synthesis of sio2 xerogels and pd/sio2 cogelled xerogel catalysts from silylated acetylacetonate...
TRANSCRIPT
www.elsevier.com/locate/jnoncrysol
Journal of Non-Crystalline Solids 343 (2004) 109–120
Synthesis of SiO2 xerogels and Pd/SiO2 cogelled xerogelcatalysts from silylated acetylacetonate ligand
Stephanie Lambert a, Luigi Sacco b, Fabrice Ferauche a, Benoıt Heinrichs a,*,Alfred Noels b, Jean-Paul Pirard a
a Laboratoire de Genie Chimique, B6a, Universite de Liege, B-4000 Liege, Belgiumb Centre d’Enseignement et de Recherche des Macromolecules (CERM), B6a, Universite de Liege, B-4000 Liege, Belgium
Received 7 October 2003
Abstract
SiO2 xerogels and Pd/SiO2 cogelled xerogel catalysts have been prepared in a mixture of tetrahydrofurane (THF) and ethanol
containing tetraethoxysilane (TEOS), and an aqueous ammonia solution of 0.18mol/l, from synthesized new silylated acetylaceto-
nate ligands, respectively, 3-[3-(trimethoxysilyl)propyl]-2,4-pentanedione (MS-acac-H), 2,2,6,6-tetramethyl-4-[3-(trimethoxysi-
lyl)propyl]-3,5-heptanedione (MS-dPvM), and 1,3-diphenyl-2-[3-(trimethoxysilyl)propyl]-1,3-propanedione (MS-dBzM), able to
form a chelate with a metal ion such as Pd2+. All samples form homogeneous colored gels. The resulting catalysts are composed
of palladium crystallites with a diameter of about 3.5nm, located inside primary silica particles exhibiting a monodisperse micro-
porous distribution as well as large palladium particles from 20 to 50nm, situated outside the silica aggregates. The silylated organic
ligand has a strong influence on the textural properties of xerogels and catalysts, both before and after calcination and reduction
steps. Changing the nature of the silylated ligand permits tailoring textural properties such as pore volume, pore size and surface
area. Although small palladium crystallites are located inside the silica particles, their complete accessibility, via the micropore net-
work, has been shown. 1,2-Dichloroethane hydrodechlorination over Pd/SiO2 catalysts mainly produces ethane and the reaction
rate increases linearly with palladium dispersion. Hydrodechlorination over Pd/SiO2 cogelled xerogel catalysts is a structure insen-
sitive reaction compared to the ensemble size concept.
� 2004 Elsevier B.V. All rights reserved.
1. Introduction
Viable preparative routes to silylated organic mole-
cules initially emerged in the 1960s, largely pioneered
by a group at Degussa [1]. Development of these com-
mercial-scale procedures was originally driven by the
interest in the bifunctional reagents as adhesion promot-
ers for inorganic–organic polymer composites such as
glass fiber-reinforced materials and mineral-filled elas-
0022-3093/$ - see front matter � 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2004.07.049
* Corresponding author. Tel.: +32 4 366 3505; fax: +32 4 366 3545.
E-mail address: [email protected] (B. Heinrichs).
tomers. Further developments arose from the realization
that attachment of catalytically active metal complexesto silica gels via silylated ligands afforded supported
complexes, whose properties combined the positive attri-
butes of both heterogeneous and homogeneous catalysts.
Immobilization of a sylilated ligand or metal complex
during a sol–gel reaction is typically a straightforward
manner of introducing the metal complex into the sol–
gel mixture, either at the same time as other reagents
or after the other precursors have been allowed to reactto form a sol. One of the most common reasons cited for
the use of sol–gel processing is the intimate molecular-le-
vel mixing of components that can be obtained at the
outset of the reactions [2,3]. In this way, several authors
110 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
developed the cogelation sol–gel process to disperse me-
tal particles in a silica matrix [4–6]: organically substi-
tuted alkoxides of the type (RO)3Si–(CH2)3–A in
which an organic amine group A, able to form a chelate
with a cation of a metal such as palladium, silver, cop-
per, etc., is linked to the hydrolysable silyl group(RO)3Si– via an inert and hydrolytically stable spacer
–(CH2)3–. The co-condensation of such molecules with
a network-forming reagent such as tetraethoxysilane
(TEOS), Si(OC2H5)4, results in materials in which the
metal is anchored to the SiO2 matrix. Heinrichs et al.
[7,8] and Lambert et al. [9–11] used this cogelation
method to synthesize mono- and bimetallic aerogel
and xerogel catalysts. By changing the molar ratiobetween the functionalized amino-alkoxide and TEOS,
textural properties such as pore volume, pore size and
surface area can be tailored. Although metal particles
are located inside the silica particles, their complete
accessibility, via the micropore network, after drying,
high temperature calcination and reduction of the gels,
has been shown [7–11].
The main purpose of the present work is to synthesizenew silylated acetylacetonate ligands, able to form a
chelate with a metal ion such as Pd2+ and able to react
with a silica-forming reagent such as TEOS to form
homogeneous silica gels containing highly dispersed org-
ano-metallic complexes. So, after vacuum drying, these
silica xerogels could be used as catalysts in several or-
ganic reactions such as cyclopropanation, Kharasch
reactions, etc. Furthermore, the samples are also cal-cined and reduced to analyze the dispersion, localiza-
tion, and accessibility of palladium and to compare
their catalytic activity to previously synthesized Pd/
SiO2 cogelled xerogel catalysts used for hydrodechlorin-
ation [9,10].
2. Experimental
2.1. Preparation of silylated acetylacetonate ligands and
palladium complexes
Synthesis of silylated ligands and metal complexes
was performed under an atmosphere of dry argon using
standard Schlenk techniques. Argon was purified by pas-
sage through columns of BASF R3-11 catalysts and 4Amolecular sieves. Tetrahydrofurane (THF) was freshly
distilled under argon from benzophenone and sodium
prior to its use. Dichloromethane was distilled under ar-
gon from calcium hydride. Sodium acetylacetonate was
synthesized from acetylacetone (Fig. 1, scheme 1, struc-
ture 1a) according to Charles [12] and dried at a temper-
ature of 100 �C and under a pressure of 0.07Pa. The
above-described synthesis method is also applied to dip-ivaloylmethane (Fig. 1, scheme 1, structure 1b) and dib-
enzoylmethane (Fig. 1, scheme 1, structure 1c).
Acetylacetone functionalized by an alkoxysilane
group is obtained by reaction of sodium or potassium
acetylacetonate with the adequate halogenocompound
in dry dimethylsulfoxide (DMF) or N,N-dimethylform-
amide (DMSO). Indeed, the kind of halogen plays a cru-
cial role in the C- and O-alkylation competition [13].Using (3-iodopropyl)-trimethoxysilane in THF, 1H
NMR and gas chromatography analyses show that the
O-alkylation is drastically minimized to the benefit of
the C-alkylation. This way is also applied to dipivaloy-
methane and dibenzoylmethane (Fig. 1, scheme 1).
Synthesis of 3-[3-(trimethoxysilyl)propyl]-2,4-pen-
tanedione, MS-acac-H (Fig. 1, scheme 1, structure 2a).
