Screening of catalysts for the oxidative dehydrogenation of ethyl lactate to ethyl pyruvate, and optimization of the best catalysts

Article abstract of DOI:10.1016/j.apcata.2020.117619

This study reports on a new industrial gas-phase process to produce ethyl pyruvate, using the oxidative dehydrogenation of ethyl lactate. In an initial approach, twenty catalysts were screened under reaction conditions required for industrial applications. When evaluated in terms of their maximum selectivity to ethyl pyruvate and their productivity, the best catalyst was found to be a phosphomolybdic polyoxometalate, with iron counter-cations: FePMo12O40. To improve this catalyst, a study was carried out by including cesium counter-cations. The new catalysts exhibited higher yields because of higher Mo active surface site density related to higher specific surface area and although their intrinsic activity was lower. Their ethyl pyruvate selectivity increased with iron content but never overpass that of FePMo12O40. These results are explained in terms of changes in structure and iron counteraction coordination leading or not to electron exchange between iron and molybdenum and influencing the acidity of the catalyst's surface.

Full text of DOI:10.1016/j.apcata.2020.117619

Journal Pre-proof
Screening of catalysts for the oxidative dehydrogenation of ethyl lactate
to ethyl pyruvate, and optimization of the best catalysts
M. Huchede, D. Morvan, V. Bellie`re-Baca, J.M.M. Millet
PII:
S0926-860X(20)30212-X
DOI:
https://doi.org/10.1016/j.apcata.2020.117619
APCATA 117619
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
7 January 2020
26 April 2020
3 May 2020
Please cite this article as: { doi: https://doi.org/
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
©
2020 Published by Elsevier.

Screening of catalysts for the oxidative dehydrogenation of ethyl
lactate to ethyl pyruvate, and optimization of the best catalysts.
1
2
2
1*
M. Huchede , D. Morvan , V. Bellière-Baca , J.M.M. Millet
1
Univ Lyon, Université Claude Bernard Lyon 1, CNRS, IRCELYON - UMR 5256, 2 Av.
Albert Einstein, 69626 Villeurbanne (France)
2
Adisseo, Antony Parc 2, 10 Place Général de Gaulle, 92160 Antony, France
*
Corresponding author: jean-marc.millet@ircelyon.univ-lyon1.fr
Graphical abstract
Highlights
Twenty catalysts were screened for the oxidative dehydrogenation of ethyl lactate into ethyl
pyruvate

1



Iron-substituted phospho-molybdic heteropolyacids were the catalysts with highest efficiency
Catalyst activity depends on Mo surface sites density and is decreased when Fe substitutes
protons in supported acids


Fe has a positive effect only in FePMo12
O40 because it induced a charge transfer improving
redox mechanism
Ethyl pyruvate selectivity depends on the catalyst acidity, strong acid sites promoting side
reactions like hydrolysis
Abstract
This study reports on a new industrial gas-phase process to produce ethyl pyruvate, using the oxidative
dehydrogenation of ethyl lactate. In an initial approach, twenty catalysts were screened under reaction
conditions required for industrial applications. When evaluated in terms of their maximum selectivity
to ethyl pyruvate and their productivity, the best catalyst was found to be a phosphomolybdic
polyoxometalate, with iron counter-cations: FePMo12O40. To improve this catalyst, a study was
carried out by including cesium counter-cations. The new catalysts exhibited higher yields because of
higher Mo active surface site density related to higher specific surface area and although their intrinsic
activity was lower. Their ethyl pyruvate selectivity increased with iron content but never overpass that
of FePMo12O40. These results are explained in terms of changes in structure and iron counteraction
coordination leading or not to electron exchange between iron and molybdenum and influencing the
acidity of the catalyst's surface.
Keywords: Ethyl lactate, Ethyl pyruvate, Oxidative dehydrogenation, Molecular oxygen, Screening of
catalysts, Polyoxometalate catalysts
1. Introduction
Lactic acid and lactates have been the precursors of choice for the production of industrial
chemicals. Indeed, they can be sustainably produced from biomass, by fermentation [1]. One of the most
relevant and economically interesting applications for these compounds could be their transformation
into other important and valuable compounds, such as pyruvic and acrylic acids, or propanediol [2]. The
latter compounds are widely used in the pharmaceutical and cosmetics industry, and in food processing.
There is thus renewed interest in the development of efficient catalysts for the oxidative
dehydrogenation of lactic acid to pyruvic acid, or for its dehydration to acrylic acid [3-5].
2

The main difficulty encountered in this type of research is the low selectivity obtained, which
is due firstly to the hydrolysis of esters and their further transformation, and secondly to over-oxidation
of the reaction product or intermediates. Moreover, both lactic acid and pyruvic acid lack in stability.
Lactic acid decomposes easily to acetaldehyde, propanoic acid and CO
x
. In addition, it tends to
polymerize at high temperatures [6]. Pyruvic acid decomposes also to CO
x
and hydroxyethylidene,
which isomerizes to acetaldehyde. One of the techniques used to at least partially solve this problem
involves dehydrating or oxidatively dehydrogenating the alkyl lactates to the corresponding alkyl
acrylates or alkyl pyruvates [7-9]. The resulting esters are more stable from the thermal point of view,
and do not form oligomers. Furthermore, as they are more volatile they are easier to spray and do not
need to be diluted in water, as in the case of the related acids. This last point is quite important when
considering the productivity of an industrial process.
Among the alkyl lactates (methyl, propyl, isopropyl, ...), ethyl lactate is generally preferred, as
it is considered to be a commodity chemical and is therefore inexpensive. Indeed, ethyl lactate can be
used as a food additive and in the manufacture of fragrances [10]. However, it is used mainly as a
solvent, with a particularly attractive application in the coatings industry due to its high solvation power
[
11]. Although the cost of ethyl lactate currently ranges between $ 0.75 and $ 1.00 per kilogram, a drop
in price of up to ~ $ 0.50 per kilogram is projected, which would allow it to compete directly with
petroleum-derived solvents. The ethyl lactate oxidative dehydrogenation reaction can be carried out in
the gas or liquid phases. The latter type of reaction leads to processes that make it possible to avoid
over-oxidation of the pyruvate, but are less well adapted to production on an industrial scale [12-15].
The reaction requires expensive noble metal catalysts for the activation of oxygen, and in general
requires the presence of an excess of a base such as NaOH or LiOH. It is interesting to note the
interesting results reported in two very recent studies [16, 17], the first of which describes the apparently
efficient use of titanosilicalite (TS-1) in the presence of hydrogen peroxide [16]. Oxidation was carried
out in a closed reactor, with one equivalent of H
2
2
O and 20% by weight of TS-1 relative to ethyl lactate.
After 5 h at 70 °C in tert-butanol, the ethyl pyruvate yield was 79%. The second and more recent study
proposed the use of a mesoporous vanadium titanium mixed oxide catalyst [8]. The liquid-phase aerobic
3

