Low-loaded Pd/α-Al2O3 catalysts: Influence of metal particle morphology on hydrogenation of buta-1,3-diene and hydrogenation and isomerization of but-1-ene

Article abstract of DOI:10.1006/jcat.1996.0393

Buta-1,3-diene hydrogenation and but-1-ene hydrogenation and isomerization were studied on low-loaded Pd/α-Al₂O₃ catalysts (0.1-0.3 wt.%) prepared from palladium acetylacetonate (Pd(C₅H₇O₂)₂). Deuterium tracer study, hydrogen chemisorption, transmission electron microscopy, and X-ray photoelectron spectroscopy analysis were used to establish the relationships between metal-support interactions, particle shapes, and buta-1,3-diene and but-1-ene hydrogenation and isomerization mechanisms. It was found that the hydrogenation reaction rates (turnover frequencies) are similar for buta-1,3-diene and but-1-ene, but 10 times lower for the 0.1% Pd catalyst compared to the 0.3% Pd catalyst. However, the 0.1% Pd catalyst has a high activity for the isomerization reaction which leads to 98% selectivity in isomers for the but-1-ene reaction. This unusual specific activity is explained considering that the 0.1% Pd catalyst contains flat particles in strong interaction with the support, as was deduced from several characterization methods, while the 0.3% Pd catalyst has more rugged bulk type particles after H₂ treatments because they are not interacting with the support.

Full text of DOI:10.1006/jcat.1996.0393

JOURNAL OF CATALYSIS 164, 369–377 (1996)
ARTICLE NO. 0393
Low-Loaded Pd/�-Al O Catalysts: Influence of Metal Particle
2
3
Morphology on Hydrogenation of Buta-1,3-diene and
Hydrogenation and Isomerization of But-1-ene
�
�
,1
J. Goetz, M. A. Volpe,† and R. Touroude
�
Laboratoire d’Etudes de la R e´ activit e´ Catalytique, des Surfaces et Interfaces, URA 1498 du CNRS, Universit e´ Louis Pasteur, 4 rue Blaise Pascal,
F-67070 Strasbourg Cedex, France; and †Planta Piloto de Ingeneria Quimica (UNS-CONICET),
1
2 de Octobre 1842, 8000 Bahia Blanca, Argentina
Received February 14, 1996; revised May 8, 1996; accepted August 21, 1996
buta-1,3-diene, and isoprene hydrogenation, performed in
Buta-1,3-diene hydrogenation and but-1-ene hydrogenation the liquid phase, exhibited a clear dependence on the parti-
and isomerization were studied on low-loaded Pd/�-Al2O3 cata- cle size. The catalytic activity decreased rapidly for particle
lysts (0.1–0.3 wt.%) prepared from palladium acetylacetonate
diameters smaller than 3–5 nm (dispersion � 20% ). The au-
(Pd(C5H7O2)2). Deuterium tracer study, hydrogen chemisorption,
thors proposed an intrinsic modification of the properties of
the metallic sites, specifically a stronger adsorption of the
unsaturated hydrocarbons on the smaller particles which
leads to a decrease of the turnover frequency. Other authors
obtained results which show the sensitivity of the hydro-
genation reaction to the Pd particle size using the same type
of alumina-supported palladium catalysts (2, 3) or other
transmission electron microscopy, and X-ray photoelectron spec-
troscopy analysis were used to establish the relationships between
metal–support interactions, particle shapes, and buta-1,3-diene and
but-1-ene hydrogenation and isomerization mechanisms. It was
found that the hydrogenation reaction rates (turnover frequencies)
are similar for buta-1,3-diene and but-1-ene, but 10 times lower for
the 0.1% Pd catalyst compared to the 0.3% Pd catalyst. However, the
0
.1% Pd catalyst has a high activity for the isomerization reaction types of palladium catalysts (4, 5), studying either the same
which leads to 98% selectivity in isomers for the but-1-ene reac- (3) or different alkadienes and alkynes (2, 4, 5). The core
tion. This unusual specific activity is explained considering that the level binding energy of the palladium 3d electrons, exam-
0
.1% Pd catalyst contains flat particles in strong interaction with
ined by X-ray photoelectron spectroscopy, has been found
to be higher for small particles supported on alumina (3),
silica (6), and carbon (7), compared to bulk palladium, but
it has been found to be lower when particles were supported
on pumice (5). Except for Pd/pumice, in which Na ionscould
play an important role in electronic and then catalytic prop-
erties of small Pd particles, it has been suggested that the
low density of states at the Fermi level for small particles
the support, as was deduced from several characterization methods,
while the 0.3% Pd catalyst has more rugged bulk type particles af-
ter H2 treatments because they are not interacting with the support.
�
c 1996 Academic Press, Inc.
INTRODUCTION
It is well known that Pd/Al2O3 catalysts present high ac- would lead to a strong adsorption of the alkyne or diene
tivity and selectivity for the semihydrogenation of alkynes which are electron-rich molecules and thus would give rise
and alkadienes; this specificity is used in many industrial to a decrease of the hydrogenation rate (1, 3, 8, 9).
processes for the selective hydrogenation of acetylenic and
diolefinic compounds in the presence of large amounts of obtained by the procedure described by Boitiaux et al. (1).
olefins.