To a suspension of 5g of sodium acetylacetonate in30ml of dry THF, 11.9g of (3-iodopropyl)-trimethoxysi-
lane were added. The resulting mixture was stirred under
reflux for five days and the reaction was monitored by
gas chromatography. The solvent was then removed un-
der reduced pressure and the product extracted with dry
dichloromethane. The slightly viscous colorless liquid
obtained by distillation (101–104 �C/0.07Pa) can be
indefinitely stored at a temperature of �15 �C under ar-gon. Yield: 55%.
Synthesis of 2,2,6,6-tetramethyl-4-[3-(trimethoxysi-
lyl)propyl]-3,5-heptanedione, MS-dPvM (Fig. 1, scheme
1, structure 2b). The same procedure described above
was followed to obtain a viscous colorless liquid by dis-
tillation (115–118 �C/0.07Pa) and stored at a tempera-
ture of �15 �C under argon. Yield: 45%.
Synthesis of 1,3-diphenyl-2-[3-(trimethoxysilyl)pro-
pyl]-1,3-propanedione, MS-dBzM (Fig. 1, scheme 1,
structure 2c). The same procedure described above was
followed to obtain a viscous orange liquid stored at a
temperature of �15 �C under argon. Yield: 80%.
Although palladium acetate in the presence of acetyl-
acetone readily gives palladium acetylacetonate in a
wide variety of solvents, this method does not work with
the functionalized diones. A pre-reaction between thedione and potassium t-butoxyde is necessary, which
then yields the palladium complex with 3-[3-(trimeth-
oxysilyl)propyl]-2,4-pentanedione, Pd(MS-acac)2, 2,2,6,
6-tetramethyl-4-[3-(trimethoxysilyl)propyl]-3,5-heptane-
dione, Pd(MS-dPvM)2, and 1,3-diphenyl-2-[3-(trimeth-
oxysilyl)propyl]-1,3-propanedione, Pd(MS-dBzM)2(Fig. 1, scheme 2).
Synthesis of palladium complexes from 3-[3-(tri-
methoxysilyl)propyl]-2,4-pentanedione, Pd(MS-acac)2,
from 2,2,6,6-tetramethyl-4-[3-(trimethoxysilyl)pro-
pyl]-3,5-heptanedione, Pd(MS-dPvM)2 and from 1,
3-diphenyl-2-[3-(trimethoxysilyl)propyl]-1,3-propanedi-
one, [Pd(MS-dBzM)2]. To a fine suspension of 1.1g of
potassium t-butoxyde in 150ml of THF was added in a
first time, either 3.4g of 3-[3-(trimethoxysilyl)propyl]-
2,4-pentanedione or 3.4g of 2,2,6,6-tetramethyl-4-[3-(trimethoxysilyl)propyl]-3,5-heptanedione or 3.4g
of 1,3-diphenyl-2-[3-(trimethoxysilyl)propyl]-1,3-prop-
R
ONaOH
Na+
R
O
Si(OCH3)3Si(OCH3)3 Si(OCH3)3
I
MeOHH2O
Na+
R
O
Na2CO3THF
R
O
R
O OH
Si(OCH3)3
R
O
+
2a, 2b, 2c
1a, 1b,1c
R
O
(H3CO)3Si
Pd(OAC)2KO tBu
THF
R
R
O
O
(H3CO)3Si
Pd 2+
2
2a , 2b , 2c3a , 3b, 3c
Ra -CH3
b -C(CH3)3
c -Ph
R
R
R
R
R
R
O
O
O
O
O
O
R
Scheme 1
Scheme 2
Fig. 1. Synthesis schemes of silylated acetylacetonate ligands and palladium complexes.
S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120 111
anedione and in a second time a solution of 1g of palla-
dium acetate in 10ml of THF. After 20min under stir-
ring, the brown suspension was filtrated and the
filtrate was evaporated under vacuum. A very viscousbrown solid was obtained and stored during three days
at a temperature of �15 �C under argon. The yield of
complexes Pd(MS-acac)2, Pd(MS-dPvM)2 and Pd(MS-
dBzM)2 is 50%. The structures of these silylated com-
plexes are presented in Fig. 1 (scheme 2, structure 3a
for Pd(MS-acac)2, structure 3b for Pd(MS-dPvM)2 and
structure 3c for Pd(MS-dBzM)2).
2.2. Characterization of silylated acetylacetonate ligands
and palladium complexes
1H NMR and 13C NMR spectra were recorded on a
Bruker 400 spectrometer with TMS as internal standard.
Infrared spectra (thin film on NaCl pellets) were re-
corded on a Perkin–Elmer FTIR 1720X. GC analysis
was carried out using a Varian Star 3400CX.
2.3. Synthesis of SiO2 xerogels and Pd/SiO2 cogelled
xerogel catalysts
Sample designation. Three silica xerogels and four Pd/
SiO2 cogelled xerogel catalysts were prepared and their
synthesis parameters are presented in Table 1. Each
sample is named by the corresponding silylated acetyl-
acetonate ligand or palladium complex. The names of
palladium samples are followed by either the lettersTE, for a solvent mixture of THF and ethanol, or by
the letters AE, for a solvent mixture of acetone and
ethanol.
Gel synthesis. Three SiO2 gels were synthesized from
the corresponding silylated acetylacetonate ligand –
MS-acac-H or MS-dPvM or MS-dBzM–, TEOS
Table
1
SynthesisoperatingvariablesofxerogelsandPd/SiO
2cogelledxerogel
catalysts
Sample
nma
(mmol)
nligandb
(mmol)
nTEOS
(mmol)
nH
2O
(mmol)
Solvent
mixture
ndiluantc
(mmol)
nC
2H
5OH
(mmol)
Gelation
behavior
Theoreticalmetal
loading(w
t%)
Weightloss
d±2%
(wt%
)Actualmetal
loading±0.1%
(wt%
)
Pd(M
S-acac)
2TE
0.266
1.57
44.5
230
THF-ethanol
123
343
Black
gel
1.0
16
1.2
Pd(M
S-acac)
2AE
0.273
1.68
45.7
237
Acetone-ethanol
136
343
Black
gel
1.0
14
1.2
Pd(M
S-dPvM) 2TE
0.267
1.59
44.5
230
THF-ethanol
123
343
Greygel
1.0
01.0
Pd(M
S-dBzM
) 2TE
0.255
1.48
44.6
230
THF-ethanol
123
343
Black
suspension
1.0
––
MS-acac-H
–2.93
48.5
253
Ethanol
–514
Whitegel
–0
–
MS-dPvM
–3.12
48.3
253
Ethanol
–514
Colorlessgel
–0
–
MS-dBzM
–2.91
48.5
253
Ethanol
–514
Light-orangegel
–0
–
anmthesilylatedacetylacetonate
palladium
complex,iseither
Pd(M
S-acac)
2,orPd(M
S-dPvM) 2,orPd(M
S-dBzM
) 2.
bnligandthesilylatedacetylacetonate
ligand,iseither
MS-acac-H,orMS-dPvM,orMS-dBzM
.cndiluantthediluantsolvent,isTHForacetone.
dWeightloss
=100·(m
th�
ma)/m
th,wherem
thisthetheoreticalmass
andm
aistheactualcatalyst
mass
measuredafter
drying,calcinationandreductionsteps.
