Selective oxidation of propene to acrolein on FeMoTeO catalysts: determination of active phase and enhancement of catalytic activity and stability

M. Tonelli a, M. Aouine a, L. Massin a, V. Belliere Baca b and J. M. M. Millet *a
aInstitut de Recherches sur la Catalyse et l'Environnement de Lyon, CNRS, 2 Avenue A. Einstein, Villeurbanne F-69626 Cedex, France. E-mail: jean-marc.millet@ircelyon.univ-lyon1.fr; Fax: (+33)472445399; Tel: (+33)472445317
bAdisseo France SAS, Antony Parc 2, 10 Place du Général de Gaulle, F-92160, Antony, France

Received 3rd August 2017 , Accepted 31st August 2017

First published on 31st August 2017


Multicomponent FeMoTeO catalysts have been synthesized and studied for mild propene oxidation to acrolein. Incorporation of silica into the catalyst at a precise stage in the synthesis process allowed dispersing the active phase without modifying it and keeping the active sites accessible. The results showed that catalysts with selectivity higher than 83% at 80% conversion could be obtained while the catalytic reaction temperature was decreased from 450 to 360 °C. In addition to these remarkable properties, the catalysts appear, after an initial decrease of activity, to be stable over several hundred hours. The characterization of the catalysts showed that the active phase corresponds to an amorphous layer of molybdenum oxide containing Te and Fe, while the Fe2(MoO4)3 phase only has a supporting role allowing stabilization of the amorphous layer with the appropriate composition. These results stimulate renewed interest in this catalytic system and open new perspectives on new applications.


1. Introduction

The oxidation of propene to acrolein and acrylic acid using molecular oxygen has attracted renewed interest from industry due to the current pressure on propene production and pricing. The market demand for propene is high and will continue to increase in the coming years especially in Asia. At the same time, it is anticipated that the supply of propene that is mainly produced as a by-product either from steam crackers for ethylene production or from fluid catalytic crackers, may not be fulfilled. This situation will drive the prices up and is currently boosting research on both new ways to produce propene or directly acrolein and acrylic acid. New processes of production of propene have been studied and are today implemented industrially on a large scale; examples of such industrial production are the propane dehydrogenation process (PDH) and the methanol to propene process (MTP).1–3

On the other hand new processes of production of acrolein and acrylic acid from biomass feedstock are developed but are still at laboratory testing.4–7 A final solution is to go back to the standard method of oxidation of propene to acrolein and to optimize it by exploiting the progress made in our understanding of oxidation catalysts. However the benchmark nowadays to develop a new catalyst is the multicomponent bismuth molybdate system, which has been used all over the word for forty years and it is very difficult to compete with such a system. Catalytic systems similar to those developed decades ago are studied again. For example copper on silica or BiVMoO catalysts have recently been investigated.8,9 Mixed tellurium-metal oxides, which exhibited promising properties in terms of selectivity to acrolein have not yet been revisited. The first patents on mixed tellurium-metal oxides relevant to their use in the allyl oxidation or ammoxidation of propene were filed at the beginning of the 1960s.10–16 Accordingly, in the next decade, a number of studies appeared in the literature exploring the catalytic properties of such Te-doped oxidation catalysts. It was generally found that binary or ternary oxide compositions including tellurium were characterized by a much better selectivity towards partial oxidation products than the simple metal oxides. For instance, Robin et al. showed that the addition of tellurium to Ti, V, Mo, Sn, Sb or W oxides provided catalysts exhibiting high selectivity to acrolein (75–96%) at low conversion of propene.17 Forzatti and Trifirò studied the influence of the addition of tellurium to Cd-, Mn-, Zn- and Co-molybdates, obtaining four isostructural MIITeMoO6 phases that gave a selectivity to acrolein around 90% at 60% propene conversion (GHSV = 2400 h−1T = 470 °C). These authors also noticed that the corresponding undoped MIIMoO4 compounds were poorly active with respect to propene oxidation and mainly produced deep oxidation products.18 Several authors observed an even more striking variation of the catalytic properties upon addition of tellurium to ferric molybdate; indeed, Fe2(MoO4)3 which is industrially employed for the oxidation of methanol to formaldehyde, exhibits a quite low selectivity to acrolein when employed as a catalyst in the oxidation of propene. However, addition of only about 2.5 weight% tellurium oxide to ferric molybdate resulted in the creation of new active sites that increased its intrinsic activity and made it very selective for acrolein (S > 93%).19–22 These results indicate that tellurium in some way mimics the role of bismuth in Bi2(MoO4)3.23,24 Kinetic studies on Te-containing catalysts have shown that, similar to bismuth molybdates, the rate-limiting step of the reaction is the α-hydrogen abstraction. The oxidation of propene occurs via a Mars–van Krevelen reaction mechanism where lattice oxygen is inserted into the activated hydrocarbon molecule, as indicated by the independence of the rate of acrolein formation from the partial pressure of oxygen. Isotopic studies with 18O2 showed that 16O was still incorporated into both acrolein and CO2 even if the oxide surface was almost completely substituted with 18O atoms, indicating high lattice oxygen mobility in Te-doped metal oxides.25,26

Similar studies were performed by Ueda et al. on Te-doped molybdates and it was found that catalysts exhibiting the highest mobility of lattice oxygen were also the most active and selective for the formation of acrolein.20,22 Two main issues have limited the use of tellurium in industrial catalytic applications: its price and its volatility under operational conditions that causes loss of tellurium from the catalytic bed in the form of metallic Te.27,28 Catalysts for selective oxidation almost invariably need the presence of metals with a covalent character to perform mild oxygen insertion in the activated hydrocarbon molecule. Such elements, however, are much more prone to volatilization and the problem of a slow but continuous loss of the active metal is typically encountered in other catalytic compositions containing Sb, Se, Bi or Mo. In the case of tellurium one practical industrial solution has been to add small amounts of volatile tellurium precursors to the feed (on a continuous or intermittent base) in order to keep an ideal concentration of tellurium at the surface of the catalysts to provide the desired catalytic performances. Alternatively this challenge may be overcome by stabilizing tellurium on the catalyst by adding a powerful redox couple in the catalytic system, which is able to rapidly re-oxidize tellurium to its higher stable oxidation state. In this regard, addition of iron (Fe3+/Fe2+), cerium (Ce4+/Ce3+) or copper (Cu2+/Cu+) has proven to be very efficient in suppressing or reducing Te volatilization.19,20,29,30

In this paper we describe the results of studies, where we have revisited Te-containing iron molybdate catalysts in order to gain further insight and better knowledge of the active phase as well as to try to improve its activity without losing its high selectivity to acrolein. So far the highest conversions reported on these catalysts never have exceeded 40% at 470 °C; this limits the future of this system for industrial applications. We undertook to increase the conversion rate by incorporating during synthesis silica to disperse the active phase without modifying it and keeping the active sites accessible. We also focused our attention on the study of the stability of the catalytic system since earlier studies always reported a loss of activity without however quantifying it. Several Te-containing iron molybdate based catalysts have been synthesized, tested as catalysts and characterized before and after testing using various techniques like X-ray diffraction, XPS spectroscopy and transmission electron microscopy in order to understand the nature of the active sites and of the effect of silica.

