Theoretical and experimental investigations on site occupancy for palladium oxidation states in mesoporous Al-MCM-41 materials

Article abstract of DOI:10.1016/j.jcat.2012.02.014

The site occupancy of the various oxidation state of palladium supported on mesostructured porous aluminosilica of Al-MCM-41 type was studied combining density functional theory (DFT), FTIR, and Raman investigations. The study focuses on the usual +2 oxid

Full text of DOI:10.1016/j.jcat.2012.02.014

Journal of Catalysis 289 (2012) 227–237
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Theoretical and experimental investigations on site occupancy for palladium
oxidation states in mesoporous Al-MCM-41 materials
Anis Gannouni a, , Xavier Rozanska b,1, Belen Albela b, Mongia Saïd Zina a, Françoise Delbecq b,
,
⇑
⇑
Laurent Bonneviot b, Abdelhamid Ghorbel a
a Laboratoire de Chimie des Matériaux et Catalyse, Faculté des Sciences de Tunis, Campus Universitaire, Tunis 2092, Tunisia
b Université de Lyon, Institut de Chimie de Lyon, École Normale Supérieure de Lyon and CNRS, 46 Allée d’Italie, F-69046 Lyon, France
a r t i c l e i n f o
a b s t r a c t
Article history:
The site occupancy of the various oxidation state of palladium supported on mesostructured porous
aluminosilica of Al-MCM-41 type was studied combining density functional theory (DFT), FTIR, and Raman
investigations. The study focuses on the usual +2 oxidation state highly dispersed during palladium
incorporation via template-ion exchange (TIE) but also on other oxidation states. The simulations of the
IR spectra using silica clusters of different sizes confirm the Pd2+ preference for sites contained in 6- or
8-membered rings (band at 930 cmꢀ1). The calculations suggest that, after calcination, isolated oxidized
palladium species such as [PdO]2+, [Pd(OH)2]2+ for the +4 oxidation state, as well as [PdOH]2+ for the +3
EPR active oxidation state may coexist. The match between experimental and calculated vibrational
frequencies relies on the two intense Raman bands at 270 and 630 cmꢀ1. All these oxidized species are
preferentially located on 8-membered rings and linked to three framework oxygen atoms.
Ó 2012 Elsevier Inc. All rights reserved.
Received 1 November 2011
Revised 29 January 2012
Accepted 18 February 2012
Available online 2 April 2012
Keywords:
Palladium
Al-MCM-41
DFT
IR
Raman
1. Introduction
alumina or even on the poorly dispersing pure silica supports, iso-
lated species and unusual reduced or oxidized states can form like
Palladium supported on acid silica–alumina amorphous
mesoporous support, such as MCM-41 or SBA-15, is an excellent
heterogeneous catalyst for several industrial hydrogenation and
oxidation reactions [1,2]. Because of this, supported palladium
oxides have been widely studied by many researchers [3–11].
The initial deposition of Pd+2 species on amorphous mesoporous
supports leads after calcination to the formation of PdO particles
located inside the pores [3]. The distribution and dispersion of
these particles have been shown to be function of the Si/Al ratio
[4–6]. Indeed, substitution of Si with Al in a SiO2 framework leads
to a charge defect, which is compensated by a proton henceforth
forming a Brønsted acid site. The oxidized palladium species pref-
erentially anchors to these acid sites. The palladium dispersion
increases with the number of Al atoms in the framework walls of
the mesoporous support [4–6]. Contrary to zeolites that favor the
isolation of metal cations on specific sites and windows, mesopor-
ous materials may offer a larger variety of sites for isolated, small
clusters, and also bulky oxide structures. Nonetheless, a high
dispersion can be generated using a careful preparation method
like cation exchange or template ion exchange (TIE) [4]. Then, on
in zeolite [12,13]. Therefore, on larger surface area mesoporous
support, such as Al-MCM-41 site, isolation is expected. The present
work investigates on the existence of such isolated species on this
type of silica–alumina support since they are claimed to play an
important role in oxidation reactions catalyzed by palladium
supported on various siliceous supports.
The variety of interactions between the support and the metal
particles and the additional heterogeneity present on amorphous
systems justify the diversity of the theoretical approaches already
applied on more defined zeolitic supports. In particular, the inter-
action of a palladium ion with Brønsted acid sites of zeolites and
the catalytic reactivity have been studied by Harmsen et al. [7].
The species M2+, MO, and [M(OH)]2+ of several divalent metal cat-
ions (i.e., Co, Cu, Fe, Ni, Pd, Pt, Ru, Rh, and Zn) were investigated by
Rice et al. [8] in terms of coordination and trends on ZSM-5 support
models. Wang et al. [9] studied similar structural and electronic
properties than Rice et al., although focusing preferentially on pal-
ladium. Grybos et al. [10,11] were interested to calculate the inter-
action between palladium Pd2+ ions on Mordenite models. Their
models were described under periodic conditions, whereas the
previously mentioned studies used gas phase cluster models. Fur-
thermore, the combination of both theoretical and experimental
studies to identify metal sites from their vibrational characteristics
is a very useful approach for catalytic materials that eventually ex-
hibit a complex mixture of phases or sites [14,15].
⇑
Corresponding authors.
E-mail addresses: anis.gannouni@gmail.com (A. Gannouni), francoise.delbecq@
ens-lyon.fr (F. Delbecq).
1
Present address: Materials Design, 18 rue de Saisset, 92120 Montrouge, France.
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.02.014

228
A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
In the present work, such a combined theoretical and experi-
recovered by filtration, washed thoroughly with 150 ml of
deionized water, and dried in vacuum at 313 K for 24 h to give
the as-synthesized sample Pd/Al-MCM-41. Elemental analyses:
molar ratio Si/Al, 10.33 and Pd, 1.51 wt.%.
mental investigation is proposed to characterize isolated oxidized
species of palladium supported on Al-exchanged MCM-41 with a
Si/Al ratio of 10.3, which are of particular interest in catalytic
methane combustion [2,5]. A detailed vibrational study by IR and
Raman spectroscopy is presented to get insight into the Pd species
structures, stability, and vibrational characteristics. Calculations
were performed using the density functional theory (DFT). The
support modeling was based on a local approach using clusters
of various sizes to take into account the different possibilities of
alumina–silica rings that occur on an amorphous surface as ex-
pected for MCM-41 support. The different positions of Si-substitu-
tions by Al atoms have also been considered. The reaction energies
of formation of Pd2+ species and also those of the corresponding
oxidized species have been calculated, and their IR and Raman
spectra have been simulated and compared with experimental
data, to identify their nature and their preferred location. Indeed,
on an amorphous silica–alumina support like MCM-41 that pre-
sents a large surface area favorable for high metal dispersion, a
lot of different isolated species can be formed [16,17].
2.1.3. Synthesis of PdO/Al-MCM-41
Pd/Al-MCM-41 sample was calcined at 823 K for 6 h in oxygen
flow (30 ml/min) to give PdO/Al-MCM-41. Elemental analyses:
molar ratio Si/Al, 10.33 and Pd, 1.99 wt.%.
2.2. Sample characterization
The IR spectra were recorded on samples diluted in KBr pressed
into pellets in the 400–4000 cmꢀ1 region using a Perkin–Elmer
FTIR paragon 1000 PC spectrometer. Raman spectroscopy was per-
formed with a Horiba Jobin–Yvon LabRam HR800 apparatus. Light
source was a mixed argon–krypton–ion laser to achieve a 514.5-
nm wavelength. An Olympus BX30 open microscope equipped
with ꢂ50 and ꢂ100 Olympus objectives was used coupled to spec-
trometer to focus the laser beam into to 1 lm spot diameter. The
Raman signal was collected in the backscattered direction. Acquisi-
tion timespan was about 300 s, with a laser output power on sam-
ple adjusted to 6 mW. The spectral resolution was of 1 cmꢀ1 using
1800 gr/mm gratting. XRD patterns were recorded on an automatic
2. Experimental section
2.1. Catalyst preparation
diffractometer (Philips Panalytical) using the CuK radiation
a
0
(k = 1.54 ÅA) and a nickel monochromator. Solid-state 27Al-MAS-
Cetyltrimethylammonium p-toluenesulfonate, (CTATos; >99%
Merck), Ludox HS-40 (40% SiO2, Aldrich), sodium hydroxide (Acros),
ammonium nitrate (95%, Alfa Aesar), aluminum sulfate–octadeca-
hydrate (98%, Aldrich), tetraammine palladium (II) chloride
(99.9%, Alfa Aesar), and ethanol (96%, Elvetec) were used as
received. The sodium silicate solution was prepared as follows:
Ludox HS-40 (187 ml) was added to sodium hydroxide (32 g) in
deionized water (800 ml) and stirred at 313 K until clear.
NMR spectra were recorded at a frequency of 104.22 MHz, spin-
ning rate of 8 kHz, pulse length of 1.0 ls, and delay time of 0.2 s.
