You are on page 1of 10

Journal of Catalysis 289 (2012) 1–10

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Identification of the active species in oxidation reactions on mixed oxide catalysts:


Supra-surface or bulk surface species
B. Mehlomakulu a,1, T.T.N. Nguyen b, P. Delichère b, E. van Steen a, J.M.M. Millet b,⇑
a
Centre for Catalysis Research, Department of Chemical Engineering, University of Cape Town, Rondebosch 7701, South Africa
b
Institut de Recherches sur la Catalyse et l’Environnement de Lyon, UMR5256 CNRS-Université Claude Bernard Lyon 1, 2 Avenue A. Einstein, F-69626 Villeurbanne Cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: Mixed metal oxides have been used as catalysts for the partial oxidation of ethanol to acetaldehyde. The
Received 17 November 2011 solids were shown to be active and selective for the reaction at relatively low temperature. Various phys-
Revised 3 January 2012 ical techniques have been used to establish the active sites for this reaction and understand the role of the
Accepted 4 January 2012
different elements. The results obtained showed that Brønsted acidic sites were involved in the formation
Available online xxxx
of diethyl ether, acetic acid, and ethyl acetate and that vanadium was the active site, whereas the Fe3+/
Fe2+ redox couple did not play any role. The most important outcome of this study is the observed activity
Keywords:
dependence on surface V5+ cations content. These surface V5+ cations, which were formed as a result of
Ethanol
Oxidation
the evolution of the surface composition of the solid solution in oxidative atmosphere, should not be con-
Acetaldehyde sidered as part of the bulk oxide, but as supra-surface species. When the solid solution is vanadium rich,
Acetic acid the substitution mechanism in the bulk structure generates cationic vacancies that inhibited the forma-
Ethyl acetate tion of these surface species likely by blocking their diffusion to the surface. This study underlines the
Vanadium antimonate importance of both the supra-surface species and the bulk structure of oxidation catalysts. While it con-
Iron antimonite firms that the formation of supra-surface species occurs in mixed oxides oxidation catalysts, it also dem-
Metal oxides catalysts onstrates that this formation strongly depends on the bulk structure.
Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction [5–7]. In parallel, other studies have investigated the effect of iron
additive [8]. The characterization of the catalysts carried out using
Mixed metal oxides have been extensively used in the oxidation several techniques has shown that a continuous solid solution was
of light alcohols to different value-added products [1]. One of the formed between VSbO4 and FeSbO4. It was demonstrated that at
most important industrial applications of metal oxides in the oxi- low iron loading, Fe3+ substituted for V4+ in the cationic-deficient

dation of alcohols is the use of iron molybdate oxide catalysts, used structure of VSbO4 (V4þ 3þ
0:64 V0:28 Sb0:92 h0.16O4 with h cationic vacan-
industrially for the oxidation of methanol to formaldehyde since cies) generated very active sites for alkane activation. At high iron
1933 [2]. With the increasing availability of ethanol, the partial loading, almost only V3+ was present in the bulk of the solids, and
oxidation of ethanol is a potential alternative to the Wacker pro- Fe3+ only contributed to the isolation of the vanadium sites leading
cess currently used for the production of acetaldehyde. However, to a positive effect on only the selectivity to acrylonitrile [8].
less costly catalysts with a high catalytic activity and high selectiv- In this study, we have tested the FeVSbO mixed oxides for the
ity toward acetaldehyde for the low temperature (<230 °C) oxida- oxidation of ethanol. It has been demonstrated that for several
tion of ethanol are still lacking [3]. two-component metal oxide used as catalysts for the oxidation
The objective of this study was to investigate mixed vanadium of alcohol, one metal oxide may be present as an atomically dis-
and iron antimonates for the partial oxidation of ethanol to acetal- persed phase over the metal oxide structure [9]. It was, thus, inter-
dehyde. Vanadium antimonates are known to be active and selec- esting to know whether this feature could be extrapolated to
tive catalysts for the ammoxidation of alkanes [4]. Isolation and ternary systems and if so, which element/redox couple is involved.
dispersion of the vanadium sites at the surface of the VSbO4 phase Furthermore, additional fundamental insights about the nature of
have been identified as key factors required necessary to achieve the active surface species and their interaction with the bulk of
high activity and, most importantly, high acrylonitrile selectivity the catalyst could be gained. Several techniques including X-ray
diffraction, X-ray photoelectron and Mössbauer spectroscopy, and
pyridine adsorption followed by infrared spectroscopy (FTIR) have
⇑ Corresponding author. Fax: +33 72445399. been used to characterize the compounds of the FeSbO4–VSbO4 so-
E-mail address: Jean-marc.millet@ircelyon.univ-lyon1.fr (J.M.M. Millet). lid solution, which were tested as catalysts in an attempt to estab-
1
Fax: +27 21 650 5501. lish structure–activity relationships.

0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.01.001
2 B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10

