Low temperature oxidation of methanol to methyl formate over Pd nanoparticles supported on γ-Fe2O3

Article abstract of DOI:10.1039/c3cy00859b

Pd nanoparticles supported on γ-Fe₂O₃ (2 wt.%) were synthesized using the water-in-oil microemulsion method (using hydrazine as a reductant agent). Materials were characterized by N₂-BET at low temperature, XRD, XPS, Raman, and FTIR and tested in the gas phase reaction of oxidation of methanol. The direct formation of methyl formate (MF) from methanol was observed. Supported palladium catalysts produced methyl formate at low temperature (<150 °C) with a relatively high selectivity depending on the Fe⁺²/Fe⁺³ ratio (2:1, 1:1, 1:2) used for the preparation of the supports. Methyl formate is already formed at 50 °C with the maximum at about 80 °C. At higher temperature, methyl formate is no longer formed and the oxidation to CO₂ and CO occurs. Raman studies indicated the changes in the structure of the Fe₂O₃ support in the case of the 1:2 sample after chemical reduction with hydrazine.

Full text of DOI:10.1039/c3cy00859b

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Low temperature oxidation of methanol to
methyl formate over Pd nanoparticles
supported on γ-Fe2O3
Cite this: Catal. Sci. Technol., 2014,
4, 738
*
R. Wojcieszak, M. N. Ghazzal, E. M. Gaigneaux and P. Ruiz
Pd nanoparticles supported on γ-Fe2O3 (2 wt.%) were synthesized using the water-in-oil microemulsion
method (using hydrazine as a reductant agent). Materials were characterized by N2-BET at low temperature,
XRD, XPS, Raman, and FTIR and tested in the gas phase reaction of oxidation of methanol. The direct forma-
tion of methyl formate (MF) from methanol was observed. Supported palladium catalysts produced
methyl formate at low temperature (<150 °C) with a relatively high selectivity depending on the Fe+2/Fe+3
ratio (2: 1, 1 : 1, 1 : 2) used for the preparation of the supports. Methyl formate is already formed at 50 °C
with the maximum at about 80 °C. At higher temperature, methyl formate is no longer formed and the
oxidation to CO2 and CO occurs. Raman studies indicated the changes in the structure of the Fe2O3
support in the case of the 1 : 2 sample after chemical reduction with hydrazine.
Received 31st October 2013,
Accepted 7th December 2013
DOI: 10.1039/c3cy00859b
www.rsc.org/catalysis
in the oxidation of methanol to methyl formate, but only at
low temperature. At higher temperature, methyl formate is
1. Introduction
The catalytic properties of supported Pd nanoparticles in the
methanol oxidation reaction have been studied in the past
years.1–3 It was shown that this reaction may lead to formalde-
hyde (HCHO, FA), dimethoxymethane (CH3OCH2OCH3, DMM),
methyl formate (HCOOCH3, MF) or hydrogen. Oxidative
routes to HCHO are predicted on silver based and iron
molybdate catalysts.4 DMM can be produced in a two-step
process involving methanol oxidation to HCHO followed by
acetalization of HCHO–CH3OH mixtures with liquid or solid
acids. Hydrogen can be produced using Cu/ZnO5 catalysts and
methyl formate is produced via CH3OH dehydrogenation on
CuO or Pd supported nanoparticles.1,6
However, to our best knowledge, the use of palladium
nanoparticles supported on Fe2O3 for low temperature oxida-
tion of methanol to methyl formate has not been reported yet
in the literature. The Fe2O3 support is rather used for the
hydrogen or formaldehyde production from methanol than
for the direct selective methyl formate formation.
no longer formed. In this case, the formation of CO2 and
CO is promoted. Catalysts were characterized using several
physico-chemical properties (chemical analysis, N2-BET, Raman,
XRD, XPS, and FTIR study). The results shown here could
provide insight into the mechanism of the reaction at low
temperatures (50–80 °C).
