Platinum, palladium and nickel supported on Fe3O4 as catalysts for glycerol aqueous-phase hydrogenolysis and reforming

Article abstract of DOI:10.1016/j.apcata.2017.07.023

Aqueous-phase hydrogenolysis (APH) and reforming (APR) are intertwined reactions with great interest for biomass valorization, as APR generates hydrogen in situ at preferred APH reaction conditions instead of using external sources. We investigated iron oxide-supported Pt, Pd and Ni as catalysts for these reactions using glycerol as starting material. Catalysts were characterized by ICP, N2 physisorption, H2 chemisorption, XRD, TPR, CO2-TPD, FTIR of pyridine adsorption, XPS and TEM. The catalyst tests were performed in an autoclave extended with an in situ pre-reduction setup. Hydrogen spillover was found from hydrogen chemisorption of Pt catalysts, and after the development of a custom-made methodology, the Pt particle size was found to be 1.8 nm for the 2.5Pt/Fe₃O₄ catalyst, which agrees with TEM analysis (1.56 nm). Both for hydrogenolysis and APR, the activity order was Pt > Pd > Ni, directly related to the stability of reduced small nanoparticles, which has an impact on acid-base properties of the support. Pt showed the best performance in hydrogenolysis even at a low content (81% conversion, 79% 1,2-propanediol selectivity for 1.0Pt/Fe₃O₄). The 2.5Pd/Fe3O4 catalyst showed the highest hydrogen yield in APR (69.2% at 513 K and 78.9% conversion). First order kinetic fits were made for 2.5Pt/Fe₃O₄, from which an apparent activation energy of 61.1 kJ mol^?1 was obtained and an initial turnover frequency (TOF₀) of 0.121 s^?1.

Full text of DOI:10.1016/j.apcata.2017.07.023

Accepted Manuscript
Title: Platinum, Palladium and Nickel supported on Fe O as
3
4
catalysts for glycerol aqueous-phase hydrogenolysis and
reforming
Authors: Andr e´ Von Held Soares, Hanan Atia, Udo
Armbruster, Fabio Barboza Passos, Andreas Martin
PII:
S0926-860X(17)30331-9
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2017.07.023
APCATA 16328
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
8-4-2017
10-7-2017
14-7-2017
Please cite this article as: Andr e´ Von Held Soares, Hanan Atia, Udo Armbruster,
Fabio Barboza Passos, Andreas Martin, Platinum, Palladium and Nickel supported on
Fe3O4 as catalysts for glycerol aqueous-phase hydrogenolysis and reforming, Applied
Catalysis A, Generalhttp://dx.doi.org/10.1016/j.apcata.2017.07.023
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Platinum, Palladium and Nickel supported on Fe
aqueous-phase hydrogenolysis and reforming
3
O
4
as catalysts for glycerol
a,b
c
c
a
André Von Held Soares , Hanan Atia , Udo Armbruster , Fabio Barboza Passos* , Andreas
Martinc
aDepartamento de Engenharia Química e de Petróleo, Universidade Federal Fluminense, Rua Passo da Pátria,
1
56, Niterói, RJ, CEP 24210-240, Brazil
bInstituto Federal do Rio de Janeiro – IFRJ, Avenida República do Paraguai, 120, Sarapuí– Duque de Caxias,
RJ, CEP: 25050-100, Brazil
cLeibniz-Institut für Katalyse e.V. an der Universität Rostock, Albert-Einstein-Str. 29a, D-18059 Rostock,
Germany
*
Corresponding author. E-mail address: fbpassos@vm.uff.br
Graphical abstract
Highlights for “Platinum, Palladium and Nickel supported on Fe3O4 as catalysts for glycerol
aqueous-phase hydrogenolysis and reforming” by Soares et al.




Glycerol hydrogenolysis and reforming were studied on Pt, Pd and Ni/Fe
3
O .
4
Both for glycerol hydrogenolysis and APR, the activity order was Pt > Pd > Ni.
Glycerol hydrogenolysis and reforming shared some of the reaction steps.
2.5Pt/Fe
3
4
O was active and stable.
ABSTRACT
Aqueous-phase hydrogenolysis (APH) and reforming (APR) are intertwined reactions with
great interest for biomass valorization, as APR generates hydrogen in situ at preferred APH
1

reaction conditions instead of using external sources. We investigated iron oxide-supported
Pt, Pd and Ni as catalysts for these reactions using glycerol as starting material. Catalysts
were characterized by ICP, N
2
physisorption, H
2
chemisorption, XRD, TPR, CO -TPD, FTIR
2
of pyridine adsorption, XPS and TEM. The catalyst tests were performed in an autoclave
extended with an in situ pre-reduction setup. Hydrogen spillover was found from hydrogen
chemisorption of Pt catalysts, and after the development of a custom-made methodology, the
Pt particle size was found to be 1.8 nm for the 2.5Pt/Fe
3
O4 catalyst, which agrees with TEM
analysis (1.56 nm). Both for hydrogenolysis and APR, the activity order was Pt > Pd > Ni,
directly related to the stability of reduced small nanoparticles, which has an impact on acid-
base properties of the support. Pt showed the best performance in hydrogenolysis even at a
low content (81% conversion, 79% 1,2-propanediol selectivity for 1.0Pt/Fe
.5Pd/Fe catalyst showed the highest hydrogen yield in APR (69.2% at 513 K and 78.9%
conversion). First order kinetic fits were made for 2.5Pt/Fe , from which an apparent
activation energy of 61.1 kJ·mol was obtained and an initial turnover frequency (TOF ) of
3
O4
). The
2
3
O4
3O4
-1
0
-
1
0
.121 s .
Keywords: Glycerol hydrogenolysis, APR, platinum, palladium, iron oxide, in situ pre-
reduction
1
. Introduction
Within the conceptual framework of biomass valorization, the transformation of
glycerol into specialty and commodity chemicals cannot be overstated. Aqueous-phase
hydrogenolysis (APH) and reforming (APR) of glycerol are intertwined reactions that differ
from one another regarding the origin of hydrogen. While H
2
is externally fed and consumed
in APH, H is produced in situ from glycerol conversion in APR. The connection between
2
these two reactions lies in the fact that they can be operated at the same hydrothermal
conditions (200-250 °C, 20-50 bar), and by coupling both reactions, APR may serve as
hydrogen source for APH, finally forming hydrocarbons, alcohols, polyols and carbonylated
structures.
2

Hydrogenolysis is defined as the process of cleaving a C-X bond between a carbon
and a hetero-atom or another carbon atom, followed by the addition of hydrogen to the
cleaved fragments [1]. The probability of C-C bond cleavage also leads to formation of
smaller molecules. The mechanism for hydrogenolysis of polyols may involve, however,
stable intermediates, for which dehydration steps play an important role, such as, for example,
acetol and acrolein in the case of glycerol conversion. The main liquid phase products of
glycerol APH are 1,2-propanediol (1,2-PD), 1,3-propanediol (1,3-PD), 1-propanol (1-PrOH),
iso-propanol (2-PrOH), hydroxyacetone (acetol), ethylene glycol (EG), ethanol and methanol.
On the other hand, APR was developed as a process that generates hydrogen as a value-added
product from aqueous carbohydrate solutions, especially from waste-water effluents [2].
Given that glycerol stems almost exclusively from biodiesel and soap production as a cheap
by-product, its valorization may happen through both reactions using different catalysts [3,4].
Iron(III) oxide (Fe
splitting [6], water treatment [7], photocatalysis [8], dehydrogenation [9,10] and Fischer-
Tropsch synthesis [11], to name a few. Fe is active for glycerol dehydration and hydrogen
2
O
3
) has many uses in catalysis, such as gas sensors [5], water
2
O
3
transfer, producing allylic alcohol, acetol and acrolein as the main products [12].
Catalysts that have platinum as the active metal and iron oxide as the support have
been described in literature for many purposes, including electrocatalytic oxidation of ethanol
[13], n-heptane isomerization [14] and the oxidation of formaldehyde [15]. In accordance to
other studies on Pt-Fe catalysts applied to the reactions of the present work [16-18] we
previously showed [19] that low contents of Fe enhance the activity of alumina-supported Pt
catalysts via bimetallic particle formation.
The present work is a comparative study on the activity of platinum, palladium and
nickel as active metals supported on iron oxide as catalysts in APH and APR of glycerol.
Particular attention is given herein to Pt/Fe
3
4
O . Although all metals belong to group 10 of the
3

