Kinetics and mechanisms of n-propylbenzene hydrodealkylation reactions over Pt(Sn)/SiO2 and (Cl-)Al2O3 catalysts in reforming conditions

Article abstract of DOI:10.1006/jcat.2002.3691

Kinetics of n-propylbenzene hydrodealkylation (HDA) was studied over Pt/SiO₂ and PtSn/SiO₂ and Al₂O₃ and 1 wt % Cl-Al₂O₃ under reforming conditions (773 K, 5 bars, and H₂/HC = 5:

Full text of DOI:10.1006/jcat.2002.3691

Journal of Catalysis 210, 431–444 (2002)
doi:10.1006/jcat.2002.3691
Kinetics and Mechanisms of n-Propylbenzene Hydrodealkylation
Reactions over Pt(Sn)/SiO and (Cl–)Al O Catalysts
2
2
3
in Reforming Conditions
∗
,
∗
∗
∗
St e´ phanie Toppi, † Cyril Thomas, C e´ line Sayag, Dominique Brodzki, Fabienne Le Peltier,†
1
Christine Travers,† and G e´ rald Dj e´ ga-Mariadassou
∗
Laboratoire de R e´ activit e´ de Surface, UMR CNRS 7609, Universit e´ Pierre et Marie Curie, 4 Place Jussieu, 75252 Paris Cedex 05,
France; and †Institut Fran c¸ ais du P e´ trole, 1 et 4 Avenue de Bois-Pr e´ au, 92852 Rueil-Malmaison Cedex, France
Received March 11, 2002; revised May 24, 2002; accepted May 24, 2002
high aromatic content; this unit is called an aromizing unit.
Kinetics of n-propylbenzene hydrodealkylation (HDA) was in- The refiners’ interest is therefore to control and to orien-
vestigated over model catalysts under reforming operating condi- tate these HDA reactions depending on the aromatic levels
tions (773 K, 5 bars, and H2/HC = 5). Silica-supported platinum
desired.
catalysts (Pt/SiO2 and PtSn/SiO2) and alumina materials (Al2O3
In both cases (reforming or aromizing), typical reforming
catalystsconsistofplatinumsupportedonachlorinatedalu-
mina carrier with either tin or rhenium added as promoters.
and 1 wt% Cl–Al2O3) were chosen to evaluate the influence of the
metallic and acidic phases on the HDA reactions, respectively. The
kinetic study showed (i) that the HDA processes occurred to a sig-
Except for few recent patents (1–4), HDA reactions have
nificant extent over both catalytic phases under identical operat-
ing conditions and (ii) that hydrodealkylated products (benzene,
scarcely been studied either on bifunctional catalysts or un-
der reforming operating conditions. Indeed, dealkylation
reactions have been studied as main reactions in the pres-
toluene, and ethylbenzene) were formed through concurrent path-
ways whatever the nature of the catalytic phase. In addition, prod-
uct intermediates were identified and mechanisms were suggested ence of either steam or hydrogen using supported group
to explain the HDA reactions over either different patches of metal- VIII metals as catalysts and toluene as a model molecule
lic sites or different acid sites (Brønsted and Lewis). Finally, as (5–18). Moreover, all these experiments were carried out
compared to toluene and ethylbenzene, the formation rate of ben-
zene was the highest over the acidic function and the lowest over
at atmospheric pressure and at temperatures ranging from
573 to 773 K (5–18), whereas typical reforming processes
Pt/SiO2.
ꢀc 2002 Elsevier Science (USA)
are performed from 743 to 803 K with a total pressure rang-
ing from 2.5 to 25 bars. In addition, relatively few studies
using alkylbenzenes other than toluene as model molecules
are reported in the literature (19–21).
Key Words: kinetics; mechanisms; n-propylbenzene; hydrode-
alkylation; hydrogenolysis; cracking; Pt/SiO2 and PtSn/SiO2 cata-
lysts; chlorinated alumina; reforming; benzene.
In the work of Grenoble (7), the kinetic study of HDA
was considered as a selective hydrogenolysis reaction over
alumina-supported platinum catalysts. However, for re-
1
. INTRODUCTION
Content of aromatic compounds in gasoline has in- forming catalysts that present an acidic function brought
creased both because of the high octane number of aro- about by the chlorinated alumina, the catalytic cracking re-
matic compounds, and because use of tetraethyl lead has action cannot be ruled out.
been restricted due to environmental concerns. These aro-
In the current paper, the main objective is to evaluate the
matics (such as benzene, toluene, and xylenes) are mainly role of metallic and acidic functions in the kinetics and the
provided in the refinery by the reforming unit through the mechanisms of HDA of n-propylbenzene, (n-PB) used as a
hydrodealkylation (HDA) reaction which consumes hydro- model molecule. For this purpose, in contrast with previous
gen. However, recent specifications for gasoline drastically published results (5–21), this work was carried out under in-
limit its aromatic content and new processes have been de- dustrial conditions (773 K, 5 bars total pressure, H /HC = 5)
2
veloped, leadingtoanaromatic-freegasolinebase. Inpetro- over model catalysts. Silica-supported platinum catalysts
chemistry, the demand for aromatics remains strong, and in with or without tin added as a promoter (Pt/SiO and
2
this case, this operating conditions are adapted to produce PtSn/SiO ) were chosen to evaluate the effect of the metal-
2
lic function on the aromatic distribution, whereas 0 and
1 wt% Cl–Al O catalysts were used for the acidic cata-
2 3
lysts. It is worth noting that on acid sites HDA is defined
1
To whom correspondence should be addressed. Fax: 33 1 44 27 60 33.
E-mail: djega@ccr.jussieu.fr.
431
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

