Hydrodesulfurization of dibenzothiophene over siliceous MCM-41-supported catalysts: I. Sulfided Co-Mo catalysts

Article abstract of DOI:10.1006/jcat.2000.3148

The HDS of dibenzothiophene (DBT) over Co-Mo catalysts supported over siliceous MCM-41 was studied. A radioisotope tracer method was used to trace the behavior of sulfur during HDS to elucidate the mechanism of HDS reaction. The extremely high surface are

Full text of DOI:10.1006/jcat.2000.3148

Journal of Catalysis 199, 19–29 (2001)
doi:10.1006/jcat.2000.3148, available online at http://www.idealibrary.com on
Hydrodesulfurization of Dibenzothiophene over Siliceous
MCM-41-Supported Catalysts
I. Sulfided Co–Mo Catalysts
Anjie Wang, Yao Wang, Toshiaki Kabe,† Yongying Chen, Atsushi Ishihara,† and Weihua Qian†
State Key Laboratory of Fine Chemicals, Dalian University of Technology, 158 Zhongshan Road, Dalian 116012, People’s Republic of China;
and †Department of Applied Chemistry, Faculty of Technology, Tokyo University of Agriculture and Technology,
2-24-16 Nakacho, Koganei, Tokyo 184, Japan
Received June 14, 2000; revised December 11, 2000; accepted December 11, 2000; published online March 7, 2001
W-based catalysts supported by -Al2O3. They are active
in converting thiophene and benzothiophenes, but not ac-
tive enough to desulfurize efficiently the most refractory
sulfur-containing polyaromatic compounds, e.g., diben-
zothiophene (DBT) and its alkyl-substituted derivatives.
Whitehurst et al. concluded, based on the composition of a
typical gas oil and the first-order kinetics for hydrodesulfu-
rization of different sulfur-containing compounds, that the
reaction time will have to be increased by a factor of about 4
when using conventional catalysts to reduce the sulfur con-
tent from 2500 to 500 ppm (1). Thus, deep hydrodesulfuriza-
tion becomes of great research interest, aiming at removing
DBT and its derivatives.
A deep hydrodesulfurization (HDS) catalyst, which was consid-
erably active to desulfurize dibenzothiophene (DBT), was prepared
by depositing Co–Mo species over siliceous MCM-41. The extremely
high surface area of MCM-41 favors the dispersion of the active
species, resulting in very high HDS activity. The optimal Co/Mo
atomic ratio for this series of catalysts is 0.75, higher than the
conventional -Al2O3-supported catalysts. This is attributed to the
higher dispersion of the active species and more Co–Mo pairs gen-
erated on the surface of MCM-41. In the liquid product obtained at
temperatures over 280 C, 70–80% BP was generated and its selec-
tivity changed a little with increasing HDS reaction temperature.
The selectivity of cyclohexylbenzene (CHB) decreased whereas the
selectivity of benzene and cyclohexane increased when increasing
the reaction temperature. It is shown that HDS of DBT over MCM-
41-supported Co–Mo sulfides includes mainly three reaction path-
ways: hydrogenolysis, prehydrogenation followed by desulfuriza-
tion, and hydrocracking of CHB. The hydrogenolysis of DBT, i.e.,
direct extraction of sulfur atom from DBT molecule, predominately
progresses in the HDS reactions, while benzene mainly results from
hydrocracking of CHB. 35S isotope tracer investigation revealed
that sulfur atoms retained on the surface could be released only by
the introduction of a sulfur-containing compound, indicating that
sulfur atom exchange between sulfur-containing compounds and
the active sites is involved in the HDS reaction. A reaction mecha-
nism for HDS is proposed, which is in accordance with the isotope
tracer experiment results, the well-established rim–edge model, and
c
Mesoporous MCM-41 has been the focus of much
research interest since its discovery because it offers a
˚
uniform pore of 15 to 100 A and a very high surface area
(2). The unique properties of MCM-41 provide a potential
for catalytic application to the bulky molecule involved
reactions. The extremely high surface area favors a high
dispersion of active species and is often conducible to high
catalytic activity. The large pore channels may reduce pore
diffusion limitation and allow the effective use of the active
sites on the surface of the pore wall by the reactants. The
MCM-41-supported catalysts have been developed for a
variety of reactions (3), including hydrodesulfurization of
petroleum fractions (4).
Shortly after the discovery of MCM-41, Corma et al. (4)
used Al–MCM-41 to support Ni–Mo species in order to
prepare catalysts for hydrocracking of vacuum gas oil. The
prepared catalysts were investigated in terms of the activi
ties of hydrodesulfurization (HDS), denitrogenation
(HDN), and mild hydrocracking. They found that Al–
MCM-41-supported catalysts were more active in HDS,
HDN, and hydrocracking than those supported over USY
or -Al2O3. Reddy et al. (5) supported Co–Mo and Ni–Mo
over Al–MCM-41 to investigate the hydrodesulfurization
of petroleum residues and atmospheric and vacuum
the DFT calculations.
Key Words: deep hydrodesulfurization; MCM-41; catalyst; mech-
anism.
2001 Academic Press
1. INTRODUCTION
The increasing concerns of global air environment have
stimulated the demand for cleaner and cleaner engine fu-
els. In modern refinery, hydrotreating and hydrocracking
have become key processes to meet this demand. Conven-
tional catalysts used in the process include mainly Mo- or
19
0021-9517/01 $35.00
c
Copyright
2001 by Academic Press
All rights of reproduction in any form reserved.

