Vous êtes sur la page 1sur 10

JOURNAL OF CATALYSIS 63, 285-294 (1980)

Exchange and Desulfurization of Thiophene over Synthetic and

Commercial CO/MO Catalyst Systems


Catalyst and Petroleum Process Research, National Iranian Oil Cornpuny, Rey, Tehrcrn. Irun

Received May 19, 1979: revised December 12, 1979

Thiophene was desulfurized and exchanged with deuterium over laboratory-synthesized Co/y-
A&O,, MO/~-AltO:,, and Co-MO/~-A&O, catalysts and over commercial Co-MO/~-A&O, catalysts.
Magnetic properties of the Co/y-A&O, catalysts were measured, desulfurization activities of all
catalysts were measured, and the deuterium exchange patterns of the desorbed thiophenes were
measured. For Co/y-A&O,, no clear correlation was found between magnetic properties and
desulfurization activity, but a crude correlation was found between percent d2 and desulfurization.
For molybdenum-containing catalysts, desulfurization activity was high and accompanied by
random exchange. Also d2 was high. For the commercial catalysts high d,, generally correlated
with high desulfurization activity. The results can be rationalized on the basis of a modified
Komarewsky mechanism.

INTRODUCTION cobalt-molybdenum-on-y-alumina cata-

lysts. During the course of these investiga-
Hydrogen-deuterium exchange has been tions we noticed unique exchange patterns
used by several investigators for mechanis- for thiophene desorbing from different co-
tic investigation. Kieren and Kemball (I) balt-molybdenum-on-y-alumina catalyst
first determined that molybdenum disulfide systems. Because unique exchange pat-
catalyzes the exchange of thiophene prefer- terns might be characteristic of catalyst
entially in the o-positions. Later we found activity, we examined exchange patterns
that cobalt-molybdenum on y-alumina cat- from a variety of catalysts. In this paper we
alyzes three kinds of exchange in thiophene report the results of these experiments.
(2). Two of them are random exchange and
multiple exchange caused by alumina, and EXPERIMENTAL
the third is a-exchange caused by molybde-
num. We correlated multiple exchange with Catalyst Preparation
hydrogenation/dehydrogenation activity Four series of catalysts, Co/y-Al,O,,
and correlated o-exchange with desulfur- MO/Y&O,, MO-CO/~-A&O,, and Co-
ization. These correlations supported Desi- MO/~-A&O,, were prepared for this work.
kan and Ambergs (3) theory of two differ- An alumina support (E. 5015) supplied by
ent catalytic sites on cobalt-molybdenum Pro Catalyse, France, was used for all
desulfurization catalysts. They also sup- preparations. Physical analysis of this alu-
ported the Komarewsky mechanism (4-9), mina revealed a surface area of 216 m/g
which involves C-S bond breakage as the and a total pore volume of 0.473 ml/g.
first step. However, they did not support The Co/y-A&O, series (Table 1, Catalyst
the Kolboe mechanism (IO), which in- Nos. l-10). The alumina support was im-
volves an intramolecular hydrogen shift pregnated with cobalt, using solutions of
and was supported by Heineman and co- cobalt(I1) acetate (analytical grade, Fluka
workers (II). Recently we extended our AG Chemische Fabrik, Switzerland) and
thiophene investigations to cobalt-on-y- cobalt(I1) nitrate (analytical grade, E.
alumina, molybdenum-on-y-alumina, and Merck Ag, W. Germany). The method was
Copyright @ 1980 by Academx Press, Inc.
All rights of reproductmn I any form reserved.

