Hydrodesulfurization of methyl-substituted dibenzothiophenes: Fundamental study of routes to deep desulfurization

Article abstract of DOI:10.1006/jcat.1996.0083

Four catalysts, Co-Mo and Ni-Mo, designed to cover a wide range of activity for aromatic ring hydrogenation and two catalysts, zeolite-Co-Mo, with different cracking activities were tested, in the hydrodesulfurization of dibenzothiophene (DBT) and 4,6-dimethyl-DBT (DMDBT) in a fixed-bed reactor at 360°C and a hydrogen pressure of 5.4 MPa. Increasing the hydrogenation activity of the catalysts increased the mole ratio, α, of cyclohexylbenzenes to biphenyls produced from both reactants. Although α had little effect on the hydrodesulfurization (HDS) rate of DBT, the HDS rate of DMDBT increased significantly with α, approching the same level as for DBT at α=2. Introduction of zeolite H-ZSM-5 to the Co-Mo -Al catalyst increased the HDS rate of DBT at a lower value of α and decreased that of DMDBT. No cracking reactions of DBT and DMDBT were detected. A Co-Mo-Al catalyst containing HY zeolite displayed significant cracking activity of DMDBT defined by two routes: demethylation of benzenic rings and scission of the C-C bond connecting the benzenic rings. It resulted in increasing HDS rates of DMDBT by about threefold relative to the Co-Mo-Al catalyst, yielding toluene and benzene, at a ratio of about 3, as the main HDS products. About 80% of the DMDBT desulfurized with HY-Co-Mo-Al catalyst was converted through "cracking" intermediates, 90% of those intermediates being produced through the scission of the C-C bond connecting the benzenic rings.

Full text of DOI:10.1006/jcat.1996.0083

JOURNAL OF CATALYSIS 159, 236–245 (1996)
ARTICLE NO. 0083
Hydrodesulfurization of Methyl-Substituted Dibenzothiophenes:
Fundamental Study of Routes to Deep Desulfurization
1
M. V. Landau, D. Berger, and M. Herskowitz
Blechner Research Center for Industrial Catalysis and Process Development, Ben-Gurion University of the Negev, Beer-Sheva, Israel 84105
Received June 30,1995; revised September 26, 1995; accepted October 13, 1995
spheric gas oils remove the DBT sulfur at normal operating
�
Four catalysts, Co–Mo and Ni–Mo, designed to cover a wide conditions in hydrodesulfurizers (350–360 C) (5–8). Deep
range of activity for aromatic ring hydrogenation and two cata- desulfurization requires new catalytic routes for efficient
lysts, zeolite-Co–Mo, with different cracking activities were tested,
desulfurization of methylated DBT.
It has been proved (9) that methyl substituents in the 4-
and 6-positions of DBT increase its chemisorption through
the � electrons of the aromatic rings in flat mode parallel
in the hydrodesulfurization of dibenzothiophene (DBT) and 4,6-
�
dimethyl-DBT (DMDBT) in a fixed-bed reactor at 360 C and a hy-
drogen pressure of 5.4 MPa. Increasing the hydrogenation activity
of the catalysts increased the mole ratio, � , of cyclohexylbenzenes to
to the catalyst surface. It was supposed (9) that they hinder
biphenyls produced from both reactants. Although � had little ef-
fect on the hydrodesulfurization (HDS) rate of DBT, the HDS rate of
DMDBT increased significantly with � , approching the same level
sterically the DBT adsorption on the catalyst surface by
a sulfur-bonded mode preventing the C–S bond scission.
However, according to theoretical calculations (10) both
as for DBT at � =2. Introduction of zeolite H-ZSM-5 to the Co–Mo
Al catalyst increased the HDS rate of DBT at a lower value of DBT and methylated DBT adsorb flat on the catalyst sur-
and decreased that of DMDBT. No cracking reactions of DBT face. Hence the decrease in reactivity of the DBT molecule
–
�
and DMDBT were detected. A Co–Mo–Al catalyst containing HY in the presence of methyl groups in the 4- and 6-positions
zeolite displayed significant cracking activity of DMDBT defined
by two routes: demethylation of benzenic rings and scission of the
C–C bond connecting the benzenic rings. It resulted in increasing
HDS rates of DMDBT by about threefold relative to the Co–Mo–Al
hydrogenation of benzenic rings, demethylation, scission
catalyst, yielding toluene and benzene, at a ratio of about 3, as the
of the single C–C bond in the thiophenic ring, and isomer-
main HDS products. About 80% of the DMDBT desulfurized with
HY–Co–Mo–Al catalyst was converted through “cracking” inter-
mediates, 90% of those intermediates being produced through the
could be a result of hindering the access of hydrogen to
the C–S–C bridge after flat adsorption. Potential catalytic
routes that involve some intermediate transformations like
ization of methyl groups from 4,6- to 1,9-positions could
activate the methylated DBT.
