Introduction

Acetonitrile (AN) market has witnessed a significant growth over the past decade. In fact, a recent report entitled “Acetonitrile Market: Global Industry Trends, Share, Size, Growth, Opportunity and Forecast 2019–2024” published by IMARC Group estimates that the global AN market had reached 113 kt in 2018. However, this research report anticipates the market to reach 143 kt by 2024, exhibiting a compound annual growth rate (CAGR) of ~ 5% during 2019–2024 [1]. Such an annual growth rate can be accredited to the versatile application of AN as solvent for high-performance liquid chromatography [2] and intermediate in the synthesis of pyrimidine derivatives [3]. Nonetheless, due to its high dielectric constant (ɛ = 35.85 at 25 °C under 1 bar [4]), stability and ability to dissolve electrolytes, acetonitrile is also used in the manufacturing of batteries. In effect, the strong demand for lithium batteries, as a result of their growing usage in electronic devices, is further propelling the market growth [5]. As a matter of fact, the synthesis routes of AN as the main reaction product have been sought and an efficient atom economy could be achieved by using C2 substrates such as acetic acid [6], ethanol [7], ethylene [8] and ethane [9].

Ethane ammoxidation (Eq. 1) is an alternative route for the manufacture of AN. Indeed, all the commercially produced acetonitrile is obtained as a co-product from the ammoxidation of propylene into acrylonitrile (Eq. 2) over Bi9PMo12O52-50 wt% SiO2 material [10]:

$${\text{C}}_{2} {\text{H}}_{6} + {\text{NH}}_{3} + 1.5{\text{O}}_{2} \to {\text{CH}}_{3} {\text{CN}} + 3{\text{H}}_{2} {\text{O}}$$
(1)
$${\text{CH}}_{2} = {\text{CH}} {-} {\text{CHO}} + {\text{NH}}_{3} + 0.5{\text{O}}_{2} \to {\text{CH}}_{2} = {\text{CH}} {-} {\text{CN}} + 2{\text{H}}_{2} {\text{O}}$$
(2)

Although no C2H6 ammoxidation process is as yet commercial, the advances in catalysis may make it an economic alternative for AN production in the near future [11]. In this context, we reported the successful application of beta zeolite modified with cobalt in ethane ammoxidation into AN at a contact time of 130 ms [12, 13]. Additionally to acetonitrile, the primary products in ethane ammoxidation over Co-exchanged beta zeolite are ethylene and CO2 [12, 13]. Nevertheless, interesting results have been obtained at a very low contact time (80 ms) over molybdenum-exchanged ZSM-5 zeolites prepared by solid-state ion exchange, which consists of mixing the zeolite and the precursor in a mortar and subsequently heating the powder under helium stream at 500 °C for 12 h [9, 14].

Interestingly, several methods were developed to introduce Mo into zeolites, which include sublimation [15,16,17], impregnation [18, 19] and sonochemical method [20]. The latter method is considered to be costly, while the inexpensive solid-state ion exchange (SSIE) has a number of advantages, e.g. achieving a high metal exchange degree in one step. However, SSIE also has major disadvantages since, on heating, some precursors undergo complex changes [21]. For example, the thermal decomposition of MoCl5 in the presence of zeolite may be accompanied by the melting of the molybdenum salt, involving the participation of several intermediates and products (MoOCl4, MoOCl3, MoO3 and MoO2(OH)2 [22]).

Although wetness impregnation has been found to be an interesting method, the use of an excess of solution is not preferred. Indeed, Mo6+ ion forms in aqueous medium the stable tetraoxidomolybdate(2–) complex ([MoO4]2−) which is easily protonated and polymerized, giving rise to very complex systems of simultaneous equilibria [23, 24]. As for the sublimation method (i.e. chemical vapour deposition, CVD), the primary disadvantage lies in the properties of the precursors. Ideally, the precursors need to be volatile at near-room temperatures. Nevertheless, this is not trivial for a number of elements in the periodic table although the use of metal-organic precursors (e.g. Mo(CO)6 [25]) has eased this situation. The precursors used for sublimation or CVD can also be highly toxic (Ni(CO)4), explosive (B2H6) or corrosive (SiCl4), while the by-products can be hazardous (H2, HF or CO) [26].