112 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
(Si(OC2H5)4) and aqueous ammonia. So a 0.18N NH3
aqueous solution in half of the total ethanol volume
was added, under stirring, to a mixture containing
TEOS, MS-acac-H or MS-dPvM or MS-dBzM, and
the rest of ethanol. The molar ratio silylated ligand/
TEOS must be equal to 0.06 to allow gelification. FourPd/SiO2 gels were prepared from the corresponding sil-
ylated acetylacetonate palladium complex –Pd(MS-
acac)2 or Pd(MS-dPvM)2 or Pd(MS-dBzM)2–, TEOS
and aqueous ammonia. In these syntheses, Pd(MS-
acac)2 or Pd(MS-dPvM)2 or Pd(MS-dBzM)2 was
dissolved in a mixture of 10ml of ethanol and 10ml of
acetone or THF, and in which some amount of silylated
acetylacetonate ligand corresponding to the used palla-dium complex was present to allow gelification. So the
molar ratio silylated ligand/TEOS is equal to 0.05 in pal-
ladium gels. After addition of TEOS, a 0.18N NH3
aqueous solution in the remaining ethanol was added
to the mixture under vigorous stirring. The volume of fi-
nal solutions was 30ml. The hydrolysis ratio, which is
the molar ratio H = [H2O]/([TEOS] + 3/4 [sylilated lig-
and]), and the dilution ratio, which is the molar ratioR = [ethanol]/([TEOS] + [sylilated ligand]) were kept
constant at values of 5 and 10 respectively for all sam-
ples. The vessel was then tightly closed and heated up
to 80 �C for 10 days (gelling and aging [14]).
Drying. The wet gels were dried under vacuum
according to the following procedure: the flasks were
opened and put into a drying oven at 80 �C, and the
pressure was slowly decreased (to prevent gel bursting)to reach the minimum value of 1200Pa after 90h. The
drying oven was then heated at 150 �C for 72h. The
resulting samples are xerogels [14].
Calcination. The calcination conditions for Pd/SiO2
catalysts were as follows: the sample was heated up to
400 �C at a rate of 120 �C/h under flowing air
(0.02mmol/s); this temperature was then maintained
for 12h in air (0.1mmol/s). All dried SiO2 xerogels wereheat treated in air at 450 �C for 72h to remove organic
residues.
Reduction. The Pd/SiO2 catalysts were heated up to
350 �C at a rate of 350 �C/h under flowing H2
(0.23mmol/s) and maintained at this temperature for
3h (same flow).
2.4. Characterization of SiO2 xerogels and Pd/SiO2
cogelled xerogel catalysts
Nitrogen adsorption–desorption isotherms were mea-
sured at 77K on a Fisons Sorptomatic 1990 after out-
gassing for 24h at ambient temperature. After a 2-h
outgassing at ambient temperature, mercury porosi-
metry measurements were performed with sample mon-
oliths using a manual porosimeter from 0.01 to 0.1MPaand a Carlo Erba Porosimeter 2000 from 0.1 to
200MPa.
S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120 113
SiO2 and metal particles sizes were examined by
transmission electron microscopy (TEM) with a Philips
CM100 microscope. All samples were impregnated with
an epoxy resin (EPON 812) to which an amine was
added to serve as a hardener. Hardening goes on for
72h at 60 �C and 60nm slices were then cut up with aReichert-Jung Ultracut E microtome. Finally, these
slices were put on a copper grid.
Metal dispersion in Pd/SiO2 cogelled xerogel catalysts
was determined from CO chemisorption at 30 �C on a
Fisons Sorptomatic 1990 device. Before measurements,
the calcined sample was reduced in situ in flowing H2
(0.003mmol/s) at 350 �C for 3h. Afterwards, this sample
was outgassed under vacuum at 340 �C for 16h. A dou-ble adsorption method was used: (i) first adsorption iso-
therm was measured, which includes both physisorption
and chemisorption; (ii) after a 2h outgassing at 30 �C, asecond isotherm was measured, which includes physi-
sorption only. Both isotherms were determined in the
pressure range of 10�8–2 · 10�1kPa. The difference be-
tween first and second isotherms gave the CO chemi-
sorption isotherm. The latter theoretically exhibits ahorizontal linear region corresponding to the complete
coverage of metallic sites by a monolayer of adsorbate.
However the linear region of experimental chemisorp-
tion isotherms often exhibits a slope and the monolayer
uptake was obtained by back extrapolation of the linear
region to zero pressure [15].
The Pd/SiO2 cogelled xerogel catalysts were tested for
1,2-dichloroethane hydrodechlorination, which wasconducted in a stainless steel tubular reactor (internal
diameter: 10mm) at a pressure of 0.3MPa. The reactor
was placed in a convection oven. A constant flow of
each reactant was maintained by a Gilson piston pump
for ClCH2–CH2Cl and Brooks mass flow controllers for
H2 and He. The effluent was analyzed by gas chroma-
tography (ThermoFinnigan with FID) using a Porapak
Q5 packed column. Prior to each experiment, the Pd/SiO2 catalysts were reduced in situ at atmospheric pres-
sure in flowing H2 (0.023mmol/s) while being heated to
350 �C at a rate of 350 �C/h and were maintained at this
temperature for 3h. After reduction, the Pd/SiO2 cata-
lysts were cooled in H2 to the desired initial reaction
temperature of 300 �C. The total flow of the reactant
mixture was 0.45mmol/s and consisted of ClCH2–
CH2Cl (0.011mmol/s), H2 (0.023mmol/s), and He(0.42mmol/s). The temperature was successively kept
at 300 �C (10h to allow activity stabilization after fast
initial deactivation), 350 �C (2h), and 300 �C (2h). The
effluent was analyzed every 15min and eight analyses
were made at each temperature (40 for the first level).
For each catalytic experiment, the mass of catalyst pel-
lets, sieved between 250 and 500lm, was equal to
2 · 10�4kg. Thanks to the very open structure of Pd/SiO2 cogelled xerogel catalysts, reaction rate values ob-
tained in the present study were not falsified by diffu-
sional limitations [16]. Thus, these values reflect the
intrinsic chemical kinetics.
3. Results
3.1. Characterization of silylated acetylacetonate ligands
and palladium complexes
The IR spectrum of MS-acac-H (Fig. 1, scheme 1,
structure 2a) is given in the literature [13]. Broad signals
are observed due to the C–H elongation at 2969 and
2841cm�1 for MS-dPvM (Fig. 1, scheme 1, structure
2b), and 2942 and 2840cm�1 for MS-dBzM (Fig. 1,scheme 1, structure 2c). Phenyl groups in MS-dBzM
are responsible for the broad signal at 3056cm�1 (C–H
elongation). Characteristic absorptions of carbonyl
groups occur at 1713cm�1 for MS-dPvM, and 1698
and 1672cm�1 for MS-dBzM. Characteristic elonga-
tions of Si–OCH3 in MS-dPvM and MS-dBzM are pre-
sent at 1199 and 1087cm�1. The absorptions of these
functional groups occur at identical frequencies for pal-ladium complexes, while the carbonyl groups are more
sensitive to coordination by the metal and resonate at
1535cm�1 for Pd(MS-acac)2 (Fig. 1, scheme 2, structure
3a), 1575cm�1 for Pd(MS-dPvM)2 (Fig. 1, scheme 2,
structure 3b), and 1595cm�1 for Pd(MS-dBzM)2 (Fig.