The Catalyst Samples Prepared using 25 g of the Alumina Support

Catalyst Source of Amount taken % Co adsorbed % MO adsorbed

No. Co(B) (g/50 ml I-&O)

(a) The Co/A1203 series

1 nitrate 0.50 0.42 0
2 nitrate 1.0 0.75 0
3 nitrate 2.15 1.22 0
4 nitrate 4.31 1.79 0
5 nitrate 5.99 2.30 0
6 nitrate 7.20 2.80 0
7 acetate 1.00 0.86 0
8 acetate 2.52 1.76 0
9 acetate 3.99 2.31 0
10 acetate 4.49 3.95 0

(b) The MO/A&O, series

11 acetate 2.61 0 7.80
12 acetate 4.21 0 9.0
13 acetate 7.36 0 12.93
14 acetate 9.20 0 14.33

(c) The MO-Co/Al,O,, series

15 nitrate 5.12 1.29 7.80
16 nitrate 5.12 1.29 9.0
17 nitrate 5.12 1.28 12.93
18 nitrate 5.12 1.29 14.33

(d) The Co-Mo/Al,O, series

19 nitrate 1.0 nitrate 0.75 12.1
5.5 molybdate
20 acetate 1.O acetate 0.86 12.74
5.5 molybdate
E, Cobalt Molybdate 0.20 0.74

similar to that employed by Tomlinson et series. The method of preparation was the
al. (12). In each impregnation the support same as in the Co/y-AlzOa series and anal-
(25 g) first was suspended in the impregnat- yses of the individual samples are presented
ing solution (50 ml) by stirring for 92 hr. in Table 1.
Then it was filtered under vacuum and The Mu-Co/y-A&O3 series (Table 1, Cat-
dried for 2 hr under flowing air at 110C. alyst Nos. 15-18). To prepare this series,
Finally the samples were calcined under a the MO/~-Al,O, catalysts were impreg-
flow of dry air (30 liters/hr) at 500C for nated with cobalt nitrate (see Table 1).
3 hr. The concentrations of solutions and The Co-Mu/y-A&O, series (Table 1, Cat-
the percentages of cobalt absorbed by the alyst Nos. 19-20). This series was prepared
support were determined by atomic ab by impregnating two of the Co/y-A&O,
sorption in each case and are shown in samples with ammonium heptamolybdate.
Table 1. A special experimental sample, designated
The Mo/y-A1203 series (Table 1, Catalyst E,, was prepared by impregnating the alu-
Nos. 11-14). Ammonium heptamolybdate mina support with a solution of cobalt mo-
(analytical grade, E. Merck AG., W. Ger- lybdate (analytical grade, K & K Laborato-
many) was used for the preparations of this ries, Inc., Calif.). Due to the very low