periodic system, the catalytic activity varies greatly and pertinent characterization of materials
is provided in order to clarify the connection between the catalyst’s structure and its activity.
2
. Experimental
.1. Catalyst preparation
Active metals (Pt, Pd, and Ni) were added through incipient wetness impregnation
using H PtCl ·6H O (Alfa Aesar®), PdCl (Alfa Aesar®), and NiCl (Alfa Aesar®). For iron
oxide preparation, 227.59 g of Fe(NO ·9H O were dissolved in 241 mL of H O, and 0.807
mL of an aqueous solution of 50 wt% of Al(HPO (Alfa Aesar®) were added. After mixing,
2
2
6
2
2
2
3
)
3
2
2
)
3 3
water was distilled away under vacuum until a brown gel-like material was obtained. The gel
was then calcined under 150 mL/min of air flow at 653 K for 1.5 hour, which was reached
through heating at a rate of 2 K/min. After impregnation of the desired metals onto Fe
2
O , the
3
catalysts were once again calcined at 673 K for 4 hours with 4 K/min. All catalyst reductions
at 523 K or lower were made using the autoclave setup described below. Reduction of the
catalysts above 523 K was performed in a separate tubular quartz reactor. Fe
Sigma Aldrich®) was used as a reference material for BET surface area.
Throughout the present text, fresh calcined catalysts are labelled as X.XM/Fe
where X.X is the metal content in wt%, and M the impregnated metal (Pt, Pd or Ni). Reduced
catalysts are addressed as X.XM/Fe and were prepared using the autoclave in situ
and 2.5Ni/Fe , which were reduced ex situ under hydrogen
2
O
3
nanocat
(
2
3
O ,
3
O
4
methodology, except for Fe
flow at 673 K.
3
O
4
3
O
4
2
.2. Catalyst characterization
Chemical analysis was done by inductively coupled plasma-optical emission
spectroscopy, using ICP-OES spectrometer Optima 3000XL (Perkin Elmer). Samples were
4

digested in a microwave oven (Anton Paar Multiwave from Perkin Elmer) with a mixture of
aqua regia and HCl.
Surface area, pore volume and pore size distribution were determined by N
2
physisorption at 77 K, using an ASAP 2010 instrument from Micromeritics. Prior to the
measurement, the samples were evacuated at 473 K and 0.1 mbar for 4 h. Surface area was
calculated by BET method. Pore size distributions were calculated by the BJH method applied
to the desorption step of the isotherm.
Hydrogen chemisorption was measured with an Autochem II 2920 equipment from
Micromeritics. A tailor-made methodology was conceived and performed in order to assess
and correct the spillover effect on 1.0Pt/Fe
3
O
4
and 2.5Pt/Fe
3
O . The methodology and its
4
hypotheses are described in detail in the Supplementary Informarion document.
Temperature Programmed Reduction (TPR) was conducted with Autochem AC2920
from Micromeritics. After drying the catalyst under Ar flow, the reduction was performed
with 50 mL/min of 5%H
2
/Ar at a rate of 10 K/min. The consumption of H was monitored
2
with a thermal conductivity detector.
X-ray powder diffraction (XRD) patterns were obtained using a STADI P automated
1
transmission diffractometer (STOE, Darmstadt) with CuKα radiation and Ge
monochromator. Scans were performed for 2θ within the range of 15-80° (step width 0.5°,
1
00 s/step) and recorded with a STOE position sensitive detector.
X-ray photoelectron spectra (XPS) were obtained with an ESCALAB 220iXL
(ThermoScientific) with monochromatic AlKα radiation. The peak for C1s at 284.8 eV was
used for calibration, and the spectra were fitted with Gaussian-Lorentzian curves for
determination of the peaks’ maxima and areas. All elements were analyzed from their
principal peak.
5

Transmission electron microscopy (TEM) was performed using a Philips CM200
microscope equipped with a LaB6 filament and a supertwin lens operating at 200 kV. Bright-
field images were acquired using a CCD camera (TVIPS GmbH). Samples of the catalyst
powder were dispersed in ethanol with ultrasound, and drops of the dispersion were deposited
on copper grids coated with amorphous carbon films. Particle were counted using ImageJ®
software after setting the scale, selecting an evenly illuminated region of the micrograph,
applying a bandpass filter and adjusting the threshold.
Basic sites of the catalysts were determined by temperature-programmed carbon
dioxide desorption (CO
The samples were pretreated under 50 mL/min of He, and heated to 523 K at a rate of 10
K/min. Then they were cooled to 373 K and exposed to 50 mL/min of 1.2%CO /He for 90
min. After this, the samples were flushed with He still at 373 K for 30 min and then cooled to
43 K, the starting temperature of the analysis. Heating to 973 K was done at a rate of 10
K/min.
2
-TPD) using an Autochem II 2910 equipment from Micromeritics.
2
3
For acidity characterization pyridine was used as probe molecule. The measurements
in transmission mode were carried out on a Bruker Tensor 27 FTIR spectrometer equipped
with a heatable and evacuable homemade reaction cell with CaF windows connected to a
2
gas-dosing and evacuation system. The sample powders were pressed into self-supporting
wafers with a diameter of 20 mm and a weight of 50 mg. Before pyridine adsorption, the
samples were pretreated by heating in synthetic air up to 673 K for 10 min, subsequent
cooling to room temperature and evacuation. Pyridine was adsorbed at room temperature until
saturation. Then the reaction cell was evacuated to remove physisorbed pyridine. The
desorption of pyridine was followed by heating the sample in vacuum up to 673 K and
-1
recording spectra every 50 K. The spectra were measured with 2 cm resolution and 64 scans.
6

2
.3. Catalyst tests
All tests were performed in a 300 mL benchtop autoclave from Parr Instruments, equipped
with a 40 mL catalyst stationary basket. Temperatures were maintained within ± 2 K from the
target temperature. In a typical reaction, glycerol (>98%, Roth®) was used in aqueous
solutions with a concentration of 10 wt%. Argon (Air Liquide 5.0) was used for drying the
catalysts and to pressurize the autoclave in APR experiments, and hydrogen (Air Liquide 5.0)
was used for catalyst pre-reduction and APH tests. Hydroxyacetone (90%, Sigma Aldrich)
was used to prepare an aqueous solution for acetol hydrogenation. The setup used for catalyst
pre-reduction and evaluation, the methodology for in situ pre-reduction and for the
experiments are all described in detail in the Supplementary Information document.
The liquid phase products were analyzed by GC-FID (Shimadzu GC-17A) with
autosampler (Chrompack FFAP column, 25 m × 0.32 mm × 0.3 µm), carrier He and 1,4-
butanediol (99.9% from Sigma Aldrich) as internal standard. Initial method conditions were T
=
313 K, p = 0.5 bar, vHe = 30 cm/s, split ratio = 50, and calibrations were done for glycerol,
1
,2-PD, 1,3-PD, 1-PrOH, 2-PrOH, EG, acetone, acetol, ethanol, methanol, acetic acid,
propionic acid, acrolein and allyl alcohol. The carbon balance in liquid phase was cross-
checked with Total Organic Carbon analysis (TOC) using a TOC-V CPN equipment from
Shimadzu. For product identification, GC-MS analysis was performed in a Shimadzu 2010
furnished with a CP-SIL 5 CB column (60 m × 0.32 × 8 µm). Gases were analyzed by an
Agilent Technologies 7985A GC, equipped with TCD and FID detectors calibrated for
permanent gases and hydrocarbons, furnished with CP Molsieve (25 m × 0.53 mm × 20µm)
and PoraPLOT Q-HT (25 m × 0.53 mm × 20 µm) columns.
Glycerol conversion and selectivity were calculated on carbon basis. H
the percentage of produced H moles over the initial quantity of glycerol moles in the feed
solution, and the initial turnover frequency (TOF ) was calculated using the number of active
2
yield is defined as
2
0
7

