On the radical cracking of n-propylbenzene to ethylbenzene or toluene over Sn/Al2O3-Cl catalysts under reforming conditions

Article abstract of DOI:10.1016/j.jcat.2004.11.026

Two series of chlorided-alumina-supported Sn catalysts were synthesized with different precursors, SnCl₂ or SnBu₄, and with various contents of Sn. The acidity of the catalysts including 0.2 wt% Sn was characterized by FTIR adsorptio

Full text of DOI:10.1016/j.jcat.2004.11.026

Journal of Catalysis 230 (2005) 255–268
www.elsevier.com/locate/jcat
On the radical cracking of n-propylbenzene to ethylbenzene or toluene
over Sn/Al2O3–Cl catalysts under reforming conditions
Stéphanie Toppi a,b,1, Cyril Thomas a,∗, Céline Sayag a, Dominique Brodzki a, Katia Fajerwerg a,
Fabienne Le Peltier b,2, Christine Travers b, Gérald Djéga-Mariadassou a
a
Laboratoire de Réactivité de Surface, UMR CNRS 7609, case 178, Université Pierre et Marie Curie, 4 Place Jussieu, 75252 Paris, France
b
Institut Français du Pétrole, 1 et 4 Avenue de Bois-Préau, 92852 Rueil-Malmaison cedex, France
Received 10 November 2004; accepted 24 November 2004
Abstract
Two series of chlorided-alumina-supported Sn catalysts were synthesized with different precursors, SnCl or SnBu , and with various
2
4
contents of Sn. The acidity of the catalysts including 0.2 wt% Sn was characterized by FTIR adsorption–desorption of 2,6-dimethylpyridine
(Brønsted acidity) and pyridine (Lewis acidity) and compared with that of Al O –Cl. The catalytic activity of the synthesized materials
2
3
was investigated for the transformation of n-propylbenzene under reforming conditions. The results show that the incorporation of Sn into
Al O –Cl is not harmless. Neither the nature of the Sn precursor nor the content of Sn evenly affects the acidity and the distribution of
2
3
the isomerized and cracked products. The correlation between the product distribution obtained for the transformation of n-propylbenzene
and the acidity of the Al O –Cl and the 0.2Sn /Al O –Cl catalysts supports the formation of isopropylbenzene and benzene catalyzed
2
3
SnCl2
2 3
by Brønsted acid sites via carbenium ion chemistry. In contrast, the production of toluene and ethylbenzene occurs via radical chemistry.
The formation of these products is assumed to be catalyzed by Lewis acid sites with different strengths. The stability of the proposed radical
intermediate species is consistent with the involvement of stronger Lewis acid sites in the production of toluene compared with those involved
in that of ethylbenzene. The catalytic cycles responsible for the formation of toluene and ethylbenzene via radical pathways over an Al–O
pair are reported. Finally, it is worth noting that benzene is always the major product of the cracked compounds.
2004 Elsevier Inc. All rights reserved.
Keywords: Chlorided-alumina-supported tin catalysts; Hydrodealkylation; FTIR; 2,6-Dimethylpyridine; Pyridine; Lewis acidity; Brønsted acidity;
Mechanisms; SnBu ; SnCl
4
2
1. Introduction
zene is a side product of alkylaromatic hydrodealkylation.
Typical reforming catalysts, which consist of platinum sup-
ported on a chlorided-alumina carrier with either tin or rhe-
nium added as a promoter, are bifunctional catalysts [1–3].
One of the aims of the studies initiated by our group was
to gain a better understanding of such a side reaction un-
der reforming conditions and more particularly to determine
whether hydrodealkylation reactions occurred preferentially
on the metallic function or on the acidic one. To achieve this
goal, model monofunctional catalysts were studied [4,5].
These studies showed that hydrodealkylation reactions ef-
fectively occurred on both functions. Moreover, we showed
that the dispersion and the Sn promotion of the Pt metallic
New environmental regulations on benzene content in
gasoline require a crucial control of its formation in the
course of the reforming process, provided that the feed-
stocks do not contain n-C6. In this refining process, ben-
*
Corresponding author. Fax: +33 1 44 27 60 33.
E-mail address: cthomas@ccr.jussieu.fr (C. Thomas).
Present address: Gaz de France, 361 Avenue du Président Wilson,
1
93211 Saint-Denis La Plaine cedex, France.
2
Present address: Axens, 89 Bd Franklin Roosevelt, 92500 Rueil Mal-
maison, France.
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.11.026
2004 Elsevier Inc. All rights reserved.

