The continuous conversion of ethanol and water mixtures into hydrogen over FexOy/MoO3 catalytic system—XPS and M?ssbauer studies

Article abstract of DOI:10.1016/j.molcata.2016.06.002

A preparation of the new catalytic systems consisting of ultrafine nanoparticles of Fe_xO_y (below 10?nm) supported on MoO₃, active in the generation of hydrogen from ethanol and water mixtures, is described. The catalysts were obtained as a result of thermal decomposition of iron dodecamolybdophosphates FeHPMo₁₂O₄₀ and FePMo₁₂O₄₀. Detailed characterization of catalysts was performed by several methods, especially by the M?ssbauer and XPS spectroscopies. It was shown that studied catalysts can convert both components forming the ethanol-water mixture to hydrogen with the total yield of ca. 74% at 773?K. The mechanism of this continuous process on Fe_xO_y/MoO₃ catalytic system is also proposed. The ethanol plays a role of reducing agent of MoO₃ because the latter is stable under hydrogen atmosphere up to 823?K. The reduced molybdenum can be re-oxidized by an electron transfer to iron oxides and then iron can be re-oxidized by water producing hydrogen. The water may origin from the used mixture as well as from the dehydration of ethanol. Small amount of oxygen strongly decreases the activity of the catalysts when high-concentrated ethanol is applied. In the presence of low-concentrated ethanol the oxygen does not affect the activity of the catalyst up to 723?K.

Full text of DOI:10.1016/j.molcata.2016.06.002

Accepted Manuscript
Title: The continuous conversion of ethanol and water
mixtures into hydrogen over Fe O /MoO catalytic system −
x
y
3
XPS and M o¨ ssbauer studies
Author: M. Zimowska K. Ł a˛ tka D. Mucha J. Gurgul L.
Matachowski
PII:
S1381-1169(16)30215-1
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.molcata.2016.06.002
MOLCAA 9908
To appear in:
Journal of Molecular Catalysis A: Chemical
Received date:
Revised date:
Accepted date:
11-2-2016
18-5-2016
1-6-2016
Please cite this article as: M.Zimowska, K.Ł a˛ tka, D.Mucha, J.Gurgul, L.Matachowski,
The continuous conversion of ethanol and water mixtures into hydrogen over
FexOy/MoO3 catalytic system − XPS and M o¨ ssbauer studies, Journal of Molecular
Catalysis A: Chemical http://dx.doi.org/10.1016/j.molcata.2016.06.002
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.

The continuous conversion of ethanol and water mixtures into hydrogen over
Fe /MoO catalytic system - XPS and Mössbauer studies
x
O
y
3
a
b
a
a,*
a,*
M. Zimowska , K. Łątka , D. Mucha , J. Gurgul , L. Matachowski
aJerzy Haber Institute of Catalysis and Surface Chemistry Polish Academy of Sciences, ul.
Niezapominajek 8, 30-239 Kraków, Poland
bM. Smoluchowski Institute of Physics, Jagiellonian University, ul. Łojasiewicza 11, 30-348
Kraków, Poland
*Corresponding authors. Tel.: +48 126395186, Fax: +48 124251923, E-mail addresses:
ncgurgul@cyf-kr.pl.edu (J. Gurgul), ncmatach@cyf-kr.edu.pl (L. Matachowski)
Graphical abstract
1

Highlights



Mechanism of continuous H generation from ethanol and water was described.
2
New catalytic systems were characterized by Mössbauer and XPS spectroscopies.
Catalytic activity strongly decreased with increasing amount of oxygen.


In the presence of water, small amount of oxygen does not affect the activity.
Catalysts have exhibited stable yield to hydrogen for 3 hours of reaction.
ABSTRACT
A preparation of the new catalytic systems consisting of ultrafine nanoparticles of Fe
x
O (below
y
1
0 nm) supported on MoO
mixtures, is described. The catalysts were obtained as a result of thermal decomposition of iron
dodecamolybdophosphates FeHPMo12 40. Detailed characterization of catalysts
3
, active in the generation of hydrogen from ethanol and water
O40 and FePMo12O
was performed by several methods, especially by the Mössbauer and XPS spectroscopies. It was
shown that studied catalysts can convert both components forming the ethanol-water mixture to
hydrogen with the total yield of ca. 74% at 773 K. The mechanism of this continuous process on
Fe
x
O
y
/MoO
3
catalytic system is also proposed. The ethanol plays a role of reducing agent of
MoO
3
because the latter is stable under hydrogen atmosphere up to 823 K. The reduced
molybdenum can be re-oxidized by an electron transfer to iron oxides and then iron can be re-
oxidized by water producing hydrogen. The water may origin from the used mixture as well as
from the dehydration of ethanol. Small amount of oxygen strongly decreases the activity of the
catalysts when high-concentrated ethanol is applied. In the presence of low-concentrated ethanol
the oxygen does not affect the activity of the catalyst up to 723 K.
2

Keywords: iron oxides, hydrogen production, ethanol, XPS spectroscopy, Mössbauer
spectroscopy
1
. Introduction
Currently, the hydrogen is mainly used in the chemical industry, but becomes more
significant as fuel for power generation. Hydrogen fuel cells are highly efficient electrochemical
systems converting the chemical energy into electricity. It is expected that the use of hydrogen in
fuel cell vehicles and in stationary fuel cell as well as the demand for it will grow rapidly in near
future.
The hydrogen is mostly extracted from natural gas, coal and liquid hydrocarbons but
production using the catalytic steam reforming of ethanol (1) is being intensively investigated [1-
9
]. Steam reforming does not require oxygen, has lower operating temperature and therefore it is
preferred in the industry [10]. However, a thermal decomposition of ethanol (2) has been also
proposed [11].
C
C
2
H
5
OH + 3H
2
O → 2CO
2
+ 6H
2
(1)
(2)
2
H
5
OH → CH
4
+ CO +H
2
Moreover, the low cost of hydrogen’s production is very important. From an economical point of
view, steam reforming or thermal decomposition of bio-ethanol seems to be the optimal choice
because of its low-toxicity, easy and low-cost production from renewable resources, and
relatively high content of hydrogen. Due to the potential of both processes, (1) and (2), a wide
range of the catalysts including solid oxides, transition and noble metals, and mixed metal-oxide
systems have been studied recently [3-14]. Depending on the nature of the catalyst many
3