20
WANG ET AL.
residues. They also prepared -Al2O3-supported catalyst was evaporated at ambient temperature and dried for 24 h
using the same procedure for comparison. The Al–MCM- in vacuo to obtain 35S-labeled sulfur. With the 35S-labeled
41-supported catalyst was superior in HDS activity to its - sulfur, [35S]DBT was synthesized according to the method
Al2O3-supported counterpart, but not as active as the com- in the literature (9).
mercialcatalyst. In the same group, Songand Reddy(6) also
investigated the hydrodesulfurization of DBT over Co–
Mo/Al–MCM-41 at both high and low metal loading. Co–
Mo/Al–MCM-41 catalyst at normal metal loading level was
less active in converting DBT than the -Al2O3-supported
one at 350 or 375 C. However, The Al–MCM-41-supported
catalyst at high metal loading was substantially more ac-
tive than its -Al2O3-supported counterpart in DBT con-
version. They concluded that sulfided Co–Mo/Al–MCM-
41 catalysts are promising for deep hydrodesulfurization
of distillate fuels under relatively mild conditions. It is in-
teresting to note that all the research on MCM-41 type
supports in HDS catalyst preparation has focused on Al–
MCM-41, probablyhopingthat the strongacidityofthe sup-
port may help to crack the polyaromatic sulfur-containing
compoundsso asto improve the HDSactivity. Nevertheless,
no great improvement in HDS activity has been achieved
on this type of support.
Recently, Ram´ıre et al. (7) reported that strong acidity
of the support (Al2O3 + MCM-41) was detrimental to the
HDS activity of the catalysts. Only the weak and medium
acids on the surface favor the formation of HDS active sites.
In the present study, the HDS of DBT over Co–Mo cata-
lysts supported over siliceous MCM-41 has been investi-
gated. And a radioisotope tracer method was used to trace
the behavior of sulfur during HDS in order to elucidate the
mechanism of HDS reaction.
2.2. Catalysts
Co–Mo/MCM-41 catalysts were prepared in the follow-
ing way. MCM-41 was impregnated with an aqueous solu-
tion of(NH4)6Mo7O24 4H2O and Co(NO3)2 6H2O for 12h
at room temperature, followed by the evaporation of sol-
vent, an oven drying at 120 C for 5 h, and calcination in air
at 450 C for 5 h. A 20 wt% MoO3 loading level was chosen
for preparing this series of catalysts. The content of cobalt
was determined by varying the atomic ratio of Co/Mo in the
range of 0–1.0. The catalysts were denoted as Co–Mo(x).
The value in parentheses represents the atomic ratio of Co
to Mo.
The X-ray diffraction (XRD) patterns of the MCM-41
and the supported catalysts were measured on a Rigaku
RAD-2X diffractometer using nickel-filtered CuK radia-
tion at 35 kV and 20 mA. Nitrogen adsorption measure-
ments were performed using a Coulter SA 3100 adsorption
analyzer, which reports adsorption isotherm, BET specific
surface area, and pore volume automatically. The samples
were outgassed for 4 h at 300 C before the adsorption.
The temperature programmed reduction (TPR) patterns
of the Co–Mo catalysts were measured on a Chembet-3000
analyzer. Before the measurement, the sample has been
treated in argon at 200 C for 2 h. A gas mixture of H2 and
Ar (5 vol% H2 in argon) was used as the reacting agent.
The patterns of the catalysts were measured from room
temperature to 930 C at 10 C/min.
2. METHODS
2.3. System and Procedure for HDS and Isotope Tracer
2.1. Materials
HDS of DBT was carried out in a fixed-bed stainless
Both ammonium heptamolybdate tetrahydrate [(NH4)6 steel tubular reactor having a dimension of 8.0 mm in in-
Mo7O24 4H2O] and cobalt nitrate hexahydrate ternal diameter. Catalysts were pelleted and then crashed
[Co(NO3)2 6H2O] were of commercial GR grade and screened to 20–35 mesh. Catalyst particles (0.2 g) were
(Kishida Chemicals). Dibenzothiophene ([32S]DBT) was diluted with quartz sand (30–50 mesh) to a constant volume
the product of Acros Organics. Decalin was obtained from of 1.5 cm3 before being charged into the reactor. The HDS
Kishida Chemicals and used as solvent without further activities of the prepared catalysts were evaluated using a
purification. Hydrogen (99.99% ) was supplied from Tohei model fuel: 1 wt% DBT in decalin.
Chemicals, and hydrogen sulfide in hydrogen (H2S 5.0% )
The catalysts were presulfided before HDS reaction by
was from Takachio Chemicals. MCM-41, as the catalyst means of a mixture of H2S/H2 (5 vol% H2S) at atmo-
support, was prepared following the procedure reported spheric pressure and 400 C for 3 h. After the presulfida-
in the previous paper (8). The preparation of the support tion, the reactor was cooled to the HDS reaction tem-
included the hydrothermal reaction of a silica gel mixture perature in H2S/H2 and then pressurized to 5.0 MPa by
of pH 10.5 at 130 C for 12 h and postsynthesis heat treat- H2. During the cooling period, the model fuel was fed
ment at 600 C for 12 h in a N2 flow followed by calcination into the reactor to wet the catalyst bed by a high-pressure
in air at 540 C for 6 h. [35S]DBT was synthesized by the metering pump (Kyowa Seimitsu KHD-16). The reaction
following method: a commercial toluene solution of 35S conditions for the HDS reaction were temperature, 260–
(total radioactivity, 1 mCi; Amersham Co., Ltd.) was 380 C; total pressure, 5.0 MPa; H2/feed ratio, 1200 Nm3/
mixed thoroughly with 32S powder and then the solvent m3; and WHSV, 45 h 1. The reaction products were

SULFIDED Co–Mo CATALYSTS
21
separated in a gas–liquid separator to collect the liquid
products. The H2S produced was absorbed by bubbling
through a commercial basic scintillator solution (Carb-
sorb, Packard Co. Ltd.). Sampling of both liquid and gas
products was started 3 h after the reaction conditions
had been achieved. For each run, three to five pairs of
the liquid and absorbed H2S solution samples were col-
lected at intervals of 20 min. The liquid products were
analyzed by a gas chromatograph equipped with an FID
detector (Shimazu 17A, Shimazu Co. Ltd.) using a com-
mercial capillary column (DB-1).