2. Experimental the pre-heater leg filled with washed sea sand (dp > 500 lm). The
reaction conditions were kept uniform for all the runs, that is,
Ternary metal oxides catalysts, VxFe1xSbO4 with 0 6 x 6 1, WHSV of 2.44 gethanol/(gcatalyst h), Pethanol = 0.27 bar, PO2 = 0.3 bar,
were prepared using two methods depending on the vanadium balanced by N2 to a total pressure of 2 bar. The reaction tempera-
and iron content [10,11]. Briefly, for iron-rich compounds ture was varied in 10 °C increments between 200 and 350 °C. Tem-
(x < 0.5), Fe(NO3)39H2O (Saarchem, 98%) was heated to 80 °C, perature of the lines was kept at 140 °C. Methane gas (13 ml(NTP)/
and Sb2O3 (Strem, 99.9%) and NH4VO3 (Strem, 99%) were then min) was used as an internal standard. Samples were collected
added under stirring. The solution was neutralized using an aque- using an ampoule sampling technique. The organic compounds
ous ammonia solution (Saarchem, 25% NH3). The resulting slurry were separated by gas chromatography and analyzed using a FID.
was filtered and dried at 110 °C for 24 h. The solid obtained after CO and CO2 were analyzed in a CO/CO2 IR analyzer.
drying was calcined in air at 500 °C for 20 h and 900 °C for 2 h. The selective oxidation of ethanol was modeled as a first-order
For the vanadium-rich compounds (x > 0.4), NH4VO3 and Fe(- reaction [12,13]:
NO3)39H2O were stirred in 150 ml of distilled water at 80 °C.
r EtoH ¼ k  pEtOH
Sb2O3 was added to the NH4VO3 and Fe(NO3)39H2O solution. The
resulting slurry was then stirred under reflux at 110 °C for 18 h. The change in the partial pressure to expansion can be neglected,
The solvent was then evaporated, and the resulting mixture was since the nitrogen content in the feed gas is high.
dried at 110 °C for 24 h. The solid obtained after drying was cal-
r EtoH ¼ k  pEtOH;feed  ð1  X EtoH Þ
cined in air at 200 °C for 3 h and further at 700 °C for 2 h. The cat-
alysts are designated according to their relative iron content. The The reactor was approximated as a plug flow reactor, and thus, the
effect of washing of the catalyst after calcination was investigated rate constant k is obtained from:
by washing it in water at 80 °C for 3 h. Finally, two catalysts with
lnð1  X EtoH Þ
Fe substituted by Al and Ga (Al0.3V0.7SbO4 and Ga0.3V0.7SbO4) have k¼ W
also been prepared for comparison using the same protocol as for  pEtoH;feed
F EtoH;feed

the corresponding iron-containing catalyst.


The activity of the catalysts was determined as a function of tem-
The chemical composition and BET surface area determinations,
perature. The activation energy and pre-exponential factor were ob-
X-ray diffraction analyses, and 57Fe Mössbauer spectroscopy were
tained by the best fit of the measured conversion as a function of
performed as described previously [8]. Additional characterization
temperature in the range of conversion between 5 and 30% mol.
studies were performed as follows: IR spectra were recorded using
The selectivity for a certain product p (Sp) was calculated on the ba-
a Bruker Vector 22 Fourier transform spectrometer. Self-support-
sis of carbon efficiency using:
ing disks were prepared by pressing the calcined samples (about  
70 mg). Adsorption experiments were carried out in a quartz infra- Np ap
Sp ¼  100%
red cell. The pressed catalyst samples were mounted into the cell 2ðNi  Nf Þ
and connected to a conventional vacuum line maintained at a base
where Ni and Nf were, respectively, the number of mole of ethanol
pressure of 5  105 Torr with a diffusion pump. The samples were
in the gas feed before and after catalytic reaction, and Np was the
heat-treated under vacuum for 1 h at 220 °C to remove any physi-
number of mole of the p product with ap carbons.
sorbed molecules. The background spectra were measured at 25 °C.
Pyridine was adsorbed at room temperature for 3 min. A spectrum
was then taken upon evacuating the cell for 20–30 min at 25 °C. 3. Results
The cell was then sequentially heated and evacuated at 100, 150,
200, 220, and 250 °C for 1 h, followed by spectra recording (200 3.1. Characterization of the fresh and used catalysts
scans, resolution: 1 cm1, range of acquisition: 1000–4000 cm1).
All spectra were recorded at room temperature. The spectra pre- Table 1 summarizes the results obtained from the chemical
sented in this work were obtained by subtracting the spectrum analysis and the BET surface area measurements of synthesized
of the catalyst sample prior to adsorption from that of the sample VFeSbO catalysts. The formulae for these catalysts were calculated
on which pyridine was adsorbed at 25 °C and desorbed for 1 h at a using the chemical analysis data and results of characterization ob-
particular temperature. tained with other techniques not shown in this study. More details
XPS measurements were performed using a Kratos Axis Ultra on samples characterization can be found in the original paper [8],
DLD spectrometer. The base pressure in the analysis chamber cited throughout in the text. The X-ray diffraction patterns of the
was better than 5  108 Pa. XPS spectra of V2p, Fe2p, Sb3d3/2, solids showed only a single well-crystallized rutile-type phase for
O1s, and C1s levels were measured at 90° (normal angle with re- all prepared samples (Fig. 1).
spect to the plane of the surface) using a monochromated Al Ka The cation-deficient composition of VSbO4 was confirmed, and
X-ray source with a pass energy of 20 eV and a spot size aperture other compositions could be calculated on the basis of a fully oxi-
of 300 lm. Binding energies were corrected relative to the carbon dized rutile structure except at low iron content, for which the
1s signal at 284.6 eV. The signal intensities of Fe2p3/2, V2p3/2, charge discrepancy introduced by the iron substitution has to be
Sb3d3/2, and O 1s were measured using integrated areas under balanced by the presence of anionic vacancies or eventually a slight
the detected peak. The O1s signal intensity was obtained by sub- excess of antimony. Upon introduction of iron, a continuous de-
traction of the calculated Sb3d5/2 signal intensity to the total inten- crease in the V4+ content took place until a rutile structure with
sity of the peak due to the superposition of the O1s and Sb3d5/2. only V3+ cations was observed and almost no cationic vacancy. It
The experimental precision on XPS quantitative measurements can be recalled that the presence of these cationic vacancies was
was considered to lead to precision on the calculated cationic ratio confirmed by the observation of bands at 1015 and 880 cm1 in
of 0.15. the DRIFT spectra and that the linear relationship of the band at
The partial oxidation of ethanol was investigated in a fixed bed 1015 cm1, and the number of cationic vacancies was confirmed
reactor. 0.5 g of catalyst (dp < 106 lm) was mixed with 4.0 g of sil- [8]. V4+, balanced by cationic vacancies, should form locally in
icon carbide (dp = 200–450 lm) to minimize temperature gradient. chains of positively charged defects that could correspond to two
The catalyst mixture was packed in a U-shaped Pyrex reactor, with V4+ octahedra sharing an edge. This is consistent with the increase
B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10 3

Table 1
Bulk and surface cationic ratios calculated from chemical analyses and XPS, specific surface areas (SSAs) calculated from BET measurements and
calculated phase stoichiometry.

