Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136

Contents lists available at SciVerse ScienceDirect

Journal of Molecular Catalysis A: Chemical

journal homepage: www.elsevier.com/locate/molcata

Pt monometallic and bimetallic catalysts prepared by acid sol–gel method for

liquid phase reforming of bioglycerol
M. El Doukkali a,∗ , A. Iriondo a , J.F. Cambra a , L. Jalowiecki-Duhamel b , A.S. Mamede d ,
F. Dumeignil b,c , P.L. Arias a
a
University of the Basque Country (UPV/EHU), Engineering School, C/Alameda Urquijo s/n, 48013 Bilbao, Spain
b
Université des Sciences et Technologies de Lille, CNRS UMR8181, Unité de Catalyse et de Chimie du Solide (UCCS), F-59655 Villeneuve d’Ascq Cedex, France
c
Institut Universitaire de France, 10 Boulevard Saint-Michel, 75005 Paris, France
d
Ecole Nationale Supérieure de Chimie de Lille, CNRS UMR8181, Unité de Catalyse et de Chimie du Solide UCCS, F-59655 Villeneuve d’Ascq Cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: PtFe, PtCo, PtNi, Ni and Pt-based ␥-Al2 O3 catalytic systems were prepared using the acid sol–gel method
Received 3 October 2012 and characterized by appropriate techniques. Their catalytic activities in glycerol APR for H2 production
Received in revised form 1 December 2012 were investigated, and the most significant results will be fully discussed in this paper. The effect of
Accepted 9 December 2012
operating conditions, the influence of different non-noble incorporated metals, the interesting promoting
Available online 20 December 2012
effect between Pt and Ni and their dispersion effect on the catalytic performance have been the main focus
of this research. The analyses by TPR-H2 , XRD, XPS and TEM/EDX techniques confirmed high Pt and Ni
Keywords:
dispersions and the promoting effect between both elements. All this explains the best catalytic activity
Platinum–nickel
Acid sol–gel
of PtNiAl(SGA) catalyst, especially at moderate temperature/pressure reaction conditions. Highly dispersed
Promoting effect Pt/Ni and Pt spillover effects were the key for obtaining high H2 production. The Pt monometallic catalyst
Aqueous phase reforming recorded lower conversions but fairly stable activity, while the Ni-containing catalysts recorded higher
Glycerol conversions, but some initial deactivation until steady state was reached.
Hydrogen © 2012 Elsevier B.V. All rights reserved.

1. Introduction considerable attention because of its advantages: (i) low tempera-

Fossil fuels have been the main resources used to meet energy
demands all over the world. However, the use of large quantities of
fossil fuels is causing the depletion of their reservoirs, global warm-
ing and climate change [1], as well as damaging the environment
[2]. All these drawbacks in the use of fossil fuels have generated
increasing interest in new sustainable and renewable clean fuels.
In this sense, hydrogen is considered one of most promising future
clean energy vectors.
Over the past decade, hydrogen generation from biomass-
derived products and by-products has been considered one of
the best alternatives when compared to traditional methods of
hydrogen production, which use fossil fuels as raw materials. The
conversion processes of biomass-derived materials, by contrast,
present environmental benefits, as the CO2 produced is absorbed
during biomass growth, providing a neutral CO2 balance [3–5].
One of the processes for producing hydrogen from biomass-derived
products [6–8], namely, the aqueous phase catalytic reforming pro-
cess [9–11] developed by Dumesic and co-workers, has received
∗ Corresponding author. Tel.: +34 946 017 297; fax: +34 946 014 179.
E-mail address: mohamed.eldoukkali@ehu.es (M. El Doukkali).
ture, thereby reducing energy consumption because the feed needs
not to be vaporized; (ii) limited decomposition reactions; (iii) the
production of a gas stream rich in H2 and CO2 and very poor in
CO in a simple chemical reactor. This last characteristic of the gas
produced renders it suitable for use in fuel cells [5].
As reported in the literature [9–14], many aqueous solutions
of oxygenated compounds have been used as feeds in aqueous
phase reforming (APR). However, the aqueous solution of glycerol
is widely used due to the possibility of revalorizing this by-product
of biodiesel production through the transesterification of fats from
biomass, as only small amounts of this glycerol can be absorbed by
its traditional markets [15].
During the APR process, apart from the ideal reforming reaction
(C3 H8 O3 + 3H2 O → 7H2 + 3CO2 ) [2], other reactions take place, such
as water-gas-shift (WGS), methanation or Fischer–Tropsch ones to
produce alkanes, partial cracking of C C/C H/C O bonds, dehydro-
genation and dehydration/hydrogenation [9–13]. In the case of C O
bond cleavage, this reaction leads to diol and alkane production and
is favored by catalyst acidity [16]. Moreover, the dehydrogenation
and dehydration/hydrogenation of glycerol also generate diols [16].
The APR process therefore includes many reaction pathways that
are highly dependent on the catalyst used [6]. Thus, it is important
to develop an active catalyst for C C/C H bond cleavage and WGS