reactions can occur simultaneously, other than those mentioned above: ethanol dehydrogenolysis,
dehydration, dehydrogenation or acetone formation. Moreover, the reactions between the
products originating from mentioned processes may be possible [6,8]. As was described
previously [9,15], among the catalytic systems used for the production of hydrogen from ethanol,
the redox cycle involving iron oxides and metallic iron have a high potential. In the first step,
ethanol reduces iron oxide (e.g. Fe
2
O ) to a pure metal and in the second step the hydrogen is
3
produced, as a result of iron oxidation by water. According to the best of our knowledge, the
application of iron oxides in the continuous process of hydrogen generation from ethanol was not
described so far.
On the other hand, the use of bio-ethanol is always connected with the presence of water and,
in fact, the mixture of water and ethanol (high or low-concentrated) can be taken into account.
Hence, primarily the steam reforming of ethanol should be considered. Our research is focused
on the finding a catalyst active in the reaction, when low or high-concentrated ethanol is applied.
In present paper the new catalysts obtained by thermal decomposition of iron
dodecamolybdophosphates, composed of ultrafine nanoparticles of Fe
x
O
y
on MoO , were
3
investigated. Their catalytic activity was tested in the continuous process of hydrogen generation
from the mixtures containing various amounts of ethanol and water. The role of water content
and the influence of oxygen on the hydrogen yield is considered. The mechanism of reaction
carried out on Fe
x
O
y
/MoO catalytic systems is also proposed.
3
2
. Experimental
.1. Catalysts preparation
2
4

Iron dodecamolybdophosphates FeHPMo12
addition of stoichiometric quantities of FeCl and Fe
of H 40 acid (Sigma-Aldrich, all reagents). The syntheses were carried out at ambient
O40 and FePMo12O40 were obtained by the
2
2
(SO
4 3
)
, respectively, to an aqueous solution
3
PMo12O
temperature without stirring and the resulting precipitates were dried in air at 363 K. The Keggin
structure of prepared heteropolysalts was verified by FT-IR method (Supplementary material,
Fig. S1), while the oxidation state of iron was determined by Mössbauer spectroscopy. Then,
both heteropolysalts were annealed at 773 K in air atmosphere for 50 hours. These decomposed
materials (denoted as FeHMo-cal and FePMo-cal, respectively) have been characterized by
number of techniques and afterwards used in the catalytic experiments.
2
.2. Catalysts characterization
.2.1. FT-IR spectroscopy
The infrared spectra were registered with a Nicolet 380 FTIR spectrometer in the mid-infrared
2
-1
(
MIR) region 4000 - 400 cm . The samples (1 mg) were mixed with 200 mg of spectroscopically
pure KBr and pressed into disks before the spectra were recorded. Spectra were obtained by co-
-1
-1
addition of 64 scans at a resolution of 4 cm and next were normalized to 870 cm band using
the Thermo Scientific OMNIC™ software package.
2
.2.2. Mössbauer spectroscopy
The local properties of FePMo12O40 and FeHPMo12O40 precursors were investigated by means
of Mössbauer spectroscopy employing the 57Fe gamma resonance transition. A Mössbauer
System that consists of the Janis top loaded liquid helium cryostat (Janis Research Company,
Wilmington, MA 01887 USA) integrated with a conventional constant-acceleration spectrometer
(Science Engineering & Education Co. USA) of the Kankeleit type in transmission geometry was
57
used. During measurements, 100 mCi Mössbauer Co(Rh) γ-ray source and the absorbers were
5

kept at room temperature. The absorbers were made of the fine powdered materials placed in a
thin-walled (~0.1 mm) cylindrical plastic containers. The used absorber thicknesses of about 40
2
mg/cm were calculated from the optimisation procedure [16]. The resonance of 14.4 keV gamma
rays (for a given measurement and the energy scale calibration) were detected simultaneously by
means of two independent LND Kr/Co proportional gas counters attached at opposite sides of
2
57
the driving system. The drive velocity calibration was performed with a second Co(Rh) source
against a standard metallic iron foil at ambient temperature.
The Mössbauer spectra were analyzed numerically by fitting a hyperfine parameter
distribution (HPD) using the Voigt-line-based method of Rancourt and Ping [17]. In this method,
the HPD for a given crystal site corresponding to similar structural, chemical and magnetic
properties is constructed by a sum of Gaussian components for the quadrupole splitting (QS)
distributions and, if necessary, the magnetic hyperfine field Bhf distributions. The isomer shift
(
IS) can be linearly coupled to the primary hyperfine parameters (QS, Bhf).
.2.3. DSC/TG method
The DSC/TG measurements of FeHPMo12 40 and FePMo12O40 heteropolysalts were
2
O
performed in the air flow using Netzsch STA 409 PC Luxx apparatus (Supplementary section,
Fig. S2). The samples of about 40 mg were heated in air at the temperature range of 293-1000 K
with the heating rate of 10 K/min.
2
.2.4. X-ray diffraction
The X-ray diffraction patterns for FeHMo-cal and FePMo-cal catalysts were recorded at
ambient temperature within the range of 2 = 5-90°, step size 0.02° and 12 s/step using the
Siemens D5005 diffractometer (CuKα radiation, 40 kV, 40 mA) equipped with the secondary
beam graphite monochromator. Before measurements the catalysts were annealed at 363 K for 2
6

