K- and Ca-promoted ferrosilicates for the gas-phase epoxidation of propylene with O2

Article abstract of DOI:10.1016/j.apcata.2017.03.031

In the propylene epoxidation reaction with Fe-SiO₂ catalysts the presence of iron oxide particles has a detrimental effect due to the total combustion of propylene on these iron species. Thus, the complete elimination of the iron oxide particles is presented as a preliminary strategy in order to increase the selectivity towards propylene oxide in iron-based catalysts. In this sense, a simple post-treatment of the catalysts with alkali or alkaline-earth elements (such as K or Ca, respectively) has proven effective in the total elimination of these iron oxide particles. Furthermore, the addition of K and Ca has modified the physico-chemical properties of the catalysts, decreasing their superficial acidity and (for higher K or Ca loadings) masking/blocking the active sites responsible for the catalytic reaction. With all this, it is shown that K has a higher efficiency removing the iron oxide particles compared with Ca (for the same molar ratios) and that a higher amount of K (compared to Fe) is required for the complete elimination of the iron oxide particles. A considerable propylene oxide selectivity enhancement (up to 65%) has been obtained for the K-promoted Fe_0.005SiO₂ and Fe_0.01SiO₂ catalysts using O₂ as sole oxidant.

Full text of DOI:10.1016/j.apcata.2017.03.031

Applied Catalysis A: General 538 (2017) 139–147
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
K- and Ca-promoted ferrosilicates for the gas-phase epoxidation of
propylene with O2
Jaime García-Aguilar, Diego Cazorla-Amorós, Ángel Berenguer-Murcia∗
Materials Science Institute and Inorganic Chemistry Department, Alicante University, Ap. 99, E-03080 Alicante, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
In the propylene epoxidation reaction with Fe-SiO2 catalysts the presence of iron oxide particles has a
detrimental effect due to the total combustion of propylene on these iron species. Thus, the complete
elimination of the iron oxide particles is presented as a preliminary strategy in order to increase the
selectivity towards propylene oxide in iron-based catalysts. In this sense, a simple post-treatment of the
catalysts with alkali or alkaline-earth elements (such as K or Ca, respectively) has proven effective in
the total elimination of these iron oxide particles. Furthermore, the addition of K and Ca has modified
the physico-chemical properties of the catalysts, decreasing their superficial acidity and (for higher K or
Ca loadings) masking/blocking the active sites responsible for the catalytic reaction. With all this, it is
shown that K has a higher efficiency removing the iron oxide particles compared with Ca (for the same
molar ratios) and that a higher amount of K (compared to Fe) is required for the complete elimination of
the iron oxide particles. A considerable propylene oxide selectivity enhancement (up to 65%) has been
obtained for the K-promoted Fe0.005SiO2 and Fe0.01SiO2 catalysts using O2 as sole oxidant.
© 2017 Elsevier B.V. All rights reserved.
Received 22 December 2016
Received in revised form 17 March 2017
Accepted 19 March 2017
Available online 23 March 2017
Keywords:
Sol-gel
Propylene epoxidation
Alkali promotion
Ferrosilicate
Dyoxygen
1. Introduction
H2. However, it is mandatory to increase the reaction temperature
(compared with the abovementioned Au/Ti-SiO2 catalysts which
The production of propylene oxide (PO) by gas-phase catalysis
is one of the most important challenges for the chemical companies
due to its commercial interest. PO is a precursor of polyether polyols
for urethanes, propylene glycols, and glycol ethers. At present, the
production of this compound is mainly performed in liquid phase
carried out by the chlorohydrin (in which Cl2 is used as oxidant)
process with approximately 40% of PO production [1–3].
requires H2/O2 mixtures) to activate the gas molecules and obtain
the desired compound. In this sense, Khatib et al. [3] have recently
reported a review summarizing the catalysts used for PO forma-
tion using gaseous O2. Pure Ru, Ag and Cu (or mixed with other
metals)-based catalysts must be mentioned as thoroughly studied
due to their similar or higher propylene conversion than Au/Ti-SiO2
catalysts but with slightly lower PO selectivity [10–18].
Performing the epoxidation of propylene by heterogeneous
catalysis in gas phase is nowadays one of the most interesting
research topics in order to obtain an industrial catalyst towards
a more environmentally friendly and applicable PO synthesis [4].
Despite the high price of gold and the need to use H2/O2 gas mix-
tures to feed the reactor, the Au/Ti-SiO2 based catalysts are still
under intense study due to high selectivities towards PO (at rel-
atively low temperatures). This has given rise to recent studies
in which the researchers have paid special interest to the reac-
tion mechanism using this type of catalysts [5–9]. Other studies
were focused on the synthesis of PO using only O2 and propy-
lene in the gas stream composition aimed at avoiding the use of
Recently, our research group reported the synthesis of PO using
dioxygen and ferrosilicate catalysts prepared in a sol-gel one-step
synthesis of well-dispersed iron oxide species into the silica struc-
ture [19]. In these materials, the propylene epoxidation reaction
took place on the iron sites dispersed into the silica framework
while the total combustion (towards CO2) was preferentially occur-
ring on the small iron oxide particles/clusters decreasing the PO
selectivity [20]. Due to the synthesis procedure, even when the
iron loading is 0.005 (Fe/Si molar ratio), the formation of extra-
framework small iron oxide particles cannot be avoided. Addition
of some alkali and alkaline earth elements after the synthesis
of the catalysts reduces the presence of iron oxide particles as
reported in the literature [21]. This strategy was also used in the
above-mentioned catalysts (Ag, Cu or Ru) to increase PO selectivity,
reaching the best improvements with the K and Ca-based catalysts
[11,22–24]. Even for Fe-SiO2 catalysts (where N2O is used as oxi-
∗
Corresponding author.
E-mail address: a.berenguer@ua.es (Á. Berenguer-Murcia).
http://dx.doi.org/10.1016/j.apcata.2017.03.031
0926-860X/© 2017 Elsevier B.V. All rights reserved.

140
J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
dant), this addition has beneficial effects in the selectivity towards
PO in spite of a reduction in propylene conversion [25–29].
This work is focused on the reduction and/or removal of the
small iron oxide particles/clusters observed in iron-based catalysts
(Fe0.0XSiO2) in order to improve their catalytic performance. In
order to increase the selectivity towards PO, the Fe0.0XSiO2 solids
were impregnated after their synthesis with the nitrate salts of
K and Ca to introduce the needed amount of the element. The
removal of the small iron oxide particles was determined by dif-
ferent spectroscopy techniques such as UV–Vis, UV-Raman or FTIR.
The promoted Fe0.0XSiO2 catalysts were tested in the propylene
epoxidation reaction using O2 as oxidant (avoiding the use of dan-
gerous and harmful agents such as N2O) with very promising results
in terms of PO selectivity.