oxidation of ethyl lactate to ethyl pyruvate was carried out with O
Although a selectivity to ethyl pyruvate greater than 90% was reported, the best yield never exceeded
5% after 5 h of reaction.
2
in diethyl succinate at 130°C.
5
Ethyl lactate can be efficiently oxidized in the liquid phase, but from an industrial point of
view it is undeniably more advantageous to run the reaction in the gas phase, especially if production
units of several hundred thousand tons per year are envisaged. For this reason, the present study focuses
on this technology, in particular on the development of the most suitable catalyst for the reaction. Until
now, two main types of catalytic system have been studied: mixed oxides or supported transition metal
oxides, and silica-supported silver catalysts. If we focus on the most efficient catalysts of the first type,
two of them stand out in in the literature: VPO
x
/SiO
2
and TeO
2
-MoO , [17-18]. Although the yield in
3
ethyl pyruvate obtained over these catalysts was always higher than 40%, none are found to be totally
satisfactory.
The best patented catalyst is certainly VPO
x
/SiO (molar ratio P / V = 10), using air as the
2
oxidant and diluent [17]. The best performance reported with this catalyst was achieved at 200 °C, with
conversion around 90% and a selectivity of 60%. However, the reaction was run with an extremely
diluted flux of ethyl lactate gas, which is hardly compatible with an industrial application. Furthermore,
catalyst losses phosphorus over time and it is necessary to often add some to the reaction stream.
The TeO
2
-MoO catalyst, which main active component was the phase gives 90% selectivity to ethyl
3
pyruvate for 80% of conversion at 300°C [18]. However, deactivation of this catalyst was reported,
arising from the reduction of tellurium to metallic tellurium, thus limiting its potential for industrial
applications.
The second family consisting of metallic silver, in bulk or supported form, are industrial
catalysts of choice for dehydrogenation reactions. They have been used for the production of
formaldehyde from methanol, and are very effective for the conversion of allyl alcohol to acrolein, and
ethanol to acetaldehyde [19-22]. The use of silica-supported metallic silver for the oxidative
4

dehydrogenation of ethyl lactate has been patented [23]. The best yield, equal to approximately 80%,
was obtained on a catalyst having a silver content varying between 20 and 40% by mass, doped with
0.2% Pd. However, it should be noted that the reported catalytic properties were obtained with a gas
mixture lying within the explosivity range of ethyl lactate and with a low productivity, thus
compromising its application to industrial processes.
Since none of the effective catalysts were found to be suitable for industrial applications, we
screened them in an effort to reveal any new, potentially efficient catalyst for the oxidative
dehydrogenation of ethyl lactate. Several oxidative dehydrogenation catalysts were thus selected,
synthesized and characterized. The choice of catalysts was made on the basis of their oxidation-
reduction and acid-base properties, as well as the catalytic characteristics previously demonstrated in
oxidative dehydrogenation reactions with similar compounds. This initial screening phase corresponded
to the first objective of our study. The best candidate resulting from this selection process was a poly-
oxometallate based catalyst. The second aim of the study was then to analyze this compound in greater
detail, in an attempt to understand how this catalyst works and optimize its composition.
2. Experimental
2.1.Catalysts synthesis
FeVO
mmol of NH VO
containing 6 mmol of Fe(NO
was stirred for 20 min. The pH was then adjusted to 4 by adding NH
for 2 h at 50°C under stirring. The solid was finally isolated by centrifugation, washed with distilled
water (2 × 30 ml), dried for 12 h at 90°C and then calcined for 3 h at 600°C. Iron molybdate, Fe (MoO
was synthesized from an aqueous solution (100 ml) containing 2.2 mmol of (NH Mo 24 and another
O. The two solutions were mixed and
4
, was synthesized by co-precipitation from an aqueous solution (20 ml) containing 6
maintained at 50°C under stirring [24]. This solution was added to a similar solution
.9H O. The pH was adjusted to 1 with HNO (65%) and the suspension
(32%) and the suspension heated
4
4
3
)
3
2
3
3
2
4 3
) ,
4
)
6
7
O
aqueous solution (50 ml) containing 10.0 mmol of Fe(NO
3
)
3
.9H
2
5

vigorously stirred for 1 h at 25°C. The water was then evaporated under vacuum (10 mbar) at 70°C. The
solid formed was dried for 12 h at 90°C and calcined for 3 h at 500°C. Iron and vanadium antimonate,
Fe0.4
V
0.6SbO
4
, (FeVSbO
and 5.3 mmol of Fe (NO
are added and the resulting suspension is refluxed for 18 h under stirring. The water is then
evaporated under vacuum (10 mbar) at 70°C. The solid was dried for 12 h at 90°C and then calcined for
h at 200°C. and then for 2 h at 700°C. under air. The mixed oxide Te MoO was synthesized by
precipitation from an aqueous solution (20 ml) containing 0.3 mmol of (NH Mo O, to which
was added a similar solution containing 4.3 mmol of H TeO . The solution is then kept at 60°C, with
4
) was synthesized from an aqueous solution (150 ml) containing 8 mmol of
4
NH VO
3
3
)
3
.9H O, maintained at 80°C [25]. After complete dissolution, 7 mmol
2
of Sb
2
O
3
3
2
7
4
)
6
7
O24.4H
2
6
6
stirring, until a white paste is obtained. The paste was dried for 12 h at 90°C. and then calcined for 2 h
at 470°C under air [26].
Molybdenum oxide was synthesized by precipitation from an aqueous solution (10 ml) containing 1
mmol of (NH
mmol of citric acid was added to the previous solution and the reaction medium was heated for 6 h at
0°C. The precipitate formed was isolated by centrifugation, washed with distilled water (2 × 30 ml),
dried for 12 h at 90°C and then calcined for 3 h at 500°C in air.
FePO was prepared using a protocol described elsewhere [28]. The ferric salt of phosphomolybdic acid,
40, was prepared by precipitation [29,30]. An aqueous solution (5 ml) containing 1 mmol of
.9H O was added dropwise to a similar solution containing 1 mmol of H 40 (Aldrich
4
)
6
Mo
7
O
24.4H
2
O maintained at 60°C with stirring [27]. A similar solution containing 1
9
4
FePMo12
O
Fe (NO
3
)
3
2
3
PMo12O
ref. 431400). The phosphomolybdic acid being hydrated, its hydration rate was determined by
thermogravimetric analysis prior to this operation. It was of 15.3 molecules per Keggin unit (K.U.). The
resulting yellow suspension was maintained at 50°C for 1 h. The water was then evaporated under
vacuum (10 mbar) at 70°C. The solid obtained was dried for 12 h at 90°C. and calcined for 5 h at 350°C
in air. The synthesis of the acids substituted by iron and cesium (Cs
and 0.33 and named Cs Fe ) was carried out according to the same protocol using a solution of iron
nitrate and cesium nitrate.
The VTiPO mixed oxide was prepared by hydrothermal synthesis from an aqueous solution (10 ml)
containing 21.2 mmol of TiCl , the pH of which was adjusted to 8 with NH (32%) [31]. Then, 61.4
2
Fe
x
PMo12O40 with x=0.0, 0.1, 0.2
2
x
x
4
3
6