Alternative exchange methods have to be investigated in order to introduce Mo into zeolites including sublimation (e.g. gMoCl5, and Mo(CO)6 which evaporates at low temperature, ca. 200 °C [25]) and organic-medium impregnation (e.g. MoO2(C5H7O2)2 in CH2Cl2 [27] and MoCl5 in CH2Cl–CH2Cl [28]).

In view of the great practical importance of Mo-exchanged zeolites, sublimation and organic-medium impregnation are expeditious alternative methods to prepare solid catalysts with definable structure and composition. In this work, we prepared Mo-exchanged ZSM-5 zeolite catalysts by solid-state ion exchange (solid–solid interface: MoCl5(s) + hydrated zeolite), sublimation (gas–solid interface: dehydrated zeolite + MoCl5(g) and hydrated zeolite + Mo(CO)6(s)) and organic-medium impregnation (solid–liquid interface: dehydrated zeolite + MoCl5(s) dissolved in CCl4). We have used several spectroscopic techniques (e.g. X-ray photoelectron and optical spectroscopy) in order to characterize the prepared catalysts, namely those active in ethane ammoxidation into acetonitrile.

Experimental

Catalysts preparation

The Mo-containing solids, denoted as Mo-P, were prepared according to the following protocols:

Solid-state ion exchange

This method consists of mixing the NH4+-ZSM-5 zeolite powder (Si/Al = 26, Zeolyst) with the desired quantity of MoCl5 (Mo/Al molar ratio = 1, i.e. 6 wt% of Mo) in a mortar. The mixture was subsequently heated under helium stream (30 cm3 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 solid was stored and labelled as Mo-SSIE where SSIE stands for solid-state ion exchange.

Impregnation in organic medium

For organic-medium impregnation method, solid MoCl5 was dispersed in anhydrous CCl4 (wt MoCl5/wt CCl4 = 2 × 10−3) under continuous stirring for 1 h at room temperature. The mixture was then transferred in a round-bottom flask containing the NH4+-ZSM-5 zeolite powder (Si/Al = 26) and connected to a rotary evaporator. It is worth mentioning that the zeolite was previously dehydrated under vacuum at room temperature for 12 h. The contact of the dehydrated zeolite with MoCl5/CCl4 mixture was performed by the rotation of the flask for 3 h inside a water bath heated at 70 °C. After the evacuation of the solvent, the mixture (6 wt% of Mo) was dried in an oven at 100 °C for 12 h and then treated under helium stream at 500 °C (2 °C min−1, 30 cm3 min−1) for 12 h. Thereafter, it was treated under pure O2 (30 cm3 min−1) at 500 °C (2°C min−1) during 3 h before being stored and labelled as Mo-IMP, where IMP stands for impregnation.

Exchange by sublimation

Two different precursors were used for the exchange by sublimation: MoCl5 and Mo(CO)6. Starting from MoCl5, we used a specific reactor (Scheme 1) in order to avoid the hydrolysis of the metallic precursor during the in-situ dehydration of the zeolite.

Scheme 1
scheme 1

The set-up used for the exchange of Mo into zeolite by MoCl5 sublimation

Full size image

Firstly, the NH4+-ZSM-5 zeolite powder (Si/Al = 26) was dehydrated at 550 °C under helium (30 cm3 min−1) for 10 h. Subsequently, the reactor was cooled to 50 °C and an excess of salt was introduced (22 wt%, i.e. 7.2 wt% of Mo). The reactor was then heated under argon to 550 °C (2 °C min−1) and was kept for 12 h at the same temperature. The prepared sample was labelled as Mo-SUB where SUB stands for sublimation.

Starting from Mo(CO)6 (see Scheme 2), the exchange consists of mixing the fresh NH4+-ZSM-5 zeolite powder (Si/Al = 26) with the desired quantity of Mo(CO)6 (6 wt% of Mo) in a mortar. The mixture was transferred in a Teflon liner that is sealed and placed in the oven at 100 °C for 12 h. Following the sublimation, the liner was cooled and the obtained powder was heated under helium stream (30 cm3 min−1) between room temperature and 500 °C (2 °C min−1) and then isothermally treated at 500 °C for 12 h. The obtained material was labelled as Mo-CVD where CVD stands for chemical vapour deposition.

