(7) Chesalova, V. S., Boreskov, G. K., Doklady Akad. Nauk S.S.S.R.

Table IV. Chemisorption of Carbon Monoxide on 85, 377 (1952).

Chromia-Alumina ( 8 ) Eischens, R. P.. Pliskin, W. A., “Advances in Catalysis,”
Vol. X, p. 19, Academic, New York, 1958.
CO Adsorption, (9) Eischens, R. P., Francis, S. A , : Pliskin, TV. A., J . Phys. Chem.
Pretreatment pMoles/Gram 60, 194 (1956).
Reduced 2 hours in hydrogen at 900” F. 46.6 (10) Eischens, R. P., 2. Elektrochem. 60, 782 (1956).
Oxidized 2 hours in air at 900” F. 2 0 (11) Emmett, P. H., Brunauer, S., J . Am. Chem. Soc. 59, 310
(1937).
( i i ) z b L i , p. 1553.
(13) Emmett, P. H., Skau, N., Ibid., 65, 1029 (1943).
(14) Farnsworth, H. E., TVoodcock, R . F., IND.ENG. CHEM.49,
I n contrast, a n oxidized chromia surface chemisorbs very little ~ 258 (1957).
CO. T h e authors are not yet able to interpret the CO adsorp- (15) Farnsworth, H. E., “Advances in Catalysis,” Vol. IX, p. 123,
Academic, New York, 1957.
tion in terms of absolute chromia surface, but plan to ac- (16) Gomer. R., Hulm, J. K., J . Chem. Phys. 27, 1363 (1957).
complish this by comparing nitrogen areas and CO adsorption (17) Gomer, R., Tl’ortman. R., Lundy, R., Ibzd., 26, 1147 (1957).
(18) Gruber, H. L., Ramser, J. H., Abstracts, p. 22-1, 13;th
values for pure chromia gel samples. T h e flow adsorption Meeting, ACS, Atlantic City. N. J., September 1959.
method is a useful tool for measurement of chromia surfaces 119) Gwathmev. A. T.. Record Chem. Proer. (Kresee-Hooker Sci.
U 1 Y

of butane dehydrogenation catalysts and other supported ‘ Lib.) 14, 117’(1953). ’

(20) Langmuir, I., J . Am. Chem. Sac. 40, 1361 (1918).
chromia catalyst$. I t has also been applied to other types of (21) Mignolet, J. C. P., Bull. soc. chim. Belges 67, 358 (1958).
supported transition metal oxide catalysts, including those ( 2 ~ ! n k ~ e r s o nL., H., Swearingen, L. E., J . Phys. Chem. 31, 88
( 1 Y LI ) .
containing molybdenum oxide and vanadium oxide.
(23) Spaadel, L., Boudart, M., Ibid., 64, 204 (1960).
(24) Srinivasan, V., Krishnaswami, K. R.. Current Sci. (India) 25,
literature Cited 328 (1956).
(25j-F$eisz,’P. B., Prater, C. D., “Advances in Catalysis,” Vol. 9,
(1) Anderson, R. B., Hall, 1%’. K., Hofer, L. J. E., J . Am. Chenz 575, .Academic, New York, 1957.
SOC.70, 2465 (1948). (26) LVortman, R., Gomer. R.. Lundy, R., J . Chem. Phys. 27,
(2) Benton, A. F., Zbzd.. 48, 1850 (1929). 1099 (1957).
(3) Boreskov, G. K., Karnaukhov, A. P., Zhur. Fzz. Khzm. 26, (27) Zwiebel. N.. Turkevich. J.. Miller. TV. TV., J . ilm. Chem. Sac.
1814 (1952). 1949).’
(4) Brunauer. S., Emmett, P. H., J . Am. Chem. Soc. 62, 1732 RECEIVED for review October 23, 1959
(1940) . RESUBMITTED January 2. 1962
(5) Burshtein, R. Kh . Frumkin, A. N., Trans. Faraday Sod. 28, ACCEPTED January 16. 1962
273 (1932).
(6) Calvin, M., Heidelberger. C., Reid, J. C., Tolbert, B. M., Division of Petroleum Chemistry. Symposium on Instrumental
Yankwich, P. F., “Isotopic Carbon,” pp. 163-4, Tl’iley, New Techniques in Study of Catalysis Mechanism, 135th Meeting,
York, 1949. ACS, Boston, Mass., April 1959.

COKE FORMATION IN CATALYTIC CRACKING

W . G. A P P L E B Y , l J . W. GIBSON, AND G. M . GOOD
Erneryville Research Center, The Shell Development Co., Emeryville, Calif.

Coke formation in the catalytic cracking process represents both a loss of product and a source of heat
through its combustion. It is important to know what causes coke as well as the final molecular structure it
assumes on the catalyst. Coke formation in catalytic cracking proceeds through intermediate aromatic struc-
tures of increasing size and complexity. The process appears to b e dependent on the molecular size of the
aromatic, but molecular structure is important, particularly in tricyclic and higher aromatics. The coke-
forming tendency of many aromatics and their alkyl derivatives correlates well with their basicity. The
mechanism of coke formation is discussed in terms of the formation of intermediate carbonium ions.

IN CATALYTIC CRACKING a significant portion of the feedstock This work has been devoted mainly to a study of the influence
is converted to a carbonaceous (coke) deposit on the catalyst. of hydrocarbon molecular structure and molecular Lveight on
It was necessary to characterize the nature of this process coke formation over several typical aluminosilicate catalysts.
because of its commercial importance and because the inter- T h e relative importance in coke formation of the various
action between hydrocarbons and a solid, acidic surface is of hydrocarbon structures present in cracking feedstocks is in-
fundamental interest in the field of heterogeneous catalysis. dicated by the data (3, 79) of Table I for the cracking of the
hydrocarbon-type components of fresh feed at constant con-
ditions.
T h e aromatic structures in typical feeds are mainly benzene,
1 Present address, Shell Chemiczl Co., Xew York, N. Y. naphthalene, phenanthrene. chrysene, and pyrene nuclei