2. Experimental

In a typical synthesis of the FeMoTeO catalysts, a solution was prepared by dissolving 2.15 mmol of ammonium heptamolybdate, (NH4)6Mo7O24, in 100 ml of distilled water (solution 1) and another was prepared by dissolving 10 mmol of iron nitrate, Fe(NO3)3·9(H2O), and varying amounts of telluric acid, H6TeO6, in 50 ml H2O (solution 2). In the syntheses where silica was employed as a dispersant agent, 5 mmol of a commercial colloidal silica solution (Ludox AS-30) was added to solution 1. Solution 2 was then mixed with solution 1 under vigorous stirring, resulting in the immediate precipitation of a yellow precipitate. Chemical and XRD analysis of this fine precipitate showed that it corresponded to an amorphous precursor of Fe2(MoO4)3. The solid was kept in the mother solution under stirring for 1 h at room temperature. Finally, the suspension was evaporated under vacuum and the obtained crystals were dried overnight at 120 °C and calcined for 3 h at 500 °C. The catalysts were designated according to their tellurium content and to the presence or absence of added silica. For example, Si-FeMoTe0.1 denotes a solid with the stoichiometry Fe2(MoO4)3 doped with 0.10 tellurium and containing SiO2.

The elemental composition of the catalysts was determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) on an ACTIVA JOBIN YVON spectrometer. The specific surface area of the samples was determined by nitrogen physisorption at −196 °C on a Micromeritrics ASAP 2020 instrument using the Brunauer–Emmett–Teller (BET) method. Prior to the measurement, the samples were desorbed in a vacuum at 300 °C for 3 hours. Samples containing tellurium were outgassed at 250 °C to avoid the possible sublimation of metallic tellurium at higher temperature. XRD diffraction patterns were recorded on a Bruker D8A25 X-ray diffractometer operating at 50 kV and 35 mA (Cu Kα radiation at 0.154184 nm), equipped with a Ni filter and a 1-D fast multistrip detector (LynxEye, 192 channels on 2.95°). The diffraction patterns were collected at 25 °C in the 4–80° 2θ range, with a step size of 0.02° 2θ and a dwell time of 2 s per increment.

Mössbauer spectra have been recorded with a homemade apparatus described elsewhere.31 Isomer shifts (IS), given with respect to αFe, and quadrupolar splitting (SQ) were calculated with a precision of 0.02 mm s−1. Scanning electron microscopy (SEM) analysis was performed on an ESEM-FEG FEI XL 30 microscope, using secondary electrons for imaging purposes. Prior to analysis, powders were coated with a film of sputtered gold in order to avoid charging effects. Transmission electron microscopy (TEM) images were acquired on a FEI TITAN ETEM G2 microscope, equipped with an objective Cs aberration corrector and operating at 80–300 kV. An energy-dispersive X-ray (EDX) analyzer from Oxford Instruments TM allowed elemental chemical analysis. Before each analysis, the catalysts was crushed and dispersed ultrasonically in pure ethanol. Then a drop of the suspension was deposited on 2–3 mm diameter carbon-coated copper grids (100–400 mesh). Ultramicrotomed cuts have also been realized for specific analyses and structure analyses. XPS measurements were carried out using a KRATOS AXIS Ultra DLD spectrometer, equipped with a detection system combining a hemispherical analyzer and a detector of the type “Detector Delay Line”. The XPS spectra of the Fe2p, Fe3p, Mo3d, Te3d, O1s and C1s levels were measured at 90° (normal angle with respect to the plane of the surface) using a monochromated Al Kα X-ray source with a pass energy of 20 eV and a spot size aperture of 300 × 700 μm. The measured binding energies (BE) were corrected with respect to the carbon 1 s signal at 284.6 eV. The Fe2p signal is composed of a number of broad and ill-defined satellites that make accurate determination of the background difficult.32 Deconvolution of the Fe 2p3/2 peak was however accomplished considering only the shake-up satellite occurring in Fe3+.33 The experimental precision of XPS quantitative measurements of Fe3+/(Fe3+ + Fe2+) has thus been considered to be only around 20%. The X-ray absorption spectroscopy (XAS) study was performed at the Synchrotron facility of the Paul Scherrer Institute in Zurich. The experimental conditions were set to acquire XAS spectra of Fe, Mo at K-edges and Te at L1-edges. In a typical acquisition, the samples were diluted with amidon (50% w/w), pelletized and analyzed in transmission mode for Fe and Mo and in fluorescence mode for Te. For in situ experiments, 10–20 mg of the catalyst were charged in a glass capillary under a C3H6[thin space (1/6-em)]:[thin space (1/6-em)]O2[thin space (1/6-em)]:[thin space (1/6-em)]N2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]1.5[thin space (1/6-em)]:[thin space (1/6-em)]8.6 gas mixture (total flow: 10 ml min−1). The temperature was fixed to 360 °C by means of a heat gun installed above the sample. In this case, acquisition of the XAS spectra was performed in the fluorescence mode for all the edges.

The mild oxidation of propene was performed in a fixed bed reactor operating at atmospheric pressure and under isothermal conditions using a glass plug-flow reactor. The oxidation was performed between 360 and 400 °C with the catalyst amounts varying from 200 to 3000 mg and the total flows varying from 6 to 125 mL min−1. The feedstock compositions were C3H6/O2/N2 + He = 30/20/50 and under standard testing conditions the gas hourly space velocity (GHSV) used was equal to 296 h−1. Propene (N25) was purchased from Air Liquide and used without further purification. The reaction products (acrolein, acetaldehyde, acrylic acid, allyl alcohol, CO, CO2) and unconverted propene were analyzed online using a gas chromatograph equipped with a TCD detector and either Porapak-Q and CP Molsieve 5A columns or a Carboxen 1010 column. A test of the empty reactor at 450 °C showed that the conversion of propene remained below 1.5% and that the main products of reaction were CO2 and CO with a selectivity of 38% and 15%, respectively. The main oxygenated products were acetaldehyde, allyl alcohol, and acrolein, with a selectivity of 13% 10% and 8%, respectively. The carbon balance was for all testing higher or equal to 98%.