Pt sites on the surface (Ptsurf) as obtained from chemisorption analysis. After the reaction,
catalysts were separated from the solution using a magnet and filtered off, and the final
solution was weighted and submitted to TOC analysis.
3
. Results and discussion
.1. Catalyst characterization
3
Table 1 shows the results from ICP analysis for each as-synthesized as well as spent
platinum catalyst. The results indicate satisfactory impregnation of all the metals. Although
Al(HPO
3
3
) was used for preparation of the solids, its contribution to the composition of the
catalysts is negligible, and will thus be ignored in the discussion.
Texture and chemisorption results are available in Table 2. The samples show
characteristic type II [20] isotherms corresponding to relatively small surface areas. The shape
of the BET isotherms points to materials constituted mainly of separate particles with
interstitial volume. BET area for the as-synthesized Fe
Fe nanocat used as reference (32 m²/gcat). The reduction of the support causes a decrease
in surface area, which remains stable during subsequent glycerol conversion. A much larger
surface area drop from 160 to 16 m²/gcat was reported for high surface area-Fe used for
2
O
3
support is similar to commercial
2
O
3
2
O
3
dehydration and subsequent hydrogen transfer from glycerol to other products [12]. This fact
is favors the presence of Pt particles over the synthesized support for catalyst stability
regarding surface area.
Hydrogen chemisorption analysis was made for Fe
Table 2). The tailor-made measurement protocol was used for the assessment of Pt particles,
considering a correction for the hydrogen spillover phenomenon, since the amount of
3
O
4
, 1.0Pt/Fe
3
O
4
and 2.5Pt/Fe
3
O
4
(
st
nd
hydrogen uptake initially measured (1 and 2 measures) was considerably larger than
necessary for the impregnated amount of Pt. Previous tests revealed that only under vacuum
8

-2
(
10 mbar) at a reduction temperature of 523 K it was possible to remove efficiently all the
chemisorbed hydrogen from partially reduced iron oxide support. The hydrogen spillover
effect over Pt/Fe is responsible for the reduction of Fe to Fe in the presence of Pt at
3
O
4
2
O
3
3
O
4
low temperatures [21,22], which is also discussed below along with the TPR results. As
described in the experimental section, one of the hypotheses for the chemisorption
methodology is that the hydrogen desorption energy for Pt particles differs from the
desorption energy for sites where hydrogen atoms are spilled over, i.e., diffuse over the
surface. It was also assumed that the flow of inert argon at the reduction temperature of the
catalyst supplies enough energy to displace the hydrogen atoms attached only to the sites of
lowest desorption energy, but not enough to displace hydrogen chemisorbed on the highest
desorption energy sites.
The results revealed that these hypotheses were in accord with reality. Firstly, as seen
in Table 2, there is no doubt that chemisorption occurs on the iron oxide support, but only in
the presence of Pt particles, since no hydrogen uptake was measured for Fe
3
4
O . The second
rd
interesting result was that the dispersion of Pt particles calculated from the 3 chemisorption
uptake for 1.0Pt/Fe and 2.5Pt/Fe perfectly agrees with literature [23,24] and with the
particle sizes observed in the TEM results of this work (see below). Therefore, it is reasonable
to state that the adsorption energy of the hydrogen atoms on the support Fe is higher than
the adsorption energy on the active Pt metal particles. After correcting the spillover effect,
catalysts 1.0Pt/Fe and 2.5Pt/Fe displayed metal dispersions and average particle sizes
3
O
4
O
3 4
3
O
4
3
O
4
O
3 4
of 71% and 1.6 nm and 63% and 1.8 nm, respectively. The increase in Pt content causes a
moderate drop in dispersion and little change in particle size. However, hydrogen spillover
has a significant impact on catalytic activity, as discussed below. Figure 1 illustrates the steps
and findings of the chemisorption methodology applied to quantify and correct the hydrogen
spillover effect.
9

The procedure is not free from objections. For instance, the sample handling in order
to evacuate bears the risk of exposing the reduced sample to atmospheric air, which can lead
to deviations from real dispersion and particle size. However, this is possible for every ex situ
reduction prior to characterization, but does obviously not seem to impede the procedure.
Furthermore, the consistency of the particle diameters obtained from chemisorption (1.8 nm)
and TEM studies (1.8 nm) is remarkable for 2.5Pt/Fe
3
4
O . A second possible objection refers
to the argon purge, asking whether the flow would be able to selectively remove the
chemisorbed hydrogen from Pt species. Once again, the results justify the assumptions. The
difference between the third uptake measure and the first two is larger than any difference
between the first and the second uptake measures. If purging would remove all hydrogen, the
third uptake measure should be the same as the second or slightly lower. Therefore three
measurements were performed for each material. Nonetheless, there is some potential for
improvement by flow adjustments and sample evacuation.
Figure 2 shows the TPR profiles for the Fe
.0Pt/Fe and 2.5Pt/Fe catalysts. Table 3 displays the respective hydrogen uptake for
reduction of the materials. In all profiles the hydrogen uptake is in agreement with the
stoichiometry of the reduction of the support, Fe + 3H → 2Fe + 3H O. The excess of
consumed H in comparison to the calculated values suggests the reduction of oxychlorated
artefacts from the impregnation process. The first peak in the reduction of Fe at 630 K was
assigned to the transition Fe →Fe . , The first peak for the Pd and Pt catalysts
comprehends the reduction of the active metals and the iron oxide transition (Table 3).
2
O
3
support and 2.5Ni/Fe
2
O
3
, 2.5Pd/Fe
2
O
3
,
1
2
O
3
2
O
3
O
2 3
2
2
2
2
O
3
2
O
3
O
3 4
Consequently, the active Pd and Pt catalysts reduced at 523 K are labeled as M/Fe
Furthermore, the presence of group 10 metal shifts the support reduction to lower
3
4
O .
temperatures around 510 K, whereas only a slight shift can be observed for Ni. There are
smooth and subtle humps around 430 K for Pt and Pd catalysts, which cannot be seen for
1
0