mmol of H
3
3
PO
4
(85%) were added to the previous solution. An aqueous solution containing 10.4 mmol
of NH VO
4
and 20 ml of an equivolumic mixture of lactic acid and distilled water was maintained at
80°C. This solution was added to the titanium solution with stirring. The suspension obtained was then
-1
placed in a 30 ml microwave reactor and heated at 175°C for 15 min (700 W, 35°C.min ). The solid
formed was isolated by centrifugation, washed with distilled water (2 × 30 ml) and dried for 12 h at
9
0°C. The solid W0.75
synthesized hydrothermally was according to the synthesis method [32]. To an aqueous solution (40 ml)
containing 4.9 mmol of (NH Mo O was added a similar solution containing 2.4 mmol of
VOSO .xH O. The resulting suspension was stirred under nitrogen for 20 min (20 ml.min ) to remove
V
0.16Nb0.09
O
3
(MoVNbO) and MoV0.33
O
x
(MoVO) mixed oxide were also
4
)
6
7
O24.4H
2
-1
4
2
dissolved oxygen, and pH was adjusted to 2.6 The mixture was placed in a 100 ml Teflon autoclave and
heated for 48 h at 175°C. The resulting solid was then isolated by centrifugation, washed with distilled
-1
water (2 × 30 ml), dried for 12 h at 90°C, then calcined for 2 h at 500°C, under nitrogen (100 ml.min ,
-1
-
5
°C.min ). The final solid (1g) was suspended in 25 ml of an aqueous solution of oxalic acid (0.4 mol.l
1)
and maintained for 30 min at 60°C, with stirring. The solid was isolated by centrifugation, washed
with distilled water (2 × 30 ml) and dried for 12 h at 90°C.
, (MoVTeNbO) more commonly called M1 phase, was obtained
by precipitation [33]. A first aqueous solution containing (NH Mo O, NH VO and H TeO
was stirred at 80°C until complete dissolution. A second solution containing Nb .xH O and oxalic
The mixed oxide MoV0.30Nb0.17Te0.14
O
x
4
)
6
7
O
24.4H
2
4
3
6
6
2
O
5
2
acid (molar ratio Nb / oxalate = 3) was heated to 100°C until complete dissolution. The latter was then
added to the first solution, resulting in a ratio Mo/V/Te/Nb = 1 / 0.27 / 0.16 / 0.14. After stirring for 30
min at 25°C, the water was then evaporated under vacuum at 80 °C. and the solid is dried for 12 h at
-1
-1
9
0°C. and then calcined for 4 h at 300°C in air (100 ml.min , 5°C.min ), then 4 h at 600°C under
-1
nitrogen (100 ml.min-1, 5°C.min ). The final solid (1 g) was suspended in 10 ml of a hydrogen peroxide
solution (30%) and kept at 60°C for 2 h. The solid was isolated by centrifugation, washed with distilled
water (2 × 30 ml) and dried for 12 h at 90°C.
Titanium pyrophosphate Ti
2
P
2
O
7
was prepared from an aqueous solution (70 ml) containing 6 ml of
[34]. To this solution were added,
lactic acid (85%), 0.6 ml of ammonia (32%) and 16.7 mmol of TiCl
4
with stirring, 33.4 mmol of H
3
PO
4
(85%). The water was evaporated at 50°C. at atmospheric pressure
7

and the solid calcined for 6 h at 300°C., then 6 h at 450°C. and finally 6 h at 600°C, under air (100
-1
-1
ml.min , 5°C.min ).
The mixed hydroxyphosphate Fe
3
(PO
4
)
2
(OH)
2
, was prepared by hydrothermal synthesis. In 20 ml of
PO (85%), 1.3 mmol of FeCl and 0.67 mmol of
O [35]. The solution was then placed in a 30 ml microwave reactor and heated at 200°C for
0 min. The solid was isolated by centrifugation, washed with distilled water (2 × 30 ml) and then dried
distilled water were dissolved 2.6 mmol of H
FeC .2H
3
4
3
2
O
4
2
1
for 12 h at 90°C.
The vanadium oxide supported on mesoporous silica, VO
x 2
/SiO , at 1% by weight, was prepared by
dissolving 0.34 mmol of NH
4
VO
3
, 1 g of CTAB and 11 g of NH
4
Cl in 65 ml of distilled water [36]. The
pH of the solution was then adjusted to 5.5 before addition of 31.3 mmol of TEOS. After 24 h at 40°C,
the precipitate was isolated by centrifugation, washed with distilled water at 80°C (2 x 30 ml) and then
stirred for 2 h under reflux in ethanol. The solid was again isolated by centrifugation, dried for 12 h at
-
1
-1
9
0°C and calcined for 6 h at 650°C in air (100 ml.min , 5°C.min ). The copper oxide supported on -
alumina (20%) CuO/Al was prepared by impregnating 1 g of -alumina with 2 ml of an aqueous
solution containing 1 mmol of Cu(NO .2.5H O [37]. The paste formed was left to stand for 1 h at
5°C. under a humid atmosphere, then dried for 24 h at 120°C and calcined for 9 h at 500°C in air (100
2
O
3
3
)
2
2
2
-1
-1
ml.min , 5°C.min ).
The silver supported on silica Ag/SiO
2
was prepared by impregnation of 5 g of SiO
2
(Cab-O-Sil) with
an aqueous solution (10 ml) containing 1.6 mmol of AgNO
3
[21]. The paste was then held for 30 min at
-1
-1
2
5°C, then dried for 24 h at 120°C and calcined for 3 h at 350°C, under air (100 ml.min .5°C. min ),
-1
then reduced under hydrogen for 5 h at 350°C (5 ml.min ).
Finally, the α and β silicon carbides (and SiC) were respectively provided by Sigma Aldrich and the
company SICAT. The solids were used without further treatment. Carbon nanotubes (CNT) were
provided by Pyrograf Products Inc. (PR-24-XT-PS). Before use, the nanotubes were washed for 2 h in
-1
nitric acid (100 ml.g CNTs) at reflux. They were isolated by centrifugation and then washed with
distilled water until a supernatant of pH = 6 was obtained and dried for 12 h at 120°C.
8

