Introduction

Acetonitrile (AN) is used as a starting material in the manufacture of active pharmaceutical ingredients and intermediates, namely aminated flavones (Schmitthenner and Falls 1988), vitamins, cortisone and drugs. Moreover, AN is used as a solvent in the synthesis of DNA and in the production of insulin and antibiotics. The main AN producers are INEOS in the United States, Asahi Kasei in Japan and CNPC Jilin Chemical Group in China, with market shares of 26, 20 and 11%, respectively, and 64% of the global capacity (IHS Markit 2014).

AN is typically produced during propylene ammoxidation into acrylonitrile (ACN) (Rojas et al. 2009). For example, in the Montedison process, propylene is ammoxidized in a fluidized-bed reactor at 420–460 °C over SiO2 supported Te-, Ce- and Mo-oxides (Weissermel and Arpe 1997). From 1200 kg of propylene and 560 kg of ammonia, one ton of ACN, 50 kg of hydrogen cyanide and 25 kg of AN were produced (Weissermel and Arpe 1997). Nevertheless, considering the fact that the excess of ammonia is eliminated by an aqueous solution of sulphuric acid, 425 kg of ammonium sulfate were also produced as impurity (Weissermel and Arpe 1997), which has an adverse impact on the purity of the final product.

The pharmaceutical industry consumes approximately 70% of the world’s supply of AN (Bonilla 2010). In this context, laboratories tend to use AnalaR grade AN while industrial applications favour the technical grade due to its lower cost. In both cases, the solvent is very highly refined; typically more than 99.9% and 99.5% of purity were reached, respectively, for AnalaR and technical grade (McConvey et al. 2012).

Several alternative routes for the direct synthesis of AN have been reported in the literature. Specifically, new insights on the gas‒phase ammoxidation of ethanol to AN over palladium (Hamill et al. 2015) and vanadium (Folco et al. 2017) based catalysts were obtained recently. Nevertheless, taking into account the price of ethanol, the cheap and abundant C2 hydrocarbons are convenient feedstocks to produce AN via ammoxidation.

Cobalt-exchanged zeolites (Li and Armor 1998a, b; Bulánek et al. 2002; Mhamdi et al. 2006; Mies et al. 2007a; Essid et al. 2018) have been evaluated in C2 hydrocarbons ammoxidation after varying several preparation parameters, e.g. the zeolite structure, the cobalt precursor nature and the metal exchange level. As such, this process is not definitely commercialized and it would be interesting to investigate further d-block elements.

More recently, our research group (Ayari et al. 2012a, b, 2013) reported the efficiency of Cr-exchanged zeolites (Cr/zeolites) in ethylene ammoxidation. However, due to the strong oxidizing power of Cr6+ ions, ethane molecules were converted into CO2 during ammoxidation (results not published). In a similar way, we (Ayari et al. 2009) experienced ethylene ammoxidation over sol‒gel issued cobalt and chromium alumina catalysts (Co/Al2O3, Cr/Al2O3). When compared to Cr/zeolite systems, alumina-based catalysts exhibited very low catalytic activities despite their developed specific surface areas. Apparently, continuous‒surface metal oxides are inefficient in ammoxidation.

Lately, we tested Mo-exchanged ZSM-5 catalysts (Si/Al = 26, Mo/Al = 1, i.e. 6 wt. % of Mo) in ethane and ethylene ammoxidation (Mannei et al. 2017) after varying the precursor’s nature. We concluded (Mannei et al. 2017) that MoCl5 precursor stabilized essentially dimeric species, which catalyze the ammoxidation reaction. However, the concentration of these active species depended on the exchange stoichiometry and it would be interesting to investigate the effect of Mo content (Mo/Al molar ratio) on the physicochemical and catalytic properties of Mo/ZSM-5 catalysts. In effect, the contribution of Mo oxide, mono- and polymolybdate species in the catalytic activity has not yet been evocated.

The aim of this work is to study the effect of Mo/Al molar ratio on the physicochemical and catalytic properties of Mo/ZSM-5 catalysts in ethane and ethylene ammoxidation. The low- and over-exchanged solids were prepared by solid‒state ion exchange using MoCl5 precursor and characterized by several techniques. For each catalyst, the catalytic activity per individual Mo specie molecule was also determined.

Experimental

Catalysts preparation

In this study, different solids were prepared by solid-state ion exchange from \({\text{NH}}_{4}^{ + }\)–ZSM-5 zeolite powder (Si/Al = 26, furnished by Zeolyst) and MoCl5 precursor (from Alfa Aesar, conserved under nitrogen in a carefully closed glass bottle). The zeolite was mixed and ground in a mortar with the desired quantity of MoCl5 (Mo/Al molar ratios = 0.2, 0.4, 0.5, 0.6, 0.8, 1 and 1.5) and subsequently heated under helium stream (30 mL min−1) from room temperature to 500 °C (heating rate 2 °C min−1) and then isothermally treated for 12 h at 500 °C. The prepared solids were labelled as Mo-x, where x stands for the Mo/Al molar ratio.

A reference sample was prepared by mixing the \({\text{NH}}_{4}^{ + }\)–ZSM-5 zeolite (Si/Al = 26) with the desired quantity of MoO3 (M = 143.94 g mol−1, from Merck, molar ratio Mo/Al = 1, i.e. 9 wt. % of MoO3) and stored without undergoing thermal treatment. This reference material was labelled as Ox‒Z (m.m.), where m.m. stands for mechanical mixture and Z for the zeolite support. However, after in situ thermal treatments, \({\text{NH}}_{4}^{ + }\)–ZSM-5 sample was denoted as H+–ZSM-5.

