A kinetic investigation of the effects of fluorine and nickel on the HDN of toluidine on fully sulfided tungsten sulfide catalysts

Article abstract of DOI:10.1006/jcat.2001.3297

The effects of fluorine and nickel on the hydrodenitrogenation of o-toluidine on alumina-supported tungsten catalysts were studied in a continuous-flow reactor at 320 to 370°C and 3.0 MPa. The catalysts were prepared from ammonium tetrathiotungstate and were fully sulfided. The kinetic data were obtained by varying the initial partial pressure of the reactant and the reaction temperature; Langmuir-Hinshelwood models were used to fit the kinetic data. The simultaneous reactions of cyclohexene and o-toluidine enabled us to study the inhibition effect of o-toluidine on the hydrogenation of cyclohexene and to determine the difference between the sites for these reactions. The kinetic data suggest that the formation of toluene and 2-methylcyclohexylamine from o-toluidine occurs through a common partially hydrogenated intermediate, dihydrotoluidine, which leads to toluene by elimination and 2-methylcyclohexylamine after further hydrogenation. Fluorination changes neither the activation energies for the hydrogenation reaction nor the heat of adsorption of o-toluidine. Fluorine thus does not change the intrinsic properties of the active sites but does affect the apparent activity by influencing the number of active sites. Only a change in the morphology of the metal sulfide surface, by stacking, can explain the fluorine effect. Addition of nickel changes the nature of the active site by creating a nickel-associated sulfur vacancy, which is highly active for hydrogenation.

Full text of DOI:10.1006/jcat.2001.3297

Journal of Catalysis 203, 192–200 (2001)
doi:10.1006/jcat.2001.3297, available online at http://www.idealibrary.com on
A Kinetic Investigation of the Effects of Fluorine and Nickel on the HDN
of Toluidine on Fully Sulfided Tungsten Sulfide Catalysts
Mingyong Sun and Roel Prins1
Laboratory for Technical Chemistry, Swiss Federal Institute of Technology (ETH), 8093 Zurich, Switzerland
Received April 9, 2001; accepted June 7, 2001; published online August 28, 2001
elementary reactions involved in the HDN of o-toluidine
The effects of fluorine and nickel on the hydrodenitrogena- over unpromoted and nickel-promoted W/Al2O3 catalysts
tion of o-toluidine on alumina-supported tungsten catalysts were (8) and found that fluorine increased the activity of the hy-
�
studied in a continuous-flow reactor at 320 to 370 C and
3
drogenation of the aromatic ring and the breaking of the
C(sp )–N bond. Complementary TPS, XPS, QEXAFS, and
.0 MPa. The catalysts were prepared from ammonium tetrathio-
2
tungstate and were fully sulfided. The kinetic data were obtained
by varying the initial partial pressure of the reactant and the
reaction temperature; Langmuir–Hinshelwood models were used
to fit the kinetic data. The simultaneous reactions of cyclohex-
ene and o-toluidine enabled us to study the inhibition effect of o-
toluidine on the hydrogenation of cyclohexene and to determine the
difference between the sites for these reactions. The kinetic data
XANES investigations showed that fluorination affected
the sulfidation process and the composition of the sulfided
catalysts (9, 10).
Reliable methods for the determination of the disper-
sion of metal sulfides are not available yet. Consequently,
it is very difficult to conclude that the difference in cata-
suggest that the formation of toluene and 2-methylcyclohexylamine lytic activity is due either to a better dispersion or to a
from o-toluidine occurs through a common partially hydrogenated difference in the catalytic sites. Kinetic studies might re-
intermediate, dihydrotoluidine, which leads to toluene by elimi- veal a difference in the intrinsic properties of the active
nation and 2-methylcyclohexylamine after further hydrogenation.
Fluorination changes neither the activation energies for the hy-
drogenation reaction nor the heat of adsorption of o-toluidine.
Fluorine thus does not change the intrinsic properties of the ac-
rameters include the number of sites, the latter do not de-
tive sites but does affect the apparent activity by influencing the
pend principally on the number of sites but rather on the
number of active sites. Only a change in the morphology of the
metal sulfide surface, by stacking, can explain the fluorine effect.
Addition of nickel changes the nature of the active site by creating a
sites on different catalysts. In principle, kinetic studies de-
termine two types of parameters, i.e., rate constants and
equilibrium adsorption constants. Whereas the former pa-
quality of the sites. Therefore, investigating the effects of
fluorination and promoter incorporation on the kinetic pa-
rameters is an alternative way for gaining a better under-
standing of the promotional effect of fluorine and of the
nickel-associated sulfur vacancy, which is highly active for hydro-
genation. �c 2001 Academic Press
Key Words: hydrodenitrogenation; fluoride effect; nickel; tung- promoter on the activities of HDS and HDN. The kinetic
sten sulfide; kinetics; o-toluidine; cyclohexene.
investigation of the effects of fluorine and nickel on the
HDN of 2-methylcyclohexylamine showed that fluorine did
not change the intrinsic properties of the active sites but
decreased their number and that nickel slightly increased
the adsorption of 2-methylcyclohexylamine on the active
sites (11). The work presented here is a continuation of our
earlier work, the aim of which was to gain a fundamental
understanding of the role of fluorine and nickel in HDN
catalysts.
Although it is well known that the addition of nickel
or cobalt to molybdenum and of nickel to tungsten-
hydrotreating catalysts strongly enhances their activity,
several fundamental questions regarding the promotional
effect of cobalt and nickel remain unanswered (12, 13). Sev-
eral models have been proposed to explain the promotional
effect of cobalt on molybdenum, the most widely accepted
model being the so-called “CoMoS” phase model (14, 15).