102 I & E C PROCESS D E S I G N A N D D E V E L O P M E N T

 Nature of Coke Deposit
Table 1. Relative laboratory Conversion and Coke
Formation from Components of Fresh Feed T h e coke deposit on a spent cracking catalyst from the re-
Conoersion Coke actor is a heterogeneous mixture of a solid coke deposit left
Molecular weight 250 400 250 400 from prior regeneration; a growing solid or semiliquid mixture
Paraffins 1 0 1 2 1 0 1 4 of polynuclear aromatic molecules-e.g.! dimers and trimers
Naphthenes of naphthalene, phenanthrene, chrysene, etc.-and the more
Aromatics strongly adsorbed components of the reactor product. I n
the spent catalyst stripper, adsorbed product components are
partially volatilized, the extent depending on the efficiency
of stripping. T h e coke formation may occur ( 7 7 , 78) only
rying alkyl substituents and fused naphthene rings-for during the cracking process and involves condensation-hy-
example, alkyltetrahydronaphthalenes. On the cracking drogen elimination reactions in which the H/C ratio of the
catalyst, the side chains (>ethyl) and the naphthene rings are coke and its degree of unsaturation [as determined ( 7 7 ) by
removed, leaving relatively bare (only 1 to 3 methyl and ethyl iodine titration] decrease as the time spent by the catalyst
substituents) aromatic nuclei corresponding mainly to naph- particle in reaction increases-i.e., throughout the holding
thalene, phenanthrene, pyrene. chrysene, and higher. These time in the riser-reactor-stripper zones of the cracking unit.
smaller aromatic molecules thus become important participants To obtain some information on the structure of the actual
in the coke-forming reactions occurring in the reactor zone and coke deposit on the cracking catalyst, the coke deposits ob-
are the aromatic components of the recycle oil in the cracking- tained in the cracking of anthracene, phenanthrene, naph-
process. Aromatics composition of a typical recycle oil is thalene plus benzene, decahydronaphthalene: and quinoline
showm in Table 11. T h e monoaromatics listed are essentially over fresh silica-alumina catalyst, and of cetane over a par-
Tetralin and indane types. the diaromatics are mainly naph- tially deactivated silica-alumina catalyst were examined by
thalenes, the triaromatics are mainly phenanthrenes, and x-ray diffraction. The cokes examined contained about 2 to
those higher contain pyrene, chqsene, perylene, etc., nuclei, 8 weight 76 ash and were concentrated from as low as 1% coke
the predominant structure being pyrene in oils which have by weight on the catalyst by dissolving the catalyst with sodium
been analyzed in more detail. hydroxide. The x-ray data were interpreted as follows:
The large contribution of these materials to coke formation
is clearly shown by the high coke yield on catalytic cracking Small-angle scattering patterns were obtained, using filtered
copper radiation ; the coke samples gave visually similar small
of the recycle oil relative to that obtained after substantial angle patterns, except that from cetane; estimates of the average
hydrogenation of the heavier aromatics (dicyclic and higher). particle size indicate a radius for assumed spherical particles
All of the foregoing data and discussion, along with the work of less than 100 A., the value for cetane coke being smaller
of Gladrow and Kimberlin ( 6 ) on gas oil and pure hydrocarbon by a factor of about 0.75.
Wide-angle scattering patterns \vere also obtained photo-
cracking, lead to the conclusion that the aromatic components
graphically with filtered copper radiation and showed several
of the catalytic cracking feedstock are the outstanding source features: lines of a-alumina from the residual catalyst, bands
of coke in catalytic cracking. corresponding to the turbostratic graphite structure observed
Relatively high partial pressures of olefins-e.g., C3H6. in carbon black by Biscoe and Warren (Z), and a very diffuse
C4H8, and CeHlo---also exist in the reactor. These components scattering of the type observed from liquids. The ratio of the
diffuse intensity to the intensity in the turbostratic bands
are less likely to be adsorbed on the catalyst surface (at cracking appeared to be highest in the patterns of the quinoline and
temperatures around 450' to 500' C. and partial pressures of phenanthrene cokes, next highest for decahydronaphthalene
about 0.3 atm.) than aromatics such as phenanthrene, pyrene? and anthracene. and least for cetane coke. There was little
chrysene, and their heterocyclic analogs. Some of these difference among the samples in the general visual appearance
latter are actually near their boiling points and a t pressures of the turbostratic pattern. The results obtained are consistent
with the conclusion that these cokes contained, as a major
in the range of capillary condensation under cracking con- component, condensed systems of fused aromatic rings.
ditions. Since the olefins are reactive compounds, however:
they may contribute to coke formation by reacting with ad- These observations are consistent with those of Haldeman
sorbed species Lvhich are undergoing condensation and hy- and Botty ( 9 ) on coke deposited in the cracking of Mid-Con-
drogen elimination reactions. tinent gas oil. However, catalysts coked by cracking of n-
These laboratory studies have been devoted to the investiga- butene and phenanthrene in our work showed losses in surface
tion of coke formation on typical catalysts from a large number area of about 22 and 337,, respectively, for 2.8 and 10.4 weight
of aromatic hydrocarbons, several of their heterocyclic analogs, % coke. This indicates blocking off of pores and is contrary to
and several olefins. observations of Haldeman and Botty. Surface areas (74) of
recovered coke derived from n-butylene and phenanthrene lvere
<0.5 sq. meter per gram, indicating the possible agglomera-
tion of the coke particles in the recovery procedure.
Table II. Aromatics Composition and Coke Formation from The important results of this work are the similarity of the
a Recycle Oil before and after Hydrogenation patterns for cokes from different hydrocarbons and the ap-
Recycle Oil pearance of the turbostratic layer structure in coke which has
Hydrogenated been involved only in the cracking reaction; no partial re-
Hz added, cu. ft./bbl. 0 745 generation of these catalysts was carried out prior to isolation
Aromatics content, mmoles/100 g. and x-ray study of the coke. This suggests that, during crack-
Monoaromatics 27 102
Diaromatics 39 15 ing, condensation-hydrogen elimination reactions proceed
Triaromatics 73 8 to the extent that condensed aromatic nuclei of high molecular
Higher 23 5 weight are formed along with or as precursors of the turbostratic
Coke yield in catalytic cracking, wt.
7 0 of feed, a t 50% conversion 13 4 graphite structure. Thus, coke formation appears to involve
a: omatics a t all stages frcm the newly adsorbed materials

