New approaches to the Pt/WOx/Al2O3 catalytic system behavior for the selective glycerol hydrogenolysis to 1,3-propanediol

Article abstract of DOI:10.1016/j.jcat.2014.12.028

Although the hydrogenolysis of glycerol to 1,2-propanediol is already well developed, the production of the more valuable 1,3-propanediol is still a challenge. To achieve this aim, it is essential to design catalysts showing high selectivity toward the CO cleavage of the secondary hydroxyl group in glycerol. In this work, two different series of Pt/WO_x/Al₂O₃ catalytic systems were studied for the selective hydrogenolysis of glycerol to 1,3-propanediol. The results reveal the necessity to control the tungsten surface density in order to obtain highly dispersed polytungstate species, which are able to produce Br?nsted acidity and are involved in the selective formation of 1,3-propanediol. After optimization of the tungsten surface density, the effect of platinum content was also studied. It was found that by improving the interactions between platinum and tungsten oxides, it is possible to increase the selectivity toward 1,3-propanediol. Under optimized conditions, a selectivity toward 1,3-PDO of 51.9% at 53.1% glycerol conversion was obtained. Based on the characterization and activity test results, a reaction mechanism for the Pt-WO_x catalytic system in glycerol hydrogenolysis to 1,3-propanediol was also proposed.

Full text of DOI:10.1016/j.jcat.2014.12.028

Journal of Catalysis 323 (2015) 65–75
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
New approaches to the Pt/WOx/Al2O3 catalytic system behavior for the
selective glycerol hydrogenolysis to 1,3-propanediol
a
a
a
b
b
a
S. García-Fernández a, , I. Gandarias , J. Requies , M.B. Güemez , S. Bennici , A. Auroux , P.L. Arias
⇑
a School of Engineering (UPV/EHU), Alameda Urquijo Street s/n, 48013 Bilbao, Spain
b Université Lyon 1, CNRS, UMR 5256, IRCELYON, Institut de Recherches sur la Catalyse et l’Environnement de Lyon, 2 Avenue Albert Einstein, F-69626 Villeurbanne, France
a r t i c l e i n f o
a b s t r a c t
Article history:
Although the hydrogenolysis of glycerol to 1,2-propanediol is already well developed, the production of
the more valuable 1,3-propanediol is still a challenge. To achieve this aim, it is essential to design cata-
lysts showing high selectivity toward the CO cleavage of the secondary hydroxyl group in glycerol. In this
work, two different series of Pt/WOx/Al2O3 catalytic systems were studied for the selective hydrogenol-
ysis of glycerol to 1,3-propanediol. The results reveal the necessity to control the tungsten surface density
in order to obtain highly dispersed polytungstate species, which are able to produce Brönsted acidity and
are involved in the selective formation of 1,3-propanediol. After optimization of the tungsten surface den-
sity, the effect of platinum content was also studied. It was found that by improving the interactions
between platinum and tungsten oxides, it is possible to increase the selectivity toward 1,3-propanediol.
Under optimized conditions, a selectivity toward 1,3-PDO of 51.9% at 53.1% glycerol conversion was
obtained. Based on the characterization and activity test results, a reaction mechanism for the Pt–WOx
catalytic system in glycerol hydrogenolysis to 1,3-propanediol was also proposed.
Received 3 June 2014
Revised 9 December 2014
Accepted 25 December 2014
Keywords:
Glycerol
Hydrogenolysis
1,3-Propanediol
Tungsten oxide
Surface density
Bifunctional catalyst
Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction
substitute of ethylene glycol as antifreeze fluid in food applications
due to its non-toxicity [2]. Interestingly, 1,3-PDO compound has an
The development of new transformation processes to convert
biomass into fuels and high added value chemicals is a real neces-
sity in order to reduce the petroleum dependence and the carbon
footprint of modern societies. In this respect, glycerol appears as
one of the top 12 biomass-derived building blocks in the biorefin-
ery industry for the production of a wide range of commodity
chemicals [1]. In the last years, glycerol availability has immensely
increased, promoted by the biodiesel production via transesterifi-
cation of vegetable and animal oils which generates large amounts
of glycerol as by-product (10.5 kg of glycerol for 100 kg biodiesel).
The huge amount of glycerol placed in the market cannot be
absorbed by its conventional uses and ends up becoming a residue.
Therefore, there is a great interest in the valorization of the
biomass-derived glycerol.
even greater added value than 1,2-PDO, as it is used as a monomer
together with terephthalic acid to produce polytrimethylene tere-
phthalate (PTT), a polymer whose unique properties make it very
attractive in a wide range of uses.
There is a considerable research dealing with the catalytic con-
version of glycerol to PDOs in which high yields of 1,2-PDO have
been reported, using typically bifunctional systems formed by a
hydrogenation metal and an acid or base cocatalyst [3–7]. Never-
theless, the production of 1,3-PDO is more challenging. The control
of the 1,2-PDO/1,3-PDO ratio, which requires highly selective C–O
cleavage of the primary or secondary hydroxyl groups of glycerol,
is strongly dependent on the catalyst used. Thus, there is a need
for the development and understanding of new selective catalytic
systems in order to obtain the more valuable 1,3-PDO.
The hydrogenolysis of glycerol to obtain propanediols (PDOs),
both 1,2-propanediol (1,2-PDO) and 1,3-propanediol (1,3-PDO),
has attracted a great deal of attention in the recent years.
1,2-PDO is widely used in the manufacture of a broad array of
industrial and consumer products, and it has become the
Chaminand et al. [8] found that the addition of H2WO4 to the
sulfolane media in the glycerol hydrogenolysis using Rh/SiO2
enhanced the yield of 1,3-PDO (4%) at 200 °C and 80 bar and also
in aqueous media (3%). However, from an industrial and environ-
mental perspective, the use of heterogeneous catalysts is preferred.
Kurosaka et al. [9] obtained a 24% yield of 1,3-PDO in 1,3-dimethyl-
2-imidazolidinone (DMI) solvent over Pt/WO3/ZrO2 at 170 °C and
80 bar. Huang et al. [10] achieved a 27% yield of 1,3-PDO in vapor
phase and aqueous media using Cu–H4Si12O40SiW/SiO2 catalyst at
⇑
Corresponding author.
E-mail address: sara_garcia@ehu.es (S. García-Fernández).
http://dx.doi.org/10.1016/j.jcat.2014.12.028
0021-9517/Ó 2015 Elsevier Inc. All rights reserved.

66
S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
5 bar and 210 °C. A significantly higher 1,3-PDO yield (56%) was
attained by Oh et al. [11] with Pt deposited on the super-acid sul-
fated ZrO2 support, in DMI solvent at 170 °C and 73 bar. All these
previous works present some drawbacks, such as the use of organic
solvents or the work in gas phase, which will greatly reduce the
environmental and economic viability of the process. Water is
the ideal solvent for the process as glycerol is obtained in aqueous
phase after the transesterification reaction. [12].