256
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
phase [4,5] and the addition of Cl to an alumina carrier [4]
markedly influenced the selectivity of the hydrodealkylated
compounds. In the latter work, the model catalysts consisted
of Al2O3 and Al2O3–Cl. For Pt–Sn/Al2O3 catalysts, how-
ever, it has been reported that not only was Sn closely as-
sociated with Pt, as PtxSny alloys, but it also coexisted as
nonreducible SnOx species, either in contact with the Pt par-
ticles or isolated on the alumina carrier [6]. This observation
led us to study model acidic chlorided-alumina-supported
Sn catalysts, which we believe to be most representative of
the acidic function of the typical reforming catalyst. Sur-
prisingly, as far as we know, such catalysts have scarcely
been studied [7,8]. In these studies, Kirszensztejn et al. [8]
used a rather unusual approach to introduce chloride into
their Al2O3–SnO2 samples by reacting them with gas-phase
CCl2F2/H2 at 300 ◦C. It must be noted that this treatment
led to the introduction of fluoride into the samples as well,
and that fluoride strongly influences the acid–base properties
of alumina [9–11]. The acid–base properties of chlorine-free
[12–15] and chlorided [7,8] alumina modified by tin oxides
have been reported. Overall, these studies concluded that the
acidity of alumina-supported Sn catalysts decreases with in-
creasing amounts of tin.
The synthesis of a series of SnSnCl /Al2O3–Cl materials,
2
with various contents of Sn, was close to that performed for
the chlorided alumina catalyst. After being in contact with
HCl for 45 min at RT, an appropriate amount of SnCl2,
diluted in the 1.2 wt% HCl solution, was introduced. The
solution was then maintained under stirring for 4 h at RT.
After filtering, the catalyst was exposed to an aqueous so-
lution of hydrogen peroxide for 30 min at RT to allow for
the oxidation of SnII to SnIV. The catalyst was then filtered
and washed with distilled water and finally dried in air at
120 ◦C for 12 h. The decomposition of the Sn precursor was
−1
performed in air (1 L h
g
−1) at 520 ◦C for 2 h. The H2O2
cat
oxidation step was performed to allow for comparison with
the catalysts synthesized with SnBu4; the oxidation state of
Sn in the latter series was SnIV.
The synthesis of the SnSnBu /Al2O3–Cl series of catalysts
4
differed substantially from that of the SnSnCl /Al2O3–Cl se-
2
ries. Solutions of SnBu4 in n-heptane (Aldrich), containing
appropriate amounts of SnBu4, were dripped into the Al2O3
support and kept for 1 h under continuous stirring at RT. n-
C7 was then evaporated under air at 120 ◦C for 15 h. The
decompos−it1ion of the Sn precursor was performed in air
−1
(1 L h
g
) at 150, 350 and 520 ◦C for 2 h at each tempera-
cat
The aim of this work was to investigate the influence
of Sn addition to Al2O3–Cl (Sn/Al2O3–Cl catalysts) on the
acidity of the catalytic materials and their catalytic activity in
the transformation of n-propylbenzene (n-PB) under reform-
ing conditions. For this purpose, two series of Sn/Al2O3–Cl
catalysts were synthesized with different Sn precursors. The
results reveal striking differences in the acidity and the cat-
alytic activity of the materials, depending on the nature of
the Sn precursor. In addition, the change in the catalytic ac-
tivity of Al2O3–Cl with tin addition from the SnCl2 precur-
sor could be correlated with acidity characterizations. The
mechanisms of the formation of the hydrodealkylated com-
pounds are also discussed.
ture. The catalyst was then acidified with the same procedure
as that used for Al2O3–Cl and SnSnCl /Al2O3–Cl, except that
2
the catalyst was kept in contact with the solution for 24 h un-
der stirring at RT. After filtering, the material was dried in−a1ir
−1
at 120 ◦C for 12 h and calcined in flowing air (1 L h
at 520 C for 2 h.
g
)
cat
◦
Prior to catalytic measurements, the catalysts were cal-
−1
cined in situ in flowing air (1 L h−1
g
−1) at 500 ◦C for
cat
2 h and then reduced in H2 (1 L h g−ca1t ) at 500 ◦C for
2 h. The composition of the synthesized catalysts is listed
in Table 1. This table shows that the applied protocols al-
lowed for the preparation of the desired catalysts. Al2O3–
Cl and SnSnCl /Al2O3–Cl catalysts contained 1 0.1 wt%
2
Cl. Note that the SnSnBu /Al2O3–Cl series of catalysts con-
4
tained slightly lower amounts of Cl (0.8–0.9 wt%) than that
2. Experimental
the Al2O3–Cl and SnSnCl /Al2O3–Cl catalysts did. The two
2
0.2Sn/Al2O3–Cl catalysts exhibited identical Sn (0.19 wt%)
and Cl (0.9 wt%) contents.
2.1. Catalyst preparation
The chlorided alumina material (1 wt% Cl) was pre-
pared by impregnation of γ -Al2O3 (GOD 200; surface area
200 m2 g−1, porous volume 0.6 cm3 g−1) with an aqueous
chlorhydric acid solution (1.2 wt% HCl, Merck) for 45 min
with stirring at room temperature (RT). After filtering, the
material was dried in air at 120 ◦C overnight and calcined in
Table 1
Composition of the synthesized catalysts determined after calcination (air,
◦
◦
500 C, 2 h) and subsequent reduction (H , 500 C, 2 h)
2
Catalysts
Sn
Cl
(wt%)
Sn
(wt%)
precursor
−1
◦
) at 500 C for 2 h.
cat
−1
flowing air (1 L h
g
Al O –Cl
–
1.0
1.0
0.9
1.1
–
2
3
0.1Sn
0.2Sn
0.5Sn
/Al O –Cl
SnCl
SnCl
SnCl
0.13
0.19
0.48
SnCl2
SnCl2
SnCl2
2
3
2
2
2
Sn/Al2O3–Cl catalytic materials were prepared with ei-
ther SnCl2 or SnBu4 as a precursor (Merck). These materials
are referred to as xSn(Sn precursor)/Al2O3–Cl, where x is the
Sn content (wt%) of the synthesized catalyst. It must be em-
phasized that, because of the obviously different natures of
the precursors, different synthesis routes were used.
/Al O –Cl
2
3
/Al O –Cl
2
3
0.1Sn
0.2Sn
0.4Sn
/Al O –Cl
SnBu
SnBu
SnBu
0.8
0.9
0.8
0.11
0.19
0.40
SnBu4
SnBu4
SnBu4
2
3
4
4
4
/Al O –Cl
2
3
/Al O –Cl
2
3

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
257
2.2. Characterization of the catalysts
Blank experiments, in which the reactor contained no
catalyst, indicate that thermal cracking was negligible com-
pared with that observed in the presence of the catalysts.
Reactions were carried out with 0.8 g of catalyst (0.16–
0.20 mm) in a fixed-bed microreactor, at a temperature of
500 ◦C and a total pressure of 5 bars. Catalytic measure-
ments were performed on stabilized catalysts. The con-
tact time varied from 0.8 to 2.8 s, during which a hy-
drogen/hydrocarbon molar ratio of 5 was maintained. The
reaction products were analyzed with an on-line gas chro-
matograph (HP 4890, FID), and the identification of the
products was confirmed by GC-MS (HP 5890-HP 5971A)
analysis and injection of the standards. GC and GC-MS
analyses were performed with a PONA (Paraffins–Olefins–
Naphtenes–Aromatics) capillary column (Hewlett–Packard,
50 m long, inner diameter 0.20 mm, film thickness 0.5 µm).
2.2.1. Adsorption–desorption of probe molecules followed
by infrared spectroscopy (FTIR)
FTIR spectra of adsorbed probe molecules on Al2O3–Cl
and 0.2Sn/Al2O3–Cl were collected on a Bruker Vector 22
FTIR spectrometer equipped with a DTGS detector and a
data acquisition station. The sample was pressed into a self-
supporting wafer of 8–12 mg cm−2. The wafer was loaded
into a conventional pyrex glass cell sealed with KBr win-
dows that was connected to a vacuum system, which pro-
duced a dynamic vaccum of at least 5 × 10−8 bar. Prior to
adsorption of the probe molecules, the catalyst was submit-
ted to oxidizing and reducing pretreatments at atmospheric
pressure identical to those described in Section 2.1. After
pretreatment, the catalyst was evacuated at 120 ◦C to re-
move residual traces of water. The catalyst temperature was
then decreased to RT under dynamic vacuum (5×10−8 bar).
Pyridine and 2,6-dimethylpyridine (2,6-DMP) were dried in
vacuo and stored in the presence of a 3A molecular sieve.
Pyridine and 2,6-DMP were adsorbed at RT under saturated
atmospheres of the probe molecules. After being exposed to
the probe molecules for 1.5 h, the catalysts were evacuated
at 25, 150, and 300 ◦C for 15 min, and the FTIR spectra were
recorded at RT; 128 scans were accumulated with a spectral
resolution of 4 cm−1. The spectrum of the pretreated cata-
lyst was used as a reference and subtracted from the spectra
of the catalysts exposed to the probe molecules, which were
subsequently evacuated at different temperatures.
As already discussed in a previous paper from our labo-
ratory [9], we followed the Lewis acidity by observing the
area of the pyridine band at ca. 1450 cm−1, assigned to
pyridine coordinated with Lewis acid sites (L-Py, υ C–N or
υ19b) [11,16]. Because of the low intrinsic Brønsted acid-
ity of alumina materials, pyridine is not the most suitable
probe molecule for characterizing Brønsted acid sites [11].
Therefore we followed the Brønsted acidity by observing the
adsorption bands from 1650 to 1615 cm−1 attributed to the
υ C–C vibration of the 2,6-DMP protonated probe molecule,
as shown by Corma et al. [10,11].
3. Results
3.1. Adsorption–desorption of probe molecules followed by
FTIR spectroscopy
3.1.1. Adsorption of 2,6-dimethylpyridine
Fig. 1 shows infrared spectra for 2,6-DMP, adsorbed at
RT, after desorption at 25, 150, and 300 ◦C under vacuum
over Al2O3–Cl and 0.2Sn/Al2O3–Cl. Adsorption bands at
1580 cm−1 and about 1600 cm−1, assigned to the interac-
tion of 2,6-DMP with Brønsted acid sites [10,11], rapidly
decrease with increasing desorption temperature, except for
the 0.2SnSnCl /Al2O3–Cl catalyst (Fig. 1b). Over Al2O3–
2
Cl (Fig. 1a) and 0.2SnSnCl /Al2O3–Cl (Fig. 1b), adsorption
2
bands in the 1615–1640 cm−1 region are less affected by
the desorption treatment than the 1650 cm−1 band, which
rapidly vanishes upon desorption. For these two catalysts,
a shift of the 1615 cm−1 band to higher frequency is ob-
served. As already suggested by Corma et al. [10], consid-
ering a wide distribution of OH groups with different acid
strengths, an increase in the desorption temperature pro-
vokes the desorption of 2,6-DMP, starting from the weakest
(lower frequency) acid centers, and shifts the 1615 cm−1
band toward higher frequencies (strongest acid centers).
Table 3 lists the Brønsted acid site distributions, as de-
duced from the 1615–1650 cm−1 adsorption bands (Table 2).
It is worth emphasizing that the introduction of Sn via SnCl2
significantly increases the total amount of Brønsted acid
sites, whereas that via SnBu4 decreases it very slightly. The
distribution of the Brønsted acid sites is also strongly af-
fected by SnCl2, with a significant increase in the amounts
of both weak and strong acid centers and a decrease in that
of the medium acid sites. Conversely, the distribution of the
Brønsted acid centers is scarcely affected when SnBu4 is
used as a Sn precursor (Figs. 1a and 1c).
The amount of Lewis and Brønsted acid sites (a.u.) re-
ported in the present work was normalized to a disk of
5 mg cm−2. Given that the extinction coefficients were not
determined, Lewis and Brønsted acidities could not be quan-
titatively compared with each other.
2.2.2. Catalytic activity
Catalytic measurements were carried out with commer-
cial n-propylbenzene (Fluka, purum ꢀ 98%) as a reactant
without further purification. Liquid n-propylbenzene was
delivered to the catalytic device [4,5] with a high-pressure
piston pump (Gilson 307). The hydrogen flow and the to-
tal pressure were controlled with a mass-flow controller
(Brooks 5850 TR) and a back-pressure regulator (Brooks
5866), respectively.

