Chloropentafluoroethane hydrodechlorination over tungsten carbides: Influence of surface stoichiometry

Article abstract of DOI:10.1006/jcat.2001.3491

The relationship between the mode of activation and the catalytic properties of tungsten carbides was studied. Tungsten carbides were prepared by temperature-programmed reaction of WO₃ with methane/hydrogen mixture and activated according to several thermal treatments by varying the temperature or the feed composition. The acid-base characteristic increased with the extent of decarburization of the surface. The surface properties of WC could be adjusted toward the formation of unsaturated compounds or alternatively toward hydrogenolysis compounds. However, the convergence observed in the temperature-programmed selectivity curves, whatever the initial C/W ratio, showed that the sample surfaces reach similar compositions by recarburization during reaction. Such acidic properties were involved in the dehydrohalogenation of chloropentafluoroethane responsible mainly for the formation of unsaturated compounds.

Full text of DOI:10.1006/jcat.2001.3491

Journal of Catalysis 206, 358–362 (2002)
doi:10.1006/jcat.2001.3491, available online at http://www.idealibrary.com on
RESEARCH NOTE
Chloropentafluoroethane Hydrodechlorination over Tungsten Carbides:
Influence of Surface Stoichiometry
1
1
L. Delannoy, J.-M. Giraudon, P. Granger, L. Leclercq, and G. Leclercq
◦
Universit e´ des Sciences et Technologies de Lille, Laboratoire de Catalyse, UPRESA n 8010 B aˆ timent C3, 59655—Villeneuve d’Ascq Cedex, France
Received April 19, 2001; revised December 10, 2001; accepted December 10, 2001; published online February 13, 2002
sites. Presently, the origin of such acid–base sites is not well
Tungsten carbides were prepared by temperature-programmed understood. Several hypotheses can be considered: (i) their
reaction of WO3 with methane/hydrogen mixture and activated promotion could occur during the reaction by fluorine or
according to several thermal treatments by varying the tempera-
ture or the feed composition. Modification of these parameters was
oxygen accumulation at the catalyst surface; (ii) the car-
burization procedure could also develop such an acidity
expected to change the surface composition (C/W). Such surface
modifications have been related to acid–base properties observed in
the decomposition of isopropanol. Finally, it was found that such
acidic properties are probably involved in the dehydrohalogenation
of chloropentafluoroethane responsible mainly for the formation of
δ+
δ−
related to the partial ionicity of the M –C bond in the
bulk; (iii) more probably the presence of acid–base sites
could be related to the surface carbide stoichiometry since
we have shown that the proportion of dehydrohalogena-
tion products increased when the ratio C/M decreased (4).
Generally the activation step under pure hydrogen, which is
unsaturated compounds.
ꢀc 2002 Elsevier Science (USA)
Key Words: tungsten carbide; catalytic hydrodechlorination;
chloropentafluoroethane; X-ray photoelectron spectroscopy; acidic always performed before catalytic experiments to remove
properties of WC; decomposition of isopropanol.
excess polymeric carbon, which is frequently deposited at
sample surfaces during the carburization step, results in par-
tial decarburization of the carbide surface (5, 6), the extent
of which depends on the temperature, the hydrogen flow
rate, and the duration of the activation step. Consequently,
changes in the activation conditions could be an interesting
way to modify the surface carbide composition and then to
study the relationship between this composition and reac-
tion selectivity. To check whether acid–base sites are related
to a surface substoichiometry, we prepared three samples
of WC with various surface stoichiometries by changing the
activation conditions.
Earlier studies showed that group VI transition metal
carbides, in comparison to noble metals, exhibit interest-
ing catalytic properties in C–Cl bond cleavage (1–3). Re-
cently the hydrodechlorination of chloropentafluoroethane
(
CFC-115) into pentafluoroethane (HFC-125) was investi-
gated on carbide catalysts MC (with M = W or Mo) (4).
This reaction is potentially interesting for practical applica-
tions, particularly to substitute hydrofluorocarbons (HFCs)
for chlorofluorocarbons (CFCs), HFCs being less harmful
to stratospheric ozone. Among the carbides studied, it was
found that the activity of WC was lower than that of a ref-
erence Pd-based catalyst; on the other hand, their selec-
tivities were comparable for the production of HFC-125.