FIG. 2. Reaction path network for hydrodesulfurization of diben-
zothiophene.
For 35S-labeled isotope radioactivity tracer experiments,
radioactivities of the unreacted [35S]DBT in liquid prod-
ucts and the [35S]H2S in the absorbed solution were mea-
sured using a liquid scintillation counter (Beckman LS-
6500, Beckman Co., Ltd.) after adding proper scintillator
solvent (Permaflour or Instafluor, Packard Japan Co. Ltd.)
to each fraction. Literature is available for the liquid scin-
tillation counting (10).
To investigate the effect of H2S on the HDS activities
in situ, a pulse tracer method was developed. A pulse in-
jection system is inserted in the stream of H2, as shown in
Fig. 1. After the HDS conversion became constant, 5.4 ml of
[35S]H2S was introduced into the reaction bed by a six-way
valve. The [35S]H2S absorbed in the Carbsorb solution was
collected every 15 min, and its radioactivity was determined
as mentioned above.
DBT). The conversion of DBT represents the activity
of HDS for the catalysts, and the selectivity of BP is a
measure of DBT hydrogenolysis. Since the chromato-
graphic peak of benzene and cyclohexane is difficult to
deconvolute, the formation of CHB is used to describe
hydrogenation of DBT while the total of benzene and cy-
clohexane is used to represent of the hydrocracking of DBT.
The rate constant of pseudo-first-order reaction of HDS
of DBT, kHDS (atom/nm2 s), was determined by (11)
ln(1 x) m F
kHDS
=
,
A
where x is the conversion of DBT, m is the charge weight
of the catalyst (g), F is the feed rate of DBT (atom/s), and
A is the specific surface area (nm2/g). Similarly, the rate
constants of hydrogenolysis, hydrogenation, and hydro-
cracking of DBT are calculated as
2.4. Measurement of Catalyst Activity
kHYG = kHDS SBP
kHYD = kHDS SCHB
The activities of HDS, hydrogenolysis (HGY), hy-
drogenation (HYG), and hydrocracking (HYC) were
estimated by the analysis of the liquid products, based
on the well-established reaction path network shown in
Fig. 2 (6). In the present study, the detectable com-
pounds in the liquid samples include biphenyl (BP),
cyclohexylbenzene (CHB), benzene, cyclohexane, and
a trace amount of tetrahydrodibenzothiophene (TH-
kHYC = kHDS
SB+C/2,
where SBP, SCHB, and SB + C are selectivities of BP, CHB,
and the total of benzene and cyclohexane, respectively.
To judge the limiting step involved in HDS reaction, the
Weisz modulus (8) was used since the intrinsic reaction
rate was not available. In the calculation, it was assumed
that the HDS reaction was of first order with respect to
DBT because H2 was in large excess. The catalyst particles
were approximately regarded as spherical, and the maxi-
mum particle size (0.84 mm) was used in the calculation. As
the pore diffusion belongs to the Knudsen diffusion range
for the mesoporous material, DK was used as the mass trans-
fer coefficient in the calculation.
3. RESULTS
3.1. Characterization of MCM-41 and Its
Supported Catalysts
FIG. 1. Schematic diagram of the HDS reaction system with an iso-
tope tracer pulse injection unit. 1, gas cylinder; 2, regulator; 3, stop valve; 4,
pressure gauge; 5, six-way valve; 6, sampler loop; 7, trap bottle; 8, reactor;
9, gas–liquid separator; 10, stop valve; 11, needle valve; 12, flowmeter; 13,
trap bottle.
The X-ray diffraction patterns of the synthesized
MCM-41 and the supported Co–Mo catalysts are shown in
Fig. 3. It can be seen that there wasa strongreflection at 2 =
1.92 , indicating that the mesostructure is well developed.

22
WANG ET AL.
TABLE 1
Physical Properties of the Prepared Catalysts
BET surface Pore volume Mo content Co content
Catalyst
(m2/g)
(cm3/g)
(atom/nm2) (atom/nm2)
Mo/MCM-41
Co–Mo(0.25)
Co–Mo(0.50)
Co–Mo(0.75)
Co–Mo(1.00)
778
790
685
608
531
0.57
0.58
0.55
0.49
0.38
1.08
1.06
1.22
1.38
1.58
—
0.27
0.61
1.04
1.58
depositing the metal species. The decreases in specific sur-
face area and pore volume result partly from the density
increase by depositing the metal species and partly from the
pore blocking by the species. It can be seen that the pore
volume decreases remarkably with increasing Co loading
level, suggesting that pore blocking may be enhanced at
high Co-loading levels.
FIG. 3. XRD patterns of the synthesized siliceous MCM-41 and Co–
Mo/MCM-41 catalysts. (a) Mo/MCM–41, (b) Co–Mo(0.25), (c) Co–Mo
(0.50), (d) Co–Mo(0.75), (e) Co–Mo(1.00).
