Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites of the Support

Article abstract of DOI:10.1007/s10562-017-2183-5

Abstract: A series of 1?wt% AuPt (6:4) catalysts were prepared by sol immobilization using acidic [TiO₂, H-Mordenite, SiO₂, MCM-41, sulphated ZrO₂ (S-ZrO₂)] and one basic (MgO) oxide supports. A careful characterization of the catalyst was performed by HRTEM and FTIR spectroscopy. EDX analysis showed that on all catalysts only alloyed AuPt nanoparticles are present, but the final size of AuPt particles is significantly affected by the support. Indeed, on TiO₂ the mean AuPt nanoparticles diameter is 3.7?nm, whereas for all the remaining supports, larger AuPt nanoparticles with diameter of 6–7.5?nm were obtained. AuPt catalysts result very active in catalyzing the liquid phase hydrogenolysis of glycerol to 1,2-propandiol (68% conversion after 16?h) and 90% selectivity with ethylene glycol, 1-propanol and 2-propanol as main by-products. The role of the support has been highlighted in terms of acidic properties. The medium strength of Lewis acid sites of TiO₂ leads to the best performance in terms of activity, selectivity and stability of the catalytic system. Graphical Abstract: [Figure not available: see fulltext.].

Full text of DOI:10.1007/s10562-017-2183-5

Catal Lett
DOI 10.1007/s10562-017-2183-5
Diols Production From Glycerol Over Pt-Based Catalysts: On
the Role Played by the Acid Sites of the Support
1
2
3
4
Alberto Villa · Maela Manzoli · Floriana Vindigni · Lidia E. Chinchilla ·
4
1
Gianluigi A. Botton · Laura Prati
Received: 31 May 2017 / Accepted: 25 August 2017
Springer Science+Business Media, LLC 2017
©
Abstract A series of 1 wt% AuPt (6:4) catalysts were
Graphical Abstract
prepared by sol immobilization using acidic [TiO , H-Mor-
2
denite, SiO , MCM-41, sulphated ZrO (S-ZrO )] and one
2
2
2
basic (MgO) oxide supports. A careful characterization of
the catalyst was performed by HRTEM and FTIR spectros-
copy. EDX analysis showed that on all catalysts only alloyed
AuPt nanoparticles are present, but the final size of AuPt
particles is significantly affected by the support. Indeed,
on TiO the mean AuPt nanoparticles diameter is 3.7 nm,
2
whereas for all the remaining supports, larger AuPt nano-
particles with diameter of 6–7.5 nm were obtained. AuPt
catalysts result very active in catalyzing the liquid phase
hydrogenolysis of glycerol to 1,2-propandiol (68% conver-
sion after 16 h) and 90% selectivity with ethylene glycol,
1
-propanol and 2-propanol as main by-products. The role of
the support has been highlighted in terms of acidic proper-
ties. The medium strength of Lewis acid sites of TiO leads
2
to the best performance in terms of activity, selectivity and
stability of the catalytic system.
*
Laura Prati
Keywords Glycerol · Hydrogenolysis · Processes and
Laura.Prati@unimi.it
reactions · Gold platinum · Alloy · Support effect
1
2
Dipartimento di Chimica, Università degli Studi di Milano,
via Golgi 19, 20133 Milano, Italy
Dipartimento di Scienza e Tecnologia del Farmaco and NIS
1
Introduction
–
Interdepartmental Centre for Nanostructured Interfaces
and Surfaces, Università degli Studi di Torino, Via P. Giuria
, 10125 Torino, Italy
The use of biomass for the production of renewable raw
materials and their conversion to high value chemicals and
materials show a significant potential [1–3]. The utilization
of vegetable oil for the production of biodiesel has led to an
increase in glycerol production, as it constitutes the main co-
product (about 20 wt%). Glycerol is a highly functionalized
9
3
4
Dipartimento di Chimica and NIS – Interdepartmental
Centre for Nanostructured Interfaces and Surfaces,
Università di Torino, Via P.Giuria 7, 10125 Torino, Italy
Canadian Centre of Electron Microscopy and Department
of Materials Science and Engineering, McMaster University,
1
280 Main Street West, Hamilton, ON L8S 4M1, Canada
Vol.:(0
1
123456789)
3

A. Villa et al.
molecule, which is recognized as a promising chemical
building block for the synthesis of fine chemicals [1–6]. One
attractive route involves the catalytic hydrogenolysis of glyc-
erol to diols and alcohols (as reported in Scheme 1) [7–9].
can significantly affect the catalytic performance [21, 22].
In particular, the acid base properties have been reported
to modify the activity and, primarily, the selectivity in the
glycerol oxidation [20]. Therefore, this study encompasses
the investigation of the effect of the acid base properties
of the support in AuPt catalyzed glycerol hydrogenolysis
to 1,2 PD.
1
,2 Propandiol (1,2-PD), is commercially produced from
propylene oxide and it is a valuable chemical for the produc-
tion of polyesters, resins, and polyurethanes [10]. 1,3-PD is
an important monomer utilized for the production of poly-
ester fibers and films [11]. Ethylene glycol (EG) is produced
via hydration of ethylene oxide and it is used as antifreezing
agent [12].
Many mechanisms corresponding to glycerol hydrogen-
olysis to 1,2 PD, in which the acidic/basic properties seem
to have a key role in determining the final pathway, have
been proposed. Three main pathways (Scheme 2) are gen-
erally accepted: dehydration–hydrogenation, dehydrogena-
tion–dehydration–hydrogenation and direct-hydrogenolysis
[9].