2.2. Catalysts characterization
The nature and purity of the synthesized phases was checked by X-ray diffraction using a
BRUKER D5005 diffractometer with a Ni-filtered CuK (0.15418 nm) radiation and by atomic
absorption (ICP) for chemical analysis. The specific surface areas were measured applying the BET
method with nitrogen adsorption. The conventional constant acceleration Mössbauer spectroscope used
5
7
to characterize iron in the catalysts was home-made [38]. It used a Co/Rh -X-ray source and allowed
spectra recording at room temperature. The quantitative analysis of the spectra was made integrating the
areas under the individual peaks and assuming equal recoil-free fraction for all species.
2.3.Catalytic testing
The oxidative dehydrogenation of ethyl lactate in ethyl pyruvate has been conducted on all
the synthesized compounds in the same conditions using a setup developed in previous studies [39]. In
order to develop a competitive industrial process, air was used as oxidizer and the gas stream
composition was above the upper explosive limit of ethyl lactate in the air. It corresponded to a volume
ratio of ethyl lactate / O / inert = 12.3 / 18.4 / 69.3. The reaction temperature was varied between 200
2
-1
and 350°C and the catalyst mass and total gas flow were respectively equal to 250 mg and 65 ml.min .
Any change in these operating conditions will be specifically mentioned. The selectivity and activity
data were given after 8 h on stream and stability was evaluated over 22 h.
The condensable reaction products were separated using two traps connected in series, filled with
acetonitrile and placed in an ice bath at 0-2°C. The output of the last trap was connected to a μGC-TCD
allowing online analysis of gaseous products that cannot be trapped. The μGC was SRA R3000
chromatograph equipped with a molecular sieve, a Porapak Q and a TCD detector. Nitrogen was used
as an internal standard. The analysis of the condensed products was carried out on a Shimadzu GC-2014
chromatograph, equipped with a Shimadzu AOC-20i autosampler. The column used was a Nukol type
-
1
(
30 m x 0.53 mm x 0.5 μm) polar capillary column with a flow rate of 3.0 ml.min and a split of 50, for
an injection of 1 μL. The temperature of the injector and of the FID detector was 220°C. Tetrahydrofuran
THF) was selected as an external standard. The carbon balances for all the catalytic properties evaluated
(
9

were greater than 98%. Ethyl lactate was not converted between 250 and 350°C in the absence of
catalyst.
3. Results and discussion
3.1. Catalyst screening
The twenty prepared catalysts were tested between 200 and 400 °C. By studying the variations
in their catalytic properties as a function of reaction temperature, the maximum ethyl pyruvate yields,
which can be obtained under the aforementioned experimental conditions, were determined. In general,
an increase in reaction temperature leads to an increase in conversion and to a decrease in selectivity to
ethyl pyruvate, with a preference for over-oxidation products, thus allowing the maximum yield
temperature to be determined.
When designing a catalyst screening test for a given reaction, one of the greatest difficulty
encountered is that of selecting the parameters most relevant to the comparison of catalytic
performances. Potential parameters include the temperature, gas flow composition, and contact time,
under which the catalysts are tested. In order to achieve fair comparisons in the present study, the 20
different catalysts were tested under highly similar operating conditions (flow of reagents, catalyst mass,
etc.), which were also chosen in view of their potential industrialization. For each catalyst, the reaction
temperature was varied, since it is known that from one catalytic system to another, selectivity to ethyl
pyruvate can increase, decrease, or pass through a maximum, as a function of temperature. For this
reason, the temperature range over which the best yield is achieved varies for each different catalyst.
Two criteria were chosen for the evaluation of each catalytic system’s properties: maximum
selectivity to ethyl pyruvate, and maximum yield. The highest selectivity obtained with a catalyst is an
important parameter, especially in the case of a process including one or more recycling sequences.
Maximum yield and productivity are also very useful parameters for the assessment of a catalyst’s
1
0

efficiency, and its potential use in an industrial process. The results of the screening evaluation are thus
presented in the form of two separate figures, which for each catalyst (over the studied temperature
range) show in the first one the highest achievable selectivity, as a function of the conversion rate at
which this selectivity occurs and in the second, the catalyst’s selectivity as a function of the conversion
rate at which the maximum yield occurs. The temperature at which each of the maximum values of
selectivity was obtained is noted in the first figure, close to the corresponding data point, and the reaction
-1
-1
temperature, the maximum yield and the productivity expressed in mol.s .g , are indicated in the second
figure. Finally, in order to improve the presentation of these results, the catalysts are grouped into two
distinct families, according to whether or not they contain V and Mo. The results obtained for the
families of catalysts that contain and do not contain V or Mo are respectively presented in Fig. 1 and 2.
All the catalysts tested in this study appear to be active and selective for the reaction of interest, and in
all cases ethyl pyruvate was the main product. The main by-products were systematically acetaldehyde,
CO , ethylene and ethanol. In certain cases, small amounts of acetic acid were detected. A reaction
2
scheme, with reaction paths for each of these by-products, is proposed in Fig. 3.
In this reaction scheme, the main secondary reaction is the hydrolysis of esters to acids and ethanol,
which takes place mainly on strong acidic sites. The resulting lactic and pyruvic acids then break down
into CO and ethanol or acetaldehyde. Very small amounts of lactic acid and pyruvic acid were detected,
2
but not quantified. The ethanol produced in this reaction can be dehydrated to ethylene on acid sites, or
oxidized to acetaldehyde. The acetaldehyde can be oxidized to acetic acid.
The results obtained with V or Mo-containing catalysts show that the highest selectivity to
ethyl pyruvate was obtained with FePMo12
respectively 40% and 37% conversion at 250 °C and 300 °C (Fig. 1a). The best yields were obtained on
MoO at 300°C with 63% and on the two former catalysts with respectively 57 and 56% at 275 and
00°C (Fig. 1b). On catalysts containing no V or Mo, lower selectivities and maximum yields were
O
40 and FeVSbO
4
catalysts with 94% and 90%, with
3
3
systematically observed. Although a selectivity of 84% for ethyl pyruvate was observed at 295 °C on
silver supported on silica, the maximum yield with this catalyst was much lower than that observed with
the best catalysts from the first series. This type of silver catalyst was also found to be less efficient than
1
1

what has previously been reported in the literature [23]. The reaction conditions used in the latter study,
although certainly more appropriate for this type of catalyst and reaction, would not be suitable for an
industrial application, and were different to those used in the present study.
The studied catalysts were stable on stream for about 22 h, with the exception of Te
2
Mo
2
O
7
and MoO .
3
In the first case, deactivation has already been reported for the same reaction, and attributed to the
reduction and volatilization of Te under the reaction flow, even in the presence of oxygen [18]. In the
second case, it is thought that Mo was lost through the gas volatilization of molybdenum oxy-hydroxides
species formed at high temperatures in the presence of water. Indeed, Mo was observed at the cold spot
at the end of the reactor, when it was run for a long time at high temperatures.
With a selectivity of 95% at 30% conversion, together with a productivity similar to that
observed with other catalysts described in the literature, FePMo12O40 appears to be the most selective of
the twenty catalysts examined in this study. The remainder of this study is thus devoted to a more
detailed analysis of its catalytic properties and to their improvement. Furthermore, this type of
polyoxometalate is well suited to the use of a wide variety of different compositions obtained by
changing the nature of the counter-cations while keeping the same primary structure and its catalytic
properties could potentially be improved by such modification.
3.2. Optimization of phosphomolybdic polyoxometalate catalysts
Although the highest selectivity to ethyl pyruvate was observed with FePMo12O40 , and this
could hardly be improved, its activity over the optimal temperature range needs to be improved.
40 has a low specific surface area, which is certainly responsible for this low activity and all
FePMo12
O
attempts to improve it by changing the synthesis conditions, have failed. However, a high, stable specific
surface area can be achieved with polyoxometalates, by synthesizing them in the form of an alkaline
salt [40]. The effect was the most important in the case of cesium salt since it exhibited the highest
specific surface area [41,42]. For this reason, a series of phosphomolybdic polyoxometalate catalysts
was prepared, using both cesium and iron as counter-cations: two Cs cations per Keggin unit were used,
with variable amounts of iron introduced as counter-cations.
1
2