The TPR patterns of MCM-41 and its supported
catalysts are illustrated in Fig. 5. The positions of the main
TPR maxima are summarized in Table 2. It is indicated that
the patterns are different from the normal Co–Mo/Al2O3
catalysts (12). The main maxima for Co–Mo/MCM-41 cata-
lysts with high Co-loading correspond to the pattern of
CoMoO4, while a shoulder peak appears in the vicinity
of 600 C. The TPR patterns of both Co/MCM-41 and Mo/
MCM-41 are different from those of Co/Al2O3 and Mo/
Al2O3. It is indicated that the interaction between the metal
species and MCM-41 is quite different from that between
the species and Al2O3.
In the XRD patterns of the MCM-41-supported catalysts,
no distinct reflection was observed except a small peak in
the vicinity of 2 = 26 which is assigned to CoMoO4 crystal
(Card No. 21-868). It is indicated that the CoO and MoO3
species are highly dispersed on the surface of MCM-41.
The N2 adsorption–desorption isotherm of the synthe-
sized MCM-41 is shown in Fig. 4. It is a typical
type IV isotherm of mesoporous materials. The N2 adsorp-
tion amount increases sharply in the range of P/P0 = 0.3–
0.4. The sharp increase in uptake results from the capil-
lary condensation of N2, indicating that uniform mesopores
are present in the siliceous MCM-41. From the isotherm,
the structural parameters of the synthesized MCM-41 were
3.2. Hydrodesulfurization of DBT
A model fuel containing 1 wt% DBT in decalin
was used to investigate the HDS activities for Co–
Mo sulfided catalysts supported over MCM-41. The
main products are BP, CHB, benzene, and cyclohex-
ane. In the temperature range of 260–380 C, the yield
of TH-DBT decreased from trace to zero with increas-
ing reaction temperature. Figure 6 shows the conver-
sion of DBT over Co–Mo bimetallic catalysts supported
over MCM-41 during hydrodesulfurization, compared
with Co/MCM-41 (CoO, 20 wt% ) and Mo/MCM-41
evaluated: BET specific surface area, 1046 m2/g; pore size,
3
˚
46 A; and pore volume, 1.00 cm /g.
The physicalpropertiesofthe prepared catalystsare sum-
marized in Table 1. It is shown that both BET specific sur-
face area and pore volume of catalysts were reduced after
TABLE 2
Main Reduction Temperature in the TPR Patterns
of MCM-41-Supported Catalysts
Catalyst
Main reduction temperature (K)
Co/MCM-41
Co–Mo(0.25)
Co–Mo(0.50)
Co–Mo(0.75)
Co–Mo(1.00)
Mo/MCM-41
773
817
828, 874
813, 862
849, 911, 1103
831
FIG. 4. N2 adsorption–desorption isotherms of the siliceous MCM-41.

SULFIDED Co–Mo CATALYSTS
23
FIG. 6. Variation of DBT conversion with temperature during HDS
over MCM-41-supported catalysts and a commercial deep HDS catalyst.
(᭹) Mo, (ᮀ) Co–Mo(0.25), (♦) Co–Mo(0.50), (᭺) Co–Mo(0.75), (4) Co–
Mo(1.00), (᭡) Co.
part of DBT was converted into TH-DBT at temperatures
below 280 C, but no TH-DBT was detected in the HDS
reaction products obtained above 280 C. Since HDS of gas
oil is generally carried out at temperatures >300 C and for
ease of data treatment the selectivity data were curtailed to
include only those in the temperature range of 280–360 C.
As shown in the Fig. 8, the selectivity for BP formation
changes a little for each catalyst and predominates in the
liquid products (70–80% ), indicating that HDS of DBT
over MCM-41-supported Co–Mo catalysts mainly takes the
route of hydrogenolysis. The selectivity of CHB decreases,
whereas the selectivity of benzene and cyclohexane in-
creases with increasing temperature, suggesting that high
temperature favors the hydrocracking of CHB and that the
FIG. 5. TPR patterns of MCM-41 and its supported catalysts. (a)
MCM-41, (b) Mo/MCM-41, (c) Co–Mo(0.25), (d) Co–Mo(0.50), (e) Co–
Mo(0.75), (f) Co–Mo(1.00), (g) Co/MCM-41.
(MoO3, 20 wt% ). It is shown that all the Co–Mo cata-
lysts showed very high activity for converting DBT into
hydrocarbons, yieldingalmost complete conversion ofDBT
at temperatures over 320 C. Moreover, Mo/MCM-41 also
showed relativelyhigh HDSactivityat high temperatures. It
is shown that the Co–Mo(0.75)/MCM-41 catalyst exhibited
the highest activity for DBT hydrodesulfurization. The op-
timal Co/Mo atomic ratio for MCM-41-supported catalysts
is different from that for conventional -Al2O3-supported
catalysts, which generally falls in the range of 0.2–0.5.
Figure 7 shows changes in HDS reaction rate with tem-
perature for this series of catalysts. It is indicated that higher
Co-loading favors the HDS reaction rate whereas HDS rate
reaches a maximum at a Co/Mo atomic ratio of 0.75.
The variations of HDS selectivity with temperature for
the Co–Mo catalysts are shown in Figs. 8a–8d. A small
FIG. 7. HDS reaction rates vs reaction temperature for Co–
Mo/MCM-41. (᭹) Mo, (ᮀ) Co–Mo(0.25), (♦) Co–Mo(0.50), (᭺) Co–
Mo(0.75), (4) Co–Mo(1.00).

24
WANG ET AL.