The total number of scans was 150, and the line broadening was
50 Hz. Chemical shifts were reported in ppm and referenced to alu-
minum nitrate. The ESR spectra were recorded on the powder sam-
ples introduced in 4-mm quartz tube using a Bruker Elexsys e500
X-band (9.4 GHz) spectrometer with a standard cavity operated
with a liquid nitrogen cryostat regulated at 118 K.
2.1.1. Synthesis of Al-MCM-41
3. Theoretical approach
Al-MCM-41 catalyst was prepared according to method de-
scribed by some of us [18]. A sodium silicate solution (80 ml)
was stirred at 333 K for 1 h. A second solution of hexadecyltrime-
thylammonium p-toluenesulfonate (CTATos, 3.2 g) in deionized
water (135 ml) was stirred for 30 min at 333 K, and Al2(SO4)3ꢁ
18H2O (3.5 g) was then added to the template solution under
stirring for 1 h at 333 K. The first solution was added dropwise to
the second one and then stirred at 333 K during 2 h. The resulting
gel was heated in an autoclave at 403 K during 20 h. After filtration
and washing with deionized water (approximately 150 ml), the as-
synthesized solid was dried at 353 K. The Si/Al ratio in the synthe-
sis gel was equal to 12.5. Note that this ratio is about ten times
smaller than the previous study using the same TIE techniques
(12.5 instead of 150) [4]. Therefore, a higher metal dispersion is ex-
pected in the present study. A partial removal of the surfactant
molecules (CTA+) present in the as-made material was carried
out using an alcoholic solution of ammonium nitrate [19]. The mix-
ture was stirred at 353 K for 45 min. The amount of NH4NO3 corre-
sponded to a NHþ4 =CTAþ molar ratio of 0.71. The above treatment
was repeated twice. The final solid was recovered by filtration,
washed with ethanol, and dried at 353 K. The final solid was de-
noted as Al-MCM-41 and presents a partial removal of the surfac-
tant before template ion exchange in order to facilitate the
diffusion of the palladium complex inside the channel during the
deposition of the metal and to optimize the dispersion.
3.1. Theoretical models
The geometry and energy of the systems obtained by incorpo-
rating Pd2+ and its oxidized form into amorphous silica are inves-
tigated using the density functional theory (DFT).
Several patterns of acid silica–alumina (Hn+2Al2Sinꢀ2O3n/2) have
been considered to model the various topological forms that can be
met on an amorphous surface. The models are built using struc-
tures of silsesquioxane type that simulates the 4-, 5-, 6-, and 8-
membered rings in which two Si atoms are substituted by two Al
atoms. This requires the introduction of two protons for charge
neutralization. The termination of these clusters of atoms is
achieved preferentially by HASi„ groups instead of HAOASi„
groups to avoid additional intramolecular hydrogen bonds and
unrealistic physico-chemical properties [22–24]. The equation of
Pd2+ deposition is written as a cation exchange reaction of palla-
dium for proton as follows:
nþ2Al2Sinꢀ2O3n=2 þ PdCl2ðgÞ ¼ ½Pd2þ ꢁ HnAl2Si
þ 2HClðgÞ
O
3n=2ꢃ
2ꢀ
H
nꢀ2
ð1Þ
The oxidized form of Pd2+ deposit is obtained as follows:
½Pd2þ ꢁ HnAl2Si
O
ꢃ þ 1=2O2ðgÞ ¼ ½PdO2þ ꢁ HnAl2Si
O
nꢀ2
3n=2ꢃ ð2Þ
2ꢀ
2ꢀ
nꢀ2
3n=2
2.1.2. Synthesis of Pd/Al-MCM-41
In this study, the acid silica–alumina aggregates (H2MS) are desig-
2þ
This solid was then added into an aqueous solution of tetraam-
mine palladium (II) chloride (2 wt.% Pd) [4,20,21]. The mixture was
vigorously stirred at 353 K for 1 h. The obtained solid was
nated by ½Hꢃ2 ½XðY ꢀ MRÞꢃ2ꢀ, where X is the ring size of TO on which
the palladium atom is deposited, and Y ꢀ MR is the size of the ring
subjacent to the X-membered ring (Table 1). Both X and Y ring sizes

A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
229
Table 1
Clusters nomenclature of silica–alumina acid.
H2+nAl2Sinꢀ2O3n/2
Si/Al
H6Al2Si4O9
2
H8Al2Si6O12
3
H10Al2Si8O15
4
H12Al2Si10O18
5
H12Al2Si10O18
5
H16Al2Si14O24
7
2þ
2ꢀ
½Hꢃ2 ꢁ ½4ð8 ꢀ MRÞꢃ
2þ
2ꢀ
2ꢀ
2þ
2ꢀ
½Hꢃ2 ꢁ ½4ð6 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃa
2þ
2ꢀ
2þ
2ꢀ
2þ
2ꢀ
2ꢀ
2þ
2ꢀ
2ꢀ
2þ
2þ
2ꢀ
½Hꢃ2 ꢁ ½4ð3 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½4ð4 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½4ð5 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½4ð4:5 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½6ð6 ꢀ MRÞꢃa
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃb
2þ
2þ
2þ
2ꢀ
2þ
2ꢀ
½Hꢃ2 ꢁ ½5ð5 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½5ð4:5 ꢀ MRÞꢃ
½Hꢃ2 ꢁ ½6ð6 ꢀ MRÞꢃb
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃc
vary from 4 to 8 TO (excluding seven-membered rings). Fig. 1 shows
the structures of the clusters and the various sites of the aluminum
atoms. In the adopted models, all the isomers of position have been
considered. The protons are bound to oxygen atoms in bridge posi-
tion between Al and Si atoms with HAO distances of about 97 pm.
For the sake of simplification, H terminating atoms were not shown
on the figures.
the pseudopotential of Hay and Wadt has been used together with
the LANL2 basis sets as implemented in Gaussian 03 code. The
optimized structures were characterized by frequency calculations,
and all energies include the zero-point vibrational energies unless
otherwise stated. The frequencies were calculated at three levels,
PBE, B3LYP//PBE, and B3LYP. The two latter sets of values do not
differ much and are in good agreement with the experimental val-
ues, whereas the former is farther (see Tables S7, S8 and Fig. S10).
B3LYP is indeed known to give reliable frequencies [32]. The Natu-
ral Bond Orbital (NBO) charges were calculated [33]. The spin-elec-
tronic configurations singlet and triplet were considered for the
palladium deposits. The most stable states are singlets for H2MS
and PdAMS and triplets for PdOAMS. For the calculations of the
vibration modes, an isotopic mass change for terminal hydrogen
was carried out to decouple unrealistic vibrations. A weight of
1000 was assigned to the hydrogen atoms.
3.2. Methods
All calculations were done within the Density Functional Theory
(DFT) using the PBE [25] and B3LYP [26] functionals with Turbo-
mole 5.7 code [27–29] and Gaussian 03 [30]. The triple-f plus
polarization basis set (TZVP) [31] was used for the following
atoms: silicon, oxygen, aluminum, and hydrogen. For palladium,
4. Results
4.1. Preparation and characterization of the catalyst
The 2D hexagonal mesostructured aluminosilicate of MCM-41
type used as a support was prepared from a gel with a molar Si/Al
ratio of 12.5 leading to Si/Al ratio of 10.33 in the final materials.
The palladium was incorporated in the support from of a Pd(NH3)4
Cl2 aqueous solution using the so called template ion exchange
method to produce a high palladium dispersion through the chan-
nels of the support [4,20,21]. The Pd content was 1.51 and 1.99 wt.%
in the as-synthesized Pd/Al-MCM-41 and calcined PdO/Al-MCM-41
materials, respectively.
The as-synthesized Pd/Al-MCM-41 is a white material that
turns into light orange after calcination at 823 K. According to pre-
vious works, this color could be due to Pd4+ ions in an isolated
PdO2+ species [34], a PdO2 phase [35] or a [HOAPdAOAPdAOH]
cluster [36]. The hexagonal structure of the support is maintained
in all the materials as shown in the XRD patterns (Fig. S1.A). In
addition, a peak observed around 2h = 34° (Fig. S1.B) evidences also
the formation of PdO particles in PdO/Al-MCM-41. The 27Al MAS-
NMR spectra of Al-MCM-41and Pd/Al-MCM-41 materials (Fig. S2)
show a single peak at 52 ppm attributed to tetrahedral aluminum
(Td-Al) which is in agreement with a four-coordinated framework
Al3+. After calcination, an additional peak at 0 ppm is also present
in PdO/Al-MCM-41 sample accounting for octahedral aluminum
species (OhAAl). The latter is much narrower, which indicates
the presence of less octahedral species than tetrahedral ones. Note
that no OhAAl sites are observed in absence of palladium (Fig. S2),
suggesting that these marginal sites are generated during the pal-
ladium incorporation. Hence, the TdAAl sites remain predominant
after calcination even in the presence of palladium.