solubility of cobalt molybdate in water, very Surface area and pore size distribution
little molybdenum was deposited on the measurements. Surface areas and pore size
support even after five successive impreg- distribution were measured on the commer-
nations. cial catalysts and the alumina support prior
Nine commercial catalysts were used in to impregnation. These measurements were
this work. They are listed in Table 2. obtained with a Carlo Erba Sorptomatic
apparatus at liquid-nitrogen temperature in
Physical Measurements an atmosphere of pure nitrogen. In addition
Magnetic measurements. The magnetic the cumulative surface area and the pore
susceptibilities of catalyst samples l-10 volume of pores having radii less than 150
(Table 1) were determined at room temper- 8, were calculated by the method of Craus-
ature by the Faraday method. The determi- ton and Inkley (Z-5). The average pore ra-
nations were made on a Newport Instru- dius was calculated in the range of O-150 A.
ments Variable Temperature Gouy Balance All of these data are shown in Table 5.
System calibrated with cobalt mercury
thiocyanate (13). From their magnetic sus- Pulse Experiments
ceptibilities the magnetic moments of the The experimental apparatus has been de-
cobalt ions were calculated. Correcting for scribed previously (4). In a typical experi-
the diamagnetic contributions of alumina ment 0.2 g of catalyst (crushed and sieved
and Co2+ to the moments are as follows: to 60-80 mesh) was packed in the reactor
between two glass-wool plugs and heated
Diamagnetic susceptibility to 400C under flowing hydrogen (97% Hz,
(lOBerg Oe~*mole-) Ref. 2.5% methane, and 0.5% C,-C, hydrocar-
bons from the Tehran Refinery) overnight.
co+ -12 (15) Then 500 ~1 of thiophene (99% pure, Fluka
ALO:, -0.3 (18)
AG Chemische Fabrik, Switzerland) was
injected through the reactor to sulfide the
TABLE 2 catalyst. Next the system was adjusted to
The Commercial Catalysts reaction conditions, the carrier/reactant
gas was changed to deuterium (99.7% pure,
Catalyst Suggested Application % Co % MO Matheson Gas Products), and samples of
thiophene were injected. The desorbed
G HDS, HDN, & Hydro- 3.8 16.0
genation of lights and
thiophene was then trapped for mass spec-
mid-distillates tral analysis by condensation in liquid-
G HDS of light & heavy 2.2 18.0 air-cooled glass traps.
petroleum feeds
C3 HDS of light & mid- 3.1 15.0 Activity Tests
G HDS of light & mid- 2.8 12.0 Activity tests were carried out in a pres-
distillates surized continuous-flow reactor system de-
C5 HDS of light & heavy 4.0 13.0 scribed previously (16). Catalysts were
petroleum feeds sulfided with a 25% solution of carbon
G HDS of light & mid- 2.4 9.8
disulfide in n-heptane as before (16). A 12%
G Desulfurization of gas- 1.5 11.3 solution of thiopene in n-heptane was em-
eous feeds ployed for the HDS reactions, which were
G Desulfurization of light 2.2 11.5 carried out under the following conditions:
to mid-distillates LHSV, 4.5 liters/liter/hi-; GHSV, 400
G HDS of light & heavy 3.0 14.3 liters/liter; pressure, 30 atm; temperature,
petroleum feeds (with
drawn from market) 300C. Each catalyst was run for a week
and samples were taken every day to assure

the maintenance of steady-state conditions. TABLE 3

Thiophene desulfurization activity (average Magnetic Properties of the Co/A&O, Catalysts
value of thiophene converted to hydrocar-
bon) for each catalyst was determined from Catalyst No. co (VA%,) Molar Magnetic
susceptibility moment
three consecutive samples taken during (lOWemu) (PB)
steady-state conditions.
Prepared from
Analyses cobalt nitrate
1 0.42 0.19 4.03
Gas chromatography. Conversions of 2 0.75 0.64 4.20
3 1.22 1.69 4.77
thiophene to hydrocarbons in the activity 4 1.79 2.16 4.39
tests were determined by gas chromatogra- 5 2.30 3.08 4.54
6 2.8 2.72 3.89
phy with a Varian Model 920 gas chromatog- Prepared from
raph. The gas chromatography column cobalt acetate
7 0.86 1.09 4.77
was 10 ft x 4 in. and contained 15% iso- 8 1.76 1.99 4.27
octadecene on 60- to go-mesh Chromosorb- 9 2.31 3.19 4.60
10 3.95 4.56 4.15
P. It was used at 70C with a hydrogen flow
rate of 60 ml/min at 40 psig.
Mass spectrometry. Deuterium distribu- second maximum occurs in the magnetic
tions of recovered substrates in the pulse moments at about 2.3% cobalt (catalyst
experiments were determined by mass No. 5 in Table 3). This arises for samples
spectrometry. A Varian Model MAT CH-5 prepared from both cobalt(I1) nitrate and
single-focusing 90 magnetic sector mass acetate (see catalyst No. 9 in Table 3).
spectrometer was used. All spectra were Impregnation with cobalt(I1) acetate
run at constant ion pressure and an ionizing gives catalysts with slightly different mag-
voltage of 11-70 eV. Parent peak regions netic properties from those prepared from
were scanned at low electron voltage (de- cobalt(I1) nitrate. At the lower cobalt con-
termined by running standard thiophene) to centrations the catalysts prepared from
eliminate any hydrocarbon fragments. The cobalt(I1) acetate have higher magnetic mo-
peak heights of each deuterated compound ments (see catalysts Nos. 2 and 7 in Table
were accurately measured and deuterium 3). This indicates a higher proportion of
distributions calculated according to previ- octahedral cobalt(I1). At high cobalt con-
ously reported procedures (2). centrations (about 2.5%) the samples pre-
pared from cobalt(I1) acetate also have
higher magnetic moments (compare cata-
lysts Nos. 9 and 10 with catalysts Nos. 5
1. Magnetic Measurements and 6) indicating less oxidation to co-
Results of the magnetic measurements on balt(II1).
the Co/y-AlzOs catalysts are shown in Ta-
ble 3. Initially the magnetic moments rise as 2. Pulse Experiments
the amounts of cobalt increase. This rise is Results from the deuterium exchange of
attributed to the increasing proportion of thiophene over catalysts l-20 (Table 1) and
octahedrally coordinated cobalt relative to over the commercial catalysts (Table 2) are
the tetrahedrally coordinated form (I 7). shown in Table 4. In addition thiophene
Following the initial rise at about 1% cobalt exchanges were also run over the alumina
(catalyst No. 3 in Table 3) the magnetic support, the alumina support impregnated
moments start to decrease. This decrease is with small amounts of cobalt molybdate
due to the presence of diamagnetic Co3+ (before and after calcination), cobalt mo-
ions formed by aerial oxidation of Co2+ions lybdate, and cobalt aluminate.
during calcination of the catalysts. Finally a Very little exchange occurs over the lat-
Results of Deuterium Exchange Experiments at 300C on the Commercial and Prepared Co/AI,O,, Mo/AI,O,, MO-Co/Al,O,, and Co-Mo/Al,O,