VOL. 1 NO. 2 APRIL 1 9 6 2 103

through intermediate ring systems to the polycyclic layer A low alumina content silica made by Shell Development Co.
structures which are aromatic precursors of graphite. (AI less than O.OOl%, C a 0.02%).
Definition and Determination of Coke. Coke was de-
Experimental fined as the carbonaceous material remaining on the catalyst
General Procedure and Apparatus. Essentially the same after hydrocarbon flow was stopped, the system allowed to
experimental procedure, involving a fixed bed bench scale drain for 1 minute. and the reactor either purged with nitrogen
reactor, was used as for previous studies of the catalytic crack- or evacuated to remove unreacted feed and volatile products.
ing of pure hydrocarbons (7, 70) with a few modifications.
Feedstocks solid a t room temperature were fed into the reactor T h e procedure used in the nitrogen purge involved (1)
from a heated container by gravity, augmented by nitrogen passage of 0.1 cu. foot of nitrogen, over a period of several
pressure. Flow was observed from burets, or by a rotameter minutes, through the hot reactor, thence through a COY
acetone-cooled condenser and product kettle (dry ice-cooled)
operated by displacement nitrogen. Distillation of products
into the product gas holder, (2) continuation of the S2 flow
boiling above room temperature derived from solid feeds was for at least 20 minutes thereafter to ensure adequate purging
limited to flashing to about the original boiling point; gen- but with diversion of the flow, following the reactor, to the
erally, little lower boiling material was found. Material regeneration system to minimize dilution of the products with
balances were 97% or better in all cases except where high N?. Any liquid material purged during (2) was trapped in an
ice-cooled receiver and added to the remainder of the liquid
melting solids were fed-e.g., chrysene. I n these cases inputs product. T h e small amount of gaseous product obtained
were not accurately measured and material balances are during 2 was measured (after burning in a furnace filled with
therefore not accurately known. Data are presented on a no- copper oxide a t 600' to 650' C.) as COZand HtO in absorbers
loss basis for comparative purposes unless indicated to be per containing soda lime mixed with caustic and Drierite, respec-
cent by weight of charge. tively. This gaseous product amounted to less than 0.1 weight
% of charge. Following the passage of nitrogen through the hot
Feedstocks. Materials were used as received. with p.xity reactor, the catalyst was brought from test temperature (usu-
in most cases of 95% or better. allv 500' C.) to 550' C., the temperature of regeneration.
Catalysts. Synthetic silica-zirconia-alumina (86.2,/9.4/4.3 In'the regeneration step (following t'he total purgey a mixture
wreight %) cracking catalyst, made by Universal Oil Products of air and nitrogen was used to avoid temperatures above
575' C. T h e regeneration gases were conducted through the
Co. (UOP Type B) with cracking characteristics similar to copper oxide furnace mentioned above co to
the synthetic SiO?-AlgOa catalysts in general commercial use. CO?.
Synthetic silica-alumina (90/10 weight %), made by Ameri- An evacuation procedure was also tested. After 0.1 cu.
can Cyanamid Co. I t had about 12070 of the activity, basis foot of r\Ts was passed through the reactor system, the pressure
was reduced to mm' Of Hg' The reactor system was
space velocity, of the UOP Type B catalyst for the cracking of then repressured with nitrogen and this procedure was re-
cetane at 500' c. and l-hour process period; product dis- peated several times. About the same amount of coke of
tributions were similar. approximately the same hydrogen content was found as by the

Feedstock Sources and Analyses

Compound Source
1,3-Butadiene Shell Development Co., 98.3% conjugated dienes
n-Butenes From 2-butanol. Stock A, 92.0% 2-, 8% I-CaHs; Stock B, 64.8% 1-,
33.0% 2-, 0 . 5 % i-C4Ha; 0 . 9 % i-, 0 . 8 % n-CdHlo
n-Butane Shell Oil Co., 98% n-, 0 . 5 % i-CdH1O
1-Pentene Phillips Petroleum Co. (technical), 95% dO : 0.6411
Benzene Stock A, General Chemical Co., 9 9 . 9 ( 5 ) % ; Stock B, J. T. Baker Co. (c.P.)
Decahydronaphthalene Eastman Kodak Co., ng, 1,4755, Br. No. 0.04, dO : 0.8838
Cetene Connecticut Hard Rubber Co., Br. No. 70.1, nZ,O 1 ,4415
Cetane E. I. du Pont de Nemours & Co., d i 0 0,7741, n*$ 1.4345, Br. No. 0 . 2
Melting Elementary Analysis, Wt. 70
Compound Source Point, ' C. Carbon Hydrogen Sulfur Nitrogen
Naphthalene
Stock A J. T. Baker Chemical Co. 79 93.4 5.3(6) <0.02
Stock B J. T . Baker Chemical Co. 79.9 ... ..
Methylnaphthalenes (41 wt. yo
50 wt. % 8) Reilly Tar and Chemical Co. ... 91.6 7.0 ... 0.2
Quinoline
Stock A Eastman Kodak Co. -17.7 83.1(4) 5.5(2) 10.4(9)
Stock B Eastman Kodak Co. ... 83.2 5.5 10.3
Acenaphthene Reilly Tar and Chemical Co. 93 93.4(7) 6.5(3) <o .02
Diphenyl Eastman Kodak Co. 70-71 93.59 6.60 ... <0.02
Fluorene Reilly Tar and Chemical Co. 113.1-113.4 93.67 6.1 0.02 <0.01
Carbazole Eastman Kodak Co. ... 86,29 5.53 0.01 8.1
Dibenzofuran Reilly Tar and Chemical Co. 83 85.6(8) 4.7(4) ... 0.13
Dibenzothiophene Eastman Kodak Co. 98.2-98.4 78.2(9) 4.4 17.4 0.01
Phenanthrene0 Reilly Tar and Chemical Co. 100-103 94.09 5.71 0.16 0.01
Anthracene Reilly Tar and Chemical Co. 93.47 5.60 0.12 0.60
Acridine Eastman Kodak Co. iojliio 87.1 5.12 0.39 7.5
9,lO-Dihydroanthracene Shell Development Co. 97-101 92,83 6.6 0.01 0.4
Dibenzyl Eastman Kodak Co. 51.2-52.1 92.3(0) 7.7(7) 0.02 0.02
Fluoranthene Reilly Tar and Chemical Co. 106.4-109.4 94.9 4.99 0.01 0.02
Pyrene Reilly Tar and Chemical Co. 146.8-147 94.53 5.0 0.14 0.02
Chrysene Paragon Testing Labs. 254-264 94,09 5.31 0.3 0.08
Terphenyls Monsanto Chemical Co. 131-1 50 93.9 6.09 0.07 0.02
a Contains 4.8% anthracene.