Catalysts characterization

The elemental content of Mo and Al was estimated by inductively coupled plasma‒optical emission spectrometry (ICP–OES) on an Optima 3300 DV Perkin-Elmer spectrometer. Energy dispersive X-ray (EDX) analysis was performed using a JEOL–JEM 2100F instrument equipped with an X-MAX detector and operating under an accelerating voltage of 200 kV and an energy resolution of 20 eV. Nitrogen adsorption‒desorption isotherms were determined with an automatic ASAP 2020 apparatus from Micromeritics. Before the nitrogen adsorption measurements, the samples were outgassed at 250 °C until a static vacuum of 3 × 10‒5 bar was reached. BET model was used to calculate the specific surface area, while the universal t-plot method was used to determine the external surface area and micropore volume. Pore volumes were calculated at the end of the step corresponding to the filling of the pores (at P/P° = 0.98). 27Al magic angle spinning nuclear magnetic resonance (27Al MAS NMR) spectra were recorded at 78.20609 MHz on a Bruker WB spectrometer using AlCl3·6H2O reference material. X-ray diffraction (XRD) measurements were performed on a Siemens D‒500 diffractometer with CuKα radiation (λ = 1.54060 Å), generator setting of 40 kV and 40 mA, a scanning speed of 0.2° min−1 and a scanning region of 2‒70°. Temperature-programmed reduction by hydrogen (H2-TPR) profiles was obtained from a Micromeritics Autochem 2920 analyzer. The sample (70 mg) was pretreated under air flow (30 mL min−1) at 450 °C for 1 h and then cooled and flushed with helium at room temperature before being reduced from 50 to 720 °C (5 °C min−1) with H2 (5%)/Ar flow (30 mL min−1). Diffuse reflectance infrared Fourier transformed (DRIFT) spectra were recorded on a Bruker IFS 55 spectrometer equipped with a Thermo Spectra Tech reacting cell. The samples were activated in situ at 500 °C (5 °C min‒1) under helium (30 mL min‒1). Temperature-programmed desorption of ammonia (NH3-TPD) was performed with the same H2-TPR equipment. The sample (40 mg) was first exposed to an air flow (30 mL min−1) at 450 °C for 1 h, then saturated with flowing ammonia at 100 °C and flushed with helium at the same temperature. Finally, the temperature was ramped to 550 °C (10 °C min−1). UV/Visible diffuse reflectance spectroscopy (UV/Vis DRS) spectra were recorded on a Perkin Elmer Lambda 45 spectrophotometer using BaSO4 as a reference material. The absorption intensity was obtained from the Schuster–Kubelka–Munk function (Eq. 1).

$$F(R_{\infty } ) = \frac{{(1 - R_{\infty } )^{2} }}{{2R_{\infty } }}.$$
(1)

Here, R∞ is the reflectance from a semi-infinite layer.

Catalytic tests were carried out in a Pyrex reactor. A series of tests were carried out for ethane ammoxidation using 200 mg of catalyst and the following gas composition: 10% of C2H6 (Air liquid 99.95%), 10% of NH3 (Air liquid 99.96%) and 10% of O2 (Air liquid 99.95%). The total flow rate was maintained at 100 mL min−1 by balancing with helium, which corresponds to a gas hourly space velocity (GHSV) of 42350 h−1 (contact time = 0.08 s). Another series of runs was done for ethylene ammoxidation using 50 mg of catalyst (GHSV = 42350 h−1, contact time = 0.02 s) and the same gas composition. The calculation of GHSV values required the determination of the density of a typical catalyst with a pycnometer at 22.5 °C using H2O. It was found that the densities of \({\text{NH}}_{4}^{ + }\)–ZSM-5 zeolite and Mo-1 catalyst are similar (1.42 and 1.40 g mL−1, respectively). The mass of the catalyst as well as the total flow rate was carefully optimized in order to avoid internal and external diffusions. The analysis of the outlet flow was recorded online by two chromatographic units (Intersmat, Delsi), one operated with a Porapak Q column (100‒120 mesh) and a flame ionization detector (FID), while the other was equipped with a Molecular Sieve 13X column and a thermal conductivity detector (TCD).

The conversion (Eq. 2), selectivity (Eq. 3) and activity (Eq. 4) were defined as follows:Conversion of hydrocarbon H (C2H6 or C2H4):

$$X_{H} = \frac{{\sum\limits_{i} {y_{i} n_{i} } }}{{y_{E} n_{E} + \sum\limits_{i} {y_{i} n_{i} } }} \times 100,$$
(2)

i = AN, C2H4 and CO2 (for H = C2H6) and i = AN and CO2 (for H = C2H4).

Selectivity of product Pi (carbon basis: i = AN, C2H4 and CO2):

$$S_{i} = \frac{{y_{i} n_{i} }}{{\sum\limits_{i} {y_{i} n_{i} } }} \times 100.$$
(3)

It is worth to note that yi and yE are the mole fractions of product Pi (AN, C2H4 and CO2) and reactant C2H6, respectively, while ni and nE are the number of carbon atoms in each molecule of product Pi and reactant C2H6, respectively.

Activity of Pi product (Aci), i.e. the rate of Pi formation:

$${\text{Ac}}_{\text{i}} ( {\text{mol}}\;{\text{s}}^{ - 1} {\text{g}}^{ - 1} )= \frac{{{\text{Hydrocarbon}}\;{\text{flow}}\; ( {\text{mL}}\,s^{ - 1} )\times X_{\text{H}} \times S_{\text{i}} }}{{{\text{Molar}}\;{\text{volume}}\; ( {\text{STP)}} \times 1 0 0\times 1 0 0}} \times \frac{ 1}{{{\text{Catalyst}}\;{\text{weight}}}}.$$
(4)

The turnover frequency (TOF), i.e. the activity per each Mo specie molecule, is determined as follows (Eq. 5):

$${\text{TOF}}\;\left( {{\text{s}}^{ - 1} } \right) = \frac{{{\text{Ac}}_{\text{Product}} \;\left( {{\text{mol}}\;{\text{g}}^{ - 1} \;{\text{s}}^{ - 1} } \right)}}{{{\text{Amount}}\;{\text{of}}\;{\text{Mo}}\;{\text{specie}}\;{\text{by}}\;{\text{gram}}\;{\text{of}}\;{\text{sample}}\;\left( {{\text{mol}}\;{\text{g}}^{ - 1} } \right)}}$$
(5)

The major products of ammoxidation reactions are AN, C2H4 and CO2. However, very small amounts of CO, CH4 and NO were also produced. The quantity of these by-products is insignificant compared with that of the major products (< 1%) and, thus, they were not included in the calculations.