2.2. Characterization
The catalysts were analyzed by Transmission Electron
Microscopy (TEM, JEOL JEM-2010) with a coupled EDX detec-
tor. The microscope operates at 200 kV with a space resolution
of 0.24 nm. For the TEM analysis, the sample was suspended in
ethanol and sonicated for a few minutes. The suspension (one
drop) was deposited onto a 300 mesh Lacey copper grid and left to
dry at room temperature. This technique allows the observation
of the iron particles in the catalysts and the detection of the
metal (by EDX) if it is not possible to see the particles. The iron,
potassium and calcium loadings were measured by inductively
coupled plasma-optical emission spectroscopy (ICP-OES), in a
Perkin–Elmer Optima 4300 system. A few milligrams of the cata-
lyst were dissolved in 0.1 ml of HF (5 vol. %) at room temperature,
in order to ensure the total dissolution of the catalyst, and then
diluted to the linear element detection range (0.05–10 ppm). The
untreated and K and Ca-treated iron based catalysts were also
characterized by N2 adsorption-desorption isotherm at −196 ◦C
and by CO2 adsorption at 0 ◦C (Quantachrome, Autosorb 6B) to
analyze the porous texture of this kind of materials.
2. Experimental
2.1. Catalyst preparation
2.1.1. Preparation of the Fe0.0XSiO2 catalysts
Some spectroscopic techniques were used to analyze the physic-
ochemical properties of the catalysts after K and Ca modification.
UV–Vis analysis (V-670, JASCO) equipped with a double monochro-
mator system, a photomultiplier tube detector and an integrating
sphere (ISN-723 UV–visible-NIR, JASCO) was used in transmittance
mode to determine the speciation of the iron. The incorporation of
the iron was analyzed by UV-Raman (NRS-5100, Jasco). The cat-
alysts were irradiated for 30 min with a UV laser (HeCd, 325 nm,
1 mW). The equipment is fitted with a Thorlabs 20× UV objective.
Fourier Transform Infrared Spectroscopy (FTIR-4100, JASCO) was
also used. The catalysts were diluted (using 1:50 wt. ratio) in KBr
and dried at 120 ◦C for 12 h before the analysis. Pyridine adsorption
was performed in order to characterize the modification of the sur-
face acidity using a reported procedure [30]. The catalysts, diluted
in KBr using the same weight ratio, were calcined at 500 ◦C. Then
they were exposed to pyridine vapours for 3 days in a vacuum des-
iccator. After pyridine adsorption, the physically adsorbed pyridine
was removed at 100 ◦C for one day. The FTIR spectra presented are
the result of the subtraction of the spectra of the pyridine-treated
to the spectra of the untreated catalysts.
Hierarchical Fe-doped silica catalysts were prepared following
the synthesis described in [19]. In a typical synthesis, 0.400 g of
surfactant (Pluronic® F127, Sigma-Aldrich), 0.452 g of urea (Sigma-
Aldrich) and 5.052 g of an acetic acid solution (0.01 M) were mixed
under vigorous stirring for 80 min, the final pH of the resulting solu-
tion being around 4. Then, the necessary amount of iron precursor
(iron (III) nitrate nonahydrate, Fe(NO3)3·9H2O, 99.99%, Sigma-
Aldrich) was added in the solution and the mixture was stirred
for 1 h. Subsequently, the solution was cooled in an ice-water bath
under stirring and the silica precursor was added dropwise (2.030 g
Tetramethyl orthosilicate, TMOS, Sigma-Aldrich). This solution was
kept under stirring for 40 min at 0 ◦C.
Finally, the sol was introduced in a Teflon autoclave and heated
at 40 ◦C for 20 h to produce the aging of the sol (the pH after this
step remained around 4). After this, the sample was submitted to
a hydrothermal treatment at 120 ◦C for 6 h, to decompose the urea
(the final pH of the supernatant liquid being around 9–10). As a
final step, the monolith was calcined at 550 ◦C for 6 h under static
atmosphere to eliminate the surfactant and the rest of unwanted
precursors.
Three samples with different Fe/Si molar ratios were prepared
Fe0.005SiO2, Fe0.01SiO2 and Fe0.02SiO2. The samples in this work
were named according to the nominal molar Fe/Si ratio in each case,
for example, Fe0.01SiO2 corresponds to the sample with 1 mol% Fe
in the oxide (with respect to Si moles).
2.3. Propylene epoxidation tests
The catalysts were tested under steady-state conditions at ambi-
ent pressure for at least 4 h. The standard temperature of the
analysis was 350 ◦C, but to obtain reliable catalytic results, in some
catalysts it was necessary to increase the temperature up to 400
or 450 ◦C. A WHSV of 10,000 ml g−1 h−1 and a gas composition
of 10% C3H6, 10% O2, 80% He was used in the catalytic tests. The
gas composition was analyzed with a GC chromatograph (Agilent
7820A) equipped with two columns, PoraBond Q (Agilent) and CTR-
I (Alltech), for the separation of the organic (propylene, propane,
propylene oxide, acetaldehyde, acetone) and the inorganic (mainly,
O2 and CO2) compounds respectively. Propylene conversion, PO
yield and PO selectivity were calculated using the following equa-
tions and with the respective calibration of each compound in order
to describe the catalytic behaviour of the samples.
2.1.2. Preparation of the K and Ca-promoted Fe0.0XSiO2 catalysts
Once the iron-based silica samples were calcined and milled, the
samples were impregnated in excess volume using aqueous solu-
tions containing the necessary amount of alkali (KNO3) and alkaline
earth (Ca(NO3)2·4H2O) precursors. The stirring was maintained for
2 days, and then, the solvent was evaporated under stirring, heating
the suspension at 80 ◦C keeping the magnetic stirring until com-
plete evaporation. Finally, the powder was calcined again in an oven
at 550 ◦C for 6 h. The samples are labelled as K0.01, for the sample
with a K/Si molar ratio of 0.01, before the Fe0.0XSiO2. For compari-
son purposes, Fe-free K and Ca-treated solids (with a molar ratio of
0.01) were prepared using pure silica and impregnating the alkali
and alkaline earth salts as described above. In addition, a Fe0.005SiO2
sample was treated with a 1 M HCl solution overnight, filtered the
solid, washed until neutral pH and then dried at 100 ◦C overnight.
After the acidic treatment, a portion of the catalyst was treated with
KNO3 (using the necessary amount for a K molar ratio of 0.01) and
calcined as described above.
CC −in − CC
H
−out
H
3
6
3
6
Propylene Conversion(%) =
× 100
CC
H
−in
3
6
CPO−out
PO Yield(%) =
× 100
CC
H
−in
3
6

J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
141
oxide clusters and/or extraction of the octahedrally coordinated
iron species by the K treatment. These effects are observed also in
the progressive increase of the Eg values obtained from the UV–Vis
spectra, from the initial 3.91 eV for the raw Fe0.01SiO2 to the final
4.46 eV for the K0.02Fe0.01SiO2 catalyst. In the case of Ca-based cat-
alysts, the same trend is observed but it is not possible to remove
all the particles of iron oxide produced during the synthesis of the
material. All the samples, even the Ca0.02Fe0.01SiO2, present a small
but significant absorption at wavelengths higher than 300 nm and
the increase of the Eg values is lower in the Ca samples than their
K-based counterparts. The differences (in UV–Vis characterization)
between the K and Ca-doped samples, when the small iron oxide
particles and octahedral iron species removal is studied, allows us
to confirm that for this kind of samples impregnation with K has
a higher efficiency in reducing the number of iron oxide particles,
as other authors have observed previously [27–29]. For this reason,
the impregnation of the Fe0.005SiO2 and Fe0.02SiO2 samples was
only studied with K.