(0.2 g catalysts, 60 cmJ/min D, at 60 psi, 5 ~1 sample size)

Catalyst 1 2 3 4 5 6 I 8 9 10 11 12 13 14 15 16 17 18 19 20

do 31.1d (23.3) 5.1 10.6 23.1 8.7 9.5 5.0 5.5 31.7 9.8 13.6 6.4 7.5 5.9 4.4 1.4 1.3 2.7 11.4 5.0
4 31.9 (41.0) 15.6 19.7 30.3 35.4 15.4 9.4 14.7 34.7 24.6 16.1 11.8 23.6 12.1 18.2 12.2 6.4 12.2 24.0 19.3
4 24.3 (27.0) 27.4 31.1 31.3 37.1 33.1 22.4 29.8 27.5 45.5 39.6 36.9 42.9 40.5 50.2 45.9 31.2 37.1 36.6 39.2
4 10.0 ( 7.9) 26.7 23.9 11.8 18.8 24.4 31.8 29.6 5.4 16.4 21.1 26.3 21.0 28.9 20.2 29.0 36.8 30.7 20.4 27.9
2.8 ( 0.9) 25.2 14.7 3.6 0.0 17.7 31.4 20.5 0.7 3.7 9.6 18.6 5.0 12.0 7.0 11.6 24.3 17.4 1.6 8.7
1.22 2.51 2.12 1.42 1.66 2.25 2.72 2.45 1.09 1.80 1.97 2.39 1.93 2.28 2.07 2.37 2.76 2.48 1.90 2.16

Catalyst Cobalt Cobalt Alu- CoMoO,/Al,O~ CoM00JA1203~

molybdate aluminate mina (uncakined) (calcined

77.7 87.0 18.4 30.2 30.3

17.3 8.3 34.4 28.4 26.8
5.8 5.0 31.3 24.8 22.3
0.8 0.2 12.6 11.9 13.6
0.0 0.0 3.2 4.8 7.1
0.27 0.18 I .48 1.33 1.40

Catalyst Cl G C3

3.6 12.4 1.8 15.5 6.5 2.2 33.5 22.4 23.0

14.4 22.3 7.3 12.9 7.3 7.0 31.6 37.9 27.4
51.1 36.6 46.4 36.0 35.0 26.2 27.8 35.2 34.2
24.4 21.1 30.7 21.2 30.7 34.6 1.4 4.2 12.3
6.6 7.7 13.9 14.4 20.6 29.9 0.2 0.3 3.1
2.16 1.90 2.47 2.06 2.52 2.83 0.96 1.22 I .45