In the past decade, glycerol hydrogenolysis has been
extensively studies using different metal based catalysts,
including Pt, Ru, Ni and Cu [7–9]. Among all noble metal
based catalysts, Pt has been recognized as one of the most
effective in the glycerol hydrogenolysis [7–9, 13–15]. For
example, commercial Pt/C showed high selectivity to 1,2-
PD (82.7%) but only at low conversion (35%) [16]. Similar
In the first path, glycerol is initially dehydrated through
an acid-catalyzed reaction to form an intermediate, which
is subsequently hydrogenated to generate propanediol [23].
Following the second possible path, which usually occurs in
neutral and alkaline conditions, glycerol is dehydrogenated
to glyceraldehyde with a consecutive dehydration–hydro-
genation to 1,2-PD [24]. The last possibility is the direct-
hydrogenolysis that was initially reported for Ir catalysts
[25]: glycerol is adsorbed on the metal surface to form 1,3
dihydroxyisopropoxide that is successively transformed to
1,2-PD. According to this mechanism, the low stability of
the intermediate dihydroxyisopropoxide can result in the
formation of over-hydrogenolysis products (1-propanol and
2-propanol). In all the possible pathways, the role of the
support properties appears fundamental and yet not fully
understood.
results were obtained using PtSn/SiO with a good selec-
2
tivity to 1,2-PD (84%) but even at lower conversion (16%)
[
17]. Nevertheless, the drawback of most of the monome-
tallic Pt systems seems to be the low resistance to deac-
tivation and the low capacity to maintain high selectivity
at high conversion. We recently reported that the addi-
tion of Au to Ru improves the stability of the catalysts,
maintaining a good selectivity also at high conversion
[
18]. Moreover, the beneficial effect of Au addition to Pt
catalysts in terms of activity, selectivity and stability was
already verified in the case of oxidation of glycerol [19,
2
0]. Recent papers illustrated that the nature of the support
-
H O
+ H2
H O
2
O
H O
propanal
1-PrOH
-H O
-H O
H O
2
HO
OH
-
H O
2
+
H2
1,3-PD
1,2-PD
2
O
OH
2-PrOH
C-O cleavage
O H
propane
acetone
H O
OH
glycerol
LEGEND
1
,2-PD= 1,2-propanediol
-PrOH= 1-propanol
C-C cleavage
1
2
-PrOH= 2-propanol
H O
H O
, CO , CH
+
CH OH
3
2
4
EG= ethylene glycol
MeOH= methanol
OH EG
MeOH
ethanol
Scheme 1 Products obtained by glycerol hydrogenolysis
1
3

Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites…
H O
H O
2.2 Catalyst Preparation
+
H2
O
OH
2.2.1 Monometallic Pt Catalyst
acetol
1,2-propanediol
dehydraꢀon-hydrogenaꢀon
Solid K PtCl (0.051 mmol) and PVA (1 wt%) solution (Pt/
2
4
PVA 1/0.5 wt%) were added to 100 mL of H O. After 5 min,
2
-
H O
2
H was bubbled (50 mL/min) under atmospheric pressure
2
H O
O H
O H
H O
and room temperature for 2 h. The colloid was immobilized
by adding the support under vigorous stirring. The amount
of support was calculated as having a total final metal load-
ing of 1 wt%. After 2 h the slurry was filtered, the catalyst
washed thoroughly with distilled water (neutral mother liq-
uors) and dried at 80 °C for 4 h.
+
H2
-
H O
2
OH
OH
1
,2-propanediol
-
H2
direct hydrogenolysis
+2H2
H O
H O
O
2
.2.2 Au–Pt Bimetallic Catalysts
-
2
OH
OH
glyceraldehyde
1,2-propanediol
For the sake of clarity, the main steps of the preparation of
the bimetallic catalysts are summarized in Scheme 3. The
bimetallic AuPt catalysts were prepared by sol immobiliza-
tion through a two steps procedure, which ensures the forma-
tion of alloyed bimetallic particle [19, 20].
dehydrogenaꢀon-dehydraꢀon-hydrogenaꢀon
Scheme 2 Proposed reaction pathways of glycerol hydrogenolysis to
,2-PD
1
In detail, NaAuCl ·2H O (Au: 0.031 mmol) was dis-
4
2
solved in 60 mL of H O, and PVA (1%, wt%) was added
2
Therefore, to clarify the possible synergy between the
(Au/PVA=1:0.5 wt/wt). The yellow solution was stirred for
acid base properties of the support and AuPt nanoparticles,
different catalysts have been prepared supporting preformed
AuPt NPs on acidic (TiO , H-Mordenite, SiO , MCM-41,
3 min, after which 0.1 M NaBH (Au/NaBH =1:4 mol/mol)
4
4
was added under vigorous magnetic stirring. The ruby-red
Au(0) sol was formed immediately. Within a few minutes of
sol generation, the gold sol was immobilized by adding the
support (acidified to pH 2 by sulphuric acid) under vigorous
stirring. The amount of support was calculated as having
a gold loading of 0.60 wt%. After 2 h, the slurry was fil-
tered and the catalyst washed thoroughly with distilled water
(neutral mother liquors). The Au/support was dispersed in
40 mL of water, therefore K PtCl (Pt: 0.021 mmol) and
2
2
S-ZrO ) and basic (MgO) oxides. High-resolution trans-
2
mission electron microscopy (HRTEM) analysis of the
AuPt catalysts was performed to investigate the morphol-
ogy as well as the composition of the nanoparticles. The
nature (Lewis and/or Brønsted) and acid site density were
determined by a quantitative analysis of the FTIR bands of
adsorbed 2,6 dimethyl-pyridine (2,6-DMP). The catalysts
have been then tested in glycerol hydrogenolysis.