The UV–Vis results of the K-impregnated catalysts in the
Fe0.005SiO2 and Fe0.02SiO2 materials are presented in Fig. 1C and
D, respectively. As it was previously studied, for all the raw
catalysts, the amount of small iron oxide particles/clusters in
extra-framework positions increases with the Fe loading of the
samples [19]. A progressive reduction in the absorption for the
K0.0XFe0.005SiO2 series at 280 nm and above 300 nm (correspond-
ing to the aforementioned species) is observed as the K content is
increased in the catalysts. In the K0.02Fe0.01SiO2 catalyst, when the
K content is higher than the Fe molar content in the silica based
catalysts (as with samples K0.01Fe0.005SiO2 and K0.02Fe0.005SiO2),
an almost total quenching of the absorption at wavelengths above
300 nm is observed, demonstrating the total removal of the iron
oxide particles formed in the raw catalysts synthesis. The removal
of the iron oxide particles produces the increase of the band gap
energy up to 4.50 and 4.52 eV in the case of samples K0.01Fe0.005SiO2
and K0.02Fe0.005SiO2, respectively. Finally, in the catalysts prepared
with the highest iron content (Fe0.02SiO2) the addition of K has
a different result, as it is possible to check in Fig. 1D. Even with
the increase of the Eg up to 4.36 eV for the sample K0.02Fe0.02SiO2
mainly due to the reduction in the absorption band located at
280 nm, there is a slight increase of the UV–Vis absorption in the
300–400 nm range. It is possible that the amount of iron oxide
particles is too high for the K to remove all of them. All these
UV–Vis results are corroborated by the colour of the catalysts and
by the TEM micrographs presented in the SI. As K or Ca loading
increases in the catalysts, the initial colour of the Fe-SiO2 (slightly
brown for Fe0.01SiO2 and Fe0.02SiO2) gradually disappears until
total decolouration of the catalysts for all the Fe0.005SiO2 samples
and the K0.02Fe0.01SiO2.
CPO−out
PO Selectivity(%) =
× 100
CC
− CC
H
−out
H
−in
3
6
3
6
3. Results and discussion
3.1. Characterization of the catalysts
For all the catalysts prepared in this work, the metal loadings
were analyzed by ICP-OES. As in the previous work [19], it was
observed that in the one-step preparation of the catalysts, not all
the iron can be incorporated into the silica framework. The real Fe/Si
molar ratios for the prepared catalysts were 0.0043 (Fe0.005SiO2),
0.0095 (Fe0.01SiO2), and 0.0186 (Fe0.02SiO2). From our observations,
approximately 7% of the iron was removed in the supernatant liquid
at the end of the sol-gel process. On the contrary and due to the
methodology employed for their addition, the real loadings for K
and Ca calculated by ICP-OES were in all cases quite close to the
desired nominal loadings.
TEM coupled with EDX characterization of the representative
catalysts is presented in the Supplementary Information. From the
TEM micrographs it is possible to see that no large iron oxide
clusters/agglomerations are formed in most samples. Only for
some samples in which the alkali loading is lower than that of Fe
(K0.005Fe0.01SiO2 and K0.005Fe0.02SiO2) some small particles can be
observed. Despite the fact that it is not possible to see the possi-
ble K, Ca or Fe oxide species on the silica surface, the EDX analysis
confirmed, in all cases, the presence of these elements. From the
textural characterization of the samples (see Fig. S1 and Table S1
in the Supplementary Information) it is not possible to observe any
significant modification in the N2 isotherm shape or the textural
properties (i.e., specific surface area, micropore volume or pore size
distribution) after the treatment of the Fe-SiO2 samples with K or
Ca.
The spectroscopic techniques, such as UV–Vis, UV-Raman, and
FTIR, are very useful to analyze the iron speciation in this kind of
materials. Studying the UV–Vis absorption of the solid catalysts it
is possible to determine the presence of iron incorporated into the
silica framework and the existence of small particles of iron oxide
deposited on the surface of the silica. In this sense, previous works
reported the absorption of the incorporated iron in tetrahedral or
pseudo-tetrahedral position into the silica framework in two bands
located at around 220 and 260 nm (45,000 and 38,000 cm−1, respec-
tively). While the iron oxide particles present UV–Vis absorption
near the 370 and 500 nm range (27,000 and 20,000 cm−1, respec-
tively). This effect allows the interpretation of the band gap energy
(Eg) as a measure of the iron dispersion in this kind of materials
[29,31]. It is very important to remark that some works attribute the
absorption at around 280 nm to the octahedral coordinated Fe(III)
species in this kind of materials [32–34]. The UV–Vis results (and
each corresponding Eg values) of the K and Ca-promoted Fe-SiO2
catalysts are presented in Fig. 1.
In order to select the appropriate promoter (K or Ca) which can
yield a better removal of the iron particles and dispersion of the
iron species, the Fe0.01SiO2 catalyst was impregnated with the 3
different K or Ca/Si ratios (0.005, 0.01 and 0.02) and analyzed by
UV–Vis reflectance (results presented in Fig. 1A and B). For the K-
based catalysts, a substantial decrease in absorbance is observed for
wavelengths above 300 nm. This can be associated to the removal
of the iron oxide particles deposited on the silica surface. The total
elimination of the iron oxide particles is reached for all samples
in which the K content exceeded that of Fe. On the other hand,
a reduction in the absorbance of the UV–Vis spectra near 280 nm
after the addition of the alkali or alkaline earth based salts is also
observed. This fact can be related to the removal of the small iron
Another complementary technique that can be used to analyze
the structural modifications produced by the K and Ca addition
is UV-Raman spectroscopy. The results of the prepared catalysts
together with the blank silica and the K- and Ca-impregnated sil-
ica are shown in Fig. 2. Different regions of the spectra can be
clearly distinguished depending on the species involved. The sym-
metric Si
O
Si stretching appears at 485 cm−1 for pure silica [35].