See Table 2.
See Table 1.
Deuterated thiophene; e.g., d, represents thiophene containing no deuterium, d, containing one deuterium, d2 two, etc.
ri Percent deuterated thiophene species after deconvolution of mass spectrum; see Ref. (2).
r Random distribution calculated from binomial expansion of (H + fIY: such random distributions for other results may be calculated by computing vaules for H and D from da,
as follows: His (4 minus d,,) divided by 4, D is I minus H.
Average number of deuteriums per thiophene molecule.
s Catalyst E,; see Table 1.

ter two catalysts (Table 4), and the molyb- 4. Activity Tests
date compound exhibits somewhat greater The results of the activity tests carried
and more random exchange than the alumi- out for catalysts l-20 (Table 1) and the
nate. In contrast to previously reported commercial catalysts (Table 2) are shown in
results of exchange of thiophene over a Table 6. Catalyst C, showed considerable
different alumina catalyst (2), deuterated cracking activity in addition to desulfuriza-
species containing more than four deute- tion. However, it is possible to diminish
rium were not detected over our alumina this cracking activity while still maintaining
support (Table 4). Even so, the exchange the HDS activity at the 100% level by
pattern supports our previous suggestion diluting the catalyst bed with glass beads
(2) of exchange sites existing on y-alumina (18). Each activity experiment was run for
catalyst. about 7 to 10 days. During this period the
activity of all catalysts reached an equilib-
3. Pore Diflusion rium value. However, the activity of the
We considered the possibility that pore laboratory-prepared catalysts started to de-
diffusion might modify the results of the crease from their equilibrium values toward
exchange experiments, and, therefore, we the end of each run, but we did not attempt
carried out surface area and pore size mea- to determine the life of each individual
surements on some of the commercial cata- catalyst. On the other hand the commercial
lysts (Table 5). If pore diffusion were a catalysts were more stable even though
problem, we would expect thiophene to some of them had lower equilibrium activi-
remain in small pores longer than in larger ties than the synthetic catalysts.
pores and, as a result, expect multiple These activity tests show first that the
exchange to be greater in smaller pores. activity order for the prepared catalysts
However, catalysts C, and C, in Table 5, was MO-Co/yAl,O, = Co-MO/~-Al,O, >
which have similar average pore radii, ex- MO/~-A&O, > Co/y-A&O,. Second, this
hibit different exchange patterns (Table 4), order confirms the general belief that cobalt
while catalysts C, and Cg, with widely is not the desulfurization species in this
ditferent average pore radii, exhibit very type of catalyst (2, 3, 19). And, finally, in
similar exchange patterns. In fact Cg, with agreement with previous observations (2)
the smaller average pore radius, exhibits the unimpregnated alumina support (Table
less rather than more multiple exchange. 6) exhibits negligible activity.
We suggest, therefore, that pore diffusion
modification of these exchange patterns is DISCUSSION
not as important as other factors. I. The Alumina Support Impregnated
with Cobalt Molybdate
TABLE 5 The exchange pattern of thiophene over
Results of Surface and Pore Measurements Carried the alumina support is nearly random (Ta-
Out on Some Commercial Catalysts ble 4). Upon impregnating the support with
small amounts of cobalt molybdate the ran-
Catalyst BET Total Average
surface area pore volume pore radius dom character of the exchange diminishes
W/g) (ml/g) (A) (CoMoO,/r-Al,O,, uncalcined , Table 4)
and the desulfurization activity increases
Cl 318 0.64 35 (Table 6). Calcination of this catalyst pro-
G 270 0.50 31 duces very small effects on the exchange
G 215 0.48 21
165 0.65 75 patterns but almost doubles the activity.
We conclude, therefore, that structural
a See Table 2. changes in the cobalt and molybdenum are