((Pt(NH3)4(NO3)2, Sigma–Aldrich, P99.995%)) as precursor. The
resulting catalysts were dried and calcined as above. These Pt/
WO3/Al2O3 samples are denoted as xPtyW, where x refers to the
platinum content in weight percent (wt%) in the final catalyst
and y to the tungsten content related to the alumina support (in
terms of wt% of W/c-Al2O3), both of them measured by ICP. The
tungsten surface density (expressed in W atoms nmꢁ2 of support)
was calculated based on the following equation,
ꢀ
ꢁ
Up to date, the most effective approach in the production of 1,
3-PDO in aqueous phase has shown to be the use of heterogeneous
catalysts formed by a noble metal (Ir, Rh, or Pt) combined with
oxophilic metals such as Mo, Re, and W [13]. In fact, one of the
most complete works was carried out by Tomishige’s group using
ReOx-modified Rh/SiO2 [14] and Ir/SiO2 [15–17] catalytic systems,
obtaining with this last catalyst a 38% 1,3-PDO yield at 120 °C
and 80 bar, using small amounts of H2SO4 as an additive. However,
the weak interactions of ReOx species with the silica support and
the high solubility of these species in water favor Re leaching
under the reaction conditions, which compromises the stability
of such catalysts [18]. Thus, more robust and stable tungsten Pt–
WO3 based catalytic systems appear as a better option. In 2010,
Quin et al. [12] reported a high yield of 1,3-PDO (32%) using a Pt/
WO3/ZrO2 catalyst at 130 °C and 40 bar. Zhu et al. [19] investigated
the performance of zirconia supported bifunctional catalysts con-
taining Pt and heteropolyacids, obtaining 31% of 1,3-PDO yield
with Pt–HSiW/SiO2 catalyst at 180 °C and 50 bar. More recently,
Arundhathi et al. [20] obtained the highest 1,3-PDO yield (66%)
reported to date using Pt/WO3/‘‘AlOOH’’ catalytic systems. The
high yield obtained was attributed by these authors to the plentiful
Al–OH groups in the boehmite support, but the high temperatures
used in the catalyst pretreatment and the XRD results indicate a
xW
MW
NA
qW½W atoms=nm2 of supportꢂ ¼
ꢄ
ꢂ
ꢃꢅ
SAAl
1 ꢁ xWO þ xPtO
2O3
3
where xW is the mass fraction of the W species in the final catalyst,
NA is the Avogadro number, MW is the W atomic weight, and SAAl
2O3
is the initial BET surface area of the calcined
c-alumina support
used. In this case, it was assumed that all the tungsten oxide species
presented in the catalyst were WO3 and the platinum oxides were
PtO.
2.2. Catalyst characterization
2.2.1. Chemical analysis
The chemical analysis of the catalyst was carried out by induc-
tively coupled plasma atomic emission (ICP-AES) using a Perkin–
Elmer Optima 2000 instrument. Previously to the analysis, the
solid samples were digested in a microwave oven in a mixture of
HF, HCl, and HNO3 heating from room temperature up to 180 °C
during 30 min.
2.2.2. N2 physisorption
Textural properties (surface area, pore volume, and pore size
distributions) were obtained by N2 physisorption at ꢁ196 °C using
a Quantachrome AUTOSORB-1C-TCD instrument. All samples were
dried at 300 °C overnight under high vacuum prior to the physi-
sorption measurements. The surface area was calculated using
the Brunauer, Emmett, and Teller (BET) method, and the pore size
distributions were obtained using the Barrett–Joyner–Halenda
(BJH) method applied to the desorption branch of the isotherms.
different alumina structure than boehmite (possibly c-Al2O3) and
therefore the absence of many hydroxyl groups. In spite of these
previous works, the overall C–O hydrogenolysis mechanism as well
as the role of tungsten and platinum still remains unclear [21].
In this study, two different series of Pt/WOx/Al2O3 catalysts
were prepared, with different WOx and Pt contents, in order to
understand the role of each active phase. The obtained glycerol
conversion and PDOs selectivity values were related to the physi-
cochemical properties of the catalysts measured by N2 physisorp-
tion, H2 temperature programmed reduction (H2-TPR), Fourier
transform infrared spectroscopy (FTIR) of adsorbed pyridine,
Raman spectroscopy, X-ray diffraction (XRD), X-ray photoelec-
tronic spectroscopy (XPS), NH3 adsorption calorimetry, transmis-
sion electron microscopy (TEM), and CO chemisorption
2.2.3. CO chemisorption
The measurements were performed in an AUTOSORB-iQ equip-
ment. Prior to adsorption, all samples were outgassed in He flow at
120 °C for 3 h and subsequently reduced at 450 °C under a stream
of pure H2 for 1 h (reaction conditions). The samples were cooled
down to room temperature and evacuated under He flow for 2 h.
After that, CO chemisorption uptakes were measured by pulses of
pure CO at 40 °C.
techniques. The results from this research allow
a better
understanding of the behavior of Pt/WOx/Al2O3 catalytic systems
in glycerol hydrogenolysis.
2.2.4. TEM
TEM images were obtained in a Philips SuperTwin CM200 appa-
ratus operated at 200 kV and equipped with LaB6 filament and
EDAX EDS microanalysis system. The reduced samples (at reaction
conditions) were prepared via dispersion into ethanol solvent and
placed on a carbon-coated copper grid (300 Mesh) followed by
drying under vacuum.
2. Experimental
2.1. Catalyst preparation
Pt/WOx/Al2O3 catalysts were prepared by sequential wetness
impregnation method. The typical procedure followed for the
2.2.5. XRD
preparation of the catalysts is detailed below.
c-Al2O3 (Sigma–
XRD studies of the reduced catalyst were recorded on an Xpert-
Pro instrument with a PW3050/60 goniometer and a Cu anode at
current of 40 mA and voltage of 40 kV, in a 2h range from 10° to
90° with a 0.026° step size. The patterns were compared with
the power diffraction files (PDF) by Xpert-Pro High Score tool.
Aldrich, P99.9%) was used as support, and it was impregnated
using the appropriate amounts of ammonium metatungstate
((NH4)6(H2W12O40)ꢀnH2O, Sigma–Aldrich, P99.99%)) dissolved in
deionized water.
Impregnated samples were dried at 110 °C overnight and subse-
quently calcined in air from room temperature up to 450 °C at a
heating rate of 2 °C minꢁ1, maintaining this temperature for 4 h.
Pt was then loaded on supported tungsten oxide catalysts by
wetness impregnation using tetraammineplatinum(II) nitrate
2.2.6. Raman spectroscopy
The Raman spectra of the calcined catalysts were determined at
ambient conditions using samples in powder form, using
a

S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
67
Renishaw InVia microscope with Ar ion laser (Modu-Laser)
2.3. Activity test
employing 514 nm laser excitation, in order to determine the spe-
cies present in the samples and the interaction between the plati-
num and tungsten species. The spectra were recorded at ambient
temperature within the 150–1500 cmꢁ1 region.
Activity tests were carried out in a 50-mL stainless steel auto-
clave with a magnetic stirrer at 550 rpm. Typically, 0.35 g of cata-
lyst powder (300–500 lm) was introduced into a catalytic basket
inside the autoclave. Prior to the activity test, the catalyst was in
situ reduced under a stream of pure H2 (100 mL/min) at 450 °C
for 1 h. After reduction, the reactor was cooled down to the reac-
tion temperature (200–220 °C) and the H2 pressure was increased
up to 20 bar. The reactant solution (42 mL of 5 wt% glycerol aque-
ous solution), placed into a feed cylinder at 50 bar H2 pressure, was
fed into the reactor taking advantage of the pressure difference and
by circulating a continuous flow of H2 from the cylinder to the
reactor for about 5 min. Thereafter, the pressure was increased
up to the 45 bar operation pressure.
2.2.7. XPS
XPS spectra of the reduced samples (fresh and some after use)
were obtained on a VG Escalab 200R spectrometer equipped with
a
hemispherical electron analyzer an Mg Ka (hm = 1253.6 eV)
X-ray source. Binding energy (Eb) values were referred to the C1s
peak at 284.6 eV. The recorded XPS signals were analyzed using a
nonlinear Shirley-type background and decomposed into subcom-
ponents by Gaussian–Lorentzian functions.
2.2.8. H2-TPR
Along the first 7 h of reaction time, 4 liquid samples were taken
and a last sample was taken at 24 h reaction time. These samples
were analyzed using 1,4-butanediol (Sigma–Aldrich, 99.99%) as
internal standard with a gas chromatograph (Agilent Technologies,
7890 A) equipped with a Meta-Wax capillary column (diameter
0.53 mm, length 30 m), a flame ionization detector (FID), and a
thermal conductivity detector (TCD). After 24 h of reaction, the
gas phase was collected in a gas bag and analyzed with another
GC-TCD-FID equipped with a molecular sieve column (HP-MOLESI-
EVE, diameter 0.535 mm, length 30 m) and a capillary column
(HPPLOT/Q, diameter 0.320 mm, length 30 m).