1381-1169/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.molcata.2012.12.006
126 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136
reaction in order to increase the selectivity toward H2 production,
while the parallel and consecutive reactions are either limited [5]
or the intermediate products they generate fully reformed.
According to the literature [17,18], catalyst activity is highly
influenced by the preparation method used. Conventional prepara-
tion methods, such as incipient wetness impregnation, also called
wet impregnation, commonly produce active phase agglomeration
on the support used. This phenomenon, which is attributed to large-
sized particle formation, occurs especially when high contents of
metal precursors are incorporated [17,18]. However, the sol–gel
method presents some advantages when compared with impreg-
nation methods, such as higher active metal dispersion, great
thermal resistance to sintering and low deactivation rate [17–23],
thus can lead to more stable catalysts [17]. Moreover, this method
allows a better control of catalyst preparation via various parame-
ters, such as: temperature, type and concentration of the alkoxide,
the solvent, and the pH of the preparation medium. Therefore, vary-
ing these factors allows a more flexible control of the catalysts
textural properties, such as surface area, pore size distribution and
metal dispersion [23].
A variety of catalytic systems have been described in the
literature for aqueous phase reforming of glycerol and other
biomass-derived products [1,2,5,6,9–13,15,24–26]. These catalytic
systems are based on Pt, Ni, Cu, Pd or Co monometallic catalysts
and PtRe and PtNi bimetallic catalysts. Pt metal phase is usually
used in both monometallic and bimetallic catalysts because of
its high activity for C C/C H bond cleavage and WGS reaction,
and low selectivity toward methane production via the metha-
nation reaction [4]. According to Huber et al. [9], Pt/Al2 O3 SiO2
catalyst recorded higher activity in the APR of ethylene glycol
than other silica-supported Group VIII metal catalysts. A similar
conclusion was reached by Iriondo et al. [15], whereby both glyc-
erol conversion and gaseous products increased when Pt and Ni
were combined on ␥-Al2 O3 and ␥-Al2 O3 La2 O3 supported cata-
lysts. They concluded that Pt catalyst activity can be improved by
the addition of Ni, Co and Fe.
Based on this background, and consistent with our previous
research work [26], the main objective here is to study new
Pt/transition-metal (Fe, Co and Ni) catalysts prepared via sol–gel
method under acid conditions (SGA) and compare their catalytic
activities and stabilities with those obtained using similar Pt/Ni
catalysts prepared by sol–gel under basic conditions (SGB) and
incipient wetness impregnation (IWI), which have been published
in a previous work [26]. In fact, the selected catalytic systems con-
sist of ␥-Al2 O3 with PtNi, PtCo and PtFe as active phases. Pt and
Ni monometallic catalysts supported on ␥-Al2 O3 were also pre-
pared for both comparative purposes and better understanding of
the properties of PtNi catalyst and their influences on catalytic
activity, as commented in Section 2.3. To facilitate a comparison
between the results reported in this work and those shown in
[26], all the parameters and the operating conditions remained
unchanged. This research aims at the development of new cata-
lysts for the aqueous phase reforming of glycerol, and therefore it
investigates their catalytic activity and stability in order to max-
imize glycerol conversion, especially to gaseous products, and to
maximize H2 yield.
2. Experimental procedures
2.1. Catalyst preparation and reactants used
As mentioned above, PtFe, PtCo, PtNi, Ni and Pt-based ␥-Al2 O3
catalysts were prepared by the SGA method following a modified
preparation procedure as described in [17–20,23,26].
H2O/AIP solution (Molar ratio = 100)
HNO3 (pH controlled at 3.5) Heating to 40 ºC
and stirring for 1 h
Transparent alumina Sol
Vigorously stirring for 1 h at Fe, Co, Ni or/and Pt
40 ºC + 30 min in ultrasound precursors + C2H5OH
Gel precipitate
Solvent evaporation
Gellation
at 40 ºC for 24 h
Gel => rigid solid
Drying in an oven at
100 ºC for 12 h
Dry solid
Calcination at 550 ºC
for 6 h under air
Fresh Catalysts; PtFe, PtCo, PtNi, Ni
and Pt-based γ-Al2O3
Fig. 1. Schematic flow of the methodology used to prepare the catalysts by the acid
sol–gel method.
First, aluminum isopropoxide (AIP) Al[OCH(CH3 )2 ]3
(Sigma–Aldrich 99.995%) was hydrolyzed at 40 ◦ C, using a
molar ratio H2 O/AIP of 100. This solution was then peptized
by adding HNO3 to adjust the pH at 3.5. This pH was chosen
to be under acid conditions while the preparation took place.
This plays an important role in the evolution of the hydrolysis
reaction [23]: H3 O+ ions (acid medium) and OH− (basic medium)
do not have the same influence. The H3 O+ cations, attracted by
oxygen, facilitate the substitution of the groups O CH(CH3 )2
and, therefore, hydrolysis, while the OH− anions attracted by the
electronegative metal M favor the formation of M O M clusters
by condensation. As a result pH selection will be decisive to
determine the final texture of the materials prepared [23]. At the
same time, another solution was prepared using absolute ethanol
as solvent and the salt precursor (H2 PtCl6 ·xH2 O (Sigma–Aldrich,
99.99%), Ni(NO3 )2 ·6H2 O (Sigma–Aldrich, 99.99%), FeCl3 ·6H2 O
(Panreac, 99.5%) and CoCl2 ·6H2 O (Quimivita)) as solute. The metal
concentration in the precursor solution was selected to obtain
3 wt% of Pt and 10 wt% of Ni, Co or Fe. This salt solution was
added to the AIP-derived alumina sol at 40 ◦ C and the mixture was
vigorously stirred. After this, the sol formed was placed in ultra-
sound equipment for 30 min. The alcoholic solvent was removed
by drying at 40 ◦ C for 24 h. The rigid solid materials obtained by
the SGA method were dried under air at 100 ◦ C overnight and
then calcined at 550 ◦ C with a temperature ramp of 2 ◦ C/min for
6 h. The calcination procedure was carried out under an air flow
of 2 L min−1 g−1
cat . The preparation procedure described above is
schematically described in Fig. 1.
2.2. Catalyst characterization techniques
In order to understand the catalytic behavior of the tested cata-
lysts, their physicochemical properties were examined at calcined
state using different characterization techniques. The catalysts pre-
senting the best performances were also characterized at their
reduced state to get a better understanding of their functionalities.
Specifically, calcined catalysts were characterized by induc-
tively coupled plasma atomic emission spectroscopy (ICP-AES),
N2 adsorption–desorption isotherms at 77 K, temperature pro-
grammed reduction (TPR-H2 ), X-ray powder diffraction (XRD) and
 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136
X-ray photoelectron spectroscopy (XPS). The last two characteriza-
tion techniques were also used to characterize reduced catalysts.
The reduction treatment was carried out in situ within XRD and
XPS chambers using similar conditions to those used in the catalytic
activity tests (Section 2.3).
The chemical composition of calcined catalysts was determined
by the ICP-AES technique using a Perkin-Elmer Optima 3300DV
apparatus. Around 100 mg of solid samples were first dissolved in
an acid solution mixture [HNO3 (3 mL), HCl (2 mL) and HF (3 mL)]
using a microwave with a temperature profile of 7 ◦ C/min up to
180 ◦ C for 45 min. The solution obtained was diluted in water to
concentrations within the instrument’s detection range.
The textural properties were evaluated using the N2
adsorption–desorption technique in an Autosorb-1C instrument
(Quantachrome). Prior to the analysis, the samples were outgassed
for 12 h at 150 ◦ C. Specific areas were calculated from these
isotherms using the BET method and taking a value of 0.162 nm2
for the cross-section of the physically adsorbed N2 molecule.
Pore distribution and pore volume were estimated using the BJH
method (desorption data).
The same Autosorb-1C instrument equipped with a TCD detec-
tor was used to analyze the reducibility of the catalysts using the
TPR-H2 technique. Prior to reduction experiments, around 50 mg
of solid sample were thermally treated under a He stream at 300 ◦ C
in order to remove impurities, such as water and other contami-
nants. TPR profiles were obtained heating the sample from 25 to
1000 ◦ C at a linearly programmed rate of 10 ◦ C/min, while a mix-
ture of 10 vol% H2 /Ar flowed at 50 mL/min through the sample. A
cold trap was used to remove water from the H2 /Ar effluent before
using TCD to measure the amount of hydrogen consumed during
catalyst reduction.
X-ray powder diffraction patterns of calcined catalysts were
recorded with a D5000 Siemens diffractometer (D8 Advance Bruker
AXS) using a copper target (Cu K␣1 = 1.5418 Å ) and a secondary
beam monochromator. Another D5000 Siemens diffractometer
equipped with chamber analysis (Anton Paar XRK900) was used to
characterize the catalysts reduced in situ. Both apparatuses worked
with X-ray tube at 40 kV and 40 mA, and could detect crystalline
phases with average sizes above 2 nm. The scans were taken in a
2 = 10–80◦ range with a step size of 0.04◦ /s of acquisition time. The
crystallite size values were obtained using the Scherrer equation.
The XPS spectra of Pt and Ni elements were recorded using
a Kratos Axis UltraDLD electron spectrometer equipped with a
monochromatic Al K␣ X-ray source (1486.6 eV) and a hemispheri-
cal electron analyzer operating at constant pass energy (40 eV). The
apparatus operated with a power of 150 W, 10 mA and 15 kV. Bind-
ing Energy (BE) values were referred to the C1s peak (284.6 eV)
for the calcined and reduced catalysts. The C1s, O1s, Al2p, Al2s,
Pt4d and Ni2p core-level spectra were recorded. Superficial quan-
tification was carried out by the level Al2s to avoid the influence
of any overlap between the peaks of Pt4f and Al2p. Therefore,
Pt4d was used here instead of Pt4f. The calcined samples were
degasified at 10−8 bar and transferred immediately to the ion-
pumped analysis chamber, whose residual pressure was kept below
(7–9) × 10−12 bar during data acquisition. Catalyst reduction was
carried out under the abovementioned conditions in the spectrom-
eter’s preparation chamber, directly connected to the ultra-high
vacuum analysis system. CasaXPS software was used for XPS
spectra treatment. Different peak areas were estimated by cal-
culating the integral of each peak after subtracting the Shirley
background and fitting the experimental peak to a combination of
Lorentzian/Gaussian lines of variable proportions. The normaliza-
tion of the estimated areas of the corresponding species allowed
calculating the relative atomic ratios.
In addition to the analysis carried out by the techniques
mentioned above, the reduced catalysts were also analyzed by
127
transmission electronic microscopy, using a TEM/EDX FEI Tecnai
G2 20 equipped with a LaB6 filament, a super twin lens operating
at 200 kV and EDX beam of 5 nm. Bright field images were acquired
using a CCD camera (TVIPS GmbH). Powders of the reduced catalyst
samples were dispersed in ethanol with ultrasound, and drops were
deposited on copper grids coated with amorphous carbon films. The
TEM images were taken using the high resolution mode of trans-
mission microscopy. The energy dispersion X-ray was used by an
EDX detector to analyze the local chemical composition of some
catalyst particles.
2.3. Activity tests
The catalytic activities of PtFe, PtCo and PtNi catalysts were
measured in the bench-scale fixed-bed reactor of a Microactivity
modular laboratory system (PID Eng. & Tech.). In order to under-
stand the influence of the different active phases of the best Pt–Ni
bimetallic catalysts, Pt and Ni monometallic catalysts were also
prepared and tested under the same operating conditions.
The fixed bed reactor was tubular (316-stainless steel), where
1 g of catalyst, sieved to 0.42–0.5 mm, was diluted in inert SiC.
The catalyst/SiC ratio was 1:3 (w/w). Prior to the catalytic tests,
the Pt and PtFe catalysts were reduced at 350 ◦ C, while the PtCo,
PtNi and Ni catalysts were reduced at 750 ◦ C. The temperature
was increased from ambient temperature to the desired reduction
temperature (TR ) using 150 mL/min of N2 over 90 min. The tem-
perature was then maintained at TR for 2 h using 100 mL/min of
H2 /N2 mixture (75 vol% of H2 ). After this in situ reduction treat-
ment, the system was purged with N2 to remove the remaining H2
and the H2 O formed during the reduction treatment. The catalytic
tests were performed using an aqueous solution of 10 wt% of glyc-
erol/water mixture. This solution was directly fed into the reactor
−1
with a HPLC pump maintaining a WHSV at 2.6 g glycerol g−1 cat h .
The temperature and pressure fixed in the activity test were 250 ◦ C
and 50 bar for bimetallic catalysts. The catalyst with the best per-
formance was also tested at 230 ◦ C and 30 bar. The operating
temperature/pressure conditions were chosen to operate under
APR conditions. Aspen Plus® simulation was used to estimate the
temperature and pressure values for which the mixtures used are
in the liquid phase. It is important to note that a new charge of
the same catalyst was prepared for each operating condition. The
high pressure used in this kind of process was maintained with N2 ,
which was directly fed into the gas–liquid phase separator to avoid
increasing the space velocity. In addition, to avoid any mixing of
N2 (used to entrain the gaseous products from the Peltier) with the
reactants in the reactor, a “non-return valve” was placed at the inlet
of the Peltier (see apparatus set-up in [26]).
The stream gas produced was periodically analyzed with an
on-line ␮GC (TCD/SM). The oxygenated hydrocarbon products and
non-reacted glycerol were condensed at 15 ◦ C (vapor–liquid sepa-
rator), collected and finally analyzed by GC equipped with FID and
TCD detectors.
Catalytic activity was evaluated using the following activity
parameters [10,15,26–28].
• Glycerol conversion:
in
NGlycerol out
(mol/min) − NGlycerol (mol/min)
GC (%) = × 100 (1)
in
NGlycerol (mol/min)
• Conversion into gas:
Catomes in gas products (mol/min)
CIG (%) = × 100 (2)
Catomes in gas products (mol/min)
128 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136