2. Experimental
2.1 Catalyst preparation
Palladium(II) chloride (from Aldrich) was used as the Pd pre-
cursor. γ-Fe2O3 was obtained by co-precipitation of ferric and
ferrous salts in adequate ratio. Demineralized water was used
for the nanoparticle preparation and degassed with argon gas
for half an hour, in order to avoid the oxidation of the ionic
salts and the final products. Two solutions were prepared,
1 M FeSO4 (FeSO4·7H2O; >99.0%, Sigma-Aldrich) and 1 M
FeCl3 (FeCl3·6H2O; ≥98%, Sigma-Aldrich), by dissolving the
iron salts in 2 M HCl solution. The pH of the ferric and
ferrous salts was maintained under 0.5. The highly acidic
medium prevented cation precipitation and enables a control
of the expected crystalline phase of the iron oxide. In a
typical experimental procedure, variable ratios of Fe2+ : Fe3+
(Table 1) were obtained by adjusting the mixed volume of a
1 M FeSO4 and 1 M FeCl3 solution under an argon atmosphere.
The pH of the system was controlled during the synthesis
process and adjusted to more than 11.5 by adding dropwise
sodium hydroxide solution (8 M). The reaction was allowed
to proceed for 2 h at 70 °C under constant stirring. Finally,
In this work, palladium nanoparticles were prepared using
the water-in-oil microemulsion method.7,8 An important
aspect is the reaction temperature. In order to preserve the
nanosize character of the particles, a low reaction tempera-
ture of methanol oxidation to methyl formate in the presence
of oxygen was studied. It is demonstrated that supported
nanoparticles of palladium are active and selective catalysts
MOST, Institute of Condensed Matter and Nanosciences (IMCN), Université
catholique de Louvain, Croix du Sud 2, L7.05.15, 1348 Louvain la Neuve, Belgium.
E-mail: robert.wojcieszak@univ-lille1.fr
738 | Catal. Sci. Technol., 2014, 4, 738–745
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Table 1 Variable molar ratio used for nanoparticle synthesis
the phases was achieved by using the ICDD-JCPDS database.
Size of particles was estimated from the X-ray diffraction line
broadening analysis using a Scherrer formula. This analysis
has been widely used for characterizing supported metal crys-
tallites. A perfect crystal would extend in all directions to
infinity so no crystal is perfect due to its limited size. Such a
deviation from perfect crystallinity leads to broadening of the
diffraction peak. The Scherrer equation relating the crystallite
size (d) to the peak width is described by d = Kλ/β cosΘ, where
K is the Scherrer constant, λ is the wavelength of the
X-radiation and β is the integral breadth of a reflection
(in radians) located at 2Θ.9
Vol Fe3+
1 M (ml)
Vol Fe2+
1 M (ml)
Fe3+ : Fe2+
Moles Fe3+
Moles Fe2+
1 : 2
2 : 1
1 : 1
20
40
30
0.02
0.04
0.03
40
20
30
0.04
0.02
0.03
the precipitates were cooled to room temperature. During the
reaction, the solution color could be seen changing from
orange to black, leading to a black precipitate. The solution
containing the precipitate was cooled down at room tempera-
ture and the mixture was then dropped into a flask. During
the experiment, argon was continuously passed through the
solution to prevent the oxidation of the Fe2+ in the system.
The solution was finally lyophilized under freezing condi-
tions. The freezing conditions were selected to keep the
iron ion oxidation to a minimum.
The organic surfactant (sodium bis(2-ethylhexyl)
sulfosuccinate (AOT)) used in this work was purchased from
Sigma. The cyclohexane used as the organic solvent was
purchased from Fluka (purity 99.5%). The hydrazine hydrate
solution (80%) was purchased from Fluka (purity 99%). All
reagents were used without further purification.
Catalysts were prepared as follows. i) precursor solution: an
appropriate quantity of PdCl2 was dissolved in 5 ml of
distilled water in the presence of NaCl (Merck, 99.5%). Then
the solution was evaporated at 60 °C until only a volume of
1 ml remained; ii) microemulsion: it was formed using 50 ml
of cyclohexane, 1 ml of Pd precursor solution and the appro-
priate quantity of AOT surfactant. The microemulsion was
then heated to 50 °C and then 2 g of γ-Fe2O3 were incorpo-
rated to the reactant flask under magnetic stirring. After
30 minutes, the appropriate quantity of hydrazine was injected.
The reduction was carried out in a thermostated water bath at
50 °C. The time of reduction of 30 min was measured from
the hydrazine injection to the microemulsion. The reaction
was carried out under nitrogen atmosphere. The reactant
solution changed colour from light red to black, indicating
the reduction of palladium. The solution was then filtered,
washed with acetone (VWR, 99%) and distilled water for
60 min and dried at 100 °C for 60 minutes.
2.2.3. N2 physisorption. The textural properties of the
support and reduced materials were studied by means of N2
adsorption/desorption at −196 °C with
a Micromeritics
ASAP2000 apparatus. Low temperature nitrogen adsorption
isotherms enable the BET calculation of the specific surface
area and the BJH calculation of the pore volume and the pore
diameter.
2.2.4. X-ray photoelectron spectroscopy. XPS analyses were
performed on an SSI X-probe (SSX-100/206) spectrometer
from Surface Science Instruments. The analysis chamber
was operated under ultrahigh vacuum with a pressure close
to 5 × 10−9 Torr and the sample was irradiated with mono-
chromatic Al Kα (1486.6 eV) radiation (10 kV; 22 mA). Charge
compensation was achieved by using an electron flood gun
adjusted at 8 eV and placing a nickel grid 3.0 mm above the
sample. Pass energy was set at 50 eV for recording high reso-
lution peaks. In these conditions, the full width at half maxi-
mum (FWHM) of the Au 4f7/2 peak of a clean gold standard
sample was about 1.1 eV. For these measurements, the bind-
ing energy (BE) values were referred to the C–(C,H) contribu-
tion of the C 1s peak at 284.8 eV. The surface atomic
concentrations were calculated by correcting the intensities
with theoretical sensitivity factors based on Scofield cross-
sections.10 Peak decomposition was performed using curves
with an 85% Gaussian type and a 15% Lorentzian type, and a
Shirley non-linear sigmoid-type baseline. The following peaks
were used for the quantitative analysis: O 1s, C 1s, N 1s, Fe
3d and Pd 3d. Based on the XPS analysis, the XPS surface
ratio of a given element is defined as the atomic concentra-
tion of the element (%) divided by the atomic concentration
of Fe (%).