Scheme 2
scheme 2

Exchange of Mo into NH4+-ZSM-5 by sublimation of Mo(CO)6 (denoted as chemical vapour deposition in order to discard the ambiguity with the sublimation of MoCl5)

Full size image

Catalysts characterization

The different characterization techniques including inductively coupled plasma optical emission spectrometry (ICP-OES), energy-dispersive X-ray (EDX), thermal analysis coupled to mass spectrometry (TA/MS), nitrogen adsorption–desorption at − 196 °C, X-ray photoelectron spectroscopy (XPS), 27Al magic angle spinning nuclear magnetic resonance (27Al MAS NMR), X-ray diffraction (XRD), temperature-programmed reduction under hydrogen (H2-TPR), temperature-programmed desorption of ammonia (NH3-TPD), diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS) and UV/visible diffuse reflectance spectroscopy (UV/Vis DRS) were previously described [9, 14, 22, 24, 29] and reported in the electronic supplementary material.

Catalytic tests

Gas-phase ammoxidation was carried out at 500 °C using a catalyst weight m = 200 mg and the following gas composition: 10% C2H6, 10% NH3, 10% O2 and 70% He. The total flow rate was maintained at 100 cm3 min−1 which corresponds to a contact time equal to 80 ms. The outlet gases were analysed by two chromatographic units (flame ionization and thermal conductivity detectors). The reaction products are essentially AN, C2H4 and CO2. However, insignificant amounts of CO, CH4 and NO by-products were also produced but were not included in the calculations.

The conversion (Eq. 3), selectivity (Eq. 4) and activity (Eq. 5) were defined as follows:

$${\text{Ethane}}\;{\text{conversion}}\;\left( {X_{\text{E}} } \right):X_{\text{E}} = \frac{{\sum\nolimits_{i} {y_{i} n_{i} } }}{{y_{E} n_{E} + \sum\nolimits_{i} {y_{i} n_{i} } }} \times 100$$
(3)

where i stands for AN, C2H4 and CO2.

$${\text{Selectivity}}\;{\text{of}}\;{\text{product}}\;{\text{P}}_{i} \left( {i = {\text{AN}},\;{\text{C}}_{2} {\text{H}}_{4} \;{\text{and}}\;{\text{CO}}_{2} } \right):{\text{S}}_{i} = \frac{{y_{i} n_{i} }}{{\sum\nolimits_{i} {y_{i} n_{i} } }} \times 100$$
(4)

Here, 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, is:

$${\text{Ac}}_{i} \left( {{\text{mol}}\;{\text{s}}^{ - 1} \;{\text{g}}^{ - 1} } \right) = \frac{{{\text{Hydrocarbon}}\;{\text{flow}}\;\left( {{\text{cm}}^{3} \,{\text{s}}^{ - 1} } \right) \times X_{\text{E}} \times S_{i} }}{{{\text{Molar}}\;{\text{volume}}\;\left( {{\text{STP}},\;{\text{cm}}^{3} \;{\text{mol}}^{ - 1} } \right) \times 10^{4} \times m\;({\text{g}})}}$$
(5)

The turnover frequency (TOF), i.e. the activity per each Mo specie molecule, is determined as follows (Eq. 6):

$${\text{TOF}}\;\left( {{\text{s}}^{ - 1} } \right) = \frac{{{\text{Ac}}_{i} \;\left( {{\text{mol}}\;{\text{s}}^{ - 1} {\text{g}}^{ - 1} } \right)}}{{{\text{Amount}}\;{\text{of}}\;{\text{Mo}}\;{\text{specie}}\;{\text{by}}\;{\text{gram}}\;{\text{of}}\;{\text{sample}}\;\left( {{\text{mol}}\;{\text{g}}^{ - 1} } \right)}}$$
(6)

Previously, we used the TOF concept in order to classify several catalytic systems in terms of activity recorded under steady-state conditions [12,13,14].

Results

Chemical and thermal analyses

The chemical analyses results obtained by ICP and EDX are compiled in Table 1.