• Flow of total gas formed: PtOx

Neighboring
−1 −1 measured volume (mL) 273 K Fe-oxides
FT (mL(STP)) min gcat. = × (3)
time (min) × Mcat. (g) Tambient
Fe-aluminate
The gaseous stream flow was normalized by catalyst mass to PtFeAl(SGA)
render it easier to compare with similar data published in the lit- PtOx CoAlxOy
Co3O4
erature. CoCl2, 6H2O CoAl2O4

CoO
• H2 selectivity: PtCoAl(SGA)

N(Hout (mol/min)

Intensity (a.u.)
2) 3 NiO-Al2O3
NiAlxOy
S(H2 ) (%) = out
× × 100 (4) NiAl2O4
NC in gas phase (mol/min) 7 PtOx-NiO

3/7 is the reforming ratio of CO2 /H2 for glycerol taking into
account the occurrence of the WGS reaction after glycerol decom- PtNiAl(SGA)
position. NiAlxOy
NiO-Al2O3
• Yield gas of H2 and alkanes: NiAl2O4

out (mol/min)
N(i) NiO
Y (i) = × 100 (5) NiAl(SGA)
in
Nglycerol (mol/min)
PtOx
where (i) are H2 or alkane products such as: CH4 , C2 H6 and C3 H8 .
PtAl(SGA)

• Yield of oxygenated hydrocarbon compounds (OHCs):

out
Nliquid (mol/min) (exp.)
products
Y OHCs (j) =
in
Nglycerol (mol/min)
where (j) are C1 (methanol), C2 (acetaldehyde, acetic acid, ethyl-
ene glycol) or C3 (propanal, 1&2-propanol, acetol, 1,2-propanediol,
excluding unconverted glycerol).
3. Experimental results
3.1. Catalyst characterization results
3.1.1. Chemical composition and textural properties
Chemical composition, as determined by the ICP-AES technique,
and data about textural properties obtained by N2 adsorption–
desorption isotherms for calcined catalysts are presented in Table 1.
The metal amounts incorporated are slightly different from the
nominal values, while the catalysts have a fairly similar chemical
composition.
According to SBET values, in all cases the metal incorporation
reduced the specific surface area of the catalysts when compared to
the bare support. This suggests that the metal particles block espe-
cially small pores, as the pore diameter data confirm, resulting in an
increase in the average pore diameter (Table 1). Panyad et al. [29]
observed a similar effect among fresh and used sol–gel catalysts,
which was attributed to the plugging of small pores. The bimetal-
lic catalysts had a lower specific surface area than NiAl and PtAl
monometallic catalysts. This observation suggests that the higher
the metal amount incorporated, the higher the SBET loss related to
small pore blockage.
3.1.2. TPR-H2 analysis of catalysts
As commented previously, TPR-H2 measurements were carried
out to investigate the reducibility of the calcined PtFe, PtCo, PtNi,
Ni, Pt-based ␥-Al2 O3 catalysts, as well as to examine the interaction
between metal species and alumina support in the catalysts. The
TPR-H2 profiles obtained are shown in Fig. 2.
In the case of the PtFeAl(SGA) catalyst, the reducibility profile
revealed two components. Gaussian deconvolution was used to
decompose each component into two peaks. The first overlapping
0 150 300 450 600 750 900 1050
Temperature (ºC)
(6)
Fig. 2. TPR-H2 profiles of PtFe, PtCo, PtNi, Ni and Pt-based ␥-Al2 O3 catalysts.
peaks in a low temperature range (130–320 ◦ C) were assigned to
the reduction of PtOx and, possibly, neighboring Fe oxide, which
was reduced by hydrogen spillover at low temperatures. The
second and ill-defined high temperature hump, between approx-
imately 680 ◦ C and 1030 ◦ C could also be decomposed into two
peaks, which can be attributed to the presence of trace amounts of
Fe strongly interacting with ␥-Al2 O3 originating a phase similar to
iron–aluminate [30,31].
For the PtCoAl(SGA) catalyst, two broad components were
observed in its TPR-H2 profile. The first component was decom-
posed into four overlapping peaks between 50 ◦ C and 500 ◦ C. The
second component was also decomposed into two other peaks
at high temperatures (T > 500 ◦ C). The shoulder at 160 ◦ C can be
attributed to the decomposition of residual cobalt chloride to
Co3 O4 . This shoulder depends mainly on the catalyst preparation
method and calcination conditions. The following two peaks, at
around 260 ◦ C, are due to the reduction of relatively large PtOx
particles and Co3 O4 to CoO. The peak at approximately 330 ◦ C can
be attributed to the reduction of CoO to metallic Co. The com-
paratively sharp peaks appearing at around 650 ◦ C and 770 ◦ C can
be attributed to the reduction of cobalt–aluminate compounds
CoAlx Oy and CoAl2 O4 , respectively. These two phases require high
temperatures for their reduction because they strongly interact
with ␥-Al2 O3 . This TPR-H2 diagram is fairly similar to analogous
ones reported in the literature for similar systems [32,33].
For Pt and PtNi catalysts, H2 consumption can be observed
in their reduction profiles between 200 ◦ C and 300 ◦ C. The cor-
responding peak can be assigned to the reduction of Pt species
interacting with the oxygen of ␥-Al2 O3 . The explanation for Pt
oxide detection is related to the ability of ␥-Al2 O3 to stabilize small
particles of PtOx scattered widely over its surface, as described
elsewhere [15,26,34].
For Ni-containing catalysts, two clearly reducing compo-
nents were observed in their TPR-H2 profiles, which have been
decomposed into four peaks by Gaussian deconvolution. As the
 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136 129

Table 1
Chemical composition and N2 isotherms of calcined catalysts.

Symbol Catalyst formula Composition (wt%) N2 isotherms

Pt Ni Co Fe SBET (m2 /gcat. ) dPore (nm) VPore (cm3 /gcat. )

Al(SGA) ␥-Al2 O3 – – – – 303 2.8 0.22

PtNiAl(SGA) PtNi/␥-Al2 O3 3.2 8.0 – – 215 3.3 0.19
PtCoAl(SGA) PtCo/␥-Al2 O3 2.8 – 7.2 – 220 3.4 0.19
PtFeAl (SGA) PtFe/␥-Al2 O3 2.9 – – 6.9 225 3.5 0.20
PtAl(SGA) Pt/␥-Al2 O3 2.6 – – – 270 2.9 0.19
NiAl(SGA) Ni/␥-Al2 O3 – 8.0 – – 250 3.4 0.21