2.2.5. Raman spectroscopy. Raman spectra were measured
using a Thermo Scientific microscope using high resolution
radiation at 633 nm from a 35 mW He–Ne laser. The laser
beam was focused on the sample by a ×50 lens to give a spot
size of 50 μm in slit mode. The laser power was always
kept at 0.9 mW at the sample to avoid sample degradation
or phase transformation. After each spectrum had been
recorded, a careful visual inspection was performed using
white light illumination on the microscope stage in order to
detect any change that could have been caused by the laser.
2.2.6. FTIR study. Fourier transform infrared spectroscopy
(FTIR) spectra were recorded in a Brucker EQUINOX 55
spectrometer. The spectra (for which 200 scans were averaged
Comparing with the catalysts, the bare γ-Fe2O3 were also
treated with hydrazine using the same conditions as the catalysts.
2.2. Characterization of physical and chemical properties
2.2.1. ICP study. ICP-AES (Inductively Coupled Plasma-
Atomic Emission Spectroscopy) analysis was used to deter-
mine the chemical composition of the prepared catalysts and
it was performed on a Thermo Jarrel ASH IRIS Advantage
analyzer.
2.2.2. X-ray diffraction. X-ray diffraction study was carried
out on
a Siemens D5000 diffractometer using the Kα
radiation of Cu (1.5418 Å). The 2Θ range was scanned
between 2 and 65° at a rate of 0.01° s−1. The identification of
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 738–745 | 739

View Article Online
Paper
Catalysis Science & Technology
with a resolution of 4 cm−1) were recorded under static air.
FTIR spectra were obtained in the transmittance mode.
through the reactor with a total flow rate of 100 cm3 min−1.
The standard catalytic test was carried out at various tempera-
tures (50, 80, 100, 150 and 200 °C). The sample was heated or
cooled at a rate of 10 °C min−1. Methanol, oxygen and
the reaction products were analyzed every 30 min using a
CP3800 Varian gas chromatograph, operated at a programmed
2.2.7. H2-chemisorption.
A
hydrogen chemisorption
method at 35 °C was used for the determination of dispersion
and the average size of palladium particles supported on
γ-Fe2O3. The amounts of chemisorbed hydrogen were deter-
mined at 35 °C on a Micrometrics device. The following
procedure was applied in the measurements: preliminary
evacuation and heating at 300 °C for 1 h in a He flow, then
a flow of hydrogen (Praxair, 99.98%) at 300 °C for 1 h, evacua-
tion at 300 °C for 2 h, cooling down to 35 °C, evacuation
under He at 35 °C for 1 h, leak test, and finally measure-
ment of hydrogen uptake at 35 °C. A double adsorption
method was used: (i) first the adsorption isotherm was mea-
sured, which includes both physisorption and chemisorp-
tion; (ii) after a 2 h outgassing under He atmosphere at
35 °C, a second isotherm was measured which includes
physisorption only. Both isotherms are determined in the
pressure range from 0.5 to 700 mmHg. The difference
between the first and second isotherms gave the H2 chemi-
sorption isotherm. Assuming a spherical shape, an average
crystallite size d (given in nanometers) was calculated based
on the irreversible adsorption isotherm of hydrogen,
according to the following: d6000/SF, where S is the surface
temperature and with
a TCD detector. CP-Poraplot Q
(25 m × 0.53 mm) and CP-Molsieve 5A (25 m × 0.53 mm) col-
umns were used for separation of the reaction products.
The following molecules were analysed during the reaction:
H2, O2, CH4, CO, CO2, methanol, methyl formate and H2O.
3. Results
3.1. ICP analysis and N2 physisorption
The results of ICP analysis are given in Table 2. The mass
content of Pd is between 1.8 and 2% which was smaller than
the nominal expected (2 wt.%). However, small differences
are probable due to the method of preparation.