TABLE 3
Reaction Constants for HDS, HYD, HYG, and HYC during HDS
of DBT over Co–Mo Catalysts
Co–Mo Co–Mo Co–Mo Co–Mo
Catalysts
(0.25)
(0.50)
(0.75)
(1.00)
300 C
kHDS
kHYD
kHYG
kHYC
320 C
kHDS
kHYD
kHYG
kHYC
(
(
(
(
10 4 atom/nm2 s)
10 4 atom/nm2 s)
10 4 atom/nm2 s)
10 4 atom/nm2 s)
11.27
2.29
7.56
0.71
15.30
2.59
10.91
0.90
18.06
3.61
14.45
0.00
12.31
1.98
8.70
0.81
(
(
(
(
10 4 atom/nm2 s)
10 4 atom/nm2 s)
10 4 atom/nm2 s)
10 4 atom/nm2 s)
21.72
3.21
16.35
1.69
20.89
2.88
15.33
1.34
35.60
5.48
27.41
1.36
23.05
2.83
16.59
1.81
hydrocracking of CHB mainly accounts for the formation
of benzene and cyclohexane.
When the selectivities are plotted against HDS reaction
conversion, the selectivities did not change markedly until
the conversion exceeded 95% . At high conversions, the se-
lectivity of BP changed a little while the selectivity of CHB
decreased and the selectivity of benzene and cyclohexane
increased with increase of conversion.
The reaction constants of HDS, hydrogenolysis, hy-
drogenation, and hydrocracking at temperatures of 300
and 320 C are summarized in Table 3. It is shown that
hydrogenolysis reaction rate is the highest, one order of
magnitude higher than other reactions.
The ratios of CHB selectivity to BP selectivity (SCHB/SBP
)
are plotted against HDS reaction temperature for MCM-
41-supported Mo, Co, and Co–Mo catalysts, as shown in
Fig. 9. It is shown that Mo/MCM-41 and Co/MCM-41 give
FIG. 9. SCHB/SBP vs reaction temperature for Mo, Co, and Co–Mo
catalysts. (᭡) Mo, (᭹) Co, (᭺) Co–Mo(0.25), (♦) Co–Mo(0.50), (4) Co–
Mo(0.75), (ᮀ) Co–Mo(1.00).
FIG. 8. Variation of selectivities for BP (♦), CHB (᭺), and benzene
and cyclyhexane (4) with temperature during HDS of DBT. (a) Co–
Mo(0.25), (b) Co–Mo(0.50), (c) Co–Mo(0.75), (d) Co–Mo(1.00).

SULFIDED Co–Mo CATALYSTS
25
the highest and the lowest SCHB/SBP ratio and the ratios for
Co–Mo/MCM-41 catalysts fall between them.
The calculation of the Weisz modulus, 8, for Co–
Mo/MCM-41 catalysts in the temperature range of 240–
360 C revealed that all the Weisz modula were below 0.04,
indicating that intraparticle mass transfer is not the limit-
ing step in HDS reaction in the reaction temperature range.
Since the high space velocity during the reaction enhanced
the external mass transfer around the particle, the reac-
tion is not likely subject to mass transfer limitation. Conse-
quently, the surface reaction should be the limiting step in
the HDS reaction.
FIG. 11. Output profile of[35S]H2Swith time after a pulse input during
HDS reaction of DBT. (᭹) blank run, (᭡) HDS.
3.3. Sulfur Behavior during HDS by Isotope
Tracer Method
The investigation of sulfur behavior during HDS of
DBT was conducted over sulfided Co–Mo(0.50)/MCM-41
catalyst at 280 C and 5.0 MPa. Figure 10 shows a typical
profile of radioactivity with reaction time during the HDS
of DBT by the isotope tracer method. A decalin solution
of 1 wt% [32S]DBT was pumped into the reactor to run
the HDS reaction. After the conversion of [32S]DBT be-
came constant, a flow of 1 wt% [35S]DBT in decalin was
introduced into the reactor to replace the [32S]DBT solu-
tion. The radioactivity of the unreacted [35S]DBT in the
liquid product increased and approached a steady state im-
mediately after the introduction, whereas there was a de-
lay for the radioactivity of the released [35S]H2S to reach
a steady state. After both the radioactivity of the un-
reacted [35S]DBT and that of the released [35S]H2S be-
came constant, a flow of decalin switched to replace the
[35S]DBT solution. Both the radioactivity of unreacted
[35S]DBT and that of released [35S]H2S decreased imme-
diately. Little [35S]DBT was detected during the period of
purging with decalin and hydrogen, suggesting that there
is no absorbed H2S on the surface of the catalyst. Af-
ter a long period of purging, a flow of 1 wt% [32S]DBT
solution was introduced again to replace decalin. No
radioactivity change was detected in the liquid product, but
a peak of 35S radioactivity in the gas phase appeared in the
form of [35S]H2S.
The retarding effect of H2S on HDS was reported by
many researchers (13, 14). To investigate the behavior of
the sulfur atom during HDS in the presence of H2S, a pulse
H2S tracer method was developed to trace the dynamics
of the sulfur atom. The experiment was conducted over
the Co–Mo(0.50) catalyst at 320 C and 5.0 MPa. Figure 11
shows the comparison between the output profile of the
blank run and that of the HDS run. For the blank run,
there is a sharp peak in the output profile immediately after
the introduction of [35S]H2S. For the HDS run, there was
some delay for the radioactivity peak; moreover, the peak
was broadened and was tailing.
4. DISCUSSIONS
4.1. Experimental Results
It is generally accepted that small cobalt sulfide species
are located at the MoS2 edges in the active phase (15).