The reducibility study of the surface species was carried out on
a Quantachrome AUTOSORB-1C-TCD apparatus. TPR profiles were
obtained using a catalyst powder amount of 0.3 g. The samples
were reduced in a flowing gas of 5 vol% H2 in Ar (50 mL minꢁ1 of
total flow rate) increasing the temperature from 40 °C up to
1100 °C at 10 °C minꢁ1. A TCD detector downstream of the sample
monitored changes in the concentration of H2.
2.2.9. NH3 adsorption microcalorimetry
Adsorption microcalorimetry measurements were performed at
80 °C in a volumetric line linked to a heat-flow microcalorimeter
(Tian-Calvet type, C80 from Setaram) in order to measure the
whole acidity as well as the strength of the catalytic sites present
in the samples. Ammonia (NH3) was used as a basic probe mole-
cule to titrate the surface acid sites. An amount of 0.1 g of powder
calcined sample was pretreated under vacuum at 250 °C overnight
(about 12 h) and subsequently evacuated at the same temperature
for 1 h prior to the measurements. Small successive NH3 injections
were sent onto the sample until a final NH3 equilibrium pressure of
about 67 Pa. For each dose, the adsorption heat and the pressure of
the adsorbate, this last measured by means of a differential Barocel
capacitance manometer (Datametrics), were recorded by a com-
puter. Following the first adsorption isotherm, the sample was out-
gassed under vacuum for 30 min at the same temperature and a
second adsorption isotherm was performed until the equilibrium
pressure of 27 Pa was attained. The irreversible volume of NH3
(Virr), associated with the strongly chemisorbed NH3, was calcu-
lated from the difference between the first and second volumetric
isotherms at 27 Pa.
The conversion of glycerol and the selectivity of liquid products
were calculated based on the following equations:
P
C-based mol of all liquid productt¼t
Conversion of glycerol ð%Þ ¼
C-based mol of glycerolt¼0
C-based mol of the product
C-based mol of all liquid products
P
Selectivity of liquid product ð%Þ ¼
3. Results and discussion
3.1. Catalyst characterization
3.1.1. N2-physisorption
The textural properties of the calcined catalysts are shown in
Table 1. The catalysts surface area was reduced in all the catalysts
as compared to the
tungsten surface density increased. This could be due to the block-
ing or filling of -alumina pores when the tungsten oxide species
c
-alumina support (SABET of 140 m2 gꢁ1) as the
c
were added. The reduction in pore volume of the catalysts in com-
parison with the bulk support (pore volume of 0.23 cm3 gꢁ1) can be
explained through the same assumption. The incorporation of the
platinum oxide on the tungstated support had the same effect.
2.2.10. FTIR of adsorbed pyridine
FTIR of adsorbed pyridine analysis was carried out in order to
study the nature (type and strength) of acid sites on the surface
of different WOx/Al2O3 and Pt/WOx/Al2O3 samples. Infrared trans-
mission spectra were recorded on Nicolet 8700 FTIR spectrometer
equipped with an extended KBr beam splitter and a mercury cad-
mium telluride (MCT) detector. The samples were pressed into a
30–35 mg of self-supporting wafers, having a surface area of about
2 cm2, which were placed into an infrared quartz cell, with CaF2
windows, connected to a vacuum line. Pellets were pretreated
under vacuum (about 10ꢁ5 Pa) with an oxygen flow and by a slow
heating of 2 °C minꢁ1 up to 350 °C. Then, they were outgassed at
the same temperature for 1 h. A known amount of pyridine was
introduced into the cell at room temperature. Spectra were
recorded after pyridine desorption by evacuation for 15 min under
vacuum at different temperatures (200 °C and 300 °C) with a spec-
tral resolution of 2 cmꢁ1 and 32 scans per spectra. All reported
spectra were obtained by subtracting the spectrum of the activated
catalyst (after pretreatment but before pyridine adsorption) from
those after pyridine adsorption.
3.1.2. CO chemisorption
The tungsten oxides present in the Pt/WOx/Al2O3 catalysts do
not adsorb CO [22]; therefore, CO chemisorption was used to mea-
sure the platinum particle size and metal dispersion on platinum
loaded samples. Depending on the type of adsorption sites, the
average of CO/Pt adsorption stoichiometry usually varies from 1
to 2 because CO chemisorbs in various forms [23]. The most widely
CO/Pt adsorption stoichiometry used for the calculation of plati-
num size and metal dispersion is a ratio of 1 [24,25]. In order to
facilitate the comparison with other works, the same value was
assumed in this article. The results are shown in Table S1. It can
be noted that increasing the platinum loading, the metal size nota-
bly increased, from 4 to 26 nm for 2Pt8W and 9Pt8W catalyst,
respectively, and with the resultant decline in platinum dispersion
(from 27% to 4%, respectively).

68
S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
Table 1
Textural properties of the catalyst and its activity results obtained after 24 h reaction time, 220 °C, and 45 bar pressure.
Catalysta
SABET (m2 gꢁ1
)
Pore volume (cm3 gꢁ1
)
Glycerol conversion (%)
Product selectivity (%)
1,3-PDO
1,2-PDO
1-PO
Degradation productsb
1Pt1W (0.2)
1Pt5W (1.0)
1Pt9W (2.4)
1Pt14W (3.9)
1Pt19W (5.9)
2Pt8W (2.2)
4Pt8W (2.2)
9Pt8W (2.2)
9Pt8W (2.2)c
128
111
104
96
79
105
–
0.24
0.22
0.19
0.18
0.15
0.19
–
25.6
22.2
16.7
11.6
5.5
25.6
42.1
60.3
53.1c
0.0
3.1
78.2
76.3
57.5
60.9
59.5
39.3
25.9
11.0
9.5c
1.9
3.5
19.9
17.2
11.3
7.5
0.0
6.1
16.4
10.9
11.2
25.7
27.8
31.2
51.9c
12.4
20.7
29.3
21.5
25.9
35.6
24.8c
7.8
81
81
0.15
0.15
12.0
1.2c
Tungsten surface density in brackets (W at. nmꢁ2).
Main degradation products: ethylene glycol, ethanol, methane.
Results obtained for 9Pt8W (2.2) catalyst at 200 °C reaction temperature.
a
b
c
Fig. 1. TEM images of fresh reduced (a) 1Pt9W and (b) 9Pt8W catalysts.
3.1.3. TEM
around 5 nm at high tungsten surface density values. These results
are in good agreement with the Raman spectroscopy and XPS
results. The absence of platinum diffraction peaks revealed the
homogeneous dispersion of this metal on the catalyst.
Fig. 1(a) and (b) shows the TEM images of 1Pt9W and 9Pt8W
catalysts. Although the TEM images do not allow a visual clear dis-
crimination among platinum and tungsten oxide clusters (neither
by shape nor contrast), it is interesting to note that only particles
with diameters 65 nm were found. EDS microanalysis on different
specific particles allowed to differentiate certain platinum particles
from the tungsten oxides at low platinum content. In Fig. 1(a), a
platinum particle (circle) of a diameter lower than 1 nm sur-
rounded by tungsten oxides well dispersed in the catalyst surface
can be seen. For 9Pt8W sample, the differentiation was absolutely
impossible even by EDS microanalysis, which reveals the closer
proximity between these species at higher platinum contents.
Both, platinum and tungsten, characteristic X-ray peaks were
obtained around throughout the sample investigated (Fig. 1(b)).
The particle sizes obtained by TEM are commonly smaller than
those from chemisorption uptakes as Iglesia et al. reported [26]. In
this case, the platinum particle size estimated from TEM (5 nm)
and CO chemisorption (26 nm) measurements is significantly dif-
ferent for the 9Pt8W catalyst.