Several studies have been published recently in this field.
Kabe et al. measured the product distribution in HDS of
DBT, 4-Me-DBT, and 4,6-DMDBT with a Co–Mo–Al cat-
alyst (9). They found that the desulfurization route by
intermediate hydrogenation of the aromatic ring was not
inhibited by addition of methyl substituents to DBT. How-
ever, only 10–15% of desulfurized methylated DBT was
converted through the hydrogenation route, probably be-
cause of a low hydrogenation activity of the catalyst.
Takaaki et al. (11) reported that, using a sulfided Co–Mo–
Al catalyst, the contribution of the hydrogenation route in
the HDS of DMDBT, 4-Me-DBT, and DBT was 94, 37,
and 12%, respectively. That could be the reason for the
higher activity of Ni–Mo–Al compared with Co–Mo–Al in
theHDSofmethylatedDBTreportedbyMaetal. (7). How-
ever, thereisstillnoevidencethatincreasingthecatalysthy-
drogenation activity accelerates the DMDBT HDS. Other
routes, intermediate isomerization of substituted DBT and
demethylation as a result of reaction with polyaromatic
molecules, catalyzed by zeolites Y and ZSM-5, were tested
scission of the C–C bond connecting the benzenic rings. �c 1996
Academic Press, Inc.
INTRODUCTION
A growing interest in the deep desulfurization of diesel
fuels has been recorded as a result of new regula-
tions designed to limit the sulfur content to 0.05 wt%
(1, 2). 4-Methyldibenzothiophene (4-MeDBT) and 4,6-
dimethyldibenzothiophene (DMDBT) werereported to re-
act at a low rate in the deep hydrodesulfurization (HDS)
of middle distillate feedstocks such as light oil (3–5). Their
HDS rates on standard Co–Mo–Al hydrodesulfurization
catalysts are lower by a factor of 10–15 compared with
dibenzothiophene (DBT) (6, 7) which translates into high
�
temperatures (more than 380 C) for this process (5). Cur-
rentCo(Ni)–Mo–AlcommercialcatalystsforHDSofatmo-
1
To whom correspondence should be addressed.
236
0021-9517/96 $18.00
Copyright �c 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

HYDRODESULFURIZATION OF METHYLDIBENZOTHIOPHENES
237
by Rollmann et al. (12). No information was given about the isotherms. TEM micrographs were produced on a JEOL
effectiveness of those routes in HDS of methyl-substituted Co. microscope 2000-FX, 200 kW.
DBT.
The purpose of this study is to investigate two routes for Preparation of DMDBT and Reaction Solvent
desulfurization, intermediate hydrogenation and cracking
of DMDBT, to enhance the rate of the HDS reaction.
DMDBT was synthesized by alkylation of DBT (Merck,
8
20409) with dimethylsulfate (Merck, 803071) in the pres-
�
ence of butyl lithium (Aldrich, 109-72-8) at 0 C according
to a published method (16). Fractional crystallization with
hot methanol was used for DMDBT separation. The prod-
METHODS
�
uct, with a melting point of 151 C, was obtained at about
Catalyst Preparation and Characterization
2
5% yield.
Six catalysts were used in this study, commercial Co–Mo–
Al (I) and Ni–Mo–Al (II) and four proprietary catalysts.
Ni–Mo (III) (13) and Ni–Mo–Si (IV) (14) were prepared
by coprecipitation. NiMoO4 was precipitated from a mixed
water solution of nickel nitrate and ammonium paramolyb-
date (concentrations of NiO and MoO3 were both 80 g/
liter) by addition of ammonium hydroxide at pH 6.84 and
The solvent nature had a significant effect on the ki-
netics of DBT HDS because of the competitive adsorp-
tion of its components on the catalyst surface (17). There-
fore, a model solvent that contained a mixture of paraf-
finic, naphthenic, and aromatic compounds was prepared.