Table 1 ICP and EDX results
Full size table

According to Table 1, the sample prepared by the sublimation of Mo(CO)6 (i.e. Mo-CVD solid) exhibited a significant metal weight loss (50%) which could be originated from the evaporation of Mo(C≡O)6 precursor either inside the Teflon liner (in the oven) or during the thermal post-treatment. In effect, our thermal analysis results (unpublished work) revealed that the complete evaporation of Mo hexacarbonyl precursor occurred between 50 and 125 °C. However, in the case of Mo-SSIE solid, the Mo weight loss (49%) would originate from the evaporation of MoCl5 decomposition products (e.g. MoO2(OH)2) during the solid-state ion exchange. Certainly, the formation of gaseous MoO2(OH)2 complex takes places during the decomposition of MoCl5 into MoO3 which reacts with H2O molecules (issued from the zeolite dehydration) according to Eq. 7 [24, 29, 30]:

$${\text{MoO}}_{{3({\text{s}})}} + {\text{H}}_{2} {\text{O}}_{{({\text{g}})}} = {\text{MoO}}_{2} \left( {\text{OH}} \right)_{{2({\text{s}})}} = {\text{MoO}}_{2} \left( {\text{OH}} \right)_{{2({\text{g}})}}$$
(7)

If compared with solid-state ion exchange, the loss of molybdenum over Mo-IMP solid is less pronounced due to lack of water molecules in the medium (i.e. the reaction in Eq. 7 is less displaced to the right side). In fact, the totality of water molecules was evacuated below 50 °C as revealed by the evolution of the MS fragment intensity of H2O (red curve in Fig. 1).

Fig. 1
figure 1

Evolution of the signal intensity of H2O+ MS fragment during the thermal treatment of MoCl5/NH4+-ZSM-5 mixtures prepared under solid-state ion exchange and impregnation conditions

Full size image

In the case of Mo-SUB solid (Table 1), we noticed a significant metal loss (76%) despite the use of an excess of Mo during the preparation procedure (7.2 instead of 6 wt%). The effect of water on the metal weight loss will be thoroughly dealt with in the Discussion section.

Textural and XPS analyses

The textural analysis results, i.e. BET (SBET) and microporous (Smicro) areas, microporous (Vmicro) and porous (Vp) volumes, are compiled in Table 2.

Table 2 Textural analysis results
Full size table

Generally, the areas and the volumes relative to NH4+-ZSM-5 zeolite decrease upon the exchange evidencing that metallic clusters clogged the channels and made some pores inaccessible to N2 molecules. However, the micropore blocking effect [14] was estimated by the calculation of the normalized microporous area given in Eq. 8:

$${\text{Normalized}}\;S_{\text{micro}} = \frac{{S_{\text{micro}} \;{\text{sample}}}}{{S_{\text{micro}} {\text{zeolite}}\; \times (1 - y)}}$$
(8)

Here, y stands for the Mo amount present at the surface, i.e. determined by EDX.

According to Table 2 (last column), Mo-CVD solid exhibited the lowest normalized Smicro value (0.64 vs. 1 for NH4+-ZSM-5) which indicates a pronounced obstruction of the zeolite micropores. In order to understand such a phenomenon, we performed XPS analysis over the solids which exhibited the highest and the lowest normalized Smicro values, i.e. Mo-SSIE and Mo-CVD, respectively. The results are compiled in Fig. 2 and Table 3 (see also Fig. S1 in electronic supplementary material).

Fig. 2
figure 2

a Deconvolution of the XP spectrum of Mo-CVD solid and b XP spectra of Mo-SSIE and Mo-CVD solids

Full size image
Table 3 XPS analysis results
Full size table

In the Mo 3d XP spectrum of Mo-CVD solid (Fig. 2a), two sets of Mo 3d doublets representing Mo6+ in bulky MoO3 (binding energy (BE) at 233.5 and 237.0 eV) and Mo6+ in MoOx moieties (BE at 232.4 and 235.2 eV) were observed. Moreover, the XP spectrum of Mo-SSIE solid exhibited the same sets of doublets (see the positions in Fig. S1).

When compared with Mo-SSIE, the concentration of Mo at the surface of Mo-CVD solid is lower (Table 3) as also revealed by the intensities of the XPS doublets in Fig. 2b. This is likely explained by the diffusion of Mo into the inner pores of Mo-CVD solid as previously reported for Mo/ZSM-5 [31] and Cr/ZSM-5 [8] systems. Nevertheless, Mo-CVD solid loaded low amounts of MoO3 at the surface (71.2 vs. 81.8% for Mo-SSIE solid) which would explain the significant drop in the normalized Smicro value (Table 2).

Structural studies by 27Al MAS NMR and XRD

The 27Al MAS NMR spectra of selected Mo-P solids and NH4+-ZSM-5 zeolite are illustrated in Fig. 3.