1, scheme 2, structure 3c).1H and 13C NMR shift assignments of MS-acac-H
(Fig. 1, scheme 1, structure 2a) are reported in the liter-ature [13]. The 1H NMR spectra of MS-dPvM (Fig. 1,
scheme 1, structure 2b) and MS-dBzM (Fig. 1, scheme
1, structure 2c) show no major difference for the unsub-
stituted dione and the (trimethoxysilyl)propyl-substi-
tuted moiety. Only the two singlets corresponding to
the two methyl groups in MS-acac-H are replaced by
either one singlet at 0.92ppm for MS-dPvM (18H; t-bu-
tyl groups) or one multiplet at 7–8ppm for MS-dBzM(10H; phenyl groups). The 1H NMR spectra of palla-
dium complexes (Fig. 1, scheme 2, structures 3a, 3b
and 3c) and of the respective free ligands are quite sim-
ilar, although the peaks are slightly broader and the tri-
plets at about 4ppm are missing in the three palladium
complexes, indicating that deprotonation in a-positionhas occurred to produce the metal complexes.
In the 13C NMR spectra, the absorption due to themethyl groups in MS-acac-H (Fig. 1, scheme 1, structure
2a) is replaced by absorptions at 27.5 and 44.2ppm (t-
butyl groups) for MS-dPvM (Fig. 1, scheme 1, structure
2b), or 129.4, 129.8, 134.4 and 137ppm (phenyl groups)
for MS-dBzM (Fig. 1, scheme 1, structure 2c). Carbonyl
groups resonate at 210.4ppm in MS-dPvM and 197ppm
in MS-dBzM. Palladium complexes give similar 13C
spectra, with the difference that the a-carbon absorption(seen at about 51ppm for the free ligands) is shifted
downfield to 110ppm for Pd(MS-acac)2 (Fig. 1, scheme
114 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
2, structure 3a), to 121ppm for Pd(MS-dPvM)2 (Fig. 1,
scheme 2, structure 3b), and to 145ppm for Pd(MS-
dBzM)2 (Fig. 1, scheme 2, structure 3c), while C–O
absorption is shifted to 186ppm for Pd(MS-acac)2, to
160ppm for Pd(MS-dPvM)2, and to 185.5ppm for
Pd(MS-dBzM)2.
3.2. Characterization of xerogels and Pd/SiO2 cogelled
xerogel catalysts
For all samples, except Pd(MS-dBzM)2TE, homoge-
neous gels are obtained after about 7h in the oven at
the temperature of 80 �C. No gel is obtained with the
sample Pd(MS-dBzM)2TE because a black suspensionappears when the complex Pd(MS-dBzM)2 is dissolved
in the mixture of 10ml of ethanol and 10ml of THF.
The actual palladium loading in cogelled catalysts is
generally higher than theoretical loading because the ac-
tual catalyst mass after drying, calcination and reduc-
tion steps, is lower than theoretical mass. In fact, some
TEOS often remains unreacted and is volatilized during
vacuum drying. This theoretical mass (mth) is calculatedfrom Eq. (1):
mth ¼ nmMm þ ðnTEOS þ nligand þ 2nmÞMSiO2; ð1Þ
where nm is the amount of palladium complex in the gel
(mmol); Mm is the palladium atomic weight; nTEOS and
nligand are respectively the amounts of TEOS and syli-
lated acetylacetonate ligand in the gel (mmol); MSiO2is
the molecular weight of SiO2, 60.085g/mol. In this equa-
tion, it is assumed that all TEOS and ligand moleculesare converted into SiO2. Results in Table 1 shows that
there are only weight losses for samples Pd(MS-
acac)2TE and Pd(MS-acac)2AE after vacuum drying,
calcination and reduction steps.
The results of textural properties are presented in Ta-
ble 2 both for samples after vacuum drying step (the
Table 2
Sample textural properties
Sample Pt (MPa) Vv (
Pd(MS-acac)2TE dried 10 3.6
Pd(MS-acac)2TE 8 4.2
Pd(MS-acac)2AE 16 3.7
Pd(MS-dPvM)2TE dried –b 1.6
Pd(MS-dPvM)2TE 170 2.1
MS-acac-H dried –b 1.6
MS-acac-H 165 2.1
MS-dPvM dried –b 1.3
MS-dPvM –b 1.9
MS-dBzM dried 24 2.4
MS-dBzM 16 3.1
Pt: pressure of change of mechanism during mercury porosimetry; Vv: total c
BET method; dSiO2: silica particle diameter measured by TEM.
a Not measurable.b Not applicable, only collapse.
sample names are followed by the word �dried� in Table
2) and for the same samples after vacuum drying, calci-
nation, and reduction steps.
Submitted to an increasing mercury pressure, the
samples Pd(MS-acac)2TE, Pd(MS-acac)2TE dried,
Pd(MS-acac)2AE, Pd(MS-dPvM)2TE, MS-acac-H,MS-dBzM and MS-dBzM dried exhibit two successive
behaviors [17,18]. The samples Pd(MS-acac)2TE and
MS-dBzM are shown as examples in Fig. 2(a) and (b),
respectively. At low pressure, the sample collapses under
the isostatic pressure, and above a pressure of transition
(Pt), which is characteristic of the material composition
and microstructure, mercury can enter the network of
small pores not destroyed during compression at lowpressure. This appears in the pressurization curve by a
sudden change of slope at Pt on the curve of pressure in-
crease. The curve of pressure decrease can also be di-
vided into two distinct parts separated by an abrupt
change of slope. At high pressure, there is large volume
variation reversibility with a hysteresis, whereas at low
pressure, there is only a very limited volume variation
indicating the mainly irreversible nature of the crushingphenomenon that occurred during pressurization. Two
models can then be used in order to calculate the pore
size distribution from mercury porosimetry: the mercury
intrusion in small pores above Pt described by Wash-
burn�s model [19] and the collapse of larger pores below
Pt described by Pirard�s model [17,18].
Submitted to an increasing mercury pressure, the
samples Pd(MS-dPvM)2TE dried, MS-acac-H dried,MS-dPvM and MS-dPvM dried only collapses under
mercury pressure, as observed on aerogels [18,20] and
xerogels for which the silylated amine ligand/TEOS ra-
tio is particularly high [17,21]. The samples Pd(MS-
dPvM)2TE dried and MS-acac-H dried are shown as
examples in Fig. 2(a) and (b), respectively. Furthermore,
these samples do not contain any trace of trapped mer-
cm3/g) SBET (m2/g) dSiO2(nm)
66 –a
415 16.9 ± 1.6
700 14.5 ± 1.3
491 –a
820 11.2 ± 1.0
471 –a
890 15.0 ± 1.2
624 –a
1085 8.7 ± 0.9
432 –a
825 37.2 ± 5.8
umulative specific pore volume; SBET: specific surface area obtained by
0
1
2
3
4
5
0.01 0.1 1 10 100 1000
Pressure (MPa)
VH
g(c
m3 /g
)
(a) Pd/SiO2 xerogel catalysts
0
1
2
3
4
5
0.01 0.1 1 10 100 1000
Pressure (MPa)
VH
g(c
m3 /g
)
(b) SiO2 xerogels
Fig. 2. Mercury porosimetry curves (volume variation as a function of
mercury pressure) of (a) samples Pd(MS-acac)2TE (m and n) and
Pd(MS-dPvM)2TE dried (j and h), of (b) samples MS-acac-H dried
(m and n) and MS-dBzM (� and �). The dark points represent the
curve of pressure increase and the empty points represent the curve of
pressure decrease.