A computer simulation showed that a mixture containing
2
0% n-decane, 50% n-octadecane, and 30% tetralin yielded
�
temperature 80 C. The same technique was used for precip-
physical properties and boiling point distribution similar
to those of gas oil. n-Decane and tetralin were purchased
from Merck (820383, 809733) and n-octadecane from the
Humphrey Chem. Co. Inc. (A-18/140/PL). Mixtures were
prepared containing 1 wt% DBT or DMDBT, which corre-
spond to 0.17 or 0.15 wt% sulfur, respectively, as normally
found in petroleum atmospheric gas oils (3, 4).
itation of NiMoO4 on the silica support; following Ref. (14)
the silica powder was added to the solution of Ni and Mo
salts before introducing the precipitation agent. Both bulk
and supported NiMoO4 catalysts III and IV were dried at
�
3
93 K and calcined in air at 300 C, a procedure which gives
the highest hydrogenation activity of the catalyst after sulfi-
dation (13). Co–Mo–HY–Al (V) and Co–Mo–HZSM-5–Al
(VI) were prepared by impregnation of a zeolite–alumina Kinetic Measurements
support with cobalt nitrate, ammonium paramolybdate in
ammonia water solution. The zeolite–alumina support was
made by coextrusion of aluminum hydroxide (prepared by
a pH oscillating precipitation method from an aqueous so-
lution of aluminum nitrate with ammonium hydroxide (15))
with zeolite powders. All salts were purchased from Fluka.
Zeolite HY LZ-Y62 was manufactured by Linde AG and
HZSM-5 (CBV-3020) by PQ Corp. The extrudate diameter
of catalysts I, II, V, and VI was 1.3–1.5 mm. Coprecipitated
powder oxide catalysts III and IV were pelletized by press-
ing and crushing to a fraction of 0.5–1.0 mm. Before test-
ing, all the catalysts were sulfided by excess of elemental
sulfur in hydrogen flow. Elemental sulfur particles (2 g, 20–
The kinetic measurements were carried out in a high-
pressure minipilot plant, automatically controlled and de-
signed for unattended continuous runs. A detailed descrip-
tion of the system, including the hardware and software, is
given elsewhere (18). A quantity of 5 g of catalyst mixed
3
with 10 cm silica was loaded into the isothermal zone of
3
the reactor between two layers of 10 cm silica. Prior to
performing the kinetic measurements all catalysts were sul-
fided. The sulfidation procedure included pretreating with
elemental sulfur, as described before, then stabilization in
the reactor for 24 h with a 2 wt% dimethylsulfide–toluene
�
1
�
mixture pumped at LHSV = 3 h , 320 C, hydrogen pres-
3
3
sure of 5.4 MPa, and hydrogen to liquid ratio 500 Nm /m .
All the kinetic measurements were carried out at 360 C,
5
0 mesh fraction) were loaded in a tubular reactor above a
�
3
layer of 5 cm of catalyst particles. The reactor was purged
with hydrogen at 200 ml/min at room temperature for 0.5 h
hydrogen pressure of 5.4 MPa, and hydrogen to liquid ra-
3
3
� 1
tio 500 Nm /m . LHSV was varied from 10 to 200 h . The
liquid product was sampled and analyzed by GC or GC-MS.
and then the temperature was increased in three steps, 120–
�
1
80–240 C, 1 h at every step.
The composition of the catalysts was measured by EDAX
microscope JEM-35, JEOL Co., link system AN-1000,
Product Analysis
(
Si–Li detector). The surface area and pore size distribution
The liquid products were analyzed by GC (Chrompack
were measured by BET (ASTM 3663-84) and Hg porosime- 9001), equipped with a flame ionization detector and CP-
try (Amin Co. porosimeter, ASTM 4284-83), respectively. Sil-5 CB capillary column, 10-m long and 0.25-mm i.d. The
The pore volume including micropores for the oxide cata- raw data were processed by the Mosaic GC control soft-
�
lysts was estimated on the basis of water adsorption at 20 C. ware. GC-MS analyses (GC-HP-5890, MSD-5971A) were
TheporevolumeofpeletizedsulfidecatalystsIIIandIVwas performed with the HP-1 capillary column (12.5-m long,
measured on the basis of nitrogen adsorption–desorption 0.2-mm i.d.) to identify and quantify by-products.