4
32
TOPPI ET AL.
as a catalytic cracking reaction whereas on metallic sites and a mean particle size diameter could be estimated from
it is defined as a hydrogenolysis reaction. Particular atten- the following formulae: D = 0.9/d, where D and d are the
tion is paid to the influence of either the metallic function metallic dispersion (%) and the mean particle diameter
(
Pt/SiO2, PtSn/SiO2) or the acidic one (Al2O3, Cl–Al2O3) (nm), respectively (22). This therefore corresponds to a
on the formation of benzene.
mean particle diameter of 2.4 nm for the unpromoted cata-
lyst. According to the work of Humblot et al. (23), a slight
increase of the particle size is expected for the tin-promoted
catalyst since this latter catalyst was synthesised with the
silica-supported platinum catalyst as a starting material,
and on which the formation of PtSn alloys was reported in
previous studies (23, 24). This latter alloying phenomenon
is accountable for the observed decrease in the Pt% ME.
2
. EXPERIMENTAL
2
.1. Materials
GOD 200 γ -Al2O3 (BET specific surface area 198 m g ,
grain size 1.0–1.5 mm) and Grace X254 silica (BET specific
2
−1
2
−1
surface area 540 m g , grain size 1–3 mm) were used as
supports. The Pt/SiO2 catalyst (0.5 wt% Pt) was prepared
according to the excess impregnation technique for the sil-
ica carrier with an ammonia solution of platinum dihydrox-
2.3. Kinetic Measurements
A commercial n-PB (Fluka, 98% pure) was used as a re-
otetramine (Pt(NH3)4(OH)2, Johnson Matthey). After dry- actant without any further purification. Liquid n-PB was
ing at 353 K overnight, the catalyst was calcined in flowing delivered to the catalytic device through a high-pressure
−
1
−1
air (1 L h
g
) at 673 K for 2 h and then reduced in piston pump (Gilson 307). The hydrogen flow and the total
catalyst
flowing hydrogen at 773 K for 2 h. The PtSn/SiO2 catalyst pressure were controlled by a mass flow controller (Brooks
0.5 wt% Pt and 0.2 wt% Sn) was prepared by impregnation 5850 TR) and back pressure regulator (Brooks 5866), re-
of the Pt/SiO2 catalyst with an aqueous solution of Me3SnCl spectively.
(
(
Aldrich). After drying at 353 K overnight, the catalyst was
Reactions were carried out using 0.8 g of sample in a
) at 773 K for 2 h fixed-bed microreactor at a temperature of 773 K and a
−1
−1
calcined in flowing air (1 L h
g
catalyst
and then reduced in flowing hydrogen at 773 K for 2 h. The total pressure of 5 bars. Kinetic measurements were per-
wt% chlorinated alumina catalyst was prepared by means formed on the stabilised catalysts. The contact time was
1
of an excess impregnation of the alumina carrier with an varied from 0.8 to 4.0 seconds, keeping a hydrogen-to-
aqueous chlorhydric acid solution (1.2 wt% HCl, Merk). hydrocarbon molar ratio of 5. The products of reaction were
After drying at 393 K overnight, the catalyst was calcined analysed by means of an online gas chromatograph (HP
−
1
−1
in flowing air (1 L h g
) at 773 K for 2 h and then 4890, FID) and the product identification was confirmed by
catalyst
reduced in flowing hydrogen at 773 K for 2 h.
both GC-MS (HP 5890-HP 5971A) analysis and injection
of the standards. GC and GC-MS analyses were performed
using a PONA (Paraffins-Olefins-Naphthenes-Aromatics)
capillary column (length 50 m, inner diameter 0.20 mm, film
thickness 0.5 µm).
The absence of diffusion effects on kinetics was verified
by the Koros–Nowak criteria (22). Concerning external dif-
fusion, no conversion variation was observed between runs
performed with 0.3 and 0.8 g of catalyst with the same con-
tact time. Regarding internal diffusion, no conversion vari-
ation was observed by varying the catalyst grain size (0.125–
2
.2. Catalyst Characterisations
Catalyst characteristics are reported in Table 1. The plat-
inum, tin, and chlorine contents were determined by X-ray
fluorescence(PhilipsPW1400). Theplatinumpercentageof
metal exposed (Pt% ME) measurements were carried out
by hydrogen–oxygen titration using a conventional pulsed
technique (χ-Sorb Licence IFP). From Table 1, it is shown
that the applied protocols allowed the preparation of the
desired catalysts and that Pt% ME was 37.5 and 10.5% for
Pt/SiO2 and PtSn/SiO2, respectively. In the case of Pt/SiO2,
the Pt% ME also corresponds to the platinum dispersion,
0.160 mm; 0.160–0.200 mm; 0.200–0.250 mm) for a constant
weight of catalyst.
3. RESULTS
TABLE 1
3
.1. Product Distribution over Pt/SiO2, PtSn/SiO2, Al2O3
and 1 wt% Cl–Al2O3 Catalysts
Composition and Platinum Percentage of Metal Exposed
(Pt% ME) of the Synthesised Catalysts
First, numerous reaction products, for which molecular
Pt
(wt%)
Sn
(wt%)
Cl
(wt%)
Pt% ME
(%)
formulae are reported in Fig. 1, were observed for the n-
PB conversion (of about 10%), whatever the nature of
the catalyst. Using completely different experimental con-
ditions, Shephard and Rooney showed comparable reac-
tion products when studying n-PB transformation over an
alumina-supported platinum catalyst (19). These products
Catalysts
Pt/SiO2
PtSn/SiO2
Al2O3
0.48
0.48
—
—
0.23
—
—
—
—
37.5
10.5
—
Cl–Al2O3
—
—
1.00
—

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
433
FIG. 1. Molecular formulae of the relevant products observed during n-PB conversion at 773 K, 5 bars, and H2/HC = 5.
are gathered into four different family reactions as also obtained on all catalysts through a C–C bond scission
indicated in Table 2. From this table, it is shown that n- of the alkyl group. On metallic sites, these three products
PB alkyl chain cyclisation products such as indane and in- are formed by a hydrogenolysis reaction. In this latter case,
dene are mainly observed on the metallic catalysts and, benzene exhibits the lower selectivity on the monometal-
more particularly, over the unpromoted silica-supported lic Pt/SiO2 catalyst, whereas its selectivity is significantly
platinum catalyst. In contrast, isopropylbenzene, formed increased on the promoted catalyst (PtSn/SiO2). On acid
from n-PB alkyl chain isomerisation, appears as one of the sites, the HDA products are formed by catalytic cracking
major products formed on acid sites. Indeed, this isomeri- reactions and in this case benzene is the major product over
sation is a well-known acidic reaction (25). Concerning the the chlorinated alumina catalyst.
dehydrogenated products (phenyl-2-propene, Z-phenyl-1-
The formation of dehydrogenated products (phenyl-2-
propene, and E-phenyl-1-propene), one can see that they propene, Z- and E-phenyl-1-propene) and HDA products
are formed on both metallic and acidic functions. The HDA (benzene, toluene, and ethylbenzene) are more specifically
reaction products (benzene, toluene, and ethylbenzene) are studied in the following sections.