2
.5Ni/Fe
2
3
O . Hence, the reduction of Ni seems to occur first after Fe
2
3
O gets reduced. After
the reduction of Pt or Pd, the chemisorbed hydrogen remains available on the active metal
sites and diffuses to the surface of the support which is then reduced. According to literature
[15,25-27], Pt/Fe
2
O
3
and Pd/Fe
2
3
O have low reduction temperatures in agreement with our
results. These low reduction temperatures are prerequisite for in situ pre-reduction in the
autoclave due to allowed maximum temperature. It should be also noted that a low content of
1
.0 wt% of Pt is sufficient to shift the transition Fe
Figure 3 shows XRD patterns for as-synthesized Fe
reduced 2.5Pt/Fe , and for spent 2.5Pt/Fe . The calcined Fe
2
O
3
→Fe
3
O
4
to 510 K.
, for calcined 2.5Pt/Fe
and 2.5Pt/Fe
2
O
3
2
O
3
, for
3
O
4
3
O
4
2
O
3
2
O
3
are
shown in the lower part of the figure (a and b). The middle displays the patterns of the 2.5Pt
catalysts after reduction at 423 K, 473 K and 523 K (c, d and e, respectively). The upper part
presents the patterns for spent 2.5Pt/Fe
support and the calcined catalysts are similar to maghemite (-Fe
impregnation of Pt does not modify the lattice structure. No reflection for PtO
3
O
4
(f and g). The patterns for the as-synthesized
), which indicates that
(for the
2
O
3
2
calcined samples) or Pt (for the reduced samples) planes was observed, which is evidence for
well dispersed particles, as corroborated by hydrogen chemisorption and TEM micrographs.
Except for the 2.5Pt/Fe
Figure 3 present patterns that match Fe
and the conclusion that the transition Fe
impregnated catalysts. Therefore, Fe
pre-reduction was done at 523 K. The spent 2.5Pt/Fe
2
O
3
sample reduced at 423 K (c), all the reduced materials displayed in
. This agrees with the TPR results for 2.5Pt/Fe
→Fe can be shifted to 510 K only for the
is the actual support of the catalysts, since the in situ
(f and g) catalysts’ diffractograms
3
O
4
2
3
O ,
2
O
3
O
3 4
3
O
4
3
O
4
also did not show any characteristic reflections for Pt planes, implying a stable state for the
catalyst at the reaction temperature of 513 K.
Micrographs of Pt and Pd catalysts can be seen in Figures 4, 5 and 6, while Figures 7
and 8 show the metal particle size distribution for 2.5Pt/Fe
3
O
4
and 1.0Pt/Fe
3
4
O , respectively
1
1

The metal particles are shown in clear contrast and the elements were determined by EDX.
From the TEM images, the Pt nanoparticles of 1.0Pt/Fe and 2.5Pt/Fe present
3
O
4
O
3 4
dimensions similar to those obtained by hydrogen chemisorption. Figure 5a shows the lattice
orientation of the support and the plane distance was measured as 0.295 nm, which is
consistent with the interplanar distance of (220) planes of Fe
for 2.5Pd/Fe from TEM analysis was not as good as for the Pt catalysts, but the
micrographs clearly indicate the existence of many Pd particles around 1 nm or even smaller
Figure 6a and b). However, the average metal particle size of 2.5Pd/Fe is notably larger
than of Pt catalysts, since particles with diameters of 20 nm were found (Figure 6c).
Assuming a gaussian distribution, the mean diameter for 2.5Pt/Fe is 1.56 nm ± 0.47 nm
Figure 7). The mean diameter for 1.0Pt/Fe was 0.77 ± 0.2 nm (Figure 8). These values are
3
O
4
[28]. The contrast obtained
3
O
4
(
3
O
4
3
O
4
(
3
O
4
consistent with the chemisorption analysis (1.8 nm for 2.5Pt/Fe
obtained from the literature (1.5 nm [13], 2.0 nm [15]).
3
4
O ) and with some values
Figures from the XP spectra can be found in the Supplementary Information
document, alongside with details of the results. By contrasting the obtained results with the
reference literature [29-50], it can be said the 2.5Pt catalyst shows a more stable reduction
behavior than 2.5Pd, which is more stable than 2.5Ni. This order of reduction is in full
agreement with the order of the catalyst activity. The XP spectra reveal that Pt particles are
practically fully reduced prior to the reaction and are very mildly oxidized after the reaction,
while Pd presents two distinct chemical states after reduction, and Ni is not reduced, even at
6
73 K, when there is a transformation from Fe
2
O
3
to Fe
3
O
4
. The Fe
3
4
O -supported particles are
3
+
2+
formed in such a way that Pt-Fe or Pd-Fe interactions stabilize adjacent Fe and Fe sites,
which determine the overall dimensions of the particle and its d band energy. This effect is
more pronounced in Pt catalysts, provided that Pt particles are smaller and more stable than
Pd particles. This heterogeneous structure displays acid and basic sites, side by side with
1
2

nanoparticles that show a high potential to chemisorb glycerol and cleave its bonds by back
donation. Acidity and basicity are considered to play an important role for the activity of the
catalysts in APH of glycerol, and it has been recently reported that the impregnation of a
noble metal increases greatly the acidity of the support [25]. From the changes in binding
energies of the metals towards lower values after reduction, it follows that some acid-base
properties of the catalyst should also be expected to alter after catalyst reduction. Active
Lewis acid sites may interact with the lone electron pairs of oxygen atoms in hydroxyl groups,
and Brønsted acid sites may protonate the hydroxyl, thereby catalyzing dehydration. In turn
basic sites are able to dehydrogenate glycerol by means of electron donation to the hydrogen
atom, thereby cleaving the O-H bond.
Temperature-programmed desorption of CO (TPD) was undertaken to assess whether
2
there is any relation between the catalysts’ activity and their basic properties, and Figure 9
shows the results obtained for the materials. All materials possess basic sites with varying
characteristics, and the impregnated metal enhances the support’s basicity. Two peaks were
found for the calcined Fe
2
O
3
, and their intensity substantially increased after reduction to
Fe and by the addition of an active metal. By correlating the strength of the basic sites to
3
O
4
the desorption temperature in a qualitative scale from weak (273-465 K), moderate (465-640
K) to strong (640-900 K) [26,51], the lower temperature peak is found around 570 K and
ascribed to moderate basic sites, while strong basic sites generate signals at 680 K. As seen
from Figure 9, the reduction of Fe
2
O
3
to Fe
3
4
O increases the basicity of the material, which is
2
+
mainly attributed to the presence of Fe ions on the surface, which are able to donate
2
+
electrons. The presence of Fe on the surface agrees with the XPS results for Fe2p discussed
above. 2.5Pd/Fe presents the largest number of moderate sites, and 2.5Ni/Fe has the
overall largest number of basic and strongest sites. The signals of the strong basic sites are
overlapped by tailing of the moderate sites for the 2.5Pt/Fe catalyst, and are seen as a
3
O
4
3
O
4
3
O
4
1
3

distinct shoulder for the 2.5Pd/Fe
3
4
O catalyst. Table 4 shows that the basic sites are a
characteristic feature of the support, whereas the impregnated metals change the number of
sites only to a limited extent. However, metal addition alters profoundly the nature of the
basicity of the material. Compared to the Fe
decrease in the amount of strong basic sites, but generate an even higher amount of moderate
sites, which does not happen for 2.5Ni/Fe . On the contrary, Ni impregnation increases the
number of strong basic sites. Since basic sites are already present in Fe , and their number
3
O
4
profile, Pt and Pd impregnation cause a
3
O
4
3
O
4
is lower than the molar quantity of impregnated metal, one can argue that the basicity stems
from an effect the metals have on the support, rather than directly from the metallic sites. One
possibility to explain the change in nature of the basic sites from Fe
3
O
4
to 2.5Pt/Fe
3
O and 2.5
4
2
+
Pd/Fe
3
O
4
is that Pt and Pd particles are formed on top of Fe ions, therefore blocking them
and in turn promoting milder sites on the particle-support interface.
Figure 10 shows the IR spectra for pyridine adsorption onto Fe
3
O
4
and 2.5Pt/Fe
3
4
O .
-1
-1
For the presented spectra, wavenumbers around 1604 cm and 1445 cm indicate Lewis acid
sites. The typical band of pyridine adsorption on Brønsted acid sites shows signals around
-1
1
540 cm [52], which are missing in the measured spectra. The fact that the Brønsted acid
sites are negligible on Fe is in agreement with studies on the nature of the acidity of iron
oxide materials [53-55]. The presence of Brønsted sites on Fe is due to impurities [54], as
in the presence of an active metal like Mo, for example [55]. Although
3
O
4
2
O
3
2
-
is the case of SO
4
hydration of the support surface is expected to generate Brønsted acid sites in aqueous
medium, it is likely to be in very limited extent. In light of the hydrogen chemisorption
analysis, the present IR spectra indicate that the spillover effect does not generate protons on
the surface of the support able to act as Brønsted sites. Although an absolute quantification of
-1
acid sites is difficult via IR spectroscopy, the absorbance for the Pt catalyst at 1445 cm is
1
4