Raman and IR bands of the silica mesoporous supports are well
documented [37–43]. In general, the
m
m
asym(SiOSi) bands range
sym(SiOSi) vibrations are
between 1000 and 1300 cmꢀ1 and the
observed between 800 and 850 cmꢀ1. The bands around 550–
Fig. 1. PBE/TZVP-optimized structures of the silica–alumina clusters (H2MS) with
aluminum and silicon atoms only. (Oxygen atoms omitted and positions of
hydrogen atoms described in the text.)
600 cmꢀ1 and 270–425 cmꢀ1 are assigned to
d(SiOSi), respectively. The characteristic bands of MCM-41 are
m(SiO) + d(OSiO) and

230
A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
observed at 791, 825, 1051 (intense), 1040, 1140, 1219 (shoulder)
cmꢀ1 [37,43]. On the other hand, peaks at 1075 and 1182 cmꢀ1 are
assigned to the four TO cycles (T = Si, Al) and peaks at 1033 and
1245 cmꢀ1 to six TO cycles [39]. It has been shown that the
interaction of Pd+2 ions with the support induces an absorption
band at 928 cmꢀ1, attributed to the stretching vibrations of the
TAO bonds [41]. These bonds are elongated by the presence of
Pd, which shifts their vibration frequencies otherwise observed
between 1000 and 1200 cmꢀ1 [44,45]. After calcination, the
[46]. The band at 950 cmꢀ1 corresponds to silanols band (SiAOH)
[2,42,43]. After palladium exchange, the spectrum exhibits subtle
changes such as a small bump at 930 cmꢀ1 corresponding to the
masym(SiOSi) perturbed by Pd, as previously proposed [41]. This
new feature is better evidenced on the spectrum obtained by differ-
ence of spectrum b and spectrum a (Fig. 2c). It further relates the
disappearance of vibrational resonances at ca. 1009 cmꢀ1. The Ra-
man spectrum of samples calcined at 823 K exhibits two bands at
274 and 630 cmꢀ1 (Fig. 2d). The latter is ascribable to the Raman ac-
tive B1g vibrational mode of the PdO phase that is usually found at
651 cmꢀ1 for single crystals or PdO foils [47] or between 626 and
640 cmꢀ1 for oxidized Pd deposited on alumina or zirconia
[48–50]. The broad bands at 608 and 410 cmꢀ1 are attributed to
the three- and four-Si siloxane rings, respectively [38], while the
bands at 395, 330, and 780 cmꢀ1 are assigned to siloxane linkages
(SiAOASi) and surface silanol groups of MCM-41 [39,40]. The band
at about 545 cmꢀ1 can be assigned to a motion of the bridging oxy-
gen atoms perpendicular toward the bisecting line connecting the
two T atoms of a TAOAT bridge [39].
masym(SiOSi) band is slightly displaced owing to the formation of
new SiAOASi and SiAOAAl bridges [42].
The experimental IR spectrum of Pd/Al-MCM-41 is presented in
Fig. 2b, together with that of the as-synthesized support Al-MCM-41
(Fig. 2a). The main peaks described in the literature for MCM-41 are
found in our samples, particularly the peak at 791 cmꢀ1 and those at
1055 and 1219 cmꢀ1 attributed to four TO cycles or to six TO cycles.
The band at 904 cmꢀ1 can be ascribed to the stretching vibration of
SiAOꢀ interacting with CTA+ (cetyltrimethylammonium template)
In the following, the different possibilities for Pd2+ site locali-
zation in the acid alumina–silica model clusters depicted in
Fig. 1 are analyzed in terms of structure and energy of formation.
The same procedure is also applied to the oxidized species in
PdO/Al-MCM-41.
4.2. Formation of [Pd2þ ꢁ HnAl2Sinꢀ2O2ꢀ
]
3n=2
The energies of formation of [Pd2þ ꢁ HnAl2Si
O
3n=2] are calcu-
2ꢀ
nꢀ2
lated using Eq. (1), and the Gibbs free energies of reaction are ob-
tained from:
DrG ¼ DrEele
þ ZPE ꢀ RTln Q; with
D
Q ¼ qtrans qrot qvib and q the partition functions:
The Gibbs free energies of reaction at T = 300, 500, and 700 K are
given in Table 2. The values change considerably depending on the
structure onto which Pd2+ adsorbs. For instance, they vary from
ꢀ135 to 92 kJ molꢀ1 at 500 K. Adsorption of Pd2+ into a four-mem-
bered ring is always unfavorable since DrG is positive whatever the
considered temperature. In these complexes, Pd2+ is located outside
the ring formed by the four oxygen atoms (R angles are between
326° and 347°), and the distances between Pd and the oxygen atoms
are in the range 204–212 pm (Table S1). The formation of all the
other PdAMS complexes is favorable with the exception of
2ꢀ
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa that shows slightly positive or negative Gibbs
free energies of formation as a function of temperature. The
distances between oxygen and palladium are slightly longer than
before, that is, between 205 and 211 pm, but the Pd atom is now
Table 2
Formation electronic energy DrEelec (kJ molꢀ1) (Eq. (1)) and formation Gibbs free
energies DrGT (kJ molꢀ1
temperature T, 300, 500, and 700 K at the PBE/TZVP level.
)
of the clusters [Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ as
a
function of
[Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ
DrEelec
DrG300
DrG500
DrG700
[Pd]2+ꢁ[4(3 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(4 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(5 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(4.5 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(6 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(8 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[5(5 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[5(4.5 ꢀ MR)]2ꢀ
127
106
106
123
171
161
ꢀ5
73
58
57
58
96
110
ꢀ54
ꢀ38
ꢀ51
ꢀ123
5
51
41
39
28
58
29
23
21
ꢀ3
19
92
73
ꢀ73
ꢀ62
ꢀ81
ꢀ135
ꢀ1
ꢀ74
ꢀ35
ꢀ93
ꢀ86
ꢀ111
ꢀ147
ꢀ7
ꢀ87
ꢀ52
16
2ꢀ
½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa
12
2ꢀ
½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃb
ꢀ87
34
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa
2ꢀ
2ꢀ
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb
ꢀ23
27
ꢀ62
ꢀ18
Fig. 2. FTIR spectra of (a) as-synthesized Al-MCM-41; (b) as-synthesized Pd/Al-
MCM-41, (c) difference spectrum (b–a) and (d) Raman spectrum of PdO/Al-MCM-
41.
2ꢀ
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃc

A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
231
coplanar with the four oxygen atoms that belong to the 5-, 6-, or 8-
membered ring into which Pd is adsorbed: the angles are always
R
near 360° (Table S1). It is well known that diamagnetic ML4 com-
plexes are square planar for d8 transition metal [51,52], which is
the case of Pd2+. This is consistent with the present simulation show-
ing that the reaction energy of formation for Pd2+ꢁMS2ꢀ is more
favorable for the 5- and 8-membered rings and, even more for the
six-membered ring ([Pd]2+ꢁ[6(6 ꢀ MR)]2ꢀ
)
where the planar
arrangement is obtained with less ring distortion (Fig. 3). This con-
clusion is in agreement with previous experimental data concerning
zeolite supports [41].
The Gibbs free energies of the Pd2+ exchange reaction follow the
order 6TO > 5TO > 8TO > 4TO as function of the aluminosilica ring
size. They become more favorable at high temperature because
two gas phase molecules (HCl) are formed while one molecule
(PdCl2) is consumed (Table 2). Note that the temperature depen-
dence of the Gibbs free energy of formation varies differently from
2ꢀ
one system to another (Table 2). ½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃb is indeed the
most stable complex, however the second most stable one is either
2ꢀ
2ꢀ
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb or ð½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa depending on the
temperature. This stems partly from the entropies of formation
that are different depending on the ring into which Pd2+ is inserted
(Table S2). It results that the distribution of Pd2+ may likely depend
on the temperature of deposition but also on the temperature of
observation.
In agreement with the theoretical work of Toulhoat and co-
workers [10] and the experimental results of Bell and co-workers
[53], the energies of formation of Pd2+ꢁMS2ꢀ are found more
favorable for a larger number of basic oxygen atoms interacting
2ꢀ
with Pd2+ (compare for example ½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa with b and
2ꢀ
½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃc with a or b). The Mullliken and Natural Bond
Orbital (NBO) charges of both Pd2+ꢁMS2ꢀ and MS2ꢀ clusters (Table
S3) were computed to better illustrate this case. We indeed
observe a rough relation between the variations of the charges, that
is, the charge transfers and the formation energies of the Pd2+ꢁMS2ꢀ
complexes (Figs. S3 and S4). However, this type of interactions can-
not account alone for the reaction energies. In complement, the
deformation energy of the palladium complexes (PBE/TZVP level)
were also investigated according to the following formula:
2ꢀ
Edef ¼ E½XðY ꢀ MRÞꢃ2ꢀ=½Pdꢃ2þ ꢁ ½XðY ꢀ MRÞꢃ2ꢀ ꢀ E½XðY ꢀ MRÞꢃ
;
ð3Þ
where E[X(Y ꢀ MR)]2ꢀ/[Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ is the electronic energy
of [X(Y ꢀ MR)]2ꢀ in the optimized geometry of [Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ
and E[X(Y ꢀ MR)]2ꢀ is the optimized electronic energy of
[X(Y ꢀ MR)]2ꢀ in gas phase. The results are provided in Table 3.