h to remove physically adsorbed water. The crystallite sizes of studied samples as well as the
MoO and Fe standards were calculated by ‘Full-Pattern Decomposition’ method, which was
3
O
2 3
completed by TOPAS R2.1 software [18]. The fundamental parameters approach was also
applied [19].
2
.2.5. H TPR method
2
The reducibility profiles of FeHMo-cal and FePMo-cal catalysts were obtained using the
quartz gas flow-through microreactor at the temperature range of 293-1100 K. A gas mixture of
5
% H
TPR experiments both catalysts were kept in a stream of helium at 373 K for 1.5 h to remove
physically adsorbed water. Subsequently, each catalyst was cooled down to ambient temperature
and the H TPR measurement was performed with the heating rate of 10 K/min and the flow rate
of reducing mixture of 30 cm /min. The H
recorded using a TCD detector were compared with those obtained for spectrographically
standardized MoO and Fe (Johnson Matthey Chemicals Ltd., both).
.2.6. XPS spectroscopy
2
in Ar as a reducing agent was passed through the catalyst of ca. 0.025 g. Before the
H
2
2
3
2
TPR profiles of FeHMo-cal and FePMo-cal catalysts
3
O
2 3
2
The X-ray Photoelectron Spectroscopy (XPS) measurements were carried out with a
hemispherical analyzer (SES R4000, Gammadata Scienta, pass energy 100 eV). The

unmonochromatized AlK X-ray source (1486.6 eV, 12 kV, 15 mA), with no charge neutralizer,
was applied to generate core excitation. The system was calibrated according to ISO 15472:2001.
The energy resolution of the system, measured as a full width at half maximum (FWHM) for Ag
3d
5/2 excitation line, was 0.9 eV. The powder samples were pressed into indium foil and mounted
on a dedicated holder then UHV evacuated. During the measurements, the base pressure in the
-9
analysis chamber was about 10 mbar. The area of the sample analysis was approximately 3
7

2
mm . Because of insulating character of FeHPMo12O40 and FePMo12O40 salts, their binding
energy (BE) values were charge-corrected to the carbon C 1s excitation set at 285.0 eV. The
Shirley-type background subtraction was used to the spectra prior to fitting procedure where
Voigt line shape i.e. Gaussian/Lorentzian functional (70:30) was applied. The Fe 2p core
excitations were deconvoluted with relative intensity ratio of 2p3/2 and 2p1/2 lines fixed to 2:1.
2
.2.7. Scanning electron microscopy and EDS analysis
Morphology of the solids was carried out by means of JEOL JSM – 7500F Field Emission
Scanning Electron Microscope equipped with the EDS (energy dispersive spectra) detection
system of characteristic X-ray radiation.
2
.3. Catalytic tests
The experiments were carried out using a quartz gas flow-through microreactor. The nitrogen
was passed over solution containing 96 vol% of ethanol + 4 vol% of water or 48 vol% of ethanol
52 vol% of water, denoted here as (96E+4W) or (48E+52W), respectively. The concentration
+
-
3
-3
of ethanol in the flowing nitrogen was c.a. 15 g·m and 10 g·m , respectively. The influence of
oxygen on the catalytic activity of catalysts was also studied. The reaction was performed in the
-1
temperature ranging from 573 to 923 K and GHSV=10000 h . Prior to the reaction, 0.46 g of
catalysts (0.25-0.50 mm fraction) were standardized in the flow of nitrogen at 473 K for 2 hours.
The products were analyzed using the SRI 8610C gas chromatograph equipped with the HayeSep
D column and the TCD and HID detectors arranged in series.
3
. Results and discussion
8

The iron dodecamolybdophosphates FeHPMo12O40 and FePMo12O40 were chosen as
precursors of the catalysts, because as it was previously shown the surface of heteropolysalts with
molybdenum as addenda atoms does not change with increasing relative humidity [20]. It was
expected that these properties are maintained in the FeHPMo-cal and FePMo-cal catalysts,
obtained after total decomposition of iron molybdophosphates, what was essential for the
hydrogen generation from the mixtures containing various amounts of water and ethanol. On the
other hand, the DSC results (Supplementary material, Fig. S2) showed that, before the
decomposition of both salts only endothermic effects related to the loss of crystallization water
and the water hydrated protons, are observed. In turn, the TG results have shown that the
decomposition of iron molybdophosphates took place without the loss of their mass. It indicates
that the atomic compositions of new catalysts are practically the same as the composition of
precursors. The catalysts were characterized with several methods such as FT-IR, XRD, H TPR,
2
Mössbauer and XPS spectroscopies. The BET method described previously [21], showed that the
2
FeHPMo-cal and FePMo-cal catalysts have a small specific surface area 1.5 and 2.2 m /g,
respectively. Hence, it is assumed that the studied reaction mainly proceeds on the external
surface of the catalysts.
As seen in Figure 1, the diffraction patterns of FeHPMo-cal and FePMo-cal samples after
calcination at 773 K show the reflections characteristic for MoO phase (PDF 00-035-0609) [22].
3
Both catalysts exhibit main Bragg peaks at 2Θ ≈ 12.78º, 23.34º, 25.88º, 27.34º, 33.76º and 38.97º
that can be indexed as (020), (110), (120), (021), (111) and (060) reflections associated with the
Pbnm symmetry of MoO
3
orthorhombic phase. Nevertheless, FeHPMo-cal sample is
phase in comparison to FePMo-cal catalyst. It was
characterized by better crystallinity of MoO
3
confirmed by the calculation, using the method described in the Experimental section, which
9

showed MoO crystallite sizes equal to 47 and 38 nm, respectively. The patterns of studied
3
samples displayed also the additional reflections (see inset in Fig. 1, marked as asterisk), which
according to the XRD database can be attributed to the iron oxides phases. Unfortunately, due to
the weak and broadened reflections as well as to their overlapping, the identification of iron
oxides and determination their crystallite sizes was impossible. However, taking into account the
atomic ratio of Mo:Fe=12:1 in studied samples, the formation of well-dispersed Fe
x
O
y
crystallites
of much lower sizes compared to MoO crystallites seems to be very probable. This suggestion
3
correlates well with the results obtained from Mössbauer spectroscopy described below. Thus, it
can be accepted that the XRD reflections represent a mixture of iron oxides with very low
crystallite sizes.
5
4
3
2
1
000
000
000
000
000
0
o
o
o
B
MoO3
* Fe Oy
o
x
4
00
o
oo
*
o
o
o
o
200
o
o
o
*
*
*
*
o
o
0
30
o
40
50
60
o
o
o
oo
o
o
o
o
o
o
o
o
o
B
A
10
20
30
40
50
60
70
80
90
2
Theta
Fig. 1. The XRD patterns of: A/ FeHPMo-cal and B/ FePMo-cal catalysts. The reflections
originating from MoO are marked by circles while those from Fe by asterisks.
3
x
O
y
1
0