Results and discussions

Catalytic tests

Figure 1 represents the catalytic behaviour of Mo-x solids (x = 0.2, 0.4, 0.5, 0.6, 0.8, 1 and 1.5) in ethane ammoxidation into AN as a function of reaction temperature. Figs. S1‒S3 in electronic supplementary material (ESM) illustrate, respectively, the C2H6 conversion and C2H4 selectivity and activity as a function of temperature.

Fig. 1
figure 1

Ethane ammoxidation into acetonitrile: evolution of a acetonitrile selectivity, b CO2 selectivity, and c acetonitrile activity as a function of temperature (catalyst weight = 200 mg, gas composition 10% C2H6, 10% NH3, 10% O2 and 70% He, total flow: 100 mL min‒1, contact time: 0.08 s)

Full size image

Figure 2 illustrates, respectively, the evolution of hydrocarbon conversion, acetonitrile selectivity and activity as a function of temperature in ethylene ammoxidation.

Fig. 2
figure 2

Ethylene ammoxidation into acetonitrile: evolution of a ethylene conversion, b acetonitrile selectivity, and c acetonitrile activity as a function of temperature (catalyst weight = 50 mg, gas composition 10% C2H4, 10% NH3, 10% O2, 70% He, total flow: 100 mL min‒1, contact time: 0.02 s)

Full size image

In the studied reactions, the data were collected under stationary conditions after a stabilization period of 2 h. In this context, the zeolite support does not exhibit any activity, while the Mo containing solids are stable even after 11 h on stream at 500 °C. Moreover, the tests of reproducibility do not reveal any significant change in activity.

The ethane ammoxidation (Eq. 6) results from numerous subsequent steps, i.e. Eqs. (7), (8) and (9). As a matter of fact, it could be considered as a complex process since each step can affect the final hydrocarbon conversion.

$${\text{C}}_{ 2} {\text{H}}_{ 6} + {\text{NH}}_{ 3} + 1. 5 {\text{ O}}_{ 2} \to {\text{CH}}_{ 3} {\text{CN}} + 3 {\text{ H}}_{ 2} {\text{O,}}$$
(6)
$${\text{C}}_{ 2} {\text{H}}_{ 6} + 0. 5 {\text{ O}}_{ 2} \to {\text{C}}_{ 2} {\text{H}}_{ 4} + {\text{H}}_{ 2} {\text{O,}}$$
(7)
$${\text{C}}_{ 2} {\text{H}}_{ 4} + {\text{NH}}_{ 3} \to {\text{CH}}_{ 3} {\text{CH}}_{ 2} {\text{NH}}_{ 2} ,$$
(8)
$${\text{CH}}_{ 3} {\text{CH}}_{ 2} {\text{NH}}_{ 2} + {\text{O}}_{ 2} \to {\text{CH}}_{ 3} {\text{CN}} + 2 {\text{ H}}_{ 2} {\text{O}} .$$
(9)

However, the ethylene ammoxidation (Eq. 10) is a less complex process, in which only Eqs. (8) and (9) occur.

$${\text{C}}_{ 2} {\text{H}}_{ 4} + {\text{NH}}_{ 3} + {\text{O}}_{ 2} \to {\text{CH}}_{ 3} {\text{CN}} + 2 {\text{ H}}_{ 2} {\text{O}} .$$
(10)

In the ammoxidation process, carbon dioxide, the main undesired product, was produced following ethane and ethylene combustion (Eqs. 11 and 12). In effect, the nitrogen oxides, which would be originated from the oxidation of aminated hydrocarbons (i.e. ethylamine and acetonitrile), were not detected.

$${\text{C}}_{ 2} {\text{H}}_{ 6} + 3. 5 {\text{ O}}_{ 2} \to 2 {\text{ CO}}_{ 2} + 3 {\text{ H}}_{ 2} {\text{O,}}$$
(11)
$${\text{C}}_{ 2} {\text{H}}_{ 4} + 3 {\text{ O}}_{ 2} \to 2 {\text{ CO}}_{ 2} + 2 {\text{ H}}_{ 2} {\text{O}} .$$
(12)

In ethane ammoxidation, the selectivity towards CO2 decreased with increasing temperature (check Fig. 1b), which is the opposite of what one would expect from oxidation reactions. A similar behaviour has been reported by Li and Armor (1997) in ethane ammoxidation over Co exchange Beta zeolite.

Despite their different exchange levels, Mo-0.2 and Mo-0.4 catalysts exhibited in ethylene ammoxidation a similar catalytic behaviour (see Fig. 2a, c) regardless of the reaction temperature. These results allow us to exclude Mo-0.4 catalyst from this study. Moreover, in the following discussion, we will discriminate the inefficient ethane ammoxidation catalysts which exhibit:

  1. (i)

    Low \(X_{{{\text{C}}_{ 2} {\text{H}}_{ 6} }}\) conversion values at low (425 °C) and high (500 °C) temperatures. We will consider that low \(X_{{{\text{C}}_{ 2} {\text{H}}_{ 6} }}\) value is inferior to half of the highest conversion value, while medium\(X_{{{\text{C}}_{ 2} {\text{H}}_{ 6} }}\) value is closer to half of the highest one. Nonetheless, high \(X_{{{\text{C}}_{ 2} {\text{H}}_{ 6} }}\) value is superior to half of the highest conversion value.

  2. (ii)

    The lowest selectivity towards acetonitrile at low and at high temperature;

  3. (iii)

    The lowest rate towards the formation of the desired product (AN) and the highest rate towards the formation of the undesired intermediate (ethylene) at low temperature;

  4. (iv)

    The highest selectivity towards the undesired product (CO2) at low and at high temperatures.

The results are compiled in Table 1.

Table 1 Summary of the catalytic properties of Mo-x solids (x = 0.2, 0.5, 0.6, 0.8, 1 and 1.25) in ethane ammoxidation. Bold character corresponds to the highest value
Full size table

According to Table 1, Mo-0.6 catalyst could be excluded from the list as we noticed the highest S CO2 values at 425 and 500 °C (38 and 82%, respectively). In the following sections, we will focus on a limited number of solids, in particular, Mo-0.2 (low-exchanged solid leading to low activity), Mo-0.5 (over-exchanged solid leading to low activity), Mo-1 (over-exchanged solid leading to high activity) and Mo-1.5 (over-exchanged solid leading to moderate activity). However, the behaviour of Mo-0.8 catalyst is worth studying in order to explain the highest S C2H4 value at 425 °C (28%, Table 1).

