Evolution of the surface sulfur composition of a Ru sulfide particle during CH3SH condensation reaction: Sulfur migration from the bulk to the surface

Article abstract of DOI:10.1039/b005909i

The condensation of methanethiol (CH₃SH) into dimethylsulfide (CH₃SCH₃), which is a useful test reaction for probing the acid-base character of transition metal sulfide catalysts for carbon-heteroatom hydrogenolysis reactions, shows a systematic deactivation process at the beginning of the reaction before reaching steady-state activity. The solid surface of RuS₂ catalyst in equilibrium with the gas phase was examined by various techniques with the aim of understanding the mechanism responsible for this loss of activity, e.g. coke formation, poisoning by the reactant or the products, change in particle size, etc. X-Ray diffraction and elemental analysis showed neither the formation of carbonaceous deposits nor a sintering effect during the catalytic run. In contrast, a modification of the surface composition of the catalyst induced by the reaction was observed using temperature programmed reduction (TPR). This technique showed that large amounts of H₂S are detectable after performing the catalytic test. Sulfur mass balance analyses demonstrate that these sulfur species are an integral part of the total sulfur content of the ruthenium sulfide particles. These surface modifications arise from sulfur migration from the bulk to the surface of the particles during reation. Based on these experimental data and on a crystallographic model developed for RuS₂, it is proposed that the methanethiol condensation reaction proceeds on monovacant Ru sites.

Full text of DOI:10.1039/b005909i

View Article Online / Journal Homepage / Table of Contents for this issue
Evolution of the surface sulfur composition of a Ru sulÐde particle
during CH SH condensation reaction: sulfur migration from the bulk
3
to the surface
Gilles Berhault and Michel Lacroix*
Institut de Recherches sur la Catalyse, (CNRS UPR 5401) 2, Avenue Albert Einstein, 69626
V illeurbanne cedex, France. E-mail: lacroix=catalyse.univ-lyon1.fr; Fax: ]33 (0)4 72 44 53 90
Received (in Strasbourg, France) 20th July 2000, Accepted 3rd November 2000
First published as an Advance Article on the web 4th January 2001
The condensation of methanethiol (CH SH) into dimethylsulÐde (CH SCH ), which is a useful test reaction for
3
3
3
probing the acidÈbase character of transition metal sulÐde catalysts for carbonÈheteroatom hydrogenolysis
reactions, shows a systematic deactivation process at the beginning of the reaction before reaching steady-state
activity. The solid surface of RuS catalyst in equilibrium with the gas phase was examined by various techniques
2
with the aim of understanding the mechanism responsible for this loss of activity, e.g. coke formation, poisoning by
the reactant or the products, change in particle size, etc. X-Ray di†raction and elemental analysis showed neither
the formation of carbonaceous deposits nor a sintering e†ect during the catalytic run. In contrast, a modiÐcation of
the surface composition of the catalyst induced by the reaction was observed using temperature programmed
reduction (TPR). This technique showed that large amounts of H S are detectable after performing the catalytic
2
test. Sulfur mass balance analyses demonstrate that these sulfur species are an integral part of the total sulfur
content of the ruthenium sulÐde particles. These surface modiÐcations arise from sulfur migration from the bulk to
the surface of the particles during reation. Based on these experimental data and on a crystallographic model
developed for RuS , it is proposed that the methanethiol condensation reaction proceeds on monovacant Ru sites.
2
The industrial hydrotreating process is currently used in all
reÐneries for cleaning up the various oil fractions by removing