Comparing the FTIR spectra recorded for the new catalysts (Figs. 2C,D) with those for ‘as
synthesized’ heteropolysalts presented in Supplementary material (Fig. S1), it can be concluded
that in both cases a total decomposition of the Keggin structure occurs. The FeHPMo-cal and
FePMo-cal catalysts exhibit similar bands to those recorded for molybdophosphoric acid
H
3
PMo12
O
40 after its decomposition at 773 K (Fig. 2B, HPMo-cal) and also for MoO
It indicates that both studied catalysts are decomposed into materials with the structure similar to
that of MoO . This result is in accordance to the results published recently [23,24] and confirms
conclusion derived from XRD analysis.
3
(Fig. 2A).
3
6
29
870
9
97
D
C
1068
651
592
B
A
1200
1000
800
600
400
-1
Wavenumber, cm
Fig. 2. The FT-IR spectra of: A/ MoO , B/ HPMo-cal, C/ FePMo-cal and D/ FeHPMo-cal
3
catalysts.
1
1

However, the analysis of FTIR spectra in the region of 1200-400 cm−1 reveals some
differences in the structure of formed MoO phase present in the investigated catalysts. Basing on
the knowledge that MoO is formed by the MoO
3
3
6
octahedra, for which stretching vibrations occur
-1
-1
in the range of 1000-500 cm , one can distinguish the characteristic bands at 997, 870, 592 cm
-1
(
Fig. 2A). The band at 997 cm originates from the terminal Mo=O symmetric stretching mode
-1
[
25-27], while the dominant band at 870 cm is associated with the vibration of the Mo-O-Mo
-1
bridging bonds. The band at 592 cm is characteristic for the vibration of the Mo
formed by edge shared MoO polyhedra building the orthorhombic MoO structure. As seen in
Fig. 2, this band is shifted to 651 cm in the spectra of HPMo-cal and FePMo-cal samples while
2
O entity
2
6
3
-1
-1
it shifts to 629 cm in the case of FeHPMo-cal catalyst. It may be attributed to the alternating of
the bond lengths in the MoO octahedra associated with a vibration of the apical Mo-O bond due
to the presence of P species in the lattice of MoO . The shift in the spectrum of FeHPMo-cal
catalyst (Fig. 2D) is the smallest suggesting less affected network of MoO by phosphorous ions
in comparison to other samples. However, the broadening of the band and the presence of the less
6
3
3
-1
intensive shoulder at 651 cm in FeHPMo-cal catalyst suggests also the formation of the partially
affected MoO octahedra by P species, similarly as in HPMo-cal or FePMo-cal samples. Thus, in
FeHPMo-cal catalyst two forms of MoO lattice, substantially and less affected by P ions, exist.
The lower amount of phosphorous in the latter phase is caused by the formation of distinct phase
6
3
-1
containing P-O bonds. It is evidenced by the band at ca. 1068 cm , attributed to the stretching
vibration of P-O bond in the PO with tetrahedral coordination, which appears in the spectrum of
4
FeHPMo-cal catalyst. The FT-IR results also show that the shift of the band originating from Mo-
O bond strongly depends on the kind of precursor, from which catalyst was obtained. This
1
2

indicates some difference in the structure of prepared catalysts which can influence their catalytic
activities in the studied reaction.
FeHPMo O 'as synthesized'
FePMo O 'as synthesized'
1
2
40
12 40
2
2
2
15
14
13
2
2
2
69
68
67
after calcination at 773K
after calcination at 773K
112
2
2
2
2
66
64
62
60
1
1
1
11
10
09
FeHPMo-cal
FePMo-cal
-6
-4
-2
0
2
4
6
-6
-4
-2
0
2
4
6
Velocity (mm/s)
Fig. 3. The Mössbauer spectra of ‘as synthesized’ FeHPMo12
O
40 and FePMo12O40 salts, and
FeHPMo-cal and FePMo-cal catalysts.
The Mössbauer spectra recorded at room temperature for the ‘as synthesized’ FeHPMo12
and FePMo12 40 salts and for the catalysts obtained after calcination at 773 K are presented in
Fig. 3. As expected, the spectra of ‘as synthesized’ heteropolysalts differed each other
significantly (top of Fig. 3). The Mössbauer spectrum of ‘as synthesized’ FeHPMo12 40 salt can
be well described by three symmetric doublets with hyperfine parameters listed in Table 1. The
O
40
O
O
1
3

2+
simple analysis shows that 75% of salt contains Fe ions, whereas the remaining iron exists in
higher oxidation state. The component with IS = 0.19 mm/s and QS = 0.41 mm/s (see Table 1) is
characteristic for Fe3+ in tetrahedral surroundings [28-31]. Iron (II) spectral contribution is
described by two components, which parameters are well agreed with iron coordination number
of four (Td) (IS = 0.95 mm/s, QS = 2.01 mm/s) and six (Oh) (IS = 1.17 mm/s, QS = 3.01 mm/s).
2+
This can be linked to the presence of [Fe(H2O)6] species in the ‘as synthesized’ FeHPW12
O
40
precursor, as reported recently [31].
3+
In the case of ‘as synthesized’ FePMo12
O40 salt, only the octahedral Fe can be distinguished.
The main phase (over 93% of total spectrum area, Table 1) can be explained by the presence of
3
+
[Fe(H O)
2 6
] countercations localized between the Keggin units [32,33]. A minor, but noticeable
component of very high QS can also be related to ferric cations, which are probably related to the
3+
highly defected surroundings of Fe ions on the surface.
Table 1. The hyperfine parameters obtained from Mössbauer spectroscopy measurements at
room temperature. For details see the text.
-1
QS (mm s-1)
w (mm s-1)
Sample
IS (mm s )
Area (%)
Site
3
+
FeHPMo12
O
40
0.19
0.95
0.41
2.01
3.01
0.30
0.22
0.15
25
6
d
Fe (T )
2
+
‘
as synthesized’
d
Fe (T )
2
+
1
.17
69
Fe (O
h
)
3
+
FeHPMo-cal
0.16
.36
0.63
0.73
0.15
0.20
70
30
Fe (T
d
)
3+
0
Fe (O
h
)
3
+
FePMo12
O
40
0.32
0.49
0.51
1.90
0.17
0.41
93
7
Fe (O
h
)
3+
‘
as synthesized’
Fe (Oh-def)
3
+
FePMo-cal
0.17
0.64
0.70
0.14
0.21
66
34
Fe (T )
d
3+
0
.37
h
Fe (O )
1
4