temperature increases from 300 ◦ C to 1000 ◦ C; NiO species, NiO
weakly interacting with ␥-Al2 O3 , non-stoichiometric and stoichio-
metric Ni-spinels were reduced [26,35,36]. These Ni peaks are large
and broad, indicating that these species were formed with different
sizes and various degrees of interaction with the support.
The oxidized Ni species of PtNiAl(SGA) are reduced at lower tem-
perature when compared to the ones of NiAl(SGA) . This is explained
by the H2 spillover effect [15,26]. In fact, Pt combined with Ni pro-
motes the reduction of most of the Ni at lower temperatures. It
is also interesting to note that in the PtNiAl(SGA) catalyst the NiO
reduction peak overlaps those of PtOx suggesting a possible interac-
tion between these two components (PtOx –NiO). This may generate
a possible promoting effect by electron transfers between Pt–Ni
metals after the reduction treatment [15,26].
3.1.3. XRD characterization of catalysts
XRD analyses were carried out to identify the crystalline phases
in the prepared catalysts after their calcination. The X-ray diffrac-
tograms are presented in Fig. 3.
The patterns are broad and their intensity is very weak, imply-
ing that the phases are not highly crystalline. All catalysts recorded
low reflections at 37.2◦ , 46.2◦ and 66.8◦ , which can be assigned to
␥-Al2 O3 phase (JCPDS 29-063). Reflections at 39.6◦ , 46.3◦ and 67.8◦ ,
corresponding to Pt/PtOx phase (JCPDS 4-802), were also registered
in the Pt-containing catalysts (except in PtFeAl(SGA) ). The Pt/PtOx
peaks of PtCoAl(SGA) catalyst were more intense and thinner, indi-
cating that they correspond to bigger particles (around 46 nm).
These peaks were not detected in the PtFeAl(SGA) catalyst, probably
due to the corresponding phase being present at very low quanti-
ties or in a more amorphous structure, since Fe species reflections
were not detected either. The reflection lines corresponding no Co
species were registered in the PtCoAl(SGA) catalyst, and this could
indicate that Co species were very small (2 nm is the limit of XRD
sensitivity) and highly distributed.
The Ni-containing catalysts presented reflections at 36.4◦ , 44.4◦
and 64.5◦ , which were attributed to NiAl2 O4 (JCPDS 01-1299)
phases. These components are formed due to the strong Ni O Al
interaction caused by the sol–gel preparation in the acidic medium.
This high interaction caused a very good dispersion of Ni species
on these catalysts. As TPR-H2 results indicated, NiO bulk was also
formed, especially in the NiAl(SGA) catalyst. However, reflection
lines related to NiO oxide were not detected, indicating that if these
species are present, they are quite small (below 2 nm) or in very low
quantities. This interpretation is in agreement with other authors
who have studied similar catalyst systems [3,10,11,15,27].
To gather more information about the active phases dispersion
and the average size of Pt and Ni particles on the PtNiAl(SGA) cata-
lyst, it was decided to analyze the reduced Pt, Ni and PtNi catalysts
by in situ X-ray diffraction (i.e., at a similar state to the one just
before the activity tests were begun). These X-ray diffractograms
are presented in Fig. 4. The X-ray patterns reveal that the reduc-
tion treatment slightly increased the crystallinity of the alumina
support of these catalysts, but their active phases remained well
dispersed. This may indicate that the surface area and porosity
of ␥-Al2 O3 were not significantly affected by the reduction treat-
ment. However, new diffraction lines were registered at 44.5◦ , 51.8◦
and 76.6◦ . These lines can be associated to metallic Ni0 (JCPDS
04-850) [3,10,11,15,27]. It should be noted that the Ni0 peaks of
PtNiAl(SGA) are more intense than those detected in NiAl(SGA) . This
indicates that Pt promoted Ni reduction, taking into account that
both catalysts have similar Ni loadings (8 wt%), as indicated by the
ICP analysis. This observation is also corroborated by previously
mentioned TPR-H2 results.

Fig. 4. X-ray patterns of in situ reduced Pt, Ni, PtNi-based ␥-Al2 O3 catalysts and their
corresponding average size particles at calcined and in situ reduced state obtained
Fig. 3. X-ray patterns of calcined Pt, Ni, PtNi, PtFe and PtCo-based ␥-Al2 O3 catalysts. by the Scherrer equation.
130 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136
To estimate the effect of the reduction treatment on metal
particle sizes, the average sizes of Pt or/and Ni species were cal-
culated from the XRD data for calcined and reduced catalysts. This
calculation involved the Scherrer equation and the most intense
reflections at 39.6◦ for Pt0 /PtOx, 36.4◦ for NiAlx Oy and 44.5◦ for Ni0 .
In fact, a careful deconvolution of overlapping peaks was used to
carry out a good estimation of the particle sizes. These results are
presented in Fig. 4.
The average Pt particle size did not change for Pt-containing
catalysts, indicating that the SGA effectively yields stable Pt parti-
cles even at high reduction temperatures (750 ◦ C for PtNiAl(SGA) ).
However, for Ni-containing catalysts, Ni particle sizes increased,
indicating that after the reduction treatment the Ni particles
suffered some sintering. This increase is lower for PtNi cata-
lyst, which suggests that Pt combined with Ni increases the
resistance of PtNi catalysts to this Ni particles sintering phe-
nomenon. This enhancement may be related to the spillover
effect, since this phenomenon led to Ni reduction at lower
temperatures.
3.1.4. XPS analysis of Pt, Ni and PtNi catalysts
In order to study the physicochemical change experienced by
different elements during the reduction treatment, the calcined
(Cal.) and in situ reduced (Red.) Pt, Ni and PtNi catalysts were
analyzed by XPS technique. The binding energy (BE), chemical
state and the superficial amounts of Pt or/and Ni species, as well
as Pt/Al and Ni/Al atomic ratios, are summarized in Table 2. The
Lorentzian/Gaussian deconvolutions of Pt4d5/2 , and Ni2p3/2 core-
level spectra are also illustrated in Fig. 5.
The results obtained from XPS analysis showed the existence
of a band at 74.4 eV (±0.4 eV) for all calcined and reduced cata-
lysts, which is characteristic of Al3+ belonging to ␥-Al2 O3 [37,38].
The deconvolution of Pt4d5/2 core-level spectra for calcined Pt and
PtNi catalysts showed the coexistence of at least two Pt species:
one corresponds to Pt metallic at 314.3 and 314.9 eV, respectively.
Another component that can be related to Pt oxide was detected at
317.8 and 317.4 eV, respectively [39]. The detection of PtOx species
may be associated to the capacity of ␥-Al2 O3 to stabilize some PtOx
particles on its surface due to the Pt O Al interaction caused by
the SGA preparation method, as TPR-H2 results corroborated. The
BEs of the Ni2p3/2 core level recorded for the calcined Ni and PtNi
catalysts are located at 855.6 and 855.8 eV, respectively. This BE is
typically assigned to Ni2+ species formed due to the strong Ni O Al
interaction forming non-stoichiometric NiAlx Oy and stoichiomet-
ric NiAl2 O4 , symbolized as NiT 2+ (pseudo tetrahedral coordination)
[40]. For NiAl(SGA) catalyst, another BE was detected at around
853.6 eV, which can also be assigned to Ni2+ in the form of NiO,
symbolized as NiO 2+ (pseudo octahedral coordination) [40].
A comparison of the BEs of the Pt4d5/2 core level of calcined Pt-
containing catalysts showed that the BE of Pt0 was slightly higher
for PtNiAl(SGA) than for PtAl(SGA) catalyst (see Table 2). A similar
observation was reported by Pawelec et al. [41], being attributed to
a possible electron transfer due to higher Pt O Al interactions. For
calcined Ni-containing catalysts, the PtNiAl(SGA) catalyst recorded a
higher proportion (100%) of NiT 2+ than the NiAl(SGA) (79%) catalyst.
This higher population of NiT 2+ on the catalytic surface indicates
that the Ni species are well dispersed and, as a result, Ni surface
exposure was also higher, since the XPS atomic ratio of Ni/Al in
PtNi catalyst (0.424) is higher than in the Ni catalyst (0.325). This
confirms that the Ni in bimetallic catalysts ends up being better
dispersed on the ␥-Al2 O3 surface. In addition, the Pt/Al and Ni/Al
atomic ratios estimated from XPS-derived data are higher than
those calculated using bulk compositions from chemical analysis
(see Table 2). This means the catalyst outer surface is enriched in
Pt and Ni species.
After the reduction treatment, BEs for Pt0 species were detected
at 314.4 for PtAl(SGA) and at 315.0 eV for PtNiAl(SGA) catalysts. In
the latter, apart from this BE energy, an additional BE was detected
at around 317.4 eV and this indicates that Pt was not completely
reduced. A comparison of the BE of the Pt4d5/2 core level for reduced
PtNi and Pt catalysts showed that BE is slightly higher for PtNi than
for Pt catalysts. Bakar et al. [42,43] attributed this BE increase to the
formation of metallic bands or alloying of the metals involved. Some
Ni atoms are probably interacting with Pt modifying its electronic
environment via electron transfer, which can led to a promoting
effect between both elements. This result is also supported by TPR-
H2 results, where the overlapping PtOx NiO peaks indicate some
possible Pt Ni interaction, which can occur after the reduction of
PtNiAl(SGA) catalyst.
The Ni2p3/2 core level for reduced Ni-containing catalysts
records a peak at 852.6 and 852.2 eV, which is characteristic of
Ni0 [44–46]. Another component was detected at 855.1–855.4 eV,
which can be related to the non-reduced nickel ions strongly
interacting with alumina lattice oxygen [44–46]. This indicates
that nickel species are not completely reduced. However, as com-
mented previously, the amount of reduced Ni was higher in the
PtNiAl(SGA) catalyst (60%) as compared to its monometallic NiAl(SGA)
counterpart (37%). This is mainly due to H2 spillover favored by
well-dispersed Pt on this catalyst.
A comparison between the Pt/Al and Ni/Al atomic ratios
(reduced catalysts with their calcined counterparts) reveals that
these ratios are lower in reduced catalysts than in their calcined
counterparts. This indicates that the reduction treatment at high
temperature (750 ◦ C) partially sintered Pt and Ni particles. This
result is in agreement with the results obtained from XRD analy-
sis, as the literature [41] also indicates. The decrease in the surface
(XPS) Ni/Al ratio after reduction could also be due to the incorpora-
tion of metal species into the porous structure of the support, as the
values for pore volumes derived from N2 adsorption–desorption
isotherms indicated.
3.1.5. TEM/EDX analysis of Pt, Ni and PtNi catalysts
The reduced Pt, Ni and PtNi-based ␥-Al2 O3 catalysts were exam-
ined by transmission electron microscopy coupled with an energy
dispersive X-ray detector (TEM/EDX) (Fig. 6).
Fig. 5(a–e) illustrates the textural morphology of reduced bulk
␥-Al2 O3 (a), Pt/␥-Al2 O3 (b), Ni/␥-Al2 O3 (c) and PtNi/␥-Al2 O3 (d)
catalysts and the corresponding EDX spectra (e) at three different
points of PtNi/␥-Al2 O3 catalysts. These points were chosen to deter-
mine the local chemical composition of particles dispersed on the
surface of PtNi/␥-Al2 O3 catalysts.
The image (a) shows that the ␥-Al2 O3 support of the catalysts
prepared by the acid sol–gel method presents a laminated tex-
ture with high micro-porosity, which confirms that preparation
by this procedure leads to the formation of a longitudinal fiber
morphology. This is consistent with the information derived from
adsorption/desorption results. The TEM images (b–d) also show
that Pt or/and Ni catalysts have well-dispersed Pt or Ni species on
the ␥-Al2 O3 surface. These good dispersions of Pt and Ni particles
make it difficult to obtain a good estimate of the particle size distri-
bution, but it is clear from TEM images that the particles sizes are
very small. It can also be appreciated that Pt and Ni clusters have a
spherical form (uniformly distributed) with an approximate aver-
age diameter of 1.6 nm and 11.5 nm for Pt/␥-Al2 O3 and Ni/␥-Al2 O3 ,
respectively. The image (d) of PtNi/␥-Al2 O3 shows that the average
diameters of Pt (1.4 nm) and Ni (7 nm) particles are similar to Pt and
Ni particle diameters in the monometallic catalyst counterparts.
The average diameters of Ni particles are consistent with those esti-
mated by XRD (Table 2), while those of the Pt particles are smaller.
This may be due to the X-ray technique’s limitation for detecting
 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136 131