the S, N, O and the metals present in organic molecules.
Depending on the nature of the heteroatom, hydrotreating
refers to hydrodesulfurization (HDS), hydrodenitrogenation
that this reactant dissociates upon chemisorption, leading to
the formation of a proton and a thiolate-type moiety. The het-
erolytic character of the adsorption implies the presence of
both a Lewis acidic site to Ðx the organic fragment and the
presence of a basic sulfur center to trap the positively charged
hydrogen species. The acidic character of the sulfur vacant
sites was conÐrmed by pyridine chemisorption experiments
and a nice correlation was obtained between the Lewis acid
site density and the initial rate of the MeSH condensation
reaction.10 However, the evolution of the condensation rate
vs. time on stream shows that the catalyst deactivates at the
beginning of the reaction. In heterogeneous catalysis, catalyst
deactivation often arises from coke deposition formed by
oligo- or polymerization of unsaturated hydrocarbons at the
surface of the particle.11
(
(
HDN), hydrodeoxygenation (HDO), and hydrodemetallation
HDM) reactions, which operate at temperatures ranging from
5
73 to 673 K and under several atmospheres of hydrogen.1
Supported nanoparticles of transition metal sulÐdes are effi-
cient systems for catalyzing all these reactions. Noble metal
sulÐdes, namely RuS and Rh S , are more active solids than
MoS -based industrial catalysts.2,3 Fundamental work based
on the determination of the catalytic properties of supported
2
2 3
2
or bulk Mo or Ru sulÐdes has shown that a fully sulÐded
material is not active in catalysis and that sulfur vacancies
must be created at the surface of the particle to provide cata-
lytically active sites.4h6 In a Ðrst approach, these results may
be interpreted in terms of SabatierÏs principle for which the
metalÈsulfur bond strength should be neither too weak nor
too strong so that the surface can easily form and regenerate
surface vacancies during the reaction cycle.7,8 However, all the
data were obtained using frozen states of reduced catalysts
and testing their catalytic properties in model reactions such
as the H ÈD isotopic exchange and oleÐn or dioleÐn hydro-
In our case this type of surface poisoning appears unlikely,
taking into account the light molecules involved in the reac-
tion and the relatively low reaction temperature (673 K).
However, as sulfur-containing molecules are used in this
model reaction, H S chemisorption, resulÐdation or solid
2
reconstruction may modify the surface composition, leading to
a diminution of the number of active sulfur vacant sites and
therefore a decrease of the catalytic activity. The aim of this
contribution is to discriminate between these various pos-
sibilities by controlling the sulfur-to-metal ratio of several
reduced states of Ru sulÐde. This could be achieved by mea-
suring the sulfur mass balance during or at the end of the
reaction using the temperature programmed reduction tech-
nique equipped with speciÐc and sensitive detectors.
2
2
genation. As these reactions proceed at temperatures lower
than those required to adjust the catalyst composition upon
hydrogen treatment, it was assumed that, in the absence of
any sulfur atoms in the feed, the catalyst composition was not
a†ected by the catalytic run.
The chemical properties of these vacant sites were recently
characterized using the condensation of methanethiol (CH SH
3
Experimental
The catalyst was prepared by impregnation of a 300 m2 g~1
BET area silica (Grace Davison 432) with an aqueous solution
or MeSH) into dimethyl sulÐde (CH SCH or MeSMe) as the
3
3
model reaction.9 IR spectroscopic studies of the interaction of
MeSH with various reduced RuS states have demonstrated
2
308
New J. Chem., 2001, 25, 308È312
DOI: 10.1039/b005909i
This journal is ( The Royal Society of Chemistry and the Centre National de la Recherche ScientiÐque 2001