2
4
PVA solution (Pt/PVA = 1:0.5 wt/wt) were added. H was
2
bubbled (50 mL/min) under atmospheric pressure and room
temperature for 2 h. After an additional 18 h, the slurry was
filtered and the catalyst washed thoroughly with distilled
water. ICP analyses were performed on the filtrate using
a Jobin Yvon JV24 instrument to verify the metal loading
on the support. The total metal loading was 1 wt%. AuPt
2
Experimental
2
.1 Materials
nanoparticles were supported on five acidic oxides, i.e. TiO ,
2
NaAuCl ·2H O, and K PtCl were from Aldrich (99.99%
H-Mordenite, SiO , MCM-41, S-ZrO , and a basic support
4
2
2
4
2
2
2
purity), TiO P25 (SA = 48 m /g) was from Evonik and
(MgO). The obtained catalysts are listed in Table 1.
2
2
H-Mordenite (SA = 450 m /g) was from Degussa. Sul-
2
2
phated zirconia (SA = 78 m /g), SiO (SA = 148 m /g)
2.3 Catalytic Tests
2
2
and MgO (SA=38 m /g) were from Alfa Aesar. MCM-41
2
(
SA = 980 m /g) have been prepared following the proce-
Glycerol hydrogenolysis was performed at 150 °C, using a
stainless steel Parr reactor (50 mL capacity), equipped with
heater, mechanical stirrer, gas supply system and thermome-
ter. The glycerol solution (30 mL; 0.3 M) was added into the
reactor and the desired amount of catalyst (glycerol/metal
ratio=1000, mol/mol) was suspended in the solution. The
dure reported in Mori et al., [26]. NaBH of purity >96%
4
from Fluka, polyvinylalcohol (PVA) (Mw=13,000–23,000,
8
7–89% hydrolyzed), from Aldrich were used. A 1 wt% PVA
solution in water was prepared. Gaseous oxygen from SIAD
was 99.99% pure.
1
3

A. Villa et al.
+
support
+
PVA
+ NaBH4
NaAuCl ·2H O
Au0
0.60 wt %
Au/support
4
2
aqueous solution
filtering,
washing
K Pt Cl + PVA
2
2
4
1
wt %
solution
AuPt/support
bubbling H
(2 h, r.t.)
filtering,
washing
2
AuPt/support
Au/support
slurry solution
slurry solution
Scheme 3 Synthesis of the bimetallic catalysts
Table 1 Catalytic evaluation in
a
Activity A Selectivity (%)c
b
Catalyst
Activity
AuPt size
(nm)
the glycerol hydrogenolysis
−1
mol
mol
(
h
)
N × h
1,2-PD 1-PrOH 2-PrOH EG MeOH
mol ×h
AuPt/MgO
AuPt/TiO2
AuPt/MCM-41
AuPt/SiO2
AuPt/H-Mordenite 14
AuPt/S-ZrO2
Pt/TiO2
34
56
22
21
176
165
111
117
65
52
119
2
93
90
52
35
30
31
86
–
–
–
–
–
8
10
–
–
–
–
–
12
11
–
3
4
2
2
6.9
3.7
6.7
7.5
6.2
6.9
3.6
3.5
22 16
28 22
20 16
18 15
3
–
10
41
1
4
–
Au/TiO2
–
–
Reaction conditions: glycerol=0.3M, pH =7 bar, T=150 °C, metal/glycerol=1/1000 mol/mol. EG ethyl-
2
ene glycol, 1,2-PD 1,2 propandiol, 1-PrOH 1-propanol. The carbon balance around 90–96% observed in all
the reactions was ascribed to the formation of CH and CO not detectable in liquid phase
4
2
a
Mol of glycerol converted per hour per mol of metal, calculated after 0.5 h of reaction
Mol of glycerol converted per hour per fraction of atoms lying at the surface, A [A=(Ns/NT)×100) after
b
0
.5 h of reaction
Selectivity at 60% of conversion
c
autoclave was purged three times with nitrogen before charg-
done by comparison with the original samples. Activity was
calculated on the total mol of metal and, alternatively, on the
total amount of exposed surface atoms. Calculations of the
number of exposed surface atoms were performed by assum-
ing that all the nanoparticles had cuboctahedral morphol-
ogy with cubic close-packed structure in this size range, the
model of full-shell nanoparticles was adopted [27]. The total
ing 10 bar of H . The mixture was heated to the reaction
2
temperature, 150 °C, and mechanically stirred (1250 rpm).