INTRODUCTION
Fluorination of alumina increases the activity of alumina-
supported molybdenum (1–4) and tungsten (5, 6) cata-
lysts for hydrodesulfurization (HDS) and hydrodenitro-
genation (HDN). These higher activitieshave been ascribed
to changes in the dispersion of the metal species in the
oxidic precursor of the catalyst (3), to enhanced acidity
(
2, 5–7), and to increased sulfidability and to changes in
the morphology of the WS2 crystallites in the sulfided cata-
lyst (5, 6). We investigated the effect of fluorination on the
1
To whom correspondence should be addressed. Fax: 41-1-6321162.
E-mail: prins@tech.chem.ethz.ch.
192
0021-9517/01 $35.00
Copyright �c 2001 by Academic Press
All rights of reproduction in any form reserved.

HDN KINETICS OF o-TOLUIDINE ON Ni-W/Al2O3(F)
193
Analogously, the promotional effect of nickel on molybde- products were analyzed with an on-line gas chromatograph
num and tungsten catalysts is attributed to the formation equipped with a 50-m CP Sil-5 fused silica capillary column
of “NiMoS” and “NiWS” phases, respectively. However, and a flame ionization detector. Samples were taken after
why the nickel-promoted molybdenum or tungsten cata- 20 h on stream when the activity of the catalyst was stable.
lysts have higher activity is not yet clear. Theoretical calcu- At the end of the experiments, the initial reaction condi-
lations suggest that the promoters weaken the sulfur bind- tions were repeated and the activities of the catalysts were
ing energy, decrease the equilibrium sulfur coverage, and found to differ by less than 5% .
hence increase the number of vacancies, which are the ac-
tive sites (16, 17). Because the loss of hydrogenation activity
of a Ni–Mo catalyst in the HDN of indoline was related to
RESULTS
a measured decrease in the nickel content on the catalyst
surface, Zhang et al. suggested that sulfur vacancies asso-
HDN of o-Toluidine
ciated with nickel in the NiMoS phase are responsible for
most of the hydrogenation activity (18).
Classically prepared tungsten catalysts are only partially
sulfided and are a mixture of W(VI) oxysulfides and WS2
The HDN of TOL proceeds via two pathways: the di-
2
rect C(sp )–N bond breaking to toluene (T) and the hy-
3
drogenation of the phenyl ring followed by C(sp )–N bond
breaking of 2-methylcyclohexylamine (MCHA) to methyl-
cyclohexane (MCH) and methylcyclohexene (MCHE)
(
9, 10, 19–21). To interpret the kinetic data more precisely,
we used catalysts prepared from ammonium tetrathio-
tungstate, which can be fully sulfided (9, 10), to study the
kinetics of the HDN of o-toluidine. Preliminary results of
investigations of the effect of fluorine on the kinetics of the
HDN of o-toluidine on the unpromoted tungsten catalysts
have been reported elsewhere (22). We extended the study
of tungsten catalysts to include nickel-promoted tungsten
catalysts.
(
Fig. 1) (8, 23). Under the present reaction conditions, the
hydrogenation intermediate MCHA was not detected in
the HDN products of TOL on any of the catalysts, because
the hydrogenation of TOL to MCHA is slow relative to the
subsequent reactions of MCHA to MCHE and MCH.
The products, MCHE and MCH, will be referred
to as ‘the hydrogenation products,’ because they are formed
2
via path 2. It was claimed that C(sp )–N bond breaking
and the hydrogenation of the phenyl ring occur on differ-
ent catalytic sites (18, 24–27); therefore, if this is true, then
different equilibrium adsorption constants of TOL should,
EXPERIMENTAL
The catalysts used in this study were prepared from am- in principle, be considered for these two active sites. The
monium tetrathiotungstate (AT T). The preparation and yield of the hydrogenation products increased linearly with
the characterization results of the catalysts AT T /Al2O3, the conversion of TOL for all catalysts, as shown for the Ni–
AT T /Al2O3 –F, Ni–AT T /Al2O3, and Ni–AT T /Al2O3 –F ATT/Al O catalyst in Fig. 2. Therefore, the selectivities of
2
3
were presented elswhere (9, 10). All the catalysts contained the hydrogenation products and of toluene do not depend
0 wt% of tungsten; the nickel-promoted catalysts con- on the initial partial pressure of TOL. This means that it
1
tained 1 wt% of nickel, and the fluorine content of the fluo- is possible to use the same adsorption constant of TOL for
rinated catalysts was 1 wt% . Prior to the reaction, the cata- both reaction paths. The adsorption of the nitrogen-free
lysts were treated in situ with a mixture of H2S (10 mol% ) hydrocarbon products and ammonia is much weaker than
�
and H2 (90 mol% ) at 400 C and 1.5 MPa for 4 h. Then, that of a nitrogen-containing reactant (23, 28) and can be
�
the temperature was lowered to 370 C, the pressure was ignored in the Langmuir–Hinshelwood rate equation for
increased to 3.0 MPa, and the liquid feed was introduced
to the reactor by means of a high-pressure pump. The reac-
tants were o-toluidine (TOL) and cyclohexene (CHE) with
n-octane as the solvent and n-heptane as the internal stan-
dard. Dimethyl disulfide was added to the feed to generate
H2S (6 kPa) in the reaction stream. The partial pressure
of TOL varied from 1 to 9 kPa, while the partial pressures
of CHE, n-octane, and n-heptane were kept constant at
9
, 75, and 5 kPa, respectively. The partial pressure of H2
was varied slightly from 2900 to 2920 kPa so that the total
pressure remained constant at 3.0 MPa. The amount of the
catalyst varied from 0.1 to 1.2 g to obtain different levels of
conversion of TOL and CHE. We checked that the conver-
sion of TOL on a catalyst was the same at the same weight
time when using different amounts of catalyst. Reaction
FIG. 1. HDN network of o-toluidine.