Their palladium contents are low (0.1–0.3 wt.% ) and their
2 3
Pd/�-Al O catalysts examined in this paper have been
Palladium being a precious metal, high dispersion on the dispersion varies from 54 to 23% . From a previous study
support is desired for the preparation of industrial cata- (10), two types of catalysts emerge: one presenting par-
lysts. Boitiaux et al. (1), using palladium acetylacetonate ticles with a strong interaction between metal and sup-
(
Pd(C5H7O2)2) as a catalyst precursor, were able to con- port, the other being less interactive. These catalysts have
trol the particle size by varying the reduction temperature; been tested for buta-1,3-diene hydrogenation and but-1-
they obtained supported palladium catalysts with disper- ene hydrogenation and isomerization. In addition the but-
sions ranging from 9 to 100% . The activity of but-1-yne, 1-ene reactions have been followed by deuterium tracer
study. This work concentrates on the relationships between
1
metal–support interactions, particle shapes, and mecha-
nisms of hydrogenation and isomerization reactions.
To whom correspondence should be addressed. Fax: (33)88416147.
E-mail: touroude@chimie.u-strasbg.fr.
369
0021-9517/96 $18.00
Copyright �c 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

3
70
GOETZ, VOLPE, AND TOUROUDE
METHODS
X-Ray Photoelectron Spectroscopy
The measurements were performed using a VG Esca
III spectrometer with AlK� radiation (1486.6 eV) as in-
cident beam without a monochromator, as described pre-
viously (10). Before conducting the X-ray photoelectron
spectroscopy (XPS) analysis, the samples, already reduced
Catalyst Preparation
The catalysts were prepared by wet impregnation of
2
�
-Al2O3 (Rh oˆ ne Poulenc, 10 m /g, precalcined at 773 K)
with benzene solutions of palladium bis-acetylacetonate
Pd(C5H7O2)2), in accordance with the procedure de-
scribed by Boitiaux et al. (1). Benzene solutions containing
.7 and 3.33 g Pd/liter were used to obtain samples with Pd
(
at 573 K, were treated in situ at 573 or 773 K for 1 h under
5
1
atm of hydrogen (1 atm = 1.013 � 10 Pa) to avoid possi-
0
ble air contamination, which is known to influence the XPS
palladium lines in supported palladium catalysts (12).
Due to charging effects on the �-Al2O3 samples, the XPS
peaks were found to shift toward higher BE. Consequently,
the Al 2s BE was used as a reference to correct the XPS
peak position in all samples. It was fixed at 119.6 eV since
L e´ gar e´ and Fritsch (13) found this value referenced to the
Fermi level of Pd obtained by metal vapor deposition on a
single �-Al2O3 crystal. On this sample the Pd 3d5/2 BE was
found to be 335.2 eV.
loadings of 0.09 and 0.30 wt.% , respectively, keeping the
solution to support ratio constant. The resulting mixtures
were maintained at room temperature in closed flasks for 72
h with occasional stirring. The samples were then filtered,
dried overnight at 383 K, and calcined in air at 573 K for
2
h. Finally they were reduced in flowing hydrogen at the
same temperature and for the same period of time. Subse-
quently, a certain amount of each catalyst was reduced at
773 K for 1 h in order to increase the Pd particle size by
metal sintering. Metal loadings were determined by atomic
absorption spectroscopy (AAS).
It will be noted that using Al 2s at 119.6 eV to calibrate
the binding energies, the C 1s BE was found at 285.6 eV
whatever the sample; i.e., the charging correction using an
internal element of the support, or contaminating carbon,
leads to the same BE of Pd.
Transmission Electron Microscopy
The particle size distribution was determined by trans-
mission electron microscopy (TEM) using a Jeol 100
CX instrument operated at 100 kV (Centro Regional
de Investigaciones Basicas y Aplicaas de Bahia Blanca,
Catalytic Tests
Apparatus, procedure, and analysis of catalytic reac-
Argentina). The Pd catalysts were ground and dispersed tions. The catalytic reactions were carried out in a stain-
onto holey carbon-coated Cu grids for direct observation. less steel reactor operating under differential conditions
Over 100 particles were examined on the TEM micro- at atmospheric pressure. the total gas flow (H2 + He +
graphs and the average particle size has been calculated as hydrocarbon), controlled by a flowmeter, was varied by
3
2
d = 6ni d /6ni d .
changing the pumping rate at the end of the flow line. A
known quantity of hydrocarbon was carried over the cata-
lyst by the hydrogen flow at a constant hydrocarbon pres-
sure: 6.7 Torr and 7.2 Torr for buta-1,3-diene and but-1-ene,
i
i
Sorption Measurements
The fraction of exposed palladium atoms was derived respectively. The stability of the hydrocarbon pressure and
from hydrogen sorption measurements performed at room the reaction time was controlled by two katharometers in-
temperature in a static volumetric apparatus, following the serted upstream and downstream with respect to the re-
�
1
double isotherm method proposed by Benson et al. (11). actor, enabling the hydrocarbon flow rate (mol s ) to be
Prior to sorption measurements, the samples were treated measured.
in a stream of hydrogen at 573 K for 1 h and evacuated at the
Rectangular pulses of hydrocarbon carried by H2 (D2)–
same temperature overnight. The first isotherm, measured He mixture (38/722 Torr), were used for each experiment.