As it can be observed in the XRD patterns of the catalysts with a
constant tungsten surface density of about 2.2–2.4 W at. nmꢁ2, for
platinum loadings higher than 2 wt%, large and narrow peaks’
characteristics of metallic platinum (2h = 39.9°, 46.4°, 67.6°) [29]
started to be observed. There was no evidence of the presence of
platinum oxides, which means that the complete reduction of the
catalyst was attained at the reduction conditions used. As the plat-
inum content raised from 2 wt% to 4 wt%, the platinum crystallite
size also increased considerably but remained almost constant for
higher platinum contents (9 wt%). This trend was supported by CO
chemisorption and TEM technique results. However, the platinum
particle size values obtained by XRD are considerably higher, prob-
ably due to the most dominant contribution of some big platinum
particles formed on this bulk type analysis and to the very small
clusters being undetectable because of the detection limit of this
technique was 3 nm.
3.1.4. XRD
3.1.5. Raman spectroscopy
The structure of the reduced (at reaction conditions) fresh Pt/
WOx/Al2O3 catalysts was analyzed by XRD (Fig. S1). The XRD pat-
terns of the catalysts with a constant Pt content (1 wt%) and differ-
ent WOx loading reveal that only diffraction peaks related to
Tungsten surface density determines the different types of
tungsten oxides species that are deposited on the surface of the
catalyst, namely surface monotungstate species (coordinated
structure: WO4), polytungstates (coordinated structure: WO5/
WO6), and WO3 crystalline nanoparticles (NPs) [30].
crystalline
c-alumina (2h = 37.5°, 45.4°, 66.9°) [27] were observed
for catalyst with
a
tungsten surface density lower than
Raman spectroscopy was used to differentiate the species pres-
ent on the catalysts surface, and the spectra of calcined catalysts
are shown in Fig. 2. The c-alumina support has no Raman active
2.4 W at. nmꢁ2. However, diffraction peaks’ characteristic of mono-
clinic-WO3 phase (m-WO3) (2h = 23.5, 33.3) were observed for a
tungsten surface density of 3.9 W at. nmꢁ2 [28]. This indicates
the formation of WO3 crystalline nanoparticles (WO3 NPs) of
modes [30]. The Raman bands of Pt/WOx/Al2O3 catalysts appearing
in the 970–1008 cmꢁ1 region can be attributed to W@O vibration

S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
69
For tungstated samples (without platinum), Al2p and W4f were
used to study the Al and W elements, respectively. Fig. 3 shows
the intensity ratio of the W4f and Al2p photoelectrons, (IW/IAl)XPS
,
compared to the bulk W/Al atomic ratio, (W/Al)bulk. The Kerkhof–
Moulijn model (K–M model) has been used mainly to estimate the
maximum concentration of tungsten at which a deviation from
monolayer deposition occurs [36]. It can be observed that experi-
mental and predicted by K–M model (black straight line) values
almost match for tungsten surface density values below
2.4 W at. nmꢁ2
. The fact that the sample with 14 wt% W
(3.9 W at. nmꢁ2) slightly deviates from the linearity reveals that
the WO3 is not only forming a two-dimensional phase in the catalyst
surface, but also a small fraction of WO3 begins to form three-dimen-
sional clusters (WO3 NPs). These results agree with previously
shown XRD results and show that WO3 NPs are formed at tungsten
surface density values below the theoretical monolayer content of
7 W at. nmꢁ2 [33]).
For Pt/WOx/Al2O3 samples, due to the overlapping of Al2p and
Pt4f lines, the Pt4d lines were used to study the platinum species
on the catalyst surface, although this line is weaker. The Eb
obtained in the Pt4d5/2 (314.8–315.9 eV) spectral region for the
investigated samples (see Table 2) shows that the platinum species
are completely reduced under the reduction conditions used [37].
Moreover, the Eb of W4f7/2 peaks can be associated with the exis-
tence of both WO3 [38] and aluminum tungstates (Al2(WO4)3)
[39]. However, due to the relative low calcination temperature
used in catalyst preparation (450 °C), it can be assumed that alumi-
num tungstates were not formed or in negligible proportion.
The XPS peak intensities were used to calculate the atomic rel-
ative contents of the Pt/WOx/Al2O3 reduced samples, which are
given in Table 2. These surface atomic ratios of Pt and W related
to Al were there compared with the bulk ratios obtained from
ICP-AES. For fresh reduced catalysts with low platinum contents
of 1 wt%, it can be observed that Pt/Al surface ratios are rather sim-
ilar to the bulk ratios, which suggests a good platinum dispersion,
even at high tungsten oxide contents (1Pt14W sample). For higher
platinum loadings, considerably smaller atomic surface ratios than
bulk ratios were obtained, which might be a consequence of a
worse platinum dispersion.
Fig. 2. Raman spectra of the calcined catalysts. Tungsten surface density
(W at. nmꢁ2) in brackets.
bands of mono- and polytungstate species [30]. The progressive
shift of the position of this band to higher wavenumbers, from
971 cmꢁ1 to 1008 cmꢁ1, as the tungsten density increases from
0.2 W at. nmꢁ2 up to 3.9 W at. nmꢁ2, can be ascribed to an increase
of the polytungstates to monotungstates ratio [31]. Typical WO3
NP bands at 800, 700, and 250 cmꢁ1 could be observed at densities
P3.9 W at. nmꢁ2, which are assigned to W–O stretching, W–O
bending, and W–O–W deformation modes of octahedrally coordi-
nated WO3 crystals [32]. These NPs correspond to m-WO3 phase
as the XRD results revealed. The formation of WO3 NPs on the cat-
alyst surface is due to the aggregation and sintering of tungsten
oxides before reaching the theoretical c-alumina monolayer cover-
age (7 W at. nmꢁ2). This value was calculated by Barton et al. at
[33] assuming corner sharing WOx octahedra with bond distances
derived from m-WO3. These data suggest that there are a fixed
XPS can provide further information about the metal anchoring
sites. Kwak et al. [40] observed that when platinum was supported
number of binding sites for tungsten oxides on
It is widely suggested this species anchor to the
c
c
-alumina surface.
-alumina support
by titrating the surface hydroxyls [34]. One possible assumption is
that WO3 NPs start forming once all the surface hydroxyls are
titrated. However, we cannot exclude the possibility that the for-
mation of NPs starts before all the anchoring sites are occupied.
At the highest tungsten surface density of 5.9 W at. nmꢁ2, the m-
WO3 is the predominant phase detected by Raman spectroscopy.
For the 9W catalyst, the band corresponding to polytungstates
appears at 973 cmꢁ1. When 1 wt% of platinum was incorporated
into WOx/Al2O3 (1Pt9W), a shift to a higher wavenumber occurred
(from 973 cmꢁ1 to 1003 cmꢁ1). This shift might be due to the inter-
action between platinum and surface tungsten oxides that affects
the distortion of the oxotungsten species. This approach was also
found over Pd–WOx/TiO2 system by Taylor et al. [35]. In Pt/WOx/
Al2O3 catalytic systems, it appears that the transfer of electrons
from the metallic platinum to the WOx increases the W@O bond
strength.
3.1.6. XPS
XPS measurements were carried out to identify the surface
chemical species, its oxidation state and to estimate the metal dis-
persion on the catalyst surface for WOx/Al2O3 and Pt/WOx/Al2O3
reduced samples.
Fig. 3. Intensity ratio, (IW4f/IAl2p)XPS, of the W4f and Al2p photoelectrons against the
bulk atomic ratio, (W/Al)bulk. The black line corresponds to Kerkhof–Moulijn model.
Tungsten surface density (W at. nmꢁ2) in brackets.

70
S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
Table 2
Bulk atomic ratio obtained from ICP-AES, surface atomic ratio obtained from XPS and
BE of the Pt and W elements.