Products analysis: the reaction mixture, after separation
from the catalysts by filtration, was analysed using high per-
formance liquid chromatography (HPLC). Samples were
removed periodically (0.5 mL) under stirring and analysed by
HPLC using a column (Alltech OA-10308 300 mm×7.8 mm)
with UV and refractive index (RI) detection in order to ana-
lyse the product mixtures. H PO 0.1 wt% solution was used
number (N ) of M atoms in the cluster for a given cluster size
t
can be calculated using the following Eq. (1):
3
4
1
∕3
as the eluent. The identification of the possible products was
dsph = 1.105 d N
at
(1)
T
1
3

Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites…
where d is the mean diameter of the Au, Pt or AuPt parti-
acquired using FEI TIA software and Oxford INCA
sph
cles obtained from transmission electron microscopy (TEM)
microanalysis software was used for XEDS acquisition
and analysis. The atomic fractions of gold and platinum
were quantified by the Cliff–Lorimer method on relative
intensities of the Pt-Lα and Au-Lα peaks using k-factors
provided by the XEDS system manufacturer. The AuPt
particle size distribution and the total metal dispersion
were determined by counting 250 particles in HAADF-
STEM images using GAUSS software.
analysis and d is the atom diameter of Au (0.288 nm), Pt
at
(
0.256 nm) and AuPt (0.264 nm), respectively. The number
of surface atoms (N ) and n can be calculated from Eqs. (2)
s
and (3), based on the values of N :
T
ꢁꢂ
ꢀ
3
2
N = 10n − 15n + 11n − −3
3
(2)
(3)
T
2
N = 10n − 20n + 12
s
c. FTIR spectra were taken on a Perkin-Elmer 2000 spec-
trometer (equipped with a MCT detector) with the sam-
ples in self supporting pellets introduced in cells allow-
ing thermal treatments in controlled atmospheres and
spectrum scanning at room temperature (r.t.) in vacuum
or in the presence of probe gases. The 2,6-dimethyl-
pyridine (2,6-DMP) adsorption/desorption experi-
ments were performed on the samples after activation
in vacuum at 393 K for 1 h. The following procedure
was adopted: (1) inlet of an excess dose of 2,6-DMP
vapor (~2 Torr), and equilibration at r.t. for 10 min; (2)
evacuation at r.t. for 30 min. From each spectrum, the
spectrum of the sample before the inlet of the probe
was subtracted. The spectra have been normalised with
respect to the weight of the pellets and to the surface
areas of the different supports.
The activity based on the surface atoms can then be cal-
culated as follows:
fraction of atoms lying at the surface:
ꢀ
ꢁ
A = N ∕N × 100
s
T
Activity based on N =activity (calculated for total metal
s
mols)/A.
Recycling tests were carried out under the same experi-
mental conditions. The catalyst was recycled in the subse-
quent run after filtration without any further treatment.
2
.4 Characterisation
a. The metal content was checked by Atomic Absorption
Spectroscopy (AAS) analysis of the filtrate, on a Perkin
Elmer 3100 instrument.
b. Samples for TEM characterization were prepared by
depositing an ethanol suspension of the catalyst onto
lacey carbon coated 300 mesh copper grids. The par-
ticle morphology of the supported nanoparticles, spe-
cifically the nanoparticle size, was first investigated by
The processing of the FTIR spectra of adsorbed 2,6-DMP
was performed by using Fityk 0.9.8, an open-source curve-
fitting and data analysis software, that allows to obtain accu-
rate curvefits (either Gaussian or Lorentzian curves were
used). As output files, the software permits the realisation
of one figure each time. Each obtained figure contains the
files related to one specific sample in the optimised spectral
range, therefore it was not possible to plot all files in the
same figure.
TEM using a Philips LaB electron microscope, operat-
6
ing at 200 kV and equipped with a Gatan CCD camera.
Detailed high resolution High Angle Annular Dark Field
Scanning TEM (HAADF-STEM) imaging and energy-
dispersive X-ray spectroscopy (XEDS) analyses were
carried out using FEI Titan3 microscope operated at
2
00 kV accelerating voltage for a deeper investigation
3 Results and Discussion
of the AuPt alloy structure of AuPt/TiO catalyst. This
2
microscope is equipped with double aberration correc-
The catalysts were evaluated in glycerol hydrogenolysis
tors, providing ultrahigh-resolution HAADF-STEM
(0.3 M glycerol, glycerol/metal=1000 mol/mol, 10 atm. H ,
2
images, and an Oxford Inca energy dispersive X-ray
T=150 °C) and both activity and selectivity, at isoconver-
2
(
EDX) spectrometer equipped with a 30 mm ultrathin
sion (60%) are reported for each system in Table 1.
window Si/Li X-ray detector. XEDS data were collected
either as spectrum images, in which a focused electron
probe was scanned across a region of interest during
data collection, or in stationary spot mode, where an
emitted X-ray spectrum from 0 to 20 kV is acquired
from a specific point on a particle using a probe size
The activity was calculated firstly basing on the moles
of glycerol converted per hour per total mol of metal. By
using this approach, AuPt/TiO showed a high activity (56
2
converted mol of glycerol (AuPt mol)/h), superior than AuPt
supported on MgO (34 converted mol of glycerol (AuPt
mol)/h) and the other acidic oxides (<22 mol of glycerol
<
0.5 nm. Spectra were acquired with a probe current
(AuPt mol)/h for AuPt/MCM41, AuPt/SiO , AuPt/H-Mor-
2
of approximately 0.5 nA and dwell times between 200
and 400 ms/pixel, in the case of maps, and 20–30 s
per analysis in spot mode. STEM digital images were
denite, AuPt/S-ZrO , accordingly). Moreover, monometal-
2
lic Pt/TiO and Au/TiO were also tested as reference cata-
2
2
lysts: it was found that monometallic Pt showed intermediate
1
3

A. Villa et al.
were carried out by filtering and using the catalyst in the
next run without any further purification. The catalyst was
stable for six run. Therefore, these results demonstrate that
when TiO is used as support the addition of Au to Pt is
2
beneficial in terms either of activity or of stability in glycerol
hydrogenolysis, similarly as previously observed for glycerol
oxidation [19, 20]. Similar findings were achieved by some
of us when Au was added to Ru supported on active carbon
(
AC), with AuRu/AC displaying an higher activity than Ru/
AC in glycerol hydrogenolysis to 1,2-PD [18].