View Article Online
of RuCl É xH O (Johnson Matthey).9 The impregnated silica
Results
3
2
was dried at 383 K and sulÐded at 673 K in a 15% H SÈ85%
2
N atmosphere in order to avoid the formation of an interme-
2
Reduction and characterization of the catalysts
diate metallic phase that is difficult to sulÐde.12 After this acti-
Elemental analysis indicates that the unreduced solid contains
.4 wt.% of sulfur, which corresponds to an initial S : Ru ratio
vation procedure, the catalyst was cooled to room
temperature under the sulÐding atmosphere, Ñushed with an
oxygen-free nitrogen Ñow and stored in sealed bottles. The Ru
loading was 7.5 wt.%.
6
of 2.7. As reported in Table 1, this overstoichiometric sulfur is
rather strongly bonded to the surface since its complete
removal is observed only after heating the solid under hydro-
gen at 523 K while hydrogen treatment at 673 K is required to
completely reduce the sulÐded phase.
Hydrogen reductive treatments and catalytic activity mea-
surements were performed in the same Ñow microreactor con-
nected to a gas chromatograph (HP 5890A) equipped with a
Ñame photometric detector (FPD) and a Ñame ionization
X-Ray di†raction patterns are reported in Fig. 1. The large
peak centered at 2h \ 20¡ and the shoulder at 2h \ 9¡ are due
to the silica support and the Ðlm used to protect the sample
from air oxidation during the measurement. Besides these dif-
fracting lines the unreduced solid presents three other peaks
located at 2h \ 32, 46 and 54¡, respectively assigned to the
detector (FID) used, respectively, for the detection of H S and
2
hydrocarbons. After loading the sample into the reactor, the
solid was Ñushed under nitrogen for 15 min and then con-
tacted with a hydrogen Ñow of 100 cm3 min~1 at room tem-
perature. The reactor was then heated up to the desired
[
200], [220] and [311] faces of the RuS pyrite phase. A
reduction temperature (T ) using a heating rate of 2 K min~1
2
r
hydrogen treatment performed up to 573 K modiÐes neither
the position nor the broadening of these di†raction lines. Thus
the structure and the particle size of the solid (ca. 50 AŽ ) appear
to be stable, in fairly good agreement with previously reported
and left at this temperature for 2 h. The amount of H S rel-
2
eased by the solid during the reduction process was quantiÐed
by calibrating the FPD detector with a known concentration
of H S (1153 ppm) diluted in hydrogen.
2
high resolution electron microscopy results9 (see Table 1). At
After this in situ reduction step, hydrogen was replaced by
623 K, the appearance of a new peak at 2h \ 43¡, correspond-
nitrogen and the reactor temperature was Ðxed at 473 K. The
reduced catalysts were then submitted to a Ñow containing 5.3
mol% of MeSH diluted in nitrogen. A HP-5 capillary column
ing to the [101] plane of the metallic phase, shows that the
system becomes biphasic. At higher reduction temperatures,
the di†raction lines of the sulÐde phase disappear and only
those corresponding to metallic Ru are detected at 2h B 38.5
(
crosslinked Ph-Me silicon, 50 m ] 0.32 mm ] 0.25 lm) main-
tained at 333 K allowed the analysis of the reaction mixture.
[
100], 42.2 [002], 44.0 [101], 58.4 [102] and 69.5 [110]. From
Besides the reactant (MeSH), only H S and dimethyl sulÐde
2
these data, it is clear that the silica-supported ruthenium
were detected under these experimental conditions. Conver-
sions were kept lower than 10% by adjusting the total Ñow
rate in order to avoid possible mass and heat transfer limi-
tations.
X-ray di†raction measurements were performed to deter-
mine the structural stability as well as the modiÐcation of the
particle size of the various reduced solids, either after
reduction or after the solid had catalyzed the condensation
reaction. XRD patterns were collected in the 2h domain
ranging from 5 to 80¡ using a Siemens D500 di†ractometer
together with a Cu-Ka radiation source operating at 45 kV
and 35 mA (j \ 1.5418 A
Ž
). Particle size (U) was determined
using the Scherrer equation:
K ] j ] 57.3
U \
(1)
cos h ] Jb2 [ b2
where b is the width at half-height of the experimental line.
The shape factor K was taken equal to 0.9 according to a
previous electron microscopy study showing that the sulÐded
RuS particles are nearly spherical.13 The correction factor b
2
was determined using quartz powder as the internal standard
(
b \ 0.11¡). The reduced solids were transferred to the sample
holder in a glove box and adhesive tape was used to protect
them from air oxidation.
Fig. 1 XRD patterns recorded for the catalyst reduced at various
temperatures.
Table 1 Degree of reduction and mean particle size after hydrogen treatment of RuS ÈSiO at various reduction temperatures (T ). Data in
2
2
r
parentheses correspond to the standard deviation of the particle size distribution
Mean particle size/A
Ž
T /K
Evolved H S/lmol g~1
S : Ru
XRD
HREM9
r
2
None
None
241
400
521
902
1644
2005
2.7
2.38
2.16
2
1.49
0.49
0
52
52
54
ND
53
60
80È100
36(7)
ND
37(8)
4
4
5
5
6
6
23
73
23
73
23
73
ND
43(11)
65(16)
110(28)
New J. Chem., 2001, 25, 308È312
309