104 l & E C PROCESS D E S I G N A N D D E V E L O P M E N T

 Table 111. Coke Formotion from C, Aliphatics
Pressure : Atmospheric
UOP-B n
n-Bu- n-Bu-
tenes, 1,3- Buta- tenes, v
Catalyst Feed n- Butane Stock A diene Stock B tCARB4ZOLEl 5.5

Process period, min. 35 30 60 15

Temp., O C. 500 400 400 500 TERPHESYL. 8 . 2 PHENAVTHRESE. 6.8

Flow rate, moles/l./hr. 8.3 8.3 7.3 6.8 " , R C I I O F L R A ~ I4.8

Product, wt. %
Lower boiling plus carbon 0.5 10.5 21.2 38.7
Higher boiling 0.0 20.3 37.0 21.6
Carbon, wt. % :,,tiI\LOTHIOPHLYE.
a , I C \IIYITI.:
3 2
PROCESS PER103
ri
17. IC-DIH,URO~,;IR*CE\L1" '3 L
C K R Y I E ~ E . 15 6 PYRESE. :.) 2
Feed 0 6.1 14.6 14.8
Lower boiling products Figure 1. Coke formation from polycyclic and heterocyclic
plus carbon 0 58 69 38
Catalys 0 1.7 7.2 1.8 aromatics over fresh silica-alumina catalyst
Temperature. 500' C.
Process period. 15 minutes
Pressure. Atmospheric
Flaw rates. Ca. 10 to 13 moles/liter/hour, except for compounds named
in parentheses
Numbers indicate coke as weight per cent of catalyst
above procedure. The nitrogen purge alone was used for
most of the work.
Carbon and hydrogen values reported in the tables were
determined by combustion of the coke, unless otherwise noted.
products amounted to 21.6 weight % of charge. Aromatic
I n a few cases, carbon alone was determined. I t serves as a
production in the CSto C Srange was 3.6 weight '$&,consisting
good indication of total coke production, since the coke rarely
mainly of toluene and xylenes. The product boiling above
contained less than 92 weight 70carbon. Coke formation is re-
xylenes (5.6 weight yo)was almost entirely aromatics by refrac-
ported on the following alternative no-loss bases: weight per
tive index-e.g., fractions from 210" to 310' C. had refrac-
cent of feed; weight per cent of the converted (lower boiling
tive indices close to 1 .GO, indicating that dicyclic aromatics
than feed plus coke) products; and weight per cent of the
were formed. Coke formation was 14.8 weight % of charge
catalyst. At equal process periods, the last provides a uniform
containing 4.9 weight hydrogen as compared with 6.3 weight
basis for comparing coke yields from different feeds. Nitrogen,
oxygen, sulfur, and other elements in the coke were generally
% hydrogen in naphthalene and 14.3 weight % feed. The
hydrogen content of the n-butene coke was but little higher
not determined.
than that of most of the coke produced directly from poly-
Conversion and extent of cracking are the weight per cent
cyclic aromatics over silica-alumina (hydrogen content of
of products boiling below the feed plus coke on a no-loss basis.
coke 2.9 to 5.1 weight yo).
"Higher boiling materials" are those products boiling above
IVith butadiene a t 400' C. the liquid product was about
the feed range.
Separation of Coke from Catalyst. The catalyst was dis- 50% Cg material containing 36% olefins and about 507,
aromatics.
solved in hot concentrated NaOH, followed by filtration.
Thus, even in the reactions of such low molecular weight
T h e filter coke was washed with dilute acid and then with
materials as C J olefins and dienes, polymerization and de-
water. This procedure was repeated until the coke recovered
hydrocyclization reactions occur which lead to higher boiling
appeared to be homogeneous and upon drying approximated
aromatics and coke. Since the gaseous and liquid products
the amount expected. Three treatments were usually suffi-
also contain paraffins (a), it seems clear that a hydrogen trans-
cient to reduce ash contents to 3 to 12 weight %.
fer step is also involved in the aromatization-coke formation
sequence.
Experimental Results
Polycyclic Aromatics. T h e reactions of linked and con-
Since the physical structure and chemical composition of the densed (fused) ring unsubstituted polycyclics were studied in
cokes obtained by cracking aromatic, naphthenic, hetero- extension of previously reported work (79). Some unsub-
cyclic, and paraffinic hydrocarbons were similar, the reaction stituted heterocyclic aromatics have also been included be-
paths leading to coke might be expected to be similar. Data cause of their possible occurrence in petroleum. Data on these
on the formation of coke and higher boiling products from materials are presented in Table IV.
pure hydrocarbons and mixtures are consistent with this pos- Testing conditions were uniform except for the feed rates,
sibility. Experiments are presented illustrating the role of A range of flow rates varying only from 10 to 13 moles per liter
aromatics, heterocyclics, and olefins in determining the extent per hour was used, however, in 12 of the 17 cases. Both the
of coke formation and the nature of the intermediates in- condensed and linked ring polycyclics reacted principally to
volved in coke formation. coke, which amounted to 75 to 93% of the converted feed.
Olefins. Coke production increases as unsaturation in- T h e remainder was lower boiling material, generally too little
creases for a given aliphatic carbon skeleton. Thus, n-butane for complete analysis, and gas containing hydrogen as the major
was practically inert a t 500' C., whereas n-butenes and 1,3- component, usually accompanied by a much smaller amount
butadiene reacted extensively a t 400' C. to form coke (Table of methane. At a n a above the feedstock boiling point there was
111). To test whether small aliphatic monoolefins, a major little evidence ol any reaction product, based on elementary
product from cracking, would form appreciable amounts of analysis and physical properties. (An exception was 9,lO-
higher boiling components at commercial cracking tempera- dihydroanthracene. which readily dehydrogenated to the
tures. n-butenes were treated at 500' C. Higher boiling higher boiling anthracene.) It is noteworthy thar in 11 of