3. Results and discussion

3.1. Catalytic properties of FeMoTeO catalysts

In their respective studies on FeMoTeO catalysts, Ueda and Forzatti reported that the most active and selective composition corresponded to that of FeMoTe01.19,22 In order to reproduce the results obtained we have compared the catalytic properties of the FeMo and FeMoTe01 compounds at 450 °C (Fig. 1). The catalytic properties of both samples evolved over time. A fast decrease of the conversion of propene was observed in the first hour of operation, followed by a further but slower decrease between 20 and 50 hours on stream and finally by stabilization after approximately 70 hours. A reversed trend was observed for the evolution of the selectivity to acrolein. The addition of tellurium clearly increased the selectivity to acrolein from 10 to 93% but the conversion, which was equal at the beginning of the testing for the two solids decreased much significantly for the Te containing one to about 12%. These results are in agreement with those published earlier if it is considered that the latter were obtained after only a few hours on stream. The increase in selectivity to acrolein during the first hour on stream was mainly correlated with a decrease of the selectivity to CO and CO2 (Fig. S1).
image file: c7cy01574g-f1.tif
Fig. 1 Evolution over time on stream of the propene conversion (circles) and acrolein selectivity (triangles) for the FeMoTe0.1 (black solid line) and FeMo (blue dashed line) samples. GHSV = 295 h−1, T = 450 °C and C3H6/O2/N2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]1.5[thin space (1/6-em)]:[thin space (1/6-em)]8.6.

The Si containing catalysts have been studied by considering first the Si-FeMoTe0.1 catalyst. The catalyst appeared extremely active at 450 °C and the reaction temperature had to be lowered to 360 °C. Under these conditions the catalyst was tested with two contact times (Table 1). When tested at the same contact time as that used for FeMo, FeMoTe0.1 exhibited an initial conversion close to 100%. After 50 h on stream this conversion has decreased to 87% and the selectivity increased to 68%. Si-FeMoTe0.1 was thus twice as active as the undiluted material, even though the operating temperature was lowered by 90 °C. A second important effect of SiO2 addition is the improved stability of the catalyst over time: the conversion of propene on the Si-FeMoTe0.1 catalyst only slightly decreased (ca. 5%) over 100–200 hours on stream. In a similar manner to that previously observed for the FeMo and FeMoTe0.1 catalysts, the selectivity increased with time to become stabilized around 83%.

Table 1 Catalytic properties of the FeMoTeO catalysts (C3H6/O2/N2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]1.5[thin space (1/6-em)]:[thin space (1/6-em)]8.6) after 200 h at 450 and 360 °C of testing and for Si-FeMoTe0.1 with different GHSV values
Sample T (°C) X C3H6 (%) S acro (%) GHSV (h−1) Acrolein productivity
(10−7 mol s−1 g−1) (10−7 mol s−1 cm−3)
FeMo 450 34 10 295 1.2 2.1
FeMoTe0.1 450 13 93 295 3.1 4.7
Si-FeMoTe0.1 360 82 82 295 17.6 22.3
Si-FeMoTe0.1 360 62 88 395 19.2 24.2


By reducing by one third the contact time, the propene conversion was decreased to 62%, but at the same time the selectivity to acrolein reached 88% (Table 1). This was like before the result of a decrease in the production of deep oxidation products, in particular CO2. Stabilization also occurred more rapidly at a shorter contact time.

Within the Si-FeMoTe series, we have varied the Te content from 0.0 to 0.3 moles per mole of Fe2(MoO4)3, with the goal of understanding the change in the catalytic properties as a function of the concentration of tellurium. First the evolution over time of the catalytic properties of the samples was studied (Fig. 2). This evolution appearing similar to a rapid decrease in the conversion of propene was observed in the first 5–10 hours of reaction and the selectivity to acrolein increased to become completely stabilized only after ca. 120 hours of operation. The gain in selectivity over time was more pronounced for the samples with lower Te content whereas the decrease in conversion followed the opposite trend. These results suggested that both the activity and the selectivity of the catalyst were affected by the tellurium concentration. Indeed, this precisely emerged from Fig. 3, comparing the values of propene conversion and acrolein selectivity as a function of Te loading. From these data it is clear that addition of tellurium to silica-dispersed ferric molybdate catalysts always led to higher values of propene conversion and to a progressive increase in the selectivity to acrolein.


image file: c7cy01574g-f2.tif
Fig. 2 Evolution over time on stream of the propene conversion (circles) and acrolein selectivity (triangles) for the Si-FeMoTe0.2 catalyst (black solid line) and the Si-FeMoTe0.3 catalyst (blue dashed line). Test conditions: GHSV = 295 h−1, T = 360 °C and C3H6/O2/N2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]1.5[thin space (1/6-em)]:[thin space (1/6-em)]8.6.

image file: c7cy01574g-f3.tif
Fig. 3 Evolution of propene conversion and of selectivity to acrolein and the major by-products as a function of the Te content for the Si-FeMoTe catalysts after 120 h of stabilization. Test conditions: GHSV = 295 h−1, T = 360 °C and C3H6/O2/N2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]1.5[thin space (1/6-em)]:[thin space (1/6-em)]8.6.

However, while the selectivity to acrolein was always found to increase with the Te loading, the conversion of propene followed a less regular trend. The addition of tellurium to ferric molybdate led to a marked increase in the propene conversion, probably because of the formation of new active sites for the activation of propene.

These results indicate that the activity of the FeMoTeO catalysts went through a maximum upon Te addition and that this maximum probably was between 0.05 and 0.1 moles of Te per mole of ferric molybdate. The sharp increase in propene conversion upon tellurium addition is a clear indication that new active centers were created at the surface of the catalyst. The analysis of the selectivity trends reveals that the increase in selectivity to acrolein with Te content is principally due to the reduction of the COx formation, as the selectivity to deep oxidation products decreased continuously when the Te loading was increased. Concerning the evolution of the selectivity to acrylic acid, the third major by-product, which is the product of the consecutive reaction of oxidation of acrolein, it was expected to rise for longer contact times or for higher yields of acrolein. In this regard, it can be observed that for the Si-FeMo sample, because of the very low production of acrolein, the selectivity to acrylic acid was almost negligible. However, as soon as tellurium was added to the catalyst and new active sites for the formation of acrolein were introduced, the selectivity to acrylic acid rose to 9.4%. Finally, when the Te concentration was further increased from 0.05 to 0.1 and 0.3 moles per mole of Fe2(MoO4)3, the selectivity to acrylic acid declined steadily to about 2%. It is worth remembering that the first step of the consecutive oxidation of acrolein is its adsorption on Brønsted acid sites where it can be readily hydrolyzed and oxidized to give acrylic acid.34,35 A higher concentration of Te on the catalyst surface could account for the suppression of those acid sites on Fe2(MoO4)3 which are responsible for the production of COx and acrylic acid. It is known that tellurium plays a moderating role in the acidity of metal molybdates and MoO3, and experimental evidence has suggested that tellurium is particularly efficient in reducing Brønsted acidity.21,36,37

In order to understand the effect of tellurium on the activity of the catalyst, some important indications can be obtained by analyzing the data on Table 2, where the values of the intrinsic activity of the catalysts with and without tellurium are compared. It should be noted that these values do not represent the true intrinsic activity of the active phase, as firstly they are not measured at low conversion and secondly in the case of Si-FeMoTe catalysts, the colloidal silica contributes to the specific surface area. The data confirmed clearly that the intrinsic activity of the catalysts was significantly increased by the addition of Te. The intrinsic activity of the catalysts was also shown to decrease when the tellurium concentration was higher than 0.1 moles.