258
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
3.1.2. Adsorption of pyridine
Infrared spectra of the Al2O3–Cl and the 0.2Sn/Al2O3–Cl
samples, taken after adsorption of pyridine at RT and fur-
ther desorption under vacuum at various temperatures, are
given in Fig. 2. In agreement with a previous work [11], ad-
sorption bands of pyridine coordinated to Lewis acid sites
can be seen at 1450, 1500, and 1580 cm−1. For all cata-
lysts, the intensities of these bands decrease with increas-
ing desorption temperature. A shift of these bands to higher
wavenumbers is also observed with increasing desorption
temperature. The same explanation as that mentioned in the
case of 2,6-DMP can be given to account for this shift.
The total amount of Lewis acid sites and the distribution of
their strength were estimated with reference to the adsorp-
tion band at ca. 1450 cm−1 (Table 3). The introduction of Sn
leads to a decrease in the total amount of Lewis acid sites
for both 0.2Sn/Al2O3–Cl catalysts. Nevertheless, this de-
crease is more pronounced for the catalyst synthesized from
SnBu4 than that synthesized from SnCl2. For all catalysts,
the amount of weak Lewis acid sites is always the great-
est; the 0.2SnSnCl /Al2O3–Cl catalyst exhibits the greatest
2
amount of weak Lewis acid sites, followed by Al2O3–Cl.
The amount of medium Lewis acid sites decreases with the
incorporation of Sn. This effect is the more pronounced
for the 0.2SnSnCl /Al2O3–Cl sample. Finally, the amount of
2
strong Lewis acid sites is slightly lower over the supported
Sn catalysts compared with that over Al2O3–Cl.
3.2. Catalytic activity
3.2.1. SnSnCl /Al2O3–Cl catalysts
2
The formation rates of the isomerized and hydrodealky-
lated products (isopropylbenzene, benzene, toluene, ethyl-
benzene + styrene) of n-propylbenzene under reforming
conditions, on the SnSnCl /Al2O3–Cl catalysts, are listed in
2
Table 4. The addition of Sn promotes the formation of iso-
propylbenzene and that of the sum of ethylbenzene and
styrene. The promotional effect of Sn on the formation of
isopropylbenzene, nevertheless, is greater than that on the
sum of ethylbenzene and styrene. For both products, a max-
imum beneficial effect of Sn is obtained for a content of
0.2 wt% Sn. The incorporation of Sn also promotes the
formation of benzene for Sn contents equal or lower than
0.2 wt%. In this case, however, a monotonous decrease in
the formation rate of benzene is observed with increasing
Sn contents. Conversely, the formation of toluene is not fa-
vored by the addition of Sn, and its formation rate decreases
monotonously with increasing Sn content.
Fig. 3 shows the formation of these products as a func-
tion of contact time over 0.2SnSnCl /Al2O3–Cl. The forma-
2
tion of isopropylbenzene, toluene, and the sum of ethyl-
benzene and styrene extrapolates to positive axis ordinate
intercepts, whereas that of benzene extrapolates to a nega-
tive axis ordinate intercept (Fig. 3). In agreement with Best
and Wojciechowski [17], these results suggest that isopropy-
lbenzene, toluene, and the sum of ethylbenzene and styrene
Fig. 1. Infrared spectra of adsorbed 2,6-dimethylpyridine on calcined
◦
◦
(air, 500 C, 2 h) and reduced (H , 500 C, 2 h) catalysts with increas-
2
ing evacuation temperature: (a) Al O –Cl, (b) 0.2Sn
/Al O –Cl, and
2 3
2
3
SnCl2
(c) 0.2Sn
/Al O –Cl.
2 3
SnBu4