1
94
SUN AND PRINS
FIG. 3. Conversion versus weight time in the HDN of 9 kPa o-
toluidine on Ni–ATT/Al2O3 at 3.0 MPa.
FIG. 2. Yield of the products of the hydrogenation path versus con-
version of o-toluidine at 345 C and 3.0 MPa on Ni–ATT/Al2O3 at different
partial pressures of o-toluidine.
�
2
are parallel reactions (8, 23); the selectivity of the C(sp )–N
bond breaking is equal to k1/(k1 + k2) and that of the hy-
drogenation is equal to k2/(k1 + k2). Thus, from the overall
[1] rate constants k1 +k2 and the selectivity of the products, the
individual values of k1 and k2 were obtained (Table 1).
Figure 4 shows the temperature dependence of the rate
constants of paths 1 and 2 and of the equilibrium adsorp-
tion constant of TOL on the Ni–ATT/Al2O3 catalyst. The
activation energies are given in Table 2, with the standard
deviations in parentheses. The equilibrium adsorption con-
stant of TOL on the unpromoted tungsten catalysts hardly
changed with temperature. Therefore, the heat of adsorp-
tion of TOL on the unpromoted catalysts is too low to be
determined accurately. The heat of adsorption of TOL on
the nickel-promoted catalysts and its standard deviation are
given in Table 2.
the HDN of TOL,
d PTOL
(k1 + k2)KTOL PTOL
1 + KTOL PTOL
=
�
,
d�
where PTOL is the partial pressure of TOL, � the weigh time,
k1 the rate constant for path 1, k2 the rate constant for path
2
, and KTOL the equilibrium adsorption constant of TOL.
Direct simulation of the partial pressure of the reactants
versus weight time, using Eq. [1], did not give unique pa-
rameter values, as has been addressed before (11, 29, 30).
To obtain unique parameters, one of the parameters must
be determined independently, as for the HDN of MCHA
(
11). At a high partial pressure of TOL, KTOL PTOL � 1, and
Eq. [1] simplifies to the zero-order equation,
k1 + k2
XTOL =
� �,
[2]
0
P
TOL
where XTOL is the conversion and P0 the initial partial
TOL
pressure of TOL. The plots are linear up to a TOL conver-
sion of 30% . As an example, Fig. 3 gives the conversion of
TOL over the Ni–ATT/Al2O3 catalyst versus weight time
when the initial partial pressure of TOL is 9 kPa. In this
way, the sums of the rate constants for the overall con-
version of TOL, k1 + k2, were obtained for all the catalysts.
Once k1 + k2 was determined, the adsorption constant KTOL
could be determined uniquely by varying the initial partial
pressure of TOL (1, 2, 4, and 9 kPa) and by doing a non-
linear regression analysis of the experimental data using
MacroMath Scientist 2.0 software, according to Eq. [1]. The
results are given in Table 1. The statistical evaluation of the
goodness-of-fit gave the standard deviations of the param-
eters, which are presented in parentheses in Table 1. Under
FIG. 4. Temperature dependence of the rate constants of paths 1 (k1)
and 2 (k2) and the equilibrium adsorption constant of o-toluidine (KTOL
)
the present conditions, paths 1 and 2 in the HDN of TOL for the HDN of o-toluidine on Ni–ATT/Al2O3 at 3.0 MPa.

HDN KINETICS OF o-TOLUIDINE ON Ni-W/Al2O3(F)
195
TABLE 1
�
1
� 1
� 1
Kinetic Parameters for the HDN of o-Toluidine at 3.0 MPa (k
1
and k
2
in kPa mol min
g
and KTOL in kPa )
ATT/Al2O3
ATT/Al2O3–F
Ni–ATT/Al2O3
Ni–ATT/Al2O3–F
�
3
3
3
70 C k2/(k1 + k2)
0.93
0.93
0.92
0.02
0.22
0.72 (0.1)
0.91
0.02
0.26
0.87 (0.1)
k1
k2
0.002
0.022
1.1 (0.2)
0.003
0.033
1.3 (0.2)
K
TOL
�
45 C k2/(k1 + k2)
0.93
0.94
0.96
0.95
k1
k2
0.001
0.009
1.1 (0.3)
0.001
0.013
1.4 (0.3)
0.004
0.076
1.2 (0.3)
0.004
0.078
1.4 (1.2)
K
TOL
�
20 C k2/(k1 + k2)
0.93
0.94
0.98
0.97
k1
k2
0.0003
0.0037
1.2 (0.3)
0.0003
0.0054
1.4 (0.3)
0.001
0.023
2.5 (0.5)
0.001
0.024
2.4 (0.3)
K
TOL
Hydrogenation of Cyclohexene
actions of CHE and TOL, the decrease in the partial pres-
sure of TOL (Eq. [3]) is very small at low conversion of
TOL. The hydrogenation of CHE in the presence of TOL
was, therefore, reduced to a pseudo-first-order reaction by
making the partial pressure of TOL in the denominator of
Eq. [2] equal to the initial value. Thus, Eq. [3] was simplified
to
The elementary reactions involved in the HDN of
MCHA were studied separately, using MCHA as the model
compound (11). Another elementary reaction in the HDN
network of TOL is the hydrogenation of MCHE to MCH.
The hydrogenation of an alkene requires different active
sites than the hydrogenation of a phenyl ring on sulfide
catalysts (8, 27). Therefore, different adsorption constants
of TOL must be used in kinetic calculations of the HDN
of TOL and the hydrogenation of CHE in the presence of
TOL. The rate equation for the hydrogenation of CHE in
the presence of TOL can be expressed as
ln(1 � XCHE) = k0 �,
[4]
CHE
where
kCHE KCHE
0
k
=
.