2
38
LANDAU, BERGER, AND HERSKOWITZ
TABLE 1
Characteristics of Catalysts in Oxide Form
V
VI
I
II
III
Ni–Mo
IV
Ni–Mo–Si
Co–Mo–
HY–Al
Co–Mo–
H-ZSM-5–Al
Catalyst
Co–Mo–Al
Ni–Mo–Al
Chemical composition, wt%
CoO
NiO
MoO3
SiO2
4.2
—
25.1
—
—
3.4
24.0
4.1
—
34.0
66.0
—
—
30.2
58.8
10.1
3.8
—
20.9
23.6
4.5
—
24.8
32.5
2
Surface area (m /g)
230
0.43
0.94
140
0.30
0.95
15
71
308
0.52
0.63
243
0.34
0.58
3
0.10a
1.30
0.15a
1.25
Pore volume (cm /g)
Bulk density (g/cm )
a
3
a
In sulfide form.
RESULTS
fact that its surface area is smaller by an order of magni-
tude. This is attributed to the inhibition of hydrogenation
activity caused by the amphoteric nature of alumina (20).
Hydrogenation Route
The potential of the hydrogenation route to enhance the Silica supports, in contrast to alumina, do not reduce the
rate of 4,6-DMDBT desulfurization was measured with a hydrogenation activity of sulfided NiMoO4, thus yielding a
set of catalysts that covered an activity range in the hydro- more active catalyst as a result of a higher surface area of
genation of aromatic rings. The composition, surface area, the active phase (14).
and pore volume of the catalysts (I–IV) in the oxide form
are shown in Table 1. Ni–Mo–Al catalysts normally have a catalyst IV calcined at 300 and 900 C. After calcination
higher hydrogenation activity for aromatics than Co–Mo– at 300 C the unsupported NiMoO4 forms long cylindrical
Figure 1 presents the TEM micrographs of Ni–Mo–Si
�
�
Al (19). Bulk sulfided NiMoO4 salt is expected to display amorphous (13) particles about 100 nm in diameter but
a higher activity than the alumina-supported Ni–Mo cat- within the pores of silica particles this phase could not be
alysts in aromatics hydrogenation (20, 21), in spite of the identified on the micrograph, probably as a result of its
�
�
FIG. 1. TEM micrographs of bulk and silica-supported Ni–Mo oxide catalysts calcined at (a) 300 C and (b) 900 C.

HYDRODESULFURIZATION OF METHYLDIBENZOTHIOPHENES
239
very small dimensions. To make it clear we calcined cata-
�
lyst IV at 900 C. The unsupported NiMoO4 that was stable
at this temperature according to XRD data was converted
to rounded sintered crystals more than 500 nm in diameter
while within the pores of silica particles the same crystals
appeared at about 100 nm in diameter. This proves that a
�
part of the NiMoO4 phase in this catalyst calcined at 300 C
was indeed encapsulated in pores of the silica support, pre-
venting its sintering at elevated temperatures. Thus it pro-
duced a higher surface area of the active component com-
pared to unsupported Ni–Mo catalyst III.
Testing the efficiency of the hydrogenation route in
DMDBT HDS considered first the activity of catalysts
I–IV in the saturation of aromatic rings. The correspond-
ing tetra- and hexahydro-DBTs are unstable, reacting fur-
ther in desulfurization reactions. Therefore the desulfur-
ization of DBT and DMDBT yielded two main products
in the liquid phase: biphenyl (BPH) and cyclohexylben-
zene (CHB), dimethyl-biphenyl (DMBPH) and dimethyl-
cyclohexylbenzene (DMCHB), respectively. The aromatic
hydrogenation activity of the catalysts in this particular case
could be characterized by the ratio of the conversion to the
two products, termed the selectivity parameter �. For all
catalysts this selectivity parameter increases with increas-
ing DBT or DMDBT conversion as a result of changing the
spacing time as illustrated by the plots in Fig. 2. The effect of
conversion on the � values was insignificant compared with
the differences in selectivity parameter among the tested
catalysts. A comparison of the catalysts at about 80% con-
version level (Fig. 2) indicates that the aromatic hydrogena-
tion activity increases significantly from catalyst I to catalyst
IV. Furthermore, the � value is higher for DMDBT than for
DBT, as a result of higher DMDBT reactivity in aromatic
ring hydrogenation, in agreement with results reported by
Houalla et al. (6).
The rate of sulfur removal was assumed to be pseudo-
first-order, as proposed by Houalla et al. (6). A semilog
plot of the sulfur conversion as a function of the space time
in Fig. 3 yielded a good fit of the data. The rate constants
calculated from the slope of the plots are listed in Table 2.
The rate constants were expressed on the basis of catalyst
mass, volume, and surface area to illustrate the intrinsic
activity of the catalysts. As expected, the mass or volume
rate constants for the Co–Mo–Al catalyst were higher than
for Ni–Mo–Al and bulk Ni–Mo catalysts in the HDS of
DBT.