XRD characterization of the catalysts showed that, in contrast to FePMo12
crystallizes as a mixture of phosphomolybdic acids hydrated with either 13 or 30 water molecules (ICSD
28-0082 and 015-4751), the compounds containing cesium only have the same cubic structure as
O40, which
0
polyoxometalate salts (ICSD 014-3585) (Fig. 4). The chemical analysis and results of the specific
surface area measurements of the catalysts are presented in Table 1. Our chemical analyses showed that
their bulk compositions are closely matched to their nominal compositions. The FePMO12O40 catalyst
used corresponds to a heteropolyacid, in which all of the protons have been substituted by iron. This is
possible, thanks to the very high flexibility of the secondary structure of this acid. This type of compound
2
-1
has a low specific surface area (in the present case 2 m .g ), which is detrimental to a high level of
catalytic activity (Table 1).
Table 1: Results of specific surface area (SSA) measurements, cell parameter calculations, thermo-gravimetric
analyses (number of water molecules lost per Keggin unit) and chemical analyses (atomic ratios) of the
synthesized samples.
a
SSA
Cell parameter
H
2
O/K.U.
Catalyst
FePMo12
Cs
-
1
(
m².g )
(Å)
Cs/P
-
Fe/P
1.0
Mo/P
11.5
O
40
2
-
13.4
2
Fe0.33
35
33
15
9
11.818(3)
11.818(3)
11.817(3)
11.814(3)
-
-
2.1
2.1
2.0
2.2
0.33
0.21
0.10
-
10.4
10.5
10.3
11.8
2
Cs Fe0.2
2
Cs Fe0.1
1.8 (5.5)
Cs
2
3.5 (10.5)
a: specific surface areas were measured after 8 h catalytic testing.
As could be expected, the presence of cesium in the polyoxometalate has a considerable influence on
2
-1
the specific surface area, with values up to 35 m .g being reached. DTG analyses of the FePMo12
and Cs Fe0.1 catalysts showed that the solid containing Cs was considerably less hydrated than that with
no Cs. In the latter, despite the presence of iron cations, water molecules were observed at a
concentration almost as high as that of the pure acid (15.2 H O per K.U.). All the infrared spectra showed
only the bands corresponding to the vibrations μas P-O (around 1060 cm ), μas Mo = O (around 968 and
O
40
2
2
-1
-1
-1
9
58 cm ) and μas Mo-O-Mo (around 859 cm ), characteristics of heteropolyphosphomolybdic
compounds with Keggin-type anions (Fig. S1).
1
3

The results of the catalytic testing of the Cs-containing catalysts are compared with those
obtained with FePMo12 40 in Table 2. The catalysts containing cesium appeared to be active and
selective. Although none of them reached the record selectivity of FePMo12 40, some of them had higher
ethyl pyruvate yields: as an example, a yield of 66 % was obtained on Cs Fe0.1 at 250°C. The catalysts
were very stable with time on stream, at least over a period of 22 h as illustrated for Cs Fe0.1 in Fig. S2.
In the case of the cesium-containing series, the addition of iron was found to have a positive effect on
selectivity, with a maximum reached on Cs Fe0.2, whereas the intrinsic rate of ethyl lactate conversion
O
O
2
2
2
decreased monotonically with iron content (Fig 5).
Table 2: Catalytic performance of polyoxometalates as a function of reaction temperature under standard testing
conditions - EtPa: ethyl pyruvate, ACHO: acetaldehyde, Ethyl: ethylene, Ethol: ethanol, AcA: acetic acid.
Compound
Temp. (°C) Conv. (%)
Selectivity (%)
2
Ethyl Ethol AcA Others
EtPa Yield (%)
EtPa ACHO CO
FePMo12
O
40
225
50
75
200
225
13.4
40.3
61.2
13 .7
31.4
45.3
71.0
13.7
40.0
62.1
76.9
30.9
54.3
87.1
93.5
47.0
67.4
82.2
96.7 1.0 1.0
93.7 1.6 1.7 1.1
0.5
0.5
1.5
1.9
0.3 0.0
1.8 0.0
0.0
0.0
1.0
0.3
0.4
1.3
2.3
2.0
2.0.
3.2
2.1
2.1
2.3
3.7
1.6
1.2
2.7
3.9
1.9
1.6
3.9
12.9
37.8
55.6
2
2
90.8
95.0
88.9
87.3
82.9
95.0
2.1 1.6 1.3
0.1 1 .2 1.1
2
Cs Fe0.33
1
3.0
1.9
2.8
3.7
1.0
2.7
2.7
27.9
39.5
58.9
13.1
2
2
50
75
2.9 3.1
3.4 3.0
1.1 0.5
2.2 2.2
2.5 2.8
1.9
0.0
2.5 1.3
0.3 0.0
1.3 0.0
1.8 1.0
2.2 2.1
1.1 0.0
1.5 0.4
1.9 2.3
3.5 3.6
0.1 0.2
1.8 1.4
1.7 3.1
2
Cs Fe0.2
200
25
250
75
200
2
90.4 1.6
87.2 2.4
3
6.2
54.2
61.0
29.0
48.8
66.9
58.4
42.3
57.8
57.5
2
79.3.
3.4 4.1 5.2
Cs
2
2
Fe0.1
94.0 1.1
89.8 1.7
1.2 1.0
2.5 2.9
4.8 8.3
8.6 13.2
2.4 3.2
3.6 4.4
6.2 11.0
225
250
275
76.9
62.5
90.0
85.8
69.9
3.1
4.7
2.1
2.4
4.3
Cs
185
00
25
2
2
This result is in agreement with the findings of other studies, showing that when iron is present in the
counter-cation position, it leads to a decrease in the rates of reduction and re-oxidation of similar solids,
and that this effect is improved when the iron content is increased [29]. All the iron-containing catalysts
were stable over the studied temperature range, and the structure of the material was maintained after
1
4