Table 2
BE and population of Pt or/and Ni species in different chemical states, as well as atomic ratios for calcined (Cal.) and in situ reduced (Red.) Pt, Ni and PtNi-based ␥-Al2 O3
catalysts.

Catalyst B.E. Pt4d5/2 (eV) Pt/Al B.E. Ni2p3/2 (eV) Ni/Al

Pt0 PtOx Ni0 NiO 2+ NiT 2+

PtAl(SGA)
Cal. 314.3 (20) 317.8 (80) 0.010a – – – –
0.007b
Red. 314.4 (65) 317.4 (35) 0.010 – – – –
NiAl(SGA)
Cal. – – – – 853.6 (21) 855.8 (79) 0.325a
0.071b
Red. – – – 852.6 (37) – 855.4 (63) 0.094
PtNiAl(SGA)
Cal. 314.9 (24) 317.4(76) 0.056a – – 855.7 (100) 0.424a
0.009b 0.069b
Red. 315.0(100) – 0.015 852.2 (60) – 855.1 (40) 0.103

( ) is population of different species.

a and b are atomic ratio calculated from XPS and ICP data, respectively.

particles with diameters below 2 nm. Besides that, it is interesting
to note that Pt/Ni agglomerations formed by Pt clusters deposited
on Ni counterparts can be observed in the PtNi/␥-Al2 O3 catalyst (see
image d). This observation was confirmed by EDX analysis (see EDX
spectra (e)), although there are not many of them.
EDX analysis was carried out to check local chemical compo-
sitions. The analysis of point P1 (single thin particles) showed
Pt and Al emission lines, indicating the existence of Pt over all
the ␥-alumina matrix, as the size of the EDX beam used was ca.
5 nm, while the Pt particles have a diameter of approximately
1.4 nm. By focusing the EDX beam on the slightly higher parti-
cles (point P3 ), the analysis indicated that these are formed by
Ni. However, the EDX analysis of point (P2 ) showed the existence
of Ni and Pt emission lines in some agglomerations. This sug-
gests that these agglomerations were formed by particles with very
close Pt/Ni atomic vicinity, which can facilitate an electronic inter-
action or transfer between both elements, generating a possible
promoting effect. This Pt present in close vicinity to Ni parti-
cles confirms similar experimental findings from TPR and XPS
analysis.
3.2. Catalytic activity results
As mentioned above, PtFe, PtCo and PtNi-based ␥-Al2 O3 cata-
lysts were only tested in the glycerol APR process at 250 ◦ C and
50 bar. The parameters for catalytic activity results, such as glyc-
erol conversion (GC), conversion into gas (CIG) and the distribution
yields of gaseous products (Y(i)) and oxygenated hydrocarbon com-
pounds (Y OHCs(j)), are shown in Fig. 7 (data obtained after 32 h of
reaction, steady state). For a better understanding of the role played
by each phase in the PtNi bimetallic catalyst, the ones with the best
results – Pt and Ni monometallic catalysts – were then tested at
250 ◦ C/50 bar and 230 ◦ C/30 bar. Their catalytic behavior as a func-
tion of time-on-stream is summarized in Table 3 in terms of GC,
CIG, and the distribution yields of major OHCs in liquid stream. The
gas flow formation and H2 selectivity obtained at steady state for
Pt, Ni and PtNi-based ␥-Al2 O3 catalysts are also presented in Fig. 8.
Operating at 250 ◦ C/50 bar, the catalysts showed increasing
activity, following the order PtFeAl(SGA) < PtCoAl(SGA)  PtNiAl(SGA) .
PtFe and PtCo showed quit low activity in glycerol reforming. The
PtNi catalyst recorded the highest glycerol conversions: GC (74%)

Fig. 5. XPS spectra deconvolution of Pt4d5/2 and Ni2p3/2 core levels for calcined (Cal.) and in situ reduced (Red.) Pt, Ni and PtNi-based ␥-Al2 O3 catalysts.
132 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136

Fig. 6. (a–d) Textural morphology of reduced bulk ␥-Al2 O3 (a), Pt/␥-Al2 O3 (b), Ni/␥-Al2 O3 (c) and PtNi/␥-Al2 O3 (d) catalysts and the corresponding EDX spectra (e) of three
different points of PtNi/␥-Al2 O3 catalysts obtained from TEM/EDX analysis.

and CIG (34%). Glycerol conversion into gas when PtNiAl(SGA) cata-
lyst was used was nearly nine fold higher than the conversion rates
obtained using PtFe or PtCo catalysts. This indicated that the PtNi
catalyst was far more active and allowed higher reforming rates of
glycerol and of liquid intermediate products (OHCs) into gaseous
products. The major gaseous products formed on the catalysts were
H2 , CO2 , CH4 and smaller amounts of other alkanes (C2 C3 ). Their
relative yields varied depending on the catalyst, and the PtNi cat-
alysts provided the best H2 yield (0.51), but also higher alkane
yields; for example, the methane yield was as high as 0.44 (see
Fig. 7b). Meanwhile, several OHCs were detected in the liquid phase,
such as C1 (methanol), C2 (acetaldehyde, acetic acid, ethylene gly-
col) and C3 (propanal, 1&2-propanol, acetol, 1,2-propanediol). The
C3 compounds accounted for 70–95% of the total liquid products
formed. These compounds were typically formed by dehydration
reactions (cleavage of C O band), indicating that the acid centers
of the alumina support were also involved in the reaction, and
not only the Pt/Ni metallic centers, where C C/C H bond cracking
 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136 133

Table 3
Glycerol conversion (GC) and conversion into gas (CIG), as well as yield (Y OHCs(j)) distribution of OHCs as a function of APR time-on-stream on Pt, Ni and PtNi-based ␥-Al2 O3
catalysts at 250 ◦ C-50 bar (a)/230 ◦ C-30 bar (b) (10 wt% C3 H8 O3 /H2 O mixture, feed rate of 0.4 mL/min, 1 g of catalyst, WHSV = 2.60 h−1 ).