View Article Online
sulÐde maintains its morphology up to 573 K and when
sintering occurs, the increase in the particle size is accompa-
nied by the formation of a metallic phase. This is in agreement
surface poisoning or catalyst resulÐdation by the hydrogen
sulÐde formed in the reaction:
2
MeSH ] MeSMe ] H S
(2)
with the presence of only two phases, RuS
in the RuÈS phase diagram.
2 and metallic Ru,
2
The poisoning e†ect of H S when progressive amounts of
this gas are added in the inlet feed1 is a well-known phenome-
2
Catalytic properties
non in hydrotreating catalysis. However, in such a situation, a
di†erence between the MeSMe and the H S outlet Ñows
should be detected. Fig. 3 reports the Ñow di†erence vs. reac-
Preliminary experiments have shown that the silica carrier,
either sulÐded or reduced at a temperature as high as 773 K,
is not active in this reaction. Fig. 2 reports the evolution of the
reactant conversion as a function of time on stream for the
catalyst reduced at various temperatures. The NR curve refers
to the activity of the unreduced solid. Its activity was deter-
mined after heating the catalyst up to the reaction tem-
perature (473 K) in the presence of a nitrogen Ñow. During
this step, the solid does not release any detectable amount of
2
tion time for the catalyst reduced at 573 K. Results evidence
that only a slight deviation is observed since the integrated
area underneath the curve represents only 20 lmol g~1 of
sulfur. Lower values were observed on the samples reduced at
other temperatures. However, the possible resulÐdation
process could be slow enough so that sulfur balance deviation
can not be detected.
In order to address this point, we have undertaken a set of
consecutive temperature programmed reduction (TPR) experi-
ments. The following methodology was used.
H S, suggesting that its activity arises from the existence of
2
sulfur vacant sites already present on the unreduced sample.
Fig. 2 shows that the activity increases upon reduction up to
Step 1: the catalyst was reduced at T as described in the
T \ 473 K, and then declines for higher reduction tem-
r
r
experimental section. The amount of sulfur eliminated during
peratures. The activity enhancement could be ascribed to the
this step corresponds to the value reported in Table 1. Step 2:
the temperature was lowered to the reaction temperature and
the catalytic test was performed over 6 h. Some blank experi-
ments were also carried out by replacing the reactant mixture
with a pure nitrogen or hydrogen Ñow. Step 3: the reaction
feed was replaced by a nitrogen Ñow for 15 min. During this
step only a few ppm of MeSH and MeSMe were detected by
online mass spectroscopy. An intermediate TPR was then per-
formed by heating the solid from the reaction temperature
creation of more numerous vacant sites accessible to the thiol-
ate species formed by the heterolytic dissociation of MeSH.
For higher degrees of reduction, the amount of sulfur atoms in
the lattice decreases as well as the number of basic centers
required for trapping the H` fragment. Therefore, the higher
activity observed for T \ 473 K, leading to a degree of
r
reduction close to 20%, corresponds to a maximum in the
number of acidÈbase paired sites.9 Besides this activity depen-
dence towards solid reduction, Fig. 2 also shows distinct deac-
tivation proÐles. The unreduced catalyst as well as the solid
reduced at 423, 623 and 673 K only exhibit a slight deactiva-
tion by comparison to the other samples for which the steady-
state activity is about 25% lower than the initial activity.
However, it should be noted that this deactivation does not
(
473 K) up to T in Ñowing hydrogen and remaining at this
r
temperature until the complete disappearance of the H S
signal. Step 4: a third TPR was done by heating the catalyst
2
from T to 773 K in order to completely transform the sulÐded
r
phase into the metallic one. A schematic depiction of these
various steps is presented in Fig. 4 using as an example the
data obtained for the solid reduced at 523 K.
modify the activity ranking against T . Accordingly, the dimi-
r
nution of catalyst activity should reÑect a decrease in the
The total sulfur content is obtained by adding the amount
of sulfur liberated during the three consecutive TPRs. For the
example illustrated in Fig. 4, this amount (2024 lmol g~1) is
identical within the experimental error to that determined by
chemical analysis (2005 lmol g~1). As reported in Table 2,
similar results were observed irrespective of the degree of
reduction resulting from step 1. These data indicate that the
solid does not retain sulfur during the reaction course. Conse-
number of active sites, leading to a lower number of sulfur
vacancies per surface Ru cation.
Origin of the deactivation
The main cause of deactivation in heterogeneous catalysis is
due to sintering and/or coke formation. In our study, these
hypotheses are unlikely to be due to the low temperature of
reaction (T \ 473 K). This was conÐrmed by XRD proÐles
quently, H S poisoning and/or solid resulÐding do not explain
2
that do not di†er before and after the CH SH condensation
the observed deactivation process. If the reduced catalysts are
3
reaction and by elemental analysis, which does not reveal the
contacted with hydrogen or an inert gas during step 2, no H S
2
presence of carbonaceous deposits on the catalyst surface after
reaction at 473 K. Consequently, both sintering and coke for-
mation can be excluded as a source of deactivation. Another
possible cause of catalyst deactivation would be to envisage
is released during the intermediate TPR (step 3), suggesting
that an equilibrium between the gas phase and the solid is
attained during the reduction step. In contrast, a signiÐcant
amount of H S is eliminated in step 3 if the various solids
2
Fig. 2 Evolution of the MeSH conversion as a function of time on
stream for the catalyst reduced at various temperatures.
Fig. 3 The Ñow di†erence between MeSMe and H S vs. time on
2
stream (initial state: solid reduced at 573 K).
310
New J. Chem., 2001, 25, 308È312