FIG. 2. Effect of DBT and DMDBT conversion with catalysts I–IV
on selectivity parameter �.
The ratio of the intrinsic activity of HDS of DMDBT
and DBT, represented by the ratio kDMDBT/kDBT, increases
sharply with �DBT, representing the relative aromatic hy-
drogenation activity of the catalysts, as illustrated in Fig. 4.
�
DBT was estimated from Fig. 2 at 80% conversion. As �DBT
reaches a value of about 2, the HDS rates of the two reac-
tants are about the same. This means that the rate of hy-
drogenation of the aromatic ring in DMDBT preceding the
HDS reaction enhanced significantly its HDS rate.
Cracking Route
The intrinsic activity reflected by the rate constants on a
The efficiency of the cracking activation route in the
surfacebasis(ks)indicatesthehigheractivityofthebulkNi– desulfurization of DMDBT was tested on two Co–Mo–Al
Mo catalysts. The rate constants kw or kv are normally used catalysts, V and VI, containing zeolites HY and H-ZSM-5.
for comparison of catalysts in fixed-bed reactors reflecting Their composition, surface area, and pore volume are listed
the differences in %HDS. The HDS activity increased with in Table 1. The pore volume was higher compared with the
increasinghydrogenationactivityfromcatalystIItocatalyst basic Co–Mo–Al catalyst due to the pore volume of the
IV. The bulk catalysts III and IV were significantly more zeolites. However, the mesopore volume was reduced, al-
active than Co–Mo–Al in the case of DMDBT.
though the pore size distribution did not change as a result

2
40
LANDAU, BERGER, AND HERSKOWITZ
FIG. 3. Pseudo-first-order kinetics of hydrodesulfurization of DBT and DMDBT with catalysts I–IV.
of zeolite introduction, as illustrated in Fig. 5. The kinetic ing first-order reaction. No products other than BPH or
measurements with catalysts V and VI were carried out DMBPH and CHB or DMCHB were detected with cata-
over a period of 2 h to avoid deactivation (22).
lyst VI (Co–Mo–HZSM-5–Al). The presence of H-ZSM-5
Figure 6 presents the rate constants of DBT and DMDBT zeolite increased the HDS rate of DBT by a factor of about
HDS calculated for the three catalysts I, V, and VI, assum- 1.5, lowering its � value from 0.15 (catalyst I) to 0.06 (cata-
lyst VI). In contrast, the HDS rate of DMDBT was reduced
by a factor of about 1.3 with no significant change in the �
value.
TABLE 2
Rate Constants of Hydrodesulfurization of DBT and DMDBT
with Catalysts I–IV
The positive effect of zeolite H-ZSM-5 on the DBT HDS
ratecouldbearesultofthedirectactionofthezeolitestrong
acidic protons as demonstrated in Refs. (22, 23) for HDS
of thiophene with Ni(Co)–Mo–(Ca)–HY catalysts. The
lower � value measured with catalyst VI indicates that the
H-ZSM-5 zeolite selectively increased the rate of BPH pro-
duction that proceeded via DBT adsorption on Brønsted
acid sites followed by removal of H2S via �-elimination
(24). The negative effect of H-ZSM-5 zeolite on DMDBT
HDS could be explained considering the higher molecu-
lar dimensions compared with the DBT molecule that led
to significantly lower diffusion rates in the zeolite channels.
Catalyst
Co–Mo–Al
Ni–Mo–Al
Ni–Mo
Ni–Mo–Si
DBT
kw, (cm � g � h� 1
3
� 1
)
96.1
90.4
0.42
40.3
38.5
0.29
61.8
80.4
4.1
55.0
71.5
0.77
�
1
kv, (h
)
ks, (cm � m � h� 1
3
� 2
)
DMDBT
kw, (cm � g � h� 1
3
� 1
)
22.1
20.8
0.096
10.3
9.9
30.0
39.1
2.0
48.1
62.5
0.68
�
1
kv, (h
)
ks, (cm � m � h� 1
3
� 2
)
0.073

HYDRODESULFURIZATION OF METHYLDIBENZOTHIOPHENES
241
FIG. 4. Correlation between DMDBT/DBT HDS rates ratio and cat-
alysts hydrogenation activity (�DBT).
FIG. 5. Pore size distribution in Co–Mo-based catalysts.