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
259
Table 2
FTIR characterization of Brønsted and Lewis acid sites of Al O –Cl, 0.2Sn
/Al O –Cl and 0.2Sn
/Al O –Cl. Absorbances (per Unit Surface Area)
2 3
2
3
SnCl2
2
3
SnBu4
−1
−1
of the 1615–1650 cm
different temperatures
bands of 2,6-dimethylpyridine (2,6-DMP) and of the 1450 cm
band of pyridine adsorbed on the catalysts, after desorption at
Catalysts
Absorbances
2,6-DMP (Brønsted acid sites)
Pyridine (Lewis acid sites)
◦
Desorption temperature ( C)
25
150
300
25
150
300
Al O –Cl
2.16
3.87
2.07
1.54
2.26
1.73
0.73
1.76
0.71
3.88
3.56
3.07
1.90
1.34
1.64
1.00
0.90
0.94
2
3
0.2Sn
0.2Sn
/Al O –Cl
2 3
SnCl2
SnBu4
/Al O –Cl
2
3
Table 3
Influence of the nature of the Sn precursor on the distribution of the acid
Table 5
Influence of the Sn content on the formation rates of isopropylbenzene and
hydrodealkylated compounds (benzene, toluene and the sum of ethylben-
sites as a function of the desorption temperature (T ), as deduced from
des.
FTIR measurements reported in Table 2
Acid site distribution
zene and styrene) for catalysts synthesized with SnBu as precursor
4
a
Formation rate
Catalysts
−5
−1 −1
(10 mol L
s )
Al O –Cl 0.2Sn
/
0.2Sn
/
Al O –Cl 0.1Sn
/
0.2Sn
/
0.4Sn
/
2
3
SnCl2
SnBu4
2
3
SnBu4
SnBu4
SnBu4
Al O –Cl
Al O –Cl
Al O –Cl
Al O –Cl
Al O –Cl
2 3
2
3
2
3
2
3
2
3
Brønsted acid sites
Weak (25 > T > 150 C)
Isopropylbenzene
Benzene
Toluene
3.0
3.1
1.9
2.9
◦
0.62
1.61
0.50
1.76
3.87
0.34
1.02
0.71
2.07
9.8
2.9
2.3
6.7
2.4
3.7
5.2
1.7
3.2
3.2
1.3
2.8
des.
◦
Medium (150 > T
> 300 C) 0.81
des.
◦
Strong (T
> 300 C)
0.73
2.16
Ethylbenzene
+ styrene
des.
Brønsted total
a
Data recorded for 5 h of run.
Lewis acid sites
◦
Weak (25 > T
> 150 C)
1.98
2.22
0.44
0.90
3.56
1.43
0.70
0.94
3.07
des.
◦
Medium (150 > T
> 300 C) 0.90
des.
of the SnSnBu /Al2O3–Cl catalysts. Whereas the introduction
◦
4
Strong (T
> 300 C)
1.00
3.88
des.
of Sn from SnCl2 has a significant influence on the forma-
tion of the isomerized and hydrodealkylated products, Sn
incorporation from SnBu4 has less influence on the catalytic
transformation of n-PB over these catalysts. The formation
of isopropylbenzene is roughly constant throughout the se-
ries of catalysts, whereas that of benzene and toluene de-
creases monotonically with increasing Sn content. On the
other hand, the formation of ethylbenzene and styrene is pro-
moted slightly by the incorporation of Sn, but decreases as
the amount of Sn increases.
Lewis total
Table 4
Influence of the Sn content on the formation rates of isopropylbenzene and
hydrodealkylated compounds (benzene, toluene and the sum of ethylben-
zene and styrene) for catalysts synthesized with SnCl as precursor
2
a
Formation rate
Catalysts
−5
−1 −1
(10 mol L
s )
Al O –Cl 0.1Sn
/
0.2Sn
Al O –Cl
2
/
0.5Sn
Al O –Cl
2
/
2
3
SnCl2
SnCl2
SnCl2
Al O –Cl
2
3
3
3
Isopropylbenzene
Benzene
Toluene
3.0
16.6
13.9
2.5
20.8
12.0
2.2
13.0
5.6
1.3
2.7
9.8
2.9
2.3
Ethylbenzene
+ styrene
3.5
3.6
4. Discussion
a
Data recorded for 5 h of run.
4.1. Formation of isopropylbenzene and benzene
are primary products, whereas benzene is a stable secondary
product. It is also interesting to note that the formation of
isopropylbenzene flattens at high contact time (Fig. 3, 2.8 s).
This behavior is typical of an unstable primary product that
further transforms into a secondary product [17]. These ob-
Tables 3–5 show that the introduction of Sn from dif-
ferent precursors does not evenly affect the acidity of the
chlorided alumina catalyst and the formation of isopropyl-
benzene and benzene. It is obvious from Table 3 that the
addition of 0.2 wt% Sn from SnCl2 enhances Brønsted acid-
ity with a marked increase in the amount of strong and
weak acid sites compared with Al2O3–Cl, whereas the Brøn-
servations, described for the 0.2SnSnCl /Al2O3–Cl catalyst,
2
also hold for Al2O3–Cl and SnSnBu /Al2O3–Cl catalysts, al-
4
though they are less pronounced than for the SnSnCl /Al2O3–
sted acidity of 0.2SnSnBu /Al2O3–Cl scarcely differs from
2
4
Cl series of catalysts, because of lower conversions with the
former materials.
that of Al2O3–Cl. Moreover, the incorporation of Sn from
both precursors only slightly decreases the Lewis acidity of
the supported Sn catalysts compared with that of the chlo-
rided alumina material. It is also interesting to note that the
supported Sn catalysts synthesized from SnCl2 drastically
enhance the formation of isopropylbenzene and benzene,
3.2.2. SnSnBu /Al2O3–Cl catalysts
4
Table 5 lists the formation rates of the isomerized and hy-
drodealkylated products of n-PB with increasing Sn content

260
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
Fig. 3. Concentrations of isopropylbenzene, benzene, toluene and the sum
of ethylbenzene and styrene versus contact time, in the conversion of
◦
n-propylbenzene at 500 C, 5 bars, and H /HC = 5 over 0.2Sn
/
2
SnCl2
Al O –Cl (extrapolated values: dotted lines). Data recorded for 5 h of run.
2
3
although this only occurs up to 0.2 wt% Sn for benzene
(Table 4), whereas the formation of isopropylbenzene is not
affected by the introduction of Sn from SnBu4, and the for-
mation of benzene decreases with SnBu4 addition (Table 5).
The correlation between FTIR characterization of the acidic
properties and the reactivity of the 0.2Sn/Al2O3–Cl catalysts
supports the formation of isopropylbenzene and benzene cat-
alyzed by the Brønsted acid sites. In addition, the profile
concentration of these products as a function of contact time
suggests that n-PB first isomerizes to isopropylbenzene be-
fore cracking to benzene (Fig. 3). The formation of benzene
from n-PB is shown in Fig. 4. Dehydrogenated products of
n-PB, already reported in a previous study over alumina cat-
alysts [4], are also detected over the supported tin catalysts
(not shown). However, the formation of these compounds
is not the aim of the present section and will be discussed
later (Section 4.2.3); hence their formation from n-PB is
symbolized by a catalytic cycle in which the nature of the
involved acid sites is assumed to be a Lewis acid site (top
of Fig. 4). In Fig. 4, it is assumed that dehydrogenated prod-
ucts of n-PB are protonated to a benzylic carbenium ion;
this cation is the most stable, as it is resonantly stabilized by
the aromatic nucleus. This benzylic carbenium ion then iso-
merizes to an isopropylbenzene carbenium ion either via a
protonated cyclopropane intermediate or via a methyl shift
[18]. It is noteworthy that this isomerization step leads to
a primary carbenium ion, the formation of which is known
to be highly unfavored [18,19]. However, in this particular
case, this primary carbenium ion might readily isomerize
to the corresponding highly stable tertiary isopropylbenzene
carbenium ion via a 1,2 hydrogen atom shift; such carbe-
nium ion rearrangements have indeed been reported as facile
[20]. The hydride transfer [18,19] from a n-PB molecule
yields to isopropylbenzene and a benzylic carbenium ion;
the latter intermediate allows the izomerization cycle to turn
over.
◦
Fig. 2. Infrared spectra of adsorbed pyridine on calcined (air, 500 C,
◦
2 h) and reduced (H , 500 C, 2 h) catalysts with increasing evacua-
2
tion temperature: (a) Al O –Cl, (b) 0.2Sn
/Al O –Cl, and (c) 0.2
2 3
2
3
SnCl2
Sn
SnBu4
/Al O –Cl.
2 3