View Article Online
data calculated from the model, the estimated number of
vacancies (V ) per superÐcial Ru ion (Ru ) is 0.54 [V /Ru \
S
S
(
2.9 [ 2.7)/0.37]. This proportion of anionic vacancies might
reasonably explain the relatively high activity of the unre-
duced sample. As shown in Fig. 2, the deactivation of this
catalyst is not really important, suggesting that the V /Ru
ratio stays fairly constant during the reaction course.
S
The data reported in Fig. 2 also show that the most active
catalyst was obtained after solid reduction at 473 K. At this
temperature the composition of the sulÐded phase corre-
sponds to RuS
and the V /Ru ratio attains 2 [V /Ru \
2
.16
S
S
(
2.9 [ 2.16)/0.37] at the end of the reduction process. As the
deactivation is marked, this V /Ru value should be corrected
S
for the amount of migrating sulfur if we assume that the latter
is the main cause of activity loss. Unfortunately, the amount
of sulfur migration from the bulk to the surface is not known
for this reduced sample since its determination requires a
reduction temperature higher than that of the reaction tem-
perature. However, Fig. 2 indicates that solids reduced at 473
and 523 K exhibit the same deactivation proÐles, suggesting a
rather similar amount of migrating sulfur. With this assump-
Fig. 4 Schematic diagram representing the experimental procedure
used for calculating the amount of sulfur migrating from the bulk to
the surface of the catalyst. Step 1: reduction at 523 K; Step 2: reac-
tion at 473 K; Step 3: TPR from 473 to 523 K; Step 4: TPR from 523
to 773 K.
tion, the corrected V /Ru ratio for the catalyst reduced at 473
S
K drops to 1.15, which is twice that over the unreduced
sample. This relative proportion of vacant sites between both
solids Ðts fairly well with their relative steady state activities
since the activity of the solid reduced at 473 K is twice that of
the unreduced solid.
have catalyzed the condensation reaction. As this amount
does not alter the overall sulfur content, the composition of
the surface was modiÐed by sulfur migration from the bulk to
the particle surface during the catalytic run.
Conclusion
Discussion
The present work gives the Ðrst experimental evidence of the
migration of sulfur species from the bulk to the surface of
The structure of RuS can be depicted as a modiÐed NaCl
structure with two interpenetrating fcc sublattices, one con-
2
small supported RuS particles. This process occurrs when
taining the Ru atoms and the other the SÈS pairs.14 Taking
into account this crystal structure and the mean particle size
determined by HREM, the fraction of the ruthenium and
sulfur ions present at the surface of the supported nanoparti-
cle can be evaluated using the crystallographic model pro-
posed by Geantet et al.15h17 According to this model, based
x
partially reduced states are contacted with sulfur organic sub-
strates. The role of a reactant, the nature of the migrating
species as well as the mechanism of this sulfur mobility are
still unknown. It could be envisaged that during the approach
of the acidic MeSH molecule a change in the surface potential
may be occur. This unstable surface intermediate state may
then decrease its surface energy by reinforcing the population
of sulfur basic centers, allowing for complete stabilization of
the system by chemisorption of the reactant. Indeed, the aug-
mentation of the concentration of sulfur surface atoms is a
favorable process, which allows trapping of the proton formed
by a heterolytic dissociation of MeSH. This emphasizes the
difficulties of characterizing sulÐded catalysts under working
experimental conditions.
This study has shown that the Lewis acidic site required for
the transformation of MeSH into MeSMe is likely a mono-
vacant Ru site. With respect to spectroscopic techniques often
used to characterize the interaction of probes with solid sur-
faces, the main advantage of utilizing a reaction is to identify
the properties of solids under experimental conditions close to
those used in the industrial hydrotreating applications of these
sulÐded materials.
on a three-dimensional growth of the fcc lattice of the RuS
pyrite structure, the fraction of surface Ru cations represents
2
37% of the total Ru content. For this high surface-to-volume
ratio, the ruthenium ions located at the corners and edges and
those at the surface of the particle are sulfur deÐcient and may
possess 3, 2 and 1 unsaturations, respectively. These coordi-
natively unsaturated sites might accommodate extra sulfur
atoms to reach the six-fold coordination state during catalyst
preparation, namely because of the rich sulfur atmosphere
necessary to transform the supported Ru precursor into a sul-
Ðded state. From this model, a fully sulÐded particle of 36 A
Ž
would lead to a S/Ru ratio of 2.9. This value is higher than the
experimental S/Ru ratio determined by chemical analysis, sug-
gesting that the unreduced catalysts might possess some
vacant sulfur sites able to chemisorb electron-donating mol-
ecules such as MeSH and pyridine. Taking into account the
sulfur content determined by chemical analysis as well as the
Acknowledgements
Table 2 Amount of H S released during the consecutive TPR experi-
2
ments
This work was carried out within the scope of a scientiÐc
program of collaboration between France and China (CNRS
PICS No. 299). G. B. is greatly indebted to the French Minis-
try of Education for a Ph.D. grant.
Evolved H S/lmol g~1
2
Total S content
T /K
Step 1
Step 3
Step 4
lmol g~1
r
None
None
521
902
1186
1644
2005
È
231
464
348
246
None
detectable
È
1273
658
486
139
None
detectable
2005
2024
2026
2020
2029
2005
References
5
5
5
6
6
23
73
93
23
73
1
H. TopsÔe, B. S. Clausen and F. E. Massoth, Hydrotreating
CatalysisÈCatalysis Science and T echnology, SpringerÈVerlag,
Berlin, 1996, vol. 11.
2
3
T. A. Pecoraro and R. R. Chianelli, J. Catal., 1981, 67, 430.
M. Lacroix, N. Boutarfa, C. Guillard, M. Vrinat and M. Breysse,
J. Catal., 1989, 120, 473.
New J. Chem., 2001, 25, 308È312
311