EXAFS measurements revealed that the Co sulfide species
located at the MoS2 edges differ locally in size and ordering
(16). It is believed that only the Co–Mo–S structure is active
in HDS. However, not all of the cobalt added to the for-
mulation results in the formation of the unique Co–Mo–S
species. As cobalt is added, it preferentially interacts with
the edge of the MoS2 crystallite surface, forming the desired
active species. When all of the edges are covered, cobalt
forms a separate phase of the stable sulfide Co9S8, which
has no contribution to desulfurization. In other words, there
is an optimal Co/Mo atomic ratio for every series of HDS
catalysts.
FIG. 10. Changes in radioactivities of unreacted [35S]DBT (᭹) and
released [35S]H2S (᭺) with reaction time.
Both the conversions of DBT and HDS rates for Co–
Mo/MCM-41 catalysts indicated that the optimal Co/Mo

26
WANG ET AL.
ratio is 0.75, higher than that of the conventional -Al2O3-
Kabe et al. (18) reported similar results for the -Al2O3-
supported catalysts. It is believed that a high dispersion of supported Mo-based HDS catalysts. Moreover, they found
MoS2 and the promoters on the surface of the supports is that, in addition to sulfur, other heteroatoms with lone pairs
essential to form a monolayer or tiny Co–Mo–S crystallite such as nitrogen or oxygen also lead to the release of sulfur
structures so as to achieve a very high HDS activity (1). The atoms on the active sites when introduced into the reac-
XRD patterns of MCM-41-supported catalysts reveal that tion system. If one considers the fact that Co–Mo nitrides
the active species are highly dispersed on the surface, even are active in desulfurization (19) while Co–Mo sulfides are
at a relatively high Mo loading. Since the support provides good catalysts for denitrogenation, it may be reasonable
enough space to accommodate the Co–Mo pairs, higher to assume that both sulfur and nitrogen atoms on the ac-
HDS activity is obtained when the Co–Mo molar ratio is tive sites are exchangeable and the exchange should be
close to unit. Under such circumstances, the metal loading an important step in completing the catalytic circle. Nev-
level becomes an important factor, since the loading will ertheless, the introduction of oxygen-containing molecules
determine the population of the species pairs. Therefore, a may considerably increase the period of the catalytic cir-
high metal loading may be favored for such a high surface cle because oxygen atoms have stronger affinity to those
support.
transition metal cations. Presulfidation of the HDS catalyst
The variation of BP selectivity with reaction temperature prior to reaction provides evidence for this assumption be-
shows that HDS of DBT over MCM-41-supported Co–Mo cause the presulfurization of Co–Mo oxides always takes
catalysts mainly takes the route of hydrogenolysis. As can several hours to completely replace oxygen atoms with sul-
be seen in Fig. 9, Co/MCM-41 seems to accelerate the hy- fur atoms. In addition, Viljava et al. (20) recently studied the
drogenolysis, and the introduction of Co to Mo-based cata- effect of alternating HDS and HDO (hydrodeoxygenation)
lysts enhances the direct extraction of sulfur atoms from the cycleson the activityofa commercialpresulfided Co–Mo/ -
DBT molecule. It is assumed that Co sulfide desulfurizes by Al2O3 catalyst by circling the feed between benzothiophene
extracting sulfur atoms directly from the sulfur-containing and phenol. They found that HDS conversion decreased
molecules, perhaps due to its electronegativity higher than after a HDO period, but the HDS activity of the cata-
that of the Mo atom.
lyst was slowly recovered with time on sulfur-containing
The variations of selectivities of CHB and of benzene feed.
and cyclohexane with temperature suggest that high tem-
The results of pulse isotope tracer experiments showed
perature favors the hydrocracking of CHB and the hy- that there was some delay of [35S]H2S output for the HDS
drocracking of CHB mainly accounts for the formation run compared with the blank one. The profile for the HDS
of benzene. The analysis of bond dissociation energies for run is an additional result contributed by the residence of
BP and CHB also supports the hypothesis that the hy- H2S in the reactor and the stay of H2S on the support. The
drocracking of DBT mainly takes the route of CHB to difference suggests that the H2S introduced during HDS
benzene and cyclohexane. With lower dissociation energy not only adsorbs on the surface of the support but also
(CHB, 389 kJ/mol; BP, 418 kJ/mol) (17), the C–C bond is involved in interacting with the active site. H2S, as one
bridging the benzene ring and the cyclohexane ring in the of the sulfur-containing molecules, competes with DBT in
CHB molecule is easier to be cleaved than that bridging attacking the active site on the surface, so as to reduce the
the two benzene rings in the BP structure. Provided that overall HDS reaction rate of DBT.
benzene and cyclohexane are mainly produced via CHB
cracking, the activation energies of the parallel reactions
for BP and CHB formation might be of the same order
4.2. Mechanism of HDS over Mo-Based Catalysts
of magnitude since their selectivities change little with the
temperature.
Despite the significant progress in HDS, many funda-
mental questions regarding the details of the catalytic
cycle and the promoter action are still being debated.
Many theories and models have been put forth and ar-
gued in the literature. Most of the models are based on
the rim–edge model, in which hydrogenation sites are cor-
related with rim sites and hydrogenolysis sites are cor-
related with edge sites (1). According to the rim–edge
model, the HDS selectivity results indicate that edge sites
are predominant on the Co–Mo/MCM-41 catalyst, suggest-
ing that the MoS2 stacked lamellar crystals exist in very
small crystals or in a monolayer structure. This conclusion
is in accordance with the fact that the extremely high sur-
face facilitates the high dispersion of the MoS2 and CoS
The 35S isotope tracer experiment (Fig. 10) shows that
the sulfur atoms extracted from DBT molecules are not re-
leased directly, but accommodated on the surface. Since the
35S accommodated on the catalyst cannot be released by H2
and decalin purging for a couple of hours, the retained 35S
does not exit in the form of adsorbed H2S on the surface.