In turn, the shapes of γ-ray absorption spectra of both FeHPMo-cal and FePMo-cal catalysts
are very similar, however, some details are different (i.e. the asymmetry of doublets). The spectra
are superpositions of two subspectra, which can be fitted by symmetric quadrupole doublets
using the same numerical procedure described in Experimental section. The hyperfine parameters
and relative intensities of all components are listed in Table 1. The most characteristic feature of
doublets presented for FeHPMo-cal and FePMo-cal catalysts (Fig. 3) is their low value of isomer
3
+
shift IS  0.37 mm/s, which is typical for Fe state (Table 1). It strongly suggests mostly
tetrahedral (IS < 0.3 mm/s) than octahedral (IS > 0.3 mm/s) surroundings [30,34-36]. The
tetrahedral coordination of Fe3+ has been previously identified as to be more favorable in the
catalysts active for hydrogen production compared to Fe-species of octahedral coordination [37].
The quadrupole splitting (QS) of iron species in FeHPMo-cal and FePMo-cal catalysts are
generally in the range of 0.63-0.73 mm/s with quite reasonable values of the Mössbauer peak
width of the resonance line (w). The species with lower QS can be attributed to the well-defined
framework sites in tetrahedral coordination, with high symmetry, and low content of defects.
Such Fe3+ species may occupy the tetrahedral sites in the lattice of Fe
tetrahedral environment of MoO moieties. The species with higher QS, can be associated with
the adjacent iron ions near to defect sites and can be attributed to the asymmetric charge
O
3 4
oxide or
3
3+
5+
distribution associated with the random distribution of Fe species and the presence of Mo or
6+
Mo ions on the neighboring sites.
The knowledge about the Fe species existing in the catalysts obtained after thermal
decomposition of heteropolycompounds is very important for the understanding of their catalytic
activity [38]. The values of hyperfine parameters have rather excluded the formation of Fe2-
1
5

x
Mo
x
O spinel ferrites during this process, because they show higher isomer shift IS ~ 0.6 mm/s
4
characteristic for mixed valence systems [39]. On the other hand, the spectra obtained in ambient
temperature do not show any spectral contribution of magnetically split components, which
proves non-magnetic “bulk” state of both catalysts. It was observed previously that similar
funnel-shape of the Mössbauer spectrum is characteristic for ultrafine Fe
2
O crystallites what was
3
additionally confirmed by TEM measurements [40]. Thus, one can accept that the measured
spectra belong to the ultrafine particles of iron oxides in the superparamagnetic state. It is worth
noting, that similar results were found in iron-molybdenum oxide catalysts supported on titania or
alumina [41-43], where the “core-shell” model was proposed for the occurrence of the iron ions
both at the surface and in the bulk of particles. In this model, each doublet can be assigned to the
bulk - tetrahedral (lower QS) species and to the surface - octahedral (higher QS) species. Basing
on this model and taking into account the relative ratio of T
in the FeHPMo-cal and FePMo-cal catalysts (Table 1), we can determine the size of Fe
particles below 10 nm [44]. Basing on obtained results, we can conclude that upon heating at 773
d
:O
h
close to 70:30, which is observed
x
O
y
3
+
]2+ present in ‘as synthesized’ heteropolysalts are
K the aqua complexes [Fe(H
decomposed. Most probably, this process is accompanied by the charge transfer between iron and
molybdenum cations. As the result, the presence of various Fe oxides distributed on the
surface of MoO is observed in XRD measurements (Fig. 1). Such an explanation was proposed
2
O)
6
] , Fe(H
O)
2 6
x
O
y
3
on the basis of experimental [32] and theoretical studies [45,46] of redox properties of the Keggin
type iron molybdophosphates and is also confirmed by the XPS results presented below. Thus, it
can be stated that both FeHPMo-cal and FePMo-cal systems consist of ultrafine nanoparticles of
Fe
x
O
y
(below 10 nm) on partially reduced MoO . Such characterized catalysts were used for the
3
production of hydrogen from ethanol.
1
6

Due to the presence of hydrogen in the reaction products, its interaction with both catalytic
systems was investigated using the H TPR method. As seen in Fig. 4A, both studied catalysts
2
show lack of hydrogen consumption up to ca. 600 K. At higher temperatures the FePMo-cal
sample displays very small uptake of hydrogen with slight maximum at 748 K (Fig. 4A) and
much larger peak with maximum at 1020 K. According to the reduction temperature (760 K) of
spectroscopically standardized Fe
reduction of Fe to Fe . For comparison the curve converted into iron content in both
catalysts was shown in Fig. 4B, marked as Fe -recal. It is visible that small amount of iron
oxides in the catalysts does not allow for a precise analysis of measurement. However, it can be
suggested that the shift of reduction temperature from 760 K (standard Fe ) to 748 K (FePMo-
cal) can be caused by the lower crystallite sizes of Fe in the latter sample. It means that
crystallite sizes of iron oxide in FePMo-cal sample are lower than that calculated for standard
Fe equal to 108 nm. On the other hand, according to Smith and Rohrer [47], MoO is able to
form hydrogen bronzes H MoO , where 0.23 ≤ x ≤ 0.4. Due to this also the formation of partially
protonated phase is possible.
In turn, the FeHPMo-cal catalyst shows two small peaks, at 865 and 968 K as well as larger
peak at 1016 K. The first one can be assigned to the reduction of Fe to FeO whereas the
second to the reduction of FeO to metallic iron. It can indicate the lack of Fe in this sample or
its very low crystallite sizes. Thus, it can be concluded that FePMo-cal sample contains mostly
Fe , whereas in FeHPMo-cal catalyst mostly Fe exists. This difference in the composition
of iron oxides formed on MoO surface can affect the catalytic activity of studied samples.
O
2 3
2 3
(Fig. 4B, Fe O -s) the first one peak can be assigned to the
O
2 3
3
O
4
O
2 3
O
2 3
2
O
3
O
2 3
3
x
3
O
3 4
2
O
3
O
2 3
O
3 4
3
1
7