In fact additional by-products, mainly unsaturated com-
pounds from dehydrofluorination reactions, have also been
observed on WC compared with Pd which essentially yields
products from hydrodefluorination reactions. Such an ob-
servation suggests a bifunctional catalysis on tunsgsten and
molybdenum carbide catalysts. Indeed, hydrodehalogena-
tion reactions are activated on metallic sites while dehy-
drofluorination reactions probably take place on acid–base
All tungsten carbides were prepared using a similar con-
ventional procedure (7, 8) by temperature-programmed
carburization of WO3 (99.95% Fluka) from room tem-
◦
perature to 800 C with a carbiding mixture composed of
(
20 vol%) CH4 (99.995%) in H2 (99.995%). The catalysts
◦
were further activated in pure hydrogen at 750 or 800 C or
in a mixture composed of 1–2 vol% CH4 in H2 at 800 C for
◦
5
h(9–11)accordingtotheexperimentalconditionssumma-
rized in Table 1. Finally, the solids were cooled to room tem-
perature before reaction. The samples are labeled WCX-Y,
where X and Y represent respectively the temperature and
the composition of the feed gas used during the activation
step. WCX-H2 and WCX-CH4 stand for catalysts res-
pectively activated in pure hydrogen and in 1 to 2% CH4
in H2.
1
To whom correspondence should be addressed. E-mail: pascal.
granger@univ-lillel.fr. Fax: +33.3.20.43.65.61.
358
0
ꢀ
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.

CHLOROPENTAFLUOROETHANE HYDRODECHLORINATION OVER TUNGSTEN CARBIDES
359
TABLE 1
of carbidic carbon (13). An additional small shoulder is
observed at 284.9 ± 0.2 eV which corresponds to polymeric
carbon (14); its intensity decreases according to the se-
quence WC800-CH4 > WC800-H2 > WC750-H2. The W 4 f
Specific spectra (Fig. 1b) are similar for all the samples and show
Parameters of the Activation Treatments of the Tungsten
Carbide Samples
Total
Feed
flow Temperature Duration
composition (L/h)
area
the contributions of oxidic and carbidic tungsten species.
◦
2
Catalyst
( C)
(h)
(m /g)
The major photopeaks at 32.3 and 34.4 ± 0.2 eV corre-
WC750-H2
WC800-H2
H2
H2
8
8
8
750
800
800
5
5
5
15
14.5
15
spond respectively to W 4 f7/2 and W 4 f5/2 lines charac-
teristic of WC (14) while the W 4 f photopeak at 38.2 eV
5
/2
WC800-CH4 1–2%CH4/H2
6+
has been attributed to W species (15). This assignment
is also supported by the difference between the W 4 f7/2
and W 4 f5/2 binding energy levels, respectively, for the
more reduced component and the oxidic phase. Indeed, a
value of 5.9 eV seems to agree with the presence of WC
in the layers probed by XPS while deviations from this
value up to 6.4 eV are encountered when metal species
are still present on the surface (16). A quantitative esti-
Thecarbidiccarbonremovalduringthetreatmentinpure
hydrogen, mentioned above, is the result of two successive
steps: elimination of carbidic carbon at the surface,
→
WCsurface + (2x)H2 ← WC1−x,surface + xCH4,
[1]
followed by replenishment of the surface by carbon diffu- mation of the surface composition reflected by the atomic
sion from the bulk,
ratio C/W has been achieved considering the total surface
area of the photopic W 4 f , while only the intensity signal
→
WCbulk + WC1−x,surface ← WC1−x,bulk + WCsurface [2] related to the carbidic phase has been accounted for after
decomposition of the C 1s photopeak. The values calcu-
◦
At 800 C, the CH4 partial pressure (PCH ) at steady state
4
lated for C/W, in Table 2, emphasize previous statements
regarding the increase in carbon deficiency in the sequence
WC800-CH4 > WC800-H2 > WC750-H2. It is worth noting
that XPS analysis involves about 10 monolayers. Conse-
quently the atomic C/W ratios in Table 2 are average sto-
ichiometries which do not reflect exactly the surface com-
position of the topmost layer involved in catalysis. The
difference between the actual values of the C/W ratio of
the three samples are probably much greater than those
is slightly lower than that at equilibrium of step [1]. Con-
sequently, both steps [1] and [2] are fast, and global decar-
burization of the surface is likely to be limited. At 750 C,
◦
◦
PCH at steady state is slightly higher than at 800 C but it
4
is lower than that at equilibrium, probably because step [2]
of carbon diffusion is slower than step [1]. In such a case,
a more extensive decarburization is expected. In fact, es-
timation of the global (C/W) atomic ratios for WC800-H2
and WC750-H2, taking into account the amount of carbidic
carbon removed during the decarburization step (consid-
ering the steady-state partial pressure of methane, the flow
rate, and the duration of the treatment), led to 0.90 and
2
84.9 eV 283.3 eV
∆
E
a - C1s
b - W4f
0
.865, respectively, indicating global decarburization for
both samples, but to a slightly greater extent for WC750-
H2. Of course it is likely that the surface decarburization
involves the top surface layers more than the bulk of the
sample. On the contrary, activation with a low CH4 pressure
in hydrogen (1 to 2%) was supposed to avoid together sur-
face decarburization and polymeric carbon deposit at the
surface, considering the equilibria corresponding to tung-
sten carburization by CH4 and CH4 decomposition (12).