As for the obtained selectivity, also reported in Table 1,
it was found that AuPt supported on MgO and TiO were
2
extremely selective toward the formation of 1,2-PD (93
and 90%, respectively. All the other catalysts showed lower
selectivity to 1,2-PD, 30–50%). In particular, these catalysts
produced high amounts of EG and methanol deriving from
the C–C cleavage (see Scheme 1; Table 1). Moreover, 1-pro-
panol and 2-propanol, deriving from the over-hydrogenolysis
of 1,2-PD (Scheme 1, Table 1) were produced in the pres-
Fig. 1 Reaction profile of AuPt catalysts
ence of the AuPt/H-Mordenite and AuPt/S-ZrO catalysts.
2
A combined HRTEM and FTIR characterization was
considered out to unravel on one hand, the influence of both
AuPt morphology and size and, on the other hand, to focus
on the effect of the nature of the support on the catalytic
activity and selectivity. For the latter reason, due to the low
stability of MgO based catalyst during the reaction, the
attention will be entirely focused on the acid catalysts which
all presented a pretty good stability, as shown for TiO .
2
A detailed HRTEM characterization of the bimetallic
AuPt catalysts has been previously reported by Chan-Thaw
et al. [28] for AuPt/TiO and by Villa et al. [20] for all the
2
other catalysts. In all cases, TEM images revealed a good
AuPt dispersion and EDS analysis showed the exclusive
presence of AuPt alloyed nanoparticles [20, 28]. HAADF
images of AuPt/TiO catalyst are shown in Fig. 3 for the sake
2
Fig. 2 Stability test using AuPt/TiO2
of clarity. A representative STEM-HAADF image of AuPt/
TiO and the corresponding XEDS elemental analysis are
2
activity (41 converted mol of glycerol (Pt mol)/h), whereas
monometallic Au was inactive (1 converted mol of glycerol
also reported in Fig. 4a, b, respectively. It is clearly shown
that both Pt and Au are present in the same nanoparticle.
Moreover, the distribution profiles of the metals shown in
Fig. 4c confirmed the coexistence of Pt and Au within the
same nanoparticle, suggesting the formation of a random
gold-platinum alloy structure.
(
Au mol)/h). These findings indicate that Au has a synergist
effect on the activity of Pt, maximized in the case of TiO
support.
2
In addition, the comparison of the reaction profile of the
two most active catalysts put in evidence that AuPt/TiO
However, despite the same alloy nanostructure, the AuPt
bimetallic nanoparticles differ in term of size, as reported
2
showed a better resistance to deactivation than AuPt/MgO
(
Fig. 1). Interestingly, AAS analysis revealed the presence
in Table 1. Indeed, AuPt/TiO average diameter is 3.7 nm,
2
of Au, Pt and Mg in the reaction media as for AuPt/MgO,
whereas for all the remaining supports larger AuPt nano-
particles, with mean diameter of about 6–7.5 nm, were
observed. The two metals have been deposited on the dif-
ferent supports by following the same synthetic procedure
summarized in Scheme 3: it is evident that the main dif-
ference in the particle size was observed when Pt and Au
whilst AuPt/TiO did not show any leaching. Moreover, the
2
performance of bimetallic AuPt compared to monometallic
Pt on TiO (Table 1) showed not only the best activity of the
2
bimetallic sample but also its best resistance to deactivation
(
Fig. 1). To confirm the stability of AuPt/TiO , recycling
2
tests have been performed (Fig. 2). Recycling experiments
were supported on TiO . Therefore, in the case of AuPt/
2
1
3

Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites…
Fig. 3 HAADF images of AuPt/TiO . Images a, b, show the presence of nanoparticles ca. 3.7 nm in size
2
Fig. 4 a STEM-HAADF image and b the corresponding XEDS spectra taken from the AuPt nanoparticle. c The corresponding Au Lα and Pt
Lα linescan profile taken across the diameter of the same AuPt nanoparticle (dashed orange arrow in a)
TiO , the support played an active role during the metal
for Au clusters on the rutile TiO (110) surface [29]. In par-
2
2
insertion. Indeed, a combined scanning tunnelling micros-
copy and density functional theory study demonstrated that
bridging oxygen vacancies are the active nucleation sites
ticular, by monitoring the temperature dependence of the
cluster size distribution and the oxygen vacancy density, it
was found that a single Au atom-vacancy complex is stable
1
3

A. Villa et al.
up to room temperature and a single oxygen vacancy can
bind three Au atoms on average. For larger clusters, the
Au-substrate interface contains a high density of oxygen
vacancies, which enhances the binding of Au particles to the
substrate. A synthesis model involving the gold nucleation
and growth on the oxygen vacancies of titania during the
preparation (Scheme 3) can be proposed. These preformed
nanoparticles will act themselves as nucleation sites for the
Pt precursor, according to the EDS findings showing exclu-
sively the presence of AuPt alloyed nanoparticles within the
whole catalysts [20, 28].