in the 0 to 14-kPa pressure range, gave the total H2 uptake. The H2 (D2)–He mixture was purified by passing through a
Subsequently, the catalysts were evacuated for 30 min and tube containing Pt/Al2O3 powder heated at 543 K, followed
a second isotherm was obtained to determine the amount by a cold trap at 196 K and a zeolite trap maintained at room
of absorbed and weakly adsorbed gas removed during the temperature to remove the oxygen and water impurities.
evacuation step. The linear portion of both isotherms was
The reaction products were drawn off the flow line
extrapolated to zero pressure to estimate the correspond- at different time during the catalytic run and analyzed
ing uptake values. The extent of irreversibly adsorbed gas by gas–liquid chromatography (GLC, using a 2.5-m-long,
�
3
was obtained by subtracting the contribution of the sec- 3.175 � 10 -m o.d. dimethyl sulfolane column operating at
ond isotherm from the total uptake. Assuming that the sur- 273 K with a flame ionization detector).
face stoichiometry of the irreversibly adsorbed hydrogen is
unity, the fraction of exposed palladium atoms is given by at 573 K during 1 h (heating rate: 10 K/min) and cooled to
H/Pd)irr. the reaction temperature (273K) under H2 (D2)/He flow. At
Before each catalyticexperiment, the catalyst wastreated
(

INFLUENCE OF PARTICLE MORPHOLOGY IN Pd/�-Al2O3 CATALYSTS
371
given by
T/C = (trans-but-2-ene)/(cis-but-2-ene),
and the selectivity in olefins toward butane is given for the
buta-1,3-diene reaction by
S =
(
but-1-ene + trans-but-2-ene + cis-but-2-ene)
,
(
but-1-ene + trans-but-2-ene + cis-but-2-ene + butane)
and for the but-1-ene reaction by
(
trans-but-2-ene + cis-but-2-ene)
S = (
.
trans-but-2-ene + cis-but-2-ene + butane)
When deuterium was used in the experiments, each hy-
drocarbon was separated by GLC and further analyzed by
FIG. 1. Conversion as a function of time. (᭹) 0.30% Pd/�-alumina;
(
� ) 0.09% Pd/�-alumina. Reduction temperature, 573 K; reaction tem- mass spectrometry (using a Varian type CH7 spectrome-
perature, 273 K. Partial pressures: but-1,3-diene, 6.7 Torr; hydrogen, 38 ter) according to the proper procedures described in detail
Torr; helium, 715.3 Torr.
elsewhere (14).
RESULTS AND DISCUSSION
the initial stage of the experiment the conversion decreases
slightly to reach a steady state after 10–20 min (Fig. 1) and
reproducible results are then obtained.
Table 1 summarizes the chemisorption and TEM data as
well as XPS results for samples reduced at 573 and 773 K
The conversion values (�) used for the rate calculation
were taken in the range where the rate equation k = (�F/ω)
was verified, F being the hydrocarbon flow in mol/s, and
ω the quantity of catalyst. If ω is given in grams of cat-
alyst (metal + support), the reaction rate k is obtained in
mol/s/catalyst g; if ω is given in term of surface metal atoms
(
10). Catalysts prepared with an impregnation solution con-
taining 0.7 g Pd/liter will be designated A; these samples
exhibit a 0.09 Pd wt.% . Catalysts B are prepared with an
impregnation solution containing 3.33 g Pd/liter and exhibit
a 0.30 Pd wt.% . The subscripts 1 and 2 are used for catalysts
reduced at 573 K and 773 K, respectively.
An increase in reduction temperature leadsto an increase
in particle size, more important for the catalyst A1 than for
the catalyst B1. It is worth noting that an increase in particle
size does not affect the Pd 3d5/2 core level electron binding
energy. This phenomenon has already been described in
Ref. (10) and ascribed to an epitaxial growth of the parti-
cles of catalysts A in strong interaction with the support,
developing (111) faces. Catalysts B, which have particles
which interact less with the support, exhibit the binding
(
measured by hydrogen chemisorption), then the catalytic
activity is given by the turnover frequency (TOF). For these
two values, the index “h” indicates the hydrogenation reac-
tion and the index “i” indicates the isomerization reaction.
The selectivity in but-1-ene toward but-2-enes is given by
but-1-ene
B1/B2 = (
trans-but-2-ene + cis-but-2-ene)
,
the selectivity in trans-but-2-ene toward cis-but-2-ene is energy of bulk palladium. As has been mentioned before
TABLE 1
Chemisorption, TEM, and XPS Characterization for Pd/�-Al2O3: Catalysts with Different
Metal Loadings Reduced at 573 and 773 K [Ref. (10)]
Solution
concentration
g Pd/liter)
Pd
content
(wt.% )
Reduction
temperature
(K)
Surface
Pd atoms
(H/Pd)irr
Particle size
(TEM)
BE
Pd 3d5/2
(eV)
(
Name
(nm)
0.70
0.70
3.33
3.33
0.09
0.09
0.30
0.30
A1
A2
B1
B2
573
773
573
773
0.54
0.25
0.26
0.23
2.8
4.5
6.0
7.2
336.1
335.9
335.1
335.2
a
a
Calculated from TEM results.