[5]
CHE
CHE
0
1
+ KTOL
P
TOL
From the slopes of the plots of ln(1 � XCHE) against weight
time, the apparent rate constants for the hydrogenation of
CHE at different initial partial pressures of TOL were ob-
tained. The reciprocal of the apparent rate constant for the
d PCHE
kCHE KCHE PCHE
=
�
,
[3]
CHE
d�
1 + KTOL PTOL + KCHE PCHE
where PCHE is the partial pressure of CHE, � is the weight hydrogenation of CHE is proportional to the initial par-
CHE
time, kCHE and KCHE are the rate and equilibrium adsorp- tial pressures of TOL. Thus, values for KTOL and kCHE KCHE
CHE
tion constants of CHE respectively, KTOL is the equilibrium were estimated from the slopes and intercepts of the plots
0
adsorption constant of TOL on the CHE reaction sites, and of 1/k
against the initial partial pressure of TOL at dif-
CHE
�
1
PTOL is the partial pressure of TOL. Since KCHE � 0.01 kPa
2 3
ferent temperatures (cf. Fig. 6 for Ni–ATT/Al O ). Using
these values as starting points for the CHE rate and adsorp-
Figure 5 shows that the conversion of CHE decreases tion constants and the rate and adsorption constants for
with increasing partial pressure of TOL, showing that TOL the HDN of TOL as fixed parameters, unequivocal kinetic
inhibits the hydrogenation of CHE. In the simultaneous re- parameters for the hydrogenation of CHE in the presence
(
23) and PCHE = 9 kPa, the term KCHE PCHE was ignored.
TABLE 2
Temperature Dependence of Rate Constants and Equilibrium Constants in the HDN of o-Toluidine at 3.0 MPa (k
i
= A
i
exp(� E
i
/RT ),
K
TOL = K exp(� 1 H/RT ))
0
ATT/Al2O3
ATT/Al2O3–F
Ni–ATT/Al2O3
Ni–ATT/Al2O3–F
E1 (kJ/mol)
114 (7)
3E + 6
114 (10)
4E + 7
—
133 (15)
2E + 8
116 (10)
8E + 7
—
224 (19)
2E + 16
142 (12)
1E + 10
79 (10)
216 (10)
8E + 15
151 (10)
5E + 11
63 (10)
A1 (kPa mol min� 1
E2 (kJ/mol)
g
� 1
� 1
)
)
A2 (kPa mol min� 1
g
�
1H (kJ/mol)
�
1
K0 (kPa
)
—
—
1E � 6
1E � 5

1
96
SUN AND PRINS
3
cleavage of C(sp )–N bonds is determined by the basicity of
the nitrogen atom (24). Therefore, a possible mechanism for
the direct conversion of TOL to toluene might be the par-
tial hydrogenation of TOL to dihydrotoluidine followed by
elimination of ammonia as suggested by Girgis and Gates
(
42) (Fig. 7). A similar mechanism has also been proposed
for the conversion of dibenzothiophene (43). Another ex-
planation may be that the arylamine undergoes C–N bond
homolysis on the surface of the metal sulfide, as proposed
for the C–S bond scission in the homogeneous desulfuriza-
0
tion of thiols on Cp Mo2Co2S3(CO)4 clusters (44). In the
2
latter reaction, the thiol binds to a “latent vacancy” by reor-
ganization of the metal and sulfur atoms of the metal–sulfur
cluster (Scheme 1). The strong bonding of the sulfur atom
of the SR fragment to several metal atoms weakens the C–S
bond. This leads to C–S bond breaking and RH formation.
A similar HDN mechanism may occur by the bonding of a
RNH fragment to the metal sulfide surface.
The selectivity of the hydrogenation products hardly
changes with temperature over the unpromoted tungsten
catalysts (Table 1). This is due to the very similar activation
energies of paths 1 and 2 (within the uncertainties of the de-
termination) over the unpromoted catalysts (Table 2). This
result can be explained by the common intermediate mech-
anism, asindicated in Fig. 7. For the ATT/Al2O3 catalyst, the
ratio of the rate constants for the elimination of ammonia
from MCHA (11) and for the conversion of TOL (Table 1)
is 49. Even if we assume that the rate of elimination of am-
monia from dihydrotoluidine to form toluene is lower than
the rate of elimination of ammonia from MCHA, the rate of
elimination of ammonia from dihydrotoluidine might still
be higher than the rate of the hydrogenation of TOL. In
that case, the observed activation energy for the formation
FIG. 5. Conversion of cyclohexene versus weight time in the presence
of different partial pressures of o-toluidine at 345 C and 3.0 MPa on Ni–
ATT/Al2O3.
�
of TOL were obtained through simulation of the experi-
mental data with Eqs. [1] and [3] (Table 3).
The rate and adsorption constants of CHE cannot be sep-
arated under the present conditions. Thus, we obtained only
the sum of the activation energy and the heat of adsorption
of CHE from the temperature dependence of kCHE KCHE
Table 3). The adsorption constant of TOL on the CHE
reaction site is too small to reliably determine the heat of
adsorption of TOL on the CHE reaction sites.