Both ring sizes on which Pd2+ is adsorbed and silsesquioxane
structure impact on the deformation energy. Effectively, the calcu-
lations demonstrate that the deformation of the silica cluster of
[X(Y ꢀ MR)] type with X = 4 or 5 increases in the order Y = 4, 5, 6
and 8. Except [Pd]2+ꢁ[4(3 ꢀ MR)]2ꢀ, 4- or 8-membered rings exhibit
the largest deformation energies. As already noted above, Pd ions
cannot lie in the plan of too small 4-membered rings, which forces
distortions on the metal center. Conversely, too large 8-membered
rings undergo a costly deformation to fit with four PdO bond for-
mation. It appears that the ring deformation contributes for a large
part to the stability of Pd2+ in various rings (Table 3).
Fig. 3. PBE/TZVP-optimized structures for the configurations of [Pd]2+ in different
sites. Bond lengths PdAOMS and sum of angles in the pseudo PdO4 plan
P
(\OMSPdOMS
)
provided in parenthesis, in pm and degree, respectively, pictures
i
restricted to TO rings containing Pd2+ ion.
4.3. Formation of ½PdO2þ ꢁ HnAl2Sinꢀ2O23nꢀ=2ꢃ
All the structures of ½PdO2þ ꢁ HnAl2Si
O
3n=2ꢃ systems investi-
reaction energy as a function of the ring size is exactly opposite
to that of the Pd2+ꢁMS2ꢀ energy of formation (Tables 2 and 4).
The less stable [Pd]2+ꢁ[4(8 ꢀ MR)]2ꢀ complex is the easiest to
oxidize leading to the most stable site for the oxidized species.
Like Pd2+, PdO2+ is preferentially coordinated to oxygen atoms
2ꢀ
nꢀ2
gated by DFT are depicted in Fig. 4 and the corresponding oxidation
reaction energies (Eq. (2)) are provided in Table 4. The oxidation is
found endothermic in all cases though easier on 4-membered and
some of 8-membered rings. Indeed, the variation of the oxidation

232
A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
Table 3
Deformation energy Edef (kJ molꢀ1) of clusters [Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ at the PBE/TZVP
Level.
[Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ
Edef
[Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ
[Pd]2+ꢁ[5(4.5 ꢀ MR)]2-
Edef
[Pd]2+ꢁ[4(3 ꢀ MR)]2ꢀ
194
231
228
207
[Pd]2+ꢁ[4(4 ꢀ MR)]2ꢀ
2þ
2ꢀ
½Pdꢃ ꢁ ½6ð6 ꢀ MRÞꢃa
2þ
2ꢀ
2þ
2ꢀ
252
269
266
304
215
205
259
266
257
½Pdꢃ ꢁ ½4ð5 ꢀ MRÞꢃb
½Pdꢃ ꢁ ½6ð6 ꢀ MRÞꢃb
[Pd]2+ꢁ[4(4.5 ꢀ MR)]2ꢀ
2þ
2ꢀ
½Pdꢃ ꢁ ½8ð8 ꢀ MRÞꢃa
[Pd]2+ꢁ[4(6 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[4(8 ꢀ MR)]2ꢀ
[Pd]2+ꢁ[5(5 ꢀ MR)]2ꢀ
2þ
2ꢀ
½Pdꢃ ꢁ ½8ð8 ꢀ MRÞꢃb
2þ
2ꢀ
½Pdꢃ ꢁ ½8ð8 ꢀ MRÞꢃc
neighboring one Al3+ ion, as shown by the comparison of positional
isomers of [PdO]2+ꢁ[8(8 ꢀ MR)]2ꢀ. This is in agreement with the
theoretical studies of Grybos et al. [11] and also with EXAFS and
XANES experimental studies, which showed that the dispersion
of PdO depends on the number of acid sites, hence on the Si/Al ratio
[54,55].
Palladium is connected to four oxygen atoms, three of them
belonging to the support framework and one forming an oxo spe-
2ꢀ
cies, except in the case of ½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃb where five PdAO
bonds are formed. The distances between the palladium ion and
the framework oxygen atoms vary between 206 and 259 pm (Table
S4), in agreement with EXAFS experiments [4]. The PdAO length of
the [PdO]2+ species (oxo) stabilizes between 178 and 181 pm, ex-
2ꢀ
cept for the ½PdOꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃb where the elongation up to
185 pm compensates the formation of four PdAO bonds with the
support. Except ½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃ2bꢀ, all these complexes are of
d6 ML4 type with a triplet electronic structure. According to the
relative orbital position of this type of fragments, they also prefer
the square-planar coordination geometry [52]. Considering now
the global energy of formation at 300 K of oxidized species starting
from the corresponding H2MS cluster (sum of Eqs. (1) and (2))
(Table 5), the most stable species is ½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb2ꢀ followed
2ꢀ
2ꢀ
by ½PdOꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa or ½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa depending on
the temperature (Fig. S5). This shows that the dispersion of
[PdO]2+ depends on the oxidation temperature. This is in agreement
with the experimental data on Pd oxo species [41,56–58]. For
instance, Lamb and co-workers [56] showed that square-planar
Pd2+ oxo species are formed on the 8-membered rings of Mordenite
after calcination at 623 K under O2.
To summarize, the most energetically favorable adsorption sites
of Pd2+ and [PdO]2+ on the aluminosilica supports are the 6- and 8-
membered rings, respectively.
4.4. Comparison between theoretical and experimental vibration
spectra
In this part, the Raman and IR spectra of Pd2+ꢁMS2ꢀ and
[PdO]2+ꢁMS2ꢀ in all the adsorption sites are simulated and com-
pared with the experimental data. Experimental Raman bands for
3- and 4-membered rings are observed at ca. 600 and 485 cmꢀ1
,
respectively [38,59]. However a 560 cmꢀ1 band was also assigned
to 4-membered rings [60].
Our simulations of Raman spectra at B3LYP level for the all silica
HnSinO3n/2 clusters show an intense band well differentiated for
each cluster. These bands are at 584, 468, 445, 446 and 456 cmꢀ1
for 3-, 4-, 5-, 6-, and 8-membered silica rings, respectively
(Fig. S6). These results agree with the experimental data
[26,58,59], in particular, they confirm the assignment of the bands
at 600 and 485 cmꢀ1 to the 3- and 4-membered rings, respectively.
Raman spectra of HnSinO3n/2 and Hn+2Al2Sinꢀ2O3n/2 exhibit similar
bands with differences smaller than 51, 38, 30, 24 and 15 cmꢀ1
for 3-, 4-, 5-, 6-, and 8-membered rings, respectively. The good
agreement between the simulated and experimental Raman
Fig. 4. PBE/TZVP-optimized structures for [PdO]2+ configurations in different sites.
PdAOMS bond lengths reported in pm. Pictures restricted to rings containing [PdO]2+
species.
spectra supports the validity of our silica and aluminosilica models
and the conclusions are consistent with previous works [61,62].

A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
233
Table 4
Formation electronic energy DrEelec (kJ molꢀ1) and formation Gibbs free energies DrGT
(kJ molꢀ1) of the clusters [PdO]2+ꢁ[X(Y ꢀ MR)]2ꢀ from [Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ (Eq. (2)), as
a function of temperature T, 300, 500, and 700 K at the PBE/TZVP level.
[PdO]2+ꢁ[X(Y ꢀ MR)]2ꢀ
DrEelec
DrG300
DrG500
DrG700
[PdO]2+ꢁ[4(3 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[4(4 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[4(5 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[4(4.5 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[4(6 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[4(8 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[5(5 ꢀ MR)]2ꢀ
[PdO]2+ꢁ[5(4.5 ꢀ MR)]2ꢀ
28
42
33
35
49
49
64
57
62
70
62
79
72
79
83
74
93
86
95
95
23
33
39
44
126
109
120
145
129
137
157
142
151
169
154
165
2þ
2ꢀ
½PdOꢃ ꢁ ½6ð6 ꢀ MRÞꢃa
2þ
2ꢀ
212
42
224
57
233
66
242
75
½PdOꢃ ꢁ ½6ð6 ꢀ MRÞꢃb
2þ
2ꢀ
½PdOꢃ ꢁ ½8ð8 ꢀ MRÞꢃa
2þ
2ꢀ
77
94
106
143
116
146
½PdOꢃ ꢁ ½8ð8 ꢀ MRÞꢃb
2þ
2ꢀ
135
139
½PdOꢃ ꢁ ½8ð8 ꢀ MRÞꢃc
Table 5
Global formation electronic energy DrEelec (kJ molꢀ1) and global formation Gibbs free
2ꢀ
energies DrGT (kJ molꢀ1) of clusters [PdO]2+ꢁ[X(Y ꢀ MR)]2ꢀ from ½Hꢃ22þ ꢁ ½XðY ꢀ MRÞꢃ
(sum of Eqs. (1) and (2)), as a function of temperature T, 300, 500, and 700 K at the
PBE/TZVP level.