Catalyst V/(V + Fe, Al, Ga) bulk (Fe, Al, Ga + V)/Sb SSA (m2 g1) Formula
Bulk Surface
VSbO 1.00 1.04 1.17 3.3 5þ
V4þ 3þ
0:61 V0:32 h0.16Sb0:91 O4
10FeVSbO 0.90 1.03 1.07 10.0 5þ
Fe3þ 4þ 3þ
0:09 V0:52 V0:32 h0.18Sb0:92 O3:95
20FeVSbO 0.81 1.01 0.82 14.8 3þ 4þ 3þ 5þ
Fe0:18 V0:41 V0:33 h0.16Sb0:92 O3:88
30FeVSbO 0.69 0.95 0.92 16.1 5þ
Fe3þ V 4þ
V
0:28 0:33 0:30

h 0.11 Sb 0:92 O3:89
40FeVSbO 0.61 0.98 1.31 21.4 5þ
Fe3þ 4þ 3þ
0:38 V0:23 V0:33 h0.08Sb0:92 O3:92
50FeVSbO 0.52 1.12 1.18 16.7 5þ
Fe3þ V 4þ
V 3þ
h
0:47 0:14 0:39 0.05 Sb 0:92 O3:95
60FeVSbO 0.39 1.10 1.22 20.5 5þ
Fe3þ 4þ 3þ
0:63 V052 V0:41 Sb0:92 O4
80FeVSbO 0.19 0.98 1.38 6.6 5þ
Fe3þ 4þ 3þ
0:81 V052 V0:20 Sb0:92 O4
90FeVSbO 0.10 0.98 1.47 8.3 5þ
Fe3þ V 4þ 3þ
V
0:90 052 0:10 Sb0:92 O4
FeSbO 0.00 0.99 1.28 10.7 5þ
Fe3þ
0:99 Sb0:99 O4
20AlVSbO 0.78 1.01 – 13.4 3þ
Al0:18 V4þ 3þ 5þ
0:41 V0:33 h0.16Sb0:92 O3:88
20GaVSbO 0.81 0.96 – 25.6 5þ
Ga3þ V 4þ
V 3þ
0:18 0:41 0:33 h 0.16 Sb 0:92 O3:88

FeSbO4

V0.1Fe0.9SbO4

V0.2Fe0.8SbO4
Intensity, a.u.

V0.4Fe0.6SbO4

V0.5Fe0.5SbO4

V0.6Fe0.4SbO4
V0.7Fe0.3SbO4

V0.8Fe0.2SbO4
V0.9Fe0.1SbO4

VSbO4
25 30 35 40 45 50 55 60 65 70
Angle (2θ)

Fig. 1. X-ray diffraction pattern of the VxFe1xSbO4 catalysts series, with 0 < x < 1.

in the c parameter observed when the V4+ content decreased (for


0 6 x 6 0.5 in FexV1xSbO4). Two of these pairs would be associated
with a cationic vacancy. Since 0.66 V4+ per formula was present in
the pure vanadium antimonate (Table 1), this corresponds to 0.16
cationic defects. Taking into account the fact that this defect num-
ber was approximately the same as that of the maximum substitu-
tion of V4+ by Fe3+, one may propose that only one out of the four
V4+ cations could be substituted without destabilizing the defect.
The catalysts have further been characterized using 57Fe
Mössbauer spectroscopy at room temperature. The spectra of the
Fe0.2V0.8SbO4 sample before and after catalytic testing are pre-
sented in Fig. 2 for illustration. Scattering observed in the sample
after reaction is due to dilution of the catalyst with >80% silicon
carbide. All the spectra were fitted with one ferric doublet with
comparable isomer shifts (0.34 ± 0.2 mm s1) and quadrupolar
splittings (0.70 ± 0.2 mm s1). These parameters are typical of high
spin Fe3+ in octahedral coordination and are comparable with
those published for FeSbO4 [14]. The Mössbauer parameters of
the solids after catalytic testing were the same as before, showing
that iron was not reduced in the testing conditions. BET surface
area decreased with increasing vanadium content. The two distinct
categories of high vanadium content and high iron content catalyst Fig. 2. Mössbauer spectra recorded at 25 °C for the compound V0.2Fe0.8SbO4; (a)
are a result of the varying calcination temperature employed fresh catalyst and (b) after ethanol oxidation. Solid lines are derived from least-
(900 °C for iron-rich samples vs. 700 °C for vanadium-rich cata- square fits.
4 B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10

lysts). This implies that the crystallite size increased with increas- B L P B L LP
L
ing vanadium content.
The surface composition of the catalysts was determined using
25oC
XPS. It showed an enrichment of the surface with antimony, pre-

Absorbance
vailing throughout the catalysts series (Table 1). The surface anti- 100oC
mony enrichment might be due to the presence of fraction of
150oC
antimony as a microcrystalline or amorphous oxide, dispersed on
the rutile-type structure, as shown in previous studies [11,15]. 200oC
The effect has also been observed and reported on other antimo- 220oC
nate systems [16]. In addition, Nilsson et al. [17] observed for FeS- 250oC
bO4 that a rutile crystallite concentration gradient of catalyst 300oC
components existed, with Sb-richer zones on the outside parts of
the crystallites and consequently Fe-rich zones in the inner parts. 1700 1600 1500 1400
The surface V/(Fe + V) ratio of the catalysts was approximately Wavenumber, cm -1
equal to that expected on the basis of the bulk composition for x
Fig. 4. Pyridine desorption pattern for a high vanadium content catalyst
varying between 0 and 0.5 in FexV1xSbO4 and increased at higher (V0.7Fe0.3SbO4); Tpretreatment = 220 °C for 2 h under vacuum; Tadsorption = 25 °C; Tevac-
iron content (Fig. 3 and Table 2). The analysis of the V2p3/2 signal uation = 25 °C, 100, 150, 200, 220, and 250 °C for 1 h; spectra recorded at 25 °C. L

showed two peaks at 517.5 and 516.0 eV corresponding to the stands for Lewis, B stands for Brønsted acid sites, and P stands for physisorbed
V2p3/2 level of V5+ and V4+, respectively. At high iron content, an- pyridine.