On the other hand, these sulfur atoms could be released
by the introduction of new DBT molecules, indicating that
sulfur atom exchange between DBT molecules and the sur-
face of the catalyst may be involved during HDS of DBT
and the sulfur atoms on the surface may serve as active
sites.

SULFIDED Co–Mo CATALYSTS
27
species and the catalysts showed considerably high HDS The mechanism briefly describes the possible reaction steps
activity. involved on the local site. It is assumed that gaseous hydro-
In order to develop efficient deep HDS catalysts, much gen adsorbs dissociatively on the surface of catalysts and
attention has been focused on the theoretical aspects of the that the hydrogen species consumed in the reaction are sup-
active phase on the supports. It is commonly accepted that plemented by means of spillover on the surface (26).
the vacancy on the surface acts as HDS active sites. Kabe
During HDS over Co–Mo sulfided catalyst, the sulfur
et al. (11, 21, 22), based on the 35S tracer results similar to atom (S ) in the sulfur-containing molecules (H2S, thio-
those mentioned above, proposed that the S–C bonds in the phene, benzothiophene, dibenzothiophene, etc.) tends to
DBT molecule were cleaved by the occupation of DBT on approach the Co atom at the edge site by hydrogen bonding.
the vacancy and that H2S formation leads to the regenera- The Co–S bond, which bridges Co and Mo atoms, is weak-
tion of the new vacancy. Their model interpreted the phe- ened and then cleaved at the attack of a sulfur-containing
nomenon that the sulfur in DBT was not directly released as molecule. With the cleavage of the Co–S bond, a SH group
H2S but was initially accommodated on the surface of the is formed by reacting with adsorbed hydrogen species, and
catalysts. However, the vacancy model could not account a cationic Co is left on the surface. The two SH groups
for the 35S behavior shown in Fig. 8, because no [35S]H2S attached to the Mo atom are subsequently cleaved to re-
was formed in the absence of DBT even in a hydrogen at- lease the sulfur atom on the active site as H2S, and a double
mosphere while all the 35S atoms accommodated on the bond Mo S is formed. By reacting with adsorbed hydro-
==
surface were completely released as [35S]H2S on introduc- gen species, the double bond is broken to form a new SH
ing [32S]DBT. Therefore, more detailed reaction steps need group and a Mo cation. Both the cationic Mo and Co exert
to be clarified.
attractive forces on the sulfur atom in the sulfur-containing
Topsøe et al. (23) performed the DFT calculations in or- molecule, extracting the sulfur atom (S ) to bridge the Mo
der to find the minimum energy configuration for the MoS2 and Co atoms. Consequently, the chemical configuration on
and Co–Mo–S structures on HDS catalysts. They concluded the catalyst surface resumes, and a new H2 molecule from
that quite large surface reconstruction/relaxation must take the gas phase adsorbs on the surface to fill the vacant ad-
place during the catalytic cycle upon addition and removal sorption site left. When a new sulfur-containing molecule
of sulfur atoms. They have suggested that the many pre- approaches the active site, a new catalytic circle will begin.
vious models based on static vacancy structures should be
It may be possible to replace the sulfur atom with a
reconsidered (24). The DFT calculations (25) also revealed nitrogen atom to act as the active site. The prepared
that the most favored positions of the Co promoter atoms sulfided or nitrided Mo-based catalysts are active in both
are at the S edges, which is in accord with EXAFS results. HDS and HDN of oil fractions. In other words, HDS
A hydrogenolysis catalytic circle during HDS over sul- and HDN share the same type of active sites during the
fided Co–Mo catalysts is proposed as shown in Scheme. 1. hydrotreating process. However, the oxygen-containing
SCHEME 1. Mechanism of hydrodesulfurization of DBT over sulfided Co–Mo catalysts.

28
WANG ET AL.
SCHEME 2. Mechanism of hydrodesulfurization of DBT over sulfided Mo catalysts.
compounds or gaseous oxygen are harmful to HDS
In the liquid product obtained at temperatures over
and HDN, because the replacement of an oxygen 280 C, 70–80% BP wasgenerated and itsselectivitychanges
atom by a sulfur atom is much more difficult on in- a little with the increase in HDS reaction temperature.
troducing sulfur-containing or nitrogen-containing feed. The selectivity of CHB decreases whereas the selectivity of
benzene and cyclohexane increases when increasing the re-
action temperature. It seems that HDS of DBT over MCM-
41-supported Co–Mo sulfides includes mainly two paral-
lel pathways: hydrogenolysis and hydrogenation followed
by desulfurization. Benzene and cyclohexane in the liquid
products are mainly from the hydrocracking of CHB. It is
indicated that hydrogenolysis predominates in the HDS re-
actions of DBT.
It is worth noting that H2S, like other sulfur-containing
compounds, participates in the catalytic circle, as evidenced
by the broadening and tailing of the output peak in the
pulse tracer experiment during HDS. The competitive
adsorption of H2S results in a slowdown of the overall
HDS reaction rate.
Similarly, the HDS reaction over Mo/MCM-41 catalyst
takes the same steps. The cleavage and formation of bonds
take place between the edged Mo atom and its neigh-
boring Mo atom, as shown in Scheme 2. Since the Mo–S
bond strength is higher than the Co–S bond strength (27),
the Mo–S bond attached to the edged Mo atom is more
difficultly cleaved than the Co–S bond. Consequently, the
HDS reaction rate over Mo catalyst is lower than that over
Co-promoted Mo catalysts. It seems that the cleavage of
the Co–S bond or the Mo–S bond attached to the edged
Mo atom might be the rate-limiting step.