Catalysta
Pt/Al
Bulk
W/Al
Bulk
Eb
Surface
Surface
Pt 4d5/2
W 4f7/2
1Pt4W (1.0)
1Pt9W (2.4)
1Pt14W (3.9)
4Pt8W (2.2)
4Pt8W (2.2)b
9Pt8W (2.2)
9Pt8W (2.2)b
0.003
0.004
0.004
0.012
0.012
0.028
0.028
0.003
0.003
0.003
0.007
0.004
0.010
0.007
0.011
0.027
0.045
0.026
0.026
0.023
0.023
0.029
0.056
0.080
0.032
0.052
0.028
0.041
315.0
314.9
315.0
314.9
314.9
314.8
315.9
35.5
35.6
35.8
35.9
35.8
35.9
35.9
a
Tungsten surface density in brackets (W at. nmꢁ2).
Catalyst after use in glycerol hydrogenolysis reaction at 220 °C.
b
on bare c
-alumina, it was anchored at specific Al3+ sites. In the case
of WOx species, it is widely suggested that they anchor to the c-
alumina support by titrating the surface hydroxyls [34]. However,
we are not aware of studies concerning the sites where platinum is
incorporated on the WOx/Al2O3 surface. The W/Al surface ratio
results of the investigated samples can reveal further information
about these sites. A remarkable decrease in W/Al surface ratio
was found as the platinum loading increased from 1 wt% to
10 wt% (0.056 to 0.028) for the fresh reduced samples. This may
be an indication that the platinum could partially be anchored at
the WOx sites at high platinum loadings, in addition being
anchored to the Al3+ sites. Nevertheless, further research is
required to corroborate this hypothesis. Interestingly, XPS results
of used catalysts (reaction at 220 °C) revealed an important decline
in Pt/Al and an increase in W/Al surface ratios comparing to the
ones of the fresh catalysts, which suggests a potential dispersion
loss of the platinum on the tungsten oxide sites under the hydro-
thermal reaction conditions employed. These reaction conditions
might also affect the oxidation state of Pt species, as the shift of
the Eb of Pt 4d5/2 for the used catalyst with the highest Pt content
(9Pt8W) suggests.
Fig. 4. H2-TPR of the calcined fresh catalysts. Tungsten surface density (W at. nmꢁ2
in brackets.
)
increases, which reveals that Pt and W phases are in closer proxim-
ity, which was also observed by XPS and TEM analysis.
3.1.8. NH3 adsorption microcalorimetry
The amounts of NH3 adsorbed at 80 °C and the corresponding
interaction energies with the catalysts were determined by means
of a volumetry–calorimetry coupled technique. Table 3 gives the
total number of acid sites (estimated by Vtot), the number of strong
acid sites (Virr) as well as the initial differential heat of NH3 adsorp-
tion (Qinit).
The volumetric adsorption isotherms (NH3 uptake vs. equilib-
rium pressure) and differential heats of NH3 adsorption (differen-
tial heats vs. NH3 uptake) are reported in Fig. 5(a) and (b),
respectively. The volumetric isotherms of the studied samples dis-
played in all cases an initial vertical section proportional to the
amount of strongly chemisorbed NH3 [42]. The incorporation of
3.1.7. H2-TPR
The H2-TPR profiles of the catalysts are shown in Fig. 4. In the
temperature range studied (40–1100 °C), the
c-alumina support
cannot be reduced (not shown). Although the complete reduction
of WOx/Al2O3 catalysts requires temperatures higher than
1100 °C, it is possible to verify that the initial reduction temper-
ature decreases as the tungsten surface density increases. This
could indicate that the interaction between WOx species and
the support is stronger at low tungsten surface densities. The
tungsten oxide to the c-alumina support increases the total acidity,
as it is indicated by the total amount of NH3 retained at 27 Pa of
equilibrium pressure (Vtot), as it can be observed in Fig. 5(a) and
Table 3. It has to be noticed that the catalysts with tungsten (with
and without platinum) presented similar capacity to adsorb the
NH3 probe. The small variations might be due to differences in
appearance of a peak at 780 °C for the 14 W (3.9 W at. nmꢁ2
)
catalyst may be related to the presence of WO3 NPs as Raman
spectroscopy and XPS, among other techniques, revealed. The
4Pt catalyst TPR profile shows two main peaks at 133 and
260 °C, which are attributed to the reduction of platinum oxides
[41]. The appearance of small reduction peaks between 400 and
600 °C is probably due to platinum species highly interacting with
the support. The tungsten oxides incorporation (4Pt8W catalyst)
notably reduced the platinum reduction peaks, which were
detected with the maxima at 120, 230, and 470 °C. This decrease
in the reduction temperature of platinum oxides indicates an
interaction between platinum and tungsten oxides, which gener-
ates more reducible platinum species.
Table 3
Initial heats of adsorption, total, and irreversible amounts of NH3 adsorbed at an
equilibrium pressure of 27 Pa, measured by NH3 adsorption microcalorimetry.
Catalysta
Acidity
b
c
d
Qinit (kJ molꢁ1
)
Vtot
(l
mol NH3 gꢁ1
)
Virr (l )
mol NH3 gꢁ1
c
-Al2O3
179
235
219
287
260
278
235
120
153
147
161
129
9W (2.4)
2Pt8W (2.2) 222
4Pt8W (2.2) 224
4Pt
178
Concerning the reduction of WOx species in the catalysts con-
taining platinum, it started at a temperature around 550 °C, in con-
trast to the catalyst without Pt (9W), for which higher
temperatures were required. The spillover of the H2 molecules on
the platinum sites seems to facilitate WOx reduction. It has to be
noticed that this effect is more noticeable as the platinum content
a
b
c
Tungsten surface density in brackets (W at. nmꢁ2).
Initial heat evolved in the first dose of NH3.
Total amount of NH3 retained as determined at 27 Pa of equilibrium pressure.
Irreversibly adsorbed amount of NH3 as determined from the difference
d
between the amounts adsorbed in the first and second adsorptions isotherms at
27 Pa.

S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
71
Fig. 5. (a) Volumetric adsorption isotherms, NH3 uptake in
NH3 uptake in
mol gꢁ1. Tungsten surface density (W at. nmꢁ2) in brackets.
l
mol gꢁ1 vs. equilibrium pressure and (b) differential heats of NH3 adsorption, differential heats in kJ molꢁ1 vs.
l
the real tungsten content. The non-tungstated catalyst (4Pt) pre-
sented a whole acidity similar to the one showed by the -alumina
surface. No evidence of Brönsted characteristic bands was found
at low densities of 1.0 W at. nmꢁ2, so the acidity appears to be
c
support, which indicates that the tungsten had the most remark-
able impact on the total acidity of the catalyst.
almost solely due to Lewis acid sites. On c-Al2O3 tungstated sam-
ples, common bands related to PyH+ began to appear at densities
P2.4 W at. nmꢁ2. On 9W (2.4 W at. nmꢁ2) catalyst at 200 °C, these
bands were found at 1540 cmꢁ1 and 1638 cmꢁ1, supporting the
presence of Brönsted sites [48]. The isolated tungsten oxide spe-
cies, observed by Raman spectroscopy at low tungsten surface den-
sities (<2.4 W at. nmꢁ2), cannot delocalize the negative charge
required for the formation of Brönsted acid sites. However, at
higher densities, polytungstates started to be formed resulting in
the formation of an extended network of tungsten oxide species,
over which an excess negative charge may be delocalized [33,49].
Qualitatively, the intensities of the IR bands of Lewis sites pro-
gressively decreased as the desorption temperature increased and
shifted toward higher wavenumbers, which confirms the high
strength of the Lewis sites. The same trend was observed for Brön-
sted sites.