Table 2 Catalytic properties of the fresh catalysts (first hour of testing) and after stabilization (after 100 h of testing)
Sample Acrolein selectivity (%) Activity (10−7 molC3H6 s−1 m−2)
Fresh Stabilized Fresh Stabilized
Si-FeMo 9 9 0.1
Si-FeMoTe0.05 54 56 0.7
Si-FeMoTe0.1 72 82 0.9 1.2
Si-FeMoTe0.2 72 89 0.6 0.8
Si-FeMoTe0.3 86 91 0.7 0.9


3.2. Characterization of the catalysts

The characterization of the catalysts by XRD showed that all of them exhibited the Fe2(MoO4)3 and MoO3 phases but no Te-containing phase. Typical examples of the XRD patterns obtained for the FeMoTe samples are shown in Fig. S2.

The chemical analysis and the results of the specific surface area measurements of the prepared catalysts are gathered in Table 3. The chemical analyses indicated that the bulk compositions closely matched nominal compositions and therefore an excess of molybdenum cannot explain the presence of free MoO3 in this case.

Table 3 Results of chemical analysis, specific surface area (SSA) measurement and XRD of the synthesized samples. SSA was measured before testing (fresh) and after 100 h of testing (stabilized)
Sample Chemical analysis SSA (m2 g−1) XRD phases
Fe/Mo Te/Mo Fresh Stabilized
FeMo 0.665 5.2 Fe2(MoO4)3
FeMoTe0.1 0.671 0.033 3.5 Fe2(MoO4)3MoO3
Si-FeMoTe0.05 5.1 Fe2(MoO4)3
Si-FeMoTe0.1 0.668 0.033 24.0 16.6 Fe2(MoO4)3
Si-FeMoTe0.2 0.657 0.065 26.0 16.8 Fe2(MoO4)3MoO3
Si-FeMoTe0.3 0.673 0.097 17.8 10.9 Fe2(MoO4)3MoO3


This observation is in line with that reported by Forzatti et al. for the same type of materials; the authors explained this phenomenon by the substitution of Mo6+ with Te6+ or by incorporation of Te6+ in the interstitial position of the ferric molybdate structure.19 The specific surface areas of the FeMoTeO catalysts were always low, in agreement with previous data reported in the literature.18–20 Addition of colloidal silica always resulted in an increased specific surface area. Nevertheless, it should be noted that these values do not necessarily indicate an increase of the specific surface area of the active phase, as SiO2 nanoparticles contribute to the total surface area by themselves. To gain better insight into the role of silica in the FeMoTeO catalyst, the FeMoTe0.1 and the Si-FeMoTe0.1 samples were analyzed by means of high-resolution transmission electron microscopy (HRTEM). For both samples analyses were performed after catalytic testing in order to observe their morphological evolution once stabilization of their catalytic properties had occurred. Analysis of the morphology of the FeMoTe0.1 sample revealed particles that were larger than 1 μm (Fig. 4). They were composed of several crystal domains, which explained the difference between the calculated BET particle size and the XRD average CS. Indeed, STEM-EDX mapping of the FeMoTe0.1 sample showed that the elemental distribution of Mo, Fe and O was uniform all over the particles, and their relative ratio was coherent with the presence of the Fe2(MoO4)3 phase, which is confirmed by electron diffraction. The spatial distribution of tellurium, on the other hand, was not uniform, and seemed to accumulate at the boundary between Fe2(MoO4)3 crystal domains. Its concentration within the iron molybdate phase was extremely low (Fig. 4b).


image file: c7cy01574g-f4.tif
Fig. 4 a) HRTEM image of aggregation of particles in the used FeMoTe0.1 sample. b) STEM-EDX mapping of the elemental distribution for an examined area.

Fig. 5 clearly shows that the catalyst was characterized at the very surface by an amorphous layer of about 10 nm in thickness coating the crystalline bulk. All the particles observed by TEM exhibited this amorphous layer regardless of their shape and size. Elemental analysis indicated that this amorphous layer was richer in molybdenum and tellurium than the bulk, as proved by the increase of the Mo/Fe and Te/Fe ratios (2.3 and 8.3 times, respectively).


image file: c7cy01574g-f5.tif
Fig. 5 TEM image of the section in the FeMoTe0.1 sample showing the presence of an amorphous layer at the surface and the crystallites with EDX elemental analysis proving the enrichment in Mo and Te of the layer.

Moreover, the molybdenum-rich nature of this amorphous layer was proven by the fact that, after sufficiently long exposure under the electron beam, it crystallized into a new phase that was shown by electron diffraction to be MoO3. The formation of an amorphous layer on the surface of a pure Fe2(MoO4)3 phase has already been observed by TEM analysis on catalysts employed for methanol oxidation and have been identified as the active phase for this reaction.38–40 The detection of a similar amorphous phase on FeMoTeO catalysts is thus not necessarily caused by the presence of tellurium. On the other hand, it is clear from EDX analysis that the amorphous structure can easily host tellurium atoms and, by considering the effect of Te addition on the catalytic properties, it seems very likely that this amorphous Mo–Te–Fe surface layer is the active phase of the catalyst. From the EDX analysis it appears that the layer had the approximate composition Mo7Fe2Te0.5Ox.

The HRTEM images of the Si-FeMoTe0.1 sample after catalytic testing were different from those of the similar FeMoTe0.1 sample (Fig. 6a). They showed smaller particles which appeared to be embedded into the amorphous SiO2 phase, which as expected, prevented the formation of large crystals of iron molybdate observed in the case of the FeMoTe0.1 catalyst and therefore accounted for the improved resistance towards sintering. Because of its texture it was impossible to perform EDX analyses specific to the smallest particles. However the analyses of the larger ones consistently showed the bulk composition of FeMoTe0.1. From Fig. 6b, where the section of a slightly larger particle is presented, it is possible to observe that an amorphous phase was detected over the crystalline Fe2(MoO4)3 bulk. Similar to that reported for the FeMoTe0.1 sample, EDX analysis showed an enrichment of both tellurium and molybdenum at the surface with Mo/Fe and Te/Fe ratios increasing by 1.6 and 6.4 times, respectively. The amorphous layer was, on average, between 5 and 10 nm thick and was regularly found on all crystalline particles.


image file: c7cy01574g-f6.tif
Fig. 6 a) TEM image of the used Si-FeMoTe0.1 sample showing particles of the active phase dispersed in a SiO2 matrix. b) A higher magnification showing a larger particle exhibiting the amorphous surface film on the crystalline core.