(
DISCUSSION
The hydrogenation of the phenyl ring is the major path
in the HDN of TOL on the alumina-supported tungsten
sulfide catalyst in the presence of H2S, as already observed
in the HDN of other substituted anilines on molybdenum-
and tungsten-based sulfide catalysts (23, 24, 29, 31–34). The
hydrogenation path contributes more than 90% to the over-
all conversion of TOL for the tungsten sulfide catalysts
2
(
Table 1). It is still unclear as to how the direct C(sp )–
N bond breaking from TOL to toluene takes place. This
reaction is often referred to as hydrogenolysis (24, 31, 33),
but it is unlikely that the C–N bond breaks on sulfide cata-
lysts in the same way as the C–C bond in hydrocarbon on
metal catalysts. Many experimental results have shown that
2
the mechanism of C(sp )–N bond breaking differs from
3
C(sp )–N bond breaking (8, 24, 27, 32, 35–37), which occurs
by Hofmann-type elimination or nucleophilic substitution
2
(
23, 24, 38–41). C(sp )–N bond breaking behaves like hy-
drogenation (8); it is enhanced by hydrogen and inhibited
by H2S (37, 38). Furthermore, it depends on the aromatic-
ity of the phenyl ring of aniline-type compounds, while the
SCHEME 1

HDN KINETICS OF o-TOLUIDINE ON Ni-W/Al2O3(F)
197
TABLE 3
Kinetic Parameters for the Hydrogenation of CHE at 3.0 MPa (kCHE in kPa mol min 1
�
g
� 1
, K
CHE
TOL
� 1
and KCHE in kPa ,
�
�
k
CHE
KCHE = (k
K
) exp(� (ECHE + 1HCHE)/RT))
CHE CHE
ATT/Al2O3
ATT/Al2O3–F
Ni–ATT/Al2O3
Ni–ATT/Al2O3–F
�
3
3
3
70 C
k
K
CHE
TOL
0.014 (0.001)
0.03 (0.01)
0.022 (0.001)
0.03 (0.01)
0.011 (0.001)
0.05 (0.01)
0.0065 (0.0001)
0.08 (0.01)
78 (10)
0.33 (0.03)
0.24 (0.06)
0.18 (0.01)
0.32 (0.04)
0.088 (0.003)
0.37 (0.02)
84 (10)
0.36 (0.01)
0.32 (0.02)
0.20 (0.01)
0.37 (0.03)
0.089 (0.003)
0.41 (0.02)
89 (10)
CHE CHE
K
�
45 C
k
K
0.0075 (0.0001)
0.04 (0.01)
CHE CHE
CHE
K
TOL
�
20 C
k
K
0.0045 (0.0004)
0.06 (0.01)
CHE CHE
CHE
K
E
TOL
+ 1HCHE
72 (10)
CHE
(
kJ/mol)
of toluene would be equal to the activation energy of the the Ni–ATT/Al2O3 catalyst and the same activation energy
hydrogenation of TOL over the ATT/Al2O3 catalyst.
On the nickel-promoted catalysts, the selectivity of the
over the ATT/Al2O3 catalyst.
It is widely accepted that there are two main types of
hydrogenation products decreases with increasing reaction active sites on molybdenum and tungsten sulfide catalysts.
temperature (Table 1), which is explained by the higher Type I consists of sulfur vacancies associated with metal
activation energy for path 1 than for path 2 over the nickel- atoms, and type II consists of surface SH groups or sulfur
promoted catalysts (Table 2). The ratio of the rate con- anions (14, 15, 18, 33, 35, 45–49). The fact that H2S inhibits
2
stants for the elimination of ammonia from MCHA (11) C(sp )–N bond breaking (37, 38) suggests that sulfur vacan-
and for the conversion of TOL (Table 1) is only 6 for the ciesare necessaryfor thisreaction to occur. In general, there
Ni–ATT/Al2O3 catalyst. This is due to the much higher hy- are two kinds of sulfur vacancies on the nickel-promoted
drogenation rate of the Ni–ATT/Al2O3 catalyst than that of catalysts, one associated with molybdenum (or tungsten)
the ATT/Al2O3 catalyst at about the same elimination rate. atoms and the other with nickel in the Ni–Mo–S (or Ni–
Thus, the rate of elimination of ammonia from dihydrotolu- W–S) phase (18, 46, 49–52). It has been proposed that va-
idine might not be higher than the rate of hydrogenation of cancies involving nickel are mainly responsible for the hy-
TOL for the Ni–ATT/Al2O3 catalyst. In that case, the acti- drogenation (18, 49, 53, 54). Consistent with this proposal,
vation energy for the formation of toluene is not the same we found that the incorporation of nickel in the tungsten
as that for the hydrogenation of TOL. This may explain catalysts hardly affected the HDN of MCHA (8, 11) but
why paths 1 and 2 show different activation energies over significantly increased the HDN of TOL (Table 1).
Nickel increases both the activation energy and the
heat of adsorption of TOL (Table 1). This means that the
nickel-associated vacancies are different in nature than
the tungsten-associated vacancies. The equilibrium ad-
sorption constant of TOL hardly changes with tempera-
ture on the unpromoted tungsten catalysts, while a clear
increase in the equilibrium adsorption constant of TOL on
FIG. 6. 1/kCHE versus initial partial pressure of o-toluidine in the
hydrogenation of cyclohexene in the presence of o-toluidine on Ni–
FIG. 7. Reaction scheme for the HDN of o-toluidine via a dihydro-
ATT/Al2O3 at 3.0 MPa.
toluidine intermediate.