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
261
Fig. 4. Carbenium-ion cracking of benzene catalyzed by Brønsted acid sites.
The cracking of isopropylbenzene has largely been stud-
4.2. Formation of toluene and ethylbenzene
4.2.1. Carbenium-ion chemistry
If the formation of isopropylbenzene and benzene might
be reasonably attributed to carbenium ion chemistry, there is
a possible rebuttal to the formation of toluene and ethylben-
zene and thus styrene, which is the dehydrogenated analog
of ethylbenzene, via classic C–C β scissions of carbenium
ion chain mechanisms.
The only carbenium ion that allows for the formation of
ethylbenzene is the benzylic carbocation that is also involved
in the isomerization catalytic cycle (catalytic cycle on the
left side of Fig. 4). It is obvious that the C–C β scission
of such a carbenium ion leads to the formation of styrene
and a methenium ion (CH3+) [4], with the rather doubtful
formation of the latter from a thermodynamic point of view,
ruling out the involvement of a carbenium ion mechanism
for the formation of ethylbenzene.
On the other hand, it is worth considering the possibility
of the formation of toluene via a rather complex transfor-
mation of a carbenium ion produced by protonation of the
benzene ring at the ortho position [4]. In such an eventuality,
the formation of toluene occurs together with the formation
of a primary ethylene carbenium ion (+CH2–CH3). How-
ever, this carbenium ion cannot rearrange to a more stable
carbenium ion, as was the case in the isomerization process
of n-PB (Section 4.1, catalytic cycle on the left side of
ied in the literature as a model reaction to characterize the
strength of the sites of acidic catalysts [21–25]. It is gener-
ally accepted that the cracking of isopropylbenzene is cat-
alyzed by Brønsted acid sites [23–25] by protonation of
the ipso position of the aromatic nucleus [25] to yield the
protonated benzene ring carbenium ion (catalytic cycle on
the right side of Fig. 4). A classic C–C β scission of this
carbenium ion leads to benzene and propene after the loss
of a proton from the isopropyl carbocation. Based on the
aforementioned possible carbenium ion 1,2 H atom shift re-
arrangement in the case of the isomerization of n-PB, one
might also consider the direct C–C bond scission of the lin-
ear alkyl chain group of n-PB. This β scission would yield to
benzene and a primary carbenium propyl ion; the subsequent
rearrangement of the latter to an isopropyl cation and the loss
of a proton also leads to propene. However, benzene appears
as a stable secondary product and isopropylbenzene appears
as an unstable primary product (Fig. 3, [17]). The mecha-
nism reported in Fig. 4 for the formation of benzene is there-
fore the more likely. Moreover, this mechanism is in good
agreement with the cracking of linear alkanes, for which iso-
merization steps are known to precede C–C β scissions to
avoid the intermediate formation of primary carbenium ions
[18]. Finally, it is worth mentioning that for poorly acidic
catalysts, such as Al2O3, a radical cracking of isopropylben-
zene was also suggested to occur to a significant extent [22].

262
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
Fig. 5. Protolytic cracking of n-PB via penta-coordinated carbonium ions adapted from the work of Haag and Dessau [32].
Fig. 4). Although the stability of such a primary carbenium
ion might be greatly questionable, it is well known that alky-
lation of benzene by ethylene occurs over acidic catalysts
[26,27], suggesting that these materials are able to stabilize
the ethylene primary carbenium ion. It is generally accepted
that the mechanism of aromatic alkylation takes place via a
carbenium ion, as the formation of the latter results from the
activation of olefin by Brønsted acid sites [16,27]. The asso-
ciated base of the Brønsted acid sites catalyzing the alkyla-
tion reaction is therefore able to stabilize the ethylene carbe-
nium ion and hence performs the same role for the reverse
reaction of benzene alkylation (i.e., ethylbenzene dealkyla-
tion). In the present work, a similar path could be considered
for the hydrodealkylation of n-PB to toluene with the stabi-
lization of an ethylene carbenium ion. Nevertheless, one can
carbenium ion for C2 products. Another strikingly obvious
difficulty is that the dominant carbenium ion derived from
i-butane does not have a β C–C bond. To overcome these
−
difficulties, Haag and Dessau originally suggested the ex-
istence of a carbonium mechanism also referred to as pro-
tolytic cracking [32]. This mechanism involves the direct
protonation of the alkane molecule by very strong Brøn-
sted acid sites to form a pentacoordinated carbonium ion
(+CxH2x+3). The decomposition of this carbonium ion with
the cleavage of α bonds results in the formation of C1–C3
light alkanes and carbenium ions that further transform into
olefins with the loss of protons.
The main argument against the formation of toluene via
protolytic cracking has already been stated above (Sec-
tion 4.2.1). Indeed, it is difficult to conceive that toluene
production takes place via a carbonium ion mechanism as
its formation rate decreases (Table 4) with the introduction
of 0.2 wt% Sn to Al2O3–Cl, whereas the Brønsted acid-
observe that the Brønsted acidity of 0.2SnSnCl /Al2O3–Cl in-
2
creases sharply compared with Al2O3–Cl (Table 3), and the
formation of toluene decreases (Table 4). This suggests that,
in the present study, Brønsted acid sites do not catalyze the
production of toluene via a carbenium ion chain mechanism.
ity of 0.2SnSnCl /Al2O3–Cl is considerably greater than that
2
of Al2O3–Cl (Table 3). On the other hand, the enhanced
sum of the formation rates of ethylbenzene and styrene
with the introduction of 0.2 wt% Sn from SnCl2 (Table 4)
might be consistent with protolytic cracking of n-PB. As
already stated above, the Brønsted acidity also increases sig-
nificantly with the introduction of 0.2 wt% Sn from SnCl2
(Table 3). However, it must be emphasized that the sum
of the formation rates of ethylbenzene and styrene is in-
creased by only 50% with the introduction of Sn (Table 4),
although the concentration of the strong Brønsted acid sites
4.2.2. Carbonium ion chemistry
As reported above, none of the existing C–C bond fis-
sion schemes based on carbenium ion chemistry can easily
account for the formation of ethylbenzene and toluene in
the present study. Similarly, Notari et al. [28] stated that the
carbenium ion mechanism could not be applied for the hy-
drodealkylation of methyl- and ethyl-substituted aromatics.
Such a conclusion has also been drawn in previous works
[18,20,29–31] that focused on the cracking of small mole-
cules, such as n-butane or i-butane, to C1–C3 products over
acidic catalysts. Indeed, the cracking of n-butane via a carbe-
nium ion mechanism implies the formation of a methenium
ion for the formation of C1 and C3 compounds or a primary
over 0.2SnSnCl /Al2O3–Cl is about twice that over Al2O3–
2
Cl (Table 2). Moreover, the protolytic cracking of n-PB must
also favor the formation of benzene and toluene (Fig. 5), as
the protonation of a C–H bond is supposed to occur pref-
erentially on the most substituted carbon atom [32] or on a