Close to this band, a broad shoulder appears from 500 to 650 cm−1
,
which we previously identified in bulk Fe2O3 and appears as a broad
shoulder in samples Fe0.01SiO2 and Fe0.02SiO2 [19]. As it was previ-
ously reported, in the 800 cm−1 region for the pure SiO2 sample it
is possible to observe a low-intensity band that can be attributed
to the vibration mode of siloxane linkage [25] or to the presence of
defects [27]. In the 950–1150 cm−1 range three characteristic bands
are observed in the SiO2 and Fe-SiO2 samples corresponding to dif-
ferent vibrations; 990 (Si
O Si near iron species or other defect
sites such as silanol groups) [25,27,35], 1080 (1016 cm−1 in [35],
1025 cm−1 in [36], 1074 cm−1 in [27] and 1090 cm−1 in [25]) and

142
J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
A
C
B
Fe0.01SiO2 (3.91 eV)
Fe0.01SiO2 (3.91 eV)
K0.005Fe0.01SiO2 (4.36 eV)
Ca0.005Fe0.01SiO2 (4.15 eV)
Ca0.01Fe0.01SiO2 (4.23 eV)
Ca0.02Fe0.01SiO2 (4.29 eV)
K
0.01Fe0.01SiO2 (4.39 eV)
K0.02Fe0.01SiO2 (4.46 eV)
200
250
300
400
450
500 200
250
300
400
450
500
Wavele3n5g0th (nm)
Wavele3n5g0th (nm)
D
Fe0.005SiO2 (3.97 eV)
Fe0.02SiO2 (3.90 eV)
K0.005Fe0.005SiO2 (4.43 eV)
K0.005Fe0.02SiO2 (4.27 eV)
K0.01Fe0.005SiO2 (4.50 eV)
K0.01Fe0.02SiO2 (4.29 eV)
K0.02Fe0.005SiO2 (4.52 eV)
K0.02Fe0.02SiO2 (4.36 eV)
200
250
300
400
450
500 200
250
300
400
450
500
Wavele3n5g0th (nm)
Wavele3n5g0th (nm)
Fig. 1. Solid UV–Vis reflectance spectra (presented in Kubelka-Munk units) and their respective Eg value for the raw and post-K or Ca impregnation of the Fe0.0XSiO2 catalyst.
A) K0.0XFe0.01SiO2, B) Ca0.0XFe0.01SiO2, C) K0.0XFe0.005SiO2 and D) K0.0XFe0.02SiO2.
1135 cm−1 (1115 cm−1 in [35], 1138 cm−1 in [27] and 1150 cm−1 in
[36], not presented in Ref. [25] maybe due to the use of a 244 nm
laser irradiation), corresponding to vibrations of O3SiO− units with
the samples with the highest K loading (0.02), especially for the
Fe0.005SiO2 sample. This fact might be due to the excess potassium
used in the impregnation producing potassium oxide (as observed
for the K-SiO2 sample). This was also observed for the Ca-containing
sample.
a very strong interaction with heteroatoms such as Fe
O Si when
the Fe is well-dispersed in isolated tetrahedral or pseudotetrahe-
dral iron into the silica framework [19,25–27,36].
In all the studied series presented in Fig. 2, deposition of the
K or Ca salt on the Fe-SiO2 solids does not produce any impor-
tant change in the band located at 485 cm−1 corresponding to
The assignation of each band within the 990–1135 cm−1 region
might be done with existing literature. The band at 1165 cm−1 has
been assigned to the crystalline environment of the iron species
in Fe-ZSM-5 samples [35,37]. In our case, given the amorphous
nature of our sample, this assignment is not straightforward. When
K or Ca are loaded on the Fe-SiO2, a decrease in intensity is
observed for the 1080 and 1135 cm−1 bands (see Fig. 2) together
with the appearance of a band located at 1000 cm−1 (1010 cm−1
for the K0.02Fe0.01SiO2 and K0.02Fe0.005SiO2 and 995 cm−1 for the
Ca0.02Fe0.01SiO2). This new band was previously studied by Wang
et al. [26,27] and could correspond to a tetrahedrally coordinated
iron oxide “surrounded and stabilized by K+ or Na+” (K+ or Ca2+
in this study) species. These K- or Ca-Fe species (such as ferrates
[26,27]) can be formed during the calcination step after impregna-
tion of the K or Ca salt, produced by its reactivity with the small
particles/clusters of iron oxide on the silica surface. This hypothe-
sis is also supported by the results showed in the UV–Vis analysis,
where there was a reduction of the absorption corresponding to
the small particles/clusters of iron oxide. The formation of these
superficial species (K- and Ca-Fe species) with the well-dispersed
iron into the silica framework, which occupies pseudotetrahedral
positions cannot be ruled out in this respect. It is possible that the
iron could migrate to extra-framework positions or could be cov-
the symmetric Si
O Si stretching. This band is slightly shifted
to higher wavenumbers due to the incorporation of iron into the
silica framework [19,35] and is not affected by the deposition of
the K and Ca oxides on the silica (K-SiO2 and Ca-SiO2) indicating
that post-impregnation does not produce modifications in the silica
framework and the oxides are only deposited on the silica surface.
Important changes in the UV-Raman spectra are observed after the
modification of the iron-based catalysts. In the raw Fe0.01SiO2, a
shoulder is observed from 500 to 650 cm−1 (Fig. 2A and B) which is
not observed for the SiO2 and K or Ca-treated materials. As the K or
Ca loading increases, this band reduces its intensity down to values
similar to the pure silica solid. It is possible to assign this shoulder
to the small iron oxide clusters/particles or iron in octahedral coor-
dination in the original Fe-SiO2 materials which are dispersed and
removed after K or Ca impregnation [27–29].
In the 800 cm−1 region, a weak signal is observed for the pure
silica and the raw Fe-SiO2 samples. However, when the K or Ca
oxides are loaded on the silica surface the band becomes more
intense and its shape is modified. The same band is obtained for

J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
143
485 cm-1
A
B
1010 cm-1
990 cm-1
1080 cm-1
1135 cm-1
K-SiO2
K0.02Fe0.01SiO2
K0.005Fe0.01SiO2
Fe0.01SiO2
SiO2
9R50aman shift (cm-1) 550
1350
1150
750
350
150
485 cm-1
1080 cm-1
1135 cm-1
995 cm-1
990 cm-1
Ca-SiO2
Ca0.02Fe0.01SiO2
Ca0.005Fe0.01SiO2
Fe0.01SiO2
SiO2
1350
1150
750
550
350
150
9R50aman shift (cm-1)
485 cm-1
C
1010 cm-1
990 cm-1
1080 cm-1
1135 cm-1
K-SiO2
K0.02Fe0.005SiO2
K0.005Fe0.005SiO2
Fe0.005SiO2
SiO2
1350
1150
750
350
150
9R50aman shift (cm-1) 550
Fig. 2. Normalized UV-Raman analysis of representative catalysts prepared in this work. A) K0.0XFe0.01SiO2, B) Ca0.0XFe0.01SiO2 and C) K0.0XFe0.005SiO2 series compared with
the pure SiO2 and its respective K or Ca-SiO2 excited with a UV laser of 325 nm.
ered (as was proposed by Horváth et al. [20]) by the alkali or alkaline
earth oxide producing the decrease of its related bands at 1080 and
(Fig. 3) due to the K or Ca impregnation on the silica surface. When
K or Ca are added with the lowest amounts (i.e., M0.005Fe0.0XSiO2),
there is a reduction in the intensity of the band corresponding
to the bridged hydroxyls with Brønsted acid character (located at
3660 cm−1) [36,39]. From our observations this reduction is not
proportional to the K or Ca loading. This effect is also demon-
strated by the normalized FTIR spectra corresponding to pyridine
adsorption presented in Fig. S2 (See Supplementary Information).
By this analysis it is possible to observe a proportional decrease
in the intensity of the bands corresponding to adsorbed pyridine
at 1446 and 1596 cm−1 [40] as the K molar ratio increases (for
the Fe0.005SiO2 catalyst) or when the K or Ca are added on the
Fe0.001SiO2. In this sense, the addition of ions with a strong basic
character on the silica surface reduces the amount of acid sites in
the solid due to their exchange with the acidic proton neighbour
to the iron atom. This fact could be beneficial for the epoxidation
reaction due to the acidity reduction of the support. It is also possi-
1135 cm−1
.