0
200
400
600
800
1000
0
200
400
600
800
1000
0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
p/p0
VN
2(c
m3 /g
)
(a) Pd/SiO2 xerogel catalysts
p/p0
VN
2(c
m3 /g
) (b) SiO2 xerogels
Fig. 3. Nitrogen adsorption–desorption isotherms of (a) samples
Pd(MS-acac)2TE (m and n) and Pd(MS-dPvM)2TE dried (j and
h), of (b) samples MS-acac-H dried (m and n) and MS-dBzM (� and
�).The dark points represent the curve of pressure increase and the
empty points represent the curve of pressure decrease.
S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120 115
cury after a mercury porosimetry experiment. The entire
pore size distribution is determined by Pirard�s collapsemodel.
In Table 2, Pt evolves very strongly with the nature ofthe silylated acetylacetonate ligands and palladium com-
plexes. Furthermore, for each sample, Pt value is always
higher before calcination and reduction steps (dried
samples) than after.
The nitrogen adsorption analysis reveals the presence
of two types of isotherms. The a-type nitrogen adsorp-
tion–desorption isotherm is characterized by: (i) at low
relative pressure, a sharp increase of the adsorbed vol-ume is followed by a plateau which corresponds to type
I isotherm according to BDDT classification [19], which
is characteristic of microporous adsorbents; (ii) at high
pressure, the adsorbed volume increases quickly, like
in type II isotherm, which is characteristic of macropor-
ous adsorbents. The volume adsorbed between p/
p0 = 0.95 and 1 is large, indicating that the sample con-
tains pores of large dimensions; (iii) all samples exhibit anarrow adsorption–desorption hysteresis loop for p/p0values close to 1, and this hysteresis is characteristic of
capillary condensation in large mesopores. The b-typenitrogen adsorption–desorption isotherms have the fol-
lowing characteristics: a nitrogen adsorption–desorptiontype IV isotherm according to the BDDT classification
[19] with a broad hysteresis. This observation could be
explained by the absence of macropores and very large
mesopores and, therefore an increase of the specific sur-
face area, SBET.
The a-type nitrogen adsorption–desorption isotherm
is observed for samples Pd(MS-acac)2TE, Pd(MS-
acac)2TE dried, Pd(MS-acac)2AE, MS-dBzM andMS-dBzM dried. The samples Pd(MS-acac)2TE and
MS-dBzM are shown as examples in Fig. 2(a) and (b),
respectively. The b-type is characterized for samples
Pd(MS-dPvM)2TE, Pd(MS-dPvM)2TE dried, MS-acac-
H, MS-acac-H dried, MS-dPvM and MS-dPvM dried.
The samples Pd(MS-dPvM)2TE dried and MS-acac-H
dried are shown as examples in Fig. 2(a) and (b), respec-
tively. When each dried sample is calcined and reduced,the porosity in samples increases because at low relative
pressure, the plateau corresponding to micropores
strongly increases and the hysteresis loop shifts towards
smaller p/p0 values, which is characteristic of capillary
condensation in smaller mesopores. In Table 2, the spe-
cific surface area, SBET strongly increases when the dried
samples are calcined and reduced. Furthermore, SBET
have smaller values for Pd/SiO2 cogelled xerogel cata-
0.001
0.01
0.1
1
10
0.1 1 10 100 1000Pore size (nm)
0.1 1 10 100 1000
Pore size (nm)
Cum
ulat
ive
volu
me
(cm
3 /g)
(a) Pd/SiO2 xerogel catalysts
(b) SiO2 xerogels
0.001
0.01
0.1
1
10
Cum
ulat
ive
volu
me
(cm
3 /g)
Fig. 4. Pore size distributions of (a) samples Pd(MS-acac)2TE (m),
Pd(MS-acac)2TE dried (n), Pd(MS-acac)2AE (d), Pd(MS-dPvM)2TE
(j), Pd(MS-dPvM)2TE dried (h) and Pd(EDAS)2 (·) [18], of (b)
samples MS-acac-H (m), MS-acac-H dried (n), MS-dPvM (j), MS-
dPvM dried (h), MS-dBzM (�), MS-dBzM dried (�) and EDAS (·)[16].
116 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
lysts than SiO2 xerogels synthesized from the same silyl-
ated ligand.
Fig. 4 shows the evolution of the cumulative volume
distributions over the entire pore size range for SiO2
xerogels and Pd/SiO2 cogelled xerogel catalysts. Thesecurves were obtained by applying a combination of var-
ious methods to their respective validity domains and by
adding the porous volume distributions corresponding
to each domain [8,17]. All samples are characterized
by a steep volume increase around 0.8nm, followed by
a plateau (Fig. 4). In the range of meso- and macro-
pores, one observes that all samples exhibit a broad dis-
tribution. These samples then contain micropores(width < 2nm), mesopores (2nm < width < 50nm) and
macropores (width > 50nm).
The total cumulative specific pore volume, Vv that is,
the pore volume obtained by addition of pore volume
measured by mercury porosimetry (width > 7.5nm)
and cumulative volume of micropores and mesopores
of widths between 2 and 7.5nm measured by nitrogen
adsorption–desorption, increases when dried SiO2 xero-gels are calcined and when Pd/SiO2 cogelled xerogel cat-
alysts are calcined and reduced (Table 2).
Texture and morphology of SiO2 xerogels and Pd/
SiO2 cogelled xerogel catalysts have also been examined
by TEM and the sizes of SiO2 elementary particles, dSiO2
have been evaluated. Sizes given in Table 2 represent the
arithmetic mean on 50 particles. Samples MS-acac-H,
MS-dPvM and MS-dBzM are shown in Fig. 5(a)–(c),
and it is observed that the silica particle size, dSiO2
strongly varies when the nature of silylated acetylaceto-nate ligand changes. Furthermore, it has been shown in
[17] that the relation between the pressure of transition,
Pt and the size of SiO2 particles, dSiO2is given by the
relationship dSiO2� 1=P 0:75
t in the case of the buckling
of the brittle filaments of mineral oxide under an axial
compressive stress. In all samples, when dSiO2increases,
Pt decreases and inversely (Table 2).
Table 3 gives palladium particle size determined byTEM and CO chemisorption measurements and cata-
lytic activity for 1,2-dichloroethane hydrodechlorination
over Pd/SiO2 cogelled xerogel catalysts.
TEM analysis indicates that Pd(MS-acac)2TE and
Pd(MS-dPvM)2TE exhibit metal particles distributed
in two families of different sizes: small crystallites of
about 3nm and large crystallites between 20 and 50nm
(Fig. 5(d)). For both catalysts, the mean diameter ofsmall crystallites, dTEM1 is the arithmetic mean of 50
diameters of small palladium particles measured on
TEM micrographs. The number of larger crystallites is
much smaller than for small crystallites, and their mean
diameter, dTEM2 is then estimated from an average of fif-
teen crystallites. Concerning localization of palladium
crystallites, it appears that cogelled catalysts are com-
posed of silica particles arranged in strings or aggre-gates, and although TEM gives only a 2D view, it
seems that small palladium particles are located inside
silica particles, whereas large palladium particles are lo-
cated at their surface (Fig. 5(d)). The sample Pd(MS-
acac)2AE exhibits a broad distribution of palladium par-
ticle sizes, from 2 to 40nm.