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
263
carbon atom located in an α position with respect to a dou-
Table 6
Influence of the nature of the Sn precursor on the acidity of the catalysts
and the sum of the formation rates of ethylbenzene and styrene, and the
formation rate of toluene
ble bond [33], which is typically the case for the benzylic
carbon atom of n-PB (Fig. 5a). However, in this study, we
showed that toluene formation decreased (Table 4) when the
Catalysts
Brønsted acidity of the SnSnCl /Al2O3–Cl catalyst increased
2
Al O –Cl 0.2Sn
/
0.2Sn
Al O –Cl
2 3
/
2
3
SnCl2
SnBu4
(Table 3) compared with Al2O3–Cl. In addition to these ar-
guments, it is important to emphasize that pure protolytic
cracking is supposed to occur predominantly at conversion
approaching zero [19,20,31,32], in the presence of low con-
centrations of olefins [34], because olefins are much better
proton acceptors than alkanes [20,34,35], and at low par-
tial pressure of hydrocarbons [19,32]. Let us recall that the
present study was performed at conversions as high as 20%,
in the presence of olefins [4] and with a rather high partial
pressure of n-PB (0.8 bar).
Al O –Cl
2
3
Formation rate
−5
−1 −1
(10 mol L
s )
Ethylbenzene + styrene
2.3
2.9
3.6
2.2
3.2
1.7
Toluene
Lewis acidity (a.u.)
Weak Lewis acid sites
Medium–strong Lewis acid sites 1.90
1.98
2.22
1.34
1.43
1.64
teresting theoretical work about the mechanism and the re-
activity of alkane C–H bond dissociation on coordinatively
unsaturated aluminum ions was recently published by Fa˘r-
cas¸iu and Lukinskas [40]. In this work, the authors showed
that the reaction of propane over unsaturated aluminum ions,
which typically correspond to Lewis acid sites, consisted of
the insertion of an aluminum ion into the C–H bond, fol-
lowed by hydrogen migration from Al to O to form the
adsorbed intermediate. The elimination of hydrogen atoms
from Cβ and oxygen then gives H2 and propene. It is worth
mentioning that such a detailed mechanism corroborates pre-
vious results suggesting that the presence of Lewis acid or
electron acceptor sites of many acidic catalysts can enhance
the dehydrogenation of paraffins [35,41–43]. Based on the
work of Fa˘rcas¸iu and Lukinskas [40], the insertion of a C–H
bond of n-PB located either on the benzylic carbon atom
(Fig. 6a) or on the terminal carbon atom of the alkyl chain
group (Fig. 7) is considered. In Figs. 6a and 7, for the sake of
simplicity, the catalytic site is reduced to an Al–O pair with
an unsaturated aluminum ion. Fig. 6a shows that the reac-
tive adsorbate resulting from the adsorption of n-PB via the
C–H bond of the benzylic carbon atom might be transformed
via two different paths. Similar to Fa˘rcas¸iu and Lukinskas’
proposal [40], the elimination of the hydrogen atom from
Cβ leads to the formation of dehydrogenated compounds of
n-PB and H2 (catalytic cycle on the right side of Fig. 6a),
whereas the radical C–Cβ scission leads to the formation of
styrene and methane. The hydrogenation of styrene to eth-
ylbenzene over Lewis acid sites (Fig. 6b), the reverse of the
reaction proposed by Fa˘rcas¸iu and Lukinskas [40], requires
the dissociation of H2 over the catalytic site [44] and the
reaction of these hydrogen atoms with a π-bonded mole-
cule [40] of styrene. The hydrogenation of styrene can also
be considered over the Brønsted acid sites with the protona-
tion of styrene followed by a hydride transfer [18,19,24,34]
from n-PB and the concurrent formation of a benzylic carbe-
nium ion (Fig. 6c). The loss of a proton from this carbenium
ion leads to the formation of the dehydrogenated products of
n-PB and the restoration of the catalytic Brønsted acid site.
The existence of this hydrogenation path is supported by
the work of Holm and Blue [45], who showed that ethylene
4.2.3. Radical chemistry
As both carbenium ion and carbonium ion chemistries
do not satisfactorily predict the formation of toluene and
ethylbenzene or styrene from a mechanistic point of view,
a radical-like cracking of n-PB was considered. Such radi-
cal pathways were suggested for the cracking of both alkyl
aromatics and alkanes. As early as 40 years ago, Tung
and McIninch [22] and Notari et al. [28] suggested radical
mechanisms for the cracking of isopropylbenzene over high-
purity alumina and that of toluene over alumina-modified
catalysts, respectively. More recently, radical pathways have
also been suggested to occur, to a significant extent, in the
cracking of hexane isomers over H-mordenite [36], of n-but-
ane over H-ZSM5 [30], and of i-butane over silica- and
halided-alumina catalysts [29]. Although the last work was
the subject of great controversy [37,38] by the defenders
of the protolytic cracking mechanism [34,38]—most prob-
ably because the product distributions obtained by purely
protolytic cracking and radical mechanisms are very sim-
ilar [19]—recent reviews support the existence of radical
mechanisms for the cracking of alkanes [19,33,39]. The ini-
tiation of the radical mechanism is assumed to occur over
electron acceptor sites [19,36,37,39] such as Lewis acid
sites [19,33,39].
Table 6 lists the sum of the formation rates of ethyl-
benzene and styrene and the formation rate of toluene to-
gether with the characterization of the Lewis acidity of the
Sn/Al2O3–Cl catalysts. In this table, the sum of the amounts
of medium and strong Lewis acid sites is also reported. It
can be seen that the sum of the formation rates of ethylben-
zene and styrene and the formation rate of toluene are fully
correlated with the amounts of weak and medium–strong
Lewis acid sites with the incorporation of Sn from SnCl2
to a chlorided alumina (Table 6). Indeed, an increase in the
sum of the formation rates of ethylbenzene and styrene and
a decrease in the formation rate of toluene are linked to an
increase in the amount of weak Lewis acid sites and a de-
crease in the medium–strong sites. These results thus suggest
that these compounds are formed by a radical mechanism
involving Lewis acid sites of different strengths. A very in-

264
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
Fig. 6. (a) Proposed radical cracking and dehydrogenation of n-PB to styrene and E- and Z-phenyl-1-propene, respectively, over Lewis acid sites adapted from
the work of Fa˘rcas¸iu and Lukinskas [40]; hydrogenation of styrene to ethylbenzene (b) over Lewis or (c) Brønsted acid sites.
could be almost fully hydrogenated to ethane over alumina
at 500 ◦C. Comparable catalytic cycles such as those for the
adsorption of n-PB via the benzylic carbon atom (Fig. 6a)
are shown for the adsorption of n-PB by the terminal carbon
atom of the alkyl chain (Fig. 7). In this case, the formation of
phenyl-2-propene and H2 (catalytic cycle on the right side of
Fig. 7) and toluene and ethylene (catalytic cycle on the left
side of Fig. 7) occurs.
in the order prim C–H > sec-C–H > tert-C–H, which is in
full agreement with the stability of the corresponding radical
species. Considering this reactivity order of the C–H bonds,
the lower formation of toluene compared with the sum of
those of ethylbenzene and styrene (Table 6) might be sur-
prising, as the formation of toluene involves the insertion of
a prim C–H bond (Fig. 7), whereas that involved for the for-
mation of ethylbenzene and styrene is a sec-C–H (Fig. 6a).
However, in the case of n-PB, one must also consider the fact
that the benzylic radical (Fig. 6a) is the most stable of the
Fa˘rcas¸iu and Lukinskas [40] also concluded that the in-
sertion of the aluminum ion in the C–H bond was favored