1
98
SUN AND PRINS
the nickel-promoted catalysts is observed with decreasing the results of the simultaneous reactions of CHE and
temperature (Table 1). The reason for this behavior is not MCHA). This is in line with the stronger basicity of MCHA
clear. We made several checks to eliminate the possibil- (pKa = 10) than that of TOL (pKa = 4.4) (57, 58). There-
ities of experimental inconsistencies. It may be that the fore, it is probable that both TOL and MCHA adsorb on
degree of sulfidation of the catalyst surface is more sen- the alkene hydrogenation site via their amine groups. TOL
sitive to temperature for the unpromoted catalysts than for adsorbs on its own hydrogenation site via the aromatic moi-
the Ni-promoted catalysts. Although the H2S/H2 ratio was ety while it adsorbs on the alkene hydrogenation site via its
kept constant at 0.002 in all the experiments at all tem- amine group; this explains why the equilibrium adsorption
peratures, this does not mean that the relative amount of constants of TOL are so different on the two sites. The
sulfur (and thus of the vacancies) is the same at all temper- much larger equilibrium adsorption constants of TOL on
atures. Even at the same H2S/H2 ratio, more sulfur vacan- the CHE reaction site on the nickel-promoted catalyststhan
cies are to be expected at higher temperature because of on the unpromoted catalysts (Table 3) confirm that there
�
�
the entropy effect on the S + H2 *ꢀ H2S + equilibrium, is a fundamental difference between the nickel-associated
where S denotes the “adsorbed” sulfur atom and is a vacancy and the tungsten-associated vacancy.
sulfur vacancy. The equilibrium constant KTOL would then Many explanations have been given in the literature for
be considered to consist of the product of the real adsorp- the effect of fluorine on hydrotreating catalysts (2, 3, 5–7).
�
�
0
�
tion constant K and the constant K of this surface sulfur The enhanced acidity of fluorinated catalysts has been con-
TOL
0
0
equilibrium, KTOL = K � K�. The constant K decreases sidered to be one of the main factors that accounts for the
TOL
TOL
�
with increasing temperature, but K increases, and thus the promotional effect of fluorine on HDS and HDN (2, 5–7,
number of sulfur vacancies also increases. Theoretical cal- 59) since it was observed that fluorine enhanced the acid-
culations showed that the sulfur binding energy is higher catalyzed isomerization and cracking reactions (2, 7, 59,
3
for unpromoted sulfided molybdenum structures than for 60). Indeed, C(sp )–N bond breaking requires acidic sites
the Co–Mo–S and Ni–Mo–S structures (16, 17). Analo- as well, be it elimination or nucleophilic substitution. How-
gously, if the sulfur atoms are more strongly bonded to ever, we found that fluorine hardly influenced the elimi-
unpromoted-tungsten structures than to nickel-promoted nation but that it decreased the nucleophilic substitution in
structures, the temperature dependence of this sulfur equi- the HDN of MCHA (11). The acid sites related to fluorine
librium is stronger for the ATT catalysts than for the Ni– on alumina are responsible for isomerization and cracking
ATT catalysts. Consequently, the temperature dependency (7, 61, 62), but they are not available for isomerization
of the ATT catalysts is less than that of the Ni–ATT cata- of CHE when a basic nitrogen-containing compound is
lysts.
present in the feed. As we showed before (8), in that
The equilibrium adsorption constant of TOL on the CHE case, the acidic sites are completely inhibited by the basic
hydrogenation site is much smaller than that on its own molecules, and increased acidity cannot explain the role of
hydrogenation site (cf. Tables 1 and 3), just as the equilib- fluorine in HDN.
rium adsorption constants of quinoline and o-propylaniline
Within the uncertainty of the measurement, fluorine
on the alkene hydrogenation site are much smaller than changes neither the activation energies nor the heat of ad-
those on the site of the phenyl ring hydrogenation on sorption of toluidine (Table 2) in the ATT/Al2O3 and the
Ni–Mo/Al2O3 catalysts (27). These results prove that the Ni–ATT/Al2O3 catalysts. This proves that, in both cases,
hydrogenation of the phenyl ring occurs on a different fluorine does not change the nature of the active sites. The
site than the hydrogenation of an alkene. Because the �- promotional effect of fluorine on the HDN of TOL, there-
electron moiety interacts with the active centers in the re- fore, does not result from a change in the intrinsic activity of
action of aniline-type compounds on sulfide catalysts by flat the sites but from an increase in the number of active sites.
adsorption (24), there must be more vacancies to accommo- For classic alumina-supported tungsten catalysts, which can
�
date and activate the aromatic ring in the reaction of TOL, only be partially sulfided at 400 C, fluorine was considered
while for the hydrogenation of an alkene one vacancy may to improve the sulfidation (5, 6) and, thus, the number of
3
be sufficient. In the C(sp )–N bond breaking of MCHA, active sites for HDN. However, TPS measurements (9, 10)
however, the basicity of the amine moiety determines the and sulfur analysis (6) showed that fluorine only slightly
reactivity (24) because the adsorption of MCHA on the ac- increased the degree of sulfidation of tungsten, while XPS
tive sites occurs via its amine moiety, which is strongly basic. results showed that the fluorinated catalysts had a lower
In the presence of MCHA, the hydrogenation of CHE W(IV)/W(VI) ratio than their fluorine-free counterparts
is strongly inhibited (8), while in the presence of TOL, after sulfidation (6, 9, 10). Furthermore, in the present study
the conversion of CHE is still substantial (Fig. 5). Thus, we used fully sulfided catalysts; therefore, the extent of sul-
the inhibitory effect of MCHA on alkene hydrogenation fidation cannot be a key factor in the effect of fluorination.
�
is much stronger than that of TOL; for instance, at 370 C,
Our results unequivocally show that the character of the
CHE
TOL
� 1
CHE
� 1
> 2.5 kPa (estimated from catalytic sites is not influenced by fluorine. This was to
MCHA
K
= 0.24 kPa and K

HDN KINETICS OF o-TOLUIDINE ON Ni-W/Al2O3(F)
199
be expected because it is well known that fluorine is lo- jacent vacancies where TOL can adsorb and react, and the
cated on the alumina support and not on the active metal higher stackingfavorsthe adsorption ofTOL on the catalyst
sulfide. This means that the fluorine effect must be indirect. via the phenyl �-system. The hydrogenation of CHE does
How does fluorination of the alumina support influence the not require adjacent vacancies, and therefore, the change
number of sites of the metal sulfide phase? A higher dis- in the morphology of the catalyst surface has little effect
persion of the metal sulfide cannot be the explanation be- on it. Incorporation of nickel in the tungsten catalysts cre-
cause fluorination decreases the dispersion of the tungsten ates a new type of active site, a nickel-associated sulfur va-
phase on the catalyst surface (6, 11, 63). Transmission elec- cancy with a much higher activity for hydrogenation than
tron microscopymeasurementsshowed that fluorination in- the tungsten-associated sulfur vacancy.