A mean diameter, dchem of palladium particles has
also been derived from metal dispersion, D, measuredby CO chemisorption for Pd/SiO2 cogelled xerogel cata-
lysts as proposed by Lambert et al. [9–11]. For Pd(MS-
acac)2TE and Pd(MS-dPvM)2TE containing two fami-
lies of particle sizes, values of dchem are between values
of dTEM1 and dTEM2 For Pd(MS-acac)2AE exhibiting a
broad distribution of palladium particle sizes, dchem rep-
resents a mean diameter of all palladium particles in this
sample.Pd(MS-acac)2TE, Pd(MS-acac)2AE and Pd(MS-
dPvM)2TE mainly produce ethane, C2H6, and two sec-
ondary products are observed: ethyl chloride, C2H5Cl,
and ethylene, C2H4 as in the case of Pd/SiO2 cogelled
xerogel catalysts synthesized with sylilated amine ligands
[9,10]. Indeed, noble metals catalysts (Group VIII), and
particularly palladium, are very active for the hydrode-
chlorination reaction [22,23]. In the case of 1,2-dichloro-ethane hydrodechlorination, the noble metal participates
in a catalytic cycle, in which the reactant is dechlorinated
Fig. 5. TEM micrographs of (a) MS-acac-H (500000·), of (b) MS-dPvM (500000·), of (c) MS-dBzM (500000·) and of (d) Pd(MS-acac)2TE
(725000·).
Table 3
Palladium average particle size and catalytic activity for 1,2-dichloroethane hydrodechlorination
Sample TEM CO chemisorption r (mmol/skgPd) TOF (s�1)
dTEM1 (nm) rTEM1 (nm) dTEM2 (nm) rTEM2 (nm) dchem (nm) D (%) 250�C 300�C 250�C 300�C
Pd(MS-acac)2TE 3.2 0.6 22.5 7.0 6.5 16 71 169 0.05 0.11
Pd(MS-acac)2AE From 2 to 40 13.5 8 27 85 0.04 0.11
Pd(MS-dPvM)2TE 3.7 0.6 46.2 11.0 9.0 12 48 112 0.04 0.10
dTEM1, dTEM2: mean diameter of small and large palladium particles respectively measured by TEM; rTEM1, rTEM2: standard deviations associated
with dTEM1 and dTEM2, respectively; D: metal dispersion measured by chemisorption; dchem: mean diameter derived from dispersion D; r: con-
sumption rate of 1,2-dichloroethane; TOF: turnover frequency.
S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120 117
by chlorination of the metal surface, which is then itself
dechlorinated by reduction with hydrogen. Because of
the high reactivity of hydrogen on noble metals, the
dechlorinated organics, C2H4 in the present case, is
immediately converted into the fully hydrogenated prod-
uct, C2H6 [9,10,22,23], which is much less useful from an
industrial point of view. However, several authors dem-
onstrated the ability of bimetallic catalysts, composed ofalloys such as Pd–Ag [8,11], Pt–Cu [24,25], Pd–Cu [11],
to convert chlorinated alkanes selectively into less or
not chlorinated alkenes. So in a previous work over
Pd–Ag/SiO2 and Pd–Ag/SiO2 cogelled xerogel catalysts
[11], increasing silver or copper content in bimetallic cat-
alysts results in an increase in ethylene selectivity, and for
samples with 1.5wt% for palladium and 3.0wt% for sil-
ver or copper, this selectivity reaches 100% in the condi-tions of the catalytic test.
The consumption rate of 1,2-dichloroethane r is cal-
culated from chromatographic measurements of C2H6,
C2H5Cl and C2H4 concentrations in the reactor effluent
and from the differential reactor equation that is written
as follows:
r ¼ F A þ F Cl þ F E
WðF A0; F Cl0 and F E0 ¼ 0Þ; ð2Þ
where r is the consumption rate (mmol/kgPds), FA is the
molar flowrate of ethane at the reactor outlet (mmol/s),
FA0 is the molar flowrate of ethane at the reactor inlet
(mmol/s), FCl is the molar flowrate of ethyl chloride atthe reactor outlet (mmol/s), FCl0 is the molar flowrate
of ethyl chloride at the reactor inlet (mmol/s), FE is
the molar flowrate of ethylene at the reactor outlet
(mmol/s), FE0 is the molar flowrate of ethylene at the
reactor inlet (mmol/s) and W is the palladium mass
118 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
inside the reactor (kgPd). For all samples, r has been cal-
culated from Eq. (2) at the temperatures of 250 and
300 �C and these results are presented in Table 3. It is
observed that r increases with palladium dispersion.
4. Discussion
An important advantage of the cogelation method is
the possibility of homogeneously distributing the cata-
lytic metal through the whole material, that is, inside
the silica particles. In previous studies [7–11,21], TEM
micrographs indeed seemed to exhibit metal particles in-
side the silica matrix. It was shown that the localizationof the metal inside the silica matrix was induced by a
nucleation process initiated by the silylated amine ligand
of the metal, 3-(2-aminoethylamino)propyltrimethoxysi-
lane (EDAS). For example, when palladium is intro-
duced in xerogels synthesized from EDAS, the
PdðEDASÞ2þ2 complexes are much more reactive than
TEOS. Therefore, hydrolyzed PdðEDASÞ2þ2 complexes
can act as a nucleation agent leading to silica particleswith a hydrolyzed and condensed PdðEDASÞ2þ2 core
and a shell principally made of hydrolyzed and con-
densed TEOS [7,9,10,21]. Furthermore, in the case of
SiO2 xerogels [21], when the silylated ligand contains
methoxy groups, the nucleation phenomenon by the sil-
ylated ligand takes place in the case of TEOS as the
main silica precursor.
Samples MS-acac-H, MS-dPvM and MS-dBzM pre-sent small homogeneous entities visible by TEM (Fig.
5(a)–(c)), which seem to be silica elementary particles.
These observations show that the nucleation phenome-
non by the silylated acetylacetonate ligand takes place
as in the case of SiO2 xerogels synthesized from EDAS
[21]. On the one hand, TEM micrographs obtained for
Pd(MS-acac)2TE and Pd(MS-dPvM)2TE show small
palladium crystallites inside microporous silica particles(Fig. 5(d)). So the nucleation phenomenon is also ob-
served when palladium is complexed either by MS-
acac-H or by MS-dPvM in a THF-ethanol mixture.
Pd(MS-acac)2 and Pd(MS-dPvM)2 complexes in a
THF-ethanol mixture are more reactive than TEOS
and allow a nucleation process initiated by the silylated
acetylacetonate ligand. On the other hand, the sample
Pd(MS-acac)2AE exhibits a broad distribution of palla-dium particle sizes from 2 to 40nm and no spherical
structure of silica particles is observed. These observa-
tions could be explained by the fact that the use of ace-
tone to dissolve Pd(MS-acac)2 complex in ethanol does
not allow preserving the metal complex architecture dur-
ing sol–gel immobilization and prevents the nucleation
phenomenon by the silylated acetylacetonate ligand to
take place [26].Thanks to the large difference in steric volume of the
three silylated acetylacetonate ligands, the textural prop-
erties of SiO2 xerogels and Pd/SiO2 cogelled xerogel cat-
alysts show a large influence of the nature of silylated
acetylacetonate ligand and metal complex on the final
texture of xerogels and catalysts, both before as well
as after calcination and reduction steps.