3
72
GOETZ, VOLPE, AND TOUROUDE
TABLE 2
Activity and Selectivity for the Buta-1,3-diene Hydrogenation and But-1-ene
Hydrogenation and Isomerization
Produced
butenes
Reduction
temperature
(K)
6
6
kh � 10
(mol/s/g cat.)
ki � 10
(mol/s/g cat.)
TOFh
TOFi
S
(% )
�
1
� 1
Catalysts
(s
)
(s
)
B1/B2
T/C
Buta-1,3-diene hydrogenation
Reaction temperature: 273 K
A1
B1
573
573
1.6
32.8
0.35
4.50
1.8
2.1
8.7
9.1
99.5
99.4
Reaction temperature: 255 K
A1
A2
573
773
0.95
0.22
0.21
0.10
1.8
1.2
8.7
6.4
99.5
99.5
But-1-ene hydrogenation and isomerization
Reaction temperature: 273 K
A1
B1
573
573
1.65
23.40
99.0
33.3
0.36
3.2
21.3
4.5
2.9
1.6
98.7
59.6
Note. h, hydrogenation; i, isomerization.
(
10), for these catalysts (A and B) the Pd 3d core level elec- results indicate that the large difference observed in hydro-
tron binding energy is not related with the particle size but genation activity between A1 and B1 is not related to the
with the concentration of Pd(C5H7O2)2 in the impregnation difference in particle size.
solution.
Let us introduce here the results obtained for the but-
Table 2 presents, for catalysts A and B, the activity and 1-ene reactions in which one discriminates between the hy-
selectivity for buta-1,3-diene hydrogenation and but-1-ene drogenation and the isomerization rates. Comparing the hy-
hydrogenation and isomerization. The reaction schemesare drogenation activity for the buta-1,3-diene and but-1-ene,
illustrated in Figs. 2 and 3, respectively.
one observes that, for a given catalyst, both rates are very
Let us first analyze the results obtained for the buta-1,3- similar. Therefore, the hydrogenation rate does not depend
diene hydrogenation. Comparing the results obtained for on the hydrocarbon but depends on the catalyst.
catalysts A1 and B1, we observe that an increase in Pd con-
For the isomerization reaction, it is worth noting that
tent leads to an increase in hydrogenation rate, which is the catalyst showing the lowest hydrogenation rate (A1)
about 13-fold higher if one compares the activity per sur- exhibits an isomerization activity, TOFi, about five times
face atom (TOFh). It is noticeable that catalyst A1 has a higher than the catalyst which showsthe highest hydrogena-
very low activity compared to catalyst B1. In order to test tion activity (B1). All these catalytic results suggest that the
the effect of particle size on the hydrogenation activity, let differences in physical characteristics, revealed mainly by
us consider the results obtained for catalyst A1, reduced at the Pd 3d electron binding energy, determine the catalytic
5
73 K and A2, reduced at 773 K, having respectively a mean behavior:
Pd particle diameter of 2.8 and 4.5 nm. At 255 K, both cata-
lysts present a very low hydrogenation activity per gram
of catalyst (kh) as well as per surface atom (TOFh). These
—
catalyst A1 (Pd 3d5/2 BE = 336.1 eV) has a very low hy-
drogenation rate, both for buta-1,3-diene and for but-1-ene,
but shows a high isomerization rate for but-1-ene reaction
which leads, in but-1-ene reaction, to a high selectivity to
but-2-enes toward butane (more than 98.5% );
—
catalyst B1 (Pd 3d5/2 BE = 335.1 eV) has hydrogena-
tion and isomerization rates in the same range, and its
FIG. 2. Buta-1,3-diene hydrogenation.
FIG. 3. But-1-ene hydrogenation and isomerization.

INFLUENCE OF PARTICLE MORPHOLOGY IN Pd/�-Al2O3 CATALYSTS
373
TABLE 3
catalytic reaction occurs between the hydrocarbon and
one hydrogen or deuterium atom, both adsorbed at the
catalytic surface.
Isotopic Distribution for the But-1-ene Deuteration
Let us analyze the results obtained for these two cata-
lysts at the same conversion level (ca. 50% ): the A1 cata-
lysts, the most selective (S = 97.7% ) and the most active for
Catalysts
A1
B1
�
1
Conversion (% ):
Selectivity (% ):
T/C:
67
98.5
2.8
54
51
26
76.1
2
the isomerization reaction (TOFi = 21.3 s ), produce the
highest ratio of nondeuterated isomers (more than 95% ),
while the B1 catalyst, which is less selective (71.8% ) and
97.7
2.7
71.8
1.8
Hydrocarbons
Percentages
� 1
less active for the isomerization reaction (TOFi = 4.5 s ),
Butane
C4H10
C4H9D
C4H8D2
C4H7D3
C4H6D4
produces only 65% of nondeuterated isomers. The latter
catalyst, which is the most active for the hydrogenation re-
56
37
4
2
1
56
27
14
2
35
40
25
37
39
24
� 1
action (TOFh = 3.2 s ), produces the highest percentage of
deuterated but-1-ene (exchange reaction) and the highest
ratio of deuterated isomers. It happens as if on A1 catalyst
the hydrogen from the gas phase would have difficulties
to react, then the hydrogenation, exchange, and isomeriza-
tion with deuterium introduction would be limited, while
the isomerization, which uses no hydrogen atom from the
gas phase, would be favored.