the reaction. This was not the case for the Cs compound, which was unstable at high temperatures and
2
could not be compared to the other compounds under these conditions.
In order to understand the differences in catalytic properties observed with these compounds,
in particular FePMo12
O
40 and Cs Fe0.33, the degree of oxidation and the environment of iron were
2
analyzed by Mössbauer spectroscopy, prior to and following catalytic testing. It was not possible to
perform similar analyses on the other substituted compounds, as their iron content was very low. The
isomeric (δ) and quadrupole moment (Δ) displacement values computed from the spectra are listed in
Table 3, and their spectra are shown in Fig. 6. The Mössbauer spectra of the Cs
2
Fe0.33 compound before
and after the catalytic test were adjusted with a single doublet with approximately the same hyperfine
3+
parameters characteristic of Fe ions.
Table 3: Mꢀssbauer parameters calculated from the spectra of the FePMo12
O
40 and Cs
2
Fe0.2 samples,
recorded at 25 °C.
-
1
 (mm.s-1)
Compound
FePMo12
Site
Fe
 (mm.s )
%
100
3+
O
40
Before test
After test
0.35
0.60
Fe3+
Fe2+
Fe3+
Fe3+
0.30
1.24
0.24
0.22
0.68
2.47
0.25
0.31
60
40
Cs
2
Fe0.33
Before test
After test
100
100
On the other hand, prior to the test, the spectrum of the FePMo12O40 was fitted to a ferric
doublet, whereas that of the solid after testing was fitted to two doublets - one ferric and the other ferrous.
The hyperfine parameters of the ferric doublets computed from the spectra were characteristic of Fe3+
in an octahedral coordination [38], whereas the ferrous doublet was characterized by a large quadrupole
splitting, typical of a highly distorted environment. The Mössbauer spectra of the Cs
2
Fe0.33 compound,
before and after the catalytic test, were fitted to a single doublet with approximately the same hyperfine
3+
parameters characteristic of Fe ions and comparable to those reported in the literature for iron counter-
cations in polyoxometalates [29].
1
5

3.3.Discussion
Although most of the catalysts tested during the screening phase of our study are known to
be efficient for oxidative dehydrogenation, their performances vary a lot in terms of activity and
selectivity. In the case of the best catalysts obtained from the screening (FePMO12 40, FeVSbO and
MoO ), in addition to the Ag/SiO catalyst proposed in the literature, there does not appear to be any
O
4
3
2
obvious, common feature explaining their catalytic properties. Although in all cases the latter properties
contribute to oxidative reactions by means of a Mars and van Krevelen mechanism, they involve distinct
6+
5+
5+
4+
+
0
redox couples: Mo /Mo , V /V and Ag /Ag , respectively, as well as different bulk constituents and
structures. This could lead to the conclusion that the reaction is relatively easy to catalyse, and that
different catalysts can be active in the reaction. The best catalysts would be those which are the most
active just below the temperature of decomposition of ethyl lactate or pyruvate, i.e. around 330 °C,
which is the case for the best catalysts described in this study. Their observed differences in productivity
could then be related simply to differences in active surface site densities, or intrinsic surface site
activities. In addition, all of the best catalysts have a low surface acidity. This feature was found to be
essential, in order to avoid the hydrolysis of esters through the formation of lactic and pyruvic acids,
and their subsequent oxidation to lighter by-products. MoO
40 is an heteropolyacid, in which the protons have been completely replaced by iron cations,
thus suppressing all strong acidity, whereas the FeVSbO catalyst with the chosen Fe/V ratio (0.66) has
been shown to have only very weak acid sites [25]. Ag/SiO has also been shown to be weakly acidic
44,23]. Conversely, WVNbO and TiP , which are the least selective among the active catalysts in
3
is known to have only weak acid sites [43].
FePMO12
O
4
2
[
O
2 7
our selection, are also the most acidic [45,46].
Our efforts to improve the characteristics of the best catalyst, i.e. FePMO12O40, were
successful in so far as the ethyl pyruvate yield was increased to 67%. This was achieved by adding
cesium counter-cations. Alkali metals can also be used to substitute protons, and in the present study
were found to precipitate, forming phases with a cubic salt structure and very fine particles (ranging
from 10 to 40 nm in diameter). When the protons are not completely substituted by alkali metals, the
prepared solids correspond to the hydrated acid together with the pure salt, with the acid phase coating
1
6

the salt particles [40]. Although the acid cannot be detected by X-ray diffraction, it can be observed by
XPS or Raman spectroscopy [40, 47]. When it is added at the same time as the monovalent cation
substitute, the iron has been shown to replace the protons in the supported acid [40]. Although in both
cases, at a macroscopic level, the transition metals do not appear to have a significant influence on the
structural characteristics of the acid, other than decreasing its degree of hydration, they have a strong
influence on the catalytic properties of the solids in the oxidation and oxidative dehydrogenation
reactions. In the polyoxometalate series studied here, the protons are progressively substituted by iron
cations in the supported acid to Cs Fe0.33, which should not contain any more protons. This evolution
2
explains the observed increase in the catalyst’s selectivity as a function of increasing iron content,
although it did not reach the level of the unsupported acid as it should.
In FePMo12O40 the iron is present mainly in the form of an octahedral complex hexahydrate
Fe(H
2
O)
6
3+, located between Keggin units. Alternatively, if a water molecule is lost, it adopts the form
O) 3
+
complex bound by terminal oxygen to a heteropolyanion (Fig. 7a) [29, 47]. In this case,
of a Fe(H
2
5
DFT calculations have shown that electron transfers between iron and molybdenum can take place:
Mo6+ + Fe2+ [48]. This is the reason for which the iron is partially reduced after catalytic
Mo5+ + Fe3+
testing. This electron transfer has a noticeable influence on the redox activity of Mo, since the intrinsic
activity of FePMo12 40 is much higher than that of the iron substituted acid supported on cesium salt.
In the latter compounds, it has been shown that the iron had a different environment and was present in
O
the form of (FeOH) or (FeOH
)+ species (Fig. 7b) [48-50]. This difference may be related to changes
2
+
2
brought about by the presence of cesium salt in the structure of the supported acid phase, leading to a
strong decrease in the space between the Kegging units. This assumption is supported by the results of
the DTG analysis, which show that the hydration water content per Keggin unit of acid is much lower
for the supported acid than for the bulk acid. In this case there are no electron transfers, as confirmed by
the results of Mössbauer spectroscopy. However, it is important to note that the protons remain, even in
the Cs
2
Fe0.33 compound, thus leading to a higher acidity than in FePMo12O40. The acidity of the Cs
2
Fe
x
compounds has previously been studied using the dehydration of isopropanol to propene as model
reaction [29] (Fig. S3). The rate of propene formation and thus the acidity of the compounds decreased
when iron content increased up to Cs Fe0.2. At higher iron content, the loss of protons substituted by
2
1
7