2þ
2ꢀ
DrEelec
DrG300
DrG500
DrG700
½Hꢃ2 ꢁ ½XðY ꢀ MRÞꢃ
2þ
2ꢀ
2ꢀ
2ꢀ
155
148
139
158
220
184
121
125
132
125
76
122
122
114
120
166
143
91
113
120
111
107
141
131
84
103
116
107
92
½Hꢃ2 ꢁ ½4ð3 ꢀ MRÞꢃ
2þ
½Hꢃ2 ꢁ ½4ð4 ꢀ MRÞꢃ
2þ
½Hꢃ2 ꢁ ½4ð5 ꢀ MRÞꢃ
2þ
2ꢀ
½Hꢃ2 ꢁ ½4ð4:5 ꢀ MRÞꢃ
2þ
2ꢀ
Fig. 5. The simulated (B3LYP//PBE) (a) and experimental (b) IR spectra: (a) B3LYP IR
114
117
76
½Hꢃ2 ꢁ ½4ð6 ꢀ MRÞꢃ
2ꢀ
2ꢀ
spectra of: ½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb (brown line) ; ½Pdꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃb (blue line);
2þ
2ꢀ
2ꢀ
½Hꢃ2 ꢁ ½4ð8 ꢀ MRÞꢃ
2þ
2ꢀ
½Pdꢃ ꢁ ½6ð6 ꢀ MRÞꢃa
(green line); [Pd]2+ꢁ[5(4.5 ꢀ MR)]2ꢀ (red line) and
2þ
[Pd]2+ꢁ[5(5 ꢀ MR)]2ꢀ (black line), (b) IR spectra of as-synthesized Pd/Al-MCM-41.
(For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
½Hꢃ2 ꢁ ½5ð5 ꢀ MRÞꢃ
2þ
2ꢀ
91
80
68
½Hꢃ2 ꢁ ½4ð4:5 ꢀ MRÞꢃ
2þ
2ꢀ
86
70
54
½Hꢃ2 ꢁ ½6ð6 ꢀ MRÞꢃa
2þ
2ꢀ
101
62
98
95
½Hꢃ2 ꢁ ½6ð6 ꢀ MRÞꢃb
2þ
2ꢀ
65
68
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃa
Table 6
2þ
2ꢀ
54
32
32
29
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃb
Theoretical vibrational frequencies m
asym (cmꢀ1) (BM) perturbed by Pd calculated with
2þ
2ꢀ
162
121
108
94
B3LYP//PBE in [Pd]2+ꢁ[X(Y ꢀ MR)]2ꢀ
.
½Hꢃ2 ꢁ ½8ð8 ꢀ MRÞꢃc
[X(Y ꢀ MR)]
masym
[X(Y ꢀ MR)]
masym
[4(3 ꢀ MR)]
[4(4 ꢀ MR)]
[4(5 ꢀ MR)]
[4(4.5 ꢀ MR)]
[4(6 ꢀ MR)]
[4(8 ꢀ MR)]
[5(5 ꢀ MR)]
845
846
851
815
814
818
917
[5(4.5 ꢀ MR)]
[6(6 ꢀ MR)]a
[6(6 ꢀ MR)]b
[8(8 ꢀ MR)]a
[8(8 ꢀ MR)]b
[8(8 ꢀ MR)]c
897
892
945
900
908
900
IR spectra cannot easily be used to characterize silica and
aluminosilica modes of vibration owing to the large band rising be-
low 700 cmꢀ1. Nonetheless, they are helpful to identify the
Pd2+ꢁMS2ꢀ complexes. For the sake of rapidity, we have performed
the frequencies calculations at the (B3LYP//PBE) level. The experi-
mental and theoretical IR spectra are shown in Fig. 5 for the most
stable Pd2+ꢁMS2ꢀ clusters, that2ꢀis, ½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃ2bꢀ; ½Pdꢃ2þꢁ
Pd2+ꢁ[5(5 ꢀ MR)]2ꢀ. We focus now on the analysis of the bands
ranging between 700 and 1400 cmꢀ1. All theoretical spectra pres-
ent a shape that reproduces the experimental spectra with a shift
inherent to the systematic DFT offset. In Table 6, the wavenumbers
of the TO bond vibrations involving Pd2+ are reported for all consid-
ered clusters. They lie in the range 814–945 cmꢀ1 depending on the
ring size. Experimentally, we found wavenumbers around
930 cmꢀ1 in agreement with the literature value, 928 cmꢀ1 [41].
The values corresponding to the 4-membered rings are too small,
which confirms that Pd2+ is not likely deposited on four-membered
rings, in agreement with their low calculated stability.
½6ð6 ꢀ MRÞꢃ2aꢀ; ½Pdꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa
,
Pd2+ꢁ[5(4.5 ꢀ MR)]2ꢀ
and
Table 7
Theoretical vibrational frequencies mPdAO (cmꢀ1
)
calculated with B3LYP//PBE in
½PdOꢃ2þ ꢁ ½XðY ꢀ MRÞꢃ
.
2ꢀ
[X(Y ꢀ MR)]
mPdAO
[X(Y ꢀ MR)]
mPdAO
[4(3 ꢀ MR)]
[4(4 ꢀ MR)]
[4(5 ꢀ MR)]
[4(4.5 ꢀ MR)]
[4(6 ꢀ MR)]
[4(8 ꢀ MR)]
[5(5 ꢀ MR)]
703
699
693
692
704
685
724
[5(4.5 ꢀ MR)]
[6(6 ꢀ MR)]a
[6(6 ꢀ MR)]b
[8(8 ꢀ MR)]a
[8(8 ꢀ MR)]b
[8(8 ꢀ MR)]c
676
629
622
644
628
663
2ꢀ
2ꢀ
The oxidized form simulated so far is the [O@Pd]2+ oxo species,
where Pd is at the +4 oxidation state, as suggested in previous
experimental works [34,56]. The calculated vibrational wavenum-
bers of the PdAO bond in the corresponding set of clusters vary be-
tween 622 and 724 cmꢀ1 (Table 7) and should be compared to the
experimental value of 630 cmꢀ1 (Fig. 6). The best fit lies on
½PdOꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa;b and ½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa;b (622–644 cmꢀ1
)
whereas those obtained for 4- and 5-membered rings or ½PdOꢃ2þꢁ
2ꢀ
½8ð8 ꢀ MRÞꢃc are the worst. The two first correspond consistently
to the most stable oxidized systems.
As mentioned in Section 4.4, the vibrational frequencies of the
most interesting structures have also been calculated at two other

234
A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
Fig. 6. The simulated (B3LYP//PBE) (a, b) and experimental (c) Raman spectra: (a)
B3LYP Raman active vibrational modes of the different clusters as follows:
2ꢀ
2þ
2ꢀ
½Hꢃ22þ ꢁ ½6ð6 ꢀ MRÞꢃa
(black line); ½Hꢃ2 ꢁ ½6ð6 ꢀ MRÞꢃb
(red line); ½Hꢃ22þꢁ
½8ð8 ꢀ MRÞꢃa (green line) and ½Hꢃ22þ ꢁ ½8ð8 ꢀ MRÞꢃb (blue line). (b) B3LYP Raman
2ꢀ
2ꢀ
active vibrational modes of different clusters as follows: ½PdOꢃ2þ ꢁ ½6ð6 ꢀ MRÞꢃa
2ꢀ
Fig. 7. B3LYP/TZVP-optimized geometries of different forms of oxidized palladium
in different sites.
2þ
2ꢀ
2ꢀ
(black line); ½PdOꢃ ꢁ ½6ð6 ꢀ MRÞꢃb (red line); ½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃa (green line)
and ½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb (blue line). (c) Raman spectra of PdO/Al-MCM-41 (as in
2ꢀ
Fig. 2d). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
levels, PBE and B3LYP for comparison. The results obtained with
PBE differ from the others and are farther from the experimental
values. In the following, the wavenumbers are calculated using
only B3LYP.
Other oxidized forms of palladium in mesoporous materials
have been proposed [8,35,36,41,57]. Bell et al. suggested the pres-
ence of Pdn+ (n = 1 ꢀ 3) species on zeolite supports [53,63]. The Pd
cation may be associated with the zeolite either as a charge com-
pensating species viz. ZꢀPdn+(OHꢀ)nꢀ1 or as a neutral PdO type of
species attached to the zeolite framework via Brønsted acid sites
viz. ZꢀH+(PdO) and ZꢀH+(PdO)H+Zꢀ. Such an eventuality was
2ꢀ
investigated using the model ½PdðOHÞꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb (Fig. 7)
where Pd is formally at the oxidation state +3, consistent with a
doublet spin-electronic configuration. The calculated Raman spec-
trum (Fig. 8b) contains an intense band at 643 cmꢀ1 assigned to
the PdAOH bond stretching of this moiety linked to framework
oxygen atoms. In addition, a second less intense band at ca.