other peak at 515.5 corresponding to V3+ is needed to fit the spec-


tra. Calculated binding energies (B.E.) and quantitative surface
at 1581 and 1438 cm1 are observed, decreasing drastically be-
analysis of the catalysts as well as spectra of the Fe0.6V0.4SbO4
tween 25 °C and 100 °C (Fig. 4). The bands at 1606, 1576, 1488,
and Fe0.2V0.8SbO4 catalysts with their deconvolution are given as
and 1447 cm1 are typical of pyridine chemisorbed on Lewis sites
Supplemental information. It is noteworthy that no correlation be-
originating from coordinative unsaturated V surface cations [18].
tween vanadium oxidation states and amount of surface antimony
The presence of a peak at the frequency of 1606 cm1 is a clear
was found; this supports the presence of excess of antimony at the
indication of the presence of medium to strong Lewis acid sites.
surface as supra-surface species or separate phase rather than anti-
The observed broad peak at (1538 cm1) is typical of vanadium-
mony-rich solid solution. Finally, it is noteworthy that no change in
based (V–OH) Brønsted acid sites [19]. Fig. 5 shows the Brønsted
surface composition of VSbO4 was observed before and after etha-
and Lewis relative acidity of the catalysts, expressed as area under
nol oxidation.
the peak at 1538 and 1448 cm1, normalized by surface unit in the
The acidity of VxFe1xSbO4 catalysts has been determined using
spectra recorded after desorption at 100 °C. It appeared clearly that
pyridine as a probe molecule in IR studies. From the spectrum re-
low vanadium content (high iron content) catalysts exhibit low
corded at 25 °C, two bands corresponding to physisorbed pyridine
acidity, while high vanadium content (low iron content) catalyst
exhibits high acidity.
The electrical conductivity behavior of four antimonates cata-
lyst has been measured as a function of temperature under air, at
atmospheric pressure (Fig. 6). There is a characteristic temperature
for each catalyst where the electrical conductivity started to in-
crease. This temperature is low for iron-rich catalyst (110–
130 °C), but relatively high for the vanadium-rich catalyst (200–
220 °C).
The compounds with rutile-type structure prepared with Al or
Ga instead of Fe exhibited similar X-ray diffraction patterns and
bulk compositions (Table 1). The introduction of Al led to a lower
specific surface area, whereas the introduction of Ga led to a larger
one. Finally, the washing of the Fe0.3V0.7SbO4 catalyst prior to

0.020 Lewis acid sites


Bronsted acid sites

0.015
Intensity

Fig. 3. Variation of surface V/(V + Fe) ratio as a function of bulk V/(V + Fe) ratio for
the VxFe1xSbO4 catalysts.
0.010

Table 2
0.005
Bulk cationic ratios from chemical analyses, specific surface areas (SSAs) calculated
from BET measurements and calculated phase stoichiometry.

Catalyst V/(V + Fe) (Fe + V)/Sb SSA (m2 g1) 0.000


0.0 0.2 0.4 0.6 0.8 1.0
20FeVSbO 0.81 1.01 14.8
20AlVSbO 0.78 1.01 13.4
V/(V+Fe), mol/mol
20GaVSbO 0.81 0.96 25.6
30FeVSbO 0.69 0.95 16.1 Fig. 5. Comparison of the Brønsted and Lewis relative acidity of the VxFe1xSbO4
30FeVSbO washed 0.69 0.95 24.0 catalysts, expressed as area under the peak at 1538 and 1448 cm1 normalized by
surface unit in the spectra recorded after desorption at 100 °C.
B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10 5

300

250
Elec. Cond. (10 ohm cm ))
-1

200
-1
-8

150

100

50

0
50 100 150 200 250 300 350 400
Temperature (°C)
Fig. 7. Conversion of ethanol as a function of temperature over VxFe1xSbO4
Fig. 6. Variation in electrical conductivity r as a function of the temperature during catalysts, where 0 < x < 1. Treaction = 150–300 °C, WHSV = 2.44 gethanol/(gcatalyst h);
temperature-programmed heating of FeSbO4 (squares), Fe0.9V0.1SbO4 (triangles), Pethanol = 0.27 bar, PO2 = 0.3 bar, balanced by N2 to a total pressure of 2 bar. (For
Fe0.2V0.8SbO4 (rhombus), and VSbO4 (circles) under dynamic vacuum from 25 to interpretation of the references to color in this figure legend, the reader is referred
400 °C (heating rate 5 °C min1). to the web version of this article.)

the calcination yielded larger specific surface area (24.0 vs. des VSbO4 and FeSbO4. Yet, ternary oxides richer in iron showed
16.1 m2 s1). A surface Sb enrichment was also observed for that an activity between that of the binary oxides. Typical product
compound (V + Fe/Sb = 0.6 instead of 0.8) (see Table 3). selectivity pattern over vanadium- and iron-rich catalysts as a
function of temperature are shown in Fig. 8. Diethyl ether seemed
3.2. Catalytic properties of the catalysts to form significantly only over vanadium-rich catalysts and with a
selectivity maximum at temperature slightly lower than 200 °C.
The conversion of ethanol as a function of temperature over Acetaldehyde was the major product on the two catalysts types
VxFe1xSbO4 is presented in Fig. 7. There were two distinct group- up to about 250 °C. The selectivity to acetaldehyde remained very
ings that can be observed from the graph, iron-rich catalysts (red high on the iron-rich catalysts above this temperature, but de-
symbols) and vanadium-rich catalysts (black symbols). It can creased markedly on the vanadium-rich catalysts. This decrease
clearly be observed that vanadium-rich ternary VxFe1xSbO4 cata- was associated with high selectivity to acetic acid and ethyl acetate
lysts exhibited higher activity than the corresponding binary oxi- and carbon oxides. The selectivity patterns observed were typical

Table 3
Binding energies (B.E.) and quantitative surface analysis of the catalysts.