In order to determine the possible contribution of the K and
Ca addition to other physicochemical properties of the materials,
the normalized FTIR spectra of all the solids were analyzed and
presented in Fig. 3.
In Fig. 3 it is possible to compare the normalized (with respect
to the most intense band at 1090 cm−1, corresponding to the Si
O
stretching) characteristic vibrations of the superficial OH groups
of this kind of materials. No significant modification of the fun-
damental framework vibration modes of the Fe-SiO2 solids (810,
1000–1300 cm−1) [19,36,38] is observed after the impregnation
and calcination of the K or Ca-based salts. This fact reveals that
the impregnation of the K or Ca nitrates only produces superfi-
cial modifications without affecting the Fe0.0X-SiO2 framework as
it is previously discussed by UV-Raman spectroscopy. On the other
hand, a significant modification in the OH vibration range occurs

144
J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
Fe0.01SiO2
Fe0.01SiO2
3660 cm-1
K0.005Fe0.01SiO2
K0.01Fe0.01SiO2
K0.02Fe0.01SiO2
Ca0.005Fe0.01SiO2
Ca0.01Fe0.01SiO2
Ca0.02Fe0.01SiO2
A
B
0.05
3600
0.05
3800
3400
3200
3000 3800
3600
3400
3200
3000
Wavenumber (cm-1)
Wavenumber (cm-1)
Fe0.02SiO2
C
D
K0.005Fe0.02SiO2
K0.01Fe0.02SiO2
K0.02Fe0.02SiO2
Fe0.005SiO2
K0.005Fe0.005SiO2
K0.01Fe0.005SiO2
K0.02Fe0.005SiO2
0.02
0.04
3800
3600
3400
3200
3000
3800
3600
3400
3200
3000
Wavenumber (cm-1)
Wavenumber (cm-1)
Fig. 3. Normalized FTIR spectra of the OH region (presented in Kubelka-Munk units) for all the catalysts prepared in this work. A) K0.0XFe0.01SiO2, B) Ca0.0XFe0.01SiO2, C)
K0.0XFe0.005SiO2 and D) K0.0XFe0.02SiO2.
ble to propose the blocking/covering of some iron sites of the silica
framework due to the K or Ca oxide formation on the silica surface.
In this respect, the impregnation and calcination of the K or Ca
salts produces the corresponding oxides deposited on the silica sur-
face. For low loadings, the K or Ca ions interact with the acidic
proton reducing its signal in all the catalysts (M0.005Fe0.0XSiO2).
Then, as the K or Ca loading increases, removal of the small clus-
ters/particles (lower than 1 nm) takes place in the Fe-SiO2 samples,
as observed by UV–Vis analysis [27–29]. This removal reduces the
amount of the small iron oxide clusters/particles (responsible for
the total oxidation of propylene) [20]. As it is mentioned above, this
alkali or alkaline earth oxide can also cover the highly dispersed iron
into the silica framework (which is the main species responsible for
the propylene epoxidation reaction [19]).
As it is discussed in the spectroscopic techniques section, the
addition of small amounts of K or Ca is a simple and useful way
to reduce the amount or to remove the undesired iron oxide parti-
cles/clusters formed during the synthesis of the catalysts. However,
this modification can also produce other different effects, like
reducing the amount of bridged hydroxyls with Brønsted acid
character located near the iron atom or blocking iron active sites,
which is very interesting for this reaction due to their respective
interaction with gaseous propylene or dioxygen in the propylene
epoxidation reaction.
significant value of propylene conversion) during at least 4 h. How-
ever, the catalytic tests performed at 450 ◦C for the Fe0.005SiO2 and
Fe0.01SiO2 resulted in a fast deactivation of the catalysts due to car-
bon deposition on the surface. This effect was also observed in the
Fe0.02SiO2 sample at 400 ◦C. In all catalytic tests, except for those
just mentioned, propylene conversion, PO yield, and PO selectivity
remained constant during the reaction, and the values presented
in this section correspond to the steady-state conditions. Analyz-
ing all reaction products, it was possible to find out that the main
by-product formed was CO2, but in some samples, acetaldehyde
was also produced to a small extent (<1%). The results are shown
in Table 1.
The addition of the small amounts (only 0.005 molar ratios) of
promoters (K or Ca) produces in the Fe0.01SiO2 catalyst the decrease
of the propylene conversion down to half of its initial activity. How-
ever, the PO yield is also halved since no significant modification
of the PO selectivity is produced. Analyzing the addition of a larger
amount of K on the Fe0.01SiO2 catalyst it is possible to assess how the
incorporation of an equimolar ratio of K/Fe (K0.01Fe0.01SiO2) pro-
duces a further reduction of both C3H6 conversion and PO yield also
corresponding to a similar PO selectivity compared to the Fe0.01SiO2
sample. Nonetheless, when the K ratio is twice with respect to
the Fe amount (K0.02Fe0.01SiO2) the C3H6 conversion is too low
at 350 ◦C and a temperature increase to 400 ◦C is mandatory to
obtain a significant conversion/yield value. At this new tempera-
ture, the C3H6 conversion is the lowest of this catalyst series (1.7%)
but the PO yield is not reduced compared with the K0.005Fe0.01SiO2
and K0.01Fe0.01SiO2 catalysts. This results in a drastic increase of
the PO selectivity from 24.6–27.5% to 62.4%. The catalytic activity
3.2. Propylene epoxidation reaction
The samples were tested under isothermal conditions (at 350,
400 or 450 ◦C depending on the activity of the sample to ensure a

J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
145
Table 1
Catalytic performance of the samples prepared in propylene epoxidation by O2 molecule under steady-state conditions.
Catalyst
Temperature (◦C)
C3H6 Conversion (%)
PO Yield (%)
Selectivity (%)
PO
Othersa
CO2
Fe0.01SiO2
350
400
350
350
400
350
350
350
350
400
350
450
450
350
350
350
400
7.8
21.6
4.0
3.4
1.7
3.2
1.7
1.5
5.5
19.1
2.0
2.2
1.4
10.3
4.9
2.5
8.5
1.9
4.4
1.1
0.9
1.1
0.9
0.8
0.8
1.8
5.1
0.6
1.3
0.9
2.4
1.3
0.8
2.3
24.6
20.4
26.0
27.5
62.4
27.6
51.4
53.3
33.6
26.7
33.8
58.8
65.5
23.1
25.7
31.5
27.2
<0.5
<1.0
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<1.0
<0.5
<0.5
<0.5
<1.0
<0.5
<0.5
<1.0
∼75
∼78
∼74
∼72
∼37
∼72
∼48
∼46
∼66
∼72
∼66
∼41
∼34
∼76
∼74
∼68
∼63
K
K
K
0.005Fe0.01SiO2
0.01Fe0.01SiO2
0.02Fe0.01SiO2
Ca0.005Fe0.01SiO2
Ca0.01Fe0.01SiO2
Ca0.02Fe0.01SiO2
Fe0.005SiO2
K
K
K
0.005Fe0.005SiO2
0.01Fe0.005SiO2
0.02Fe0.005SiO2
Fe0.02SiO2
K0.005Fe0.02SiO2
K0.01Fe0.02SiO2
K0.02Fe0.02SiO2
a
Acetaldehyde is the main organic by-product obtained in the catalytic reaction.
corresponding to the non-promoted Fe0.01SiO2 at 400 ◦C shows a
quite high propylene conversion (21.6%) and very low PO selectiv-
ity (20.4%). In this sense, the addition of 0.02 molar ratio of K allows
to control the propylene combustion at 400 ◦C and to increase the
PO selectivity.
