.7 8.6 1.8
12.3 10 1 9.8 6.1 19.2 15.2 0.0 4.5 6.6 10.5 5.7
11.5 19.9 3.8 1.: 19., 6.1 0.2 6.8 7.5 20.4 2.3
8.3 10.6 63.8‘ 12.1 0.1 1.9 3.2 79-85
18.3 8.1 54.5, 15.8 0.1 1.9 1.3 69-73
30 :3 7.8
38.4 6.4
56.9 0.2 2.4 2.4 >120
50 3 0.2 3.0 1. I >96
21.6
28.8
41 7
48.8
Saturates in gas. Total olefins above Cz Expluding bmzene.
dominant influenre of molecular weight over t h a t of structure. Further study of this hydrocarbon class was directed primarily at a more exact determination of some small differences previously noted and their relation t o structural isomerism. HEXANES.T h e five isomeric hexanes ivere cracked a t 550 C. arid 0.5 LHSV (liquid hourly space velocity) with the results shown in Table 11. T h e effect of structure is clearly demonstrated by the following approximate rstents of rracking, in pcr cent b y weight:
c
c; 152
CYCLOHEXANE ASD METHYL-
These two naphthenes were cracked previously (3) at 500” C. and 1.5 LHSV, and decomposed 6 and 9 weight per cent, respecttively, on a no-loss bahis. T o magnify the cracking, t h w t tests were repeated a t hc500C!. and 0.5 LHSV with the .results shoTvn in Table 111. T h e perc.entages decomposed to lower boiling materials anti coke nerf. 22 and 29, respcctively, verifying the greater susceptibility of methylcyclopentane, which contains a trrtiary carbon atom. The chief product of molecular fission in both rases was propanepropylene. The d a t a in Table 111, and the liquid product yields given in Table IT, show that there was also c-onsidwahlc formation of higher boiling material and some isornc~rizatic~n.The percentage diwnipo*ed, as defined, does not fully portray the reactivity of these compounds; thus, isomerized material arid benzene are’ counted as part of the “decomposed” in the ease of cyclohexane but not in the case of methylcyclopentane, because of the boiling point rillations. If we include the higher boiling materials and the benzent, formed, but not the isomerized Cg naphthrnes, tht: percentages reacted on a no-loss basis are 26.0 for cyclohexane and 42.8 for methylcyclopentane. These figures are derived from thtt following breakdown of products, given in weight per w n t on a CYCLOPESTANE.
.i2 4
46 6
Cyclohexnnt.
hiethylcyclopentane
10.3 8.1 65 9 1.1 12.5 2.1 100.0
26.8 2 . 0 (estd.1 55.2 0.1 13.9 2.0 ~. 100 n
Gas and lower aliphatiis Isonieriaed naphthene Unchanged naphthene Benzene Remaining higher lmilina Coke ~
T h e modified calculation again shows the methylcyclopentane t o he more reactive. The appreciable formation of higher boiling products from the naphthenci is in contrast to the very slight formation of such materials from the paraffins. For the sake of consistency and because of the practical interest in lon-er boiling products, we shall continue t o use the percentage decomposed as defined above in discussing the reactivities of various hydrocarbons, although bearing in mind the additional reactivity of naphthcnes which leads t o higher boiling products. Refractive indices show t h a t these higher boiling materials, especially above 150”C., are rich in aromatics. TL-OCTANE AND 2,2,4TRI~fETHYLPEsTaNE. I n the first paper ( I ) n-heptane and 2,2,4trimethylpentane (iso-octane) provided
August 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
TABLEI\'.
YIELDS OF LIQUID PRODCCTS mnli CATALYTIC CRACKISG O F CYCLOHEX.4NE AND AIETHYLCYCLOPEXTASE
(Temperature, 550' C.; LHSV, 0 . 5 ; process period, 60 minutes) Hqdrocarbon Cyclohexane hlethyicyclopentane Product, w t . Yo of charge 0.2 0.7 Cs 1.4 Cs aliphaticsa 7.4 0.0 7.9 Lower boiling Ca naphthenes Benzene 0.1 1.1 63.8a 54.5 Peed boiling range 6.5 8.l h Above feed t o 150' C. 150-254O C. 4.6 7.2 Above 254O C. 1.0 0 4 Olefin content, wt. YG 14 18 Below feed boiling range 4 3 Feed range and above 'I
h
Total 42-69' C. product Excluding henzene.