700 cmꢀ1 as well as a very small hump around 281 cmꢀ1 are also
found. The first two features may nicely fit with the experimental
peak at 630 and 702 cmꢀ1 whereas the third one seems not in-
tense enough to match the experimental data at 270 cmꢀ1. To fur-
ther support such an occurrence, an ESR investigation of the
sample has been performed at 118 K. The spectrum exhibits typ-
ically a broad signal at g = 2.12 (peak to peak linewidth of 110 G)
and a very sharp one at g = 2.002 (Fig. S7). The former could be
assigned to Pd(III) though rarely observed as a monomeric com-
plexes in solution or in zeolites (average g factor ranging in the
range 2.12–2.13 and a g tensor span from 2.020 to 2.19) [64–
66]. The later is likely due to an organic radical on carboneous
residue formed during the calcination. Both types of species ac-
counts for a very low concentration of spins (<0.1 wt.% for Pd(III))
and attest for a marginal oxidation-reduction process where sin-
gle electron has been involved. The presence of [Pd(OH)]2+ species
on eight-membered rings appears here as a reasonable possibility
as a marginal product according to the rather weak ESR signal. A
XANES experiment could have been performed. However, owing
to the small quantity of the Pd3+ species, the corresponding peak
should be hardly visible [67]. Note also that in a previous study
on Pd-Al-MCM-41 using a similar method of palladium deposi-
Fig. 8. Comparison of experimental Raman spectrum of PdO/Al-MCM-41 spectrum
with simulated B3LYP Raman active vibrational modes of clusters as follows: (a)
2þ
2ꢀ
½PdOꢃ ꢁ ½8ð8 ꢀ MRÞꢃb2ꢀ, (b) ½PdOHꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb2ꢀ, (c) ½PdðOHÞ2ꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb
.
(d) Raman spectra of PdO/Al-MCM-41.
tion, there was no mention of unusual oxidation states of palla-
dium [4]. However, they did not report ESR experiments and
did not comment the XANES profiles considering only the EXAFS
analysis that hardly detects phases below 10% in concentration. It
is stressed also that the much higher aluminum loading (>10
times) is more favorable for stabilizing isolated species and the
associated unusual oxidation states in the present case.
Mobile Pd(OH)2 moieties may also form in O2 at 773 K as sug-
gested by Pommier and Gélin [41]. Their presence was evaluated

A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
235
2ꢀ
from two model clusters ½PdðOHÞꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb and ½PdðOHÞꢃ2þꢁ
in Figs. S9e and S9f. The most characteristic band for the
HOAPdAOAPdAOH moiety in [HOAPdAOAPdAOH]4+
[6(6 ꢀ
MR)]4 arises at 550 cmꢀ1 and is assigned to PdAO vibrations
involving framework oxygen atoms. For [HOAPdAOAPdAOH]4+
2ꢀ
½8ð8 ꢀ MRÞꢃd that differ one to another only from the position of
ꢁ
the Al atoms (Figs. 7 and S.8). The spin-electronic configuration is
back to a singlet state. The calculated Raman spectrum of the former
model (Fig. 8c), shows two intense bands at 612 and 653 cmꢀ1, re-
lated to OH bending modes coupled with the framework oxygen
atoms, and also an intense band around 270 cmꢀ1. This species
may explain the relatively strong experimental peak at 274 cmꢀ1
not fully explainedneither by [PdO]2+ nor by [Pd(OH)]2+ investigated
above (Fig. 8c). The two intense peaks at 612 and 653 cmꢀ1 may
account for the shoulder identified at 602 cmꢀ1 and the asymmetry
on the high energy side of the main peak at 630 cmꢀ1. For the latter
model (Fig. S.9a), there are three main calculated bands correspond-
ing to (i) the PdAO stretching mode (intense) at ca. 628 cmꢀ1, (ii) the
ꢁ
[8(8 ꢀ MR)]4ꢀ the two intense bands at 683 and 644 cmꢀ1 are asso-
ciated to the PdAOAPd vibration of the bridge oxygen and to the
bending vibration of OH in the ring plane coupled with the frame-
work oxygen atom, respectively. The 593 cmꢀ1 band was attrib-
uted to the vibration of PdAOH in interaction with framework
oxygen atoms. In all spectra, no band exists around 630 cmꢀ1 nor
around 274 cmꢀ1, and these structures can be rejected in the case
of our Al-MCM-41.
in-phase coupling
m
(PdO) stretching vibrations at 720 cmꢀ1 and, (iii)
5. Discussion
the corresponding out-of-phase coupling at 577 cmꢀ1 that is
strongly coupled with the framework vibrations. The absence of
In this study, models with ring structures that are derived from
polyhedral oligomeric silsesquioxanes (POSS) have been used as
representation of the surface of Al-MCM-41 to study the deposition
of isolated Pd species into mesoporous materials. The stability of
the Pd+2 cations in the various clusters depends on the number
of bonds between palladium and oxygen atoms directly connected
to Al. The calculations show that the Pd+2 cation is located in a
square-planar conformation with the oxygen atoms. Deviation
from this conformation destabilizes the Pd complex, which pre-
vents Pd to adsorb onto the 4-membered rings. The most stable
Pd+2ꢁMSꢀ2 complexes correspond to adsorption on 6-membered
rings ([6(6 ꢀ MR)]b. The formation of the oxidized form PdO2+ is
the most energetically favorable species for the [8(8 ꢀ MR)]b clus-
ter. The second most stable structure varies with the temperature.
Here, Pd is in a square-planar environment with an average PdAO
bond distance around 210 pm with the oxygen atoms of the ring
and 181 pm with the oxygen oxo. These results are in good agree-
ment with the experimental data that suggest that Pd deposited
into zeolites gives, after calcination, square-planar oxo species
localized in 8-membered rings with distances PdAOMS around
210 pm [36,56].
A characteristic band for ion-exchanged silica materials, called
BM, originates from the perturbation by the cation of the antisym-
metric vibrations of the framework TAO bands. The position of this
band is function of the silica ring size and of the nature of the cat-
ion [44,45,68]. A BM value of 930 cmꢀ1 is reported for Pd2+ deposits
[36,45]. In ref. 45, the palladium atom is proposed to be situated in
a 10-membered ring. Such ring size is not observed in MCM-41
[69], and we therefore did not consider it. According to the Gibbs
free reaction energies, the most stable ring to deposit Pd2+ is a 6-
membered ring ([6(6 ꢀ MR)]b). In the FTIR spectrum of our synthe-
sized catalyst Pd/Al-MCM-41, a band around 930 cmꢀ1 appears
after cation exchange (Fig. 2b). By comparison with the literature,
this shows that Pd2+ cations have effectively been incorporated in
the mesoporous support according to the model proposed here.
The calculated values for the BM band corresponding to the most
stable systems ([6(6 ꢀ MR)]b) and [8(8 ꢀ MR)]b) are between 908
and 945 cmꢀ1. The comparison between experimental and calcu-
lated IR spectra indicates that, in our Pd/Al-MCM-41, catalyst
Pd2+ is likely incorporated into 6- or 8-membered rings, in agree-
ment with the calculated energies of formation. These results are
similar to those found for SBA-15 mesoporous sieves [70]. The val-
ues obtained for the 4-membered rings (between 814 and
851 cmꢀ1) exclude again the presence of Pd in these small rings.
Fig. 6 (spectrum c) shows the Raman spectrum of the catalyst
powder PdO/Al-MCM-41 recorded at room temperature. The spec-
trum exhibits significant peaks at 630 and 274 cmꢀ1 and bands at
410, 395 cmꢀ1. The two latter bands can be assigned to modes of
the siloxane bridges and of the ring structure [39,71,40]. On the
one hand, no band at 630 nor 274 cmꢀ1 exists in the calculated
peak near 580 and 720 cmꢀ1, precludes the existence of the cluster
2ꢀ
½PdðOHÞꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃd
.
Bell and co-workers [53,63] suggested also the existence of
[Pd(OH)]+ interacting with a cycle containing only one Al atom,
which gives the clusters [Pd(OH)]+ꢁ[8(8 ꢀ MR)]ꢀ. In this species,
Pd is in the oxidation state +2 and the spin-electronic configuration
corresponds to a singlet state (structure [Pd(OH)]+ꢁ[8(8 ꢀ MR)]ꢀ in
Fig. S8). Its Raman spectrum is given in Fig. S.9c. No band around
630 cmꢀ1 is observed in the spectrum.