Percent Thiophene Desulfurized at 3OOC,over the Commercial and
the Prepared* Co/y-A&O,, MO/~-A&O,, MO-CO/~-A&O,, and Co-Mo/yAl,O, Catalysts

Catalyst % Thiophene Catalyst % Thiophene Catalyst % Thiophene

No. converted No. converted No. converted

Co/y-A&O, MO-Co/y-Al,O, Commercial

1 50. I 15 100 C, 98.3
2 7.5 16 100 G 87.3
3 19.2 17 100 G 99.4
4 34.1 18 100 G 95.0
5 31.5 Co-MO/~-A&O, G loo (plus
12% cracking)
6 26.1 19 97.2 90.2
I 4.8 20 97.5 51.6
8 38.5 Alumina support 0.61 88.0
9 9.4
10 19.5 CoMoO,/y-A&O, 28.5 G 70.6
(before calcination)
11 71.1 CoMoO,/yAl,O, 54.5
(after calcination)
12 72.4
13 98.5
14 82.7

n See Table 2.
b See Table 1.
r Catalyst E,; see Table 1.

more important for activity than for ex- to 3 and this corresponds with an increase
change. We also conclude that the active of dz formation (Table 4). Therefore dz
desulfurizing species in the Co-MO/~-Al,O,, formation in the thiophene exchange reac-
catalysts is probably not simple cobalt tion may be correlated with octahedral co-
molybdate but may be related to it. balt in the Co/y-A&O, catalysts. Second, in
contrast, catalyst No. 5 shows a large
2. State of Cobalt amount of d, and dz formation while its
Magnetic measurements are of limited exchange pattern is closer to random than
value in providing information about the any other (Table 4). But random exchange
state of cobalt in the CO/MO catalyst sys- has previously been observed over a Co/y-
tems. Calculations, based on such measure- A&O, catalyst containing mostly Co3+ ions
ments, cannot be carried beyond a cer- (2). Therefore, we suggest that catalyst No.
tain point because of interference of 5 contains CO*+ mostly in the octahedral
diamagnetic Co3+. Therefore, the second state, giving rise to large d2 formation, and
maximum in the magnetic moments of the enough Co3+ to produce random exchange.
Co/y-Al,O, catalysts is (catalyst No. 5, The ratio of Co2+ (octahedral) to Co3+ must
Table 3) not understandable by magnetic be large so that despite the diamagnetism of
measurements alone. However, on the the latter, the catalyst shows a high mag-
basis of our exchange data some tentative netic moment (Table 3).
statements can be made about the state of Finally, different values of magnetic mo-
the cobalt. First the percent of octahedral ments have been reported for Co/y-Al,O,
cobalt gradually increases from catalyst 1 catalysts containing this same range of co-