VOL. 1 NO. 2 APRIL 1962 105

 the cases in Table IV, the H/C ratio in thc coke !vas near to
or lower than the value for coronene (0.50), supporting clearly
the view of coke formation as involving condensation reactions
with hydrogen elimination in which increasingly large poly-
nuclear aromatic residues are formed, leading to the hexagon
layer lattice of the turbostratic coke observed in the x-ray
work discussed earlier.
The results in terms of coke on catalyst are set forth in Figure
1. At varying flow rates? coke formation may be compared
approximately on the basis of coke as per cent by weight of
catalyst. Although runs at the same process period and tem-
perature generally give more coke on catalyst \vith increasing
flow rates, the increase is usually rather small compared with
the decrease in coke expressed as per cent by iveighr of feed.
At flow rates above 9 moles per liter per hour (applying to 12
of the 17 tests in this series) the increase is generally very small.
Coke increases more rapidly in the condensed ring series,
benzene, naphthalene, anthracene (also phenanthrene), than
in the linked ring series, benzene, biphenyl, terphenyl. The
much greater coke formation from anthracene compared with
that from its isomer, phenanthrene, is most striking. The
tetracyclics, chrysene and pyrene, make more coke than phe-
nanthrene but still much less than anthracene.
Heterocyclic Compounds. Several heterocyclic compounds
k containing nitrogen, sulfur, and oxygen were cracked to com-
pare them with analogous hydrocarbons (Table I V and Figure
1). The chief product was coke. With the series fluorene,
carbazole, dibenzofuran, and dibenzothiophene, the hydro-
carbon prototype was the most reactive for formation of coke,
the others following in the order given.
20 % Coke formation from the nitrogen bases, quinoline and
n acridine, which are also notable cracking inhibitors (75, 79),
was 2.9 and 21.3 weight % of catalyst, respectively. However,
these two bases show little difference in their inhibition of
catalyst activity for cracking decahydronaphthalene. Naph-
thalene and anthracene, the corresponding hydrocarbon
analogs, formed 3.8 and 23.0 weight % coke on the catalyst,
respectively, at similar feed space rates.
Products obtained from the reaction of quinoline over silica-
alumina catalyst a t 500" C., shown in Tables V and VI, were
hydrogen, coke, and a material higher boiling than quinoline.
Sitrogen contents of the feed and the products, including
coke, indicate that nitrogen is retained in the ring.
Distillation of the quinoline product left a residue which,
considering the feed analysis, was in fairly good agreement in
elemental analysis (Table VII) with biquinoline. In con-
firmation, the ultraviolet spectrum of the residue resembled
the spectrum of 2,2 '-biquinoline ( 4 ) . Thus, heterocyclic
nitrogen compounds, like their hydrocarbon analogs, undergo
condensation and coupling reactions which are undoubtedly
steps in the process of coke formation.
1-PENTENE WITH METHYLNAPHTHALESES. Olefins react to
form aromatics; they may also condense with aromatics to
form more coke. Experiments were performed with an ole-
fin-aromatic mixture to obtain data on this point. As shown
in Table VIII, 1-pentene was cracked alone and in 20% mole
mixture with methylnaphthalenes.
Pure 1-pentene at 520" C. (reactor temperature rose from
500" C. because of the exothermic reactions) yielded 10.4
weight yo higher boiling material comprising alkylated ben-
zenes and higher fractions, of which 30% boiled in the range
of naphthalene and heavier polycyclic aromatics.
The products from methylnaphthalenes and their mixture
with 1-pentene in the range from about 200' to 250' C. both

106 I L E C PROCESS D E S I G N A N D D E V E L O P M E N T
Table V. Reaction of Quinoline over Fresh Silica-Alumina Table VI. Elementary Analyses of Quinoline and Products
Catalyst
Quinoline
Pressure Atmospheric Product Feed Theory
Temperature. ' C. 500 Elementary analysis, wt. yo
Process period. min. 60
Flow rate. moles 'I./hour 6.1 Carbon 84.2 82.7 83.72
Material balance, wt. % of feed Hydrogen 5.22a 5.5 5.43
Hydrogen 0.1 Nitrogen 10.60 10.3
__ ~10.85
Product above feed 4.9 Total 100.02 98.5 100.00
Feed recovered 88.2 a ,4 small quantity of water was obtained with the liquid product, SUB-
Coke 7.1 cient to account for the low hydrogen value.

Table VII. Analysis of Quinoline Product Boiling above

Feed Range
Elementary Pure
consisted mainly of naphthalene and presumably unconverted Analysis, W t . % Product Biquinoline
methylnaphthalenes; above this range, there were residues Carbon 84.5 84.4
of 4.8 and 6.0%, respectively, which were examined spectro- Hydrogen 5.0 4.7
Nitrogen 10 5 -10.9
scopically and found to differ considerably from alkylated ~

naphthalenes. A rather strong peak in the methylnaphthalene Total 100.0 100.0

residue is found also in anthracene. phenanthrene, 2!2-bi-
naphthyl? and 2-phenylnaphthalene. The ultraviolet spectra Table VIII. Coke Formation from Hydrocarbon Mixtures
for the higher boiling materials from the mixture of olefin Process period. 1 5 minutes
and aromatics and the pure feeds are similar, again suggesting Pressure. Atmospheric
that aromatics may be the common final step on the route to Catalyst. Fresh silica-alumina
catalyst coke. The coke on the catalyst from the mixture was .ifethyl- 1-Pentene
naph- plus Methyl-
greater than from either pure feed, indicating interaction be- Feed 7- Pentene thalenes naphthalenes
tween and olefin to enhance coke formation. Temperature, O C. 520 500 500
Products from Benzene-Naphthalene Mixture. It was Flow rate, moles/l./hr. 26.8 20 2 26.4
of interest to determine whether condensed aromatics are Olefin in feed, mole %
' 100 0 20 ( I 1
formed over the cracking catalyst via interaction of benzene wt. %)
Conversion, wt. % 23.9 10.9 ...
and naphthalene, which represent in nuclear type important Lower boiling, wt. % 18.2 7.9 ...
aromatic homologs present in many cracking feedstocks and Higher boiling. wt. % 10 4 4.8 6.0a
Coke, wt. % of
catalytically cracked products. Experimental data are given Feed 5.7 3.0 4.2
in Table IX. Elementary analyses were used to establish Converted products 16.6 19.1 24.8
the extent and type of condensation reactions and an analytical Catalyst 4.1 3.4 5.5
a Fraction boiling above methylnaphthalenes.
technique combining chromatography and ultraviolet absorp-
tion measurement was used to identify some of the products in
the fractions boiling above feed. Table IX. Reaction of a Benzene-Naphthalene Mixture
Products obtained were hydrogen: and small quantities of
Composition. 78 mole yo ben- Process period. 1 hour
gaseous hl-drocarbons. material boiling above the feed (mainly zene (Stock B) +
22 mole %
naphthalene (Stock B)
Pressure. Atmospheric
Flow rate. 6.0 moles/l./hr.
a residue above 492' C.)> and a similar quantity of coke on
the catalyst. Temperature. 500' C. Catalyst. Fresh silica-alumina
About 50% of the residue dissolved in benzene and was Distillate" Residue
chromatographed over alumina. Ultraviolet spectra of the Below 167- 261- above
761' 261' 492' 492'
fractions obtained showed the presence of many compounds Fraction Gas C. C. C. C. Coke
and nine were identified, amounting to 17 weight % of the Charge, wt. yo 0.01 69.52 23.79 0.18 2.26 2.63
sample. The six major compounds found are presented in Analysis
Table X. Carbon . . . . . . 93.7 9 3 . 4 95.2 85.2
Hydrogen . . . . . . 6.3 6.6 4.8 2.9
Fluoranthene, perylene, and benzofluoranthenes are pos- Ashb . . . . . . ... ... .., 11.9
sible direct coupling products of the feed components. The Loss on heating . . . . . . ... ... ... 1.5
other major compounds, 1,2-benzanthracene, chrysene, and a An effort was made to separate roughly the benzene, combined naphtha-