Actually, the zeolite component, being inactive in DMDBT
HDS, acted as an inert diluent for the active alumina sup-
port and thus hindered the formation of Co–Mo–Al–oxide
precursors.
Introduction of acidic HY zeolite (catalyst V) increased
the HDS rates of both DBT and DMDBT by factors of
1
.8 and 3.0, respectively (Fig. 6 The product distribution
of DMDBT HDS with catalyst V was different from those
�
1
obtained with all other catalysts (Table 3, LHSV = 50 h ).
The main product was toluene at about 50% and then ben-
zene at about 20%, both being the result of scission of the
C–C bond connecting the benzenic rings in the DMDBT
molecule.
The higher HDS rate of DMDBT activated by cracking
of the C–C bond in the thiophenic ring compared to
demethylation was confirmed by the absence of dimethyl-
phenyl sulfide and diphenyl sulfide in the products,
while DMDBT demethylation products (methyldiben-
zothiophene, tetrahydro-methyl-dibenzothiophene, and
tetrahydro-dibenzothiophene) were detected (Table 3). No
DMCHB was detected in the DMDBT HDS products ob-
tained with catalyst V as a result of its high reactivity in
cracking reactions producing methylcyclohexane and alkyl-
toluenes. In the case of DBT, substantial scission of the C–C
FIG. 6. Effect of zeolite addition on HDS rates of DBT and DMDBT
bond connecting the benzenic rings was also observed with with Co–Mo–Al catalyst.

2
42
LANDAU, BERGER, AND HERSKOWITZ
TABLE 3
DMDBT Conversion Products with Co–Mo–HY–Al Catalyst
Products
wt%
19.9
Benzene
Toluene
47.5
Methylcyclohexane
Methylethylbenzene
5.9
3.0
Methylpropylbenzene
Biphenyl
8.5
1.0
Methylbiphenyl
2.1
Dimethylbiphenyl
6.5
1.1
Tetrahydrodibenzothiophene
Methyldibenzothiophene
2.7
0.3
0.7
Tetrahydromethyldibenzothiophene
Tetrahydrodimethyldibenzothiophene
Co–Mo–Al–HY catalyst, producing benzene at about 30% dimethylphenyl sulfide which was not detected as a result of
of the products. No CHB was detected in the DBT HDS its high reactivity. The Co–Mo–Al catalyst displayed a low
products obtained with catalyst V as a result of its cracking activity in DMDBT hydrogenation (Fig. 3), needed for pro-
producing alkylbenzenes.
duction of those compounds by the second route, through
Figure 7 presents the scheme of DMDBT molecule trans- tetrahydro-DMDBT, which was found in the products. This
formations under HDS conditions with the Co–Mo–Al–HY last compound could also give alkyltoluenes as a result of
catalyst. The rate constants shown in Fig. 7 were calculated hydrocracking. The same arguments led to the conclusion
on the basis of the product mole fractions assuming pseudo- that benzene was mainly produced as a result of diphenyl
first-order kinetics in all the routes of DMDBT conversion. sulfide HDS rather than through the second three-stage
Intermediates that were not found in DMDBT HDS prod- route (Fig. 7), including a slow hydrogenation step. Com-
ucts are shown in parentheses. Toluene and methylcyclo- paring the values of the estimated rate constants with the
hexane were supposed to be mainly the HDS products of measured overall HDS rate constant for DMDBT showed

HYDRODESULFURIZATION OF METHYLDIBENZOTHIOPHENES
243
�
� 1
FIG. 7. Conversion of DMDBT in the presence of sulfided Co–Mo–Al–HY catalyst at 360 C, 5.4 MPa, LHSV = 50 h : the reaction network. The
�
1
numbers next to the arrows are pseudo-first-order rate constants (kv, h ).
that about 80% of the desulfurized DMDBT was converted the S atom is screened by methyl groups. The product of
through intermediates yielded from “cracking” reactions DMDBTpartialhydrogenation, 4,6-diMe-hexahydro-DBT
of which about 70% was converted through scission of the (HHDMDBT), contains a methyl group, located at an ax-
C–C bond connecting the benzenic rings.
ial position to the cyclohexyl ring, that reduces the steric
hindrance at the S–C bond scission making all the atoms in
the C–S–C bridge equally accessible to the catalyst surface.