Chemical analyses by ICP and EDX

Table 2 gives the Mo and Al contents determined by EDX. This table also includes the Mo weight loss following the solid-state ion exchange. However, Table S1 compiles additional analysis, namely those determined by ICP.

Table 2 EDX results
Full size table

The Mo weight loss would belong to the evaporation of some Mo moieties during the exchange as we previously reported in one of our recent works (Ayari et al. 2017). Moreover, for Mo-0.2, Mo-0.5 and Mo-1.5 solids, the EDX analysis performed at different surface regions revealed higher aluminium amounts when compared to the value obtained by ICP with the zeolite (1.3 wt. %, Table S1b). These results indicate the migration of Al atoms from the zeolite framework towards the surface.

Textural study by nitrogen physisorption at − 196 °C

Textural analysis results expressed in terms of specific and microporous surface areas (SBET, Sμ) and (micro)porous volume(s) (Vμ and Vp) are compiled in Table 3.

Table 3 Textural analysis results
Full size table

According to Table 3, the microporous area of \({\text{NH}}_{4}^{ + }\)–ZSM-5 zeolite decreases after Mo introduction. To assess to the micropore blocking effects, we calculated the normalized microporous area (Eq. 13) as previously reported for Mo containing MCM-41 and SBA-15 solids (Vradman et al. 2005).

Normalized

$$S_{\mu } = \frac{{S_{\mu } \;{\text{sample}}}}{{S_{\mu \;} \;{\text{zeolite}}\; \times (1 - y)}}.$$
(13)

Here, y stands for the Mo weight fraction determined by EDX.

The normalized Sμ value calculated for Mo-0.5 solid is close to 1, suggesting that metallic phases are not causing an obstruction of the zeolite micropores. This trend is confirmed by the similarity in Vμ/Vp values obtained with Mo-0.5 and \({\text{NH}}_{4}^{ + }\)–ZSM-5 samples (Table 3, see also Fig. S4). However, for the rest of the solids, the micropores blocking effect is significant since the normalized Sμ values are far below 1 and Vμ/Vp ratios are lower than 0.62.

Structural studies by 27Al MAS NMR and XRD

The NMR spectra of Mo-x (x = 0.2, 0.4, 0.6, 1 and 1.5) and \({\text{NH}}_{4}^{ + }\)‒ZSM-5 solids are illustrated in Fig. 3, while the spectrum of Mo-0.8 solid is depicted in Fig. S4.

Fig. 3
figure 3

27Al NMR spectra of \({\text{NH}}_{4}^{ + }\)–ZSM-5 and Mo-x solids (x = 0.2, 0.4, 0.6, 1 and 1.5)

Full size image

The NMR spectrum of \({\text{NH}}_{4}^{ + }\)–ZSM-5 zeolite exhibited a peak at ca. 55 ppm ascribed to four-coordinate framework Al (Yu et al. 2011), while the very weak signal at ~ 0 ppm is attributed to extra-framework octahedral Al centres (Yu et al. 2011).

The NMR spectrum of Mo-1 solid reveals a strong increase in the signal intensity of the peak situated at 0 ppm, which indicates a pronounced zeolite dealumination. However, in the spectrum of Mo-1.5 solid, an additional signal was detected at ‒14 ppm, which belongs to the Al2(MoO4)3 phase (Liu et al. 1997).

Considering the fact that Al2(MoO4)3 is originated from the reaction of MoO3 with the zeolite dealumination product (i.e. Al2O3), the absence of the NMR peak at − 14 ppm in the spectra of Mo-x solids (x = 0.2, 0.4, 0.6, 0.8 and 1) would indicate the presence of significant amounts of Al2O3 with traces of MoO3 or higher amounts of MoO3 with insignificant amounts of Al2O3.

In the NMR spectra of Mo-0.2, Mo-0.4, Mo-0.6 and Mo-1.5 solids, we noticed the presence of a signal at ~ 15 ppm ascribed to the five-coordinate extra-framework aluminium specie (Yu et al. 2011). Seemingly, the intensity of this signal increases with the increase in Mo/Al ratio, except for Mo-1 solid.

The X-ray diffraction patterns of the reference zeolite, pure α-MoO3 and Mo-x (x = 0.2, 0.5, 1 and 1.5) solids are shown in Fig. 4.

Fig. 4
figure 4

XRD patterns of \({\text{NH}}_{4}^{ + }\)‒ZSM-5 zeolite, pure α-MoO3 and Mo-x (x = 0.2, 0.5, 1.0 and 1.5) solids

Full size image

The XRD patterns exhibit narrowed and well-defined diffraction peaks similar to those of ZSM-5 zeolite (van Koningsveld et al. 1990). However, the pattern of Mo-1.5 solid clearly revealed additional peaks at 2θ = 12.79, 25.80, 27.46 and 39.10°. Based on the XRD analysis performed with pure oxide (orange curve in Fig. 4), these peaks would correspond to crystalline α-MoO3 (respectively, (020), (040), (021) and (060) planes) (JCPDS data file No. 05-050827). Apparently, Mo-1.5 solid still contains MoO3 crystallites despite the formation of Al2(MoO4)3 phase (detected by NMR, olive curve in Fig. 3). Moreover, the presence of amorphous MoO3 oxide particles at the surface of Mo-0.2, Mo-0.5 and Mo-1.0 samples is not excluded despite that they would not be detected by XRD.

H2-TPR study

The TPR profiles of Mo-x solids are depicted in Fig. 5.