Sachlter and co-workers [57] proposed the formation of a bridge
[PdAOAPd]2+ form. The [8(8 ꢀ MR)]d cluster where two aluminum
atoms are opposite one from each other in the ring allows the graft-
ing of such species (structure ½Pd ꢀ O ꢀ Pdꢃ2þꢁ ½8ð8 ꢀ MRÞꢃd
,
2ꢀ
Fig. S8). The spin-electronic configuration corresponds to a triplet
state. The calculation (Fig. S9b) shows an intense Raman band at
752 cmꢀ1, which corresponds to the vibration of the bridge oxygen.
Since such a band is not observed on the experimental Raman spec-
tra, the possibility of this species is discarded. The cluster
2ꢀ
½8ð8 ꢀ MRÞꢃd is required for the grafting of [(PdOH)2]2+ for the
same reasons as for ½Pd ꢀ O ꢀ Pdꢃ2þꢁ ½8ð8 ꢀ MRÞꢃd2ꢀ. The spin-elec-
tronic configuration corresponds to a singlet state (structure
½ðPdOHÞ2ꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃd in Fig. S8). The most intense Raman
2ꢀ
band of the calculated spectrum arises at around 564 cmꢀ1 for a
vibration of both PdAO bonds of (PdOH)2 moiety (Fig. S9d). The ab-
sence of such a band in the experimental spectrum suggests that
this oxidized form of palladium does not exist in the present PdO/
Al-MCM-41 catalyst.
Kim and co-workers [36] also proposed that Pd ion has been
substantially oxidized to Pd4+ with the existence of linear
[HOAPdAOAPdAOH]4+ clusters, the Pd4+ ions lying in six-ring
planes. The [6(6 ꢀ MR)] cluster containing four aluminum atoms,
two in each 6-membered ring in opposition one to each other,
allows the grafting of the HOAPdAOAPdAOH moiety (structure
[(PdOH)2O]4+ꢁ[6(6 ꢀ MR)]4ꢀ as depicted in Fig. S8. A good agree-
ment exists between the bond distances and angles of the opti-
mized structure and those determined experimentally [36] (Table
S6), except that the X-ray structure suggests a linear PdAOAPd
arrangement (Pd(2)AO(5)APd(2) = 180°), while it is bent in the
calculated structure (Pd(2)AO(5)APd(2) = 134°). Another possibil-
ity is to graft HOAPdAOAPdAOH inside [8(8 ꢀ MR)] (structure
[(PdOH)2O]4+ꢁ[8(8 ꢀ MR)]4ꢀ in Fig. S8). In the optimized geometry,
one of the OH groups connected to Pd lies in the ring plane while
the other is located out of the plane of the other ring. This may
be due to the disposition of the aluminum atoms: in one of the
rings, Pd is already linked to four framework oxygen atoms and
the OH group must lie out of the plane (as in ½PdOꢃ2þꢁ
½6ð6 ꢀ MRÞꢃ2bꢀ). In the other ring, Pd is linked to only three oxygen
atoms and OH can lie in the plane, giving the favorable ML4 planar
arrangement. The Raman spectra of these two structures are given

236
A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
spectra of the supports, namely ½Hꢃ22þ ꢁ ½6ð6 ꢀ MRÞꢃa2ꢀ; ½Hꢃ22þꢁ ½6ð6ꢀ
oxidation of Pd2+, three species are likely to exist, namely [PdO]2+
,
2ꢀ
2ꢀ
MRÞꢃ2bꢀ; ½Hꢃ22þ ꢁ ½8ð8 ꢀ MRÞꢃa and ½Hꢃ22þ ꢁ ½8ð8 ꢀ MRÞꢃb (Fig. 6a). On
the other hand, our simulations of the spectra corresponding to
the systems obtained when PdO2+ is deposited onto [8(8 ꢀ MR)]b,
[8(8 ꢀ MR)]a, [6(6 ꢀ MR)]b, and [6(6 ꢀ MR)]a show a good agree-
ment with the experimental spectra (Fig. 6b). Indeed, these bands
are associated with vibrational modes of PdO2+ in interaction with
[PdOH]2+, and [Pd(OH)2]2+ in 8-membered rings. Such 8-mem-
bered rings have already been invoked in Mordenite as the sites
for incorporating PdO [56]. In forthcoming contributions, we will
examine the reactivity of the oxidized species identified in the
present work for CH4 oxidation reactions and the effect of the Si/
Al molar ratio.
MS2ꢀ that at 630 cmꢀ1 to (PdO) and that at 274 cmꢀ1 to combina-
m
tions of PdO out-of-plane and in-plane bending modes coupled
with support modes. Hence, the band at 630 cmꢀ1 can be associ-
ated with the Raman active B1g vibrational modes of PdO [72].
The broadening of this band implies that the amorphous environ-
ment affects the vibrations as suggested by Demoulin et al. [72].
Moreover, among all the other Pd oxidized forms for which the
Raman spectra have been simulated, two, namely ½PdOHꢃ2þꢁ
Acknowledgments
The authors would like to thank G. Montagnac for Raman exper-
iments, Sandrine Denis-Quanquin for the solid-State NMR experi-
ments, and Lhoussain Khrouz for the ESR measurements. Grants
of computer time from Pole of numerical simulations (PSMN) in
Ecole Normale Supérieure de Lyon are gratefully acknowledged.
2ꢀ
½8ð8 ꢀ MRÞꢃb and ½PdðOHÞ2ꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃ2bꢀ, display spectra in
agreement with the experimental one. To summarize the results,
the simulated spectra of the three most likely structures are gath-
ered in Fig. 8 with the experimental one. The spectrum for
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2012.02.014.
2ꢀ
½PdOꢃ2þ ꢁ ½8ð8 ꢀ MRÞꢃb differs from that shown in Fig. 6 by the
relative intensity of the peaks because the level of calculation is
different (B3LYP vs. B3LYP//PBE). The two peaks at 606 and
References
626 cmꢀ1 correspond to
m(PdO) coupled with the SiAO elongations
of two different cycles of the cluster adjacent to the 8-membered
ring. All three simulated spectra present bands around 643–606,
390–386, and 281–270 cmꢀ1 and hence are in agreement with
the experimental data.
[1] P. Gélin, M. Primet, Appl. Catal. B 39 (2002) 1.
[2] J.A.C. Ruiz, M.A. Fraga, H.O. Pastore, Appl. Catal. B 76 (2007) 115.
[3] I. Yuranov, L. Kiwi-Minsker, P. Buffat, A. Renken, Chem. Mater. 16 (2004) 760.
[4] T. Kawabata, I. Atake, Y. Ohishi, T. Shishido, Y. Tian, K. Takaki, K. Takehira, Appl.
Catal. B 66 (2006) 151.
[5] Z.J. Wang, Y. Liu, P. Shi, C.J. Liu, Y. Liu, Appl. Catal. B 90 (2009) 570.
[6] K. Okumura, M. Niwa, J. Phys. Chem. B 104 (2000) 9670.
[7] R. Harmsen, S. Bates, R.A. van Santen, Faraday Discuss. 106 (1997) 443.
[8] M.J. Rice, A.K. Chakraborty, A.T. Bell, J. Phys. Chem. B 104 (2000) 9987.
[9] J. Wang, C. Liu, Z. Fang, Y. Liu, Z. Han, J. Phys. Chem. B 108 (2004) 1653.
[10] R. Grybos, J. Hafner, L. Benco, H. Toulhoat, J. Phys. Chem. C 111 (2007) 6454.
[11] R. Grybos, J. Hafner, L. Benco, P. Raybaud, J. Phys. Chem. C 112 (2008) 12349.
[12] L. Bonneviot, M. Che, K. Dyrek, R. Shöllner, G. Wendt, J. Phys. Chem. 90 (1986)
2379.
[13] L. Bonneviot, D. Olivier, M. Che, J. Chem. Soc., Chem. Commun. (1982) 952.
[14] M. D’Amore, C. Bisio, G. Talarico, M. Cossi, L. Marchese, Chem. Mater. 20 (2008)
4980.
[15] G. Tabacchi, E. Gianotti, E. Fois, G. Martra, L. Marchese, S. Coluccia, A. Gamba, J.
Phys. Chem. C 111 (2007) 4946.
[16] M. Hartmann, A. Pöppl, L. Kevan, J. Phys. Chem. 99 (1995) 17494.
[17] D.P. Debecker, K. Bouchmella, C. Poleunis, P. Eloy, P. Bertrand, E.M. Gaigneaux,
P.H. Mutin, Chem. Mater. 21 (2009) 2817.
In summary, the analysis of the FTIR and Raman spectra coupled
with the calculated reaction energies suggests that two oxidized
palladium species, namely [PdO]2+ in a 6- or 8-membered ring or
[Pd(OH)]2+ and [Pd(OH)2]2+ in a 8-membered ring are the species
deposited onto Al-MCM-41, which corresponds to the Pd oxidation
states +4 and +3. Nonetheless, the latter oxidation state is probably
too reactive and should remain a marginal species according to ESR
measurements. A Pd species with oxidation state +4 has also been
characterized on SnO2 as the major oxidized species after impreg-
nation by PdCl2 and calcination [73], which confirms the possible
existence of such a high oxidation state.