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
265
Fig. 7. Proposed radical cracking and dehydrogenation of n-PB to toluene and phenyl-2-propene, respectively, over Lewis acid sites adapted from the work of
Fa˘rcas¸iu and Lukinskas [40].
above-mentioned radicals because it is resonantly stabilized
by the aromatic ring [28]. This peculiarity therefore accounts
for the higher formation of the sum of ethylbenzene and
4.3. Influence of the content of Sn and the nature of the Sn
precursor
styrene than that of toluene for 0.2SnSnCl /Al2O3–Cl com-
2
Tables 4 and 5 clearly show that the content of Sn de-
posited on Al2O3–Cl and/or the nature of the Sn precursor
do not evenly affect the production of the isomerized and
cracked products and, thus, the acidity of the catalysts as
shown in Table 3 for the 0.2Sn/Al2O3–Cl materials.
For both series of supported Sn catalysts, the formation
rates of the cracked compounds decrease with increasing Sn
content (Tables 4 and 5). These results are consistent with a
decrease in the acidity of the catalysts, as already reported in
previous studies [7,8,12–14]. In contrast, the production of
pared with those found for Al2O3–Cl. It is also interesting
to note that the stability of the speculated radicals is consis-
tent with the strength of the Lewis acid sites assumed to be
responsible for the formation of the two classes of product
(Table 6). Indeed, the formation of the benzylic radical, in-
volved in the formation of ethylbenzene or styrene (Fig. 6a),
is more likely to occur on weaker Lewis acid sites than on
those engaged in the formation of toluene, the radical in-
termediate of which (Fig. 7) is less stable than the benzylic
radical intermediate. It must be added that the radical mech-
anisms suggested in the present study are nothing but radical
C–C cleavages in the β position, as already proposed by
Leveles et al. [46] for the oxidative conversion of propane.
Finally, Tung and McIninch [22] suggested an alterna-
tive path for the formation of toluene from isopropylbenzene
with the intermediate isomerization of isopropylbenzene to
n-PB, followed by decomposition of the 3-phenyl-propyl
radical, which is identical to that proposed in the present
work (catalytic cycle on the left side of Fig. 7).
isopropylbenzene is almost constant for the SnSnBu series,
4
whereas a maximum is obtained for the catalyst including
0.2 wt% Sn in the SnSnCl series. As suggested from the
2
results of the present work, the production of isopropylben-
zene and benzene occurs over the Brønsted acid sites via
carbenium ion chemistry (Section 4.1). It is interesting to
note that the ratio of the formation rate of isopropylben-
zene to that of benzene increases with increasing Sn content
for both series. A likely explanation for this trend is a de-
crease in the strength of the Brønsted acid sites with Sn
addition. This explanation is consistent with previous stud-
ies that reported a decrease in the acidity of the support with
the introduction of Sn, which resulted in higher selectivity
for isomerization and lower selectivity for the cracking of
C6–C8 alkanes over PtSn/Al2O3 catalysts [47–49]. The rel-
ative formation of toluene and ethylbenzene, attributed to
Lewis acidity via radical chemistry, is also consistent with
a decrease in the strength of the Lewis acid sites; the forma-
The correlation with the Sn catalyst synthesized from
SnBu4 is less obvious than that with 0.2SnSnCl /Al2O3–Cl
2
(Table 6). However, the distribution of the acid strength
of 0.2SnSnBu /Al2O3–Cl is very close to that of Al2O3–Cl
4
with a rather large proportion of medium Lewis acid sites
(Table 3). In this case, the arbitrary desorption tempera-
ture range used for the assignment of the strength of the
Lewis acid sites might not be adapted as done for that of
0.2SnSnCl /Al2O3–Cl.
2

266
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
Fig. 8. Reaction network of the transformation of n-PB over chlorided alumina catalysts.
tion of toluene involves Lewis acid sites stronger than those
needed for that of ethylbenzene (Section 4.2.3).
The comparison of the supported Sn catalysts with
Al2O3–Cl is more complicated, and the main differences can
be summarized as follows: (i) a drastic increase in the for-
Finally, it is worth mentioning that benzene is always the
major product of the cracked compounds for all of the cata-
lysts investigated in the present work.
4.4. Reaction network of the transformation of n-PB over
mation of isopropylbenzene over the SnSnCl series, whereas
chlorided alumina acidic catalysts
2
that of the SnSnBu series is not affected, and (ii) produc-
4
tion of benzene promoted by the incorporation of Sn up to
The results discussed in the present study allowed us to
propose the reaction network of the transformation of n-PB
over chlorided alumina acidic catalysts shown in Fig. 8. This
reaction network put a particular emphasis on the nature
of the mechanisms involved in the formation of the differ-
ent products and the nature and the strength of the required
acid sites. It is quite interesting to note that ethylbenzene
and toluene might be formed by radical monofunctional
acidic pathways over the Lewis acid sites. The production
of ethylbenzene involves Lewis acid sites weaker than those
involved in the formation of toluene. In contrast, the for-
mation of isopropylbenzene and benzene would be formed
by bifunctional acidic mechanisms, as already suggested by
Schuette and Schweizer [39]. Their production involves the
intermediate formation of dehydrogenated products over the
Lewis acid sites. These dehydrogenated compounds are then
0.2 wt% for the SnSnCl series, whereas benzene production
2
is slightly inhibited for the SnSnBu series. This study there-
4
fore reveals that the incorporation of Sn via SnCl2 strongly
promotes Brønsted acidity for Sn contents of 0.2 wt% or
less, whereas that via SnBu4 does not significantly modify
it. It must be emphasized that the promotional effect of Sn
on the Brønsted acidity had not been stated up to now. The
reasons for this are (i) most of the works investigated cata-
lysts with Sn contents higher than 0.2 wt% [7,8,12,15,50];
(ii) for a Sn content of 0.11 wt%, the catalyst studied by
Sheng et al. was chloride-free [13], whereas it is well known
that the acidity of Al2O3 is promoted by the addition of Cl−
[51–53]; and (iii) the nature of the Sn precursor was rarely
SnCl2 [7]. It is thus clear that not only the content of Sn is
important, but also the nature of the Sn precursor.