Another important aspect that must be taken into account is the
aforementioned carbon deposit produced during the reaction due
to the propylene cracking on the catalysts surface. For the catalysts
without K or Ca, a small fraction of carbon deposit is observed after
the reaction (less than a 3 wt. % determined by thermogravimetry).
However, when K or Ca is added to the Fe-SiO2 no carbonaceous
deposit is obtained after the epoxidation reaction. Only for the
K0.005Fe0.02SiO2 and Ca0.005Fe0.01SiO2 catalysts a slight deposit of
carbon material is observed that results in a slight darkening of the
materials. In this sense, the role of the alkali or alkaline earth pro-
motion is very important because it allows to control the reactivity,
decreasing the propylene conversion and its combustion and avoid-
ing propylene cracking (and subsequent carbonaceous deposit) on
the catalysts.
On the other hand, the catalysts prepared following the HCl
treatment in order to remove the iron oxide particles present a dif-
ferent catalytic behaviour for the propylene epoxidation reaction.
It must previously mentioned that the HCl treatment (Fe0.005SiO2-
HCl) was not enough for the complete elimination of the iron oxide
particles (see Fig. S3 in Supplementary Information) but removes a
significant portion of them. After the acidic treatment, the results
from the propylene epoxidation reaction at 450 ◦C (lower temper-
atures did not yield reliable results) for the Fe0.005SiO2-HCl catalyst
were: 5.8% of propylene conversion, 2% of PO yield and a PO selec-
tivity of 34.5% (with a 6 wt. % of carbonaceous deposit). These
results mean that a large portion of the iron oxide particles were
removed (compared with the 19.1% of propylene conversion for
the raw Fe0.005SiO2) but there is no significant increase in the PO
selectivity. Moreover, the high amount of carbonaceous deposit
on the catalysts surface could be attributed to the acidity of the
surface which favours propylene reactivity and cracking. For this
reason, the propylene conversion of this sample is higher than the
K-promoted Fe0.005SiO2. This catalyst was also treated with K (with
a K molar ratio of 0.01) for its comparison with the K-promoted
without the acidic treatment. As revealed by the UV–Vis analysis of
the solid in Fig. S3, the absorbance of this catalyst is quite similar to
those obtained for the K0.01Fe0.005SiO2 and K0.02Fe0.005SiO2 without
the HCl treatment. The resulting KNO3-treated catalyst presents a
propylene conversion and a PO yield of 3.3 and 1.5%, respectively
with a PO selectivity of 45%. However, no significant carbonaceous
deposit is evidenced after the epoxidation reaction at 450 ◦C. In this
sense, it must be mentioned that it was not possible to achieve the
On the other hand, in the Ca-promoted Fe0.01SiO2 series a dif-
ferent behaviour is observed when the Ca amount increases. The
addition of 0.005 Ca/Si molar ratio produces the reduction of
the propylene conversion and PO yield, more than half in both
cases without any significant modification in the PO selectivity.
However, when the Ca amount increases up to 0.01 and 0.02
(Ca0.01Fe0.01SiO2 and Ca0.02Fe0.01SiO2, respectively) a similar value
of the C3H6 conversion and PO yield is observed (1.6% and 0.8%,
respectively) obtaining PO selectivities around 52% without requir-
ing a temperature increase. Analyzing the catalytic results of the
Fe0.01SiO2 catalyst promoted with K or Ca it can be concluded that
K-promotion produces higher selectivities towards PO despite of
the higher temperatures required to reach significant propylene
conversions.
For the Fe0.005SiO2 catalyst, the addition of the equimolar K/Fe
ratio (K0.005Fe0.005SiO2) produces the lessening of the C3H6 con-
version and PO yield to 2.0% and 0.6%, respectively, with respect to
the unpromoted catalysts. When the amount of potassium added
after the synthesis of the Fe0.005SiO2 increases up to 0.01 and 0.02
(K0.01Fe0.005SiO2 and K0.02Fe0.005SiO2, respectively) it is necessary
to increase the reaction temperature up to 450 ◦C to obtain signifi-
cant conversion values. The temperature increase compensates the
reduction in C3H6 conversion (which is not measurable at 350 ◦C),
obtaining values around 2.2 and 1.4% for the K0.01Fe0.005SiO2 and
K0.02Fe0.005SiO2 samples, respectively. In addition, the PO yield
obtained in these catalytic tests is remarkably higher, reaching
PO selectivities between 58.8% and 65.5% (under the same condi-
tions, the unpromoted Fe0.005SiO2 catalyst gave 19.1% and 26.7% for
propylene conversion and PO selectivity, respectively).
The catalysts prepared with the Fe0.02SiO2 sample present a dif-
ferent catalytic behaviour (in terms of PO selectivity) compared
with the lower iron content. As it is possible to see in Table 1, as the
K loading increases, the propylene conversion decreases to 2.5%
for the K0.01Fe0.02SiO2 catalyst with only a slight increase of the
PO selectivity (31.5%). However, when the temperature reaction
is increased at 400 ◦C for the K0.02Fe0.02SiO2 catalyst the propy-
lene conversion increases up to 8.5%, diminishing the selectivity
towards PO (27.2%).

146
J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
Scheme 1. Possible interpretation of the K effects on the Fe-SiO2 surface and its influence on the PO selectivity.
selectivities of the K0.01Fe0.005SiO2 and K0.02Fe0.005SiO2 catalysts
but the propylene conversion and PO yield were not so low.
The selectivities towards PO obtained for the K0.02Fe0.01SiO2,
K0.01Fe0.005SiO2 and K0.02Fe0.005SiO2 catalysts at 400 or 450 ◦C pre-
sented in Table 1 (reaching values over 60%) are higher, but lower in
propylene conversion (4–5%), than other promoted catalysts tested
using O2 as oxidant and based on transition metals such as Ag, Cu
and/or Ru [11,22,41,42] and similar to other alkali promoted Cu
[14,43] or Ag and Mo [44] catalysts.
Comparing the catalytic performance of our samples with sim-
ilar catalysts (alkali or alkaline earth promoted iron based silica)
reported in other studies, similar [26,27,29,45] or slightly lower
[20,34] PO selectivities were obtained with similar values of propy-
lene conversion in these cases. However, in these works the use
of nitrous oxide was mandatory to produce the epoxidation of the
propylene. In this sense, the replacement of nitrous oxide by molec-
ular oxygen implies user-friendly reaction conditions.