Summarizing the above considerations, the surface C/W
ratio should decrease in the following sequence: WC800-
CH4 > WC800-H2 > WC750-H2.
WC750-H2
WC800-H2
WC750-H2
WC800-H2
WC800-CH4
WC800-CH4
X-Ray diffraction measurements show that no struc-
tural modification occurs due to changes in the decar-
burization procedure; the usual hexagonal close-packed
compact structure is preserved for all WC samples. The
surface properties were investigated by XPS analysis.
Figure 1a illustrates the C 1s spectra obtained on tungsten
carbides after passivation by exposure to 2 vol% O2 in N2.
2
92 290 288 286 284 282 280
44 42 40 38 36 34 32 30
Binding Energy (eV)
Binding Energy (eV)
FIG. 1. (a) X-Ray photoelectron spectra of C 1s energy region for
the fresh samples. Carbidic carbon is located at 283.3 eV and polymeric
carbon at 284.9 eV. (b) X-Ray photoelectron spectra of W 4 f7/2–W 4 f5/2
doublets for the fresh samples. ꢀE represents the gap in binding energy
The main contribution at 283.3 ± 0.2 eV is characteristic between W 4 f5/2 (W6 ) and W 4 f7/2 (reduced phase).
+

3
60
DELANNOY ET AL.
TABLE 2
9%
8
7
6
5
4
%
%
%
%
%
WC750-H2
WC800-H2
WC800-CH4
Surface Composition and Binding Energy Gap
from XPS Measurements
Intensity ratio
Atomic
ꢀEa
(eV)
Catalyst
(IC/IW)
(C/W) ratio
2
WC750-H2
WC800-H2
WC800-CH4
7.8 × 10−
0.74
0.80
0.97
5.9
5.8
5.8
3%
−
2
2
8.6 × 10
2
1
%
%
10.3 × 10−
a
ꢀ
E = Eb/W4 f 7/2(W+6) − Eb/W4 f 7/2(reduced phase).
0%
1
50
200
250
300
350
Isotherm
Temperature (°C)
measured in Table 2. Nevertheless, their comparison shows
FIG. 2. Temperature-programmed and steady-state conversion
curves in the hydrodechlorination of CF3CF2Cl (CFC-115) over WC750-
H2, WC800-H2, and WC800-CH4. The loading of catalyst was 1.5 g, the
◦
that thermal treatment at 750 C in pure hydrogen leads to
◦
partial carbon surface depletion, while treatment at 800 C
−
1
volumetric space velocity was 4000 h , and the molar ratio H2/CFC-115
in a mixture containing a low concentration of CH4 pre-
vents (or reduces) such surface decarburization.
◦
was fixed at 2. The temperature was raised gradually from RT to 350 C
◦
at 1 C/min.