4
34
TOPPI ET AL.
TABLE 2
Product Distribution Given as Selectivities at 10% Conversion of n-PB (moles of product
formed per 100 moles of n-PB reacted) on Stabilised Catalysts
Pt/SiO2
(1040)a
PtSn/SiO2
(770)a
Al2O3
(200)a
Cl–Al2O3
Reactions
Cyclisation
Products
Indene
(380)a
5.0
31.6
0.0
2.6
0.0
0.0
0.0
0.6
Indane
Dehydrogenation
Isomerisation
Phenyl-2-propene
Z-phenyl-1-propene
E-phenyl-1-propene
2.0
4.5
15.2
1.8
4.4
16.4
0.0
0.0
1.4
0.0
2.2
7.0
Isopropylbenzene
α-Methylstyrene
o-Xylene
1.7
0.0
0.7
2.4
4.5
0.0
1.3
0.0
13.0
0.0
0.0
1.4
6.1
2.4
2.0
0.7
2-Ethyltoluene
HDA
Methane
Ethane
Propane
Benzene
Toluene
10.5
6.4
2.9
3.5
6.3
7.4
20.6
6.1
10.6
13.0
4.7
24.9
14.9
8.6
13.0
8.9
11.1
16.5
13.2
19.7
7.5
Ethylbenzene
14.0
14.0
6.3
a
Corresponding time on stream (min) of reported data.
3
.2. Product Formation versus Time On Stream
.2.1. Metallic sites contribution (Pt/SiO2 and PtSn/
3
SiO2). The dehydrogenated and hydrogenolysis product
concentrations obtained over Pt/SiO2 and PtSn/SiO2 cata-
lysts from n-PB conversion are reported versus time on
stream in Figs. 2 and 3, respectively.
Regarding the dehydrogenated product formation, an
identical behaviour is observed in Fig. 2 whatever the cata-
lyst: (i) the concentration of the three products increases
with time onstream; (ii) the major product is E-phenyl-
1
-propene, and phenyl-2-propene is the one that displays
the lowest concentration; and (iii) the product concentra-
tions are roughly the same on both catalysts (Pt/SiO2 and
PtSn/SiO2). However, steady state is reached a lot more
rapidly on the bimetallic catalyst than on the monometal-
lic one, and such a phenomenon is also observed for the
hydrogenolysis products (Fig. 3). Concerning these latter
products, one can see that their concentrations decrease
sharply with time on stream over the unpromoted cata-
lyst (Fig. 3, panel a), due to the metallic phase deactiva-
tion, whereas on the promoted catalyst (Fig. 3, panel b),
the product concentration is already stable after several
minutes on stream. The observed deactivation is attributed
to coke deposition over the metallic function. Indeed, a
deactivation trend due to sulfur poisoning was ruled out
by an additional experiment, not reported in the present
study, in which the effect of the sulfur content of the liq-
uid feed was investigated. The comparison between the n-
PB used in this study (98% purity) and a high-purity n-PB
FIG. 2. Concentrations of dehydrogenated products versus time on
stream, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢀ, E-phenyl-
(>99%) containing 20 and 2 ppm of sulfur, respectively, was
1
-propene; ꢁ, Z-phenyl-1-propene; and ꢂ, phenyl-2-propene; (a) Pt/SiO2
carried out. This comparative study was performed over a and (b) PtSn/SiO2.

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
435
concentration increases slightly with time on stream. On
wt% Cl–Al2O3 (Fig. 4, panel b), stable concentrations
1
of E- and Z-phenyl-1-propene are observed after a rather
short period of time on stream; the major product was E-
phenyl-1-propene.
In the case of the catalytic cracking products (Fig. 5), the
aromatic distribution is quite different between the two ma-
terials. On Al2O3, the concentrations of benzene, toluene,
−
5
−1
and ethylbenzene are similar (from 3 × 10 mol L to
−
5
−1
5
× 10 mol L ) and ethylbenzene is the major product
Fig. 5, panel a), whereas over 1 wt% Cl–Al2O3 (Fig. 5,
panel b), benzene is the major product with a much higher
(
−
5
−1
concentration (11 × 10 mol L ) than those of toluene
−
5
−1
and ethylbenzene (around 4 × 10 mol L ).
.3. Kinetic Results
.3.1. Reactions over the metallic sites (Pt/SiO2 and
3
3
PtSn/SiO2). Dehydrogenated and hydrogenolysis prod-
uct concentrations obtained over Pt/SiO2 and PtSn/SiO2
catalysts from n-PB conversion versus contact time are
FIG. 3. Concentrations of products of hydrogenolysis versus time on
stream, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢃ, ethylben-
zene; ꢄ, toluene; and ꢅ, benzene; (a) Pt/SiO2 and (b) PtSn/SiO2.
bifunctional catalyst (Pt/Al2O3–Cl) to evaluate the impact
of sulfur on both metallic and acidic functions. Even with an
order of magnitude difference in sulfur content between the
two n-PB feeds, it was found that the deactivation trend and
the product selectivities were the same whatever the sulfur
content. The steady state was reached much more rapidly
over the tin-promoted catalyst and is therefore attributable
to a much lower coking of the latter catalyst as compared
to the unpromoted one. Finally, ethylbenzene formation
and especially toluene formation occur to lesser extent
on PtSn/SiO2 than on Pt/SiO2 after deactivation. Ben-
zene concentration is the same on both stabilised catalysts
(
Fig. 3).
3
.2.2. Acid sites contribution (Al2O3 and 1 wt% Cl–
Al2O3). The dehydrogenated and catalytic cracking prod-
uct concentrations obtained over Al2O3 and 1 wt%. Cl–
Al2O3 catalysts from n-PB conversion are reported versus
time on stream in Figs. 4 and 5, respectively.
Regarding the dehydrogenated products (Fig. 4), only
E-phenyl-1-propene is detected on Al2O3 under our ex-
FIG. 4. Concentrations of dehydrogenated products versus time on
stream, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢀ, E-phenyl-
1
-propene; and ꢁ, Z-phenyl-1-propene; (a) Al2O3 and (b) 1 wt% Cl–
perimental conditions. Figure 4 (panel a) shows that its Al2O3.