According to the catalytic results and the characterization
obtained for all the K- or Ca-promoted catalysts prepared in this
work it is possible to propose which are the main changes produced
by K or Ca in the iron species of the catalysts (Scheme 1). For the
lowest K and Ca loadings, an important aspect that must be taken
into account is the reduction in the PO yield but with similar PO
selectivities. As it was previously studied, the well-dispersed iron
species incorporated into the silica framework (interacting with
the molecular O2) along with a neighbour acidic proton (interact-
ing with gaseous propylene) are the main responsible species of
the propylene epoxidation reaction in this kind of catalysts [19].
So the addition of this kind of promoters, even in the 0.005 molar
ratios, is producing some changes in the well-dispersed iron species
that diminishes the PO production. This catalytic behaviour can
be explained trough the FTIR and UV-Raman results. As the FTIR
results shown, the addition of K or Ca modifies the surface prop-
erties of the silica. A significant reduction in the amount of acidic
protons with Brønsted character located at 3660 cm−1 is observed
after the addition of the smallest K or Ca molar ratio. In this sense,
its intensity decrease could be due to the substitution of the acidic
proton by the K or Ca species during the thermal treatment after
impregnation. This observation together with the PO yield decrease
(see Table 1) suggests that there is a reduction of the number of
sites that can interact with the propylene towards the epoxidation
reaction. However, for these catalysts (with low K or Ca loadings)
it is not possible to confirm the complete removal of the small
iron oxide particles by either UV–Vis or UV-Raman (due to the
formation of stabilized K- or Ca-Fe species). These particles are
responsible for the total combustion of the propylene. From our
observations, a graphical depiction of these catalysts is presented
in Scheme 1 (upper right-hand side), where the number of acidic
sites is reduced. The substitution of the acidic Brønsted proton by
K or Ca oxide also produces a significant decrease of the propy-
lene cracking on the catalyst surface thus resulting in lower carbon
deposition during the epoxidation experiments.
On the other hand, when the K or Ca loading increases different
interactions can be produced on the Fe-SiO2 surface. The picture
in the middle right-hand side of Scheme 1 shows how as the K
content is increased K-Fe species are formed in addition to the sub-
stitution of the protons in the bridged hydroxyls with Brønsted
acid character by K species. As it is commented by the UV–Vis
characterization, the addition of the K precursor produces the dis-
persion of the iron oxide particles and the octahedral coordinated
iron species. This fact is also supported by UV-Raman analysis, by
the emergence of the band at 1010 cm−1 assigned to tetrahedrally
coordinated iron oxide surrounded and stabilized by K+. In this
sense, the reduction of the small particles/clusters of iron oxide
determined by UV–Vis and TEM (which are the main responsi-
ble species for the total oxidation of the propylene towards CO2)
reduces the propylene conversion of the catalysts. Due to the fact
that propylene epoxidation mainly occurs on the well-dispersed
iron incorporated into the silica structure, the iron oxide particles
removal produces an increase of the PO selectivity as a beneficial
side effect. This behaviour is obtained for all the catalysts, where the
propylene conversion progressively decreases when small loadings
of K or Ca species are added. However, in order to achieve the com-
plete elimination of the iron oxide particles, a higher amount of K
(higher than the nominal Fe loading) is required. This is observed
for samples Fe0.005SiO2 (K0.01 and K0.02) and Fe0.01SiO2 (only for the
K0.02 catalyst) which show similar Eg values around 4.50 eV. Using
this kind of samples the propylene conversion and the PO selectiv-
ities are around 1.5% and 60%, respectively at their corresponding
reaction temperature (see Table 1). The addition of Ca as a pro-
moter does not produce the same efficiency (for the same promoter
molar ratios) in removing the iron oxide particles (as confirmed by
UV–Vis) despite the reaction temperature being 350 ◦C in all cases.
On the other hand, for the sample with the highest amount of iron
oxide particles (Fe0.02SiO2), it has not been possible to remove all
the iron clusters with the tested alkali molar ratio. By this way, the
catalytic results obtained for this sample show how the propylene
conversion (up to 8.5% when the temperature is increased) is not
as affected as in the other series.
Thus, the addition of promoters (such as K or Ca) reduces
the number of small iron oxide particles due to the formation of
K- or Ca-Fe species during the calcination step after the nitrate
impregnation. When the K molar ratio is higher than that of Fe
(K0.02Fe0.01SiO2, K0.01Fe0.005SiO2, and K0.02Fe0.005SiO2) the com-

J. García-Aguilar et al. / Applied Catalysis A: General 538 (2017) 139–147
147
plete elimination of the iron oxide particles is possible. Under these
conditions very promising PO selectivities (higher than 60%) by the
oxidation of propylene with O2 are obtained.
[7] X.-Y. Chen, S. Chen, A. Jia, J. Lu, W. Huang, Appl. Surf. Sci. 393 (2017) 11–22.
[8] A. Prieto, M. Palomino, U. Díaz, A. Corma, Appl. Catal. A: Gen. 523 (2016)
73–84.
[9] J. Chen, E.A. Pidko, V.V. Ordomsky, T. Verhoeven, E.J.M. Hensen, J.C. Schouten,
T.A. Nijhuis, Catal. Sci. Technol. 3 (2013) 3042–3055.
[10] D.O. Atmaca, D. Düzenli, M.O. Ozbek, I. Onal, Appl. Surf. Sci. 385 (2016)
99–105.
[11] P. Phon-in, A. Seubsai, T. Chukeaw, K. Charoen, W. Donphai, P. Prapainainar,
M. Chareonpanich, D. Noon, B. Zohour, S. Senkan, Catal. Commun. 86 (2016)
143–147.
[12] S. Ghosh, S.S. Acharyya, R. Tiwari, B. Sarkar, R.K. Singha, C. Pendem, T. Sasaki,
R. Bal, ACS Catal. 4 (2014) 2169–2174.
[13] Q. Hua, T. Cao, X.-K. Gu, J. Lu, Z. Jiang, X. Pan, L. Luo, W.-X. Li, W. Huang,
Angew. Chem. 53 (2014) 4856–4861.
[14] J. He, Q. Zhai, Q. Zhang, W. Deng, Y. Wang, J. Catal. 299 (2013) 53–66.
[15] W. Long, Q. Zhai, J. He, Q. Zhang, W. Deng, Y. Wang, ChemPlusChem 77 (2012)
27–30.
Finally, another undesirable scenario which cannot be ruled out
when the promoter molar ratio is increased is that suggested by
Horváth et al. [20] where there is a masking of the iron species due
to its covering by K or Ca oxide (lower right-hand side in Scheme 1).
The excess of alkali or alkaline-earth oxides or the formation of K- or
Ca-Fe species on the catalysts surface could block the accessibility
of the reactants to the active sites of the catalysts decreasing the
catalytic behaviour of the materials. This effect is confirmed by the
reduction in propylene conversion and PO yield between samples
K
0.01Fe0.005SiO2 and K0.02Fe0.005SiO2 measured at 450 ◦C.
[16] Q. Zhang, G. Chai, Y. Guo, W. Zhan, Y. Guo, L. Wang, Y. Wang, G. Lu, J. Mol.
Catal. A: Chem. 424 (2016) 65–76.
[17] G. Jin, G. Lu, Y. Guo, Y. Guo, J. Wang, X. Liu, Catal. Today 93–95 (2004) 173–182.