On the three catalysts, we studied selectivity in the hy-
drogenolysis of chloropentafluoroethane (CFC-115) and in
the conversion of isopropanol which is a good probe reac- base properties. At first glance, it may seem surprising that
tion for the study of metallic/acid–base behavior of sup- such wide differences in selectivities were observed while
ported metals (17–19); the dehydrogenation into acetone there is almost no difference in activities. This can be very
occurs on metal-like sites while the dehydration into well accounted for by considering that the primary prod-
propene or diisopropyl ether take place on acid–base uct is HCF-125, further leading to secondary elimination
sites. Consistently, this latter reaction quantifies the acid– reactions which will govern the selectivity of the reaction.
base properties of the three carbides by the selectivity in Consequently, overall conversion of CFC-115 probably de-
dehydration versus dehydrogenation. Hydrogenolysis of pends essentially on the specific surface areas, which are
chloropentafluoroethane (CFC-115) was studied under at- quite similar for all samples, while selectivity depends on
mospheric pressure in temperature-programmed experi- the proportion of acid–base sites.
ments performed in a stainless-steel fixed-bed flow reactor
As time-on-stream and temperature increase, selectiv-
described earlier (4), from room temperature to 350 C with ity in HFC-125 increases and ultimately the three catalysts
◦
◦
a constant heating rate of 1 C/min. The space velocity was exhibit similar selectivities in HFC-125, i.e., about 90%
000 h and the molar ratio H2/CFC-115 was 2. Then the (Fig. 3). This suggests that during reaction, and probably
catalysts were maintained for a few hours at 350 C. Trans- at higher temperature, surface recarburization occurs and
formation of chloropentafluoroethane leads to a complex sample surfaces reach similar compositions. Such observa-
product distribution where CF3CF2H (HFC-125), the tar- tions are in good agreement with earlier work showing that
get molecule, CF3CFH2 (HFC-134a), and CF3CH3 (HFC-
−1
4
◦
1
43a) are produced from hydrodehalogenation reactions,
1
00%
unsaturated compounds (C2H4, CF2CH2, C2F4) are pro-
duced from dehydrofluorination reactions, and alkanes
such as CH4, C2H6, and C3H8 are also formed in substan-
tial proportions. As shown in Fig. 2 and Table 3, no signif-
icant changes in steady-state activities are observable with
respect to the conditions of activation and to the surface
composition related to the atomic C/W ratio. WC800-CH4
seems to be slightly more active than the other two car-
bides. This relates to the very similar specific surface areas
of the three samples. More relevant are the changes in se-
lectivity at low temperature (Fig. 3, Table 3) with surface
C/W ratio: selectivity in HFC-125 decreases while the ra-
tio C/W decreases, whereas the formation of unsaturated
compounds and alkanes (all resulting from dehydrofluori-
nation of HFC125) follows the reverse trend. This suggests
9
8
7
6
5
0%
0%
0%
0%
0%
WC750-H2
WC800-H2
WC800-CH4
40%
30%
20%
10%
0%
1
00
150
200
250
300
350
Temperature (°C)
Isotherm
FIG. 3. Temperature-programmed and steady-state selectivity curves
in the hydrodechlorination of CF3CF2Cl (CFC-115) over WC750-H2,
that the lower the surface C/W ratio, the higher the acid– WC800-H2, and WC800-CH4.

CHLOROPENTAFLUOROETHANE HYDRODECHLORINATION OVER TUNGSTEN CARBIDES
361
TABLE 3
Conversion and Product Distribution (%) in the Hydrogenolysis of CF3CF2Cl
WC750-H2
WC800-H2
WC800-CH4
◦
◦
◦
◦
◦
◦
◦
◦
◦
2
00 C
240 C
350 C
200 C
240 C
350 C
200 C
240 C
350 C
Conversion
CF3CF2H (HFC-125)
Alkanes
0.15
39
43
0.6
45
32
7.2
72
15
0.12
44
43
0.5
52
31
7.5
73
15
0.15
72
27
0.5
75
17
8
88
5
(
CH4, C2H6, C3H8)
Unsaturated
C2H4, CF2CH2, C2F4)
Others
CF3CFH2, CF3CH3)
17
22
13
1
12
16
11
1
1
8
6
1
(
<1
<1
<1
<1
<1
<1
(
during the exposure of metallic tungsten or molybdenum (here again potential Lewis acid sites) being predominantly
to a hydrocarbon/hydrogen mixture, partial carburization exposed at the surface. Alternatively, the carbon-deficient
occurs (20, 21). Formation of a Pd–C solid solution has also carbides more sensitive to oxygen adsorption could lead to
been confirmed during the hydrogenolysis of chlorofluoro- the formation of oxicarbide species, due to the presence of
carbons on Pd-based catalyst (22, 23), suggesting that un- traces of O2 in the reagents, which also exhibit acidic prop-
der our experimental reaction conditions recarburization is erties (24–26). This hypothesis is supported by the study of
likely to occur on partially decarburized WC samples.