balt content but on a different alumina 20,7 in Table 4). It also maximizes the
support (27). Thus it appears that different exchange pattern toward d,, brings it closer
aluminas accept cobalt in different forms. to randomness, and lowers the day (the aver-
This may account for the desirability of one age number of deuteriums per molecule).
alumina preparation over another. The additional presence of molybdenum,
The Co/y-A&O, catalysts prepared from therefore, seems to increase random ex-
cobalt acetate show exchange patterns dif- change at the expense of multiple exchange
ferent from their matching pairs in the (greater than random amounts of da) on
nitrate series. Such differences might be Co/y-A&O, alone. On the other hand, co-
due to shift differences in the magnetic balt impregnated on the molybdenum-con-
properties (Table 3) resulting from some taining catalysts seems to enhance the ef-
carbonaceous species deposited on the cat- fects of molybdenum (compare catalyst
alyst surface during calcination of the ace- pairs 11,15; 12,16; 13,17; and 14,18 inTable
tate ion. For example, the exchange activ- 4). For the two lower concentration MO
ity of catalyst 7 peaks toward d3 and d4 in catalysts (11 and 12) the effect of Co is
contrast to its matching catalyst in the dramatic in two respects. Dideutero species
nitrate series, catalyst 3, for which dz for- are increased well above random and desul-
mation is maximum. Similar differences oc- furization is increased. In contrast, for the
cur for catalyst pairs 8,4 and 95. On the two higher concentration MO catalysts the
other hand, catalyst 10 shows an exchange effect of Co is less. However, this is be-
trend similar to that of its nitrate counter- cause dz and desulfurization were already
part, catalyst 6. large on the MO catalyst.
Catalysts 1, 4, 5, and 8 exhibit the highest The exchange patterns for the commer-
desulfurization activities in the Co/y-A&O, cial catalyst fall into three categories. The
series. But there does not seem to be any first pattern resembles the cobalt-on-mo-
clear correlation between desulfurization lybdenum catalysts (15-18) and is exhibited
and magnetic properties. Only a crude cor- by commercial catalysts C,-C,. It is char-
relation occurs between desulfurization and acterized by a volcano-shaped peak with a
percent dz. If catalyst 10 is excluded, the dz maximum at d2. The second pattern is ex-
goes through a maximum at catalysts 5 and hibited by C5 and C6and is characterized by
6 as desulfurization increases. Apparently large d,, with higher than random amounts
good desulfurization activity requires a cer- of dz and, especially, d4. Both of these sets
tain ratio of tetrahedral to octahedral co- of catalysts exhibit high desulfurization ac-
balt. The presence of some Co3+ does not tivity. The third pattern is exhibited by C,-
interfere with desulfurization as long as the Cg and contains low d,,.
required ratio of tetrahedral/octahedral It is difficult to correlate exchange pat-
CoZ+ exists. terns with desulfurization activity, but ref-
erence to Table 6 generally indicates that
3. Molybdenum Catalysts catalysts C,-C, are of the higher activity
The exchange patterns from the MO/~- type. More specifically, d,, of 1.9 or higher
A&O, catalysts peak toward dZ formation generally corresponds with good desulfur-
(Table 4, catalysts 11-14). This same pat- ization activity.
tern was observed for a different MO/~-
Al,O, catalyst at 200C rather than 300C 4. Mechanistic Considerations
(2). This difference might be the result of Higher d,, values, which correspond to
the different alumina supports involved. good desulfurization activity, are a result of
Molybdenum deposited on a Co/y-A&O, some multiple exchange. This suggests flat
catalyst greatly increases desulfurization adsorption of thiophene molecules on the
activity (compare catalyst pairs 19,2 and surface. Flat adsorption appears to be nec-

essary for desulfurization (2, 9), so higher catalysts (Table 4), suggests two exchange-
d,,, values appear to correlate with good able sets with an energy barrier separating
desulfurization activity. Greater than ran- them. Such an exchange pattern can be
dom amounts of d2 and d4, which are found rationalized on the basis of a modified
for nearly all the molybdenum-containing Komarewsky mechanism (5):