(lata for tlir comparison of structural isomers, sincr a small interpolaPion along the curve for normal parafins indicated that n-octane should crack to the extent of ahout 5% while iso-octane was 8.5% decomposed. In view of the great difference in structure, the similarity in cracking rates was cwnsidered noteworthy. The present tests under somewhat more severe conditions, 550 O C. and 0.6 LHSV, provide a better comparison and serve t o confirm the earlier results. iis Table I11 shows, the weight percentages decomposed of n- and iso-octane rwre about 42 and 49, respectively. Compared with the two- to threefold differences in rates observed among t,he isomeric hexanes, n- and iso-octane show similar susceptibilities to cracking. I n the hexane series tertiary carbon groupings favored cracking and quaternary groupings did the reverse. Iso-octane contains one of each type, and it is iwiclutird that they act in compensating fashion t o yield the observed rrsults. T h e gas analyses show that iso-octane produces larger quantities of methane and C, hydrocarbons than n-octane, which may be associated with the particular structure of iso-octane. I)EC.4LIS . A S D 2,7-DIMETHYLOCTANE. I n a Comparison Of hydrocarbon classes ( 4 ) , it was concluded that naphthenes crack c.onsiderably faster than normal paraffins. -1pproximate rate ratios of 2 to 3 were assigned for the CIOto CISrange of compounds. T h e explanation of this behavior m a p be sought both in the intrinsic properties of a cyclic structure and in the types of carbon groupings coninion to both aliphatic and alicyclic hydrocarbons. T o explore this question, Decalin (decahydronaphthalene) and 2,7-dimethyloctarie, each containing two tertiary carbon atoms, were cracked a t 500" C. and about 13.4 moles per liter per hour, with the results shown in Table 111. The extents of cracking were 52.4 and 46.6 weight per cent, respectively, a similarity quite at variance with the rate ratios just cited. In contrast, n-decane has a predicted esterit of cracking of only 10% under these conditions (1, Figure 2). A comparison of the distribution of carbon groupings in thrse compounds follow (primary, secondary, tcrtiary, and q u a t i w a r y groupings are designated as P, S,T, and Q) : r d e c a n r , 2P, 8s; 2,7-diniethyloctane, 4P, 4S, 2 T ; Decalin, 8S, 2T. This comparison indicates t h a t the presence of tertiary groupiIigs has such an important influence upon the cracking rate of paraffins that certain isomers will crack as readily as naphthenes of the same carbon number. The near equality in rates of cracking does not mean t h a t the products are the same; they niay be radically different, as Table V shows. HightJr boiling products were formed from Decalin in appreciable amounts but %-ere not included in the percentage decomposrd. If they are included, the total percentage reacted becomes 63.3. .is noted under cyclohexane, this basis of coniputation is not furthw used in the present discussion. BL-TYLBESZESER. Sormal, secondary, and tertiary butylbenzenes were cracked ( 4 ) at 400" C. and about 2 LHSV to the extents of 13.9, 49.2, and 80.4 neight per cent, respectively. The c*hiefreaction was the removal of the entirr hutyl group from the
1035
aromatic ring. Here the greatest cracking again occurred with all the paraffinic carbon atoms in a tertiary grouping. However, the paraffinic carbon atom linked to the ring in tert-butylbenzene also niay he viewed as quaternary, since all four valences are occupied by carbon. Therefore, a n important categorical distinction must be observed in rlassifying the carbon groupings cncountered in aromatic compounds with respect to catalytic cracking, which requires that only the paraffinic carbon atomH tie counted in assigning the groupings. Thus, the butylbeneenes have the following distrihution.7 of carbon groupings: normal, 2P, 2 s ; secondary, 2P, 2 s ; tertiary, 3P, 1T. In addition, the grouping of the carbon atom attac.hed to the aromatic ring must be taken into account, which in this case is P, S,and T, respcctively, following standard nomenclature for substituted alkanes. The observed rates of (.racking then fall in thc same order as would be predicted from the results on other hydrocarbons prrwntcd here. PROPYLBESZEXES. Further confirmation of these relations is obtained from the rrsults of cracking normal and isopropylbenzenes a t 500' C. and 1.9 LHSV ( 4 ) ,whirh decomposed 43 and 84 weight per cent, respectively. The alkyl carbon atoms h a w the same grouping in both compounds-namely, 2P, 1s; however, the groupings attachrd to the ring are P and 8, respectively, which again demonstrates the large differences in cracking rat.? produced hy the t.ype of linkage in these strurt ural isomrrs;.
TABLE
v.
h A L Y S I S O F LIQUID PRODUCTS FROM CAT.4I.YTIC C R A C K I N G O F D E C A L I N A N D 2,7-DIMETHYLOCTA?E (Ternpprature ,500' C.: process period. 6 0 minutes)
Hydrocarbon CHYV Product, wt. Oic of vhargr 42-990 C. 99-152' C.