[18] X. Zheng, Q. Zhang, Y. Guo, W. Zhan, Y. Guo, Y. Wang, G. Lu, J. Mol. Catal. A:
Chem. 357 (2012) 106–111.
4. Conclusions
Fe-SiO2 catalysts modified with small loadings of K or Ca salts by
impregnation methodology have been prepared and tested in the
propylene epoxidation reaction in gas-phase using O2 as oxidant.
The addition of the promoters led to physicochemical changes of
the surface of the iron-based catalysts, such as: the complete elimi-
nation of the small particles/clusters of iron oxide generated during
the synthesis of the materials via the formation of superficial K-
or Ca-Fe species, the substitution of the hydroxyls with Brønsted
acid character by K or Ca species and the partial iron sites blocking.
These modifications produced by the K or Ca incorporation are also
reflected in the catalytic behaviour for the epoxidation of propy-
lene. As a first consequence, the propylene conversion (towards
CO2) is drastically reduced due to elimination of the small iron
oxide particles. Furthermore, the acidity reduction and the blocking
of the catalysts produce a reduction in the PO yield and avoids the
cracking of propylene on the catalyst surface. As a result of these
processes, a very promising PO selectivity (65.5%) is obtained using
small loadings of Fe and K, and O2 as oxidant without the production
of many organic byproducts.
[19] J. García-Aguilar, I. Miguel-García, J. Juan-Juan, I. Such-Basán˜ez, E. San Fabián,
D. Cazorla-Amorós, Á. Berenguer-Murcia, J. Catal. 338 (2016) 154–167.
ˇ
[20] B. Horváth, M. Sustek, I. Vávra, M. Micˇusˇík, M. Gál, M. Hronec, Catal. Sci.
Technol. 4 (2014) 2664–2673.
[21] Y. Yang, H. Xiang, Y. Xu, L. Bai, Y. Li, Appl. Catal. A: Gen. 266 (2004) 181–194.
[22] T. Chukeaw, A. Seubsai, P. Phon-in, K. Charoen, T. Witoon, W. Donphai, P.
Parpainainar, M. Chareonpanich, D. Noon, B. Zohour, S. Senkan, RSC Adv. 6
(2016) 56116–56126.
[23] S¸ . Kalyoncu, D. Düzenli, I. Onal, A. Seubsai, D. Noon, S. Senkan, Z. Say, E.I. Vovk,
E. Ozensoy, Catal. Lett. 145 (2015) 596–605.
[24] W. Yao, Y.L. Guo, X.H. Liu, Y. Guo, Y.Q. Wang, Y.S. Wang, Z.G. Zhang, G.Z. Lu,
Catal. Lett. 119 (2007) 185–190.
[25] Y. Li, Z. Feng, Y. Lian, K. Sun, L. Zhang, G. Jia, Q. Yang, C. Li, Microporous
Mesoporous Mater. 84 (2005) 41–49.
[26] X. Wang, Q. Zhang, Q. Guo, Y. Lou, L. Yang, Y. Wang, Chem. Commun. 12
(2004) 1396–1397.
[27] X. Wang, Q. Zhang, S. Yang, Y. Wang, J. Phys. Chem. B 109 (2005)
23500–23508.
[28] S. Yang, W. Zhu, Q. Zhang, Y. Wang, J. Catal. 254 (2008) 251–262.
[29] Q. Zhang, Q. Guo, X. Wang, T. Shishido, Y. Wang, J. Catal. 239 (2006) 105–116.
[30] A. Held, J. Kowalska-Kus´, A. Łapin´ ski, K. Nowin´ ska, J. Catal. 306 (2013) 1–10.
[31] Y. Wang, Q. Zhang, T. Shishido, K. Takehira, J. Catal. 209 (2002) 186–196.
[32] M.S. Kumar, M. Schwidder, W. Grünert, A. Brückner, J. Catal. 227 (2004)
384–397.
[33] J. Pérez-Ramírez, M.S. Kumar, A. Brückner, J. Catal. 223 (2004) 13–27.
[34] B. Horváth, T. Soták, M. Hronec, Appl. Catal. A: Gen. 405 (2011) 18–24.
[35] F. Fan, Z. Feng, C. Li, Acc. Chem. Res. 43 (2010) 378–387.
[36] S. Bordiga, R. Buzzoni, F. Geobaldo, C. Lamberti, E. Giamello, A. Zecchina, G.
Leofanti, G. Petrini, G. Tozzola, G. Vlaic, J. Catal. 158 (1996) 486–501.
[37] F. Fan, K. Sun, Z. Feng, H. Xia, B. Han, Y. Lian, P. Ying, C. Li, Chem. Eur. J. 15
(2009) 3268–3276.
[38] D. Scarano, A. Zecchina, S. Bordiga, F. Geobaldo, G. Spoto, G. Petrini, G. Leofanti,
M. Padovan, G. Tozzola, J. Chem. Soc. Faraday Trans. 89 (1993) 4123–4130.
[39] J.P. Gallas, J.M. Goupil, A. Vimont, J.C. Lavalley, B. Gil, J.P. Gilson, O. Miserque,
Langmuir 25 (2009) 5825–5834.
Acknowledgements
We thank the Spanish Ministry of Economy and Competi-
tiveness (MINECO), Generalitat Valenciana and FEDER (CTQ2015-
66080-R MINECO/FEDER and PROMETEOII/2014/010) for financial
support. J.G.A. thanks the Spanish Ministry of Economy and Com-
petitiveness (MINECO) for his fellowship (BES-2013-063678).
[40] A. Penkova, L.F. Bobadilla, F. Romero-Sarria, M.A. Centeno, J.A. Odriozola, Appl.
Surf. Sci. 317 (2014) 241–251.
[41] J. Lu, M. Luo, H. Lei, C. Li, Appl. Catal. A: Gen. 237 (2002) 11–19.
[42] O. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov, R. Lambert, J. Catal. 236
(2005) 401–404.
[43] A. Miller, B. Zohour, A. Seubsai, D. Noon, S. Senkan, Ind. Eng. Chem. Res. 52
(2013) 9551–9555.
[44] A. Palermo, A. Husain, M.S. Tikhov, R.M. Lambert, J. Catal. 207 (2002) 331–340.
[45] E. Ananieva, A. Reitzmann, Chem. Eng. Sci. 59 (2004) 5509–5517.
References
[1] D. Kahlich, U. Wiechern, J. Lindner, Ullmanns Encycl. Ind. Chem. (2012)
313–335.
[2] D. Trent, Kirk-Othmer Encycl. Chem. Technol. (2001) 1–26.
[3] S.J. Khatib, S.T. Oyama, Catal. Rev. 57 (2015) 306–344.
[4] K. Singh, K. Merchant, Handb. Ind. Chem. Biotechnol. (2012) 643–698.
[5] S. Kanungo, D.M. Perez Ferrandez, F. Neira d’Angelo, J.C. Schouten, T.A.
Nijhuis, J. Catal. 338 (2016) 284–294.
[6] X. Feng, X. Duan, H. Cheng, G. Qian, D. Chen, W. Yuan, X. Zhou, J. Catal. 325
(2015) 128–135.