4
36
TOPPI ET AL.
Concerning the three products of the dehydrogenation
reaction, their concentrations decrease constantly in the
range of studied contact times (Fig. 6) and should therefore
experience maxima at lower contact times. From Table 3,
it can also be observed that the disappearance rates of the
dehydrogenated products are from two to three times lower
over PtSn/SiO2 than over Pt/SiO2.
Regarding the hydrogenolysis products (Fig. 7), it can be
seen that the rates of formation of the three products (ben-
zene, toluene, and ethylbenzene) are constant over both
silica-supported platinum catalysts. Over Pt/SiO2, toluene
and ethylbenzene are the major products and their forma-
−
5
−1 −1
tion rates are similar, respectively 2.26 × 10 mol L
s
−
5
−1 −1
and 2.25 × 10 mol L
s
(Table 3) while benzene is
the minor product with a smaller formation rate (1.43 ×
−
5
−1 −1
10
mol L s ). Over the bimetallic catalyst PtSn/SiO2
Fig. 7, panel b), the formation rates are affected by the
presence of the tin promoter. Indeed, the formation rate of
(
−
5
−1 −1
ethylbenzene decreases slightly (1.68 × 10 mol L s )
and the formation rate of toluene drops drastically (0.44 ×
FIG. 5. Concentrations of products of cracking versus time on stream,
in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢃ, ethylbenzene; ꢄ,
toluene; and ꢅ, benzene; (a) Al2O3 and (b) 1 wt% Cl–Al2O3.
shown in Figs. 6 and 7, respectively. The corresponding re-
action rates are reported in Table 3.
First, a constant global rate of consumption of n-PB is ob-
served over both Pt/SiO2 and PtSn/SiO2 catalysts (Fig. 6).
TABLE 3
Dehydrogenation, Hydrogenolysis, or Cracking and Isomerisa-
tion Rates of n-PB over Different Catalysts at 773 K, 5 bars, and
H2/HC = 5
a
−5
−1 −1
Reaction rates (10 mol L s )
Products
Pt/SiO2
PtSn/SiO2
Al2O3
Cl–Al2O3
Disappearance rate
Phenyl-2-propene
Z-phenyl-1-propene
E-phenyl-1-propene
Formation rate
Benzene
Toluene
Ethylbenzene
Isopropylbenzene
0.24
0.44
1.04
0.12
0.21
0.35
—
—
—
—
—
—
1.43
2.26
2.25
—
2.15
0.44
1.68
—
3.20
1.86
2.64
3.20
9.82
2.94
2.20
3.00
FIG. 6. Concentrations of n-PB and dehydrogenated products versus
contact time, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢆ,
n-PB; ꢀ, E-phenyl-1-propene; ꢁ, Z-phenyl-1-propene; and ꢂ, phenyl-2-
propene; (a) Pt/SiO2 and (b) PtSn/SiO2. (Kinetic data recorded for 8 h of
run.)
a
Kinetic data recorded for 8 h of run for metallic catalysts and 5 h of
run for nonmetallic catalysts.

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
437
contact times (Fig. 8, panel a). Over 1 wt% Cl–Al2O3, the
product concentrations increase and finally seem to reach
a plateau (Fig. 8, panel b).
For the catalytic cracking reactions (Fig. 9), the rates of
formation of the three products are constant for all cata-
lysts. In the range of the contact times studied, the highest
concentration is obtained for ethylbenzene and the lowest
for toluene over Al2O3. Moreover, the formation rates of
both ethylbenzene and toluene are similar (Fig. 9, panel a,
and Table 3). In addition, benzene exhibits the highest for-
−
5
−1 −1
mation rate (3.20 × 10 mol L s ). Such a behaviour
is emphasised over 1 wt% Cl–Al2O3 with a formation rate
−
5
−1 −1
of 9.82 × 10 mol L
s
and in this latter case, benzene
becomes the major product. Concerning toluene and ethyl-
benzene, the formation rates of these two products are sig-
nificantly affected by chlorine addition. Even though it is
not as spectacular as in the case of benzene, it should be
noted that the concentration of toluene is higher than that
of ethylbenzene over the chlorinated alumina whereas the
opposite is observed on alumina (Fig. 9 and Table 3).
FIG. 7. Concentrations of n-PB and products of hydrogenolysis ver-
sus contact time, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5:
ꢆ, n-PB; ꢃ, ethylbenzene; ꢄ, toluene; and ꢅ, benzene; (a) Pt/SiO2 and
(b) PtSn/SiO2. (Kinetic data recorded for 8 h of run.)
−
5
−1 −1
1
0
mol L s ), whereas the formation rate of ben-
−5
−1 −1
zene increases significantly (2.15×10 mol L s ). Over
such a promoted catalyst, toluene thus becomes the minor
product.
3
.3.2. Reactions over the acid sites (Al2O3 and 1 wt% Cl–
Al2O3). The concentrations of dehydrogenated and cata-
lytic cracking products obtained over Al2O3 and 1 wt%
Cl–Al2O3 catalysts from n-PB conversion are shown versus
contact time in Figs. 8 and 9, respectively. The correspond-
ing reaction rates are reported in Table 3.
As previously found over the metallic function, the n-
PB consumption rate is constant over both Al2O3 and
1
wt% Cl–Al2O3. The formation rates of the dehydro-
genated products (Fig. 8) are not constant versus contact
time and some discrepancies appear between the two ma-
terials in the range of studied contact times. Indeed, over
Al2O3, Z-phenyl-1-propene and E-phenyl-1-propene ex-
hibit the same behaviour with an increase of their concen-
FIG. 8. Concentrations of n-PB and dehydrogenated products versus
contact time, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢆ, n-
PB; ꢀ, E-phenyl-1-propene; and ꢁ, Z-phenyl-1-propene; (a) Al2O3 and
trations up to 1.1 s and a subsequent decrease at higher (b) 1 wt% Cl–Al2O3. (Kinetic data recorded for 5 h of run.)