Decalin 2 1
2.7-Dinieth).lor,tanc 2 6
15.4 8 2
8 7
15 3 57.8
40.4
1 7
59 6 " 1.5 Olefins )iarafbn+ Saphthenes and )iaraffin+ Rpniene Benzene .~ ~~~~. Toluene Ethllbenzenc. Ethylbenzenc. u-Xylene m-Xylene p-Xylene ~~
2
9 8
27 4 n0 n5 .A1 0 :31.0
5 0
i . o0 7
1122 .99 6 4
88 0" 0 0 6 0 0 0 0.6 3 8 1 6
By difference
Il-I)ODEC.4SE AS1) I S O D O I > E C A s E . The catalytic cracking Of these materials was studied in somr detail ( 1 ) . At 500" C. the extents of rracking were 18 and 13% and at 550" C. were 34 and 32%, respectively. The somewhat larger losses of material in the runs a t 500" C. lfad us to plarta more reliance on the runs at 550' C. The isododrcane was obtained from the hydrogmation of tri-isohutene from the cold-acid polymerization of isobutcliir,. and is considered to he chiefly 2,2,4,6,6-pr~ntamethylhcptanc:. This structure contains 7P, 2S, lT, and 2Q carbon groupings, and the results indicate that compensating influences are at work xhich reduce its rate of cracking even below that of n-dod(wnt:. I t is also evident that a highly branched parafEn cracks no more readily than the normal isomer unless the carbon groupings arc' of a particular type. I n isododeranr the two quaternary groupiugs appear completely to offset the favorable influence of the tcrtiary grouping. It seems d i f k u l t at present to assign any g ~ ~ i i ~ r a l quantitative signifirancr to the relative numbcrs of primary and secondary groupings with respect to the rate of cracking. Conipounds containing large numbers of primary groupings trnd to produce more methane than the normal isomers, which indicatcs a definite effect upon product composition.
1036 TABLEVI.
I N D U S T R I A L A N D E N (3 I N E E R I N G C H E M I S T R Y SUMMARY OF
DATAON
STRUCTURAL ISOMERISM
Hydrocarbon Cyclohexane Methylcyclopentane n-Hexane 2-Nethylpentane 3-Methylpentane 2,3-Dimethylhutane 2,2-Dimethylhutane n-Octane Iso-octane n-Decane 2,?-Dimethyloctanr Decalin n-Dodecane Isododecane n-Propylbenzene Isoprop3lbenzenc
Carbon Exrenr of Groupings Cracking, W r . 70 6s 22 lP', 45, 1T 29 2P, 4 s 14 25 3P, 2S, 1T 25 3P, 2S, 1T 4P, , , 2T 32 10 4P. 1 s . .., 1Q 2P. 69 42 49 5P, 1 s . lT,1Q ZP, 8s 10" 4P, 4S, 2T 47 . ., 8s. 2T 52 2P. 1 0 s 34 iP, 25. l T , 2Q 32 l P , l P , IS 43 2P, 84
n-Bit5 lbenzene w-Butblbenzene lert-Rotx lbenzene
1P. E, 2 s 2P, 1% 1P 3P, I T
.
E
Covditi , n -
Temp., ' C .
LHSI
.I
550 550 500 550 400
0.5 0.6 1.9-2.6 3.1-3.2 1 9-2 1
A'
B C
D
'2
14 49 80
lloles/Liter/Hr~iir 3.4- 4 . 6 3.6- 3 . 7 13.3-13.8 13.6-14.1 12 3-13 2
Conditions A A
i:
A A A A; A B B B C
C B B
D D D Prnce-
Period. AIin 60 30 60 60 60
Estimated from Figure 2 of the firat paper ( 1 ) .