View Article Online
4
5
A. Tanaka, Adv. Catal., 1985, 33, 99.
11 E. Furimsky and F. E. Massoth, Catal. T oday, 1999, 52, 381.
A. Wambeke, L. Jaloviecki, S. Kasztelan, J. Grimblot and J. P.
Bonnelle, J. Catal., 1988, 109, 320.
12 O. Knop, Can. J. Chem., 1963, 41, 1838.
13 B. Moraweck, G. Bergeret, M. Cattenot, V. Kougionas, C.
Geantet, J.-L. Portefaix, J. L. Zotin and M. Breysse, J. Catal.,
1997, 165, 45.
6
M. Lacroix, C. Mirodatos, M. Breysse, T. Decamp and S. Yuan,

in Proceedings of the 10th International Congress on Catalysis,
eds. L. Guczi, F. Solymosi and P. Tetenyi, Elsevier, Budapest,
993, pp. 597È609.


14 O. Sutarno, O. Knop and I. G. Reid, Can. J. Chem., 1967, 45,
1
1391.
7
8
9
0
H. Toulhoat, P. Raybaud, S. Kasztelan, G. Kresse and J. Hafner,
Prepr.-Am. Chem. Soc., Div. Pet. Chem., 1997, 42, 114.
P. Raybaud, J. Hafner, G. Kresse and H. Toulhoat, J. Phys.:
Condens. Matter, 1997, 9, 11107.
15 C. Geantet, C. Calais and M. Lacroix, C. R. Seances Acad. Sci.,

Ser. II, 1992, 315, 439.
16 J. A. De Los Reyes, M. Vrinat, C. Geantet and M. Breysse, Catal.
T oday, 1991, 10, 645.
G. Berhault, M. Lacroix, M. Breysse, F. Mauge
and L. Qu, J. Catal., 1997, 170, 37.
G. Berhault, F. Mauge, J.-C. Lavalley, M. Lacroix and M.

, J.-C. Lavalley
17 C. Dumonteil, M. Lacroix, C. Geantet, H. Jobic and M. Breysse,
J. Catal., 1999, 187, 464.
1

Breysse, J. Catal., 2000, 189, 431.
312
New J. Chem., 2001, 25, 308È312