4
38
TOPPI ET AL.
order could be considered because of the low conversion
of the reactant (less than 10%) over PtSn/SiO2, a true zero
order was also assumed over this latter catalyst consider-
ing both the experimental conditions (high partial pressure
of n-PB) and the zero order found over the unpromoted
catalyst.
4.1.2. Formation and desorption of phenyl propene
species. Since n-PB suffers a dehydrogenating chemisorp-
tion of the lateral alkyl chain over two adjacent Pt sites,
adsorbed phenyl propene species are therefore expected to
be the most abundant reactive intermediates (22) on plat-
inum sites. Indeed, phenyl-2-propene, Z-phenyl-1-propene,
and E-phenyl-1-propene were detected in the gas phase
over the silica-supported platinum catalysts. For the sake
of clarity, phenyl propene species will be denoted as Mi,ads
when adsorbed and Mi,gas when present in the gas phase.
Equation [1] can therefore be written as follows with k1
being the adsorption rate constant of n-PB:
k1
n-PBgas → Mi,ads + H2.
[1]
The one-way arrow presented in Eq. [1] means that the
elementary step is far from equilibrium (22), since the de-
hydrogenation reaction is thermodynamically favoured un-
der the experimental conditions used in this study (773 K,
5
bars, H2/HC = 5).
The adsorbed phenyl propene species can then desorb,
leading to the corresponding species in the gas phase as
observed in Fig. 6. Both the formation rates of Mi,gas and the
desorption rates of Mi,ads species are indeed constant due
to the saturation of all active sites [L]M by Mi,ads species, as
expected from the true zero-order reaction with respect to
n-PB. Such a behaviour indicates that Mi,ads species must be
in equilibrium at the surface of the catalyst, while desorbing
through a rake mechanism (26), as illustrated in Eq. [2]:
FIG. 9. Concentrations of n-PB and products of cracking versus con-
tact time, in n-PB conversion at 773 K, 5 bars, and H2/HC = 5: ꢆ, n-PB;
ꢃ, ethylbenzene; ꢄ, toluene; and ꢅ, benzene; (a) Al2O3 and (b) 1 wt%
Cl–Al2O3. (Kinetic data recorded for 5 h of run.)
4
. DISCUSSION
M1,gas
M2,gas
↑
M3,gas
↑
4
.1. Kinetics and Mechanisms over the Metallic Sites
Pt/SiO2 and PtSn/SiO2)
(
↑
n-PBgas → M1,ads
L1,M
[
2]
0
0
M2,ads
L2,M
M3,ads.
L3,M
In this section, the global Pt density of sites (Pt being
a zero-valent platinum atom) will be denoted as [L]M and
corresponds to the metal dispersion measured according to
hydrogen–oxygen titrations.
[
Li ]M, representing the partial density of sites occupied by
Mi,ads, should be constant and the balance of the metallic
sites can therefore be written as follows (Eq. [3]):
4
.1.1. Zero order reaction with respect to n-PB. Figure 6a
indicates that the global rate of consumption of n-PB, −d[n-
PB]/dt, leading to the various detected products, is constant
up to 22% conversion of the reactant over Pt/SiO2. This lin-
ear decrease of n-PB with contact time, up to such a high
conversion value, means that the order related to the n-PB
is a true zero order. It cannot be a degeneracy of the or-
der, which could be assumed if conversion had been less
than 10%. This result is presumably due to the experimen-
[L]M = [L ] + [L ] + [L ] .
1 M
2 M
3 M
[3]
For each Mi,ads species, the desorption pathway can there-
fore be summarised as follows:
k1
n-PBgas → Mi,ads + H2
[1]
[4]
kd,i
Mi,ads → Mi,gas.
tal conditions. Indeed, pure n-PB is sent into the feed with
0
a high partial pressure and therefore saturates all Pt acces- From the kinetic results (Fig. 6, panel b), each desorption
sible active sites as adsorbed species. Even if a degenerated rate (rd,i ) is constant with contact time and is expressed as