Catalyst Line B.E. (eV) State and content (%) V/(V + Fe) (V + Fe)/Sb
VSbO V2p3/2 517.8, 516.5 1.00 1.0
Sb2p5/2 531.5 V5+(63%)V4+(37%)
10FeVSbO Fe2p3/2 712.0
V2p3/2 517.8, 516.5 V5+(77%)V4+(23%) 0.97 0.9
Sb2p5/2 531.6
20FeVSbO Fe2p3/2 712.3
V2p3/2 517.8, 516.5 V4+(74%)V4+(26%) 0.90 1.1
Sb2p5/2 531.5
30VSbO Fe2p3/2 712.0
V2p3/2 517.8, 516.5 V4+(71%)V4+(29%) 0.86 1.0
Sb2p5/2 531.6
40FeVSbO Fe2p3/2 712.3
V2p3/2 517.7, V4+(65%)V4+(29%)V3+(6%) 0.63 0.8
Sb2p5/2 516.5515.6531.6
50FeVSbO Fe2p3/2 512.0
V2p3/2 517.7, 516.5, V5+(65%)V4+(29%)V3+(6%) 0.68 0.9
Sb2p5/2 515.4531.6
60FeVSbO Fe2p3/2 712.2 –
V2p3/2 517.8, 516.5, 515.6 V5+(64%)V4+(29%)V3+(7%) 0.71 0.8
Sb2p3/2 531.5
80FeVSbO Fe2p3/2 712.2
V2p3/2 517.6, 516.4, 515.5 V5+(53%)V4+(36%)V3+(11%) 0.45 0.6
Sb2p3/2 531.5
90FeVSbO Fe2p3/2 712.2
V2p3/2 517.6, 516.6, 515.6 V5+(50%)V4+(38%)V3+(12%) 0.40 0.6
Sb2p5/2 531.5
6 B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10

corroborates well with the Mössbauer spectroscopy study showing


that iron was not reduced during catalytic testing. The selectivity
to the different products upon oxidation of ethanol did not vary
with the substitutions (Fig. 9b). The selectivity to carbon dioxide
mirrors that of acetaldehyde and is not shown in the figure, as that
to carboxylic acetates (acetic acid and ethyl acetate), which also
varied slightly. A slight variation was observed at high temperature
with a lower selectivity to acetaldehyde and high selectivity to
other oxidation products for the gallium-containing compounds.
It might be caused by the lower acidity of Ga compared to Fe.
The washed Fe0.3V0.7SbO4 has been tested as catalysts. Very lit-
tle differences in the conversion of ethanol were observed (Fig. 10).
An interesting aspect was observed in the selectivity patterns at
high temperature with an increase of the selectivity to acetic acid
and ethyl acetate.

4. Discussion
Fig. 8. Typical product selectivity patterns over V-rich (V0.7Fe0.3SbO4) catalysts
(black circles) in comparison to Fe-rich (V0.4Fe0.6SbO4) catalysts (red circles).
Treaction = 150–300 °C, WHSV = 2.44 gethanol/(gcatalyst h); Pethanol = 0.27 bar, PO2 = 4.1. Bulk and surface structure of the FeVSbO catalysts
0.3 bar, balanced by N2 to a total pressure of 2 bar. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of Before starting to discuss the results obtained, it should be
this article.) emphasized that as it has recently been reported, the vanadium
antimonite rutile structure is flexible [20]. It accommodate large
amounts of cationic vacancies with changes in the oxidation state
of series–parallel reaction pathway known for ethanol oxidation. of vanadium leading to structural modulations and short-range or-
The primary product is acetaldehyde as well as diethyl ether, ders rather than extended structural defects and making it very
which is formed by a competing parallel reaction. Acetic acid and sensitive to the preparation method. With that respect, it should
ethyl acetate are formed from acetaldehyde, and all intermediates be mentioned that the preparation method used in this study
are susceptible to further oxidation to carbon oxides. was different from that most of time use to prepare antimonates
Finally, it can be noted that V-poor catalyst remained very catalysts that lead systematically to Sb to other cations ratios lar-
selective to acetaldehyde at very high temperature. This tends to ger than one.
show that COx is mainly formed from carboxylic acetates oxidation Characterization of the mixed oxide catalysts prepared for this
rather than from acetaldehyde oxidation. study confirmed the cation-deficient composition of VSbO4 earlier
A study was undertaken to establish the role of Fe3+ in the cat- reported by Landa-Canovas et al. [21]. Upon substitution of Fe by V,
alytic oxidation of ethanol over the catalysts by substituting Fe3+ a continuous solid solution is formed from FeSbO4 up to VSbO4;
with other trivalent cations (Al3+ and Ga3+). Fig 9a shows conver- however, as already stated in the introduction, two ranges of com-
sion of ethanol as a function of temperature over V0.8M0.2SbO4 cat- position have to be distinguished in the solid solution for which
alyst (M = Fe, Al, Ga). The conversion over the three catalysts is the substitution mechanism is different. When vanadium substi-
comparable for all three catalyst systems. An activation energy tutes iron, it is first present as V3+. This isovalent substitution takes
for the conversion of ethanol of 100 ± 2 kJ/mol was determined, place up to about 50% of substitution. V4+ has been detected in the
implying a similarity in the species and sites involved in the range by EPR, but most of the cations are V3+ [22]. Substituting
rate-determining step and indicating that the Fe3+/Fe2+ redox cou- more than 50% of the iron cations, it is substituted almost exclu-
ple is not playing a catalytic role in the oxidation of ethanol. This sively by V4+ with a resulting charge discrepancy balanced cationic