iron was compensated by hydroxyl groups bound to iron and the acidity did not decreased anymore [29].
In this respect, the best selectivity to ethyl pyruvate was obtained with Cs Fe0.2, which confirmed that
2
the less acidic catalysts were well the most selective. It is highly likely that similar phenomena would
be observed with transition metal cations other than iron, which does not allow to wait any progress
with other substitutions.
4. Conclusion
The screening of various catalysts for the oxidative dehydrogenation of ethyl lactate has led
to the identification of several candidates, which could be sufficiently efficient for use in industrial
application, especially if an easy separation and recycling of unconverted ethyl lactate is possible.
Among these catalysts, iron-substituted phospho-molybdic heteropoly acid FePMo12
O
40
appears to be the candidate having the highest efficiency. The performance of this catalyst was
optimized by means of iron and cesium co-substitution, leading to higher ethyl pyruvate yields under
similar reaction conditions. These catalysts, which correspond to iron-substituted acids supported on
cesium salt, are substantially more active. This characteristic is related to their significantly higher
specific surface area, since the intrinsic activity of their active sites is lower and the selectivity is
decreased. The higher intrinsic activity of active sites of FePMo12O40 is likely related to an electron
transfer between iron and molybdenum that is not present for iron-substituted acid supported on cesium
salt. Considering the selectivity, it has been observed that it depends highly on the acidity of the catalysts
since the later promotes side-reactions like hydrolysis that decreases the selectivity to ethyl pyruvate.
Consequently, the best catalysts appear the ones with numerous redox Mo sites and relatively low acidity
i.e. Cs
area and number of active sites and a higher selectivity to ethyl pyruvate. This is due to a lower intrinsic
activity of the active sites. Finally, FePMo12 40 appears as the most selective catalyst because of its
lowest acidity related to its lowest number of protons.
2
Fe0.1. The compounds Cs
2
Fe0.2 and Cs
2
Fe0.33 are less efficient despite a higher specific surface
O
1
8

In addition to its concrete application for the design of an efficient catalyst for the oxidative
dehydrogenation of ethyl lactate, this study provides an example of the importance of the influence of
surface acidity in oxidative dehydrogenation catalysts, and shows how this acidity can vary according
to the nature of the counter-cations used in polyoxometalate-based catalysts.
CRediT author statement
M. Huchede Visualization, Data Curation, investigation, Writing - Original Draft
D. Morvan, Project administration, Visualization, Validation, data curation
V. Belliere-Baca. Funding acquisition, Supervision
J.M.M. Millet Funding acquisition, Project administration, Supervision, Visualization,
Writing - Review & Editing, Conceptualization, data curation
Signification
This paper is original. It presents the screening of twenty catalysts for the gas phase oxidative
dehydrogenation of ethyl lactate, which are for most of them never tested in this reaction.
Ethyl lactate should be an important plateform molecule in the future for the chemical industry and it is
interesting to report a new reaction to use it industrially; The best candidate resulting from this screening
was a poly-oxometallate based catalyst, which turns out to be one of the most efficient ever proposed
for this reaction. It was shown that this catalyst could be modified and to optimized
5. Acknowledgment
The authors gratefully acknowledge the ADISSEO company for the financial support it
provided for the present study.
6. References
[
1] Y. Fan, C. Zhou, X. Zhu, Catal. Rev. 51 (2009) 293-324.
1
9

[
2] M. Dusselier, P. Van Wouwe, A. Dewaele, E. Makshina, B.F. Sels, Energy Environ. Sci. 6 (2013)
415-1442.
1
[
(
3] T. Bonnotte, S. Paul, M. Araque, R. Wojcieszak, F. Dumeignil, B. Katryniok, ChemBioEng Rev. 5
2018) 34-56.
̈
[
[
[
4] P. Maki-Arvela, I.L. Simakova, T. Salmi, D. Yu. Murzin, Chem. Rev. 114 (2014) 1909-1971.
5] E. Blanco, P. Delichere, J.M.M. Millet, S. Loridant, Catal. Today 226 (2014) 185-191.
6] D.T. Vu, A.K. Kolah, N.S. Asthana, L. Peereboom, C. T. Lira, D. J. Miller, Fluid Phase
Equilibria 236 (2005) 125-128.
[
[
7] B.M. Murphy, M.P. Letterio, B. Xu, ACS Catal. 6 (2016) 5117-5131.
8] W. Zhang, G. Innocenti, P. Oulego, V. Gitis, H. Wu, B. Ensing, F. Cavani, G. Rothenberg, N.R.
Shiju, ACS Catal. 8 (2018) 2365-2374.
[
9] E. Blanco, C. Lorentz, P. Delichere, L. Burel, M. Vrinat, J.M.M. Millet, S. Loridant, Appl. Catal. B:
Env. 180 (2016) 596-606.
[
10] Industrial Solvents Handbook" by E.W. Flick. 5th Edition. William Andrew Inc., 1998. ISBN 0-
155-1413-1, ISBN 978-0-8155-1413-8
8
[
11] S.P.J. Ahmadkalaei, S. Gan, H.K. Ng, S.A. Talib, Environ. Sci. Pollut. Res. 23 (2016) 22008-
2018.
2
[
[
12] T. Mallat, A. Baiker, Chem. Rev. 104 (2004) 3037-3058.
13] T. Tsujino, S. Ohigashi, S. Sugiyama, K. Kawashiro, H. Hayashi, J. Mol. Catal. 71 (1992) 25-35.
2
0

[
14] S. Sugiyama, T. Kikumoto, H. Tanaka, K. Nakagawa, K.I. Sotowa, K. Maehara, Y. Himeno, W.
Ninomiya, Catal. Lett. 13 (2009) 1129-1134.
[
[
[
[
[
[
15] C. Zhang, T. Wang, Y. Ding, J. Appl. Catal. A: Gen. 533 (2017) 59-65.
16] S.I. Fang, C.F. Hsieh, C.C. Shih, China Petrochemical Corp., US20140031581, 2014.
17] F. Thalhammer, K.-H. Schrank, K. Wernthaler, Sueddeutsche Kalkstickstoff, DE19756584, 1999.
18] H. Hayashi, S. Sugiyama, N. Masaoka, N. Shigemoto, Catal. Lett. 19 (1993) 273-277.
19] S. Lomate, T. Bonnotte, S. Paul, F. Dumeignil, B. Katryniok. Sustain. Chem. Proc. 3 (2015) 1-8.
20] M. Qian, M.A. Liauw, G. Emig, Appl. Catal. A : Gen. 238 (2003) 211-222.
[
21] T.T.N. Nguyen, M. Huchede, E. Blanco, F. Morfin, J.L. Rousset, L. Massin, M. Aouine, V.
Belliꢁre-Baca, J.M.M. Millet, Appl. Catal. A: Gen. 549 (2018) 170-178.
[
22] G.J. Millar and M. Collins, Ind. Eng. Chem. Res. 56 (2017) 9247-9265.
[
23] J. Shen, W. Ji, Y. Han, Y. Chen, Nanjing University CN1359893, 2002.
[
[
[
[
[
[
24] L.E. Briand, J.M. Jehng, L. Cornaglia, A.M. Hirt, I.E. Wachs, Catal. Today, 78 (2003) 257-268.
25] H. Roussel, B. Mehlomakulu, F. Belhadj, J.M.M. Millet, J. Catal. 205 (2002) 97-106.
26] H. Hayashi, S. Sugiyama, N. Masaoka, N. Shigemoto, Ind. Eng. Chem. Res. 34 (1995)135-139.
27] K. Chen, S. Xie, A.T. Bell, E. Iglesia, J. Catal. 198 (2001) 232-242.
28] J.M.M. Millet, J.C. Védrine, Appl. Catal. 76 (1991) 209-219.
29] M. Langpape, J.M.M. Millet, U.S. Ozkan, P. Delichère, J. Catal. 182 (1999) 148-155.
2
1