2
.52 times more intense than the support’s absorbance, clearly indicating that Pt impregnation
increases the total number of Lewis acid sites.
3
.2. Activity tests
The Weisz-Prater parameter was calculated using a glycerol diffusivity in water equal
-7
to 1.33 × 10 cm²/s, a tortuosity of 3.8, a porosity of 0.4 [56-61], and an average pore
diameter of 14.3 nm as given by the physisorption analysis. The value was calculated to be
-2
3
.38 × 10 << 1,which suffices for kinetic assessment.
Table 5 shows conversions and selectivities for the Fe
3
4
O -supported catalysts used in
the APH of glycerol in regular batch experiments. All materials were submitted to in situ pre-
reduction and 2.5Ni/Fe
O
3 4
was reduced by means of a prior additional reduction step ex situ
in a tubular oven under H
2
flow at 673 K. The results indicate that the platinum-based
catalysts are best performing in APH among the used materials, even at lower Pt content, as is
the case of 1.0Pt/Fe (81% conversion and 79.3% 1,2-PD selectivity), while 2.5Ni/Fe
3
O
4
3
O
4
shows insatisfactory activity as well as low 1,2-PD selectivity. From a comparison of the
obtained conversion results, the order of activity is very clear for the impregnated metals,
namely, Ni < Pd < Pt. 1.0Pt/Fe
glycerol/mcat = 10.4. A recycle experiment was performed for 2.5Pt/Fe
considerable stability under the condition of excess water. In dry environments, using
3
O
4
led to promising results at 493 K at a mass ratio of
m
3
O
4
, which showed
alcohols as hydrogen donors, the stability of iron oxide mixed with alumina as a support for Pt
is noteworthy [16]. However, in the presence of hot water, Pt catalysts are known to suffer
from deactivation processes such as leaching, sintering and acetol polymerization, in the cases
involving glycerol dehydration [25,57].
Table 6 shows the activity and selectivity of the catalysts in the APR of glycerol, and
once again 2.5Pt/Fe
3
O
4
shows the highest conversion. However, 2.5Pd/Fe
3
O presents the best
4
1
5

performance in H
45.3%). The same activity order as in APH is also found in APR: Pt > Pd > Ni. In
comparison with results for Pd/Fe (conversion of 35% at 523 K, 15 ggly/gcat from ref. [26]),
2
production (69.2% yield) and the highest selectivity towards 1,2-PD
(
2
O
3
the conversion obtained in this work reaches the same range, using however a much lower
amount of catalyst per batch (30 ggly/gcat), which seems to affect 1,2-PD selectivity, since in
this work it was about 30%. In view of the fact that the APR of glycerol is mainly done for
biomass valorization through H
ones obtained for ethylene glycol by Huber et al. [17]. The results of this work confirm that
Pd is more active for H production, but there is a discrepancy in activity, since Pt was the
2
production, it is useful to compare the results herein and the
2
most active in glycerol conversion. The interaction of the ethylene glycol molecules and the
metal particles of Pd and Pt differs from the interaction of the glycerol molecules and the
surface of the metal particles of these same metals, despite the similarities of the polyols. As it
can be seen from the carbon balance in Table 6 (only calculated from liquid phase), there is
considerable formation of CO
Although similar APR conversions were obtained for Fe
was higher for the former. The main difference in H
the ability of these metals to chemisorb hydrogen, which is more pronounced with Pd. Thus,
2
and hydrocarbons with 2.5Pt/Fe
3
O
4
and 2.5Pd/Fe
3
O
4
.
2
O
3
and 2.5Ni/Fe
3
O
4
, the H
2
yield
2
yield from Pd to Pt is probably related to
hydrogen atoms on Pd particles combine faster and H
2
is released more easily than Pt
surfaces.
The results displayed in Tables 4 to 6 evidence that the activity of the catalysts
decreases when the number and strength of basic sites increases. Carbon monoxide was either
absent or identified as trace in the analysis of the gas phase, which is due to water-gas shift
reaction as an important part of the APR reaction network.
Further hydrogenolysis experiments with 2.5Pt/Fe
3
4
O were made, since 1,2-PD yield
was highest for this catalyst. Table 7 shows conversions, selectivities, carbon balances and
1
6

turnover frequencies for temperatures from 463 to 513 K, and Figure 11 shows the Arrhenius
plot for APH using a first order apparent kinetic fit. Since hydrogen chemisorption studies
indicated spillover, the turnover frequency was calculated using the third uptake measurement
described herein, assuming a H/Ptsurf stoichiometry equal to one. The turnover frequencies for
2
.5Pt/Fe
3
O
4
are comparable to the previously presented Pt
x
y
Fe /Al
2
3
O catalysts [19]. Apparent
activation energy of 61.1 kJ/mol was obtained, which is slightly lower than for ruthenium
catalysts [58], which also agrees with the value of 63.7 kJ/mol obtained by Jin et al. for Pt/C
[59].
Hydrogen partial pressure is important as it controls solubility and concentration near
the catalyst surface and thus can influence both reforming and hydrogenolysis. It has been
reported that H pressures as low as 3.4 bar are enough to drive hydrogenolysis at 473 K with
5% conversion [60]. It is also known that H solubility drops drastically as glycerol
concentration is increased [61]. For low glycerol concentrations, as used throughout this work
10 wt%), there is very little difference from the solubility of H in pure water [61-63], Table
shows the pressure impact on 2.5Pt/Fe performance in regular batch experiments, and
2
2
2
(
2
8
3
O
4
Figure 12 shows the behavior of the reaction for two constant pressure batch experiments.
Higher pressures cause considerable loss in conversion without increasing 1,2-PD selectivity
significantly (Table 8).
As expected, acetol selectivity decreases with higher H
hydrogenation. The gauge water vapor pressure at 493 K can be estimated to approximately
5 bar. Therefore, in Figure 12, at a total pressure of (i) 35 bar, the H partial pressure is 10
bar, while at a total pressure of (ii) 60 bar, the H partial pressure is 35 bar. The total pressure
2
pressures because of carbonyl
2
2
2
was maintained constant throughout the entire reaction, and 1,2-PD selectivity stabilizes after
some time. Although both experiments displayed in Figure 12 showed similar initial
conversions at t – (i) 5.8%, and (ii) 6.0%, due to glycerol consumption in APR and
0
1
7

dehydrogenation processes prior to hydrogen uptake – they end up with different conversions
after four hours: (i) 17.5%, and (ii) 13.5%. Since 1,2-PD selectivity remains fairly stable and
its increase accompanies the acetol selectivity decrease, this is proof that acetol is being
consumed to form 1,2-PD. The only difference between the conditions of the experiments (i)
and (ii) is the pressure value, and therefore lower conversions obtained for high H
indicate that some step in 1,2-PD formation is hindered by the hydrogen excess. Since acetol
hydrogenation consumes H , it is likely that the decrease in conversion observed when more
2
pressure
2
hydrogen is available ensues from a dehydrogenation prior to acetol production.
The decrease in conversion caused by the increase in hydrogen pressure does not agree
with Le Châtelier’s principle, and, therefore, it has to be a kinetic effect. For the present Pt
catalyst, there are two main explanations for this phenomenon. The first one is that hydrogen
spillover occurs on the surface of the catalyst. Therefore, increasing hydrogen pressures cause
increasing active site blockage on the catalyst surface as a whole. This blockage impedes the
interaction of glycerol and the catalyst surface and slows down the reaction. The second
explanation stems from the reaction mechanism, since APR reactions show that H
produced from glycerol conversion.
2
is
As seen in Figure 13, there are at least three well accepted mechanisms for 1,2-PD
production from glycerol over noble metals, [60,64-71] namely (a) the dehydration-
hydrogenation route [60], (b) the dehydrogenation-dehydration-hydrogenation route [65-68]
and (c) the direct hydrogenolysis route [69,70]. Acetol is an intermediate in the production of
1
,2-PD, and its formation is accepted to run via an enol-keto tautomerization mechanism
during glycerol dehydration over Lewis acid sites [64,72], while acrolein, which is also a
product from the dehydration of glycerol, is thought to be produced over Brønsted acid sites.
As reported elsewhere [66], the formation of 1,2-PD via preceding dehydrogenation over
metal sites is more likely than from a first dehydration catalyzed by acid sites. Despite their
1
8