Fig. 9. Effect of substitution of Fe3+ with Al3+ and Ga3+ on the conversion of ethanol (a) and on the selectivity to acetaldéhyde (b) over V0.7M0.3SbO4 catalysts (Fe: N, Al: d, Ga:
). Treaction = 150–300 °C, WHSV = 2.44 gethanol/(gcatalyst h); Pethanol = 0.27 bar, PO2 = 0.3 bar, balanced by N2 to a total pressure of 2 bar.
B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10 7

Fig. 10. Ethanol conversion (a) and selectivity to acetic acid and ethyl acetate (b) as a function of the temperature on washed (black triangles) and unwashed (black squares)
V0.3Fe0.7SbO4 catalyst. Treaction = 150–300 °C, WHSV = 2.44 gethanol/(gcatalyst h); Pethanol = 0.27 bar, PO2 = 0.3 bar, balanced by N2 to a total pressure of 2 bar.

vacancies. A slight maximum in V4+ content was observed around primary products of the oxidation of ethanol and are formed by
80% of substitution. The study of surface acidity by pyridine two different pathways. It is now accepted that ether is formed
adsorption confirmed well the presence of the two substitution from the alcohol according to an associative mechanism without
ranges and showed that the change also takes place at the surface formation of an intermediate alkoxide species [24]. Two ethanol
of the solids. The acid sites are present almost only on the vana- molecules would adsorb at a Lewis basic and a Brønsted acid sites
dium-rich compounds, and their number seems to decrease with to react via a complex carbenium-like transition state. The selec-
the iron content. This tends to show that these sites may corre- tive oxidation of ethanol to acetaldehyde proceeds via the forma-
spond to ferric cations in the close vicinity of a V cation or a tion of a surface alkoxide, followed by cleavage of the b–C–H
cationic vacancy. XPS characterization showed like acidity mea- bond in the rate-determining step [26]. According to the Mars
surements differences between the two substitution ranges. and van Krevelen mechanism [27], the role of oxygen is to re-oxi-
Whereas the surface V/(V + Fe) ratio followed more or less the bulk dize the reduced catalyst and thus maintain the redox cycle. Acetic
ratio on the vanadium-rich compounds, an excess of vanadium was acid and ethyl acetate formed from acetaldehyde in arguably par-
observed for the vanadium-poor compounds. The evolution of the allel reactions, whereas carbon oxides are formed from all the oxi-
vanadium surface content could be explained from the evolution of dation reaction intermediates and products.
the structure of the bulk. In the vanadium-rich compounds, cat- The results obtained clearly show that diethyl ether was formed
ionic vacancies should block the diffusion of cations and avoid on Brønsted acidic sites, which are only present on V-rich catalysts
the enrichment of the surface in vanadium, whereas in the vana- and below 220 °C. This later temperature corresponds to that at
dium-poor compounds, this enrichment takes place. Such enrich- which the electrical conductivity of the V-rich catalysts increased,
ment is driven by the need of the solid to equilibrate with the probably because of the dehydration of the catalysts surface. This
oxidative atmosphere of catalysis or heat treatment. Such equili- temperature is also the temperature at which the acid sites on
bration could proceed simply by oxidation of constituting cations, the surface fade away. When acid sites start vanishing, surface oxy-
but this is certainly insufficient, and migration of the element sus- gen vacancies are formed leading to a strong increase in electrical
ceptible to oxidation takes place. Similar enrichment was observed conductivity and the coordinative unsaturated cationic site formed
for FeVO4 catalysts and attributed to the lower surface energy of from these vacancies might be the catalytic sites for acetaldehyde
and high mobility of VOx species compared to FeOx species [24]. oxidation to acetic acid, ethyl acetate. However, the latter products
Other examples have recently been given that outline the fre- are not formed on V-poor compounds. As a matter of fact, if the in-
quency of such phenomenon in binary oxides [25]. crease in electrical conductivity, that is, beginning of dehydration,
Over the whole range, an excess of antimony was observed and starts at the same temperature, it increases drastically for V-poor
attributed to the recurrent presence of antimony oxide on antimo- but only above 300 °C for V-rich compound. It is interesting to
nates [15]. Antimony oxide supported on SiO2 has been studied as see that these strong increases correspond to the vanishing of
catalyst for ethanol oxidation [23]. It was observed that dispersed the Brønsted acid sites on both types of compounds. This clearly
antimony species were active, but not selective, and bulk oxide shows that the mechanism of formation of carboxylic acetates
was selective to acetaldehyde, but not active. However, on both (acetic acid and ethyl acetate) needs both oxidation and Brønsted
catalysts, acetaldehyde formation rapidly decreased exponentially acid sites.
with time on stream. Thus, antimony oxide as a phase should not In an attempt to identify the components of the catalyst respon-
contribute to the activity of the catalysts, although it might play sible for the observed activity for acetaldehyde formation reaction,
a role which will be discussed later. the rate of formation of acetaldehyde was tentatively correlated
with various species contents in the catalyst. For that, the rate of
4.2. Catalytic sites and reaction mechanism formation of acetaldehyde was modeled as a first-order reaction
with respect to ethanol and zero order with respect to oxygen.
Before entering into a full discussion on the possible catalytic Based on this model, the rate constant for each of the catalysts at
sites in the catalysts, it is important to recall what is commonly 210 °C was calculated and normalized with respect to the BET sur-
agreed on the reaction mechanisms of transformation of ethanol face area. The only significant correlation found was that to the
on oxide-based catalysts. Acetaldehyde and diethyl ether are the vanadium content with an almost linear increase of the catalyst
8 B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10