In the case of SiO2 xerogels, the silylated acetylaceto-nate ligand/TEOS molar ratio is equal to about 0.06 and
the samples MS-acac-H, MS-dPvM and MS-dBzM are
compared with a SiO2 xerogel synthesized from EDAS
and TEOS and in which the EDAS/TEOS molar ratio
is also equal to 0.06 [21]. The shape of the adsorption
isotherms of the samples MS-dBzM (Fig. 3(b)) and
EDAS [21] is of type II at high relative pressure. During
mercury porosimetry measurements, both samples alsopresent a buckling-intrusion mechanism (Fig. 2(b)),
characteristic features of macroporous solids [19]. These
samples, MS-dBzM and EDAS, therefore have high val-
ues for the total specific pore volume, Vv equal to
3.1cm3/g (Table 2) and 2.3cm3/g [21], respectively. The
shape of the adsorption isotherms of samples MS-
acac-H and MS-dPvM is of type IV at high relative pres-
sure (Fig. 3(b), where MS-acac-H dried presents thesame isotherm shape) and only a collapse mechanism
is observed for both samples during mercury porosime-
try measurements (Fig. 2(b), where MS-acac-H dried
presents the same porosimetry curve). All these observa-
tions are characteristic features of mesoporous solids
[19], which therefore present smaller values of Vv than
MS-dBzM and EDAS (Table 2). So the nature of the sil-
ylated ligand influences xerogel textural properties andby changing the nature of the silylated ligand, e.g.,
EDAS, MS-acac-H, MS-dPvM or MS-dBzM, textural
properties such as pore volume, pore size and surface
area can be tailored. The same development can be per-
formed for Pd/SiO2 cogelled xerogel catalysts in which
the silylated acetylacetonate ligand/TEOS molar ratio
is equal to about 0.05. The samples Pd(MS-acac)2TE,
Pd(MS-acac)2AE and Pd(MS-dPvM)2TE are comparedwith a Pd/SiO2 catalyst synthesized from EDAS and
TEOS in which palladium loading is equal to 1.5%
and the EDAS/TEOS molar ratio is equal to 0.05 [10].
The shape of the adsorption isotherms of the samples
Pd(MS-acac)2TE (Fig. 3(a)), Pd(MS-acac)2AE and
Pd(EDAS)2 is of type II at high relative pressure, and
during mercury porosimetry measurements, all these
samples present a buckling-intrusion mechanism (Fig.2(a)), characteristic features of macroporous solids.
These samples, Pd(MS-acac)2TE, Pd(MS-acac)2AE
and Pd(EDAS)2, therefore have high values for the total
specific pore volume, Vv equal to 4.2cm3/g, 3.7cm3/g
(Table 2) and 3.9cm3/g (10), respectively. The shape of
the adsorption isotherm of sample Pd(MS-dPvM)2TE
is of type IV at high relative pressure (Fig. 3(a), where
Pd(MS-dPvM)2TE dried presents the same isothermshape) and a collapse mechanism is only observed for
this sample during mercury porosimetry measurements
S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120 119
(Fig. 2(a), where Pd(MS-dPvM)2TE dried presents the
same porosimetry curve). All these observations are
characteristic features of a mesoporous solid, which
therefore presents smaller value of Vv than Pd(MS-
acac)2TE, Pd(MS-acac)2AE (Table 2) and Pd(EDAS)2[10]. So the nature of silylated ligand also influences tex-tural properties of Pd/SiO2 cogelled xerogel catalysts.
The specific surface area obtained by BET method,
SBET presents a higher value equal to 1085m2/g for the
sample MS-dPvM than those of samples MS-acac-H,
MS-dBzM (Table 2) and EDAS [21]. Furthermore, in
Fig. 4(b), the cumulative volume associated to micro-
pores (pore size < 2nm) is more important for sample
MS-dPvM than for samples MS-acac-H, MS-dBzMand EDAS. For Pd/SiO2 cogelled xerogel catalysts, SBET
value of Pd(MS-dPvM)2TE (Table 2) and the cumulative
volume associated to micropores (Fig. 4(a)) are much
higher than those of Pd(MS-acac)2TE, Pd(MS-acac)2AE
and Pd(EDAS)2 [10]. As it is well-known that micropores
(pore size < 2nm) and small mesopores (2nm < pore
size < 10nm) contribute essentially in SBET values [19]
and the silica particle sizes of SiO2 xerogels and Pd/SiO2 cogelled xerogel catalysts are equal from 8.7 to
37.2nm (Table 2), microporosity is largely located inside
silica particles. So samples MS-dPvM and Pd(MS-
dPvM)2TE are more microporous than all other samples.
This higher microporosity with the use of silylated dipi-
valoylmethane ligand could be the result of its size of
about 1nm against sizes of about 0.6nm for the other syl-
ilated acetylacetonate ligands and of about 0.35nm forEDAS (Molecular simulation with Chem3D Ultra Ver-
sion 7.0). So greater repulsion between silylated dipiva-
loylmethane ligand molecules could generate more
spaces into silica nuclei, increasing the microporosity in-
side silica particles. The silica particle sizes are also smal-
ler for samples MS-dPvM and Pd(MS-dPvM)2 (Table 2).
This feature could also be the result of repulsion between
silylated dipivaloylmethane ligand molecules, involving asmall number of ligand molecules in silica nuclei.
Before calcination and reduction steps, all the sam-
ples already have high specific surface areas, SBET (Ta-
ble 2). When the samples are calcined in air, the
organic groups are removed by thermolysis and/or pyro-
lysis reactions. This results in an increase in porosity and
specific surface area [27,28]. For all samples, after calci-
nation and reduction steps, the total cumulative porevolume, Vv slightly increases, whereas the specific sur-
face area, SBET strongly increases, indicating that a
large percentage of micropores and small mesopores
are created by the removal of the organic groups dur-
ing calcination and reduction steps (Table 2). Moreover,
the shape of the isotherms of dried samples is not chan-
ged during calcination and reduction steps. Only micro-
pores are additionally formed, whereas the meso- andmacropores are retained after calcination and reduction
steps.
Accessibility to active sites is of crucial importance
for an efficient catalytic activity. Site accessibility will
naturally depend on the material structure, which is
dependent on the composition of the material and on
processing conditions. It is frequently found that the
specific activities of metal complexes in catalytic reac-tions are considerably diminished when the complexes
are supported on a silica gel material [26]. Such reduc-
tions in activity may arise from changes in the coordina-
tion state of the complex or other environmental
influences. In many cases, however, the reduced activity
may simply result from a large fraction of supported me-
tal complexes being rendered inaccessible to solvent or
external reagents. Because palladium is located insidesilica particles in Pd/SiO2 cogelled xerogel catalysts,
there is a risk of inaccessibility. The values of Vv in Ta-
ble 2 show that porosity remains high after drying under
vacuum, although this does not prove accessibility to
palladium. Nevertheless, it is observed in Table 3 that
dTEM1 6 dchem 6 dTEM2 for Pd(MS-acac)2TE and Pd
(MS-dPvM)2TE. So palladium particle sizes obtained
by CO chemisorption measurements are related toTEM measurements, which suggests that palladium par-
ticles located inside silica particles are completely
accessible.
In Table 3, the specific consumption rate of 1,2-
dichloroethane, r, increases with palladium dispersion
and the turnover frequency (TOF), that is, the number
of molecules consumed per surface metal atom and
per second, is equal to about 0.04s�1 at 250 �C and toabout 0.11s�1 at 300 �C. These values are identical to
those obtained for 1,2-dichloroethane hydrodechlorina-
tion over Pd/SiO2 cogelled xerogel catalysts synthesized
with EDAS as silylated ligand [9,10]. Although the
structure sensitivity of C–Cl hydrogenolysis with the
ensemble size concept has been pointed out by several
authors [22,29,30], it seems that the 1,2-dichloroethane
hydrodechlorination over Pd/SiO2 catalysts examinedin the present study, is a structure insensitive reaction,
as already previously observed [9,10].