Slight changes can be observed in the differential heats profiles
as it is shown in Fig. 5(b). In all cases, the heats of adsorption
showed a decreasing trend upon increasing coverage, as usually
observed for heterogeneous surfaces. The heterogeneity of the
studied samples is due to the presence of acidic sites of different
nature and strengths [43–46]. All the tungstated alumina investi-
gated samples exhibited high initial Qinit (>200 kJ molꢁ1) and Virr
very similar to each other, which are characteristic of the strongest
,
acid sites, whereas lower values were presented by the
support and the catalysts without tungsten.
c-alumina
It can also be seen in Fig. 5(b) that the presence of relatively
high amounts of platinum (4 wt%) enhanced the content of med-
ium strength acid sites, which seems to be independent of the
presence of tungsten.
In order to make a semi-quantitatively study of the samples
acidity, the area of 1450 cmꢁ1 peak for Lewis sites and the area
of 1540 cmꢁ1 peak for Brönsted sites were used. In Table 4, the
results of the Brönsted to Lewis ratio (B/L) at the different pyridine
desorption temperatures are shown. It can be noticed that WO3
3.1.9. FTIR of adsorbed pyridine
The effect of the tungsten oxide content (and tungsten surface
density) and the effect of the incorporation of platinum on WOx/
Al2O3 were studied by FTIR of adsorbed pyridine on the samples
after successive heating at 200 and 300 °C (Fig. S2). Brönsted and
Lewis acid sites are easily identified and distinguished by examina-
tion of the 1700–1400 cmꢁ1 range [47].
NPs formed at high tungsten surface densities (>3.9 W at. nmꢁ2
)
had not a contribution to the catalyst acidity, displaying slightly
lower B/L ratio than the sample with 2.4 W at. nmꢁ2
.
On the other hand, for 9W and 4Pt8W, samples with a close
tungsten surface density (2.4 and 2.2 W at. nmꢁ2, respectively),
the decrease of the Lewis acidity was more noticeable than the
Brönsted acidity decrease after desorption at 200 °C; therefore,
The type and strength of acidity on
of controversy for some time. In this work, the pyridine adsorbed
on the -Al2O3 support led to the following main bands at 200 °C
in this frequency range: 1614 (
8a) cmꢁ1, 1578 ( 8b) cmꢁ1, and
1450 (
19b) cmꢁ1, which can be assigned to coordinately bonded
c-Al2O3 has been a matter
c
m
m
m
pyridine on Lewis acid sites [47]. Another band at 1493 (m19a) -
cmꢁ1 was found, which is common of both vibrations due to
pyridinium ion (PyH+), produced by the reaction of pyridine with
Brönsted sites, and coordinately bonded pyridine on Lewis sites.
However, the non-appearance of another PyH+ bands seems to
Table 4
Brönsted to Lewis acid sites ratio (B/L) obtained from the FTIR of adsorbed pyridine.
Catalysta
B/L ratio at 200 °C
B/L ratio at 300 °C
c
-Al2O3
0.00
0.00
0.25
0.23
0.28
0.00
0.00
0.19
indicate the exclusive Lewis character of the
sites.
c
-Al2O3 support acid
4W (1.0)
9W (2.4)
14W (3.9)
4Pt8W (2.2)
0.15
The three characteristic bands of Lewis sites present on
c
-alu-
b
mina and the band at about 1490 cmꢁ1 have also been observed
for all the samples studied. Tungsten surface density plays a key
role in the formation of the Brönsted acid sites on the catalyst
Tungsten surface density in brackets (W at. nmꢁ2).
The area was too low to be integrated but not equal to zero.
a
b

72
S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
the B/L ratio increased. However, Brönsted sites were hardly
observed at 300 °C for the catalyst containing platinum (4Pt8W).
This could mean that the presence of platinum reduced the
strength of Brönsted acid sites.
It is worth adding that the hydrogen atoms forming by H2 dis-
sociation, during the catalyst pretreatment and along the reaction,
could be involved in the generation and maintenance of Brönsted
sites [32].
densities, the formation of crystalline WO3 NPs has a negative
effect on the 1,3-PDO yield due to the over-hydrogenolysis reac-
tions obtaining 1-PO as the main by-product. This effect cannot
be just related to the catalyst acidity, because small acidity differ-
ences were found between WOx/Al2O3 containing WO3 NPs and the
sample with the maximum polytungstate content and without
these WO3 NPs (see Table 4 entries 3 and 4). This aspect is later dis-
cussed with the proposed reaction mechanism.
Therefore, it seems that the presence of highly dispersed poly-
tungstate species in the catalyst surface, which provide Brönsted
acid sites, is necessary in order to produce 1,3-PDO by hydrogenol-
ysis of glycerol. It was found that the optimum surface density of
tungsten that maximizes the production of 1,3-PDO at these condi-
tions is 2.4 W at. nmꢁ2, obtaining a maximum 1,3-PDO selectivity
of 16.4% (2.7% yield).
3.2. Effect of tungsten surface density on glycerol conversion and
selectivity to PDOs
The effect of tungsten surface density was investigated in the
aqueous phase glycerol hydrogenolysis. It is widely suggested that
the production of 1,3-PDO from glycerol requires the presence of
Brönsted acid sites [12] but also a hydrogenation metal able to
activate the molecular H2 present in the reactive atmosphere. As
we have shown above, an interesting feature of the supported
WOx/Al2O3 catalysts is that their B/L acidity can be tuned as a func-
tion of the tungsten surface density. The activity test results are
presented in Fig. 6(a) and (b) obtained at 24 h reaction time,
220 °C and 45 bar with the catalyst containing 1 wt% of Pt and dif-
ferent tungsten surface density values.
3.3. Effect of platinum metal content on glycerol conversion and
selectivity to PDOs
The effect of platinum content was also studied on the Pt/WOx/
Al2O3 catalytic systems in the glycerol hydrogenolysis reaction. A
series of catalysts with different platinum content and similar
tungsten surface density, nearly the optimum value previously
obtained (2.2 W at. nmꢁ2), were prepared and tested.
Glycerol conversion diminished with increasing tungsten sur-
face density as it is shown in Fig. 6(a). On the other hand, it can
The activity results are represented in Fig. 7(a) and (b), and the
values are given in Table 1. As expected, when the platinum con-
tent increased, the glycerol conversion also increased, from 25.6%
for 1 wt% Pt to 60.3% for 9 wt% Pt. The hydrogen molecules present
in the reactive atmosphere might be dissociatively adsorbed on
platinum metal giving rise to hydrogen atoms that are subse-
quently involved in the activation of the adsorbed glycerol mole-
cule and in the later transformation to the products. The
electronic interactions, observed by TPR and Raman spectroscopy,
between platinum and the highly dispersed polytungstates could
also contribute to increasing the activation of H2 molecules on
the platinum active sites. These interactions were more pro-
nounced as the platinum content increased. On the other hand,
glycerol conversion does not seem to be clearly correlated to the
platinum dispersion, because the highest values were obtained at
the lowest platinum dispersions.
More interestingly, the platinum content had not only an effect
on glycerol conversion but it also affected PDOs selectivity. It must
be added that the most remarkable effect of platinum loading was
found in the 1,2-PDO selectivity: It decreased from 57.2% for 1 wt%
platinum catalyst to 11.0% for 9 wt% platinum content catalyst, as
it can be observed in Fig. 7(b). The 1,3-PDO selectivity increased
slightly and progressively with the platinum content increase.
be observed
a different trend in the PDO yields. Whereas
1,2-PDO yield decreased as the tungsten density increased, the
1,3-PDO production was increased to a maximum value and then
decreased as it can be seen in Fig. 6(b). The progressive coverage
of the hydroxyl groups of the c-alumina by the tungsten oxides,
as the tungsten surface density increases [31], may explain the
observed decrease in the glycerol conversion. It seems that tung-
sten oxide sites are less active than those of the c-alumina support
for the hydrogenolysis reaction of glycerol. This feature was not
observed when other supports such as silica or zirconia were used
[12,50]. The strong interaction between the c-alumina support and
tungsten oxide and therefore the low interaction between plati-
num and tungsten oxides, at low platinum loadings, could be the
reason for this poor glycerol conversion.