acidic properties, acetol is not produced when only a bare oxide is used, such as CeO
2
, La
2
O
3
,
ZnO [25], TiO [73] and Al [74]. This poor dehydration activity coupled with the observed
2
2
O
3
dehydrogenation strongly suggests that metallic and acid functions are required in the typical
temperature range of hydrogenolysis (453-523 K). The very low selectivity for 1,3-PD in our
APH and APR experiments is also evidence that a Brønsted acid catalyzed protonation of the
secondary hydroxyl is unlikely to occur over the iron oxide-supported catalysts. The order of
metal activity, Pt > Pd > Ni, both with and without external H , shows that steps in the
2
production of acetol are deeply connected with the bifunctional ability of the metal sites to
dehydrate and dehydrogenate glycerol.
Thus, for the present catalysts, the dehydration-hydrogenation mechanism (a) explains
the formation of acetol, but does not satisfactorily clarify the negative effect of hydrogen
pressure increase on the conversion. It requires additional explanation about the responsible
sites for the initial protonation, since pyridine adsorption IR spectroscopy revealed no
Brønsted acid sites. Moreover, acetol hydrogenation was tested over Fe
although conversion was high (79%for a 10 wt% aqueous solution with 0.5 g of Fe
bar H ), 1,2-PD selectivity was very poor (25%) with formation of a final mixture comprising
2
O
3
at 493 K, and
2
O
3
at 20
2
two phases. This means that acetol hydrogenation is not the main reaction path towards 1,2-
PD, as also pointed out in [25].
By means of GC-MS analysis of the APR reaction products, traces of epoxides like
glycidol and oxirane-methoxymethyl were found among other compounds. Moreover, when
exposed to ambient air for few hours, the final solution changes from colorless to a yellow
reddish mixture, indicating that albeit the small concentration, unstable products are present.
It has been reported that glycerol is converted into epoxides with considerable selectivity over
Ni/γ-Al
2
O
3
[75]. Glycidol should be expected to be unstable in the reaction medium with
water excess and easily hydrogenated to 1,2-PD. Part of the dehydrogenation processes in the
1
9

initial steps may also occur, in a low extent, via epoxide formation. At this point it should be
noted that none of the aforementioned mechanisms addresses the formation of epoxides.
As observed in the experiments with varying H
2
pressure over 2.5Pt/Fe
3
4
O , increasing
the pressure does not accelerate the reaction rate, probably due to site blockage. This means
the site availability has a pronounced effect in the reaction rate. From another perspective,
many papers on Langmuir-Hinshelwood Hougen-Watson mechanism show that a high
adsorption constant for H could explain the drop in conversion caused by hydrogen pressure
2
increase [59,76-79]. This can also be connected to the hydrogen spillover. The low selectivity
towards 1,3-PD in comparison to 1,2-PD can be due to the nature of the active acid sites in the
present catalysts. For instance, the reduction of WO
Brønsted acids, and these materials are used as supports of catalysts for the production of 1,3-
PD from glycerol [80-82]. On the other hand, the reduction of Fe to Fe increases the
3
produces H
x
WO
3
, which are typical
2
O
3
O
3 4
number of basic sites (Figure 9), as well as the number of Lewis acid sites (Figure 10). Thus,
although spillover occurs and the hydrogen transport could likely lead to the formation of
protons on the surface, the catalyst activity is a function of the Lewis acid sites. Therefore,
although the presence of active hydrogen is indispensable to the hydrogenation steps, it
inhibits the activation of the glycerol molecule in the slow steps of dehydrogenation and
dehydration.
It has been recently shown that incorporation of Pt increases the acidity of different
supports, causing an effect in catalyst activity towards glycerol hydrogenolysis [25], which is
corroborated by the pyridine adsorption analysis in the present work. It is shown herein that
basicity is also greatly affected by metal impregnation, and since dehydrogenation seems to
be a crucial step prior to the formation of 1,2-PD, Lewis acid sites and basic sites play a role
in activity. Lewis acid sites on the particles can thus interact with the oxygen atoms in
hydroxyl groups of glycerol while neighbor active metal atoms can further chemisorb the
2
0

hydrogen atom from the hydroxyl and from the terminal carbon acting as bases. The ability of
the impregnated metal to form stable reduced active particles at the reduction conditions is
crucial to activity, and depends on intrinsic characteristics of the element. For the present
catalysts, metal reducibility follows the order: Pt > Pd > Ni, as observed from XPS results,
which is the same order of the catalysts’ activity. However, it should be noted that under more
controlled conditions of reduction, model surfaces of Ni could perform as better catalysts then
the presently portrayed.
A straightforward interpretation of the connection between basicity and catalytic
activity is not clear. It can be said that the catalysts with higher activity are also the ones
possessing the lowest amount of strong basic sites and largest amount of moderate basic sites.
However, in view of the results, it is not possible to state that strong basic sites cause a poor
catalyst activity. Further experiments with extra iron oxide-supported materials should be
undertaken with different proportions of basic sites in order to check for an unequivocal
relationship between basic sites and their effect on APH or APR. From TEM analysis, Pt
presents smaller particles than Pd, which is evidence that smaller particles have a positive
effect on catalyst activity. It is important to note that in our previous work on Pt
x
-Fe
y
/Al
2
O
3
[19] it was found that larger Pt-Fe particles were more active. This apparent contradiction is
probably due to the absence of reduced bimetallic Pt-Fe particles in the present work, since
the reduction temperatures used in the present experiments were lower. Thus, overlayers of
0
Fe over the Pt particles are less likely to occur, favoring smaller particles.
Particle size plays an important role in hydrogenolysis reactions, considered sensitive
to the structure of the catalyst [83]. Albeit the different activity, 1,2-PD selectivity does not
change much from 2.5Pd/Fe
3
O
4
to 2.5Pt/Fe
3
4
O , which means that hydrogenation steps occur
more easily than dehydrogenation on the particles of both metals. Recent studies reveal that
dehydrogenation of hydrocarbons on Pt surfaces is favored by small particles [84,85], and
2
1

therefore, a particle distribution with larger Pd particles seems to explain in part its lower
catalyst activity. The case is not inverse for APR, but it is different, due to the dissimilar
desorption energies of H
2
on the surface of Pd and Pt particles. Given the smaller average
particle diameter of Pt particles, a larger number of active sites is expected to dehydrogenate
the glycerol molecules on 2.5Pt/Fe
the former.
O
3 4
than on 2.5Pd/Fe
3
4
O , hence the higher conversion of
4
. Conclusions
Glycerol hydrogenolysis and APR share some of the same reaction steps, which
necessarily include dehydrogenation as the initial interaction between glycerol and the
metallic nanoparticles. The Fe
activity, according to the impregnated metal, but all materials studied presented some activity.
Moreover, the fact that Fe is active at the low reaction temperature of 503 K, is highly
interesting for APR reactions conducted at higher temperatures aiming at H production,
given the fact that Fe presents virtually no cost-related issues in comparison to other
materials. As shown by XRD analysis of spent Pt containing catalysts, when used as support
for this metal, Fe does not present the drawback of being structurally changed by
hydration, as for example γ-Al , which is transformed into boehmite [57].
3
O
4
-supported Pt, Pd and Ni catalysts vary remarkably in
2
O
3
2
2
O
3
3
O
4
2
O
3
The formation of stable reduced metallic particles seems to be the greatest limiting feature
of the presented catalysts for the reaction. Further studies with higher reduction temperatures,
especially for as-synthesized Ni/Fe
2
O
3
catalysts, may show promise, since support phase
transition from Fe to FeO is expected to occur as well as metallic particle formation. The
3
O
4
in situ pre-reduction setup seems to be well suited for hydrogenolysis and APR batch
reactions, having the limitation of being only suitable for catalysts with low reduction
temperature, such as the case of supported Pt and Pd. After reduction of the catalyst, the
2
2