Catalyst Time-on-stream (h) of aqueous phase reforming reaction

Parameters 2h 8h Night 25 h 32 h
(SGA)
PtAl Conversion a/b a/b a/b a/b
CG (%) 25.5/11.4 25.0/11.1 27.6/9.9 23.4/9.3
CIG (%) 7.9/4.8 6.8/3.7 8.1/3.4 6.2/3.5

Liq. stream Y OHCs(j) × 10−2

2-Propanol 3.0/1.1 3.0/1.0 3.0/1.0 3.0/1.0
1,2-PDO 4.1/1.0 5.2/1.0 6.0/2.1 7.1/2.0
Acetic acid 0.0/3.0 0.0/3.0 0.0/3.0 0.0/3.0
Acetol 1.1/1.0 2.1/1.0 2.2/1.0 2.2/1.0
EG 0.0/0.0 1.0/0.0 2.1/0.0 2.0/0.0
Methanol 0.0/0.0 1.0/0.0 1.0/0.0 1.0/0.0

NiAl(SGA) Conversion a/b a/b a/b a/b

CG (%) 68/57 49/40 47/29 42/27
CIG (%) 22/12 10/7.5 9/5 8/4.4

Liq. stream Y OHC(j) × 10−2

2-Propanol 8.0/5.0 5.0/4.0 4.0/2.0 4.0/2.0
1,2-PDO 22.0/20.0 17.0/15.0 14.07.0 14.0/8.0
Acetic acid 0.0/0.0 0.0/0.0 0.0/0.0 0.0/0.0
Acetol 5.0/5.0 5.5/5.0 6.0/4.0 6.0/4.0
EG 5.1/4.1 4.2/3.0 3.0/1.1 2.0/1.1
Methanol 1.1/1.0 1.0/1.0 0.0/0.0 0.0/0.0

PtNiAl(SGA) Conversion a/b a/b a/b a/b

CG (%) 97/70 87/64 78/56 74/55
CIG (%) 58/36 49/29 39/25 35/22

Liq. stream Y OHCs(j) × 10−2

2-Propanol 08.1/5.3 07.1/4.2 06.1/3.0 4.4/3.0
1,2-PDO 10.3/20.0 15.1/17.0 15.0/16.0 14.8/16.0
Acetic acid 6.1/00.0 2.7/0.0 0.0/0.0 0.0/0.0
Acetol 1.5/2.0 2.2/2.0 2.7/2.0 2.6/2.0
EG 4.7/13.0 10.3/12.0 10.5/14.0 10/15.0
Methanol 3.1/3.0 3.0/3.1 2.4/2.0 2.1/2.0

uses to occur leading to gaseous products and C1 C2 compound
production.
An analysis of Table 3 and Fig. 8 shows that at both operating
conditions, PtNi bimetallic catalyst provided higher conversion and
hydrogen flow rates when compared to its Pt and Ni monometal-
lic counterparts. Not only were the conversions (CG, CIG) obtained
using the PtNiAl(SGA) catalyst higher than the sum of the ones
obtained by each PtAl(SGA) and NiAl(SGA) (Table 3), but also the
gaseous product flow was higher than the corresponding sum of
the flows generated by both monometallic catalysts (see Fig. 8).
This observation confirmed once again the important role of the
promoting effect between the two metals on the activity of PtNi
bimetallic catalyst. The indicated effect is more pronounced at
moderate temperature/pressure conditions (230 ◦ C/30 bar). More-
over, it can be observed that Pt monometallic catalyst presented
lower conversions, but its catalytic behavior was relatively sta-
ble throughout the reaction time, while the Ni-containing catalyst
counterparts recorded higher glycerol initial conversions but lower
stability (Table 3). However, considering only the Ni-containing
catalysts, the data presented in Table 3 show that its cat-
alytic behavior is relatively more stable at 230 ◦ C/30 bar than at
250 ◦ C/50 bar.
Regarding H2 selectivity, the Pt catalyst promotes H2 -rich gas,
while Ni is less selective. PtNi led to an intermediate behavior,
particularly at 230 ◦ C/30 bar. This high H2 selectivity presented by
PtAl(SGA) catalyst is related to its lower glycerol conversion into
gas (≈6%) as opposed to 35% for PtNiAl(SGA) and to its negligible
methanation activity. At higher glycerol conversions, H2 selectivity
usually fell due to the formation of several undesired by-products
via series and parallel reactions [10,11] and these intermediates are
not completely reformed. On the one hand, series reactions arose
from the hydrogenation of adsorbed CO/CO2 species on catalyst
surfaces to form alkanes, which negatively affect H2 generation.
Alkane formation can take place through different hydrogen con-
suming reactions. The hydrogen generated in the first step could
be converted into alkanes, CO2 and water by hydrogenation of the
dehydrated reaction intermediates [10,11].
CH4 can be formed via methanation from CO/CO2 (Ni sites are
active in these reactions), leading to a loss of H2 rate formation.
Other alkanes could possibly be formed by Fischer–Tropsch type
reactions [10,11]. High pressures (50 bar) favored gaseous reactions
that lower total mole number, such as the reactions generating
CH4 , C2 H6 and C3 H8 [10,11]. On the other hand, parallel reactions
proceeded through C O bond cracking followed by hydrogenation
to give diols, essentially in the form of 1,2-propanediol (recording
higher yields for PtNi and Ni and presenting higher glycerol con-
version and lower H2 selectivity – see Table 3 and Fig. 8) and
monoalcohols (2-propanol and methanol), or by rearrangement
pathways to form organic acids in liquid phase. Therefore, one of the
keys for aqueous phase reforming of glycerol is to choose a selective
and robust catalyst able to avoid, as far as possible (i) the hydro-
genation of CO/CO2 via the methanation or/and Fischer–Tropsch
reactions, and (ii) C O bond cleavage, although it must also be
very active for C H/C C bond cleavage and promote the removal
of adsorbed CO throughout the WGS reaction.
4. Discussion
The acid sol–gel preparation method used in this work gener-
ated catalysts with fibrous and laminated morphology, and high
134 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136

80
74 (a)
GC CIG
60
GC/CIG (%)

34 39 35
40

20

4 4
0
PtCoAl(SGA) PtFeAl(SGA) PtNiAl(SGA)

0.6
H2 CH4
0.51 (b)
C2H6 C3H8
0.5
0.44 Fig. 8. Gas flow formation (olive) and H2 selectivity (red) obtained from the APR test
of Pt, Ni and PtNi-based ␥-Al2 O3 catalysts at 250 ◦ C/50 bar and 230 ◦ C/30 bar (10 wt%
0.4
C3 H8 O3 /H2 O mixture, feed rate of 0.4 mL/min, 1 g of catalyst, WHSV = 2.60 h−1 , data
Y (i)

at steady state). (For interpretation of the references to color in this figure legend,
0.3 the reader is referred to the web version of the article.)

0.2 that Pt helped both an easier reduction and a better dispersion

0.04 0.05 of Ni active sites as compared to the ones observed in NiAl(SGA)
0.1 0.01 0.00 0.01 catalyst. Moreover, the high surface area and micro/mesoporosity
0.01
0.00 achieved by the SGA method facilitate the adsorption of intermedi-
0.000 0.000 0.001
0.0 ate molecules on highly dispersed Pt and Ni sites, converting them
PtCoAl(SGA) PtFeAl(SGA) PtNiAl(SGA) into gaseous products. In addition, the close vicinity of Pt/Ni leads
to a promoting effect, which was suggested by TPR-H2 results and
0.4 confirmed by XPS and TEM/EDX analyses, and could also have a
C1 (methanol) (c) crucial effect in the best catalytic activity shown by PtNiAl(SGA) cat-
C2 (acetaldehyde + ethylene glycol) alyst. This better performance can also be related to the high activity
C3 (propanal + acetol + 1&2-propanol +1,2-propanediol)
0.3 of Pt and Ni in C H/C C bond cleavage [10–12]. According to the
Y_OHCs (j)