Finally, the EDX elemental mapping presented in Fig. 7 indicates that the SiO2 nanoparticles were evidently dispersed all over the sample. Iron and molybdenum were exclusively detected in particles composed of iron molybdate; a slight tellurium signal could also be detected in areas composed only of silica particles.


image file: c7cy01574g-f7.tif
Fig. 7 a) STEM image of a section of the used Si-FeMoTe0.1 catalyst together with the elemental mapping of b) silicon, c) tellurium, d) iron 55 and e) molybdenum.

Considering that tellurium is added during the synthesis as a soluble precursor in the form H6TeO6 and that its solubility within the Fe2(MoO4)3 bulk phase is low, it seems very plausible that during calcination some Te was also deposited on the SiO2 particles. Castellan et al. showed that silica doped with 2% weight TeO2 was able to convert 35% of the propene at 440 °C with little selectivity to acrolein (25%) and the activity of the catalyst increased with tellurium concentration.41 In our case, the catalyst studied by Castellan would be stoichiometrically formed if only 8% of the tellurium initially introduced was deposited on silica. The presence of tellurium on SiO2 could thus account for the lower selectivity to acrolein of the Si-FeMoTe0.1 catalyst in comparison with FeMoTe0.1, as the Te-doped silica nanoparticles can present active sites for the total oxidation of propene. However, the fact that the Si-FeMoTe0.1 catalyst was highly active already at 360 °C, i.e. 80 °C lower than the previously mentioned TeO2/SiO2 catalyst, suggests that the amount of propene reacting on the silica nanoparticles was likely very low.

FeMoTe0.1 and Si-FeMoTe01 have also been characterized by Mössbauer spectroscopy before and after catalytic testing (Fig. S3). Both spectra were fitted with only one quadrupolar doublet, exhibiting an isomeric shift of 0.37 ± 0.02 mm s−1, and a quadrupolar splitting of 0.21 ± 0.02 mm s−1, which are characteristic of Fe3+ in Fe2(MoO4)3.42 No Fe2+ could be detected even after catalytic testing and the slight reduction of iron detected by XPS is limited to the top surface in both samples.

The XANES spectroscopy study of the catalysts before and after catalytic tests did not bring any new information and confirmed those obtained by other techniques concerning the oxidation state of the elements and the evolution of the catalyst with time on steam. The oxido-reduction of the constituting elements was certainly limited to the top surface layer of the catalysts. However important and interesting information was obtained from the comparison of the spectra recorded at the Te L1-edge of the fresh and used FeMoTe0.1 samples (Fig. 8). It can be seen that the fresh sample contained Te4+and Te6+ with peaks at 4944 and 4948 eV and only Te4+ after testing.43 The amount of Te6+ before testing represents less than 17% of the total tellurium. In order to obtain information on the surface, XPS analysis was performed on the fresh and used catalysts. The results presented in Table 4 show that in the fresh catalysts only a single species of molybdenum was present, whose binding energy of 232.4 eV (Mo3d5/2 peak) was characteristic of Mo(VI) species.44–46 Iron as expected was mainly present as Fe(III), as indicated by a typical value of the Fe2p3/2 peak at 711.5 eV; a minor contribution at 709.3 eV characteristic of Fe(II) species was always detected, both in the fresh and used catalysts.47–49 The amount of Fe(II) has a tendency to increase after testing but never exceeds 16%. Finally, tellurium was exclusively found in the Te(IV) form, as indicated by the position of the 3d5/2 peak at 576.5 eV.50,51


image file: c7cy01574g-f8.tif
Fig. 8 XANES spectra recorded for the fresh and used FeMoTe0.1 samples at the Te L1-edge.
Table 4 XPS analysis of catalysts before and after test (a.t)
Sample Mo3d5/2 Fe2p3/2 Te3d5/2 Si2p Fe2+ (%) Fe/Mo Te/Mo
FeMo 232.4 711.4 709.2 8 0.60
FeMoTe0.1 232.6 711.5 709.2 576.6 10 0.60 0.10
Si-FeMoTe0.1 232.5 711.6 709.3 576.5 103.5 11 0.56 0.10
Si-FeMoTe0.1 a.t. 232.6 711.7 709.3 576.6 103.7 16 0.54 0.10


The quantitative analysis of the XPS elemental abundance at the surface is also reported on the same table. In agreement with EDX analysis of the same materials, the Mo and Te enrichments at the surface occurs to a significant extent even in pure iron molybdate (FeMo).47 They are however not as high as those measured for the whole amorphous surface layer by EDX, which seemed to show that gradients of concentration of Te and Fe exist in the layer. The addition of Si during the synthesis did not modify the surface composition and only a slight decrease in the Fe content was observed. It is also interesting to note that the surface composition after testing is similar to that before. It can be noted in particular that the surface concentration of Te was virtually unchanged after catalytic testing. These results are in agreement with chemical analyses of the bulk of the catalysts, which showed only a slight decrease in the tellurium concentration of less than 2% after the longest catalytic testing. The comparison of the iron signals in the XPS spectra of the fresh and used FeMoTe catalysts revealed an increase in the concentration of Fe(II) species after catalytic testing (Table 4). The shoulder at 709.2 eV corresponding to the signal of Fe2+ species was always detected, even in the freshly calcinated catalysts. The appearance of this spectral feature has been already reported in a Fe2(MoO4)3 catalyst exposed to a methanol/oxygen mixture in the selective oxidation of methanol to formaldehyde.48 The authors observed an increase in the intensity of the Fe2+ signal under reducing conditions and related its occurrence to the intervention of iron sites in the hydrogen abstraction step of the mechanism. In the present study, however, there was almost no change in the intensity of the Fe2+ signal of the FeMo sample after exposure to the propene/oxygen mixture under the operating conditions, whereas a significant increase in the intensity of the Fe2+ signal was observed in the Te-containing catalysts after catalytic testing. The rise was particularly pronounced for FeMoTe0.3, showing also the presence of crystalline MoO3. Quite interestingly, a more intense Fe2+ signal in the stabilized catalyst always resulted in an increased selectivity to acrolein, suggesting that fewer deep oxidation sites were present in the partially reduced material. This observation is coherent with a study by Brazdil et al. on simple and multicomponent bismuth molybdates, where it was proposed that a correlation exists between the degree of surface reduction and the selectivity to oxygenated products.52 It was suggested that a partially reduced surface is required to achieve the highest acrolein selectivity, as the fully oxidized surface is characterized by excessively labile Mo[double bond, length as m-dash]O bonds that result in a higher production of COx. Therefore, an ideal degree of surface reaction requires the formation of enough oxygen vacancies to strengthen the remaining lattice oxygens without compromising the surface activity. This hypothesis was further supported by a detailed HRTEM study, showing the formation of a sub-stoichiometric surface composition in the industrial catalysts for the partial oxidation of propene.53 In some cases after catalytic testing, the catalysts showed other Mo species in various amounts evidencing the partial reduction of Mo6+ to Mo5+. A complete study has been undertaken in order to understand the complete surface redox dynamics at the surface.