change in the support composition, from Fe
and Pd active as nanoparticles, has a profound effect in the acid-base characteristics of the
catalyst, as well as in the hydrogen uptake and spillover ability. For the Pt catalyst, the H
spillover is likely the cause of the drawback in conversion for high hydrogen pressures. The
best performing catalyst for hydrogenolysis, 2.5Pt/Fe , remains stable even after the
reaction, and the conversion drop in catalyst recycle is relatively low. As an additional
advantage for batch operation, it should be highlighted that the Fe -supported materials can
2
O
3
to Fe
3
O
4
, in tandem with the presence of Pt
2
3
O
4
3
O
4
be (and actually were) separated from the reaction solution using a magnet, which is highly
desirable for catalyst recovery.
5
. Acknowledgements
The Brazilian authors would like to thank CAPES , CNPq and FAPERJ. The authors
would also like to thank all the colleagues in the analytical department of LIKAT.
2
3

References
[
[
(
[
1] R. Connor, H. Adkins, J. Am. Chem. Soc. 54 (1932) 4678-4690.
2] R.R. Davda, J.W. Shabaker, G.W. Huber, R.D. Cortright, J.A. Dumesic, Appl. Catal. B 43
2003) 13-26.
3] A.V.H. Soares, J.B. Salazar, D.D. Falcone, F.A. Vasconcellos, R.J. Davis, F.B. Passos, J.
Mol. Catal. A Chem. 415 (2016) 27-36.
4] A. Pamphile-Ádrian, P.P. Florez-Rodriguez, F.B. Passos, J. Braz. Chem. Soc. 27 (2016)
58-966.
5] L. Yin, D. Chen, M. Feng, L. Ge, D. Yang, Z. Song, B. Fan, R. Zhang, G. Shao, RSC Adv.
(2015) 328-337.
6] D. Lee, Y.-W. Choi, Y.-S. Na, S.-S. Choi, D.-W. Park, J. Choi, Mat. Res. Bull. 68 (2015)
21-226.
7] W. Cheng, J. Xu, Y. Wang, F. Wu, X. Xu, J. Li, J. Colloid Interface Sci. 445 (2015) 93-
01.
[
9
[
5
[
2
[
1
[8] R. Li, Y. Jia, N. Bu, J. Wu, Q. Zhen, J. Alloys Compd. 643 (2015) 88-93.
[9] D.E. Stobbe, F.R. van Buren, A.W. Stobbe-Kreemers, J.J. Schokker, A.J. van Dillen, J.W.
Geuss, J. Chem. Soc. Faraday Trans. 87 (1991) 1623-1629.
[
(
[
10] D.E. Stobbe, F.R. van Buren, A.J. van Dillen, J.W. Geuss, Stud. Surf. Sci. Catal. 75
1993) 2337-2340.
11] D.R. Milburn, K.V.R. Chary, R.J. Obrien, B.H. Davis, Appl. Catal. A 144 (1996) 133-
1
46.
12] Y. Liu, H. Tuysuz, C.-J. Jia, M. Schwickardi, R. Rinaldi, A.-H. Lu, W. Schmidt, F.
Schüth, Chem. Commun. 46 (2010) 1238-1240.
[
[
[
[
[
[
13] S. Kang, P.K. Shen, Electrochim. Acta 168 (2015) 104-110.
14] X. Xu, T. Liu, P. Xie, Y. Yue, C. Miao, W. Hua, Z. Gao, Catal. Comm. 54 (2014) 77-80.
15] N. An, P. Wu, S. Li, M. Jia, W. Zhang, Appl. Surf. Sci. 285 (2013) 805-809.
16] E.S. Vasiliadou, A.A. Lemonidou, Catalysts 4 (2014) 397-413.
17] G.W. Huber, J.W. Shabaker, S.T. Evans, J.A. Dumesic, Appl. Catal. B 62 (2006) 226-
2
35.
[
[
[
[
[
18] J. Lee, Y.T. Kim, G.W. Huber, Green Chem. 16 (2014) 708-718.
19] A.V.-H. Soares, P. Geronimo, F.B. Passos, Appl. Catal. B, 185 (2016) 77-87.
20] K.S.W. Sing, Pure Appl. Chem. 54 (1982) 2201-2218.
21] G. Fröhlich, W.M.H. Sachtler, J. Chem. Soc. Faraday Trans. 94 (1998) 1339-1346.
22] H.-Y. Liu, W.-A. Chiou, G. Fröhlich, W.M.H. Sachtler, Top. Catal. 10 (2000) 49-57.
2
4

[
(
[
[
[
23] B.J. Kip, F.B.M. Duivenvoorden, D.C. Koningsberger, R. Prins, J. Am. Chem. Soc. 108
1986) 5633-5634.
24] G.C. Bond, Platin. Met. Rev.19 (1975) 126-134.
25] M. Checa, A. Marinas, J.M. Marinas, F.J. Urbano, Appl. Catal. A. 507 (2015) 34-43.
26] J. Ge, Z. Zeng, F. Liao, W. Zheng, X. Hong, S.C.E. Tsang, Green Chem. 15 (2013)
2
064-2069.
27] F. Mauriello, A. Vinci, C. Espro, B. Gumina, M.G. Musolino, R. Pietropaolo, Catal. Sci.
Tech. 5 (2015) 4466-4473.
[
[
[
(
[
[
[
[
(
[
[
28] J. Li, H. Zeng, S. Sun, J.P. Liu, Z.L.Wang, J. Phys. Chem. B 108 (2004) 14005-14008.
29] V.I. Nefedov, Y.V. Salyn, G. Leonhardt, R. Scheibe, J. Elec. Spec. Rel. Phenom. 10
1977) 121-124.
30] P. Mills, J.L. Sullivan, J. Phys. D Appl. Phys. 16 (1983) 723-732.
31] E. Paparazzo, J. Elec. Spec. Rel. Phenom. 43 (1987) 97-112.
32] D. Brion, Appl. Surf. Sci. 5 (1980) 133-152.
33] G.C. Allen, M.T. Curtis, A.J. Hooper, P.M. Tucker, J. Chem. Soc. Dalton Trans. 14
1974) 1525-1530.
34] N.S. McIntyre, D.G. Zetaruk, Anal. Chem. 49 (1977) 1521-1529.
35] G.M. Bancroft, I. Adams, L.L. Coatsworth, C.D. Bennewitz, J.D. Brown, W.D.
Westwood, Anal. Chem. 47 (1975) 586-588.
[
[
[
[
36] J.P. Contour, G. Mouvier, M. Hoogewys, C. Leclere, J. Catal. 48 (1977) 217-228.
37] T.H. Fleisch, G.J. Mains, J. Phys. Chem. 90 (1986) 5317-5320.
38] T.H. Fleisch, G.W. Zajac, J.O. Schreiner, G.J. Mains, Appl. Surf. Sci. 26 (1986) 488-497.
39] E. Janin, M. Björkqvist, T.M. Grehk, M. Götheld, C.M. Pradier, U.O. Karlsson, A.
Rosengren, Appl. Surf. Sci. 99 (1996) 371-378.
40] J.E. Drawdy, G.B. Hoflund, S.D. Gardner, E. Yngvadottir, D.R. Schryer, Surf. Interface
Anal. 16 (1990) 369-374.
41] M. Alnot, V. Gorodetskii, A. Cassuto, J.J. Ehrhardt, Thin Solid Films 151 (1987) 251-
62.
[
[
2
[
(
[
42] C. Battistoni, A.M. Giuliani, E. Paparazzo, F. Tarli, J. Chem. Soc. Dalton Trans. 7
1984) 1293-1299.
43] C. Sleigh, A.P. Pijpers, A. Jaspers, B. Coussens, R.J. Meier, J. Elec. Spec. Rel. Phenom.
7
7 (1996) 41-57.
44] A. Tressaud, S. Khairoun, H. Touhara, N. Watanabe, Z. Anorg. Allg. Chem. 540 (1986)
91-299.
[
2
2
5