activity up to V/(V + Fe) = 0.5 and then a plateau up to V/


(V + Fe) = 1 (Fig. 11). Similar correlations have been searched for
selectivity patterns. For that, the selectivity to different products
upon the oxidation of ethanol over the VxFe1xSbO4 catalysts was
compared at 50% of ethanol conversion (Fig. 12). From Fig. 12b, it
is clear that the VxFe1xSbO4 catalysts are exceptionally selective
toward acetaldehyde. More than 90% acetaldehyde selectivity is
obtained throughout the catalyst range. The evolution of the selec-
tivity to diethyl ether, acetic acid, ethyl acetate, and COx confirms
that there are definitively two ranges of composition of the solid
solution with different properties. The limit between the two
ranges is rather marked and corresponds more or less to the mid-
dle of the solid solution (V0.5Fe0.5SbO4). It can be noted in Fig. 12d
that the proportion of ethyl acetate in the carboxylic acetates frac-
tion increased following the variation of the vanadium active sites
in the solid solution. It increased linearly below V/(V + Fe) = 0.5 and
remained constant above. The formation of acetic acid and ethyl
acetate can be attributed to the reaction of surface ethylate with
Fig. 11. First-order rate constant for the formation of acetaldehyde on the ethanol adsorbed water and ethanol, respectively. Water and ethanol
oxidation at 210 °C as a function of the vanadium content of the VxFe1xSbO4 should be in competition to adsorb on the same Brønsted acid sites,
catalysts, normalized with respect to the specific surface area; WHSV = 2.44 gethanol/
and thus, ethyl acetate-to-acetic acid ratio depends on the ethanol-
(gcatalyst h); Pethanol = 0.27 bar, PO2 = 0.3 bar, balanced by N2 to a total pressure of
2 bar. to-water ratio present in the reaction mixture. Since water is pro-

Fig. 12. Products selectivity to diethyl ether (a), products in the fraction of oxidation products (b and c), and acetic acid content of the acid + ester fraction as a function of the
vanadium content of the VxFe1xSbO4 catalysts at 50% conversion in the oxidation of ethanol. WHSV = 2.44 gethanol/(gcatalyst h); Pethanol = 0.27 bar, PO2 = 0.3 bar, balanced by N2
to a total pressure of 2 bar.
B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10 9

duced by ethanol oxidative dehydrogenation, it is consistent to find [11,34,29,35]. Such effects that seem opposite may take place,
the same dependence of the selectivity to ethyl acetate as ethanol respectively, at low and high temperature, increasing the catalysts
oxidation activity with the V content of the catalysts. activity at low temperature and increasing the selectivity to car-
It has been demonstrated from earlier studies that the active boxylic acetates and to carbon oxides at high temperature.
species in alcohol oxidation were likely V5+ species [28]. Such spe- The rutile-type bulk oxide with the supra-surface antimony
cies are not present in the bulk of the solids, but analyses of the oxide phase may constitute an efficient support for the VOx species
vanadium region of the XPS spectra showed that these are present allowing tuning their reducibility for high catalytic properties [36].
at the surface. The V5+ surface content has been calculated from However, the catalytic properties are strongly dependent on the
XPS data, considering that the ratio (V + Fe)/Sb in the mixed oxide rutile bulk oxide that controls the dispersion of the VOx species
was equal to one. It is plotted as a function of the V/(V + Fe) bulk and generates acid sites that intervene in the reactions.
ratio and compared to the ethanol conversion in Fig. 13. It can be
seen that this content increases at low vanadium content and re- 5. Conclusion
mains constant at high vanadium content in a similar manner as
the ethanol conversion. Thus, it may be postulated that V5+ surface This study showed that mixed vanadium and iron antimonates
species are the active sites for ethanol oxidation. However, it is not are efficient catalysts for partial oxidation of ethanol to acetalde-
possible to conclude how these species are distributed at the sur- hyde. The activity of the ternary mixed oxides increases with their
face. The V5+ species are the major vanadium species at the surface, vanadium content up to a composition corresponding approxi-
but there is also an excess of antimony. Because of redox equilib- mately to V(/Fe + V) = 0.5. At that point, their activity reached that
rium, both elements should not be associated in the same oxide of VSbO4. However, the ternary mixed oxide appeared much more
phase at the surface, and the VOx may be present as supra-surface selective to acetaldehyde due to the absence of Brønsted acid sites
sites more or less aggregated and isolated from by the antimony that promote the formation of acetic acid or ethyl acetate via reac-
oxide on the rutile-type phase. Interestingly, this conclusion is in tion of adsorbed acetaldehyde with water or ethanol themselves
agreement with recent studies conducted on FeVO4 and VSbO4 cat- adsorbed on the acid sites. In view of these results, it appears rea-
alysts in methanol oxidation showing that the catalytic active sites sonable to conclude that VOx species and more likely pentavalent
reside in an outermost surface VOx layer and not in the bulk lattice species are the active species in alcohol oxidation. Similar conclu-
[29–31]. Such feature may be common to a lot of vanadium mixed sions have already been reached for other mixed oxides catalysts. It
oxides based on solid solution. has not been possible to determine how these species are distrib-
The effect of washing after calcination of the catalysts and the uted at the surface, but they presumably should not be anymore
effect of surface excess of antimony can be addressed in the same considered as part of the bulk oxide. When the solid solution is rich
time, since the major consequence of washing is an increase of sur- in vanadium (0.5 < V/(Fe + V) < 1), the substitution mechanism of
face antimony content. The washed sample seemed to show higher iron by vanadium changes, and the bulk structure involves the cre-
activity than the unwashed sample at low temperature and a high- ation of cationic vacancies that inhibited the formation of these
er selectivity to carboxylic acetates at high temperature. surface species. This certainly occurred by blocking the diffusion
Several roles have been attributed to supra-surface antimony of vanadium to the surface when the solid solution equilibrates
oxide phase associated with antimonates when these solids were in oxidative atmosphere. Furthermore the defective structure gen-
used as catalysts for various oxidation or ammoxidation reactions. erates acid sites whose presence close to oxidation sites allows the
Among these roles, some could be retained for ethanol oxidation. It formation of (i) diethyl ether at low temperature directly from
can be proposed that they would contribute to the active-phase re- ethanol and (ii) carboxylic acetates at high temperature from
oxidation as likely as MoO3 bulk phase do for FeMoO catalysts in acetaldehyde.
methanol oxidation [32–34]. They could also bring about isolation Interestingly, this study shows in the same time the importance
of vanadium centers at the surface, which benefit to the selectivity of supported surface species as active species in a mixed oxide cat-
by optimizing the size of the molecular cluster formed by the sur- alyst and the importance of the bulk structure of this mixed oxide
face vanadium species or/and avoiding over-oxidation to COx catalyst, as the formation of surface species appears strongly
dependent on bulk structure.