It is well known that smaller nanoparticles are more
active than larger ones in liquid phase reactions, probably
due to the higher amount of exposed atoms [1–3]. Keeping
this observation in mind, in order to rule out the effect of the
particle size and to focus on the support effect, the activity
was calculated by normalizing with respect to the fraction
of exposed atoms (Table 1). In that case, the most active
catalyst was AuPt/MgO, in agreement with what reported in
the literature, i.e. that metal nanoparticles on basic supports
are more active than on acid ones for this reaction [7–9, 30].
The AuPt/MgO catalyst resulted also very selective toward
the formation of 1,2-PD (93%) with a small amount of EG
and methanol produced (Table 1). Under basic conditions,
it was reported that the formation of 1,2-PD proceeded via
dehydrogenation of glycerol to glyceraldehyde as firs step
Fig. 5 FTIR difference spectrum reported in the spectral region
of 8a–8b ring modes of 2,6-DMP collected exposed to 2,6-DMP
(
2 mbar) at r.t. on AuPt/MCM-41 (blue dotted curve), curvefit of the
spectrum (orange curve) and deconvoluted bands (grey curves)
(
Scheme 2) [7, 8]. The limited formation of EG can be due
to the difficulty for the retro-aldol reaction of glyceraldehyde
to take place under alkaline conditions [7].
To be noted that the value obtained for AuPt/TiO is very
2
closed to the one of AuPt/MgO (Table 1).
The normalized activity seems to decrease by increas-
ing the support activity. Therefore, to elucidate the effect
of the nature of the support on the catalytic performance,
a detailed characterization of the surface properties of the
catalysts has been performed, in particular by focusing on
the acid supports.
Fig. 6 FTIR difference spectrum reported in the spectral region of
8
a–8b ring modes of 2,6-DMP collected after the inlet of 2,6-DMP
(
2 mbar) at r.t. on AuPt/SiO (blue dotted curve), curvefit of the spec-
2
trum (orange curve) and deconvoluted bands (grey curves)
−1
1630 cm , respectively. A careful deconvolution of the nor-
malised spectra collected upon interaction with 2,6-DMP
and after outgassing at r.t. for 30 min was carried out (see
Figs. 5, 6, 7, 8, 9) and the calculated integrated areas of
the produced absorption bands (normalized to the weight
of the pellets and to the SSA) are reported in Table 2. This
allowed us to compare the obtained quantitative informa-
tion about the different acidic sites present on the catalysts
and to afford structure-activity correlations with the catalytic
results reported in Table 1.
The normalised FTIR spectra collected on the different
catalysts exposed to 2 mbar 2,6-DMP at r.t. and after 30 min
outgassing at the same temperature are shown in the wave-
−
1
number range 1500–1700 cm , (dotted curves in Figs. 5, 6,
7
, 8, 9). It is worth noting that the use of 2,6-DMP allowed
to characterise both Lewis and Brønsted acidic centers
present at the surface and offered the advantage to distin-
guish among Lewis and/or Brønsted sites of different acidic
strength [31, 32].
The former species, together with H-bonded species, gave
Complex bands related to liquid like and H-bonded 2,6-
DMP were produced upon 2,6-DMP adsorption at r.t. (see
Figs. 5, 6, 7, 8, 9). Therefore, the 2,6-DMP interaction was
given mainly via H bonding with the OH groups exposed
at the surface of the samples, suggesting that OH groups
with weak acid strength play a role in the C–C cleavage to
produce EG and methanol (see Tables 1, 2). This is not true
−
1
rise to three bands in the 1620–1580 cm range, where
the 8a and 8b ring stretching modes can be observed [33].
The presence of the latter species, i.e. Brønsted sites, was
revealed by the formation of the 2,6-dimethylpyridinium
+
species (2,6-DMPH ), that presented the 8a–8b bands of
the ring stretching modes, at 1640–1655 cm− and about
1
1
3

Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites…
Fig. 7 FTIR difference spec-
trum reported in the spectral
region of 8a–8b ring modes
of 2,6-DMP collected after
the inlet of 2,6-DMP (2 mbar)
at r.t. and inset after 30 min
outgassing, on AuPt/S-ZrO
2
(
blue dotted curves), curvefits
of the spectra (orange curves)
and deconvoluted bands (grey
curves)
Fig. 8 FTIR difference spec-
trum reported in the spectral
region of 8a–8b ring modes of
2
,6-DMP collected after the
inlet of 2,6-DMP (2 mbar) at r.t.
and inset after 30 min outgas-
sing, on AuPt/H-Mordenite
(
blue dotted curves), curvefits
of the spectra (orange curves)
and deconvoluted bands (grey
curves)
in the case of AuPt/TiO , which exposed almost exclusively
On the contrary, both Lewis and Brønsted acid sites
2
Lewis sites, according to the residual intensities of the bands
after outgassing (given by the integrated areas in Table 2).