pyrene, are dehydrogenation products which cannot be formed lene, and methylnaphthalene fractions, and higher boiling distillate. The
temperatures listed are of the vapor in a Claisen Jask. The quantities of
directly from the reactants. These compounds may have material involved were too small to avoid some remixing in the take-off
originated by cracking and rearrangement of higher molecular manifold. * By combustion of coke remainiqy after -l;aOH dissolution of
catal,yst. e In nitrogen at 200' C.
weight aromatics originally formed by polymerization steps.
Benzene, naphthalene, or higher products cracked to gaseous
and lower hydrocarbons to a minor extent only and minor Table X. Some Polycyclic Aromatics Identified in Residual
amounts of condensed aromatics were found having carbon Oil from Benzene-Naphthalene Mixture
numbers other than simple sums or multiples of the feed Relative
hydrocarbon carbon numbers. Among these were fluorene Molar
Polycyclic Quantitiej
(CH), phenanthrene (C14),and anthracene ((214) found in the
Fluoran thene
distillate from 261' to 492' C. Analysis was not complete Perylene
because of lack of sufficient reference compounds. O n a Benzofluoranrhene
weight basis of the total recovered product, 3 =t1% benzene and 1,2-Benzanthracene
Chrysene
2 i- 1% naphthalene had disappeared from the original feed Pyrene
boiling range. ~ ~

VOL. 1 NO. 2 APRIL 1962 107

 T h e foregoing results strongly suggest that coke formation
Table XI. Effect of Alkyl Substitution on Coke Formation proceeds from simpler feed molecules (or fragments thereof) via
Temperature. 500' C. polymerization and dehydrogenation reactions, which yield
Process period. 15 minutes
Pressure. Atmospheric a wide range of polynuclear aromatic intermediates, to the
Feed rate. Ca. 2.4 molesjl./hr. larger, polynuclear, turbostratic graphite structures observed
Catalyst. Fresh silica-alumina in the x-ray examination of the coke.
Number Alkyl Aromatics. '4 number of alkylbenzenes and alkyl-
of
Coke, Weight 7 0
naphthalenes were tested to investigate the effect of alkyl
Carbon
Atoms Con- substituents on coke production from aromatics. T h e results
in .41kyl verttd are shown in Table XI. Reactivity to form coke appears to
Aromatic Group Feed product Catalyst be related to feed structure and molecular weight: T h e type
Benzene 0 0.6 63 0.04 of unsubstituted aromatic nucleus determined the minimum
Toluene 1 2.9 20 0.22
Ethylbenzene 2 8.1 14 0.74 reactivity in a homologous series; coke increased according to
p-Xylene 2 8.8 26 0.81 the total number of substituted alkyl carbon atoms rather than
n-Propylbenzene 3 9.8 11 1.1 to the number of alkyl substituents on the aromatic nucleus
Isopropylbenzene 3 9.6 10 1.o
Pentamethylbenzene 5 17.6 20 2.1 (triethylbenzenes and hexaethylbenzenes were notable ex-
Triethylbenzenes 6 11.3 12 1.7 ceptions) ; and coke deposition increased \vith boiling point
Hexaethy lbenzenes 12 7.4 8 I .6 or with molecular weight.
Naphthalene 0 23.4 89 2.6
2-Methylnaphthalene 1 27.1 51 3.5 Coke Formation via Noncatalytic Thermal Processes.
2,3-Dimethylnaphthalene 2 41 . 2 57 5.5 Coke (23 weight yo)> as obtained from anthracene on the fresh
silica-alumina catalyst, i s about sufficient to form a mono-
Table XII. Comparison of Catalytic and Thermal Cracking of layer. To correlate coke formation with the acidic nature of
Aromatics the catalyst, pure silica gel (a nonacidic material of comparable
Coke Gas Yield, specific surface) was tested with anthracene at 500' and 600" C.
as wt. Moles X 703per Only 1 and 2% coke on catalyst, respectively, were round,
% of ThermalMole Charge of
Cot., Cracking Thermal Residue indicating that the neutral surface alone (free radical-type
500' C., 500" Formalion. thermal cracking processes) was not responsible for extensive
75-M i n . C. Wt. % o/ Charge
Process 475O 90 c.
475' C., 500' C.,
coke production. However, since some coke was formed
Hydrocarbon Period 60 min. min. 60 min. 90 mtn. (presumably under purely thermal conditions), such processes
Anthracene 23.0 37 44 10 3 6 1" might be involved in initiating mechanisms in the catalytic
Acridine 21.3 83. 8b 56. 2h process.
Acenaphthene 18.7 128b 15.3h A further indication that coke formation is related to the
Chr ysene 16.6
Pyrene 14.2 catalytic nature-Le., acidity-of the silica-alumina catalyst
9,lO-Dihydro- is presented in Table XII, where the coke-forming tendencies
anthracene 13.2 129b
Fluoranthene 12.5 1.7 0.0 of a number of aromatics oker the silica-alumina cracking
Fluorene 8.7 133 0.2 catalyst are compared with their tendency to form gaseous and
Phenanthrene 6.8 3 0.0 heavier products in a purely thermal conversion as studied by
Carbazole 5.5 0 25 0.0 0.0
Naphthalene 3.8 2 0.0 Roberts and Madison (77). There is clearly no relation
Dibenzothiophene 3.2 0 10 0.0 0.0 between the catalytic and thermal processes. This sub-
Quinoline 2.9 4b 49 0.0 4.9 stantiates the results shown for anthracene. Thus the coke-
Biphenyl 1.9 1.5 0.0
forming reactions of aromatics are governed by the size and
a 20 min. 30 min.
structure of the hydrocarbon and the nature of the cracking
catalyst surface. Because of the basic nature of polycyclic
+IO I
aromatics, coke formation is probably accelerated by the
carbonium ion reactions that prevail in catalytic cracking
+e - systems.
2,4 M O L E S ~ L I T E R I H O L R 1 0 - 1 3 I I O L E : ~ LITL:II H m n