S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
267
catalytically converted to isopropylbenzene and benzene by
the Brønsted acid sites. Following this reaction pathway, it
is very likely that isopropylbenzene would require Brønsted
acid sites weaker than those needed for the formation of
benzene. This conclusion is in good agreement with basic
cracking principles for which acid sites stronger than those
needed for isomerization are required [18,49]. Although the
existence of protolytic cracking of n-PB appears unlikely,
its intervention as an initiation step to produce the benzylic
carbenium ion shown in Fig. 8 cannot be ruled out.
that of Al2O3–Cl, with a rather large proportion of medium
Lewis acid sites. In this case, the arbitrary desorption tem-
perature range used for the assignment of the strength of the
Lewis acid sites might not adapted as done for the sites of
0.2SnSnCl /Al2O3–Cl.
2
Finally, it is worth noting that benzene is always the ma-
jor product of the cracked compounds for all of the acidic
catalysts investigated in the present work.
Acknowledgments
5. Conclusion
The Institut Français du Pétrole (IFP) provided financial
support for this work; the ANRT organization supported the
work of Ms. S. Toppi (CIFRE grant 128/99). We thank Dr.
C. Marcilly (IFP) for fruitful discussions and for his interest
in this work. Finally, we gratefully acknowledge P. Lavaud
for technical support.
Two series of chlorided alumina-supported Sn cata-
lysts were synthesized with different precursors, SnCl2 or
SnBu4, and with various Sn contents ranging from 0.1 to
0.5 wt%. The acidity of the catalysts including 0.2 wt% Sn
was characterized by FTIR adsorption–desorption of 2,6-
dimethylpyridine (Brønsted acidity) and pyridine (Lewis
acidity) and compared with that of Al2O3–Cl. The catalytic
activity of the synthesized materials was investigated for the
transformation of n-propylbenzene under reforming condi-
tions.
The results show that the incorporation of Sn into Al2O3–
Cl is not harmless. Neither the nature of the Sn precursor
nor the content of Sn evenly affects the acidity and the dis-
tribution of the isomerized and cracked products. Compared
with Al2O3–Cl, the introduction of Sn from SnBu4 slightly
decreases the acidity of the materials and the production of
benzene and toluene, whereas the production of isopropyl-
benzene remains constant and that of ethylbenzene increases
slightly. In contrast, the incorporation of Sn from SnCl2
drastically increases the Brønsted acidity; this effect is the
most pronounced for Sn contents of 0.2 wt% or less. In ad-
dition, the incorporation of Sn from SnCl2 decreases Lewis
acidity very slightly.
References
[1] B.H. Davis, Catal. Today 53 (1999) 443.
[2] J.H. Gary, G.E. Handwerk, in: L.F. Albright, R.N. Maddox, J.J.
McKetta (Eds.), in: Petroleum Refining: Technology and Economics,
vol. 5, Marcel Dekker, New York, 1975, p. 65.
[3] P.G. Menon, Z. Paál, Ind. Eng. Chem. Res. 36 (1997) 3282, and refer-
ences therein.
[4] S. Toppi, C. Thomas, C. Sayag, D. Brodzki, F. Le Peltier, C. Travers,
G. Djéga-Mariadassou, J. Catal. 210 (2002) 431.
[5] S. Toppi, C. Thomas, C. Sayag, D. Brodzki, F. Le Peltier, C. Travers,
G. Djéga-Mariadassou, J. Catal. 218 (2003) 411.
[6] R. Burch, J. Catal. 71 (1984) 348.
[7] S.R. Bajaj, P. Pal, J.K. Gupta, L.D. Sharma, G. Murali Dhar, T.S.R.
Prasada Rao, Stud. Surf. Sci. Catal. 113 (1998) 365.
[8] P. Kirszensztejn, B. Czajka, T.-C. Sheng, T.N. Bell, I.D. Gay, Catal.
Lett. 32 (1995) 305.
[9] M. Moreno, A. Rosas, J. Alcaraz, M. Hernández, S. Toppi, P. Da Costa,
Appl. Catal. A 251 (2003) 369.
[10] A. Corma, C. Rodellas, V. Fornes, J. Catal. 88 (1984) 374.
[11] A. Corma, V. Fornés, E. Ortega, J. Catal. 92 (1985) 284.
[12] P. Kirszensztejn, W. Przystajko, T.N. Bell, Catal. Lett. 18 (1993) 391.
[13] J. Shen, R.D. Cortright, Y. Chen, J.A. Dumesic, Catal. Lett. 26 (1994)
247.
The correlation between the product distribution obtained
for the transformation of n-propylbenzene and the acidity of
Al2O3–Cl and the 0.2SnSnCl /Al2O3–Cl supports the forma-
2
tion of isopropylbenzene and benzene, catalyzed by Brøn-
sted acid sites via carbenium ion chemistry. In contrast, the
production of toluene and ethylbenzene cannot be attributed
to carbenium ion or carbonium ion chemistries, but more
likely to radical chemistry. The formation of these products
is assumed to be catalyzed by Lewis acid sites with different
strengths. The stability of the proposed radical intermedi-
ate species is consistent with the involvement of stronger
Lewis acid sites for the production of toluene, compared
with those involved for the production of ethylbenzene. The
catalytic cycles responsible for the formation of toluene and
ethylbenzene via radical pathways over an Al–O pair are re-
ported. The correlation with the Sn catalyst synthesized from
SnBu4 is less obvious than the correlation with the Sn cat-
[14] T.-C. Sheng, P. Kirszensztejn, T.N. Bell, I.D. Gay, Catal. Lett. 23
(1994) 119.
[15] D. Sprinceana, M. Caldararu, N.I. Ionescu, A. Auroux, J. Ther. Anal.
Calor. 56 (1999) 109.
[16] L. Rodríguez, J. Alcaraz, M. Hernández, Y. Ben Taârit, M. Vrinat,
Appl. Catal. A 189 (1999) 53.
[17] D. Best, B.W. Wojciechowski, J. Catal. 47 (1977) 11.
[18] Y.V. Kissin, Catal. Rev. 43 (2001) 85.
[19] A. Corma, A.V. Orchillés, Micropor. Mesopor. Mater. 35–36 (2000)
21.
[20] F.C. Jentoft, B.C. Gates, Topics Catal. 4 (1997) 1.
[21] C.D. Prater, R.M. Lago, Adv. Catal. 8 (1956) 295.
[22] S.E. Tung, E. McIninch, J. Catal. 4 (1965) 586.
[23] R. Covini, V. Fattore, N. Giordano, J. Catal. 7 (1967) 126.
[24] M. Guisnet, Stud. Surf. Sci. Catal. 20 (1985) 283.
[25] A. Corma, J.L.G. Fierro, R. Montana, F. Tomas, J. Mol. Catal. 30
(1985) 361.
alyst from 0.2SnSnCl /Al2O3–Cl. However, the Lewis acid
2
strength distribution of 0.2SnSnBu /Al2O3–Cl is very close to
[26] J. Weitkamp, Acta Phys. Chem. 31 (1985) 271.
4

268
S. Toppi et al. / Journal of Catalysis 230 (2005) 255–268
[27] P.G. Smirniotis, E. Ruckenstein, Ind. Eng. Chem. Res. 34 (1995) 1517.
[28] B. Notari, P. Duranti, Valentini, M. Le Maldé, in: Proc. 3rd Congr. on
Catalysis, Berlin, vol. 2, Verlag Chemie, Amsterdam, 1965, p. 1034.
[29] G.B. McVicker, G.M. Kramer, J.J. Ziemiak, J. Catal. 83 (1983) 286.
[30] Y.W. Bizreh, B.C. Gates, J. Catal. 88 (1984) 240.
[31] H. Krannila, W.O. Haag, B.C. Gates, J. Catal. 135 (1992) 115.
[32] W.O. Haag, R.M. Dessau, in: Proc. 8th Congr. on Catalysis, Berlin,
vol. 2, Verlag Chemie, Weinheim, 1984, p. 305.
[42] A. Brait, A. Koopmans, H. Weinstabl, A. Ecker, K. Seshan, J.A. Ler-
cher, Ind. Eng. Chem. Res. 37 (1988) 873.
[43] T.F. Narbeshuber, A. Brait, K. Seshan, J.A. Lercher, J. Catal. 172
(1997) 127.
[44] D. Fa˘rcas¸iu, P. Lukinskas, J. Phys. Chem. A 103 (1999) 8483.
[45] V.C.F. Holm, R.W. Blue, Ind. Eng. Chem. 43 (1951) 501.
[46] L. Leveles, K. Seshan, J.A. Lercher, L. Lefferts, J. Catal. 218 (2003)
296.
[33] A. Corma, Stud. Surf. Sci. Catal. 49 (1989) 49.
[34] S. Kotrel, H. Knözinger, B.C. Gates, Micropor. Mesopor. Mater. 35–36
(2000) 11.
[35] T.F. Narbeshuber, A. Brait, K. Seshan, J.A. Lercher, Appl. Catal.
A 146 (1996) 119.
[36] D. Fa˘rcas¸iu, K.-H. Lee, J. Catal. 219 (2003) 186.
[37] G.M. Kramer, G.B. McVicker, J. Catal. 115 (1989) 608.
[38] W.K. Hall, E.A. Lombardo, J. Enghelardt, J. Catal. 115 (1989) 611.
[39] W.L. Schuette, A.E. Schweizer, Stud. Surf. Sci. Catal. 134 (2001) 263.
[40] D. Fa˘rcas¸iu, P. Lukinskas, J. Phys. Chem. A 106 (2002) 1619.
[41] S.E. Tung, E. McIninch, J. Catal. 10 (1968) 166.
[47] R. Burch, L.C. Garla, J. Catal. 71 (1981) 360.
[48] R. Bacaud, P. Bussière, F. Figueras, J. Catal. 69 (1981) 399.
[49] M.C. Rangel, L.S. Carvalho, P. Reyes, J.M. Parera, N.S. Figoli, Catal.
Lett. 64 (2000) 171.
[50] R. Srinivasan, B.H. Davis, J. Mol. Catal. 88 (1994) 343.
[51] A. Ayame, G. Sawada, H. Sato, G. Zhang, T. Ohta, T. Izumizawa,
Appl. Catal. 48 (1989) 25.
[52] T. Cai, S. Liu, J. Qü, S. Wong, Z. Song, M. He, Appl. Catal. A 97
(1993) 113.
[53] G. Clet, J.M. Goupil, G. Szabo, D. Cornet, J. Mol. Catal. 148 (1999)
253.