The specific surface area changed significantly after chemi-
cal reduction as shown by the BET results in Table 2. BET sur-
face areas of bare γ-Fe2O3 were about 11, 49, and 103 m2 g−1
for 1 : 2, 1 : 1, and 2 : 1 samples, respectively. Interestingly, a
sharp increase in the surface of about 8 times was observed for
the 1 : 2 sample (from 11 to 88 m2 g−1). It could be concluded
that reduction of the γ-Fe2O3 by hydrazine could change
the structure of the oxide or/and liberate the oxide pores
from NaCl contained in the materials after the preparation
step (see Experimental part) also observed in XRD.
area of the fraction of reduced palladium (given in m2 g−1
)
and F is the palladium density (given in cm3 g−1). S was
calculated assuming H/Pd 1 and a surface area of 7.93 Å2
per palladium atom.9
2.3. Catalytic test
3.2. XRD patterns
The reaction of partial oxidation of methanol (POMeOH) was
performed in a metallic fixed-bed microreactor made of an
inconel tube of 1 cm internal diameter (PID ENG&Tech,
Madrid, Spain), operating at atmospheric pressure. The cata-
lytic bed was composed of 100 mg of catalyst powder selected
within the granular fraction of 200–315 μm and diluted in
600 mg glass spheres, which was confirmed to be inactive. The
catalyst was placed in the reactor. Before the catalytic test,
catalysts were reduced in a flow of pure hydrogen (Praxair,
flow rate = 50 cm3 min−1) at 300 °C for 1 h and heated with a
ramp of 10 °C min−1, then the sample was cooled to 300 °C.
After the reduction, a gas mixture containing 5 vol.% of
methanol and 2.5 vol.% of O2 was prepared by passage of a
He stream through methanol (Merck, >99%) placed in a satu-
rator maintained at 5.4 °C. The reactant mixture was passed
There are many different possible iron oxide structures such
as magnetite (Fe3O4), hematite (α-Fe2O3), maghemite (γ-Fe2O3),
or ferric hydroxide (β-FeOOH). The d spacings calculated
from the patterns are close to those of magnetite and
maghemite, both of which are inverse spinel structures.
XRD analysis shows the presence of the γ-Fe2O3 structure for
all bare samples. XRD study of the catalyst indicated that
palladium is in a metallic state after hydrazine reduction
(Fig. 1b). However, the very low resolution of these reflec-
tions did not allow determining the average particle size
from the well-known line broadening analysis as proposed
by Scherrer. The XRD patterns of the supports confirmed
the presence of the NaCl in the materials (Fig. 1a). However,
the presence of NaCl after palladium incorporation and
hydrazine reduction is not observed. It could be concluded
Table 2 ICP, BET and H2 chemisorption results
Catalyst
Pd [wt.%]
BET surface area [m2 g−1
]
BET after hydrazine reduction [m2 g−1
]
Pd particle size [nm]
Fe2O3 2 : 1
Fe2O3 1 : 1
Fe2O3 1 : 2
Pd/Fe2O3 2 : 1
Pd/Fe2O3 1 : 1
Pd/Fe2O3 1 : 2
103
49
11
—
—
—
84
68
88
92
64
78
2.0
1.7
1.8
4.6
3.9
3.2
740 | Catal. Sci. Technol., 2014, 4, 738–745
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
801, 636, and 561 cm−1. The peaks centered at 883 and
801 cm−1 are consistent with the vibration due to the Fe–O in
the goethite (FeOOH) structure. The peaks centered at 636
and 561 cm−1 are characteristic of the Fe–O bond vibration in
the spinel γ-Fe2O3.
The spectra are identical regardless of the molar ratio of
Fe2+ : Fe3+ used to synthesize the iron oxide. This is in good
agreement with the previous work.11 Even after metal deposi-
tion and hydrazine treatment, the bands that originated from
γ-Fe2O3 are clearly observed. However, other peaks are not visi-
ble in the spectrum because of the overlap with the presence of
the goethite structure that could be observed at 801 and
822 cm−1. These peaks could be assigned to the N–H bond
vibration.12 Furthermore, no peaks relative to the C–H bonds
were observed in the range of 2700–2900 cm−1, indicating
that the extraction of the AOT using the washing procedure
was successfully done.
Fig. 1 XRD patterns of the supports (A) and Pd modified samples (B).
3.4.2. Raman results. Fig. 3 depicts the Raman spectra
of a freshly synthesized γ-Fe2O3 for variable molar ratios of
Fe2+ : Fe3+. In the case of 1 : 1 and 2 : 1 samples, only the
bands at 309, 531 and 667 cm−1 are observed despite the very
reasonable signal-to-noise ratio.
that the preparation method (microemulsion, washing,
drying) could totally remove NaCl from the materials.
3.3. H2-chemisorption
The observed asymmetric shape of the broad bands
suggests an overlapping of other frequencies. These could also
be induced by non-stoichiometry, the presence of vacancies,
interstitial cations, and defects, which in general may result
in an activation of new phonon modes not predicted by
group theory. However, according to the literature, even the
positions of the bands of the maghemite are most of the
time difficult to predict. Most authors take the three broad
maxima at around 360–380, 500 and 660–720 cm−1 as indica-
tive of maghemite,13 which is in accordance with our results.