[18] L. Bonneviot, M. Morin, A. Badiei, WO 01/55031 A 1, 2001.
[19] N.R. Lang, A. Tuel, Chem. Mater. 16 (2004) 1961.
6. Conclusion
[20] M. Yonemitsu, Y. Tanaka, M. Iwamoto, Chem. Mater. 9 (1997) 2679.
[21] A.R. Badiei, L. Bonneviot, Inorg. Chem. 37 (1998) 4142.
[22] J. Sauer, P. Ugliengo, E. Garrone, V.R. Saunders, Chem. Rev. 94 (1994) 2095.
[23] J. Sauer, J.R. Hill, Chem. Phys. Lett. 218 (1994) 333.
[24] M.A. Zwijnenburg, S.T. Bromley, J.C. Janses, T. Maschmeyer, Chem. Mat. 16
(2004) 12.
[25] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.
[26] A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
[27] R. Ahlrichs, M. Bär, M. Häser, H. Horn, C. Kölmel, Chem. Phys. Lett. 162 (1989)
165.
[28] O. Treutler, R. Ahlrichs, J. Chem. Phys. 102 (1995) 346.
[29] P. Deglmann, F. Furche, R. Ahlrichs, Chem. Phys. Lett. 362 (2002) 511.
[30] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785.
[31] A. Schäfer, C. Huber, R. Ahlrichs, J. Chem. Phys. 100 (1994) 5829.
[32] E.E. Zvereva, A.R. Shagidullin, A. Katsyuba, J. Phys. Chem. A 115 (2011) 63.
[33] A.E. Reed, L.A. Curtiss, F. Weinhold, Chem. Rev. 88 (1988) 899.
[34] F. Blatter, K.W. Blazey, J. Phys. Chem. Solids. 52 (1991) 629.
[35] Y. Bi, G. Lu, Appl. Catal. B 41 (2003) 279.
[36] S.H. Lee, Y. Kim, K. Seff, J. Phys. Chem. B 104 (2000) 2490.
[37] C.A. Koh, R. Nooney, S. Tahir, Catal. Lett. 47 (1997) 199.
[38] C.Y. Chen, H.X. Li, M.E. Davis, Micropor. Mater. 2 (1993) 17.
[39] P.D. Vaz, C.D. Nunes, M. Vasconcellos-Dias, M.M. Nolasco, P.J.A. Ribeiro-Claro,
M.J. Calhorda, Chem. Eur. J. 13 (2007) 7874.
The results of this combined experimental and theoretical study
have provided deeper insight into the structure of the oxidized
forms of the palladium containing Al-MCM-41 mesoporous sup-
port. The amorphous character of the material has been taken into
account in the calculations by considering several types of sils-
esquioxane complexes with various ring sizes. The structure and
the vibrational properties of the Pd2+ species and of their oxidized
forms bonded to these complexes have been determined from DFT
calculations. The energy of formation of silica alumina exchanged
with palladium follows the order 6TO > 5TO > 8TO > 4TO at high
temperature, according to the nature of the cycle. The position of
the FTIR bands (BM) of the antisymmetric TAO vibrations values
around 900–945 cmꢀ1 are in good agreement with the band at
930 cmꢀ1 in our experimental spectrum, which can exclude the
existence of systems [Pd]2+ꢁ[4(Y ꢀ MR)]2ꢀ in relation with their
lower calculated stability. The comparison between the experi-
mental and simulated value of the BM band allowed us to assume
that Pd2+ is deposited in 6- or 8-membered rings. From the global
energy of formation at 300 K of the oxidized species from the H2MS
cluster, we2ꢀ found that the most stable species is ½PdOꢃ2þꢁ
½8ð8 ꢀ MRÞꢃb , where Pd4+ is connected to four oxygen atoms, three
of them being support framework oxygen atoms, the fourth oxygen
[40] Z. Luan, P.A. Meloni, R.S. Czernuszewicz, L. Kevan, J. Phys. Chem. B 101 (1997)
9046.
[41] B. Pommier, P. Gélin, Phys. Chem. Chem. Phys. 1 (1999) 1665.
[42] G. Wang, Y. Wang, Y. Liu, Z. Liu, Y. Guo, G. Liu, Z. Yang, M. Xu, L. Wang, Appl.
Clay Sci. 44 (2009) 185.
[43] G.A. Eimer, L.B. Pierella, G.A. Monti, O.A. Anunziata, Catal. Commun. 4 (2003)
118.
[44] L. Drozdová, R. Prins, J. Dedecek, Z. Sobalik, B. Wichterlová, J. Phys. Chem. B 106
(2002) 2240.
atom forming
experimental and theoretical Raman spectra, it is shown that, after
a
[PdO]2+ oxo group. By comparison of the

A. Gannouni et al. / Journal of Catalysis 289 (2012) 227–237
237
[45] J.A.Z. Pieterse, R.W. van den Brink, S. Booneveld, F.A. de Bruijn, Appl. Catal. B 46
[60] A. Fidalgo, L.M. Ilharco, Chem. Eur. J. 10 (2004) 392.
(2003) 239.
[61] S. Chempath, Y. Zhang, A.T. Bell, J. Phys. Chem. C 111 (2007) 1291.
[46] G.A. Eimer, L.B. Pierella, G.A. Monti, O.A. Anunziata, Catal. Lett. 78 (2002) 1.
[47] J.R. McBride, K.C. Hass, W.H. Weber, Phys. Rev. B 44 (2000) 5016.
[48] A. Baylet, P. Marécot, D. Duprez, P. Castellzzi, G. Groppi, P. Forzatti, Phys. Chem.
Chem. Phys. 13 (2011) 4607.
[62] M. Corno, A. Pedone, R. Dovesi, P. Ugliengo, Chem. Mater. 20 (2008) 5610.
[63] L.J. Lobree, A.W. Aylor, J.A. Reimer, A.T. Bell, J. Catal. 181 (1999) 189.
[64] J.R. Khusnutdinova, N.P. Rath, L.M. Mirica, J. Am. Chem. Soc. 132 (2010)
7303.
[49] A.S. Mamede, G. Leclecq, E. Payen, P. Granger, J. Grimblot, J. Mol. Struct. 651
(2003) 353.
[50] S.C. Su, J.N. Cartens, A.T. Bell, J. Catal. 176 (1998) 125.
[51] T.A. Albright, Tetrahedron 38 (1982) 1339.
[65] A. Raizman, J. Barak, J.T. Suss, Phys. Rev. B 31 (1985) 5716.
[66] S.N. Gandhi, G.D. Lei, W.M.H. Sachtler, Catal. Lett. 17 (1993) 117.
[67] A. Eyssler, P. Mandaliev, A. Winkler, P. Hug, O. Safonova, R. Figi, A. Weidenkaff,
D. Ferri, J. Phys. Chem. C 114 (2010) 4584.
[52] Y. Jean, The Orbital Molecular Ones in the Complexes, Editions of the
Polytechnic School, 2003, p. 96.
[68] Z. Sobalik, J.E. Šponer, B. Wichterlová, Stud. Surf. Sci. Catal. 130 (2000) 1463.
[69] B.P. Feuston, J.B. Higgins, J. Phys. Chem. 98 (1994) 4459.
[70] Z. Wang, D. Wang, Z. Zhao, Y. Chen, J. Lan, Comput. Theor. Chem. 963 (2011)
403.
[71] K.J. Chao, C.N. Wu, H. Chang, L.J. Lee, S.f. Hu, J. Phys. Chem. B 101 (1997)
6341.
[72] O. Demoulin, M. Navez, E.M. Gaigneaux, P. Ruiz, A.S. Mamede, P. Granger, E.
Payen, Phys. Chem. Chem. Phys. 5 (2003) 4394.
[73] R.-Q. Tan, Y.-Q. Guo, J.-H. Zhao, Y. Li, T.-F. Xu, W.-J. Song, Trans. Nonferrous
Met. Soc. Chin. 21 (2011) 1568.
[53] A.W. Aylor, L.J. Lobree, J.A. Reimer, A.T. Bell, J. Catal. 172 (1997) 453.
[54] K. Okumura, J. Amano, M. Niwa, Chem. Lett. (1999) 997.
[55] K. Okumura, M. Niwa, Top. Catal. 18 (2002) 85.
[56] S.N. Reifsnyder, M.M. Otten, H.H. Lamb, Catal. Today. 39 (1998) 317.
[57] B. Wen, Q. Sun, W.M.H. Sachtler, J. Catal. 204 (2001) 314.
[58] J.H. Park, B. Kim, C.H. Shin, G. Seo, S.H. Kim, S.B. Hong, Top Catal. 52 (2009) 27.
[59] N. Das, H. Eckert, H. Hu, I.E. Wachs, J.F. Walzer, F. Feher, J. Phys. Chem. 97
(1993) 8240.