Kwon et al. (27), who observed stronger adsorption of oxy-
Theresultsobtainedintheconversionofisopropanolper- gen on a carbon-deficient vanadium carbide than on a sto-
◦
formed at 150 C (Pisopropanol = 0.2 atm, helium is the carrier ichiometric one. They concluded that imperfect carburiza-
gas) are reported in Fig. 4 and Table 4, where it can be tion of the surface might enhance its affinity toward oxygen
seen that the product distribution is strongly dependent on adsorption. This is emphasized by various studies reporting
the activation procedure. The proportion of acetone de- that changes in the surface compositions of transition metal
creases while that of the dehydration products increases carbides and nitrides might affect both activity and selec-
on the carbides in the following sequence: WC800-CH4, tivity (28–31). For instance, Choi (31) suggested that geo-
WC800-H2, WC750-H2, that is, when the surface C/W ra- metrical defects such as kinks, dislocations, and vacancies
tio decreases. This shows very clearly that the acid–base in deficient carbon tantalum carbides might affect their ac-
character increases with surface carbon deficiency.
tivity in the decomposition of ammonia. However, this au-
Acid–base properties could be a direct result of substo- thor did not rule out the contribution of surface electronic
ichiometry, the surface carbon vacancies being pair elec- modifications to the catalytic properties of these materi-
x+
tron acceptors (i.e., Lewis acid sites) or W carbide species als. Recently, Choi et al. (32) studied the influence of the
degree of carburization of molybdenum carbides on their
density of sites and hydrogenation activity. They found that
an increase in the activity per site in the hydrogenation of
benzene occurs when the degree of carburization, which is
related to the ability of the catalyst to selectively chemisorb
CO, increases. They concluded that the higher the carbidic
carbon content and the lower the residual oxygen content
in the Mo lattice, the higher “the noble metal-like” behav-
ior. The trends observed in our experiments agree well with
this hypothesis.
TABLE 4
Selectivity Ratio for the Decomposition of Isopropanol
Sel. dehydration products/Sel. acetone (SD/SA)
◦
◦
◦
Catalyst
150 C
165 C
180 C
FIG. 4. Product distribution at steady state in the decomposition of
isopropanol over WC750-H2, WC800-H2, and WC800-CH4 at 150 C. The
◦
WC750-H2
WC800-H2
WC800-CH4
1.1
0.5
0.02
2.8
1.2
0.03
4.9
1.9
0.08
loading of catalyst was 500 mg; the flow of isopropanol was diluted in
3
helium (40 cm /min) corresponding to a partial pressure of isopropanol of
0
.2 atm.

3
62
DELANNOY ET AL.
In conclusion, it seems obvious that modifying the ac-
8. Oyama, S. T., Catal. Today 15, 263 (1992).
. Volpe, L., and Boudart, M., J. Solid State Chem. 59, 348 (1985).
0. Lee, J. S., and Boudart, M., Catal. Lett. 8, 107 (1991).
9
tivation of tungsten carbide by changing the surface car-
bide stoichiometry converts the catalytic properties of WC
into either metallic or acid–base ones and subsequently its
selectivity toward the formation of pentafluoroethane or
alkanes and unsaturated by-products. More generally, this
work provides evidence of a relationship between the mode
of activation and the catalytic properties of tungsten car-
bides; the acid–base character seems to increase with the
extent of decarburization of the surface. It is obvious that
1
1
1. Boudart, M., Oyama, S. T., and Leclercq, L., Stud. Surf. Sci. Catal. 71,
578 (1981).
1
2. Leclercq, G., Kamal, M., Giraudon, J.-M., Devassine, P.,
Feigenbaum, L., Leclercq, L., Frennet, A., Bastin, J.-M., L o¨ fberg, A.,
Decker, S., and Dufour, M., J. Catal. 158, 142 (1996).
3. Giraudon, J.-M., Leclercq, L., Leclercq, G., L o¨ fberg, A., and
Frennet, A., J. Mater. Sci. 28, 2449 (1993).