In the case of the 1 : 2 sample, changes in the Raman
profile after hydrazine reduction was observed. As shown in
Fig. 3C, new peaks are observed: 227, 290, 406, 493, 609,
817 and one broad peak at 1310 cm−1. These peaks corre-
spond to the hematite (α-Fe2O3) structure.14
Low temperature H2 chemisorption permits us to determine
the dispersion of the active phase on the surface catalysts.
The results are given in Table 2. As expected, all samples
present palladium particles in the nanosize range (from 3.2
to 4.6 nm). The average particle size seems dependent on the
Fe2+/Fe3+ ratio. The smallest particles were observed for the
1 : 2 sample. This catalyst also presents a higher surface area.
The small sizes of palladium particles were expected due to
the method of preparation. Indeed, the microemulsion
method allows obtaining small metal particles depending on
the chemical composition of the primary microemulsion.
3.4. Structural and morphological characteristics of
palladium catalysts
3.4.1. FTIR results. Fig. 2 shows the IR spectra of oxide
nanoparticles in the 1000–400 cm−1 region. The IR spectra in
Fig. 2 are related to maghemite. The IR spectra of the
as-synthesized material show broad maxima at around 883,
3.5. Surface properties
XPS study is able to determine the chemical composition and
oxidation degrees of components present on the catalyst
Fig. 3 Raman spectra of the Fe2O3 2–1 (A), Fe2O3 1–1 (B) and Fe2O3 1–2
Fig. 2 FTIR spectra of the Fe2O3-1 : 2 and Pd/Fe2O3-1 : 2 catalysts.
(C) oxides and corresponding Pd catalysts.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 738–745 | 741

View Article Online
Paper
Catalysis Science & Technology
Table 3 Molar surface concentration and XPS ratios
surface. A very low contamination with carbon was observed
in all samples. The results confirmed that organic surfactant
is removed from the catalyst surface after filtration and wash-
ing steps. Moreover, the presence of S in the chemical com-
position of the AOT surfactant permits us to monitor the
presence of AOT on the catalyst surface. The results also con-
firm the presence of the palladium in metallic state (Pd0).
However, some extent of oxidized palladium (Pd+) was also
detected (Fig. 4). XPS Pd0/Pd+ atomic ratios are indicated in
Table 3.
Catalyst
Pd [%] Fe [%] C [%] Pd+/Pd0 Pd/Fe Fe/O
Fe2O3 2–1
Fe2O3 1–1
Fe2O3 1–2
Pd/Fe2O3 2–1 1.98
Pd/Fe2O3 1–1 1.52
Pd/Fe2O3 1–2 2.41
—
—
—
28.5
25.6
23.9
23.7
24.8
19.3
15.1
18.9
23.1
20.9
19.3
23.1
—
—
—
0.52
0.54
0.45
—
—
—
0.08
0.06
0.12
0.50
0.46
0.45
0.44
0.45
0.38
3.6. Catalytic activity
During the catalytic test, only methyl formate, CO2, CO and
H2O products were detected.
The γ-Fe2O3 alone, previously treated in aqueous hydrazine
media or left untreated, was inactive in the oxidation of
methanol at low temperature (below 150 °C).
Some activity was however observed at high temperature
(above 250 °C). The only product was CO2, indicating the
unselective complete oxidation of methanol.
The Pd modified catalyst showed high activity in the oxi-
dation of methanol (Fig. 5 and Table 4). Pd/γ-Fe2O3 1–2
catalyst is very active at low temperature (50–80 °C). At about
100–150 °C, conversion of methanol is practically stable at
about 75%. Under these conditions (50–80 °C), the direct for-
mation of methyl formate was observed (Fig. 5). In addition,
it is produced with a relatively high selectivity of about 84%
(Table 4). CO2 and CO as secondary products were observed.
At high temperature (>150 °C), the conversion of methanol
is high, but methyl formate does not form at all. Under these
conditions, only CO2 and CO are formed. Other catalysts are
also active but their activity is less than that of the 1 : 2 sample.
The differences in the catalytic activity could be derived
from the differences in the average Pd particle size. In addi-
tion, despite the highest Pd dispersion, the 1 : 2 sample also
Fig. 5 Conversion of methanol and product yields for Pd/Fe2O3 1–1
(A), Pd/Fe2O3 1–2 (B) and Pd/Fe2O3 2–1 (C) catalysts.
Fig. 4 XPS Pd 3d region for the Pd/Fe2O3 1–2 catalyst.
742 | Catal. Sci. Technol., 2014, 4, 738–745
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Table 4 Catalytic results for Pd/Fe2O3 catalysts
MeOH conversion [%]
MF yield [%]
MF selectivity [%]
CO/CO2 selectivity [%]
Catalyst
50 °C
80 °C
50 °C
80 °C
50 °C
80 °C
50 °C
80 °C
Pd/Fe2O3 2 : 1
Pd/Fe2O3 1 : 1
Pd/Fe2O3 1 : 2
Fe2O3 2–1
Fe2O3 1–1
Fe2O3 1–2
10
21
41
0
0
0
21
61
76
0
0
0
3
5
35
—
—
—
9
27
22
88
—
—
—
41
56
81
—
—
—
73/0
75/3
8/4
—
—
—
59/0
40/4
16/3
—
—
—
35
62
—
—
—
when the pure support is treated in the presence of N2H4.