creased the stacking and the length of WS2 slabs (5, 64). Our
QEXAFS results also showed that fluorine increased the
size of the WS2 crystallites (9). WS2 crystallites are well
REFERENCES
dispersed on fluorine-free catalysts but aggregate to bigger
particles on the fluorinated catalysts. As discussed above,
the hydrogenation of alkenes requires one vacancy, the
adsorption of MCHA is through its amine group, and the
hydrogenation of TOL requires more vacancies to accom-
modate the phenyl ring. The bigger WS2 particles may con-
tain more adjacent vacancies, which are the active centers
for TOL hydrogenation. Furthermore, the higher stacking
provides more space for the flat adsorption of TOL on
these active centers. Considering that fluorine hardly influ-
ences the hydrogenation of CHE and the HDN of MCHA
and that it increases the hydrogenation of TOL, we con-
clude that the higher stacking of WS2 is the origin of the
promotional effect of fluorine on the HDN of TOL.
1
. Muralidhar, G., Massoth, F. E., and Shabtai, J., J. Catal. 85, 44 (1984).
2. Jiratova, K., and Kraus, M., Appl. Catal. A 27, 21 (1986).
3
4
5
6
7
8
. Papadopoulou, Ch., Lycourghiotis, A., Grange, P., and Delmon, B.,
Appl. Catal. A 38, 255 (1988).
. Kwak, C., Lee, J. J., Bae, J. S., Choi, K., and Moon, S. H., Appl. Catal.
A 200, 233 (2000).
. Benitez, A., Ramirez, J., Vazquez, A., Acosta, D., and Lopez Agudo,
A., Appl. Catal. A 133, 103 (1995).
. Benitez, A., Ramirez, J., Fierro, J. L. G., Lopez Agudo, A., Appl. Catal.
A 144, 343 (1996).
. Boorman, P. M., Kydd, R. A., Sarbak, Z., and Somogyvari, A.,
J. Catal. 106, 544 (1987).
. Sun, M., Bussell, M. E., and Prins, R., Appl. Catal. A 216, 103 (2001).
9. Sun, M., Burgi, Th., Cattaneo, R., and Prins, R., J. Catal. 197, 172
2001).
(
10. Sun, M., Burgi, Th., Cattaneo, R., van Langeveld, A. D., and Prins, R.,
submitted for publication.
11. Sun, M., and Prins, R., J. Catal. 201, 138 (2001).
CONCLUSIONS
12. Topsøe, H., Clausen, B. S., and Massoth, F. E., “Hydrotreating Cata-
lysis.” Springer Verlag, Berlin, 1996.
1
1
1
1
1
1
3. Prins, R., de Beer, V. H. J., and Somorjai, G. A., Catal. Rev.-Sci. Eng.
1, 1 (1989).
4. Topsøe, H., Clausen, B. S., Candia, R., Wivel, C., and Mørup, S.,
J. Catal. 68, 433 (1981).
5. Wivel, C., Candia, R., Clausen, B. S., Mørup, S., and Topsøe, H,
J. Catal. 68, 453 (1981).
6. Byskov, L. S., Nørskov, J. K., Clausen, B. S., and Topsøe, H., J. Catal.
The fact that the two pathways in the HDN of TOL, one
leading to toluene and the other to MCHE and MCH, show
similar activation energies on unpromoted tungsten cata-
lysts suggests the existence of a common intermediate, di-
hydrotoluidine, for the formation of toluene and MCHA.
The fact that the selectivities of toluene and MCHE plus
MCH do not depend on the partial pressure of TOL also
suggests that these two pathways go through the same in-
termediate.
3
187, 109 (1999).
7. Raybaud, P., Hafner, J., Kresse, G., Kasztelan, S., and Toulhoat, H.,
J. Catal. 190, 128 (2000).
8. Zhang, L., Karakas, G., and Ozkan, U. S., J. Catal. 178, 457 (1998).
The equilibrium adsorption constants of TOL on the 19. Ng, K. T., and Hercules, D. M., J. Phys. Chem. 80, 2094 (1976).
2
0. Breysse, M., Cattenot, M., Decamp, T., Frety, R., Gachet, C., Lacroix,
M., Leclerq, C., de Mourgues, L., Portefaix, J. L., Vrinat, M., Houari,
M., Grimblot, J., Kasztelan, S., Bonnelle, J. P., Housni, S., Bachelier,
J., and Duchet, J. C., Catal. Today 4, 39 (1988).
TOL hydrogenation site and on the CHE hydrogenation
site are very different, demonstrating that the site for the
hydrogenation of a phenyl ring differs from that for the
hydrogenation of an alkene. The former site probably con-
21. Scheffer, B., Mangnus, P. J., and Moulijn, J. A., J. Catal. 121, 18 (1990).
sists of more vacancies that accommodate TOL via its �- 22. Sun, M., and Prins, R., Stud. Surf. Sci. Catal. 133, 581 (2001).
2
2
3. Rota, F., and Prins, R., Top. Catal. 11/12, 327 (2000).
4. Moreau, C., Aubert, C., Durand, R., Zmimita, N., and Geneste, P.,
Catal. Today 4, 117 (1988).
5. Okamoto, Y., Maezawa, A., and Imanaka, A., J. Catal. 120, 29 (1989).
6. Massoth, F. E., Balusami, K., and Shabtai, J., J. Catal. 122, 256 (1990).
electrons; the latter site has one vacancy only, upon which
TOL adsorbs via the amine group.