Acknowledgments

Sasol Technology R&D (South Africa) and CNRS (France) are


gratefully acknowledged for financial support and the South Africa
National Research Foundation for traveling grants.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in


the online version, at doi:10.1016/j.jcat.2012.01.001.

References

[1] G.I. Golodets, Heterogeneous catalytic reactions involving molecular oxygen,


J.R.H Ross (Ed.), Stud. Surf. Sci. Catal. vol. 15, 1983, pp. 280.
[2] V.E. Meharg, A. Adkins, U.S. Patent 1913,405, 1933.
[3] R. Tesser, V. Maradei, M. Di Sergio, E. Santacesaria, Ind. Eng. Chem. Res. 43
(2004) 1623.
[4] G. Centi, S. Perathoner, F. Trifiro, Appl. Catal. 157 (1997) 143.
Fig. 13. Variations in the surface V5+ content and in the rate constant for the [5] A.T. Guttmann, R.K. Grasselli, J.F. Brazdil, U.S. Patents 4746,641 and 4788,317,
formation of acetaldehyde at 210 °C vs. bulk V/(V + Fe) ratio for the VxFe1xSbO4 1988.
catalysts. [6] G. Centi, P. Mazzoli, Catal. Today 28 (1996) 351.
10 B. Mehlomakulu et al. / Journal of Catalysis 289 (2012) 1–10

[7] A. Andersson, S.L.T. Andersson, G. Centi, R.K. Grasselli, M. Sanati, F. Trifiro, Appl. [22] D.L. Nguyen, Y.B. Taarit, J.M.M. Millet, Catal. Lett. 90 (2003) 65.
Catal. 113 (1994) 43. [23] K. Matsuzawa, T. Shido, Y. Iwasawa, Langmuir 19 (2003) 2756.
[8] H. Roussel, B. Mehlomakulu, F. Belhadj, E. van Steen, J.M.M. Millet, J. Catal. 205 [24] S.R. Blaszkowski, R.A. van Santen, J. Am. Chem. Soc. 118 (1996) 5152.
(2002) 97. [25] S.V. Merzlikin, N.N. Tolkachev, L.E. Briand, T. Strunskus, C. Wöll, I.E. Wachs, W.
[9] I.E. Wachs, Appl. Catal. A: General 391 (2001) 36–42. Grünert, Angew. Chem. Int. Ed. 49 (2010) 8037–8041.
[10] M. Allen, R. Betteley, M. Bowker, G.J. Hutchings, Catal. Today 9 (1991) 97. [26] M. Seman, J.N. Kondo, K. Domen, C. Reed, S.T. Oyama, J. Phys. Chem. B 108
[11] R. Nilsson, T. Linblad, A. Andersson, C. Song, S. Hansen, Stud. Surf. Sci. Catal. 82 (2004) 3231.
(1994) 281. [27] C. Doonkamp, V. Ponec, J. Mol. Catal. A: Chem. 162 (2000) 19.
[12] A. Yee, S.J. Morisson, H. Idriss, J. Catal. 186 (1999) 279. [28] I.E. Wachs, Catal. Today 100 (2005) 79–94.
[13] H. Idriss, E.G. Seebauer, Catal. Lett. 66 (2000) 139. [29] K. Routray, W. Zhou, C.J. Kiely, I.E. Wachs, ACS Catal. 1 (2011) 54.
[14] F.J. Berry, J.G. Holden, M.H. Loretto, J. Chem. Soc., Dalton Trans. 7 (1987) 1727. [30] J. Nilsson, A. Landa-Cánovas, S. Hansen, A. Andersson, Catal. Today 33 (1997)
[15] G. Centi, S. Perathoner, F. Trifiro, Appl. Catal. A: General 157v (1997) 143. 97.
[16] N. Ballarini, F. Cavani, C. Guinchi, R. Catani, U. Cornaro, Catal. Today 28 (2002) [31] M. Olga Guerrero-Perez, T. Kim, M.A. Banares, I.E. Wachs, J. Phys. Chem. C 112
1. (2008) 16858–16863.
[17] J. Nilsson, A.R. Landa-Canovas, S. Hansen, A. Andersson, J. Catal. 160 (1996) [32] A.P.V. Soares, M. Farinha Portela, A. Kiennemann, L. Hilaire, J.M.M. Millet, Appl.
244. Catal. A: General 206 (2001) 221.
[18] G. Busca, Catal. Today 41 (1998) 191–198. [33] B.K. Kim, B.H. Wang, H.S. Lee, H.C. Woo, D.W. Park, Korean J. Chem. Eng. 21
[19] G. Centi, R.K. Grasselli, F. Trifiro, Catal. Today 13 (1992) 661. (2004) 104.
[20] A.R. Landa-Cánovas, F. Javier García-García, S. Hansen, Catal. Today 158 (2010) [34] J.C. Vedrine, G. Coudurier, J.M.M. Millet, Catal. Today 33 (1997) 3.
156. [35] R. Nilsson, T. Linblad, A. Andersson, J. Catal. 148 (1994) 501.
[21] A. Landa-Canovas, J. Nilsson, S. Hansen, K. Stahl, A. Andersson, J. Solid State [36] W. Zhang, J. Desika, S.T. Oyama, J. Phys. Chem. 99 (1995) 14468.
Chem. 116 (1995) 369.

You might also like