1
But-1-ene
C4H8
C4H7D
C4H6D2
99
1
<1
99
1
88
10
2
97
3
<1
trans-But-2-ene
C4H8
C4H7D
C4H6D2
C4H5D3
97
3
<1
<1
66
25
9
76
18
6
To sum up, we can say that the number of deuterium
atoms introduced in the reaction products is much lower for
the A1 catalyst than for the B1. However, even for this latter
catalyst, one introduces statistically less than one deuterium
molecule per butane molecule (6idi = 0.9; di, molecules
containing i deuterium atoms).
cis-But-2-ene
C4H8
C4H7D
C4H6D2
C4H5D3
C4H4D4
95
4
1
64
26
10
69
21
7
3
<1
Applied to but-1-ene, the well known Horiuti–Polanyi
(
HP) mechanism (15) describes together the addition reac-
tions, the exchange of but-1-ene (H to be replaced by D),
and the isomerization to deuterated products (see Fig. 4).
Analyzing the deuterium location in but-1-ene, it has
been shown (16–18) that, on palladium, the deuterium is
Note. Reaction temperature, 273K;reduction temperature,
73 K; deuterium partial pressure, 38 Torr; but-1-ene partial
5
pressure, 7.2 Torr.
selectivity to but-2-enes is quite poor (� 60% ) for the but-
1
-ene reactions.
In order to clarify the reaction mechanism, a deuterium
tracer study was performed on the two types of catalysts.
It has been performed under a deuterium/helium flow
(
D2–He/5–95% ). The comparison of the deuterium atom
distributions in the reaction products shows the relation-
ships between deuteration, isomerization and exchange re-
actions.
Table 3 indicates the distribution of the reaction products
for catalysts A1 and B1 at different conversions.
The butane distributions exhibit a high proportion of
nondeuterated molecules (35 to 56% ). These results indi-
cate that hydrogen atoms are introduced in the products
although the reaction is taking place under a deuterium/
helium flow. On the other hand, the rather low proportion
(
4 to 25% ) of di-deuterated molecules indicates that the
main process does not occur by a direct addition of the
deuterium molecule to the olefin. These results clearly
show that there is an isotopic dilution of the hydrogen
and deuterium atoms at the catalytic surface and that the
FIG. 4. Horiuti–Polanyi (HP) mechanism.

3
74
GOETZ, VOLPE, AND TOUROUDE
localized, for 90% , on carbon number 1. Hub (18) has de-
termined that this specific place was due to the preferential
formation of half-hydrogenated secondary butyl radicals;
therefore the pathway number 1, described on the Horiuti–
Polanyi mechanism (Fig. 4), is not favored on palladium.
On the other hand, to explain large amounts of d0 and
d1 butane molecules, one must suppose an isotopic dilu-
tion of hydrogen and deuterium atoms at the catalytic sur-
face when the addition reaction occurs (19). The difference
between the number of hydrogen atoms involved in the
addition process and the number of deuterium atoms sub-
stituted for the hydrogen atoms is actually positive. The
hydrogen atoms lost by the molecules are, for a large part,
involved in the hydrogenation process and are not balanced
by the deuterium from the gas phase. Therefore, when the
catalytic surface is covered with olefin, the equilibrium, (D2,
HD, H2)gaz *ꢀ (H, D)adsorbed, does not exist.
TABLE 4
But-1-ene Hydrogenation: Comparison of Experimental and
Calculated Distributions of Non-, Mono-, and Di-deuterated
Products in Butane Taking into Account an Isotopic Dilution
of Hydrogen and Deuterium Atoms at the Catalytic Surface
(
x = D/(H + D))
Catalysts
A1
B1
Calc. (% )a
x = 0.19
Calc. (% )a
x = 0.35
Isotopes
Exp. (% )
Exp. (% )
d0
d1
d2
56
27
14
56
27
3
37
39
24
37
39
10
a
The calculated data are normalized taking into account the butane
d0, d1.
If one considers an isotopic dilution of the deuterium and
hydrogen atoms on the catalytic surface and a mechanism
taking place by the addition of two adsorbed atoms, H or
D, the ratio of the different isotopes of butane can be calcu-
lated. If x is the percentage of deuterium atoms adsorbed at
the catalytic surface, (1 � x) is the percentage of hydrogen
atoms on the same surface. The probabilities to obtain the
different isotopes are given by
covered by 100% of deuterium atoms; for the B1 catalyst,
6% of the hydrogenation reaction occurs on sites contain-
ing D/H atom ratio equal to 0.54, and 14% occurs on sites
covered by 100% of deuterium atoms.
8
If the hydrogenation and isomerization reactions occur
on the same sites, by the Horiuti–Polanyi mechanism, one
can suppose that the isotopic dilution (in hydrogen and deu-
terium) is the same for both reactions and it is possible to
calculate the d0 and d1 isomerization ratio obtained by the
Horiuti–Polanyi mechanism.
As the but-1-ene isomerization products are, for more
than 90% , d0 and d1 molecules, one can neglect the partici-
pation of the polydeuterated molecules for the calculation
of the product distributions. In the Horiuti–Polanyi mecha-
nism, the formation of the half-hydrogenated radical occurs
2
C4H10(d0) = (1 � x) , C4H9D(d1) = 2x(1 � x),
2
C4H8D2(d2) = x .