Fig. 5
figure 5

TPR profiles of Mo-x (x = 0.2, 0.4, 0.6, 0.8, 1 and 1.5)

Full size image

The TPR profile of Mo-0.2 solid exhibits two shoulders. The first one, centred at 476 °C, is attributed to the reduction of MoO3 into MoO2 on the basis of the study performed with MoO3/\({\text{NH}}_{4}^{ + }\)‒ZSM-5 mechanical mixtures (Ayari et al. 2018), while the second one, centred at 590 °C, belongs to the reduction of tetrahedral Mo6+ species associated with Brønsted acid sites (Mannei et al. 2017; Ayari et al. 2018). The increase in Mo/Al ratio from 0.2 to 0.6 moderately increases the hydrogen consumption and provokes the simultaneous shift of the first peak to the right side (from 476 to 485 °C). This effect may be due to a strong interaction established between Mo and the support. However, the increase in Mo/Al ratio from 0.6 to 0.8 resulted in a drastic increase in the hydrogen consumption, evidencing that there were some portions of molybdenum species which are still available for reduction. Moreover, when compared to Mo-0.8, the profile of Mo-1 solid undergoes a prominent decrease in the intensity of the peak at 595 °C. Nonetheless, the TPR profile of Mo-1.5 sample reveals a significant hydrogen consumption with a peak centred at 535 °C ascribed to the reduction of MoO3 oxide particles (Ayari et al. 2018), which is concomitant with the subsequent reduction of Al2(MoO4)3 phase (López Cordero et al. 1990).

It can be concluded that less reducible Mo species undergo strong interaction with the zeolite lattice oxygen at low Mo/Al ratio (< 0.6). However, increasing the Mo/Al ratio from 0.6 to 1.5 favours the reduction of molybdenum oxide species, namely MoO3 and Al2(MoO4)3, previously detected by XRD and NMR. As for Mo-1 solid, the reduction of tetrahedral Mo6+ ions associated with Brønsted acid sites was seemingly inhibited.

NH3-TPD and DRIFT studies

The NH3-TPD profiles and the DRIFT spectra of Mo-x (x = 0.2, 0.5, 1 and 1.5) solids and the reference zeolite are reported in Fig. 6.

Fig. 6
figure 6

a NH3-TPD profiles, and b DRIFT spectra of the dehydrated/deammoniated zeolite (H+‒ZSM-5 sample) and Mo-x (x = 0.2, 0.5, 1.0 and 1.5) solids

Full size image

The NH3-TPD profile of H+‒ZSM-5 zeolite exhibited the typical doublet peak (Hidalgo et al. 1984; Post and van Hoff 1984) signal, respectively, known as l-peak and h-peak. The h-peak, centred at 435 °C, was associated with ammonia molecules adsorbed on strong acid sites (Si‒O+H‒Al) (Hidalgo et al. 1984; Lobree et al. 1999), while the l-peak, centred at 207 °C, was assigned to NH3 adsorbed on weak acid sites (Topsoe et al. 1981; Meshram et al. 1986; Lobree et al. 1999).

The introduction of Mo modified the shape of the NH3-TPD profile. In fact the h-peak increased in intensity at Mo/Al = 0.2 and 0.5, evidencing the generation of a strong acidity. However, for Mo-1 and Mo-1.5 solids, the intensity of h-peak drastically decreased due to the establishment of deep interaction between Mo and Brønsted acid sites.

The DRIFT spectrum of the parent zeolite (Fig. 6b) exhibits two characteristic bands in the νOH region. The first one situated at 3595 cm−1 was assigned to the vibration of the strong Brønsted acid site (Si–O+H–Al) (Jentys et al. 1997), while the second one situated at 3731 cm−1 is attributed to the terminal silanol groups (Si–OH) (Jentys et al. 1997). The exchange of Mo into zeolite involves an attenuation of the 3595 cm−1 band’s intensity, evidencing that a fraction of Mo ions was deeply exchanged with Brønsted acid sites. In particular, the spectrum of Mo-1.5 solid revealed the significant decrease in the 3730 cm‒1 band’s intensity, which could be explained by the presence of metallic particles that covered the silanol groups.

Optical spectroscopy study

The overview illustrated in ESM (UV/Vis DRS section) revealed the conflicting interpretations of DRS spectra relative to Mo containing solids. In order to step forward in the understanding of the nature of supported molybdenum species, we explored the absorption band gap in this study using the Shuster‒Kubelka‒Munk function (Eq. 1) and Eqs. (14) and (15) (Khan and Hogarth 1991).

$$F(R_{\infty } ) = \frac{{(1 - R_{\infty } )^{2} }}{{2R_{\infty } }} = \frac{{(hv - E_{g} )^{n} }}{hv},$$
(14)
$$[F(R_{\infty } ) \times hv]^{{\frac{1}{n}}} = hv - E_{g} .$$
(15)

Here, R∞ is the diffuse reflectance of an infinitely thick sample, n is an exponent which takes the values of 2, 3, 1/2 and 3/2 for indirect allowed, indirect forbidden, direct allowed and direct forbidden transitions, respectively (Khan and Hogarth 1991), hν is the photon energy and Eg is the optical energy gap of the material. Plotting \([F(R_{\infty } ) \times hv]^{{\frac{1}{n}}}\) versus hν and extrapolating to \([F(R_{\infty } ) \times hv]^{{\frac{1}{n}}} = 0\) (Eq. 16) yields to the value of Eg. The results are compiled in Figs. S6 and S7 and Table S2 for pure MoO3 and Ox/Z (m.m.) samples. For the direct allowed transition (n = 1/2, i.e. [F(R∞) × hν]2 vs. hν in Fig. S6A), indirect forbidden transition (n = 3, i.e. [F(R∞) × hν]1/3 vs. hν in Fig. S7) and direct forbidden transition (n = 3/2, i.e. [F(R∞) × hν]2/3 vs. hν in Fig. S7), the area ratio for pure and diluted MoO3 does not correspond to 9% (Table S2). However, for indirect allowed transition (n = 2), i.e. the plot representing [F(R∞) × hν]1/2 vs. hν (Fig. S6B), we determined the area under the curve ascribed to Ox–Z (m.m.) to be 0.26 a.u., while the area under the curve attributed to pure MoO3 is 2.78 a.u.. The ratio of these areas (i.e. (0.26/2.78) × 100, Table S2) is similar to the theoretical percentage of MoO3 in the mixture (i.e. 9%). Seemingly, [F(R∞) × hν]0.5 function describes our system in a similar way as in the literature works (Anwar and Hogarth 1988, 1989; Donnadieu et al. 1988; Chithambararaj et al. 2013).