35S isotope tracer investigation revealed that sulfur atoms
retained on the surface could be released only by the intro-
duction of a sulfur-containing compound, indicating that
sulfur atom exchange between sulfur-containing molecules
and the active sites is involved in the HDS reaction. Accord-
ingly, a reaction mechanism for HDS is proposed, which is
in accordance with the isotope tracer results, the rim–edge
model, and the DFT calculations.
ACKNOWLEDGMENTS
5. CONCLUSION
This research was partly supported by the Natural Science Foundation
of China (20003002). The authors are grateful to Ms. D. Wang, Mr. X. Li,
Mr. D. Han, Mr. S. Otsuki, Mr. H. Shirai, Mr. H. Sukuno, Mr. Putu, and
Mr. Saito for their kind help with the research work. They also thank the
Chinese and Japanese governments for financial support in carrying out
the research program.
In summary, siliceous MCM-41 is a promising support
for preparing deep hydrodesulfurization catalysts. The
optimal atomic ratio for Co–Mo/MCM-41 catalysts is 0.75,
higher than that of the -Al2O3-supported catalysts. It
is assumed that Co–Mo pairs have been created on the
extremely high surface area, yielding considerably high
activity for HDS of DBT.
REFERENCES
1. Whitehurst, D. D., Isoda, T., and Mochida, I., Adv. Catal. 42, 345(1998).

SULFIDED Co–Mo CATALYSTS
29
2. Beck, J. S., Vartuli, J. C., Roth, W. J., Leonowicz, M. E., Kresge, C. T.,
Schmitt, K. D., Chu, C. T.-W., Olson, D. H., Sheppard, E. W., McCullen,
S. B., Higgins, J. B., and Schlenker, J. L., J. Am. Chem. Soc. 114, 10,834
(1992).
15. Topsøe, N. Y., and Topsøe, H., J. Catal 84, 386 (1983).
16. Bouwens, S. M. A. M., Van Zon, F. B. M., Wan Dijk, M. P., Van der
Kraan, A. M., De Beer, V. H. J., Van Veen, J. A. R., and Koningsberger,
D. C., J. Catal. 146, 375 (1994).
3. Ying, J. Y., Mehnert, C. P., and Wong, M. S., Angew. Chem. Int. Ed. 36,
56 (1999).
17. Dean, J. A., “Handbook of Organic Chemistry,” pp. 3–18. McGraw–
Hill, New York, 1987.
4. Corma, A., Mart´ınez, A., Mart´ınez-Soria V., and Monto´n, J. B., J. Catal.
153, 25 (1995).
5. Reddy, K. M., Wei, B., and Song, C., Catal. Today 43(3–4), 261
(1998).
6. Song, C. S., and Reddy, K. M., Appl. Catal. A 176(1), 1 (1999).
7. Ram´ıre, I., Contreras, R., Castillo, P., Klimova, T., Za´rate, R., and
Luna, R., Appl. Catal. A: Gen. 197, 69 (2000).
8. Wang, A., and Kabe, T., Chem. Commun. 20, 2067 (1999).
9. Gilman, H., and Jacoby, A. L., J. Org. Chem. 4, 108 (1938).
10. Scheffer, B., Mangnus P. J., and Moulijn, J. A., J. Catal. 121, 19
(1990).
11. Qian, W., Ishihara, A., Okoshi, Y., Nakakami, W., Godo, M., and
Kabe, T., J. Chem. Soc. Faraday Trans. 93(24), 4395 (1997).
12. Arnoldy, P., Franken, M. C., Scheffer, B., and Moulijn, J. A., J. Catal.
96, 381 (1985).
13. Broderick, D. H., and Gates, B. C., AIChE J. 27, 663 (1981).
14. Ledoux, M. J., Huu, C. P., Segura, Y., and Luck, F., J. Catal. 121, 70
(1990).
18. Kabe, T., Qian, W., and Ishihara, A., J. Phys. Chem. 98(3), 912 (1994).
19. Nagai, M., Koyama, H., Sakamoto, S., and Omi, S., Stud. Surf. Sci.
Catal. 127, 195 (1999).
20. Viljava, T., Komulainen, S., Selvam, T., and Krause, A., Stud. Surf. Sci.
Catal. 127, 145 (1999).
21. Kabe, T., Qian, W., and Ishihara, A., J. Catal. 149, 171 (1994).
22. Qian, W., Ishihara, A., Wang, G., Tsuzuki, T., Godo, M., and Kabe, T.,
J. Catal. 170, 286 (1997).
23. Topsøe, H., Knudsen, K. G., Byskov, L. S., Nørskov, J. K., and Clausen,
B. S., Stud. Surf. Sci. Catal. 121, 13 (1999).
24. Knudsen, K. G., Cooper, B. H., and Topsøe, H., Appl. Catal. A: Gen.
189, 205 (1999).
25. Byskov, L. S., Nørskov, J. K., Clausen, B. S., and Topsøe, H., J. Catal.
187, 109 (1999).
26. Hensen, E. J. M., Lardinois, H. J., de Beer, V. H. J., van Veen J. A. R.,
and van Santen, R. A., J. Catal. 187, 95 (1999).
27. Byskov, L. S., Hammer, B., Nørskov, J. K., Clausen, B. S., and Topsøe,
H., Prepr. Am. Chem. Soc. Div. Pet. Chem 42(1), 113 (1997).