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
439
follows (step corresponding to Eq. [4]), where kd,i is the From this pathway and in agreement with experimental re-
corresponding desorption rate constant:
sults (Fig. 7), the rates of formation of all products can be
expressed as follows:
d[Mi,gas]
rd,i =
= kd,i [Mi,ads] = constant = kd,i [Li ]M. [5]
dt
rproduct,i = kr,i [Mi,ads] = constant = kr,i [Li ]M.
[7]
Equation [5] confirms that [Li ]M can be considered as
constant, due to the equilibria between surface species It has to be emphasised that in the case of zero-order reac-
Eq. [2]), since all platinum sites remain saturated under tions, reaction rates are directly proportional to the number
(
our operating conditions.
of active sites.
4
.1.3. Surface reactions of adsorbed phenyl propene
4.1.4. Elementary step sequences of hydrogenolysis reac-
species: Hydrogenolysis of C–C bonds of the alkyl chain on tions and effect of the tin promoter. The kinetic study al-
0
Pt . Surface reactions of the adsorbed phenyl propene lows us to suggest reactive intermediates corresponding to
species have to be taken into account to explain the ob- the hydrodealkylated products. From the work of Duprez
served decrease of the phenyl propene isomer concentra- et al. (20, 21) and Grenoble (7), benzene formation involves
tions reported in Fig. 6. Depending on the contact time, a unique adsorption site (Fig. 10, part a). By extrapolating
the corresponding adsorbed phenyl propene species can this afore mentioned elementary step sequence (7) to the
suffer hydrogenolysis reactions, leading to the formation formation of toluene and ethylbenzene, it is likely that two
of the dealkylated products reported in Fig. 7. However, adjacent sites are required (Fig. 10, parts b and c). This as-
the chemical structure of the phenyl propene species can sumption is fully consistent with both the observed phenyl
only explain the formation of two out of the three hy- propene species detected in the gas phase and the forma-
drodealkylated products shown in Fig. 7. Indeed, ethyl- tion rates of the hydrodealkylated products reported in
benzene and toluene are formed through the C–C bond Table 3. As expected, the formation rates of ethylbenzene
scission of adsorbed phenyl-2-propene and Z-/E-phenyl- and toluene decrease while that of benzene increases when
1
-propene species, respectively. Therefore a fourth surface tin is added as a promoter. Indeed, one of the known effects
species, whose corresponding desorbed species was not de- of tin promotion over silica-supported platinum catalysts is
tected, must be considered to explain the formation of ben- to isolate the zero-valent platinum sites by alloy formation
zene (Fig. 7). This fourth species is either adsorbed on the (24, 27, 28) which clearly favours the single-site elementary
aromatic nucleus as already suggested by Duprez et al. (20) step sequence leading, in our case, to benzene.
or adsorbed on a single site as assumed by Grenoble (7)
0
and should thus occupy an L4,M patch of Pt sites to explain 4.2. Mechanisms and Kinetics of n-PB Conversion over
the constant formation rate found for benzene (Fig. 7). In
addition, the highest concentrations of toluene and ethyl-
benzene found over the Pt/SiO2 catalyst (Fig. 7, panel a)
are in good agreement with previous studies performed by
Duprez et al. (20, 21). Indeed, for HDA of n-PB over Rh-
supportedcatalysts(20)orHDAofalkylbenzenemolecules
other than n-PB over Pt- and Ni-supported catalysts (21),
these authors also reported that the C–C bond scission was
favoured in the middle and at the end of the alkyl chain.
It is worth noting that the kinetic study of the transfor-
mation of a complex molecule such as n-PB allowed us to
demonstrate that hydrogenolysis reactions follow parallel
pathways. Indeed, ethylbenzene, toluene, and benzene are
formed concurrently with constant rates at the very begin-
ning of the reaction. In previous studies (7, 12, 13, 16), such
a conclusion was not drawn because of the use of toluene as
a model molecule which could only lead to the formation
of benzene.
the Acid Sites (Al2O3 and Cl–Al2O3)
4.2.1. AcidicmechanismproposalsoverAl2O3 and1wt%
Cl–Al2O3. As previously observed on metallic sites (cf.
Sect. 4.1.3.), hydrodealkylated products are formed con-
currently, according to the constant rates of formation re-
ported in Fig. 9. The direct formation of benzene, toluene,
and ethylbenzene on the acid sites was therefore considered
through three independent pathways.
Crackingmayoccurthroughprotolysis(carboniumions),
through a classical chain mechanism (carbenium ions), and
finally through a radical process (radical intermediates).
Cracking by protolysis occurs through a monomolecular
mechanism involving pentacoordinated intermediates (29)
and requires very strong protonic sites such as those found
in zeolite-type catalysts. Nevertheless, on Al2O3 and 1 wt%
Cl–Al2O3, it is likely that the strength of the protonic sites
is not high enough to lead to this mechanism. Therefore ei-
ther a carbenium ion chain cracking mechanism or a radical
process is considered.
The hydrogenolysis pathway can now be summarised as
follows where kr,i is the corresponding reaction rate con-
stant:
Cracking through a carbenium ion chain mechanism is
illustrated in Fig. 11 and involves strong Brønsted acid
sites. Such strong Brønsted acid sites were already re-
portedoverbothCl–Al2O3 andAl2O3 usingdifferentprobe
k1
n-PBgas → Mi,ads + H2
[1]
kr,i
Mi,ads → producti,gas
.
[6] molecules (30, 31). FTIR adsorption-desorption studies

4
40
TOPPI ET AL.
FIG. 10. Elementary step sequences leading to the formation of (a) benzene, (b) toluene, and (c) ethylbenzene. A double arrow and a single arrow
represent a global step and an elementary step, respectively.
of 2,6-dimethylpyridine (2,6-DMP) and pyridine carried
In the case of benzene, n-PB was previously isomerised to
out over these two materials confirmed the presence of isopropylbenzenewhich undergoes cracking (Fig. 11, part a,
strong and weak Brønsted and Lewis acid sites. Indeed, route 1). Note that isopropylbenzene can also be formed as
in agreement with a previous study (31), bands at 1620 and a final product over a weaker Brønsted acid site than that
−
1
1
650 cm (2,6-DMP adsorbed species) were detected for required for cracking reactions (25, 32). The one-step crack-
strong and weak Brønsted acid sites, respectively. Similarly, ing of n-PB leading to benzene might be ruled out because
−
1
bands at 1625 and 1615 cm (pyridine adsorbed species it would involve the formation of the primary carbocationic
+
(
31)) were observed for strong and weak Lewis acid sites, species CH3–CH2–CH . Based on a similar mechanism, the
2
respectively.
cracking of n-PB to toluene and ethylbenzene involves two

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
441
FIG. 11. Cracking through a carbenium ion chain mechanism proposal: (a) benzene, (b) toluene, and (c) ethylbenzene.
comparable thermodynamically unfavoured carbocationic favoured for toluene and ethylbenzene formation. In the
+
+
species, CH3–CH and CH , respectively (Fig. 11, parts b case of toluene (Fig. 12, part b), the phenyl propene radical
2
3
and c).
species could desorb as phenyl-2-propene which was not
Thus, the formation of toluene and ethylbenzene is con- detected under our experimental conditions. For ethylben-
sidered to occur through radical mechanisms involving zene, the corresponding desorbed products of the phenyl
Lewis acid sites (33) which are described in Fig. 12. The propene radical species (Fig. 12, part a) were observed as E-
radical intermediate species are very stable, especially in and Z-phenyl-1-propene (Fig. 8). Based on the afore men-
the case of the ethylbenzene formation (Fig. 12, part a). In- tioned mechanisms, one can finally add that the strength
deed, in this latter case the radical species is stabilised by the of the Lewis acid sites must reasonably be higher for the
aromatic nucleus. Such a mechanism is therefore the most formation of toluene than that of ethylbenzene.