Fig. 4. The reducibility profiles of: A/ FeHPMo-cal (black line) and FePMo-cal (red line)
catalysts and B/ Fe standard (green line), MoO (blue line) and Fe (black line) standards
recalculated on the metal amount in the catalysts.
O
2 3
3
2
O
3
The significant hydrogen uptake is observed in both studied catalysts above 800 K, attaining
the maximum ca. 1016-1020 K (Fig. 4A), due to the reduction of MoO . This finding was
confirmed by the H TPR experiment with spectroscopically standardized MoO (Fig. 4B). The
difference between curves obtained for the catalysts and standard MoO , visible in the maximum
of reduction temperature, as well as in the shape of curves, may be due to different crystallite
3
2
3
3
sizes of the samples. Indeed, the crystallite size calculated for standard MoO
is significantly differ from values 38 and 47 nm calculated for the catalysts.
3
was 128 nm, what
1
8

However, according to some reports, not only hydrogen can reduce MoO
showed that the reduction of MoO , by ethanol vapor performed at 673 K for 12 h, leads to
MoO . Also, Rosmaninho et al. [9] pointed out the possibility of iron oxides reduction by
ethanol. Thus, taking into account the temperature range of studied reaction (573 to 923 K) it can
be concluded that hydrogen formed during reaction can influence the reduction of MoO above
23 K only.
The catalytic experiments with the use of our new catalytic systems were performed in the
3
. Holze et al. [48]
3
2
3
8
flow mode for two different compositions of the reaction mixture. In the first case, a nitrogen as a
carrier gas was passed over the solution containing 96 vol% of ethanol and 4 vol% of water
(denoted here as (96E+4W)). In the second one, the solution was consisted of 48 vol% of ethanol
and 52 vol% of water (denoted here as (48E+52W)). Hence, at the same gas flow rate, the
-3
concentrations of ethanol were about 15 and 10 g·m , respectively, whereas the molar ratio of
ethanol to water was 4.2:1 for (96E+4W) and 1.3:1 for (48E+52W) solutions. In such conditions
the experiments with ethanol-water mixtures were carried out at the temperature range of 573-
923 K. Taking into account the amount of water resulting from the partial pressure of water over
the ethanol-water solutions and the concentration of ethanol in the flowing nitrogen, it was
calculated that the maximum quantity of the resulting hydrogen, originating from ethanol and
water, could attain ca. 2.3 vol% for (96E+4W) and 1.8 vol% for (48E+52W) solutions. In both
cases these amounts were taken as 100% for the calculation of the hydrogen yield. Thus, the
hydrogen yield (YH ) can be defined as:
2
1
9

To compare the yield to hydrogen, calculated from ethanol and water, with that from ethanol only
the reaction rate expressed as (molethanol/gcat·s) was introduced in the Figure 5-8.
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
14
12
10
8
A
B
C
6
4
2
0
5
50
600
650
700
750
800
850
900
950
Temperature, K
Fig. 5. The yield to hydrogen (solid symbols) and the reaction rate (open symbols) obtained using
96E+4W) solution on: A/ FeHPMo-cal, B/ FePMo-cal catalysts in N atmosphere and C/
FePMo-cal catalyst with an addition of 0.5 vol% of oxygen.
(
2
During reaction in nitrogen atmosphere with (96E+4W) solution the yield to hydrogen
increases to almost 70% at 823 K for FeHPMo-cal and 873 K for FePMo-cal (Fig. 5). This
difference observed in catalytic activities may be due to the more reduced MoO
FeHPMo-cal catalyst in comparison to FePMo-cal one. However, different composition of iron
oxides evidenced by H TPR measurements may also be important. At higher temperatures the
yields to hydrogen start to decrease for both catalysts (Fig. 4A), because occurring hydrogen was
3
present in
2
2
0

probably used for the reduction of MoO , which is visible above 823 K in Fig. 4B. It should be
3
also noted that apart from hydrogen mainly ethylene, with the yield exceeding 10 %, as well as
ethane and methanol, with a few percent yield, were produced. Apart from these products also
traces of methane, CO, CO , acetic aldehyde and dimethyl ether were detected in the product
2
mixtures with the yield below 1 %. Ethylene can be produced by the dehydration of ethanol,
ethane by the hydrogenation of ethylene whereas the appearance of methanol requires detailed
study. It should be underlined that XPS measurements showed that the amount of carbon species
on the surface of studied catalysts increased almost twice after reaction, from ca. 23-24 % to ca.
43-46 %. Therefore, the precise balance of carbon-containing compounds was not possible.
It is also interesting that in the presence of small amount of oxygen (0.5 vol%), when
(
96E+4W) solution was used, the yield to hydrogen on FePMo-cal catalyst was much lower in
the entire temperature range (Fig. 5). Similar result was obtained for FeHPMo-cal catalyst (not
shown). It seems that the oxygen partially prevents the reduction of MoO by ethanol, which is
3
probably required for the effective reaction. As seen in Fig. 5A (black lines) and Fig. 5B (red
lines), the increasing difference between both curves obtained for particular catalysts can be
observed with rising temperature. Probably, it is due to the presence of water (4 vol%)
participating in the reaction what is especially visible in higher temperatures. In turn, the
comparison of catalytic activities of FePMo-cal sample in the presence (Fig. 5C, blue lines) or
absence of oxygen (Fig. 5B, red lines) shows how strong the oxygen influences the yield to
hydrogen. However, when comparing the share of water in the reaction at the same temperature
(e.g. 823 or 873 K) it appears that the oxygen has no effect on this.
2
1