Figure 7 represents the deconvolution of the DRS spectra of Mo-x solids (x = 0.2, 0.5, 1 and 1.5), while Table 4 summarizes the corresponding quantitative study results.

Fig. 7
figure 7

Optical absorption spectra of Mo-x (x = 0.2, 0.5, 1 and 1.5) solids

Full size image
Table 4 Edge energy (Eg) values relative to each Mo specie and the area of the corresponding band
Full size table

We used a commercial software (Origin 8.0, Microcal Software Inc., USA) to decompose the spectra into Gaussian peaks (additional details are available in Mannei et al. 2016; Ayari et al. 2018).

The deconvolution of the spectra for Mo-1.5 and Mo-0.5 solids revealed the presence of four O2− → Mo6+ charge transfer bands assigned, respectively, to MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2− moieties (Ayari et al. 2018). However, the spectrum associated with Mo-0.2 solid does not reveal the presence of the band ascribed to [MoO4]2−, while Mo-1 solid contains only MoO3 and [Mo2O7]2−.

To determine the concentration of MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2− moieties, we need in the first place to determine the absorption coefficient (k’) of each Mo specie using the corresponding area (A) compiled in Table 4 and the formulas given in Eqs. (16‒18). In this context, equivalent formulas were previously used in the case of Co2+ exchanged zeolites (Essid et al. 2018; Dědeček and Wichterlová 1999).

For Mo-1 solid:

$$C_{\text{Mo}} \;({\text{mmol}}\;g^{ - 1} ) = A_{{{\text{MoO}}_{ 3} }} \times k_{{{\text{MoO}}_{ 3} }}^{'} + 2\left( {A_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }} \times k_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }}^{'} } \right).$$
(16)

For Mo-0.2 solid:

$$C_{\text{Mo}} \;({\text{mmol}}\;{\text{g}}^{ - 1} ) = A_{{{\text{MoO}}_{3} }} \times k_{{{\text{MoO}}_{ 3} \;}}^{'} + 2\left( {A_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }} \times k_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }}^{'} } \right) + 7\left( {A_{{{\text{Mo}}_{ 7} {\text{O}}_{ 2 4}^{{ 6 { - }}} }} \times k_{{{\text{Mo}}_{ 7} {\text{O}}_{ 2 4}^{{ 6 { - }}} }}^{'} } \right).$$
(17)

For Mo-0.5 and Mo-1.5 solids:

$$\begin{aligned} C_{\text{Mo}} \;({\text{mmol}}\;{\text{g}}^{ - 1} ) & = A_{{{\text{MoO}}_{ 3} }} \times k_{{{\text{MoO}}_{ 3} \;}}^{'} + 2\left( {A_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }} \times k_{{{\text{Mo}}_{ 2} {\text{O}}_{ 7}^{{ 2 { - }}} }}^{'} } \right) \\ & \quad + 7\left( {A_{{{\text{Mo}}_{ 7} {\text{O}}_{24}^{6 - } }} \times k_{{{\text{Mo}}_{ 7} {\text{O}}_{24}^{6 - } }}^{'} } \right) + \left( {A_{{{\text{MoO}}_{4}^{2 - } }} \times k_{{{\text{MoO}}_{4}^{2 - } }}^{'} } \right). \\ \end{aligned}$$
(18)

After calculation, the absorption coefficients relative to MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2− species are, respectively, 8.45 × 10‒5, 0.31 × 10‒5, 3.22 × 10‒5 and 2.71 × 10‒3 mol g‒1 (a.u.)‒1. However, using the formula given in Eq. (19), the concentration and the percentage of each Mo specie were determined (Table 5).

$$C_{\text{Mo}} = A_{\text{Mo}} \times k_{\text{Mo}}^{'} .$$
(19)
Table 5 Amount of each Mo moiety expressed in μmol g‒1 (value between brackets: the corresponding molar fraction in %)
Full size table

Here, A stands for the band’s area depicted in Table 4 and k′, given above. We used for this purpose the surface Mo concentration (i.e. CMo value determined by EDX) since the photons do not reach the internal cavities of the sample during UV/Vis DRS analysis. In fact, in our recent study (Ayari et al. 2018), we were not able to quantify by DRS and X-ray photoelectron spectroscopy the fraction of Mo species which occupied the zeolite micropores.

The data illustrated in Table 5 indicate that the molar fraction of polymolybdate species is very low. These results are consistent with those in previous work (Mannei et al. 2016), reporting that, unlike ZSM-5 (Si/Al = 15), the long distance between two neighbouring tetrahedral Al atoms disfavours the polymerization of Mo species in the case of ZSM-5 (Si/Al = 26) zeolite. Additionally, Table 5 revealed higher amounts of MoO3 over Mo-0.2 and Mo-1 solids, while Mo-0.5 and Mo-1.5 solids loaded significant amounts of monomolybdate species.

Based on H2-TPR (Sect. 3.5) and DRS (Sect. 3.7) studies, we can exclude the reducibility of polymolybdates under hydrogen due to their very low amount. However, in the case of Mo-0.2 solid, the H2-TPR peaks at 476 and 590 °C would correspond, respectively, to the reduction of MoO3 and dimolybdate species (Table 5). Nonetheless, in the H2-TPR profile of Mo-1 solid, the peak centred at 515 °C is associated with the reduction of MoO3. Apparently, the reduction of dimeric species (78.9 μmol g‒1, Table 5) was inhibited due to the presence of significant amounts of MoO3 (158 μmol g‒1, Table 5). In the case of Mo-1.5 solid, the H2-TPR peak centred at 535 °C would correspond to the reduction of MoO3 (82 μmol g‒1, Table 5). Nevertheless, the reduction of tetrahedrally coordinated Mo6+ species (i.e. 460.7 μmol g‒1 of monomolybdates and 54.7 μmol g‒1 of dimolybdates, Table 5) was inhibited (absence of the TPR peak at ~ 600 °C) due to the presence of crystalline MoO3 (detected by XRD, Fig. 4) and Al2(MoO4)3 (detected by NMR, Fig. 3).