No Lewis and Brønsted acidity was observed in the case
were observed on AuPt/S-ZrO (total normalised areas
2
1.831 and 8.368, respectively) and AuPt/TiO (total nor-
2
malised areas 4.401 and 8.472, respectively), whereas
AuPt/H-Mordenite showed almost exclusively Brønsted
acidity (2.38). Interestingly, as revealed by the deconvo-
lution of the spectra collected after outgassing for 30 min
of AuPt/MCM-41 and AuPt/SiO (the total normalised area
2
of the bands related to H-bonded and physisorbed species
was 12.783 and 18.91, respectively). The higher stability to
the outgassing shown by the bands related to AuPt/MCM-
at r.t., the amount of Lewis sites present on AuPt/TiO
2
4
1, with respect to those detected on AuPt/SiO , was due to
was higher than that obtained for AuPt/S-ZrO , where the
2
2
the presence of mesopores, where the H-bonded 2,6-DMP
can be removed only by prolonging the outgassing time (data
not shown for the sake of brevity).
sulphate groups are occupying these sites. Moreover, the
apparent absence of Lewis sites on the AuPt/H-Mordenite
catalyst was ascribed to the presence of residual water
1
3

A. Villa et al.
Fig. 9 FTIR difference spec-
trum reported in the spectral
region of 8a–8b ring modes of
2
,6-DMP collected after the
inlet of 2,6-DMP (2 mbar) at r.t.
and inset after 30 min outgas-
sing, on AuPt/TiO (blue dotted
2
curves), curvefits of the spectra
(
orange curves) and deconvo-
luted bands (grey curves)
Table 2 Position and integrated areas of the bands observed upon 2,6-DMP dosage at r.t
Band integrated areas of adsorbed 2,6-DMP
Brønsted acid sites
H-bonded
Physisorbed
H-bonded and
physisorbed
Position (cm−1
AuPt/MCM-41
AuPt/SiO2
)
1648
1640
1632
1603
2.404
5.169
1595
5.256
2.022
1581
5.123
11.719
Brønsted acid sites
Lewis acid sites and
H-bonded
Physisorbed
H-bonded and
physisorbed
AuPt/S-ZrO2a
AuPt/TiO2a
3.480
.444
3.621
.377
0.365
.334
3.186
1.702
5.832
1.043
0.429
1.908
1.688
2.384
1.046
5.318
2.443
0.9286
2.318
6.609
0.785
14.738
1.326
1.616
6
3.808
2.111
0.107
7.073
0.632
0.200
0
AuPt/H-Mordenitea
1
aThe deconvolution was performed also on the spectra collected after 30 min outgassing at r.t. (refer to the second line for each catalyst)
molecules adsorbed on H-Mordenite activated by simply
outgassing at 120 °C for 1 h.
(around 4 nm) and by the Brønsted acidity of the support. In
our case, for the AuPt/H-Mordenite catalyst, an effect on the
selectivity to 1,2-PD due to the additional presence of gold,
which acts on the electronic properties of Pt, can be inferred.
Interestingly, Brønsted sites are also present at the sur-
On the basis of the FTIR results, the low selectivity to
1
,2-PD displayed by both AuPt/S-ZrO and AuPt/H-Mor-
2
denite it has been related to the presence of Brønsted sites,
that are able to give over-hydrogenolysis of 1,2-PD, form-
ing 1-propanol and 2-propanol. An high 1,3-PD selectivity
face of AuPt/TiO , i.e. the most active catalyst (according
2
to Table 1). Our results gave precise indication that despite
(
48.6%) in the presence of a 2 wt% Pt/H-Mordenite catalyst
the amount of these sites is comparable to that detected on
at 94.9% glycerol conversion was also reported [21]. The
selectivity to 1,3-PD was influenced by the Pt dispersion
AuPt/S-ZrO , and higher than that observed for AuPt/H-
2
Mordenite, the behaviour towards the outgassing at r.t. is
1
3

Diols Production From Glycerol Over Pt-Based Catalysts: On the Role Played by the Acid Sites…
completely different. Indeed, on both AuPt/S-ZrO (Fig. 7)
3. Corma A, Iborra S, Velty A (2007) Chem Rev 107:2411–2502
2
4
.
Katryniok B, Kimura H, Skrzyńska E, Girardon J-S, Fongarland
P, Capron M, Ducoulombier R, Mimura N, Paul S, Dumeignil F
and AuPt/H-Mordenite (Fig. 8) the amount of Brønsted
sites is enhanced after outgassing (the integrated area values
move from 8.368 up to 12.276 and from 2.38 up to 3.022,
respectively) indicating that at low coverages a rearrange-
ment of the 2,6-DMP molecules on the most acidic sites has
occurred. Differently, a decrease in intensity of the bands
related to the Brønsted sites is observed in the case of AuPt/
(
2011) Green Chem 13:1960–1979
5. Davis SE, Ide MS, Davis RJ (2013) Green Chem 15:17–45
6. Villa A, Dimitratos N, Chan-Thaw CE, Hammond C, Prati L,
Hutchings GJ (2015) Acc Chem Res 48(5):1403–1412
7
8
.
.
Nakagawa Y, Tomishige K (2011) Catal Sci Technol 1:179–190
Nakagawa Y, Tamura M, Tomishige K (2014) J Mater Chem A
2:6688–6702
TiO (from 8.472 down to 2.917, see Fig. 9), giving evidence
9. Wang Y, Zhou J, Guo X (2015) RSC Adv 5:74611–74628
2
1
0. Martin A, Armbruster U, Gandarias I, Arias PL (2013) Eur J Lipid
Sci Technol 115:9–27
of the presence of less strong Brønsted sites on this catalyst.
On the contrary, the FTIR spectroscopic characterisation
indicated that the superior activity as well as the high selec-
1
1. Zhou C-H, Beltramini JN, Fan Y-X, Lu GQ (2008) Chem Soc Rev
37:527–549
tivity displayed by AuPt/TiO can be related to the presence
12. Yue H, Zhao Y, Ma X, Gong J (2012) Chem Soc Rev
2
4
1:4218–4244
of Lewis sites with moderate strength. More acidic supports
present a lower activity. Similar results were reported in the
literature by Guo et al. [34]. Indeed, they reported that Cu
immobilized on alumina with moderate acid sites is more
active than more acidic zeolite supports.