Therefore HHDMDBT is expected to be more active than
DISCUSSION
Increasing the hydrogenation activity of Co(Ni)–Mo sul- DMDBT in HDS in spite of the presence of the methyl
fide catalysts enhances the HDS rate of DMDBT. The com- group connected to the second aromatic ring.
puter simulation of DMDBT configuration (Serena soft-
The HDS rate of DBT with Ni–Mo catalysts employed
ware, PC model PI 89.0) illustrated in Fig. 8 shows that in this study, especially Ni–Mo–Al, was lower in compari-

2
44
LANDAU, BERGER, AND HERSKOWITZ
FIG. 8. Computer simulations of the DMDBT molecule structure before and after partial hydrogenation.
son with Co–Mo–Al employed in industrial desulfurization tural elements of sulfide phases on the catalyst surface,
processes (Table 2). Different active sites on the surface namely corners and edges of the Mo(W)S2 crystals (26),
of sulfided Co(Ni)–Mo(W) catalysts perform as catalytic bulk Ni–Mo particles and the Ni–Mo–Al monolayer (21,
centers for aromatic ring saturation and S–C bond scission 27), and rim and edge parts of MoS2 crystal edge planes
reactions (21, 25). It appears that an effective catalyst for (28). Optimization of such structures on the catalyst at a
deep desulfurization of petroleum gas oils has to contain proper combination could lead to improved catalysts for
an optimal concentration and ratio of hydrogenation and deep desulfurization.
desulfurization active sites. This would provide effective
This study confirmed the high efficiency of the crack-
hydrogenation of aromatic rings in 4,6-DMDBT and other ing activation route for desulfurization of DMDBT. Addi-
low reactivity substituted dibenzothiophenes as well as ef- tion of strong acidic components in Co–Mo–Al catalysts
fective desulfurization of high reactivity compounds like enhanced the demethylation the DMDBT and scission of
sulfides, thiophenes, and benzothiophenes. The nature of the single C–C bond in the thiophenic ring. Demethylation
those active sites is still not well defined (21, 25) rendering of DMDBT or scission of the single C–C bond in the thio-
the formulation of direct quantitative criteria difficult.
phenic ring would increase the HDS rate of reaction as a
The data obtained in this study provide the basis for the result of removing the screening effect of methyl groups, as
design of an optimal catalyst for deep desulfurization which illustrated in Fig. 9. In the second case the screening effect
combines desulfurization activity at the level of commercial would be removed as a result of free rotation of two methyl
Co–Mo–Al catalysts and hydrogenation activity at the level phenyl groups around the C–S–C bridge. The preparation
of � for DBT of about 2.
of HDS catalysts which contain acid sites could be a so-
Some published data indicate that hydrogenation and lution to deep desulfurization. However, zeolite catalysts
desulfurization active sites are related to different struc- are deactivated at a high rate as a result of coking (22) and

HYDRODESULFURIZATION OF METHYLDIBENZOTHIOPHENES
245
5. Ishihara, A., Tajima, H., and Kabe, T., Chem. Lett. 669 (1992).
6. Houalla, M., Broderic, D. H., Sapre, A. V., Nag, N. K., de Beer, V. H.
J., Gates, B. C., and Kwart, H., J. Catal. 61, 523 (1980).
7
8
9
. Ma, X., Sakanishi, K., and Mochida, I., Ind. Eng. Chem. Res. 33, 218
(
1994).
. Ma, X., Sakanishi, K., Isoda, T., and Mochida, I., Ind. Eng. Chem. Res.
4, 748 (1995).
. Kabe, T., Ishihara, A., and Zhang, Q., Appl. Catal. A 97, L1 (1993).
3
1
1
1
0. Lamure-Meille, V., Schulz, E., Lemaire, M., and Vrinat, M., Appl.
Catal. 131, 143 (1995).
1. Takaaki, I., Xiaoliang, X., and Isao, M., Sekiyu Gakkaishi 37, 368
(
1994).
2. Rollmann, L. D., Howley, P. A., Mazzone, D. N., and Timken, H. K.
C., AIChE 1994 Spring National Meeting, Atlanta, April 17–21, 1994,
Paper No. 53d.
1
1
3. Landau, M. V., Alexeenko, L. N., Vinogradova, O. V., Michailov, V. I.,
and Nefedov, B. K., Kinet. Katal. 30, 933 (1989). [Russian]
4. Landau, M. V., Alexeenko, L. N., Nikulina, L. I., Vinogradova, O. V.,
Nefedov, B. K., Konovalchikov, L. D., Stein, V. I., and Podobaeva,
T. P., Chem. Technol. Fuels Oils N1, 11 (1991). [Russian]
5. Wakabayashi, M., Ono, T., Togari, O., and Nakamura, M., Ger. Offen.
1
2
9326448, 1980 (CA 92P: 170026e).