Discussions

When an intimate mixture of MoCl5 and \({\text{NH}}_{4}^{ + }\)‒ZSM-5 was heated under helium stream between room temperature and 500 °C, a fraction of Mo was lost (see Table 1) following the departure of MoO2(OH)2(g) and MoOCl4(g) (Ayari et al. 2017). However, extending the exchange at 500 °C for 12 h led to the consumption of the strong zeolite acidity as revealed by NH3-TPD (Fig. 6a) and DRIFT (Fig. 6b) studies. The extra-framework Al species (Al–OH) do not contribute to the exchange as we were not able to detect their DRIFT’s band situated at 3662 cm−1 (Jentys et al. 1997). Nevertheless, at Mo/Al = 1.5, the DRIFT’s band at 3730 cm−1 drastically decreased in intensity, evidencing that silanol groups were covered by crystalline MoO3 (XRD, Fig. 4) and Al2(MoO4)3 (NMR, Fig. 3) particles present at the surface.

Once the exchange is achieved, Mo was stabilized over \({\text{NH}}_{4}^{ + }\)–ZSM-5 in the monomeric ([MoO4]2−), dimeric ([Mo2O7]2−) and polymeric ([Mo7O24]6−) forms as well as MoO3 oxide. The same species were also identified by Mosqueira (Mosqueira and Fuentes 2002) using a semi-quantitative DRS study based on the estimation of the gap energy (Mosqueira et al. 2004). In our study, we extended the DRS analysis to quantify each Mo specie using the results obtained with pure MoO3 and MoO3/\({\text{NH}}_{4}^{ + }\)–ZSM-5 mechanical mixture. The quantitative study results (Table 5) revealed that the solid prepared at Mo/Al = 1 does not contain polymeric species. Such result, supported by Raman spectroscopy (Mannei et al. 2016, 2017), was explained by the fact that the long distance between two neighbouring tetrahedral Al atoms in ZSM-5 (Si/Al = 26) disfavours the Mo polymerization (Rice et al. 1999). Nevertheless, due to imperfections in the Al distribution, namely at the surface, the formation of [Mo7O24]6− species cannot be totally discarded. In effect, the formation of very low amounts of polymolybdates occurred on the near-surface of Mo-0.2, Mo-0.5 and Mo-1.5 solids since the EDX results revealed higher Al amounts (Table S1a) compared to those obtained by ICP (Table S1b).

In Table 5, we noticed that solids prepared at Mo/Al = 0.2 and 1 stabilized dimeric species and MoO3 at the detriment of [MoO4]2‒. Moreover, at Mo/Al = 0.5, the corresponding solid exhibited low molar fraction of MoO3 (29%, Table 5), which excluded the micropores blocking effect (Table 3).

Owing to their low concentration in all the studied catalysts (Table 5), we will not consider the contribution of [Mo7O24]6− species in the catalytic activity. Therefore, we calculated the catalytic activity per each Mo specie molecule (i.e. MoO3, [Mo2O7]2− and [MoO4]2−) denoted in our study as TOF (Eq. 5). An example of calculation was illustrated in ESM (page 9).

For ethane ammoxidation, we reported in Figs. 8 and 9 the TOF values (s−1) relative, respectively, to ethylene and AN activity as a function of reaction temperature. For ethylene ammoxidation, Fig. 10 illustrates the TOF values for AN activity (check also the corresponding histograms in Figs. S8‒S10 in ESM).

Fig. 8
figure 8

Ethane ammoxidation: TOF values relative to MoO3, [Mo2O7]2‒ and [MoO4]2‒ species for ethylene activity as a function of reaction temperature

Full size image
Fig. 9
figure 9

Ethane ammoxidation: TOF values relative to MoO3, [Mo2O7]2‒ and [MoO4]2‒ species for acetonitrile activity as a function of reaction temperature

Full size image
Fig. 10
figure 10

Ethylene ammoxidation: TOF values relative to MoO3, [Mo2O7]2‒ and [MoO4]2‒ species for acetonitrile activity as a function of reaction temperature

Full size image

It is generally admitted that, for a given Mo moiety and given temperature, TOF values should be constant for a specific reaction. Nevertheless, according to Figs. 8, 9, 10, TOF values are various for each catalyst. Apparently, during a catalytic act, synergistic effects and interactions may exist either between reactant (C2H6/C2H6) or intermediate (C2H4/C2H4 or CH3‒CH2‒NH2/CH3‒CH2‒NH2 or C2H4/CH3‒CH2‒NH2) molecules (Li and Armor 1999). Hence, this set of parameters could perfectly explain these variations and would probably enlighten the discrepancy in the individual TOF values reported by Čapek (Čapek et al. 2010).

Under the assumption that the synergistic effects and the interactions are negligible, the total TOF values [TOFtotal = TOF (MoO3) + TOF ([Mo2O7]2‒) + TOF ([MoO4]2‒)] for Eqs. 6, 7 and 10 are compiled in Table 6.

Table 6 Ethane and ethylene ammoxidation: TOFtotal values at different temperatures
Full size table

Based on the data compiled in Table 6, the maximum TOFtotal in ethylene ammoxidation was 2.3 s‒1 at 500 °C (SAN = 93%, Fig. 2b, which is more than that reported for Co-exchanged Beta zeolite (0.8 s‒1) with a selectivity of 58% to AN (Mies et al. 2007b).

However, in ethane ammoxidation, the highest TOFtotal values are 40.7, 71.4, 138.8 and 234.0 h‒1, respectively, at 425, 450, 475 and 500 °C. These values are higher than those reported at 450 °C for Co/Beta solids, i.e. 24.3 h‒1 (Bulánek et al. 2002) and 30.6 h‒1 (Li and Armor 1998a; Wichterlová et al. 2000).

The ethane ammoxidation (Eq. 6) is considered as a complex process since the reaction complied in Eqs. 7, 8 and 9 may affect the final catalytic activity. Therefore, we will interpret only the catalytic results obtained at specific temperatures (500 and 425 °C).