4
42
TOPPI ET AL.
FIG. 12. Cracking through a radical mechanism proposal: (a) ethylbenzene and (b) toluene.
4
.2.2. Zero-order reaction with respect to n-PB and ki-
According to the carbenium mechanism (Fig. 11, part a,
netics of the hydrodealkylated products. Over 1 wt% Cl– route 1), isopropylbenzene is the intermediate that leads to
Al2O3, the consumption rate of n-PB is constant up to 18% benzene formation over strong Brønsted acid sites. Here it
conversion (Figs. 8b or 9b). This means, as in the case of is emphasised that isopropylbenzene does not necessarily
hydrogenolysis reactions over platinum, that the surface undergo cracking when weak Brønsted acid sites are in-
acid sites are saturated by n-PB adsorbed species (Ai,ads) volved. Finally, four reaction products are associated with
and a true zero-order reaction with respect to the reac- the four acid sites leading to the three hydrodealkylated
tant is observed. Even if a degenerated order is considered products and isopropylbenzene, this latter compound be-
because of the low conversion of the reactant (less than ing both an intermediate for the cracking reaction and a
1
0%) over Al2O3, a true zero order was assumed consider- final product for isomerisation.
ing both the previously described experimental conditions
As a consequence, Eqs. [9–11] can be written as follows
ꢁ
ꢁ
ꢁ
(
Sect. 4.1.1.) and the zero order found over the 1 wt% Cl– where k , k , and k correspond to the adsorption rate
1
d,i
r,i
Al2O3 catalyst.
constant, the desorption rate constant, and the reaction rate
As suggested before (Sect. 4.2.1.), for different acid sites, constant over acid sites, respectively:
i.e., weak and strong Brønsted acid sites and weak and
ꢁ
k
1
strong Lewis acid sites, have to be considered and the bal-
n-PBgas → Ai,ads + H2
[9]
[10]
[11]
ance over the acid sites can therefore be written as follows:
ꢁ
k
d,i
Ai,ads → Ai,gas
[
L]A = [L1]A + [L2]A + [L3]A + [L4]A
[8]
ꢁ
k
r,i
Ai,ads → producti,gas
.
where [L]A is the total density of acid sites and [Li ]A is the
partial density of sites saturated by the adsorbed interme- Finally, considering the preceding pathway (Eq. [11]) and
diates (Ai,ads).
theconstantformationrate(Fig. 9)ofthecrackingproducts,
ꢁ
From the kinetic data (Fig. 8a), it is obvious that E- and Z-
phenyl-1-propene display an intermediate behaviour over
the alumina carrier. According to the proposed mechanism
r
can thus be written as follows:
product,i
ꢁ
ꢁ
ꢁ
rproduct,i = kr,i [Ai,ads] = constant = kr,i [Li ]A.
[12]
(
Sect. 4.2.1.), these intermediates lead to ethylbenzene for-
In this equation, [Li ]A are thus constant, which confirms
that all acid sites are saturated. As in the case of the metallic
sites (Sect. 4.1.3), the reaction rates are proportional to the
number of active acid sites.
mation. In the case of the chlorinated alumina, such de-
hydrogenated intermediates are also observed. However,
they do not exhibit the expected intermediate behaviour,
notably because of the too low contact times investigated.
In the case of toluene, its corresponding desorbed in-
4.2.3. Correlation between kinetics and acidic mecha-
termediate (phenyl-2-propene) was not detected. From a nisms from chlorine addition. According to the mecha-
kinetic point of view, this suggests that the desorption rate nistic considerations mentioned previously (Sect. 4.2.1.),
constant is lower than the cracking rate constant.
benzene would preferably be formed over strong Brønsted

HYDRODEALKYLATION REACTIONS IN REFORMING CONDITIONS
443
acid sites whereas isopropylbenzene would be formed over reactions over the metallic catalyst or cracking reactions
weaker sites. In addition, toluene would be formed over over the acidic catalyst.
stronger Lewis acid sites than ethylbenzene.
By the use of a C9 aromatic molecule as reactant, the ki-
On the one hand, the formation rate of benzene is in- neticstudyshowedthathydrogenolysisandcrackingofsuch
creased by a factor of three, whereas the formation rate a model molecule lead to three hydrodealkylated products
of isopropylbenzene is almost constant after the addition (benzene, toluene, and ethylbenzene) through concurrent
of chlorine on the alumina carrier (Table 3). Kinetic data reactions rather than successive ones as might have been
thus suggest that the number of strong Brønsted acid sites expected at first.
is significantly enhanced whereas the number of weaker
In addition, whatever the nature of the active sites, a true
ones remains almost constant. Indeed, Gates et al. (34) re- zero-order reaction with respect to n-PB was observed over
ported that the addition of chlorine to an alumina carrier the whole range of contact times used in this study. As a con-
enhances the strength of Brønsted acid sites by weaken- sequence, all the active sites, either metallic or acid ones,
ing O–H bonds of adjacent hydroxyls through an inductive were saturated and the reaction rates were determined. As
effect. Moreover, Berteau and Delmon (35) reported an compared to toluene and ethylbenzene, the formation rate
increase of the total number of Brønsted acid sites.
of benzene was found to be the highest over the acidic func-
The increase of the formation rate of benzene after the tion and the lowest over Pt/SiO2.
addition of chlorine to an alumina carrier is therefore in
The kinetic study also revealed the intermediates in-
good agreement with the preceding proposed mechanism volved in the n-PB HDA leading to the corresponding hy-
(
Fig. 12, part a) showing that benzene is formed on strong drodealkylated products, and mechanisms were suggested
Brønsted acid sites through a carbenium ion chain mech- to explain their formation. These mechanisms were con-
anism. The constant formation rate of isopropylbenzene firmed by tin promotion in the case of the metallic catalyst
whatever the catalyst nature (Al2O3 or Cl–Al2O3) suggests and chlorine addition in the case of the acidic one.
that the number of weak Brønsted acid sites remains al-
As a consequence, over both Pt/SiO2 and PtSn/SiO2, two
most constant, and that the increase in the total number types of reaction sequences are proposed: (i) through ad-
of Brønsted acid sites is mainly due to the increase of the sorption on a single site for the formation of benzene, and
number of strong sites.
(ii) through adsorption on two adjacent sites for the forma-
On the other hand, the formation rate of toluene in- tion of toluene and ethylbenzene. Over Al2O3 and 1 wt%
creases significantly whereas the formation rate of ethyl- Cl–Al2O3, (i) benzene is formed on strong Brønsted acid
benzene decreases slightly over the chlorinated alumina sites via a carbocationic pathway, and (ii) toluene and ethyl-
as compared to the alumina carrier (Table 3). These ki- benzene are formed via a radical pathway on strong and
netic data are also consistent with our proposed mecha- weak Lewis acid sites, respectively.
nisms suggesting that toluene and ethylbenzene are formed
through radical processes involving strong and weak Lewis
acid sites, respetively. Indeed, Berteau et al. (30) suggested
an increase of the electron-deficient state of unsaturated
aluminium cations through an inductive effect of chlorine
leading to stronger Lewis acid sites. In addition, they re-
ported a slight increase in the number of Lewis acid sites
ACKNOWLEDGMENTS
We acknowledge the Institut Fran c¸ ais du P e´ trole (IFP) for the financial
support of this work as well as the ANRT organisation for the CIFRE grant
number128/99 allowed to Ms. S. Toppi. We also thank Dr. C. Marcilly (IFP)
for fruitful discussions and for his interest concerning this work. Finally,
we gratefully acknowledge Dr. K. Fajerwerg-Sainte-Marie for her kind
when adding chlorine to an alumina carrier. In the case of assistance in FTIR measurements.
Lewis acidity, the increase in the number of strong Lewis
acid sites thus mainly occurs at the expense of weak Lewis
acid sites since the total number of Lewis acid sites remains
almost constant.
REFERENCES
1
2
3
4
5
6
7
8
. Howley, P. A., and Shih, S., U.S. Patent No. 5,001,296 (1991).
. Wu, A., and Drake, C. A., U.S. Patent No. 5,698,757 (1997).
. Wu, A., and Drake, C. A., U.S. Patent No. 5,763,721 (1998).
. Wu, A., and Drake, C. A., U.S. Patent No. 5,945,364 (1999).
. Grenoble, D. C., J. Catal. 51, 212 (1978).
5
. CONCLUSION
. Grenoble, D. C., J. Catal. 51, 203 (1978).
. Grenoble, D. C., J. Catal. 56, 40 (1979).
The HDA of n-PB was investigated for the first time over
model catalysts (Pt/SiO2, PtSn/SiO2, Al2O3, and 1 wt%
Cl–Al2O3 under identical reforming operating conditions
. Duprez, D., Maurel, R., Miloudi, A., and Pereira, P., Nouv. J. Chim. 6,
163 (1982).
. Kochloefl, K., in “Sixth International Congress on Catalysis” (G. C.
Bond, P. B. Wells, and F. C. Tompkins, Eds.), Vol. 2, p. 1122. The
Chemical Society, London, 1976.
0. Rabinovich, G. L., and Mozhaiko, V. N., Neftekhimiya 15, 373 (1975).
1. Kasaoka, S., Omoto, M., Watanabe, T., and Yakamatsu, K., Nippon
Kagaku Kaishi 8, 1418 (1975).
9
(
773 K, 5 bars, and H2/HC = 5).
This work allowed us to evaluate separately the contri-
butions of both metallic and acidic functions to HDA reac-
tions. It was found that these reactions occur to a significant
extent over both functions through either hydrogenolysis
1
1