[45] J.M. Tura, P. Regull, L. Victori, M.D. De Castellar, Surf. Interface Anal. 11 (1988) 447-
4
49.
[
[
[
46] K.S. Kim, N. Winogard, Surf. Sci. 43 (1974) 625-643.
47] R.B. Shalvoy, P.J. Reucroft, B.H. Davis, J. Catal. 56 (1979) 336-348.
48] T. Dickinson, A.F. Povey, P.M.A. Sherwood, J. Chem. Soc. Faraday Trans. 1 73 (1977)
3
27-343.
[
49] S.O. Grim, L.J. Matienzo, W.E. Swartz, J. Am. Chem. Soc. 94 (1972) 5116-5117.
50] A.B. Mandale, S. Badrinarayanan, S.K. Date, A.P.B. Sinha, J. Elec. Spec. Rel. Phenom.
[
3
3 (1984) 61-72.
[
[
[
[
[
[
51] H. Hattori, J. Jpn. Pet. Inst. 47 (2004) 2.
52] E.P. Parry, J. Catal. 2 (1963) 371-379.
53] G. Busca, V. Lorenzelli, Mat. Chem. 6 (1981) 175-185.
54] V. Lorenzelli, G. Busca, Mat. Chem. Phys. 13 (1985) 261-281.
55] M.M. Mohamed, B.M. Abu-zied, Thermochim. Acta 359 (2000) 109-117.
56] S.R. Fogler, Elementos de Engenharia das Reações Químicas. 3rd Edition, LTC, Rio de
Janeiro, 2002, p. 697.
57] R.M. Ravenelle, F.Z. Diallo, J.C. Crittenden, C. Sievers, ChemCatChem 4 (2012) 492-
94.
58] D.G. Kovacs, J.E. Jackson, D.J. Miller, Abst. Papers Am. Chem. Soc., San Diego, CA, 1-
april, 2001, v. 221, p. U177-U178, Part 2. Abstract 457-ORGN.
[
4
[
6
[59] X. Jin, B. Subramaniam, R.V. Chaudhari, P.S. Thapa, AIChE J. 62 (2016) 1162-1173.
[60] M.A. Dasari, P.-P. Kiatsimkul, W.R. Sutterlin, G.J. Suppes, Appl. Catal. A 281 (2005)
2
25-231.
61] Physical Properties of Glycerine and its solutions, Hydrocarbon Process. 1967.
http://www.aciscience.org/docs/physical_properties_of_glycerine_and_its_solutions.pdf>.
[
<
[
[
[
62] T.E. Crozier, S. Yamamoto, J. Chem. Eng. Data 19 (1974) 242-244.
63] H.A. Pray, C.E. Schweickert, B.H. Minnich, Ind. Eng. Chem. 44 (1952) 1146-1151.
64] A. Martin, U. Armbruster, I. Gandarias, P.L. Arias, Eur. J. Lipid Sci. Tech. 115 (2013) 9-
2
7.
65] C. Montassier, J.C. Ménézo, L.C. Hoang, C. Renaud, J. Barbier, J. Mol. Catal. 70 (1991)
9-110.
[
9
[
[
[
66] A. Alhanash, E.F. Kozhenikova, I.V. Kozhenikov, Catal. Lett. 120 (2008) 307-311.
67] E.P. Maris, R.J. Davis, J. Catal. 249 (2007) 328-337.
68] E.P. Maris, W.C. Ketchie, M. Murayama, R.J. Davis, J. Catal. 251 (2007) 281-294.
2
6

[
69] Y. Nakagawa, Y. Shinmi, S. Koso, K. Tomoshige, J. Catal. 272 (2010) 191-194.
70] Y. Shinmi, S. Koso, T. Kubota, Y. Nakagawa, K. Tomishige, Appl. Catal. B 94 (2010)
[
3
18-326.
[
[
[
71] H. Atia, U. Armbruster, A. Martin, Appl. Catal. A 393 (2011) 331-339.
72] A. Alhanash, E.F. Kozhevnikova, I.V. Kozhevnikov, Appl. Catal. A 378 (2010) 11-18.
73] J.B. Salazar, D.D. Falcone, H.N. Pham, A.K. Datye, F.B. Passos, R.J. Davis, Appl. Catal.
A 482 (2014) 137-144.
[
74] X. Lin, Y. Lv, Y. Xi, Y. Qu, D.L. Phillips, C. Liu, Energ. Fuels 28 (2014) 3345-3351.
75] P.P. Florez-Rodriguez, A.J. Pamphile-Adrián, F.B. Passos, Catal. Today 237 (2014) 38-
[
4
6.
76] A. Torres, D. Roy, B. Subramaniam, R.V. Chaudhari, Ind. Eng. Chem. Res. 49 (2010)
0826-10835.
77] Y. Xi, J.E. Halladay, J.G. Frye, A.A. Oberg, J.E. Jackson, D.J. Miller, Org. Process Res.
Dev. 14 (2010) 1304-1312.
78] Z. Zhou, X. Li, T. Zeng, W. Hong, Z. Cheng, W. Yuan, Chin. J. Chem. Eng. 18 (2010)
84-390.
[
1
[
[
3
[79] R.V. Sharma, P. Kumar, A.K. Dalai, Appl. Catal. A 477 (2014) 147-156.
[80] S. Zhu, X. Gao, Y. Zhu, Y. Li, J. Mol. Catal. A 398 (2015) 391-398.
[81] S. García-Fernández, I. Gandarias, J. Requies, M.B. Güemez, S. Bennici, A. Auroux,
P.L. Arias, J. Catal. 323 (2015) 65-75.
82] J. Wang, X. Zhao, N. Lei, L. Li, L. Zhang, S. Xu, S. Miao, X. Pan, A. Wang, T. Zhang,
ChemSusChem 9 (2016) 784-790.
[
[83] J.H. Sinfelt, J.L. Carter, D.J.C. Yates, J. Catal. 24 (1972) 283-296.
[84] Y. Chu, Q. Zhang, T. Wu, Z. Nawaz, Y. Wang, F. Wei, J. Nanosci. Nanotech. 14 (2014)
6
900-6906.
85] J. Zhu, M.-L. Yang, Y. Yu, Y.-A. Zhu, Z.-J. Sui, X.-G. Zhou, A. Holmen, D. Chen, ACS
Catal. 5 (2015) 6310-6319.
[
2
7

Figure 1: Scheme for the hydrogen chemisorption analysis to measure Pt dispersion on Fe
consideration of the spillover effect.
3 4
O particles with
2
8

2
.5Pt/Fe O
2 3
1
.0Pt/Fe O
2 3
2
.5Pd/Fe O
2 3
2
.5Ni/Fe O
2
3
Fe O
2
3
400
500
600
700
800
900
Temperature (K)
(support), 2.5Ni/Fe
Figure 2: TPR profiles for Fe
2
O
3
2
O
3
2
, 2.5Pd/Fe O
3 2 3 2 3
, 1.0Pt/Fe O and 2.5Pt/Fe O .
2
9