Mo-0.2 and Mo-0.5 catalysts exhibit a similar molar fraction of dimolybdate species (~ 18%, Table 5). Nevertheless, the first solid exhibited the highest selectivity towards CO2 at 500 °C (22%, Table 1) when compared to Mo-0.5. This behaviour is apparently due to the presence of higher amounts of amorphous MoO3 oxide (79% vs. 29% for Mo-0.5 solid, Table 5), which inhibit the accessibility of reactant molecules towards the active sites (Mannei et al. 2017) and enhance, therefore, the secondary combustion reactions (Eqs. 11 and 12). This explains once again the fact that the pore blocking effect is limited over Mo-0.5 solid as revealed by the textural analysis results (Normalized Sμ and Vμ/Vp values, Table 3).

Mo-0.5 Solid contains a significant amount of monomolybdate species (50%, Table 5) and exhibits a high selectivity towards ethylene at 500 °C (22%, Table 1). It could be concluded that monomolybdate species catalyze the oxidative dehydrogenation of ethane into ethylene (Eq. 7) over Mo-0.5 catalyst. In fact, Mo-0.2 solid which does not contain [MoO4]2‒ species exhibited a low selectivity towards ethylene at 500 °C (9% in Table 1).

The same solids, i.e. Mo-0.2 and Mo-0.5, exhibited on the one hand a similar molar fraction of dimolybdate species and on the other hand a similar AN selectivity at 500 °C (67 ± 2%, Table 3). Apparently, [Mo2O7]2‒ species catalyze the ethylene ammination into ethylamine (Eq. 8) and the oxidative dehydration of ethylamine into AN (Eq. 9). This interpretation seems correct since Mo-0.2 and Mo-0.5 catalysts exhibited the highest TOFtotal values in ethylene ammoxidation.

The solid which does not contain monomeric species, i.e. Mo-1, exhibited the lowest TOFtotal value (19.8 h‒1, Table 6) in ethane oxidative dehydrogenation into ethylene (Eq. 7), which corroborates the fact that [MoO4]2‒ species catalyze this reaction. However, the lowest TOFtotal values relative to Eqs. 6 and 7 (respectively, 162.7 h‒1 and 0.6 s‒1, Table 5) are not expected since Mo-1 solid exhibited the highest molar fraction of [Mo2O7]2‒ species (33%, Table 5). These low TOFtotal values would be explained by the presence of higher amounts of MoO3 (158 μmol g‒1, 67%, Table 5). This oxide phase, undetectable by XRD (Fig. 4), would be in a highly divided crystalline state and does not catalyze the hydrocarbon combustion at 500 °C nor at 425 °C (see the S CO2 values at 500 and 425 °C for Mo-1 catalyst in Table 1). At the contrary, the secondary reactions (Eqs. 11 and 12) occurred over Mo-0.2 catalyst which contains 79% of MoO3 (Table 5) and over Mo-0.5 (29% of MoO3, Table 5) as revealed by the high S CO2 values in Table 1.

During ethane ammoxidation over Mo-1.5 catalyst, the selectivity towards ethylene is very high (30% at 500 °C, Table 1) since crystalline MoO3 (Fig. 4) and higher amounts of monomolybdates (76%) catalyze the ethane dehydrogenation into ethylene. However, despite the presence of significant amounts of MoO3 over Mo-1 catalyst (67% vs. 14%, Table 5), these catalysts exhibit different S CO2 values (68 and 28%, Table 1) at 425 °C. The presence of the undesired Al2(MoO3)4 phase over Mo-1.5 solid resulted, at low temperature, in the lowest activity towards AN in ethylene ammoxidation (Fig. 2c and Table 1) and the lowest selectivity towards AN during ethane ammoxidation (19% at 425 °C, Table 1). Such phase would inhibit the diffusion of reactants molecules towards [Mo2O7]2‒ active sites (9%, Table 5) and enhance the combustion reactions at low temperature.

In summary, the undesired Al2(MoO3)4 phase inhibits the diffusion of reactant molecules towards the active sites, while amorphous MoO3 particles catalyze the hydrocarbons combustion during ethane and ethylene ammoxidation. In effect, the poorly active Mo-0.6 solid exhibited in its TPR profile (Fig. 5, check also Fig. S11) only one peak at 485 °C ascribed to the reduction of MoO3. However, during ethane ammoxidation, [MoO4]2‒ species catalyze the oxidative dehydrogenation of C2H6 into C2H4, while [Mo2O7]2‒ active species catalyze ethane and ethylene ammoxidation into AN. Moreover, crystalline MoO3 particles (XRD-detectable) are adequate in ammoxidation namely at low temperature.

In the case of Mo-0.8 catalyst, the very high selectivity towards C2H4 exhibited at 425 °C (Table 1) would indicate the presence of a significant amount of monomeric species, reducible under H2 at 595 °C (wine curve in Fig. 5).

Conclusion

Mo/ZSM-5 solids prepared by solid-state ion exchange were evaluated in ethane and ethylene ammoxidation into acetonitrile between 425 and 500 °C at very low contact times. Starting from MoCl5 and \({\text{NH}}_{4}^{ + }\)‒ZSM-5 zeolite (Mo/Al molar ratio = 0.2), MoO3, [Mo2O7]2‒ and [Mo7O24]6− species were stabilized following the exchange. However, at Mo/Al ratio equal to 0.5 and 1.5, the corresponding solids stabilized MoO3, [MoO4]2‒, [Mo2O7]2‒ and [Mo7O24]6− species, while only MoO3 and [Mo2O7]2‒ were stabilized over the solid at Mo/Al = 1. In this study, we were able to quantify the different Mo species using optical spectroscopy and to reveal that dimolybdate species are the active species. On the other hand, the quantitative study performed by DRS revealed that the molar fraction of polymolybdates is very low and they do not play a role in catalysis. However, monomolybdate species, reducible under hydrogen above 550 °C, catalyzed the ethane dehydrogenation into ethylene, while Al2(MoO4)3 (identified by NMR) species and amorphous MoO3 and catalyzed the hydrocarbons’ combustion into CO2. In this study, the highest turnover frequency value reported for ethylene ammoxidation is 2.3 s‒1 at 500 °C. Based on this value and those obtained in ethane ammoxidation, we revealed the superiority of Mo-exchanged zeolite when compared to other systems.