FIG. 9. DMDBT molecule “activation” according to cracking routes.
1
1
6. Gerdil, R., and Lucken, E., A., J. Am. Chem. Soc. 87, 213 (1965).
7. Ishihara, A., Itoh, T., Hino, T., Nomura, M., Qi, P., and Kabe, T., J.
Catal. 140, 184 (1993).
8. Herskowitz, M., Landau, M. V., Harel, A., and Cohen, B., submitted
for publication.
poisoning with nitrogen compounds present in petroleum
gas oils (29). The published data evident for positive zeolite
1
effects on Co(Ni)–Mo–Al catalysts HDS efficiency (22, 30, 19. Stanislaus, A., and Cooper, B. H., Catal. Rev.-Sci. Eng. 36, 75 (1994).
3
1, 32) were measured in short runs with model compounds 20. Landau, M. V., Alexeenko, L. N., Nefedov, B. K., Kipnis, M. A.,
Agievsky, D. A., Chukin, G. D., Sergienko, S. A., and Kvashonkin,
in absence of nitrogen. Zeolite deactivation in long runs
with petroleum gas oils containing nitrogen compounds
could be a detrimental factor in the implementation of such
solutions to desulfurization catalysts. The high efficiency of
V. I., Kinet. Katal. 25, 934 (1984). [Russian]
2
1. Landau, M. V., Michailov, V. I., Tsisun, E. L., Samgina, T. Yu., Chukin,
G. D., Vinigradova, O. V., Nefedov, B. K., and Slinkin, A.A., Kinet.
Katal. 27, 931 (1986). [Russian]
the cracking activation route in deep HDS should support 22. Welters, W. J. J., de Beer, V. H. J., and van Santen, R. A., Appl. Catal.
A 119, 253 (1994).
further work to resolve potential deactivation problems.
2
3. Welters, W. J. J., Vorbeck, G., Zandbergen, H. W., de Haan, J. W., de
Beer, V. H. J., and van Santen, R. A., J. Catal. 150, 155 (1994).
4. Kolboe, S., Can. J. Chem. 47, 352 (1969).
5. Chianelli, R. R., and Daage, M., Adv. Catal. 40, 177 (1994).
6. Moreau, C., and Geneste, P., in “Theoretical Aspects of Heteroge-
neous Catalysis” (J. B. Moffat, Ed.), p. 256. Van Nostrand–Reinhold,
New York, 1990.
2
2
2
ACKNOWLEDGMENTS
The authors thank Mrs. Y. Press for catalyst preparation, Dr. R.
Rachman for assistance with kinetic measurements, Dr. L. Kogan for con-
ducting the GC-MS analysis, and Dr. I. Levy for conducting the TEM
analysis.
2
7. Landau, M. V., Alexeenko, L. N., Nefedov, B. K., Chukin, G. D.,
Agievsky, D. A., Kvashonkin, V. I., Michailov, V. I., Surin, S. A., and
Pavlova, L. I., Kinet. Katal. 27, 440 (1986). [Russian]
REFERENCES
28. Daage, M., and Chianelli, R. R., J. Catal. 149, 414 (1994).
9. Nat, P. J., Erdol Kohle Erdgas Petrochem. 42, 447 (1989).
2
1
2
. Unzelman, G. H., Oil Gas J. June 19, 55 (1987).
. Takatsuka, T., Wada, Y., Suzuki, H., Komatsu, S., and Morimura, Y.,
J. Jpn. Pet. Inst. 35, 179 (1992).
30. Visotsky, A. V., Chuikova, N. A., and Lipovich, V. G., Kinet. Katal. 18,
1345 (1977). [Russian]
31. Surin, S. A., Aliev, R. R., and Nefedov, B. K., Petrol. Chem. USSR 21,
385 (1981). [Russian]
3
. Kabe, T., Ishihara, A., and Tajima, H., Ind. Eng. Chem. Res. 31, 1577
(
1992).
32. Welters, W. J. J., Koranyi, T. I., de Beer, V. H. J., and van Santen, R. A.,
in “New Frontiers in Catalysis” (L. Guczi, F. Solymosi, and P. T e´ t e´ nyi,
Eds.), p. 1931. Elsevier, Amsterdam, 1992.
4
. Amorelli, A., Amos, Y. D., Haisig, C. P., Koaman, J. J., Jonker, R. J.,
de Wind, M., and Vrieling, J., Hydrocarbon Proc. June, 93 (1992).