Discussion of Results

Among the various types of hydrocarbons involved, aro-

matics and olefins have the strongest tendencies to form coke in
catalytic cracking and olefins probably form coke via the
formation of intermediate aromatics. Therefore, coke forma-
tion on acidic catalysts is discussed primarily in terms of the
reactions of aromatics.
I n any series of related hydrocarbon structure, such as
benzene, naphthalene, anthracene or benzene, diphenyl,
terphenyl, there is a marked increase in coke formation with
increase in molecular weight. T h e effect of adding one more
ring seems, however, so large that other factors may be in-
volved. This is clearly shown by comparison of the data
(Table IV) for the various tri- and tetracyclic structures
studied, wherein structural factors outweigh molecular weight
in coke formation from tricyclic and larger aromatics. 'The
interaction of the aromatic and the catalyst surface is probably
strongly dependent on the aromatic structure and may involve

108 I & E C P R O C F S S DESIGN A N D DEVELOPMENT

a proton transfer to the aromatic, an electron transfer to the is peculiar because of the peri-condensed structure of pyrene.
catalyst, or both. Therefore, coke formation is expected to be Some notion of the charge distribution on such ions can be
related to the acid base properties of the catalyst and the obtained by methods (72) used in molecular orbital calcula-
reactant aromatics. T h e reactions of anthracene over the tions for ions of alternant hydrocarbons. I n the pyrene ion
acidic ( 7 ) silica-alumina catalyst and the neutral silica gel, the charge is evenly distributed and of lower intensity a t a
discussed above, show the importance of the nature of the given carbon atom than the charge on the carbon atoms of the
catalyst surface. The influence of the basicity of the aromatic other hydrocarbons of Figure 2. The pyrene ion \vould
in determining coke formation is shown by the curves of Figure thus be expected to react with other pyrene molecules at a
2. where log basicity constant is plotted against coke formation lower rate, leading, under the same conditions, to lower coke
as weight per cent of catalyst. The data on log basicity formation.
constant were determined by Mackor, Hofstra, and van der On the basis of these data and discussion we believe that
\Vaals (73) from experiments on the partition of the hydro- coke formation proceeds on acidic, heterogeneous catalysts,
carbons between H F and n-heptane or CC11. Their data such as the cracking catalyst, mainly via the reactions of ions
seem particularly pertinent for correlation with our data, generated from aromatic structures initially present in the
showing interaction of the same materials with anacidic solid, feedstock (or formed as intermediates) and the products of the
The correlation is satisfactory for all of the compounds various cracking reactions involved. Some insight into the
shown except hexaethylbenzene, pyrene, and chrysene. I n mechanism can be gained by a discussion of the products
terms of the basicity data, hexaethylbenzene and pyrene would formed during the reaction of a mixture of benzene and
be expected to interact (given equal access to the surface) naphthalene over fresh silica-alumina catalyst (see Tables IX
more extensively with surface protons or positive sites than and X) .
any of the other hydrocarbons shown except anthracene. Naphthalene would be expected to show somewhat stronger
Coke formation, however, involves more than this. Only adsorption on the catalyst surface than benzene because of its
\vhen the reaction proceeds further, through polymerization, higher molecular weight. Following or simultaneous with
condensation, and hydrogen elimination reactions, to form adsorption the naphthalene molecule will undergo protonation
large polynuclear aromatic aggregates does true coke forma- to form positive ionic species which carry on the coking reaction.
tion actually occur. Thus. hexaethylbenzene and pyrene Naphthalene protonation is assumed to occur most readily
apparently do not proceed as easily through the intermediate on the marked atoms (5, 73) of I , to produce a n ion in which
stages of the coke-forming process as do the other hydrocarbons the charge density
shown in Figure 2, particularly chrysene. Chrysene was * *
difficult to handle experimentally because of its high melting
point, so that the chrysene coke point in Figure 2 may be less
reliable than the others. The pyrene and hexaethylbenzene
03 * I

points are accurate, however, and experimental error could

is concentrated as marked on I1 (the original H atoms and the
probably not explain the chrysene discrepancy.
conventional bond structure are eliminated for simplicity),
The formation by chrysene of too much coke for its basicity
may be related to the possibility that chrysene is the largest
of the molecules listed in Table I V and Figure 2 and its planar.
close-packed tetramer or pentamer approaches the range of
dimensions observed in turbostratic crystallite formation.
QHQ I1

In other words, fewer molecules of chrysene are needed to so that addition to benzene would produce I11 and IV.
form coke and at any stage-dimer, trimer, etc.-the chrysene
intermediates are larger and should be more reactive than
the corresponding polynuclear intermediate stages of the
other hydrocarbons of Figure 2. With increasing size of
polynuclear hydrocarbons, there is a n increase in the
maximum values of their bond orders and free valences leading
to increased reactivity in substitution and bond addition I11
reactions (76). Thus, a larger aromatic reactant in catalytic
Ring closure in I11 would produce
cracking will form larger polyaromatic units on the catalyst
for a given extent of reaction, leading, therefore, to a greater
tendency to increase still further in size via condensation
reactions a t the peripheral points of higher reactivity.
While hexaethylbenzene may relatively easily add a proton
in solution, on a solid catalyst surface this may be more difficult
and the on-going coke-forming process from the ion may be
V
retarded because of steric effects arising from the high degree of
substitution. T h e six ethyl groups, as indicated clearly with Elimination of a proton from I V and V would yield, re-
a molecular model, may actually prevent a close enough spectively,
approach of the nucleus to the catalyst surface for a charge
transfer process to occur. Only after the removal of some of
the ethyl groups can the remaining portion of the molecule
participate in coke formation and the result is more characteris-
tic of less highly substituted benzenes.
Compared to positive ions formed by protonation of the
other molecules of Figure 2, the pyrene ion charge distribution