0.23 literature [10–12], oxygenated molecules are decomposed on Pt

0.18 0.19
0.2
0.1
0.06
0.02
0.000.01 0.000.01
0.0
PtCoAl(SGA) PtFeAl(SGA) PtNiAl(SGA)
Fig. 7. (a–c) Catalytic activity of PtFe, PtCo and PtNi-based ␥-Al2 O3 catalysts at
250 ◦ C/50 bar; conversions of glycerol (a), distribution yields of gaseous products
(b) and liquid products (c) at steady state (10 wt% C3 H8 O3 /H2 O mixture, feed rate of
0.4 mL/min, 1 g of catalyst, WHSV = 2.60 h−1 and 30–32 h of APR).
surface area and micro/mesoporosity. This method allows obtain-
ing Pt and Ni particles with close vicinity and highly dispersed
leading to an increase in catalytic activity.
The higher catalytic activity of PtNi compared to bimetallic (PtFe
and PtCo) and monometallic (Ni and Pt) catalysts can be explained
by its better catalytic properties compared to the other catalysts:
higher Ni dispersion (Ni0 active phase formed from finer NiAlx Oy
spinels after reduction treatment), as indicated by the results
obtained from the analysis of TPR-H2 and XRD (higher dispersion)
and XPS (higher metallic surface exposure). A high proportion of
these thin particles were reduced at relatively lower temperatures
due to the H2 /Pt-spillover phenomenon. This effect avoids their sin-
tering, and as a result a higher number of Ni active sites on the
catalyst were available. XRD also confirmed that Pt and Ni (see
Fig. 4) were present on the catalyst surface as thin particles, and
or Ni by dehydrogenation, followed by C C cleavage to give H2
and adsorbed CO, which is readily converted to CO2 by the WGS
reaction.
The poor results recorded by PtCoAl(SGA) and PtFeAl(SGA) cat-
alysts could be related, on the one hand, to lower Pt dispersion
(bigger Pt particles, around 46 nm) for PtCo catalyst and also to
the lower temperature (350 ◦ C) used to reduce the PtFe catalyst
because it was not severe enough to reduce iron active sites.
According to Davda et al. [9], these metals record higher activity
for methanation and Fischer–Tropsch reactions and lower activ-
ity for C C/C H bond cleavage. Thus, catalyst formulations with
these metals give lower activities for glycerol APR than PtNiAl(SGA)
catalysts.
The low glycerol conversions obtained using the Pt monometal-
lic catalyst may be due to its smaller metal loading (2.6 wt%) than
Ni or Pt/Ni loadings 8 wt% for NiAl(SGA) and 11.2 wt% for PtNiAl(SGA)
catalysts, approximately. Its pronounced catalytic stability could
be due to high Pt resistance to oxidation, which is a characteristic
of noble metals (Pt active centers remained in their reduced state
because of the H2 generated throughout the APR reaction). This sta-
bility could also be related to a possible oxidation/reduction cycle
of Pt, since the XPS analysis showed that Pt was not completely
reduced just after the initial reduction treatment, and before APR
started. This part of Pt might be reduced by the H2 generated by the
reactions taking place and replace the oxidized Pt on the surface,
maintaining the catalytic activity of the catalyst relatively stable. On
the other hand, the gradual decrease in catalytic activity shown by
Ni-containing catalysts along time-on-stream, as well as the quan-
titative and qualitative changes of the OHCs yields distribution until
the overall reaction process reached its steady state might be an
 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136
indication of acid and Ni active centers being modified. The deacti-
vation of Ni species was also observed by Wen et al. [4], who found
that Ni atoms are highly susceptible to oxidation by water excess
under APR thermodynamic conditions [26]. In addition, the overall
carbon balance of these catalysts recorded a difference of 8–16%
between the carbon in the inlet and outlet streams. It is quiet prob-
able that part of this carbon ended as light carbonaceous molecules
deposited on the catalyst surface, which may have led to catalyst
deactivation.
The values of H2 selectivity registered (Fig. 8) indicate that the
catalysts generated higher H2 yields at 230 ◦ C/30 bar. This sug-
gests that operating at relatively moderate temperatures/pressures
is much more selective to H2 formation by the APR of glyc-
erol. This can be related to the effect of pressure, since lower
pressures implies more favorable thermodynamic conditions
for reactions increasing the total mole number in gas phase
(C3 H8 O3 + 3H2 O → 7H2 + 3CO2 ). Nevertheless, as it can be seen
in Table 3 and Fig. 8 the catalysts presented higher glycerol
conversions at the highest reaction temperature and its cor-
responding total pressure (250 ◦ C/50 bar) and produced higher
gas flow rates as compared to the ones obtained operating at
230 ◦ C/30 bar, which is to be expected as the APR is an endothermic
process.
As it has been showed, a comparison among Pt/Ni catalysts pre-
pared by acid sol–gel (SGA) and those prepared by basic sol–gel
(SGB) and incipient wetness impregnation (IWI), presented previ-
ously in [26], revealed that catalytic properties varied depending
on the preparation method. The textural characteristics of the
SGA-prepared ␥-alumina supported Pt/Ni catalysts are consid-
erably different as their superficial area is approximately 50%
higher than the ones of equivalent IWI-prepared catalysts and
50% lower than the ones measured for similar SGB catalysts [26].
The Pt/Ni SGA catalysts show micro/mesoporosity, while the SGB
catalysts are clearly mesoporous materials. In fact, both sol–gel
methods lead to Ni spinel-type catalysts ensuring a very good dis-
persion of Pt/Ni metallic particles, which is not achieved when
these materials are prepared by IWI method. The SGA method
ensures a better compromise between platinum–nickel interac-
tions at lower temperatures, as TPR-H2 indicated, but the SGB
catalysts are more resistant to the sintering effect after reduction
treatment at high temperatures, as corroborated by XRD and XPS
analysis.
The results obtained using PtNiAl(SGA) catalyst prepared by acid
sol–gel are much better than those recorded by a similar PtNi-based
␥-Al2 O3 catalyst prepared by incipient impregnation [15,26], but
they present slightly lower catalytic activities than those obtained
using the PtNiAl(SGB) catalyst reported in [26]. This can be directly
related to the catalytic properties developed by each preparation
method. However, the acid sol–gel method may still be an interest-
ing alternative for preparing promoted catalysts for glycerol APR.
Comparing the results reported in this paper to those presented in
several previous investigations [1,2,4–7,9–13,15,22,24], they also
represent a promising and interesting advance in the scientific
knowledge, especially if they are jointly considered with previ-
ous research published in [26]. As examples, operating at similar
operating conditions (225 ◦ C/40 bar and 240 ◦ C/40 bar), A. Iriondo
et al. [15] obtained gas flow rates of 2 and 6.5 mL(STP) min−1 g−1 cat
using an impregnated PtNi/␥-Al2 O3 catalyst (steady state (27–30 h)
and WHSV = 2 h−1 ), which are lower than those obtained in this
work (13 and 19 mL(STP) min−1 g−1 cat ) testing the PtNiAl
(SGA) cata-
lyst at a temperature/pressure of 230 bar/30 ◦ C and 250 ◦ C/50 bar,
respectively (steady state (30–32 h) and WHSV = 2.6 h−1 ). This is a
significant improvement in the development of more active and
stable catalytic systems for H2 production throughout the APR pro-
cess of glycerol.
135
5. Conclusions
The following catalytic systems: PtFe, PtCo, PtNi, Ni or Pt/␥-
Al2 O3 prepared by an acid sol–gel procedure were used for glycerol
aqueous phase reforming to produce H2 at low temperatures and
moderate pressures. The main conclusions of this study are:
PtNi bimetallic systems are clearly more active than