Fluorination does not cause significant changes in the ac-
tivation energy and the heat of adsorption of TOL on the
2
2
sites for the hydrogenation of TOL and CHE. This indi- 27. Jian, M., and Prins, R., Catal. Lett. 50, 9 (1998).
2
8. Satterfield, C. N., and Yang, S. H., Ind. Eng. Chem. Proc. Des. Dev. 23,
cates that the promotional effect of fluorine is not due to a
change in the intrinsic properties of the active sites but to an
increase in the number of sites. Fluorine apparently affects
the activity of the catalyst by modifying the morphology of
the active phases. The larger WS2 particles have more ad-
11 (1984).
2
3
3
9. Jian, M., Kapteijn, F., and Prins, R., J. Catal. 168, 491 (1997).
0. Jian, M., and Prins, R., Ind. Eng. Chem. Res. 37, 834 (1998).
1. Oliv e´ , J. L., Biyoko, S., Moulinas, C., and Geneste, P., Appl. Catal. A
19, 165 (1985).

2
00
SUN AND PRINS
3
2. Schulz, H., Schon, M., and Rahman, N. M., Stud. Surf. Sci. Catal. 27,
01 (1986).
3. Ho, T. C., Catal. Rev.-Sci. Eng. 30, 117 (1988).
51. Topsøe, H., and Clausen, B. S., Catal. Rev.-Sci. Eng. 26, 395
(1984).
52. Louwers, S. P. A., and Prins, R., J. Catal. 139, 525 (1993).
2
3
3
4. Durand, R., Geneste, P., Moreau, C., and Zmimita, N., J. Catal. 112, 53. Perot, G., Brunet, S., and Hamze, N., in “Proceedings, 9th International
4
11 (1988).
Congress on Catalysis, Calgary, 1988” (M. J. Phillips and M. Ternan,
Eds.), Vol. 1, p. 19. Chem. Institute of Canada, Ottawa, 1988.
3
3
5. Yang, S. H., and Satterfield, C. N., J. Catal. 81, 168 (1983).
6. Shanthi, K., Pillai, C. N., and Kuriacose, J. C., Appl. Catal. A 46, 241 54. Ozkan, U. S., Ni, S., Zhang, L., and Moctezuma, E., Energy Fuels 8,
(
1989).
249 (1994).
3
3
3
4
4
7. Rota, F., and Prins, R., J. Mol. Catal. A: Chem. 162, 367 (2000).
8. Nelson, N., and Levy, R. B., J. Catal. 58, 485 (1979).
9. Laine, R. M., Catal. Rev.-Sci. Eng. 35, 459 (1983).
0. Perot, G., Catal. Today 10, 447 (1991).
1. Cattenot, M., Portefaix, J. L., Afonso, J., Breysse, M., Lacroix, M., and
Perot, G., J. Catal. 173, 366 (1998).
55. Pecoraro, T. A., and Chianelli, R. R., J. Catal. 67, 430 (1981).
56. Nørskov, J. K., Clausen, B. S., and Topsøe, H., Catal. Lett. 13,
1 (1992).
57. Smith, J. W., in “The Chemistry of the Amino Group” (S. Patai, Ed.),
p. 161. Wiley Interscience, New York, 1968.
58. Yang, M. H., Grange, P., and Delmon, B., Appl. Catal. A 154, L7 (1997).
59. Boorman, P. M., Kriz, J. F., Brown, J. R., and Ternan, M., in “Proc.
4th Int. Climax Conf. on Chemistry and Uses of Molybdenum” (H. F.
Barry and P. C. H. Mitchell, Eds.), p. 192. Climax Comp., Ann Arbor,
1982.
4
4
2. Girgis, M. J., and Gates, B. C., Ind. Eng. Chem. Res. 30, 2021 (1991).
3. Meille, V., Schulz, E., Lemaire, M., and Vrinat, M., J. Catal. 170, 29
(
1997).
44. Curtis, M. D., and Druker, S. H., J. Am. Chem. Soc. 119, 1027 (1997).
45. Mcllvried, H. G., Ind. Eng. Chem. Proc. Des. Dev. 10, 125 (1971).
46. Topsøe, N. Y., and Topsøe, H., J. Catal. 84, 386 (1983).
4
60. Benitez, A., Ramirez, J., Cruz-Reyes, A., and Lopez Agudo, A.,
J. Catal. 172, 137 (1997).
7. Yang, S. H., and Satterfield, C. N., Ind. Eng. Chem. Proc. Des. Dev. 23, 61. Boorman, P. M., Kydd, R. A., Sarbak, Z., and Somogyvari, A., J. Catal.
0 (1984). 96, 115 (1985).
8. Hadjiloizou, G. C., Butt, J. B., and Dranoff, J. S., Ind. Eng. Chem. Res. 62. Boorman, P. M., Kydd, R. A., Sarbak, Z., and Somogyvari, A., J. Catal.
1, 2503 (1992). 100, 287 (1986).
9. Ozkan, U. S., Zhang, L., Ni, S., and Moctezuma, E., J. Catal. 148, 63. Ramirez, J., Cuevas, R., Lopez Agudo, A., Mendioroz, S., and Fierro,
81 (1994). J. L. G., Appl. Catal. 57, 223 (1990).
0. Satterfield, C. N., Modell, M., and Wilkens, J. A., Ind. Eng. Chem. 64. Ramirez, J., Castillo, P., Benitez, A., Vazquez, A., Acosta, D., and
Proc. Des. Dev. 19, 154 (1980). Lopez Agudo, A., Appl. Catal. 57, 223 (1990).
2
4
4
5
3
1