The incorporation of the tungsten oxide enhances the total acid-
ity of the catalyst, as observed from the NH3 adsorption calorime-
try results. B/L ratio also increased with increasing the tungsten
surface density due to the generation of polytungstate species, as
the FTIR of adsorbed pyridine (Table 4) and the Raman spectros-
copy (Fig. 2) analysis showed. This fact causes a decrease in the
1,2-PDO yield and an increase in the 1,3-PDO yield until a density
of 2.4 W at. nmꢁ2 was reached. For higher tungsten surface
Fig. 6. (a) Glycerol conversion values and (b) PDO yields obtained for Pt/WOx/Al2O3 catalysts with 1 wt% Pt as a function of the tungsten surface density, in the hydrogenolysis
of glycerol at 220 °C, 45 bar H2 pressure, and 24 h reaction time.

S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
73
Fig. 7. (a) Glycerol conversion values and (b) product selectivities obtained for Pt/WOx/Al2O3 catalysts with a constant tungsten surface density as a function of the platinum
content, in the hydrogenolysis of glycerol at 220 °C (200 °C for 9Pt8W⁄ catalyst), 45 bar H2 pressure, and 24 h reaction time.
Fig. 8. Reaction mechanism proposed for glycerol hydrogenolysis to 1,3-PDO.
For the catalyst with 1 wt% of platinum, the 1,3-PDO selectivity
was only of 16.4%, whereas the platinum content increases up to
9 wt% enhanced this value until 31.2%.
3.4. Proposed reaction mechanism
The most accepted mechanism for the formation of PDOs from
glycerol hydrogenolysis is the dehydration–hydrogenation route.
In the first step, glycerol is dehydrated to acetol or 3-hydroxyprop-
anal (3-HPA), while in the second step, these intermediates are
hydrogenated to 1,2 and 1,3-PDO, respectively. The formation of
acetol is thermodynamically favoured due to its higher stability as
compared to 3-HPA, which explains why normally 1,2-PDO is the
major product on glycerol hydrogenolysis. Our activity tests and
characterization results match with previously reported studies
that indicate that Bronsted acidity is required for the production
of 1,3-PDO: The maximum selectivity toward 1,3-PDO was obtained
at the tungstate surface density showing the highest B/L ratio (at
2.4 W at. nmꢁ2). However, there are some important evidences sug-
gesting that the formation of 1,3-PDO cannot be only related to the
amount of Bronsted acid sites.
The amount of platinum does not significantly contribute nei-
ther to the total acidity nor to the relative B/L ratio, as it was
observed by NH3 adsorption calorimetry, and FTIR of the absorbed
pyridine techniques. Therefore, the effect of the platinum loading
on the PDOs relative formation must be related to other factors.
The XPS results suggested that for high platinum contents
(>5 wt%), it anchors to the tungsten oxides sites (in addition to
the Al3+ preferential anchoring sites). Although this hypothesis
must be confirmed by other techniques, it is clear that the closer
proximity between the Pt and WOx active sites seems to enhance
the selectivity toward 1,3-PDO while decreasing the selectivity of
1,2-PDO. The main drawback is that the formation of 1,3-PDO
seems to be accompanied by the over-hydrogenolysis to 1-PO. In
order to avoid the undesirable reactions, the reaction temperature
was decreased to 200 °C for the 9Pt8W catalyst. The results, shown
in Table 1, indicate that the over-hydrogenolysis of 1,3-PDO to
1-PO was partially avoided at 200 °C, enhancing the 1,3-PDO selec-
tivity up to 51.9% (27.6% yield).
Recently, Tomishige’s group [16] studied the influence that
adding different concentrations of H2SO4 as acid co-catalyst
had on the glycerol hydrogenolysis reaction, over another
noble metal–oxophilic metal catalytic system (Ir–ReOx/SiO2).

74
S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
Interestingly, the positive effect of H2SO4 on the production of 1,3-
PDO formation over the Ir–ReOx/SiO2 catalyst was saturated at H+/
Re = 1. Both activity and selectivity were almost unchanged
between H+/Ir = 1 and 12.4. Similarly, in our results, the yield of
1,3-PDO was more than two times higher with the 1Pt9W catalyst
compared with the 1Pt14W catalyst even though both catalysts
presented similar B/L ratios. In addition to this, it was observed
that the amount of Pt and the interaction between Pt and WOx
active sites also affect the selectivity toward 1,3-PDO.
Development Fund (FEDER) (Spanish MICIN Project: CTQ2012-
38204-C03-03), and the Basque Government (Researcher Training
Programme of the Department of Education, Universities and
Research). The authors also gratefully acknowledge Prof. José Luis
García Fierro from Institute of Catalysis and Petrochemistry of
Madrid for the technical support in the XPS measurements.
Appendix A. Supplementary material
For Pt–W–Al catalysts, Kaneda et al. proposed a mechanism [12,
51] which involved an initial glycerol adsorption on Al–OH site
forming an alkoxide specie. As it has been mentioned above, the
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.12.028.
tungsten oxide species anchor to the c-Al2O3 support at its surface
References
hydroxyls [40], and therefore, an increase in tungsten surface den-
sity should imply a decrease in Al–OH sites. We suggest that glyc-
erol is adsorbed on tungsten active sites forming an alkoxide (see
reaction scheme in Fig. 8). Next, a proton, coming from a Brönsted
acid site provided by polytungstate species, protonates the second-
ary hydroxyl group of the alkoxide. After dehydration, a secondary
carbocation is formed. The adsorbed carbocation is attacked by a
hydride specie generated on the Pt active sites. Further hydrolysis
yields 1,3-PDO. The key step in this reaction seems to be the fast
hydrogenation of the secondary carbocation in order to avoid its
conversion into acrolein (precursor of 1-PO). The stabilization of
the carbocation might be crucial at this point. It was suggested that
only the polytungstate species are able to delocalize the negative
charge required for the carbocation intermediate stabilization,
instead of the monotungstate and WO3 NPs species [32]. This
explains well the reason why the best results (at similar platinum
content of 1 wt%) were obtained for the catalyst in which more
polytungstate species are generated before the WO3 NPs forma-
tion. These WO3 NPs are not only inactive for the carbocation sta-
bilization, but they also are tridimensional structures that limit the
accessibility of the glycerol to the polytungstate acid sites. Another
important factor is the close proximity between the Pt and the WOx
active sites, which might facilitate the attack of the hydride species
for the fast hydrogenation of the secondary carbocation. This
explains the higher selectivities toward 1,3-PDO that are obtained
at higher Pt contents.
[1] T. Werpy, G. Petersen, A. Aden, J. Bozell, J. Holladay, J. White, A. Manheim, D.
Eliot, L. Lasure, S. Jones, Top Value Added Chemicals from Biomass. Results of
Screening for Potential Candidates from Sugars and Synthesis Gas, vol. 1, 2004.
[2] R. Arbour, B. Esparis, Chest 118 (2000) 545.
[3] I. Gandarias, J. Requies, P.L. Arias, U. Armbruster, A. Martin, J. Catal. 290 (2012)
79.
[4] Y. Shinmi, S. Koso, T. Kubota, Y. Nakagawa, K. Tomishige, Appl. Catal. B 94
(2010) 318.
[5] M. Akiyama, S. Sato, R. Takahashi, K. Inui, M. Yokota, Appl. Catal. A: Gen. 371
(2009) 60.
[6] Z. Yuan, L. Wang, J. Wang, S. Xia, P. Chen, Z. Hou, X. Zheng, Appl. Catal. B 101
(2011) 431.