4. Discussion

The results of catalysts modified by silica can only be discussed in the context of those obtained on pure FeMoTeO catalysts. The results obtained on the later catalysts confirmed that the addition of Te to iron molybdate has an effect on both its activity and selectivity. The conversion showed a maximum around 0.1 Te whereas the selectivity continuously increased with Te content achieving a plateau at approximately 93%. This behavior is not unusual for metal molybdates doped with tellurium and was reported for CaMoO4 and CoMoO4 in the selective oxidation of butene54 and already for Fe2(MoO4)3 in the oxidation of propene, in this later case, with a more irregular trend in the activity at higher Te concentrations.55 On the other hand, it is clearly demonstrated that an excess of tellurium is detrimental to the activity of FeMoTeO catalysts, as previously reported by Ueda et al. and also confirmed in this study.55,56

In earlier studies, Forzatti and Garagiola have proposed that tellurium could enter in the interstitial positions of the ferric molybdate structure as Te6+.19,57 In fact, the Fe2(MoO4)3 structure is composed of oxygen polyhedra sharing only corners, and can accommodate interstitial foreign ions without significant distortion. The iron molybdate structure also presents channel-like cavities that allow the reversible migration from and towards the surface of these interstitial foreign ions.58,59 The presence of such a highly charged cation such as Te6+ in the interstitial position is highly doubtful and if it exists should be very limited. A direct but limited substitution of Mo6+ by Te6+ or Te4+ would be more probable. In that respect it may be noted that Te has been homogeneously incorporated into the Bi2MoO6 lattice with the substitution of Te4+ to Mo6+; generation of two holes per Mo substituted allowed maintaining the charge balance.60 However the substitution is very limited and should be the same in our case as confirmed by EDX. Indeed the results of the characterization of the catalysts by HRTEM, XANES and XPS spectroscopy show that most tellurium ions are Te4+ and located at the surface, outside the crystal domains of iron molybdate. They are accumulated either at the crystallite borders or in an amorphous phase layer coating all the ferric molybdate particles. The highly positive effect of tellurium on the selectivity to propene is likely to reside in the Te sites involved in the α-hydrogen abstraction of the alkene molecules and the formation of allylic intermediates, which is a prerequisite for oxygen insertion.59 Tellurium on the surface of the Fe2(MoO4)3 phase resulted in a drastic increase of the selectivity to acrolein which rose from about 10% on pure ferric molybdate to 93% on the FeMoTe0.1 sample. It has to be noted that the substitution of Mo by Te if it occurs, is so limited and cannot account for the presence of molybdenum oxide detected by XRD in the case of the Te containing solids. Actually both the FeMo and FeMoTe samples show the same Mo rich amorphous surface layer, with approximately the same thickness. The formation of this surface layer should be linked to the preparation method by precipitation that leaves iron in solution. The specific surface area of the FeMoO solid is higher than that of the Te containing ones and molybdenum oxide is formed in a small amount besides the coated particles of the last solids. Furthermore MoO3 is much less active than any iron containing catalysts and cannot account for the higher efficiency of the Te containing ones. With time on stream the FeMoTeO catalysts exhibited a rapid decrease in the conversion of propene over the first 20 hours of reaction followed by a very slow but constant decrease for longer reaction times. This deactivation was the results of an important sintering of the Fe2(MoO4)3 phase, as confirmed by BET analysis of the used samples. However if the intrinsic activity of the catalysts is calculated, it can be observed that it increased with time on stream. This is in disagreement with the results of Ueda et al.,20,22 who rather reported a decrease in the activity of Te-doped iron molybdates over time on stream, especially in the first few hours of reaction. However, these authors likely did not measure the difference in the surface area of the fresh and used catalysts and their observations probably referred to the specific activity rather than the intrinsic activity of the catalysts. In this case there would be no contradiction with the results of the present study, as the observed decrease in the specific activity reported by Ueda et al. was most probably due to the sintering of the Fe2(MoO4)3 phase, and not to an actual loss of active sites. Two possible explanations can account for the observed increase in the intrinsic activity of the FeMoTeO catalysts after stabilization, (i) a better dispersion of Te at the surface, with the creation of new sites for the activation of propene, as suggested by TEM results that evidenced a high mobility of tellurium species and XPS and XANES that showed their tendency to migrate in the surface layer during catalytic testing or (ii) the formation of new Lewis acid sites for propene chemisorption as a result of the partial reduction of the surface. It has been proven, in fact, that the Lewis acidity of Fe2(MoO4)3 catalysts was increased upon reduction, as a consequence of the formation of anionic vacancies at the surface.48,61 Chemisorption of propene is also possible on this kind of site and a higher number of adsorbed olefins could thus explain the observed increase in activity. The concomitant increase of selectivity to acrolein leads to favouring the first hypothesis. The selectivity to acrolein also evolved under operating conditions, exhibiting an increasing trend and was completely stabilized after approximately 70 hours of operation. This increase in the selectivity to acrolein was the result of the migration of Te species to the surface leading to active and selective sites and to the suppression of some deep oxidation sites, as indicated by the lower production of COx and acrylic acid. For Te loadings higher than 0.1 mole per mole, tellurium exhibited a moderating effect on the activity of the catalysts probably due to the covering of the active phase by MoO3 formed in a higher amount. The amorphous surface layer tends to saturate tellurium and the selectivity to acrolein reached a plateau.

The results of this study led to the conclusion that the active and selective phase of the catalysts would be an amorphous Te-doped Fe–Mo–O amorphous oxide layer. TEM studies showed that this amorphous surface layer consistently and completely covered all the particles of the catalyst so that the iron molybdate phase could not play any other role than that of a support allowing the stabilization of the amorphous surface phase containing the three elements. In that respect all attempts to reproduce as a bulk phase the supported phase with the same composition led to a crystallized phase mixture containing mainly MoO3 without stable catalytic properties. Iron in the amorphous layer should play the role of a mediator in the re-oxidation process of both Mo and Te species thus providing a fast re-oxidation route for the molybdenum active species and for the volatile reduced tellurium species. As a matter of fact, crystallized MoTeO phases are very selective by themselves for the formation of acrolein, but suffer rapid deactivation because of tellurium loss.20,22,36 The stabilization of tellurium in the surface layer might also be linked to the amorphous nature of this layer that can also further accept Fe cations as dopants. It seems likely that a solid solution containing an ideal elemental ratio of the three elements is possibly stabilized only as an amorphous phase. Therefore, the fact that Fe2(MoO4)3 presents naturally a suitable Mo-rich amorphous layer in which Te and Fe atoms can be effectively dispersed, makes iron molybdate an ideal support for the active phase.