We suggest that the results over CO/MO sheds some light on the problem. With the
catalysts arise from an energy barrier order exceptions of catalysts 5 and 10 cobalt
of6>4>2>3 = 5. Over certain CO/MO catalysts have lower d2 and higher d4 than
catalysts the barrier at 6 is lower and random. This suggests a lower barrier at 4
greater desulfurization activity occurs, but than on Mo/Al,O, catalysts, and possibly a
because the order may become 4 > 2 > 6 > process leading directly to multiple (d4)
3 = 5, less d3 and d4 and more d2 are exchange.
obtained. Poor HDS catalysts of CO/MO
type seem to have the barrier at 4 much ACKNOWLEDGMENT
higher than other barriers, so d, and d, The authors thank NIOC management for permit-
appear, but only little d3 and d4. ting the publication of this paper. Thanks are due also
to Mr. H. E. Najmabadi, Dr. A. Badakhshan, and Dr.
Certain MO/~-A&O, catalysts show ap- M. Shahab for their interest in this work. Useful
preciable desulfurization activity, but the discussions with Professor G. V. Smith are deeply
relative heights of the energy barriers are appreciated.
not so obvious. The addition of cobalt to
these catalysts invariably increases desul- REFERENCES
furization activity, i.e., lowers the barrier Kieran, P., and Kemball, C., J. Catal. 4, 394
at 6. However, the exchange patterns re- (1965).
Smith, G. V., Hinckley, C. C., and Behbahany,
main similar and suggest a concomitant F., J. Car&. 30, 218 (1973).
lowering of the other barriers. Desikan, P., and Amberg, C. H., Can. J. Chem.
The correlation between da, and desul- 42, 843 (1964).
furization over the Co/Al,O, catalysts is Sheikhrezai, Z., Djalali, M., Salajegheh, S., and
opposite to that of the CO/MO catalysts. Behbahany, F., in Catalysis in Organic Syntheses,
1977, Ed. Gerard V. Smith, New York, Academic
That is, higher desulfurization activity is Press, 1977.
found for lower da,., In these cases the Komarewsky, V. I., and Knaggs, E. A., Ind. Eng.
relative proportions of octahedral and tetra- Chem. 43, 1415 (1951).
hedral Co*+ and Co3+ may play important Owens, P. J., and Amberg, C. H., Advan. Chem.
roles since the da, of catalysts 2 through 4 Ser. 33, 182 (1961).
Owens, P. J., and Amberg, C. H., Can. J. Chem.
go through a minimum with increasing co- 40, 941 (1962).
balt concentrations. Examination of the de- Owens, P. J., and Amberg, C. H., Can. J. Chem.
viation of the patterns from randomness 40, 947 (1962).

9. Smith, G. V., and Cowley, S. W., The Hydro- 15. Crauston, R. W., and Inkley, F. A., in Advances
genolysis and Exchange of Sulfur Heterocycles in Catalysis and Related Subjects (D. D. Eley,
Present in Coal, the 41st Annual ACS Chemical W. G. Frankenburg, V. I. Komarewsky, and P. B.
Engineering Symposium, 1975, Pittsburg, Penn. Weisz, Eds.), Vol. 9, p. 143. Academic Press,
10. Kolboe, S., Can. J. Chem. 47, 352 (1969). New York, 1957.
I!. Mikovsky, R. J., Silvestri, A. J., and Heineman, 16. Salajegheh, S., Djalali, M., and Sheikhrezai, Z.,
H., J. Catal. 34, 324 (1974). Bull. Iran. Petrol. Sot. 62, 16 (1976).
12. Tomlinson, J. R., Keeling, R. O., Rymer, G. T.,
17. Ashley, J. H., and Mitchell, P. C. H., J. Chem.
and Bridges, J. M., Preprints of the 2nd Interna-
Sot. A, 2821 (1968).
tional Conference on Catalysis, paper No. 90,
Paris, 1960. 18. Djalali, M., Salajegheh, S., and Sheikhrezai, Z.,
13. Figgis, B. N., and Nyholm, R. S., J. Chem. Sot., Bull. Iran. Petrol. Sot. 64, 1 (1976).
4190 (1958). 19. Ahuja, S. P., Derrien, M. L., and Le Page, J. F.,
14. Selwood, P. W., Magnetochemistry, 2nd ed., Ind. Eng. Chetn. Prod. Res. Develop. 9(3), 272
pp. 78, 92. Interscience, London, 1956. (1970).