40
30
20
10
0
8
7
6
5
4
3
2
1
0
0,0
0,2
0,4
0,6
0,8
1,0
Oxygen concentration, vol%
Fig. 6. The yield to hydrogen (black line) and the reaction rate (red line) obtained at 773 K on
FePMo-cal catalyst with the addition of different amount of oxygen using (96E+4W) solution.
To confirm above mentioned conclusion, the experiments with the influence of oxygen
concentration on the yield to hydrogen was performed for FePMo-cal sample at 773 K. As seen
in Fig. 6, an increasing concentration of oxygen leads to the rapid decrease of the hydrogen yield
what corroborates the results presented previously in Fig. 5C. It may indicate that oxygen can
prevent the reduction of MoO
metallic iron to form hydrogen. This is also visible in Fig. 6 that the share of water in the
production of hydrogen is almost constant. Hence, it can be assumed that the reduction of MoO
is required for the electron transfer between molybdenum and iron ions during reaction. In turn,
3
by ethanol as well as may block the reaction of water with
3
2
2

the reduced iron can be re-oxidized by water forming hydrogen. It seems to be clear that the
presence of oxygen substantially affects the yield to hydrogen when (96E+4W) solution is used.
On the other hand, the decrease of hydrogen yield caused by the reaction of oxygen with
occurring hydrogen cannot be excluded. Thus, it can be stated that whether these phenomena
occur individually or together, each of them can strongly influence the mechanism of reaction.
On the other side, it is important for further applications how many moles of hydrogen are
obtained from 1 mole of ethanol. Taking into account the hydrogen yield of 70% and assuming
that all water molecules originating from the reaction mixture are converted into hydrogen,
detailed calculation shows that from 1 mole of ethanol about 2 moles of hydrogen is generated.
That is 2/3 amount of hydrogen, which could be obtained from ethanol.
Fig. 7 presents the results obtained for FeHPMo-cal catalyst, when the reaction mixture
originated from (48E+52W) solution was used, with and without the presence of oxygen. As
seen, the yield to hydrogen without the presence of oxygen is slightly higher (ca. 74% at 773 K)
than that obtained using (96E+4W) solution (ca. 70% at 823 K) (Fig. 5). Alike as previously, it
was calculated how many moles of hydrogen are obtained from 1 mole of ethanol. In case of
FeHPMo-cal catalyst, 74% of hydrogen yield was taken into account as well as the assumption
that all water molecules originating from the reaction mixture are converted into hydrogen.
Result was the same as previously i.e. from 1 mole of ethanol ca. 2 moles of hydrogen is
generated. Thus, in both cases 2/3 of the possible amount of hydrogen was obtained.
2
3

70
60
50
40
30
20
10
0
14
12
10
8
A
C
B
D
6
4
2
0
550
600
650
700
750
800
850
Temperature, K
Fig. 7. The yield to hydrogen (solid symbols) and the reaction rate (open symbols) obtained using
48E+52W) solution on FeHPMo-cal catalyst: A, C/ without presence of oxygen and B, D/ with
(
an addition of 0.5 vol% of oxygen.
The suggestion resulting from Fig. 5A,B concerning the share of water in hydrogen
production can be supported by the larger difference between curves A and C, as observed in Fig.
7
(black lines). It seems to be clear that when the concentration of water in the reaction mixture
increases from 4 vol% (Fig. 5A, black lines) to 52 vol% (Fig. 7A vs. C), the share of hydrogen
originating from water also rises. Moreover, an observed impact of oxygen on the hydrogen yield
is much weaker (Fig. 7A vs. B). The presence of larger amount of water in the reaction mixture
(48E+52W) causes that the influence of oxygen is almost negligible up to 723 K. At higher
temperature significant difference is observed. It can be suggested that up to 723 K the water
2
4

protects the surface of the catalyst from the interaction with oxygen. Similar effect of adsorbed
water/OH species on the steam reforming of ethanol was also observed in the case of Co
Fe mixed oxides [7]. However, it can be underlined that despite the presence of oxygen, the
3
O -
4
O
2 3
share of water in the reaction was still significant (Fig. 7B vs. D). This fact is very promising for
future applications because the maintenance of completely anaerobic conditions is very difficult.
The comparison of results obtained for FeHPMo-cal catalyst using two various ethanol-water
mixtures (Fig. 5 and 7) clearly shows that using (48E+52W) solution lower reaction temperature
(
773 K vs. 823 K) is required to generated 2 moles of H from 1 mole of ethanol. Additionally,
2
with low-concentrated ethanol-water mixture more water molecules are converted into hydrogen
and the catalyst is more resistant to the influence of oxygen.
45
40
35
30
25
20
15
10
5
0
9
8
7
6
5
4
3
2
1
0
A
B
0,5
1,0
1,5
2,0
2,5
3,0
Time of stream, h
Fig. 8. The yield to hydrogen (solid symbols) and the reaction rate (open symbols) on FePMo-cal
catalyst vs. time of stream using (96E+4W) solution: A/ procedure with gradually rising
temperature up to 773 K and B/ introduction of ethanol at 773 K.
2
5

Other experiments have also confirmed the impact of MoO
hydrogen (Fig. 8). The catalytic reaction was performed on FePMo-cal catalyst with the use of
96E+4W) solution. The first experiment was carried out with gradually increasing temperature,
from 573 to 773 K and then the latter temperature was maintained for 3 hours. As seen in Fig.
A, the catalyst exhibited stable catalytic activity. In the second experiment (Fig. 8B), the
3
reduction on the yield to
(
8
FePMo-cal catalyst was heated to 773 K in nitrogen and then the reaction mixture was
introduced. The reaction has also been accomplished for 3 hours.
As it is seen, the difference in the yield to hydrogen is significant (Fig. 8A vs. B). In the first
experiment, MoO
second, the initial 2.3% after 3 hours turns into 20.1%, probably due to the progressive reduction
of MoO with time of experiment. It confirms that the ethanol is the reducing agent since the
3
is already reduced and the yield to hydrogen oscillates around 40-45%. In the
3
reduction by hydrogen at this temperature was excluded (Fig. 4B). Alike, in the first experiment
the share of water in the hydrogen production, expressed as difference between curves, is stable
(Fig. 8A, black lines) whereas in the second one this share slightly increases with rising
temperature (Fig. 8B, red lines). It is worth noting that in the first experiment the yield to
ethylene, ethane and methanol decreases with time of stream which can be caused by the
formation of carbon deposit, whose twofold increase has been confirmed by XPS measurements.
However, the carburization of MoO
3
and the formation of Mo
2
C with a C
2
H
6
/H mixture cannot
2
be excluded [49]. In the second experiment, the yield to ethylene is roughly constant, while the
yield to ethane and methanol slightly increases. It may indicate that the carbon deposit is mostly
formed in lower temperatures.
2
6