[
(
30] O. Benlounes, S. Cheknoun, S. Mansouri, C. Rabia, S. Hocine, J. Taiwan, Inst. Chem. Eng. 42
2011) 132-137.
[
31] D. Nagaki, H. Weiner, J.T. Chapman, M.O. Scates, A.S. Locke, C.J. Peterson, Celanese
International Corporation, US20120277466, 2012.
[
32] A. Chieregato, F. Basile, P. Concepciꢂn, S. Guidetti, G. Liosi, M.D. Soriano, C. Trevisanut, F.
Cavani, J.M. Lꢂpez Nieto, Catal. Today, 197 (2012) 58-65.
[
[
[
[
33] B. Deniau, G. Bergeret, B. Jouguet, J.L. Dubois, J.M.M. Millet, Top. Catal. 50 (2008) 33-42.
34] I.-C. Marcu, I. Sandulescu, J.M.M. Millet, J. Mol. Catal. A: Chem. 203 (2003) 241-250.
35] D. Rouzies, J.M.M. Millet, D. Siew Hew Sam, J.C. Vedrine, Appl. Catal. 124 (1995) 189-203.
36] L.D. Nguyen, S. Loridant, H. Launay, A. Pigamo, J.L. Dubois, J.M.M. Millet, J. Catal. 237 (2006)
38-48.
[
[
37] A.E. Marcinkowsky, J.P. Henry, Union Carbide Corporation, US4220803, 1980.
38] J.M.M. Millet, C. Virely, M. Forissier, P. Bussière, J.C. Vedrine, Hyperfine Interact. 46 (1989)
619-628.
[
39] P. Lauriol-Garbey, S. Loridant, V. Belliꢁre-Baca, P. Rey, J.M.M. Millet, Catal. Comm. 16 (2011)
70-174.
1
[
[
40] M. Langpape, J.M.M. Millet, U.S. Ozkan and M. Boudeulle, J. Catal. 181, (1999) 80-90.
41] M. Mizuno, M. Misono, Chem. Rev. 98 (1998) 199-218.
[
[
42] S. Paul, W. Chu, M. Sultan and E. Bordes-Richard, Sci. Chin. Chem. 53 (2010) 2039-2046.
43] K. Routray, W. Zhou, C.J. Kiely, W. Grꢃnert, I.E. Wachs, J. Catal. 275 (2010) 84-98.
2
2

[
44] W. Janssens, E.V. Makshina, P.Vanelderen, F. De Clippel, K. Houthoofd, S. Kerkhofs, J.A.
Martens, P.A. Jacobs and B.F. Sels, ChemSusChem 8 (2015) 994-1008.
[
45] A. Chieregato, M.D. Soriano, F. Basile, G. Liosi, S. Zamora, P. Concepciꢂn, F. Cavani, J.M. Lꢂpez
Nieto, Appl. Catal. B: Env. 150-151 (2014) 37-46.
[
46] I.-C. Marcu, I. Sãndulescu, Ana. Univ. Bucureúti, Chimie, Anul XII (serie nouă), vol. I-II, pag.
09-313.
3
[
[
[
47] Staroverova I.N., Kutyrev M., Stakheev A., Kinetika i Kataliz 33 (1) (1992) 127-133.
48] S.A. Borshch, H. Duclusaud, J.M.M. Millet, Appl. Catal. A: Gen. 200 (2000) 103-108.
49] E. Etienne, F. Cavani, R. Mezzogori, F. Trifirꢄ, G. Calestani, L. Gengembre, M. Guelton, Appl.
Catal. A: Gen. 256 (2003) 275-290.
[
50] Q. Huynh, J.M.M. Millet, J. Phys. and Chem. Solids 66 (2005) 887-894.
2
3

Figure 1: Catalytic properties of catalysts containing V and Mo: (a) best selectivity to ethyl pyruvate as
a function of ethyl lactate conversion, showing the reaction temperature (°C) at which it was obtained,
(
b) selectivity to ethyl pyruvate as a function of ethyl lactate conversion at best yield, showing the
reaction temperature (°C) at which it was obtained, together with the corresponding yield in %.
Fig.1 a
Fig.1b
2
4

Figure 2: Catalytic properties of catalysts that do not contain V or Mo: (a) best selectivity to ethyl
pyruvate as a function of ethyl lactate conversion, showing the reaction temperature (°C) at which it was
obtained, (b) selectivity to ethyl pyruvate as a function of ethyl lactate conversion at best yield, showing
the reaction temperature (°C) at which it was obtained, together with the corresponding yield in %.
Fig.2 a
Fig.2 b
2
5

Figure 3: Reaction scheme for the formation of quantized by-products in the oxidative dehydrogenation
of ethyl lactate to ethyl pyruvate for all tested catalytic systems.
Figure 4 : X-ray diffraction patterns of the prepared phosphomolybdic polyoxometalates : a)
FePMo12
O40, b) Cs
2
Fe0.3, c) Cs
2
Fe0.3 and d) Cs Fe0.33.
2
2
6

Figure 5: a) Rate of ethyl lactate conversion and selectivity to ethyl pyruvate b) selectivity to by-
products as a function of the iron content of the catalysts. ethanol (), ethylene (), CO (),
2
acetaldehyde (), acetic acid () and others (). Catalytic reaction conditions: ethyl lactate / O /
2
inert = 12.3 / 18.4 / 69.3 and temperature =225 °C.
Fig.5 a
Fig.5 b
2
7

Figure 6: Mössbauer spectra of two compounds: Cs Fe0.33 a) and b) and FePMo12O40 c) and d), recorded
2
at 25 °C before and after catalytic testing; solid lines are derived from least-squares fits.
Figure 7: Model clusters representing the interaction between the iron counter-ion and the Keggin unit
in the bulk (a) and in the supported iron-substituted heteropolyacid (b).
2
8