monometallic (Pt or Ni) or PtFe and PtCo systems for APR reactions,
even if all of them were prepared using an acid sol–gel procedure
and similar metal loadings were attained.
Based on reported findings [26] and on the results of this study,
it was confirmed that the properties of the catalysts studied were
directly related to the preparation method. The physicochemical
properties of PtNi catalysts prepared by the acid sol–gel method
were greatly improved as compared to similar catalysts prepared by
conventional impregnation [1,2,4–7,9–13,15,22,24]. The prepara-
tion under acid conditions promoted hydrolysis reactions resulting
in the formation of longitudinal fibrous materials, characterized
by high surface area and micro/mesoporosity, which considerably
favored the dispersion of Pt and Ni particles. The preparation under
basic conditions studied previously [26] increased the condensa-
tion reactions and lead to cluster materials, characterized by a
spherical shape and mesoporosity.
The TPR-H2 , XRD and XPS studies of PtNi/␥-Al2 O3 catalysts sug-
gest that Pt presence improved Ni reducibility by H2 spillover, and
that part of the Pt and part of the Ni ended up interacting to gen-
erate a promoting effect between both metals, as suggested by
TPR-H2 and XPS and confirmed by TEM/EDX analysis. This effect
was clearly manifested in catalytic activity. All these factors explain
why PtNiAl(SGA) catalytic systems had interesting and promising
properties for glycerol APR. This catalyst showed better catalytic
activity than similar catalysts prepared using conventional proce-
dures, being especially active for gaseous product formation from
glycerol reforming under liquid-phase conditions. This system is
also a good catalyst for WGS reaction, as negligible amounts of
CO were detected. In fact, the PtNiAl(SGA) catalyst prepared by acid
sol–gel presented promising results, and further work on this sys-
tem could allow the development of more active and selective
catalysts for APR processes.
Acknowledgements
This work was carried out in the laboratory of the “Sus-
tainable Process Engineering” research group (University of the
Basque Country (UPV/EHU), Spain) in collaboration with the “UCCS”
research unit (University of Science and Technology of Lille, France),
supported by funds from the Spanish Ministry of Science and
Innovation (Project: ENE2011-23950) and the Basque Government.
The authors gratefully acknowledge both institutions for providing
financial support, as well as express their gratitude to L. Burylo for
XRD analysis, A. Addad for Microscopy analysis and M. Trentesaux
and A. Beaurain for XPS analysis. The European Regional Develop-
ment Fund (ERDF), CNRS, Région Nord Pas-de-Calais and the French
Ministry for Higher Education and Research are acknowledged for
funding the XPS/LEIS/ToF-SIMS spectrometers within the Regional
Platform of Surface Analysis.
References
[1] D.O. Özgür, B.Z. Uysal, Biomass Bioenergy 35 (2011) 822–826.
[2] Y. Guo, X. Liu, M.U. Azmat, W. Xu, J. Ren, Y. Wang, G. Lu, Int. J. Hydrogen Energy
37 (2012) 227–234.
[3] A. Iriondo, V.L. Barrio, J.F. Cambra, P.L. Arias, M.B. Güemez, R.M. Navarro, M.C.
Sanchez-Sanchez, J.L.G Fierro, Top. Catal. 49 (2008) 46–58.
[4] G. Wen, Y. Xu, H. Ma, Z. Xu, Z. Tian, Int. J. Hydrogen Energy 33 (2008) 6657–6666.
[5] R.L. Manfro, A.F. Da Costa, N.F.P. Ribeiro, M.M.V.M. Souza, Fuel Process. Technol.
92 (2011) 3330–3335.
136 M. El Doukkali et al. / Journal of Molecular Catalysis A: Chemical 368–369 (2013) 125–136
[6] N. Luo, X. Fu, F. Cao, T. Xiao, P.P. Edwards, Fuel 86 (2008) 3483–3489.
[7] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 9 (2006) 4044–4098.
[8] S. Adhikari, S.D. Fernando, A. Haryanto, Energy Convers. Manage. 50 (2009)
2600–2604.
[9] R.R. Davda, J.W. Shabaker, G.W. Huber, R.D. Cortright, J.A. Dumesic, Appl. Catal.
B: Environ. 56 (2005) 171–186.
[10] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964–967.
[11] G.W. Huber, J.W. Shabaker, J.A. Dumesic, Science 300 (2003) 2075–2077.
[12] J.W. Shabaker, R.R. Davda, G.W. Huber, R.D. Cortright, J.A. Dumesic, J. Catal. 215
(2003) 344–352.
[13] J.W. Shabaker, R.R. Davda, G.W. Huber, R.D. Cortright, J.A. Dumesic, Catal. Lett.
88 (2003) 1–8.
[14] I.O. Cruz, N.F.P. Ribeiro, D.A.G. Aranda, M.M.V.M. Souza, Catal. Commun. 9
(2008) 2606–2611.
[15] A. Iriondo, J.F. Cambra, V.L. Barrio, M.B. Güemez, P.L. Arias, M.C. Sanchez-
Sanchez, R.M. Navarro, J.L.G. Fierro, Appl. Catal. B: Environ. 106 (2011) 83–93.
[16] A. Wawrzetz, B. Peng, A. Hrabar, J.A. Lercher, J. Catal. 269 (2010) 411–420.
[17] Z. Wang, Q. Liu, J. Yu, T. Wu, G. Wang, Appl. Catal. A: Gen. 239 (2003) 87–94.
[18] I. Gandarias, P.L. Arias, J. Requies, M. El Doukkali, M.B. Güemez, J. Catal. 282
(2011) 237–247.
[19] I.H. Cho, S.B. Park, S.J. Cho, R. Ryoo, J. Catal. 173 (1998) 295–303.
[20] L. Hu, K.A. Boateng, J.M. Hill, J. Mol. Catal. A: Chem. 259 (2006) 51–60.
[21] A. Vazquez, T. Lopez, R. Gomez, X. Bokhimi, J. Mol. Catal. A: Chem. 167 (2001)
91–99.
[22] A. Tanksale, C.H. Zhou, J.N. Beltramini, G.Q. Lu, J. Incl. Phenom. Macrocycl. Chem.
65 (2009) 83–88.
[23] J.A. Schwarz, C. Contescu, A. Contescu, Chem. Rev. 95 (1995) 477–510.
[24] K. Lehnert, P. Claus, Catal. Commun. 9 (2008) 2543–2546.
[25] G.W. Huber, J.A. Dumesic, Catal. Today 111 (2006) 119–132.
[26] M. El Doukkali, A. Iriondo, P.L. Arias, J. Requies1, I. Gandarias, L. Jalowiecki-
Duhamel, F. Dumeignil, Appl. Catal. B: Environ. 125 (2012) 516–529.
[27] A. Iriondo, V.L. Barrio, M. El Doukkali, J.F. Cambra, M.B. Güemez, J. Requies, P.L.
Arias, M.C. Sanchez-Sanchez, R.M. Navarro, J.L.G Fierro, Int. J. Hydrogen Energy
37 (2012) 2028–2036.
[28] M. El Doukkali, A. Iriondo, P.L. Arias, J.F. Cambra, I. Gandarias, V.L. Barrio, Int. J.
Hydrogen Energy 37 (2012) 8298–8309.
[29] S. Panyad, S. Jongpatiwut, T. Sreethawong, T. Rirksomboon, S. Osuwan, Catal.
Today 174 (2011) 59–64.
[30] A. Sirijaruphan, J.G. Goodwin Jr., R.W. Rice, D. Wei, K.R. Butcher, G.W. Roberts,
J.J. Spivey, Appl. Catal. A: Gen. 281 (2005) 11–18.
[31] V. Hacker, R. Fankhauser, G. Faleschini, H. Fuchs, K. Friedrich, J. Power Sources
86 (2000) 531–535.
[32] D. Schanke, S. Vada, E.A. Blekkan, A.M. Hilmen, A. Hoff, A. Holman, J. Catal. 156
(1995) 85–95.
[33] F.B. Noronha, et al., Catal. Today 28 (1996) 147–157.
[34] T. Huizinga, J.V. Grondelle, R. Prins, Appl. Catal. 10 (1984) 199–213.
[35] N. Ichikuni, D. Murata, S. Shimazu, T. Uematsu, Catal. Lett. 69 (2000) 33–36.
[36] J.M. Rynkowski, T. Paryjczak, M. Lenik, Appl. Catal. A: Gen. 106 (1993) 73–82.
[37] M.C. Sánchez-Sánchez, R.M. Navarro, J.L.G. Fierro, Int. J. Hydrogen Energy 32
(2007) 1462–1471.
[38] J.C.S Araujo, D. Zanchet, R. Rinaldi, U. Schuchardt, C.E. Hori, J.L.G. Fierro, J.M.C.
Bueno, Appl. Catal. B: Environ. 84 (2008) 552–562.
[39] D. Briggs, M.P. Seah (Eds.), Practical Surface Analysis by Auger and X-Ray Photo-
electron Spectroscopy, vol. 1, 2nd ed., Wiley, Chichester, 1990.
[40] J. Jun, M. Dhayal, H.J. Shin, Y.H. Han, N. Getoff, Appl. Surf. Sci. 254 (2008)
4557–4564.
[41] B. Pawelec, S. Damyanova, K. Arishtirova, J.L.G. Fierro, L. Petrov, Appl. Catal. A:
Gen. 323 (2007) 188–201.
[42] N.H.H. Abu Bakar, M.M. Bettahar, M. Abu Bakar, S. Monteverdi, J. Ismail, M.
Alnot, J. Catal. 265 (2009) 63–71.
[43] N.H.H. Abu Bakar, M.M. Bettahar, M. Abu Bakar, S. Monteverdi, J. Ismail, M.
Alnot, J. Mol. Catal. A: Chem. 308 (2009) 87–95.
[44] M. García-Diéguez, I.S. Pieta, M.C. Herrera, M.A. Larrubia, L.J. Alemany, J. Catal.
270 (2010) 136–145.
[45] M. García-Diéguez, I.S. Pieta, M.C. Herrera, M.A. Larrubia, L.J. Alemany, Appl.
Catal. A: Gen. 377 (2010) 191–199.
[46] M. García-Diéguez, E. Finocchio, M.A. Larrubia, L.J. Alemany, Guido Busca, J.
Catal. 274 (2010) 11–20.