Basing on the results obtained so far, the reaction mechanism can be proposed as follows. The
is reduced by ethanol forming Mo4+ and Mo5+ cations, which are re-oxidized by an
MoO
3
electrons transfer with iron oxides. The obtained Fe reacts with water forming iron oxides and
producing hydrogen. The water may origin from the ethanol-water solution as well as from the
dehydration of ethanol. According to proposed mechanism of studied reaction the hydrogen can
be formed in continuous process. Moreover, presented mechanism would explain the difference
observed in the hydrogen yields between both catalysts.
Table 2. The BE values (eV) and relative areas of components (%) of Mo 3d5/2 and O 1s core
excitations obtained for all catalysts and for HPMo-cal sample.
Core
excitation
FePMo-cal
FePMo-cal
used
FeHPMo-cal
FeHPMo-cal
used
HPMo-cal
MoO
3
MoO
2
BE
eV)
Area
BE
Area
(%)
BE
Area
(%)
BE
Area BE
(%) (eV)
Area
BE
Area
(%)
BE
Area
(%)
(
(%)
(eV)
(eV)
(eV)
(%)
(eV)
(eV)
Mo 3d5/2 ---
---
---
230.0 24.6
230.7 17.4
231.8 26.4
233.3 24.9
235.1 6.7
---
---
---
---
229.9
230.4
231.6
233.1
235.3
---
11.9 ---
19.2 ---
---
---
---
---
---
---
---
---
229.8 23.1
--- ---
A
B
C
D
E
-
--
2
2
2
31.8 4.8
33.6 88.8
34.9 6.4
231.5 1.3
233.2 67.7
234.0 31.0
529.1 0.6
530.9 55.1
531.7 40.5
533.4 3.8
22.8 231.8 3.8
43.0 233.3 86.5
231.2 11.8
233.0 65.1
233.7 100
--- ---
529.4 0.7
--- ---
3.1
---
234.2 9.7
529.3 0.9
---
---
O 1s
529.4 0.8
---
---
528.5 0.5
530.5 48.8
531.3 39.8
532.7 10.9
5
5
5
31.2 66.8
32.3 27.9
32.8 4.5
530.7 29.2
531.8 35.9
532.8 34.9
530.7
531.8
533.4
22.4 531.1 64.4
59.1 531.9 26.0
18.5 533.2 8.7
531.4 88.4
532.6 10.9
The FeHPMo-cal and FePMo-cal catalysts, before and after catalytic tests using (96E+4W)
solution, were investigated by XPS spectroscopy (Tables 2 and 3). The XPS results of ‘fresh’ and
used catalysts were analyzed in the BE regions of Mo 3d, O 1s, C 1s, Fe 2p and P 2p. The C 1s
2
7

core line of catalysts before the catalytic tests (not shown here) is composed of three components,
at 285.0 eV (C-C structural bonds in organic contaminants), 286.3-287.0 eV (C-O groups) and
289.1-289.3 eV (C=O and carboxyl groups). After reaction, the C 1s spectra of both used
catalysts are characterized by several components (ranging from 283.6 to 291.8 eV) associated
with the reaction products. In contrast, the analysis of P 2p core excitation indicated no
qualitative differences between all catalysts, before and after reaction, and only single peak at BE
of 134.2 eV was found. It means that the PO tetrahedron originating from the center of the
4
Keggin anion remains stable independently on the changes in the surrounding octahedrons, which
is in line with previous results [16, 41, 50, 51].
The O 1s lines in the spectra of catalysts before tests can be decomposed into four
components (Table 2): lattice bridging oxygen Mo-O-Mo (530.5-531.4 eV), OH groups (531.7-
532.3 eV), non-bridging Mo=O bonds (much smaller peaks at 528.5-529.4 eV), adsorbed water
and/or oxygen of organic contaminants (BE > 532.6 eV). For comparison, the XPS measurement
was also performed on HPMo-cal sample as reference, in which the O 1s line is very similar to
these found in both FeHPMo-cal and FePMo-cal catalysts (Table 2). Used catalysts are
characterized by three components: lattice oxygen (530.7 eV), OH groups (531.8 eV) and oxygen
of organic contaminants (close to 533 eV).
Fig. 9 compares the Mo 3d XPS spectra of HPMo-cal sample and FePMo-cal catalysts before
and after the catalytic tests. One can find several different contributions in the Mo 3d spectra
obtained from the fitting procedure under only single assumption of relative intensity ratio of
3d5/2 and 3d3/2 lines fixed to 3:2. For clarity, we have denoted these contributions (Mo 3d
doublets) with A-E letters in order of BE increasing (Fig. 9 and Table 2). It is worth noting that
spin-orbit splitting of all doublets are equal to 3.13 eV, which matches well with the literature
2
8

values for Mo oxides [52,53]. Usually, the main complication with the interpretation of such data
arises from the assignment of different final states to the corresponding spectral components. In
order to solve this problem, we have measured additional XPS spectra of HPMo-cal, MoO
3
and
MoO samples. The main idea was to use these spectra as a reference to determine
2
unambiguously oxidation states of Mo ions in all investigated compounds. It should be noted that
we have applied experimentally justified fitting parameters (like lineshape, BE and FWHM of all
components) in the fitting procedure of heteropolyacid derived catalysts.
MoO3
D
HPMo
D
E
C
FePMo - cal
D
E
C
FePMo - cal used
D
E
C
A
B
MoO2
D
B
A
242
240
238
236
234
232
230
228
Binding Energy (eV)
Fig. 9. XPS spectra of Mo 3d core lines of ‘fresh’ HPMo-cal sample and of FePMo-cal catalyst
before and after reaction. For comparison, XPS spectra of Mo oxides are also added.
2
9