VOL 1 NO. 2 APRIL 1962 109

Dehydrogenation of VI and VI1 would lead to phenylnaph- The tendency to form coke increases with substitution of the
thalene and fluoranthene, respectively. The data of Table aromatic nucleus, although in cases of extensive substitution,
X indicate that formation of fluoranthene was a more prom- coke formation may actually be less than expected-e.g..
inent reaction; however, the presence of phenylnaphthalene hexaethylbenzene, in which the substitution may actually
could not be determined spectroscopically because of inter- decrease the adsorption of the aromatic on the catalyst surface.
fering substances and lack of reference spectra. The presence or formation of a peri-condensed structure
A similar series of reactions of naphthalene with itself will may also influence coke formation because ions from such
account for the formation of perylene and benzofluoranthene. structures may be less reactive. S) nergistic influences may
Ionic species such as those illustrated can continue to undergo be observed in coke formation from aromatics and olefins.
addition and dehydrogenation reactions to produce poly-
aromatic aggregates of increasing size into the coke range. Acknowledgment
Species such as VI and VI1 and their analogs of higher molecu-
lar weight can also undergo cracking reactions with formation The authors acknowledge many helpful and stimulating
of C I and higher fragments, which can participate in the coke- discussions of the work presented herein with B. S.Greensfelder.
forming process by acting as hydrogen acceptors in the de- H. H. Voge, R . M. Roberts, D. P. Stevenson, and E. R. Bell
of these laboratories. The authors also thank R . C . Archibald
hydrogenation of the growing coke deposit. Other fragments
and A. D. Reichle for the hydrogenation data in Table 11,
and small molecules can also perform this function. The
production of methane and higher gaseous products is shown in J. N. Wilson for the x-ray work, and D. D. Tunnicliff for
chromatography and ultraviolet absorption measurements on
Table IV. Products from the reaction of benzene and naphtha-
the products from the benzene-naphthalene experiment.
lene, such as chrysene and benzanthracene, may be associated
with the cracking and condensation reactions of higher molecu-
literature Cited
lar weight ions and polynuclear aggregates.
In terms of the argument presented earlier in explanation (1) Benesi, H. A., J . Phys. Chem. 61, 970 (1957).
(2) Biscoe, J., \Varren, B. E., J . Appl. Phys. 13, 364 (1942).
of the behavior of pyrene, the presence of fluoranthene and (3) Bittner, C. \V., Brewer, C. P., Good, G. M., unpublished
perylene in major concentration in the products from the data.
naphthalene-benzene experiment may be related to the peri- (4) Breckenridge, J. G., Can. J . Rrsearch 28, Section B, 593 (1950).
(5) Burkitt, F. H., Coulson, C. A., Longuet-Higgins, H. C.,
condensed structures of these materials which lead to ions of Trans. Faraday SOC.47, 553 (1951).
relatively low reactivity after protonation. Thus fluoranthene (6) Gladrow, J. M., Kimberlin, C. N., Jr., Division of Petroleum
Chemistry, Symposium on Catalysts and Catalytic Cracking.
and perylene can be relatively easily formed along with other 138th Meeting, ACS, New York, September 1960.
intermediates, but as ions they may react easily along the path (7) Good, G. M., Voge, H. H., Greensfelder, B. S., IND.ENG.
to coke. CHEM. 39, 1032 (1947).
(8) Greensfelder, B. S.. Voge, H. H., Ibid., 37, 983 (1945).
(9) Haldeman, R. G., Botty, M. C.. J . Phys. Chem. 63,489 (1959).
(10) Henrioues. H. J.. IND.ENG.CHEM.39. 1564 (1947).
Summary (11) KlimeAok,’ B. V:, Andreev, E. A,, Gordee;a. N’. A., Bull.
h a d . Scz. (C‘S.3.R.) Dto. Chem. Sa., No. 5 , 521-6 (1956).
Coke formation over acidic solids such as the silica-alumina (12) Longuet-Hiegins, H. C., J . Chem. Phys. 18, 265, 283 (1950).
catalyst in catalytic cracking occurs mainly via the reactions of (13) Mackor, E.-L.$ Hofstra. A , , van der \Yaals, J. H., Trans.
Faraday Sac. 54, 66, 186 (1958).
aromatic hydrocarbons, present as reactants or formed as (14) Meisenheimer, R. G.. unpublished BET N2 and 0 2 adsorption
intermediates or products in the process. The coke-forming data.
(15) Mills, G. A., Boedeker, E. R., Oblad, A. G., J . Am. Chem.
reactions appear to be governed by SOC.72, 1554 (1950).
Adsorption of the aromatics on the catalyst surface: which (16) Pullman, B., Proceedings of Third Conference on Carbon,
Buffalo, N. Y., 1957, Pergamon Press, New York. 1959.
may be dependent mainly on their molecular weight (17) Roberts. R. M.. Madison. J. J.. IKD.E s c . CHEM.50, 237
Reaction of the adsorbed species with the surface to form ‘ (i958). ’
aromatic or unsaturated ions: a process dependent mainly on (18) Thomas, C. L., J . A m . Chem. SOC.
66, 1586 (1944).
the basicity of the aromatic (19) Voge, H. H., Good, G. M., Greensfelder, B. S., Third World
Addition of the ions to other aromatics and unsaturates to Petroleum Congress, Sect. IV, The Hague: 1951, E. J. Brite,
form a growing polymer. a process dependent on the relative Leiden, 1951.
reactivities of the several ionic species RECEIVED for review July 19, 1961
ACCEPTEDDecember 15%1961
Dehydrogenation, cracking, and condensation reactions of
the polymer to form large aggregates of polynuclear aromatics Division of Petroleum Chemistry, Symposium on Catalysts and
which exhibit the turbostratic graphite structure of cokes from Catalytic Cracking, 138th Meeting, .4CS, New York, N. Y.,
other sources September 1960.

110 l & E C PROCESS D E S I G N A N D D E V E L O P M E N 1