EFFECT OF STRUCTURAL ISOJlERISJl
A number of consistent relations between structural i w n i e r i m and extent of cracking have been demonstrated for several groups of isomeric hydrocarbons. It n-ould be useful if these findings could provide a basis for generalizations covering a wide range of molecular weights and types for a t least the paraffins and naphthenes. Aside from the alterations in relative rates of cracking of any two hydrocarbons introduced by change of esperimental conditions, the diagrams for the effect of molecular weight on the cracking of normal paraffins (1, Figure 2) and of naphthenes (3, Figure 4) indicate t h a t these relations are not simple functions and are different for the two hydrocarbon classes. Thus, additional correlation factors would be required for t,he computation of the relative extents of cracking of either txvo members of a homologous class or of any given paraffin-naphthene pair. T h e clearest point established in these experiments is the accelerating influence of a tertiary carbon atom in paraffins, naphthenes, and alkyl groups attached t o aromatic rings, upon the estent of cracking (Table VI). T h e effect of a quaternary grouping appears to be definite for 2,2-dimethylbutane, although some doubt arises from the circumstance t h a t the largest group attached to t'he quaternary carbon has only two carbon atoms. All prior work indicates that fragments smaller than three carbon atoms are removed with difficulty. In the cases of iso-octane and isododecane, this difficulty docs not arise, and i t appears t h a t the presence of a tertiary grouping offsets the effect of the quaternary. T h e concurrent changes in the number of primary and secondary carbons seem t o be of less importance. T h e peries d e c a n e , 2,7-dimethyloctane,and Decalin strikingly illustrates the influence of tertiary groups. Comparison of t h c two latter shows that the presence of much primary carbon in 2,7-dimethyloctane has little effect. T h e relations among the alkyl aromatics tested :we particularly regular. Here, the type of paraffinic carbon grouping attached to the benzene ring (underlined in Table VI) determines the extent of cracking, increasing in the order primary, secondary. tertiary. T h e contrast between primary and secondary paraffinic carbon in these compounds may be ascribed to the fact that cracking predominantly occurs a t only one linkage in the molecule, a t
Vol. 39, No. 8
which point the grouping of the attached paraffinic carbon is of prime importance. Among the petroleum naphthenes as a class, only cyclopentane and cyclohexane are lacking in tertiary groups (all higher homologs contain a t least one, except in the special cases of geminate substitution) ; linked and condensed ring naphthenes usually contain several. For this reason, it is difficult t o make an estimate of the influence of ring structure alone upon the cracking of naphthencs. T h e results on the hexanes and 2,7-diniethyloctane clearly show that certain paraffins may crack about as well a s naphthenes of the same carbon number. From a practical standpoint the preferred position of naphthenic cracking stocks in refinery operations (4)remains rather generally true because of the large proportion of normal isomers found in the paraffinic portion of most of the more abundant petroleums. The effect of structural isomerism upon the nature of the cracked products varies in importance according to circumstanccs. Honever, examination of the size and structure of a hydrocarbon in the light, of present knowledge can lead to useful predictions of cracking behavior. For example, certain structural units tend to persist in the products of catalytic cracking, particularly if they are bound in the original molecule by a linkage unusually susceptible to cracking. Thus, both normal and isopropylbenzenes yield largely benzene and propylene. Secondary and tertiary butylbenzenes show greatly different ratios of normal to obutene in the cracked gas, despite the possibility of isomerizaon. Similarly, methylcyclopentane and iso-octane show a higher ratio of iso- t,o normal butane than do their isomers. Decalin, with a hexahydroaromatic structure, produces far more aromatics than a paraffin of the same carbon number. Aromatic rings are very stable and, regardless of the isomerism occurring in attached groups, tend to preserve their original structures, and lose only the more easily removed substituents. A large number of methyl groups in highly branched structures encourages the production of methane, although 2,3-dimethylbutanc appears t o be an exception. Other examples of persistence of structural units may be found in prior papers of this series. Thus, di- and tri-isobutenes revert primarily to the monomer isobutene ( 3 ); n-octenes yield more normal than isobut,ene. However, some of these differences may be markedly reduced if severe cracking conditions are imposed, particularly urith respect' t o the rat,io of is0 to normal aliphatics which is especially susceptible to the temperature of cracking. The cracking of 3-methylpentane to much C1 and relatively little C, or C2 is of particular interest. Most of the C3 derived from 3-methplpentane is believed to arise during the primary cracking reaction and not from the subsequent cracking of larger fragments such as C4 or C5. Therefore, rearrangement of 3methylpentane during the cracking reaction must take place t o account for the observed Ct production. It is further indicated t h a t isomerization of 3-methylpentane independently of the cracking reaction is negligible, since the composition of the recovered hexanes is almost identical with that of the feedstock within the limits of analytical accuracy. ACKNOWLEDGMENT
The authors are indebted to Mrs. S. S. Tornbom for assistance in execution of experiments and calculation of results, and to other members of the staff of Shell Development Company for aid in preparation of materials and analysis of products. LITERATURE CITED
(1) Greensfelder, B.
S.,and Tope, H. H., IND. ENG.CHEM.,37, 5 1 4
(1945).
( 2 ) I bid., 37, 953 (1945). (3) Ibid., 37, 1038 (1945). (4) Greensfelder, B. S., Voge, H. H., and Good, G. M., Ibid., 37, 1168 (1945). (5)
Voge. H. H., Good, G. M., and Greensfelder, B. S., Ibid., 38, 1033 (1946).