1
3. Zhao X, Wang J, Yang M, Lei N, Li L, Hou B, Miao S, Pan X,
Wang A, Zhang T (2017) ChemSusChem 10:819–824
14. Bhanuchander P, Shanthi Priya S, Pavan Kumar V, Hussain S,
Pethan Rajan N,, Bhargava SK, Chary KVR (2017) Catal Lett
1
47:845–855
1
1
1
1
5. Fan Y, Cheng S, Wang H, Ye D, Xie S, Pei Y, Hu H, Hua W, Hua
Li Z,a Qiao M, Zong B (2017) Green Chem 19:2174–2183
6. Dasari MA, Kiatsimkul P-P, Sutterlin WR, Suppes GJ (2005) Appl
Catal A 281:223–231
4
Conclusion
7. Barbelli ML, Santori GF, Nichio NN (2012) Bioresour Technol
1
11:500–503
Bimetallic AuPt catalysts were tested in the glycerol hydrog-
enolysis, using supports with different acid–base properties.
It was found that the smallest AuPt nanoparticles (with aver-
age diameter equal to 3.7 nm) showed a better activity than
largest ones (6–7 nm).
8. Villa A, Chan-Thaw CE, Campisi S, Bianchi CL, Wang D,
Kotula PG, Kuebel C, Prati L (2015) Phys Chem Chem Phys
17:28171–28176
1
2
9. Villa A, Veith GM, Prati L (2010) Angew Chem Int Ed
4
9:4499–4502
0. Villa A, Campisi S, Mohammed KMH, Dimitratos N, Vindigni
F, Manzoli M, Jones W, Bowker M, Hutchings GJ, Prati L (2015)
Catal Sci Technol 5:1126–1132
However, by normalizing the activity (conversion) on the
number of exposed atoms, it appears clearly that the activ-
ity and the selectivity were significantly influenced by the
acid base properties of the support. As expected, AuPt on
MgO appears the most active catalyst, despite it deactivates
quite rapidly. The activity in the case of acidic supports
2
1. Shanthi Priya S, Bjanuchander P, Pavan Kumar V, Dumbre DK,
Periasamy SR, Bhargava SK, Lakshmi Kantam M (2016) ACS
Sustain Chem Eng 4:1212–1222
22. Miyazawa T, Kusunoki Y, Kunimori K, Tomishige K (2006) J
Catal 240:213–221
2
3. Montassier C, Menezo JC, Moukol J, Naja J (1991) J Mol Catal
follows the order (AuPt/TiO ) > (AuPt/MCM41 ≥ AuPt/
2
7
0:65–84
SiO )>(AuPt/H-Mordenite>AuPt/S-ZrO ). The observed
2
2
24. Nakagawa Y, Shinmi Y, Koso S, Tomishige K (2010) J Catal
272:191–194
trend can be directly related to the acid character of the sup-
port: the higher the strength of the acidic sites, the lower
the activity. In particular, the presence of Lewis acid sites is
required for achieving good catalytic performances. Moreo-
ver, it was shown that the nature of the acidic sites plays also
an important role in tailoring the selectivity, by lowering the
production of 1,2-PD and by directing the glycerol hydrog-
enolysis reaction to EG and methanol by C–C cleavage and/
or 1-propanol and 2-propanol by over-hydrogenolysis of 1,2-
PD. Finally, it has been shown that the intermediate acidic
25. Khalil KMS (2007) J Colloid Interface Sci 315:562–569
2
6. Mori K, Hara T, Mizugaki T, Ebitani K, Kaneda K (2004) J Am
Chem Soc 126:10657–10666
2
7. Dimitratos N, Villa A, Prati L, Hammond C, Chan-Thaw CE,
Cookson J, Bishop PT (2016) Appl Catal A 514:267–275
28. Chan-Thaw CE, Chinchilla LE, Campisi S, Botton GA, Prati L,
Dimitratos N, Villa A (2015) ChemSusChem 8:4189–4194
2
9. Wahlström E, Lopez N, Schaub R, Thostrup P, Rønnau A, Africh
C, Lægsgaard E, Nørskov JK, Besenbacher F (2004) Phys Rev
Lett 303:026101-1–026101-4
30. ten Dam J, Hanefeld U (2011) ChemSusChem 4:1017–1034
3
1. Morterra C, Meligrana G, Cerrato G, Solinas V, Rombi E, Sini
MF (2003) Langmuir 19:5344–5356
character of TiO leads to an improved selectivity towards
2
1
,2-PD.
3
2. Toledo-Antonio JA, Cortés-Jácome MA, Navarrete J, Angeles-
Chavez C, López-Salinas E, Rendon-Rivera A (2010) Catal Today
1
55:247–254
3
3
3. Morterra C, Cerrato G, Meligrana G (2001) Langmuir
References
1
7:7053–7060
4. Guo LY, Zhuo JX, MaO JB, Guo XW, Zhang SG (2009) Appl
Catal A 367:93–98
1
2
.
.
Gallezot P (2012) Chem Soc Rev 41:1538–1558
Huber GW, Iborra S, Corma A (2006) Chem Rev 106:4044–4098
1
3