Moreover, changes in XRD spectra of fresh and hydrazine
treated supports were observed. It could be concluded that
hydrazine changed the chemical composition of the materials
and totally removed NaCl from the catalysts.
Contrary to the classical high temperature methods of
preparation (calcination, H2 reduction) the chemical reduc-
tion with hydrazine at low temperature in the presence of
protective agents is the ideal alternative. Indeed, as already
shown in the literature, the water-in-oil microemulsion
method permits very small metal particles to be obtained
with a very good dispersion of the active phase.18–20 In this
method, the metal precursor is reduced in solution in the
presence of protective agents such as polymers or surfactants.
In some cases, the interaction between metal particles
formed and surfactant could be very strong and the surfac-
tant molecules can stay adsorbed on the metal surface after
calcination of the sample or a cleaning step. However, in our
case, the XPS data confirmed that the AOT surfactant is
totally removed from the sample after the water and acetone
washing step (see Experimental section). The reduction of
palladium in microemulsion is very fast. The number of
nuclei formed at the very beginning of the reduction deter-
mined the number and size of the resultant particles.21 The
reduced species that formed later were involved in collisions
with the nuclei that were already formed instead of the
new nuclei formation. This allows the creation of larger parti-
cles. In our case, the hydrazine to palladium molar ratio was
100, which placed this preparation in the desired conditions
(particle size not dependent on hydrazine concentration).
These conditions ensured that the particle size would be
mainly governed by the nature of the solvent and surfactant.
After reduction with hydrazine, XPS study confirmed the
fact that in all catalysts, palladium is not fully reduced. The
presence of the oxidized palladium on the surface could be
explained by the diffusion of oxygen from the bulk of the
palladium particles towards its surface. In fact, the surface of
palladium is always formed by an oxidized layer, even after
reduction with hydrogen. Pd/γ-Fe2O3 catalyst is always in an
oxide PdO form with its surface in an intermediate state
between surface lacunary PdO and crystalline PdO species.
This phenomenon has been demonstrated experimentally
using Raman spectroscopy.22 It also cannot be excluded that
part of the presence of the oxidized palladium is also due
to the air exposure in the washing and drying steps. The
Fig.
catalysts.
6 Methanol conversion versus specific surface area of the
showed a higher specific surface area. As shown in Fig. 6, the
catalytic activity depends on the specific surface area.
4. Discussion
4.1. Physical and chemical properties of the supported
catalysts
XRD is not sensitive enough to differentiate between Fe3O4
and γ-Fe2O3, due to extreme similarity in crystal structure of
the two phases, especially if there is an intergradation, in
which Fe3O4 and γ-Fe2O3 coexist at a certain proportion.15
The difference between these two phases could be observed
in the FTIR study. Fe3O4 presents one absorption peak of
Fe–O at 561 cm−1.16,17 Contrary to that, in the case of γ-Fe2O3,
two peaks at 636 cm−1 and 561 cm−1 associated with the Fe–O
could be observed. In our case, the presence of these two
peaks observed in all samples confirmed the γ-Fe2O3 struc-
ture. Regardless of the ratio of Fe3+ : Fe2+ used for the starting
solution, the iron oxide crystallizes in the maghemite phase
γ-Fe2O3. The results suggest that the oxidation of the Fe2+
occurs for each ratio despite the precaution taken during
the preparation procedure. The modification of γ-Fe2O3
via Pd precipitation–reduction changes the adsorption iso-
therm and significantly influenced the BET surface area. The
changes in the BET specific surface area were also observed
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 738–745 | 743

View Article Online
Paper
Catalysis Science & Technology
XPS ratios of the Pd0 to Pd+ (Table 2) provide information
about the reduction degree of the palladium before and after
the catalytic test. The increase of this ratio observed after the cata-
lytic test could be explained by the reducing role of methanol.
crystallites to give CO multicoordinated to the metal surface.
Formyl species have also been found at this temperature, but
they promptly disappear when the catalyst is heated and the
formation of CO and H2 prevails.31,32
Methyl formate can form via the condensation of
adsorbed methoxides with HCHO to form methoxymethanol
intermediates (CH3OCH2OH) that then dehydrogenate to
methyl formate.1 The esterification of formic acid (HCOOH)
intermediates formed by HCHO oxidation is also possible
but the high selectivities to methyl formate reported here indi-
cate fast reactions via condensation of HCHO to methyl formate.