Thus,
d0
d1
(1 � x)
2x
d1
=
or x =
.
(2d0 + d1)
Hence one obtains the calculated distribution in d0, d1, by addition of one deuterium or hydrogen atom, on carbon
and d2 indicated by d0calc., d1calc., and d2calc., given for a statis- number 1, losing one hydrogen atom on carbon number 3.
tical distribution of deuterium and hydrogen atoms at the This mechanism leads to d0 and d1 molecules with the same
catalytic surface determined by x.
The results given in Table 4 indicate that the proportion present at the catalytic surface, and thus, d0/d1 = D/H.
of d2 is underestimated by this calculation. This means that Table 5 gives a comparison between experimental and
distribution as the ratio of hydrogen and deuterium atoms
the estimated amount of deuterium at the catalytic surface, calculated data, these latter being obtained by taking into
which is in accordance with the d0/d1 ratio, is too low to account the isotopic dilution D/H at the catalytic surface.
explain the formation of the quantity of d2 butane actually One obtains experimentally much more d0 but-2-enes than
observed (d2exp. = 24% and d2calc. = 10% for the catalyst B1). is predicted by the statistical calculation based on the iso-
The formation of an excess of d2 butane can be explained topic dilution at the catalytic surface applied to Horiuti–
if one supposes a nonhomogeneous isotopic distribution at Polanyi mechanism. These results indicate that a part of the
the catalytic surface and, for several regions of this surface, isomers has been produced without any hydrogen or deu-
a deuterium concentration of 100% . On such sites, only terium atom insertion in the molecules, which means that
the deuteration reaction would occur, leading directly to an intramolecular hydrogen shift (IM) takes place in this re-
the di-deuterated butane. This process would lead to 14% action (Fig. 5). The part of this second mechanism is given
(
= 24 � 10) of the d2 production on the catalyst B1 and to by the difference between the but-2-enes values obtained
1% (= 14 � 3) on the catalyst A1. experimentally and by calculation. Such a mechanism has
To sum up, one can describe the nonhomogeneous iso- already been proposed by Smith and Swoap (20) to explain
topic dilution by the following scheme. For the A1 catalyst, the exchange reaction H, D in cyclohexene and has been
9% of the hydrogenation reaction occurs on sites contain- demonstrated in the case of propene (21) and butene (22)
ing D/H atom ratio equal to 0.23, and 11% occurs on sites reactions.
1
8

INFLUENCE OF PARTICLE MORPHOLOGY IN Pd/�-Al2O3 CATALYSTS
375
TABLE 5
But-1-ene Isomerization: Comparison of the Experimental Re-
sults and Calculated Data Taking into Account the Isotopic Di-
lution D/H
Catalyst
A1
B1
Calc.
Calc.
x = 0.19
x = 0.35
Isotopes
Exp.
D/H = 0.23
Exp.
81
H/D = 0.54
trans-but-2-ene
d0
97
3
13
HP = 16
3
35
HP = 54
d1
19
19
d0 exp. � d0 calc.
cis-but-2-ene
d0
H⇒
IM = 84
H⇒
IM = 46
96
17
HP = 21
4
69
40
HP = 61
d1
4
21
21
d0 exp. � d0 calc.
H⇒
IM = 78
H⇒
IM = 29
Note. HP, Horiuti–Polanyi; IM, intramolecular hydrogen shift.
Scheme I indicates the distributions obtained for the dif-
ferent mechanisms of hydrogenation and isomerization for
A1 and B1 catalysts. While A1 catalyst, in the total reaction,
leads to 80% of intramolecular isomerization mechanism,
SCHEME I
B1 catalyst leads only to 30% of this mechanism. On this proposed a classification of the surface metallic sites into
latter catalyst, the Horiuti–Polanyi mechanism, which leads corner, edge, or face sites, with respectively 3, 2, and 1 un-
3
2
1
to both hydrogenation and isomerization reactions, repre- saturated degrees (called M, M, and M). The hydrogena-
sents the main part with 70% . The part of deuterium atoms tion, which leads to the reaction between one adsorbed
3
which is inserted in the molecules, compared to the hydro- olefin and two hydrogen atoms, would take place one the M
gen atoms, on B1 catalyst, is twice the quantity observed sites, the isomerization resulting from the Horiuti–Polanyi
2
on the A1 catalyst, even though the D/H ratio remains less mechanism would take place on the M sites, and the iso-
than one.
merization resultingfrom the intramolecular hydrogen shift
1
The distribution between the two isomerization mech- would take place on the M sites. Smith et al. (25) applied
anisms, by intramolecular hydrogen shift or by Horiuti– this concept to explain the modification of the isomeriza-
Polanyi mechanism, is directly related to the relative hy- tion//hydrogenation ratio observed in the apopinene (+)
drocarbon and hydrogen coverage at the catalytic surface. If reaction on palladium catalysts. If we apply this concept to
the atomic hydrogen coverage is high, the Horiuti–Polanyi our catalysts, we could assume that the A1 and A2 catalysts
1
mechanism, in which a determining step is the formation present mainly M sites (face atoms), while the B1 catalyst
2
3
of the half-hydrogenated radical, will be enhanced. If the would develop a large number of M and M sites (edge and
atomic hydrogen coverage is low, the intramolecular hydro- corner atoms), which would be in good accordance with the
gen shift, which needs no other hydrogen atom than those characterization of the catalysts (10). The B catalysts con-
present in the molecule, will be predominant.
tain rather larger particles which generally develop fewer
In order to explain the competition between hydrogena- corner and edge atoms than the smallest ones. However, for
tion and isomerization reactions, several authors (23, 24) palladium particles, the situation might be different. The
FIG. 5. Intramolecular hydrogen shift mechanism (IM).