4
44
TOPPI ET AL.
1
1
2. Mori, S., and Uchiyama, M., J. Catal. 42, 323 (1976).
3. Duprez, D., Miloudi, A., and Tournayan, L., Appl. Catal. 14, 333
1985).
24. Dautzenberg, F. M., Helle, J. N., Biloen, P., and Sachtler, W. M. H.,
J. Catal. 63, 119 (1980).
25. Guisnet, M., and P e´ rot, G., in “Zeolite: Science and Technology” (F. R.
Ribeiro, A. E. Rodrigues, L. D. Rollman, and C. Naccache, Eds.),
p. 397. Nijhoff, The Hague, 1984.
(
1
1
4. Kim, C. J., J. Catal. 52, 169 (1978).
5. Beltrame, P., Ferino, I., Forni, L., and Torrazza, S., Chim. Ind. 60, 191
(
1978).
6. Beltrame, P., Ferino, I., Forni, L., and Torrazza, S., J. Catal. 60, 472
1979).
26. Germain, J. E., “Catalytic Conversion of Hydrocarbons.” Academic
Press, New York, 1969.
27. Sexton, B. A., Hughes, A. E., and Foger, K., J. Catal. 88, 466
(1984).
28. Zhou, Y., and Davis, S. M., Catal. Lett. 15, 51 (1992).
29. Meloni, D., Martin, D., and Guisnet, M., Appl. Catal. A 215, 67
(2001).
1
1
1
(
7. Beltrame, P., Casalone, G., Forni, L., Ferino, I., Marongiu, B., and
Torrazza, S., Ind. Eng. Chem. Process Des. Dev. 23, 176 (1984).
8. Duprez, D., Pereira, P., Barbier, J., and Maurel, R., React. Kinet. Catal.
Lett. 13, 217 (1980).
1
2
9. Shephard, F. E., and Rooney, J. J., J. Catal. 3, 129 (1964).
0. Duprez, D., Miloudi, A., Delahay, G., and Maurel, R., J. Catal. 90, 292
30. Berteau, P., Kellens, M. A., and Delmon, B., J. Chem. Soc., Faraday
Trans. 87, 1425 (1991).
(1984).
31. Corma, A., Fornes, V., and Ortega, E., J. Catal. 92, 284 (1985).
2
2
2
1. Duprez, D., Miloudi, A., Delahay, G., and Maurel, R., J. Catal. 101, 56 32. Baudon, A., Ph.D. dissertation, Universit e´ de Poitiers, Poitiers, 1996.
(
1986).
33. Mc Vicker, G. B., Kramer, G. M., and Ziemiak, J. J., J. Catal. 83, 286
(1983).
2. Boudart, M., and Dj e´ ga-Mariadassou, G., “Kinetics of Heterogeneous
Catalytic Reactions.” Princeton University Press, Princeton, 1984.
3. Humblot, F., Candy, J. P., Le Peltier, F., Didillon, B., and Basset, J. M.,
J. Catal. 179, 459 (1998).
34. Gates, B. C., Katzer, J. R., and Schuit, G. C. A., “Chemistry of Catalytic
Processes.” McGraw–Hill, New York, 1979.
35. Berteau, P., and Delmon, B., Catal. Today 5, 121 (1989).