1
1
4. Leclercq, L., Provost, M., Grimblot, J., Hardy, A. M., Gengembre, L.,
and Leclercq, G., J. Catal. 117, 371 (1989).
the surface properties of WC could be adjusted toward the 15. Vedrine, J. C., Hollinger, G., and Duc, T. M., J. Phys. Chem. 82, 1515
(
1978).
formation of unsatured compounds or alternatively toward
hydrogenolysis compounds, according to the practical ap-
plication considered. Such flexibility in the preparation pro-
cedure could be of practical interest. However, the conver-
1
1
1
1
6. Devassine, P., Ph.D. thesis, Lille, 1997.
7. Hathaway, P. E., and Davis, M. E., J. Catal. 116, 263 (1989).
8. Gervasini, A., and Auroux, A., J. Catal. 131, 190 (1991).
9. Pepe, F., Angelotti, L., and De Rossi, S., J. Catal. 118, 1 (1989).
gence observed in the temperature-programmed selectivity 20. Sinfelt, J. H., and Yates, D. C. J., Nat. Phys. Sci. 229, 27 (1971).
2
2
1. Levy, R. B., and Boudart, M., Science 181, 547 (1973).
2. Juszczyk, W., Malinowski, A., and Karpinski, Z., Appl. Catal. A 166,
curves, whatever the initial C/W ratio, probably shows that
the sample surfaces reach similar compositions by recar-
burization during reaction.
311 (1998).
2
3. van de Sandt, E. J. A. X., Wiersma, A., Makkee, M., van Bekkum, H.,
and Moulijn, J. A., Appl. Catal. A 155, 59 (1997).
REFERENCES
24. Ribeiro, F. H., Della-Betta, R., Boudart, M., Baumgartner, J., and
Iglesia, E., J. Catal. 130, 86 (1991).
1
. Morikawa, S., Yoshitake, M., and Tatemasu, S., Japanese Patent
319442 (1989).
25. Iglesia, E., Baumgartner, J. E., Ribeiro, F. H., and Boudart, M.,
J. Catal. 131, 523 (1991).
1
2
3
4
. Dhandapani, B., and Oyama, S. T., Catal. Lett. 35, 353 (1995).
. Sherif, F. G., U.S. Patent 5,426,252 (1995).
. Delannoy, L., Giraudon, J.-M., Granger, P., Leclercq, L., and
Leclercq, G., Catal. Today 59, 231 (2000).
26. Keller, V., Wehrer, P., Garin, F., Ducros, R., and Maire, G., J. Catal.
166, 125 (1997).
27. Kwon, H., Thompson, L. T., Eng, J., Jr., and Chen, J. G., J. Catal. 190,
60 (2000).
5
. Frennet, A., Leclercq, G., Leclercq, L., Maire, G., Ducros, R., Jardinier-
Offergeld, M., Bouillon, F., Bastin, J.-M., L o¨ fberg, A., Blehen, P.,
Dufour, M., Kamal, M., Feigenbaum, L., Giraudon, J.-M., Keller, V.,
Wehrer, P., Cheval, M., Garin, F., Kons, P., Delcombe, P., and Binst, L.,
in “Proceedings, 10th International Congress on Catalysis” (L. Guczi,
F. Solymosi, and P. Tetenyi, Eds.), pp. 927–937. Akad e´ miai Kiad o` ,
Budapest, 1993.
28. Toth, L. E., “Transition Metal Carbides and Nitrides.” Academic Press,
New York, 1971.
29. Leclercq, L., Provost, M., Pastor, H., and Leclercq, G., J. Catal. 117,
384 (1989).
30. Chen, J. G., Fruhberger, Weisel, M. D., Baumgartner, J. E., and
De Vries, B., in “The Chemistry of Transition Metal Carbides and
Nitrides.” Blackie, New York, 1996.
6
. Giraudon, J.-M., Devassine, P., Lamonier, J.-F., Delannoy, L.,
Leclercq, L., and Leclercq, G., J. Solid State Chem. 154, 412 (2000).
. Lee, J. S., Oyama, S. T., and Boudart, M., J. Catal. 106, 125 (1985).
31. Choi, J.-S., Appl. Catal. A 184, 189 (1999).
32. Choi, J.-S., Bugli, G., and Dj e´ ga-Mariadassou, G., J. Catal. 193, 238
(2000).
7