The first reaction step was found to be the formation of
methoxy groups by dissociative adsorption of methanol on a
dual acid–base site formed by an accessible cation and a
surface oxygen ion.27 The methoxy species are further trans-
formed to different intermediates depending on the acid
strength of the site on which it is adsorbed and on the nature
of the active centers in close proximity. It could be expected
that the desorption of the reaction products will be more
favored by a weak than by a strong acid site.27 According to
this scheme, the selective formation of formaldehyde and
then methyl formate would require both weak acidic and
basic sites to limit the H abstraction and to prevent a very
strong adsorption of formaldehyde. If the acid sites are too
strong, the residence time of formaldehyde species is long
enough to form a dioxymethylene species. If both acidic
and basic sites are stronger than those needed for methyl
formate formation, the intermediates would be further
oxidized to carbon oxides.27 The total oxidation of MeOH
to CO and CO2 usually involves a radical mechanism. The
oxidation of methanol to CO2 can proceed through a
radical reaction on basic centers or step by step with the
oxidation to formaldehyde and formic acid on redox centers.33
It is important to note that at lower temperature (below
150 °C) and lower MeOH conversion, methyl formate is
formed besides CO2. It suggests the participation of the
mechanism of total MeOH oxidation proposed via formation
of formaldehyde (strongly adsorbed on the active centers)
and, next, formic acid, which is oxidized to CO2.33
4.2. Raman, catalytic activity and mechanism of the reaction
Very interesting results were obtained for the 1 : 2 sample in
the Raman and FTIR studies. It could confirm the change in
the structure of γ-Fe2O3 after chemical reduction with hydra-
zine. New peaks observed in Raman clearly indicated the for-
mation of α-Fe2O3. It was already reported in the literature
that γ-Fe2O3 (metastable phase) is transformed into α-Fe2O3
(stable) on heating at 500 °C.23,24 This transformation under
mild conditions has not been reported yet in the literature. The
low temperature used for preparation of our catalysts seems to
be not high enough to complete this transformation. Some
authors claimed that a very small percentage of doping metal
ion such as Co2+ can dramatically change the thermal stability
of magnetic nanoparticles25 and influence the temperature of
γ-Fe2O3 to α-Fe2O3 phase transformation. In some cases, the
use of a high voltage laser in Raman study could also influence
the phase transformation. However, in our case, the laser
power was always kept at 0.9 mW at the sample to avoid
sample degradation or phase transformation. More in-depth
studies are necessary to understand this phenomenon.
The reaction resulting in methyl formate formation involves
a C–O–C coupling. There is still no consensus on the role of
the nature of the active phase, the role of the support and the
reactant molecules, or the rate determining step for this
reaction. The only agreement that comes out from the litera-
ture data is that the surface reaction sequences change in a
wide range with the reaction conditions. Methanol oxidation
reactions lead to formaldehyde (HCHO), dimethoxymethane
(CH3OCH2OCH3, DMM), and methyl formate (HCOOCH3, MF)
products. Oxidative routes to HCHO are practiced on silver-
based and iron-molybdate catalysts.26,27 Methyl formate is
produced via (nonoxidative) CH3OH dehydrogenation on
CuO or carbonylation using liquid bases28–30 and DMM can
be produced in a two-step process involving methanol oxida-
tion to HCHO followed by acetalization of HCHO–CH3OH
mixtures with liquid or solid acids.28
However, studies using DRIFT in operando mode and
studies on the basicity of the catalysts are presently in progress
to understand the mechanism of methyl formate formation.
When methanol is adsorbed on the surface of Pd, its OH
bond weakens, formation of Pd–O bonds prevails, and
chemisorbed methoxy species CH3O are formed, and then
they undergo successive dehydrogenation to formaldehyde
CH2O and then to chemisorbed CO and H.31 It could be
argued that in the first step, methanol adsorbed on Pd pro-
motes the formation of CO at the interface of the γ-Fe2O3
surface with the palladium particles. At the same time, the
dehydrogenation of water occurs, giving OH and O adsorbed
species. It was shown that in the case of Pd/SiO2 catalysts,
the methanol adsorbs and reacts at 25 °C on Pd and the dis-
sociation of methanol via both O–H and C–O bond breaking is
observed. The latter scission is only detected at 380 °C. At room
temperature, adsorbed CH3OH decomposes easily on the Pd
Conclusions
In this work, active Pd/γ-Fe2O3 catalysts were prepared by
hydrazine reduction. The AOT/cyclohexane microemulsion
permitted us to obtain very small Pd particles. It was also
shown that hydrazine reduction could change the structural
properties of the Fe2O3 samples. Indeed, the increase in the
specific surface area was observed. Moreover, the changes
observed in the Raman spectroscopy for the 1 : 2 sample
indicate the structure change.
All catalysts were active in the gas phase oxidation of
methanol at low temperature. Catalytic activity depends on
the structural properties of the catalysts. The mechanism of
744 | Catal. Sci. Technol., 2014, 4, 738–745
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
formation of methyl formate from methanol via the forma-
tion of intermediate methoxides and HCHO could explain
the results; studies using DRIFT mode are ongoing.