The addition of colloidal silica during the catalyst preparation did not lead to the formation of any specific phase. The matrix of SiO2 nanoparticles that is formed provided a physical barrier to the sintering of the iron molybdate phase without modifying it and especially without affecting the amorphous layer at its surface. It seems thus logical to assume that the Mo-rich amorphous phase also detected in the Si-FeMoTe0.1 catalyst was similarly the active phase. It can thus be concluded that the presence of silica does not change the nature of the active phase and its beneficial effect on the activity of the FeMoTeO catalyst is largely due to its dispersing properties. In fact, the intercalation of SiO2 nanoparticles results in the formation of smaller iron molybdate particles during the synthesis of the catalyst and in a reduced sintering phenomenon during catalytic operation. Similar to the FeMoTe0.1 catalyst, the selectivity to acrolein was increased in the initial stages of the reaction and reached stabilization after about 70 hours of testing. However the most striking feature of these new catalysts is their activity, which increased so drastically that the Si-FeMoTe0.1 catalyst approached 87% conversion already at 360 °C. Another important effect of SiO2 addition appeared to be the improved stability of the catalyst over time on stream. Actually, characterization by XPS did not show after long testing times any change and in particular no decrease of Te surface content. Te loss seems to be annihilated or if it exists, should be very low and compensated by the migration to the surface of Te from the bulk of the amorphous layer or from the grain boundaries where it is accumulated.

5. Conclusions

The study of the FeMoTeO catalytic system presented in this paper confirmed the high efficiency of this system in the oxidation of propene to acrolein with selectivity to acrolein overpassing 93%. The characterization of the catalysts after catalytic testing revealed that it was composed of ferric molybdate particles coated by a nanometric amorphous layer enriched in molybdenum and tellurium. This amorphous layer, described for the first time in this study, constitutes the active and selective phase of the system. It is likely that active and selective sites are involving the three elements present in the layer in the close vicinity. Modeling of such sites has been undertaken to prove this hypothesis and determine the role of each element. By analogy with other molybdate based catalytic systems of the same type, tellurium should be involved in the first stage of α-hydrogen abstraction of propene molecules and iron in the reoxidation of the active sites.

The addition of silica in the synthesis protocol of the catalysts improved the catalytic activity without decreasing the selectivity to acrolein. Indeed the addition of colloidal SiO2 during the synthesis of the FeMoTeO catalysts resulted in significantly smaller particles of the active phase. This resulted in a marked increase in the specific activity that made it possible to attain more than 80% propene conversion at 360 °C. This reaction temperature is lower by 90 °C than that used with the same catalyst prepared without silica, making it possible to operate in the range of temperatures typically employed in industrial processes. Very importantly, the presence of SiO2 also resulted in an improved stability not only towards sintering as the result of a mechanical effect but also towards surface chemical composition with Te that has been strongly stabilized in the amorphous FeMoTeO surface layer. Finally these catalysts appear very efficient and stable and could be used as a reference to further design new propene oxidation catalysts.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

ADISSEO is thankfully acknowledged for financial support. The authors would like to thank O. Safonova who provided help in the recording of the XANES spectra.