[7] I. Gandarias, S. García-Fernández, M. El Doukkali, P.L. Arias, Top. Catal. (2013)
995
[8] J. Chaminand, L. aurent Djakovitch, P. Gallezot, P. Marion, C. Pinel, C. Rosier,
Green Chem. 6 (2004) 359.
[9] T. Kurosaka, H. Maruyama, I. Naribayashi, Y. Sasaki, Catal. Commun. 9 (2008)
1360.
[10] L. Huang, Y. Zhu, H. Zheng, G. Ding, Y. Li, Catal. Lett. 131 (2009) 312.
[11] J. Oh, S. Dash, H. Lee, Green Chem. 13 (2011) 2004.
[12] L. Qin, M. Song, C. Chen, Green Chem. 12 (2010) 1466.
[13] M. Chia, Y.J. Pagán-Torres, D. Hibbitts, Q. Tan, H.N. Pham, A.K. Datye, M.
Neurock, R.J. Davis, J.A. Dumesic, J. Am. Chem. Soc. 133 (2011) 12675.
[14] S. Koso, H. Watanabe, K. Okumura, Y. Nakagawa, K. Tomishige, Appl. Catal. B
111–112 (2012) 27.
[15] Y. Nakagawa, Y. Shinmi, S. Koso, K. Tomishige, J. Catal. 272 (2010) 191.
[16] Y. Amada, Y. Shinmi, S. Koso, T. Kubota, Y. Nakagawa, K. Tomishige, Appl. Catal.
B 105 (2011) 117.
[17] Y. Nakagawa, X. Ning, Y. Amada, K. Tomishige, Appl. Catal. A 433–434 (2012)
128.
[18] L. Zhang, A.M. Karim, M.H. Engelhard, Z. Wei, D.L. King, Y. Wang, J. Catal. 287
(2012) 37.
[19] S. Zhu, Y. Qiu, Y. Zhu, S. Hao, H. Zheng, Y. Li, Catal. Today 212 (2013) 120.
[20] R. Arundhathi, T. Mizugaki, T. Mitsudome, K. Jitsukawa, K. Kaneda,
ChemSusChem 6 (2013) 1345.
4. Conclusions
[21] Y. Nakagawa, M. Tamura, K. Tomishige, J. Mater. Chem. A 2 (2014) 6688.
[22] C. Pfaff, M.J. Pérez Zurita, C. Scott, P. Patiño, M.R. Goldwasser, J. Goldwasser,
F.M. Mulcahy, M. Houalla, D.M. Hercules 49 (1997) 13–16.
[23] G. Bergeret, P. Gallezot, in: G. Ertl, H. Knözinger, F. Schüth, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, Wiley-VCH, 2008, p. 738.
[24] Y. Izutsu, Y. Oku, Y. Hidaka, N. Kanaya, Y. Nakajima, J. Fukuroi, K. Yoshida, Y.
Sasaki, Y. Sekine, M. Matsukata, J. Phys. Chem. C 118 (2014) 10746.
[25] K. Jeong, H. Kim, T. Kim, J. Kim, H. Chae, S. Jeong, C. Kim, Catal. Today 232
(2014) 151.
[26] A.D. Allian, K. Takanabe, K.L. Fujdala, X. Hao, T.J. Truex, J. Cai, C. Buda, M.
Neurock, E. Iglesia, J. Am. Chem. Soc. 133 (2011) 4498.
[27] L. Karakonstantis, K. Bourikas, A. Lycourghiotis, J. Catal. 162 (1996) 295.
[28] M.I. Zaki, N.E. Fouad, S.A.A. Mansour, A.I.M. Rabee, Appl. Surf. Sci. 282 (2013)
898.
[29] M. Kim, J. Park, C. Shin, S. Han, G. Seo, Catal. Lett. 133 (2009) 288.
[30] X. Wu, L. Zhang, D. Weng, S. Liu, Z. Si, J. Fan, J. Hazard. Mater. 225–226 (2012)
146.
[31] I.E. Wachs, T. Kim, E.I. Ross, Catal. Today 116 (2006) 162.
[32] D.G. Barton, M. Shtein, R.D. Wilson, S.L. Soled, E. Iglesia, J. Phys. Chem. B 103
(1999) 630.
In this work, it has been demonstrated that the bimetallic Pt/
WOx/Al2O3 catalytic system is effective for the selective hydrogen-
olysis of glycerol to 1,3-PDO. This study proved that 1,3-PDO selec-
tivity depends on the tungsten surface density that controls the
kind of tungsten oxide species deposited on the catalyst surface.
The maximum selectivity was achieved when the highest content
of polytungstate species are formed but before the appearance of
WO3 NPs, concluding that in order to selectively obtain 1,3-PDO,
the presence of highly dispersed polytungstate species is required.
These species are able to delocalize the negative charge required
for the formation of Brönsted acid sites. However, the results from
this work also indicate that not only acidity plays a role in the
selective hydrogenolysis of glycerol to 1,3-PDO, but also the plati-
num and tungsten oxides sites interactions may be involved. The
closer proximity between platinum and tungsten oxides at high
platinum content and the electronic interactions between acid
and metallic sites could be the responsible for the enhancement
of the 1,3-PDO yields. Further research should allow improving
Pt–WOx synergies and as a result the yield to 1,3-PDO.
[33] D.G. Barton, S.L. Soled, G.D. Meitzner, G.A. Fuentes, E. Iglesia, J. Catal. 181
(1999) 57.
[34] M.M. Ostromecki, L.J. Burcham, I.E. Wachs, N. Ramani, J.G. Ekerdt, J. Mol. Catal.
A 132 (1998) 43.
[35] M.N. Taylor, W. Zhou, T. Garcia, B. Solsona, A.F. Carley, C.J. Kiely, S.H. Taylor, J.
Catal. 285 (2012) 103.
[36] V. León, Surf. Sci. 339 (1995) 931.
[37] J.Z. Shyu, K. Otto, Appl. Surf. Sci. 32 (1988) 246.
[38] S.F. Ho, S. Contarini, J.W. Rabalais, J. Phys. Chem. 91 (1987) 4779.
[39] W. Grünert, E.S. Shpiro, R. Feldhaus, K. Anders, G.V. Antoshin, K.M. Minachev, J.
Catal. 107 (1987) 522.
[40] J.H. Kwak, J. Hu, D. Mei, C.W. Yi, H. Kim do, C.H. Peden, L.F. Allard, J. Szanyi,
Science 325 (2009) 1670.
Acknowledgments
This work was supported by University of the Basque Country
(UPV/EHU), European Union through the European Regional

S. García-Fernández et al. / Journal of Catalysis 323 (2015) 65–75
75
[41] J.L. Contreras, G.A. Fuentes, B. Zeifert, J. Salmones, J. Alloys Compd. 483 (2009)
371.
[46] C. Busco, A. Barbaglia, M. Broyer, V. Bolis, G. Foddanu, P. Ugliengo,
Thermochim. Acta 418 (2004) 3.
[42] D. Sprinceana, M. Caldararu, N. Ionescu, A. Auroux, J. Therm. Anal. Calorim. 56
(1999) 109.
[43] S. Bennici, A. Auroux, in: S.D. Jackson, J.S.J. Hargreaves (Eds.), Metal Oxide
Catalysis, Wiley-VCH, 2009, p. 391.
[44] A. Auroux, D. Sprinceana, A. Gervasini, J. Catal. 195 (2000) 140.
[45] A. Auroux, Top. Catal. 4 (1997) 71.
[47] E. Parry, J. Catal. 2 (1963) 371.
[48] G. Busca, Phys. Chem. Chem. Phys. 1 (1999) 723.
[49] J. Macht, E. Iglesia, Phys. Chem. Chem. Phys. 10 (2008) 5331.
[50] L. Gong, Y. Lu, Y. Ding, R. Lin, J. Li, W. Dong, T. Wang, W. Chen, Appl. Catal. A:
Gen. 390 (2010) 119.