14 M. F. Hassan, Z. Guo, Z. Chen and H. Liu, Mater. Res. Bull.,
2011, 46, 858.
15 J. H. Li, et al. Mater. Chem. Phys., 2009, 113, 140.
16 R. Y. Hong, J. H. Li, J. Wang and H. Z. Li, China Particuol.,
2007, 5, 186.
17 R. Y. Hong, H. P. Fu, G. Q. Di, Y. Zheng and D. G. Wei,
Mater. Chem. Phys., 2008, 108, 132.
18 R. Wojcieszak, M. Genet, P. Eloy, P. Ruiz and E. M. Gaigneaux,
J. Phys. Chem. C, 2010, 114, 16677.
19 J. Eastoe, M. J. Hollamby and L. Hudson, Adv. Colloid
Interface Sci., 2006, 128–130, 5.
20 S. Eriksson, U. Nylén, S. Rojas and M. Boutonnet,
Appl. Catal., A, 2004, 265, 207.
21 A. Degan and J. Macek, Nanostruct. Mater., 1999, 12, 225.
22 O. Demoulin, M. Navez, E. M. Gaigneaux, P. Ruiz,
A. S. Mamede, P. Granger and E. Payen, Phys. Chem. Chem.
Phys., 2003, 5, 4394.
Notes and references
1 J. Lichtenberger, D. Lee and E. Iglesia, Phys. Chem. Chem.
Phys., 2007, 9, 4902.
2 N. Kakiuchi, Y. Maeda, T. Nishimura and S. Uemura, J. Org.
Chem., 2001, 66, 6620.
3 T. L. Stuchinskaya and I. V. Kozhevnikov, Catal. Commun.,
2003, 4, 417.
4 M. Bowker, C. Brookes, A. F. Carley, M. P. House, M. Kossif,
G. Sankar, I. Wawata, P. P. Wells and P. Yaseneva, Phys.
Chem. Chem. Phys., 2013, 15(29), 12056.
5 J. Agrell, K. Hasselbo, K. Jansson, S. G. Järås and M. Boutonnet,
Appl. Catal., A, 2001, 211, 239.
6 R. Wojcieszak, E. Gaigneaux and P. Ruiz, ChemCatChem,
2013, 5, 339.
7 M. Boutonnet, J. Kizling, P. Stenius and G. Maire, Colloids
Surf., 1982, 5, 209.
8 T. Hanaoka, T. Hatsuta, T. Tago, M. Kishida and
K. Wakabayashi, Appl. Catal., A, 2000, 190, 291.
9 G. Bergeret and P. Gallezot, In Handbook of Heterogeneous
Catalysis, ed. G. Ertl, H. Knozinger and J. Weitkamp, VCH,
Weinheim, 1997, vol. 2, p. 439.
10 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom.,
1976, 8(2), 129.
11 S. Alibeigi and M. Reza Vaezi, Chem. Eng. Technol., 2008,
11, 1591.
12 S. Kaval and R. V. Ramanujan, Mater. Sci. Eng., C, 2010,
30, 484.
13 I. Chourpa, L. Douziech-Eyrolles, L. Ngaboni-Okassa,
J.-F. Fouquenet, S. Cohen-Jonathan, M. Soucé, H. Marchaisb
and P. Dubois, Analyst, 2005, 130, 1395.
23 M. A. Dara, S. G. Ansarib, R. Wahabc, Y.-S. Kimd and
H.-S. Shin, Mater. Sci. Forum, 2007, 534–536, 157.
24 P. S. Sidhu, Clays Clay Miner., 1988, 36, 31.
25 S. S. Pati and J. Philip, in Nanoscience, Engineering and Technology
(ICONSET), 2011, p. 323–325, ISBN: 978-1-4673-0071-1.
26 H. Liu and E. Iglesia, J. Phys. Chem. B, 2005, 109, 2155.
27 J. M. Tatibouet, Appl. Catal., A, 1997, 148, 213.
28 E. Tronconi, A. S. Elmi, N. Ferlazzo, P. Forzatti, G. Busca and
P. Tittarelli, Ind. Eng. Chem. Res., 1987, 26, 1269.
29 M. Ai, J. Catal., 1982, 77, 279.
30 S. A. Kirillov, P. E. Tsiakaras and I. V. Romanova, J. Mol.
Struct., 2003, 651–653, 365.
31 S. P. Bates, M. J. Gillan and G. Kresse, J. Phys. Chem. B,
1998, 102, 2017.
32 G. C. Cabilla, A. L. Bonivardi and M. A. Baltanás, J. Catal.,
2001, 201, 213.
33 I. Sobczak, K. Szrama, R. Wojcieszak, E. M. Gaigneaux and
M. Ziolek, Catal. Today, 2012, 187, 48.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 738–745 | 745