3
76
GOETZ, VOLPE, AND TOUROUDE
catalysts studied here undergo hydrogen treatments which
lead to successive hydride formation and destruction for the
palladium particles. For particles less interactive with the
support, these treatments would lead to cracks in the crys-
tal network (26, 27) and thus to a rugged surface containing
2
3
M and M sites. Such sites can adsorb the hydrocarbon and
the hydrogen atoms. That is why the hydrogenation process
�
1
is encouraged on this catalyst which gives high TOFh (4.5 s
�
1
for buta-1,3-diene and 3.2 s for but-1-ene). The low iso-
merization selectivity (� 60% ) in the but-1-ene reactions
and the higher contents of deuterium atoms in the isomers
are due to a higher quantity of hydrogen (deuterium) on the
active sites. On the other hand, the A catalysts contain par-
ticles in strong interaction with the support and thus might
develop particular morphology. In fact, several studies on
palladium single crystals indicate that, in the buta-1,3-diene
hydrogenation, the (111) face is less active than the (110)
face. The TOFh, given by the authors concerned (28–30)
and standardized to our experimental conditions are of the
same order of magnitude. The A particles might contain
FIG. 7. But-1-ene hydrogenation and isomerization and buta-1,3-
diene hydrogenation sites.
act only as spectators. The presence of spectator species in
hydrogenation reactions has often been suggested and the
structure has even been defined in ethyne, ethene, and bu-
tyne hydrogenation (31–33). It is reasonable to assume that
this hypothesis is also applicable in butadiene hydrogena-
tion; however, in this case the structure of the spectator
species has still to be defined. With this interpretation, we
are able to explain the similar rates obtained for the hy-
drogenation of buta-1,3-diene and but-1-ene, for the two
types of catalysts: for both hydrocarbons, the hydrogena-
tion reaction takes place on the corner or edge atoms, for
1
mainly (111) faces. These faces ( M sites) would be cov-
ered by hydrocarbons and would have few sites able to dis-
sociate hydrogen. These surfaces are very poor in M sites
3
which would be able to dissociate hydrogen in the presence
of hydrocarbons. On such hydrocarbon-covered catalysts,
the difficulties to dissociate hydrogen would explain the low
�
1
TOFh, both for the buta-1,3-diene (0.35 s at 273 K) and
�
1
for the but-1-ene (0.36 s at 273 K). However, it is notice-
able that the but-1-ene shows a high isomerization activity
21.3 s ), which indicates the presence of two different
�
1
(
types of sites, one for the hydrogenation and isomerization
by the Horiuti–Polanyi mechanism and the other for the
isomerization which occurs mainly by intramolecular hy-
drogen shift (Fig. 6). The sites situated on the (111) faces
would be able to promote the intramolecular hydrogen shift
leading to butene isomerization reaction. When buta-1,3-
diene is the reactant, these sites are no longer active for
hydrogenation but they might be covered by species which
1
the but-1-ene, the M sites are very active in the isomeriza-
tion reaction, while for the buta-1,3-diene they are covered
by spectator species (Fig. 7). Thus, the number of sites in-
volved in the hydrogenation reaction is the same whatever
the hydrocarbon.
CONCLUSION
A coherent explanation concerning the reaction mecha-
nism for the hydrogenation of buta-1,3-diene and the hy-
drogenation and isomerization of but-1-ene on Pd/�-Al2O3
catalystsemergesfrom the comparison of reaction rates and
using deuterium for a tracer study, as well as chemisorption,
TEM, and XPS data. Two types of catalyst, prepared from
benzene solutions of Pd acetylacetonate with two different
Pd concentrations, were used for this study. From the low
concentration solution (0.7 g Pd/l) the particles obtained
show a high BE for Pd 3d5/2, 336 eV, while from the high
concentration solution (3.33gPd/liters) the Pd particle have
3
35 eV as their Pd 3d5/2 BE, similar to bulk Pd. These results
were considered to originate not from a particle size effect
but from a difference in the nature of the interactions be-
tween the metal and the support. In the first catalyst a strong
interaction induces the formation of flat monocrystalline
FIG. 6. Hydrogenation and isomerization sites on A and B catalysts.

INFLUENCE OF PARTICLE MORPHOLOGY IN Pd/�-Al2O3 CATALYSTS
377
(
111) particles, while the second catalyst contains less inter-
6. Ryndin, Yu. A., Nosova, L. V., Boronin, A. I., and Chuvilin, A. L.,
Appl. Catal. 42, 131 (1988).
active particles (10). The catalytic activity results support
these interpretations. On interactive flat particles isomer-
ization reactions would be favored while the noninteractive
particles which might present rugged (rough) surfaces af-
ter hydrogen pretreatments would preferentially be active
for the hydrogenation reaction. The deuterium tracer study
shows that the isomerization reaction occurs by two differ-
ent reaction mechanisms, the Horiuti–Polanyi mechanism
with addition of adsorbed hydrogen (deuterium) atoms to
7
8
. Ryndin, Yu. A., Stenin, M. V., Boronin, A. I., Bukhtiyarov, V. I., and
Zaikovskii, V. I., Appl. Catal. 54, 277 (1989).