5. Conclusions
The main purpose of the present work was to synthe-
size new silylated acetylacetonate ligands, able to form achelate with a metal ion such as Pd2+ and able to react
with a silica-forming reagent such as TEOS. This
purpose is attained because the use of 3-[3-(trimethoxy-
silyl)propyl]-2,4-pentanedione (MS-acac-H), or 2,2,6,
6-tetramethyl-4-[3-(trimethoxysilyl)propyl]-3,5-heptan-
edione (MS-dPvM), or 1,3-diphenyl-2-[3-(trimethoxysi-
lyl) propyl]-1,3-propanedione (MS-dBzM) in an
ethanolic solution containing tetraethoxysilane (TEOS)and an ammonia solution of 0.18mol/l gave very homo-
geneous colored xerogels, the textural properties of
120 S. Lambert et al. / Journal of Non-Crystalline Solids 343 (2004) 109–120
which depend on the nature of the silylated acetylaceto-
nate ligand.
No acetylacetonate deritative has ever been used as a
ligand for the synthesis of metal xerogel catalysts. Nev-
ertheless, in this study, the use of silylated palladium
complexes of the same silylated acetylacetonate ligands,in a mixture of tetrahydrofurane (THF) and ethanol
containing tetraethoxysilane (TEOS) and an ammonia
solution of 0.18mol/l yielded Pd/SiO2 cogelled xerogel
catalysts. These samples contain small metal particles
of about 3.5nm located inside silica particles exhibiting
a monodisperse microporous distribution and some
large metal particles from 20 to 50nm situated outside
the silica matrix.The silylated acetylacetonate ligand has a large influ-
ence on the textural properties of xerogels and catalysts,
both before and after calcination and reduction steps.
By changing the nature of the silylated ligand, e.g.,
MS-acac-H, MS-dPvM or MS-dBzM, textural proper-
ties such as pore volume, pore size and surface area
can be tailored. For all the samples, the total cumulative
pore volume, Vv slightly increases after calcination andreduction steps, whereas the specific surface area, SBET
strongly increases, indicating that a large percentage of
micropores and small mesopores are created by the re-
moval of the organic groups during calcination and
reduction steps.
Although palladium particles are located inside the
silica particles, their complete accessibility, via the micro-
pore network, is suggested. 1,2-dichloroethane hydrode-chlorination over Pd/SiO2 catalysts mainly produces
ethane, and the hydrodechlorination activity increases
with palladium dispersion. Hydrodechlorination over
Pd/SiO2 xerogel cogelled catalysts is a structure insensi-
tive reaction compared to the ensemble size concept.
Acknowledgments
S.L. is grateful to the Belgian Fonds pour la Forma-
tion a la Recherche dans l�Industrie et dans l�Agricul-
ture, F.R.I.A., for a PhD grant. The authors also
thank the Belgian Fonds National de la Recherche Sci-
entifique, the Fonds de Bay, the Fonds de la Recherche
Fondamentale et Collective, the Ministere de la Region
Wallonne and the Ministere de la Communaute Franc-aise (Action de Recherche Concertee n� 00-05-265) for
their financial support.
References
[1] U. Deschler, P. Kleinschmit, P. Panster, Angew. Chem. Int. Ed.
Engl. 25 (1986) 236.
[2] T. Lopez, A. Romero, R. Gomez, J. Non-Cryst. Solids 127 (1991)
105.
[3] R.D. Gonzalez, T. Lopez, R. Gomez, Catal. Today 35 (1997)
293.
[4] B. Breitscheidel, J. Zieder, U. Schubert, Chem. Mater. 3 (1991)
559.
[5] U. Schubert, New J. Chem. 18 (1994) 1049.
[6] A.D. Ward, E.I. Ko, J. Catal. 157 (1995) 321.
[7] B. Heinrichs, F. Noville, F.J.-P. Pirard, J. Catal. 170 (1997)
366.
[8] B. Heinrichs, P. Delhez, J.-P. Schoebrechts, J.-P. Pirard, J. Catal.
172 (1997) 322.
[9] S. Lambert, C. Cellier, P. Grange, J.-P. Pirard, B. Heinrichs, J.
Catal. 221 (2004) 335.
[10] S. Lambert, J.-F. Polard, J.-P. Pirard, B. Heinrichs, Appl. Catal.
B Env. 50 (2004) 127.
[11] S. Lambert, B. Heinrichs, A. Brasseur, A. Rulmont, J.-P. Pirard,
Appl. Catal. A General 270 (2004) 201.
[12] R.G. Charles, Org. Synth. 39 (1965) 61.
[13] W. Urbaniak, U. Schubert, Liebigs Ann. Chem. (1991) 1221.
[14] C.J. Brinker, G.W. Scherer, Sol–Gel Science: The Physics and
Chemistry of Sol–Gel Processing, Academic, San Diego, 1990.
[15] G. Bergeret, P. Gallezot, in: G. Ertl, H. Knozinger, J. Weitkamp
(Eds.), Handbook of Heterogeneous Catalysis, vol. 2, Wiley-
VCH, Weinheim, 1997, p. 439.
[16] B. Heinrichs, J.-P. Schoebrechts, J.-P. Pirard, AIChE J. 47 (2001)
1866.
[17] C. Alie, R. Pirard, A.J. Lecloux, J.-P. Pirard, J. Non-Cryst. Solids
285 (2001) 135.
[18] R. Pirard, S. Blacher, F. Brouers, J.-P. Pirard, J. Mater. Res. 10
(1995) 2114.
[19] A.J. Lecloux, in: J.R. Anderson, M. Boudart (Eds.), Catalysis:
Science and Technology, vol. 2, Springer, Berlin, 1981, p. 171.
[20] L. Duffours, T. Woignier, J. Phalippou, J. Non-Cryst. Solids 194
(1996) 283.
[21] S. Lambert, C. Alie, J.-P. Pirard, B. Heinrichs, J. Non-Cryst.
Solids 342 (2004) 70.
[22] B. Coq, G. Ferrat, F. Figueras, J. Catal. 101 (1986) 434.
[23] J.W. Bozzelli, Y.-M. Chen, S.S.C. Chuang, Chem. Eng. Comm.
115 (1992) 1.
[24] L.S. Vadlamannati, V.I. Kovalchuk, J.L. d�Itri, Catal. Lett. 58(1999) 173.
[25] V.Y. Borovkov, D.R. Luebke, V.I. Kovalchuk, J.L. d�Itri, J.
Phys. Chem. B 107 (2003) 5568.
[26] S.P. Watton, C.M. Taylor, G.M. Kloster, S.C. Bowman, in: K.D.
Karlin (Ed.), Progress in Inorganic Chemistry, vol. 51, Wiley-
VCH, Weinheim, 2003, p. 333.
[27] C. Gorsmann, U. Schubert, J. Leyrer, E. Lox, Mater. Res. Soc.
Symp. Proc. 435 (1996) 625.
[28] N.K. Raman, M.T. Anderson, C.J. Brinker, Chem. Mater. 8
(1996) 1682.
[29] P. Fouilloux, G. Cordier, Y. Colleuille, Stud. Surf. Sci. Catal. 11
(1982) 369.
[30] Y. Soma-Noto, W.M.H. Sachtler, J. Catal. 32 (1974) 315.