References

  1. K. J. Caspary, H. Gehrke, M. Heinritz-Adrian and M. Schwefer in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger and J. Weitkamp, Wiley-VCH, Weinheim, 2008, p. 3206 Search PubMed .
  2. S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke and E. V. Kondratenko, J. Catal., 2012, 293, 67 CrossRef CAS .
  3. I. Yarulina, J. Goetze, C. Gücüyener, L. van Thiel, A. Dikhtiarenko, J. Ruiz-Martinez, B. M. Weckhuysen, J. Gascon and F. Kapteijn, Catal. Sci. Technol., 2016, 6, 2663 CAS .
  4. T. T. N. Nguyen, V. Bellière-Baca, P. Rey and J. M. M. Millet, Catal. Sci. Technol., 2015, 5, 3576 CAS .
  5. P. Lauriol-Garbey, G. Postole, S. Loridant, A. Auroux, V. Belliere-Baca, P. Rey and J. M. M. Millet, Appl. Catal., B, 2011, 106, 94 CAS .
  6. P. Lauriol-Garbey, J. M. M. Millet, S. Loridant, V. Bellière-Baca and P. Rey, J. Catal., 2011, 281, 68 CrossRef .
  7. M. Dolores Soriano, P. Concepcion, J. M. Lopez Nieto, F. Cavani, S. Guidetti and C. Trevisanut, Green Chem., 2011, 13, 2954 RSC .
  8. H. Tüysüz, J. Llamas Galilea and F. Schüth, Catal. Lett., 2009, 131, 49 CrossRef .
  9. Z. Zhai, M. Wütschert, R. B. Licht and A. T. Bell, Catal. Today, 2016, 261, 146 CrossRef CAS .
  10. J. B. Bream, D. J. Hadley and J. R. Bethell, GB878803A patent, 1961 .
  11. L. C. Fetterly, K. F. Kpetitz and G. W. Conklin, FR1335423A patent, 1963 .
  12. H. H. Voge and W. E. Armstrong, FR1342963A patent, 1963 .
  13. W. R. Middleton and N. J. Wenonah, US2669586 patent, 1954 .
  14. E. Cavaterra and N. Ferlazzo, US3476810A patent, 1969 .
  15. B. W. Kiff and C. R. Norman, US3467716A patent, 1969 .
  16. L. B. Levy, US3637526A patent, 1972 .
  17. J. Y. Robin, Oxydation Du Propène Sur Oxydes Mixtes Catalyseurs à Base de Tellure, Université Claude Bernard - Lyon, 1975 Search PubMed .
  18. P. Forzatti, F. Trifirò and P. L. Villa, J. Catal., 1978, 55, 52 CrossRef CAS .
  19. P. Forzatti, P. L. Villa, N. Ferlazzo and D. Jones, J. Catal., 1982, 76, 188 CrossRef CAS .
  20. W. Ueda, Y. Moro-oka and T. Ikawa, Chem. Lett., 1982, 11, 483 CrossRef .
  21. R. Arrigada, R. Cid, J. Villaseñor and G. Pecchi, Z. Phys. Chem., 1983, 138, 117 CrossRef .
  22. W. Ueda, Y. Moro-oka and T. Ikawa, J. Catal., 1984, 88, 214 CrossRef CAS .
  23. J. F. Brazdil, L. C. Glaeser and R. K. Grasselli, J. Catal., 1983, 81, 142 CrossRef CAS .
  24. A. Getsoian, V. Shapovalov and A. T. Bell, J. Phys. Chem. C, 2013, 117, 7123 CAS .
  25. Y. N. Shyr and G. L. Price, Ind. Eng. Chem. Prod. Res. Dev., 1984, 23, 536 CrossRef .
  26. Y. N. Shyr and G. L. Price, Ind. Eng. Chem. Prod. Res. Dev., 1984, 23, 542 CrossRef .
  27. R. K. Grasselli, G. Centi and F. Trifiro, Appl. Catal., 1990, 57, 149 CrossRef CAS .
  28. Y. Takayama and T. Ikeda, Acrolein from propylene, Montecatini Edison S.p.A., 2007, FR 1524744, pp. 3–33 Search PubMed.
  29. Procédé Catalytique de Préparation D'aldéhydes Alpha, Beta-Insaturés et Catalyseur Pour sa Mise En Oeuvre, FR1524744(A), 1968.
  30. J. C. J. Bart and N. Giordano, J. Catal., 1980, 64, 356 CrossRef CAS .
  31. J. M. M. Millet, C. Virely, M. Forissier, P. Bussiere and J. C. Vedrine, Hyperfine Interact., 1989, 46, 619 CrossRef .
  32. T. Yamashita and P. Hayes, Appl. Surf. Sci., 2008, 254, 2441 CrossRef CAS .
  33. T. Yamashita and P. Hayes, J. Electron Spectrosc. Relat. Phenom., 2006, 152, 6 CrossRef CAS .
  34. R. K. Grasselli, C. G. Lugmair and A. F. Volpe Jr, Top. Catal., 2008, 50, 66 CrossRef CAS .
  35. R. K. Grasselli and J. D. Burrington, in Handb. Heterog. Catal., 2nd edn, Wiley-VHC Verlag, 2008 Search PubMed .
  36. Y. He, Y. Wu, W. Weng and H. Wan, J. Nat. Gas Chem., 2011, 20, 249 CrossRef CAS .
  37. P. Forzatti, F. Trifiro and P. L. Villa, J. Catal., 1978, 52, 389 CrossRef CAS .
  38. E. Söderhjelm, M. P. House, N. Cruise, J. Holmberg, M. Bowker, J. O. Bovin and A. Andersson, Top. Catal., 2008, 50, 145 CrossRef .
  39. K. Routray, W. Zhou, C. J. Kiely, W. Grünert and I. E. Wachs, J. Catal., 2010, 275, 84 CrossRef CAS .
  40. C. Brookes, P. P. Wells, N. Dimitratos, W. Jones, E. K. Gibson, D. J. Morgan, G. Cibin, C. Nicklin, D. Mora-Fonz, D. O. Scanlon, C. R. A. Catlow and M. Bowker, J. Phys. Chem. C, 2014, 118, 26155 CAS .
  41. A. Castellan, A. Vaghi, J. C. J. Bart and N. Giordano, J. Catal., 1975, 39, 213 CrossRef CAS .
  42. A. P. V. Soares, M. Farinha Portela, A. Kiennemann, L. Hilaire and J. M. M. Millet, Appl. Catal., A, 2001, 206, 221 CrossRef CAS .
  43. B. Deniau, T. T. Nguyen, P. Delichere, O. Safonova and J. M. M. Millet, Top. Catal., 2013, 56, 1952 CrossRef CAS .
  44. B. I. Popov, A. V. Pashis and L. N. Shkuratova, React. Kinet. Catal. Lett., 1986, 30, 129 CrossRef CAS .
  45. J. Baltrusaitis, B. Mendoza-Sanchez, V. Fernandez, R. Veenstra, N. Dukstiene, A. Roberts and N. Fairley, Appl. Surf. Sci., 2015, 326, 151 CrossRef CAS .
  46. C. V. Ramana, V. V. Atuchin, V. G. Kesler, V. A. Kochubey, L. D. Pokrovsky, V. Shutthanandan, U. Becker and R. C. Ewing, Appl. Surf. Sci., 2007, 253, 5368 CrossRef CAS .
  47. A. P. S. Dias, F. Montemor, M. F. Portela and A. Kiennemann, J. Mol. Catal. A: Chem., 2015, 397, 93 CrossRef CAS .
  48. K. Thavornprasert, M. Capron, L. Jalowiecki-Duhamel, O. Gardoll, M. Trentesaux, A. S. Mamede, G. Fang, J. Faye, N. Touati, H. Vezin, J. L. Dubois, J.-L. Couturier and F. Dumeignil, Appl. Catal., B, 2014, 145, 126 CrossRef CAS .
  49. T. Yamashita and P. Hayes, Appl. Surf. Sci., 2008, 254, 2441 CrossRef CAS .
  50. H. Hayashi, N. Shigemoto, S. Sugiyama, N. Masaoka and K. Saitoh, Catal. Lett., 1993, 19, 273 CrossRef CAS .
  51. S. Wu, Q. Kan, W. Ding, F. Shang, H. Liu and J. Guan, React. Kinet., Mech. Catal., 2012, 106, 157 CrossRef CAS .
  52. J. F. Brazdil, D. D. Suresh and R. K. Grasselli, J. Catal., 1980, 66, 347 CrossRef CAS .
  53. J. B. Wagner, D. S. Su, S. A. Schunk, H. Hibst, J. Petzoldt and R. Schlögl, J. Catal., 2004, 224, 28 CrossRef CAS .
  54. P. Forzatti, P. L. Villa, D. Ercoli, G. Ercoli, F. Gasparini and F. Trifiro, Ind. Eng. Chem. Prod. Res. Dev., 1977, 16, 26 CAS .
  55. W. Ueda, Y. Moro-oka and T. Ikawa, Chem. Lett., 1982, 11, 483 CrossRef .
  56. W. Ueda, Y. Moro-oka and T. Ikawa, J. Catal., 1984, 88, 214 CrossRef CAS .
  57. F. Garagiola, S. Mobilio, P. Villa and G. Vlaic, Chim. Ind., 1988, 70, 78 CAS .
  58. H. Chen, Mater. Res. Bull., 1979, 14, 1583 CrossRef CAS .
  59. M. H. Rapposch, J. B. Anderson and E. Kostiner, Inorg. Chem., 1980, 19, 3531 CrossRef CAS .
  60. R. K. Grasselli, in: Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger and J. Weitkamp, Wiley-VCH, New York, 1997, p. 2302 Search PubMed .
  61. G. Centi, S. Perathoner and S. Tonini, Catal. Today, 2000, 61, 211 CrossRef CAS .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7cy01574g

This journal is © The Royal Society of Chemistry 2017