. Boitiaux, J. P., Cosyns, J., and Vasudevan, S., in “Proceedings, 3rd
Int. Symp. on the Scientific Bases for the Preparation of Het-
erogeneous Catalysts, Louvain-La-Neuve, 1982,” p. 123. Elsevier,
Amsterdam, 1983.
9. Aduriz, H. R., Bodnariuk, P., Coq, B., and Figueras, F., J. Catal. 129,
47 (1991).
10. Goetz, J., Volpe, M. A., Sica, A. M., Gigola, C. E., and Touroude, R.,
J. Catal. 153, 86 (1995).
form a half-hydrogenated (-deuterated) radical, and an in- 11. Benson, J. L., Hwang, H. S., and Boudart, M., J. Catal. 30, 146
(
1973).
tramolecular hydrogen shift which occurs without addition
of hydrogen or deuterium atom; the first mechanism takes
place on corner or edge atoms and the second on face atoms.
Thus, palladium particles in strong interaction with alu-
1
1
1
2. Noack, K., Zbinden, H., and Schl o¨ gl, R., Catal. Lett. 4, 145 (1990).
3. L e´ gar e´ , P., and Fritsch, A., Surf. Interface Anal. 15, 698 (1990).
4. Mintsa Eya, V., Touroude, R., and Gault, F. G., J. Catal. 66, 412
(
1980).
mina present high activity for isomerization of but-1-ene 15. Horiuti, J., and Nakamura, T. Z., Z. Phys. Chem. 11, 358 (1957).
1
1
6. Sim, K. S., Ph.D. Thesis, Strasbourg, France (1988).
7. Boutonnet, M., Kizling, J., Mintsa Eya, V., Choplin, A., Touroude, R.,
Maire, G., and Stenius, P., J. Catal. 103, 95 (1987).
8. Hub, S., Ph.D. Thesis, Strasbourg, France (1987).
9. Burwell, R. L., Adv. Catal. 9, 13 (1957).
and low hydrogenation activity for buta-1,3-diene and but-
-ene, with an identical value for both hydrocarbons. The
numerous (111) face atoms in this type of catalyst are fully
covered by tightly adsorbed hydrocarbon molecules and ac-
1
1
1
cordingly have no places left for hydrogen dissociation; on 20. Smith, G. V., and Swoap, J. R., J. Org. Chem. 31, 3904 (1966).
2
2
2
2
2
1. Naito, S., and Tinamoto, M., J. Catal. 102, 377 (1986).
2. Touroude, R., and Gault, F. G., J. Catal. 32, 279 (1974).
3. Ledoux, M. J., Nouv. J. Chim. 2, 9 (1978).
4. Siegel, S., Gutlaw, J., and Garti, N., J. Catal. 52, 102 (1978).
5. Smith, G. V., Molnar, A., Khan, M. M., Ostjard, D., and Yoshida, N.,
J. Catal. 98, 502 (1986).
these sites but-1-ene molecules are transformed into but-
-enes (IM isomerization) while butadiene molecules are
standingasspectator molecules. On the other hand the cata-
lyst with less interactive particles presents rough surfaces
and permits high hydrogenation and comparatively lower
isomerization rates.
2
2
2
6. Nakamura, K., J. Less-Common Metal 84, 173 (1982).
7. Lynch, J. F., Clewley, J. D., Curran, T., and Flanagan, B., J. Less-
Common Metal 55, 153 (1977).
REFERENCES
28. Bertolini, J. C., Delichere, P., Khanra, B. C., Massardier, J., Noupa, C.,
and Tardy, B., Catal. Lett. 6, 215 (1990).
1
. Boitiaux, J. P., Cosyns, J., and Vasudevan, S., Appl. Catal. 6, 41
1983).
. Gigola, C. E., Aduriz, H. R., and Bodnariuk, P., Appl. Catal. 27, 133
1986).
29. Tardy, B., Noupa, C., Leclercq, C., Bertolini, J. C., Hoareau, A.,
Treilleux, M., Faure, J. P., and Nihoul, G., J. Catal. 129, 1 (1991).
30. Noupa, C., Rousset, J. L., Tardy, B., and Bertolini, J. C., Catal. Lett. 22,
197 (1993).
(
2
(
3
4
. Hub, S., Hilaire, L., and Touroude, R., Appl. Catal. 36, 307 (1989).
. Nosova, L. V., Stenin, M. V., Nogin, Yu. N., and Ryndin, Yu. A., Appl.
Surf. Sci. 55, 43 (1992).
31. Webb, G., Catal. Today, 139 (1990).
32. Cremer, P. S., and Somorjai, G. A., J. Chem. Soc. Faraday Trans. 91,
3671 (1995).
5
. Deganello, G., Duca, D., Martorana, A., Fagherazzi, G., and Benedetti, 33. Maetz, Ph., Saussey, J., Lavalley, J. C., and Touroude, R., J. Catal. 147,
A., J. Catal. 150, 127 (1994). 48 (1994).