Fig. 3
figure 3

27Al NMR spectra of Mo-SSIE, Mo-SUB, Mo-IMP solids and the corresponding support

Full size image

The NMR spectrum of NH4+-ZSM-5 zeolite exhibited a peak at ca. ~ 55 ppm ascribed to the four-coordinate Al framework, while the very weak signal at ~ 0 ppm is attributed to the extra-framework octahedral Al centres [14].

The NMR spectrum of Mo-SSIE solid revealed a strong increase in the intensity of the peak at 0 ppm, while the peak situated at 56 ppm slightly decreases in intensity which proves some preservation of the zeolite structure (see also the Al wt% values determined by EDX at the surface, Table 1). Similar phenomenon, denoted as reversible dealumination, was previously described by Iglesia and co-workers [32] as well as our group [9] over Mo/ZSM-5 solids.

It is well known that the zeolite has a very regular structure with a restricted range of T–O–T angles (here, T represents an individual SiO4 or AlO4 tetrahedron) [33]. Lippmaa et al. [34] reported the following relation (Eq. 9) between the T–O–T angle (θ’) and the 27A1 isotropic chemical shift (δcs) for zeolites:

$$\delta_{\text{cs}} \left( {{\text{A}}1} \right) = - 0.50\;\theta^{{\prime }} + 132\;\left( {\text{ppm}} \right)$$
(9)

From the NMR spectra compiled in Fig. 3, the θ′ values were calculated and then compiled in Table 4.

Table 4 T–O–T angle (θ′) determined from 27Al NMR spectra of NH4+-ZSM-5 reference material and selected Mo-P solids
Full size table

According to Table 4, the introduction of Mo by impregnation in organic medium considerably modified the \({\text{Si}} - {\hat{\text{O}}} - {\text{Al}}\) angle value (155.9° vs. 154.1° for the rest of catalysts) since CCl4 molecules (kinetic diameter d = 5.4 Å [35]) are able to diffuse during impregnation throughout the zeolite channels (dimension ~ 6 Å [30, 36]). In effect, a similar trend was reported by Fyfe et al. [37] with different organic molecules such as acetylacetone (d = 5.6 Å [38]), benzene (d = 5.8 Å [39]), pyridine (d = 5.4 Å [40]) and p-xylene (d = 5.8 Å [39]).

The XRD patterns of Mo-P solid and the corresponding support are presented in Fig. 4.

Fig. 4
figure 4

XRD patterns of NH4+-ZSM-5 zeolite and Mo-P solids

Full size image

It is possible to calculate the unit-cell volume of the orthorhombic MFI zeolite (a ≠ b ≠ c, α = β = γ = 90°) by XRD (V = a × b × c). Effectively, by choosing the appropriate Miller indexes, the unit-cell parameters could be determined by the formula in Eq. 10 [9]:

$$d_{(hkl)} = \frac{1}{{\sqrt {\frac{{h^{2} }}{{a^{2} }} + \frac{{k^{2} }}{{b^{2} }} + \frac{{l^{2} }}{{c^{2} }}} }}$$
(10)

In this context, the diffraction planes having h = k = 0 and l ≠ 0 allow the determination of c, while the planes with h or k = 0 and l ≠ 0 allow the determination of b or a. As a matter of fact, we selected the following planes (002), (012) and (312) in order to determine dhkl from Bragg’s law. The results are compiled in Tables 5 and S1.

Table 5 Crystallographic parameters and unit-cell volume relative to Mo-P and NH4+-ZSM-5 solids
Full size table

The expansion [41] and the contraction [9] of the zeolite’s lattice were previously studied by XRD. According to Table 5, the unit-cell volume of Mo-IMP is higher than those of Mo-SSIE, Mo-SUB and Mo-CVD solids which corroborates the NMR results.

In this study, the XRD patterns of Mo-P solids exhibit narrowed and well-defined diffraction lines similar to those of ZSM-5 zeolite [14] evidencing that the crystallinity is maintained upon the thermal treatment. Moreover, the diffractograms of the prepared materials do not show the lines ascribed to α-MoO3 which indicates the absence of oxide crystallites with sizes above 20 nm [30].

In Fig. 4, we noticed also a broad baseline peak at low 2θ values in the diffractograms of Mo-SSIE and Mo-CVD solids which would probably be ascribed to an amorphous phase (see Fig. 2 in [30]).

H2-TPR study

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

Fig. 5
figure 5

TPR profiles of Mo-P solids

Full size image

The TPR profile of Mo-SUB solid exhibits a shoulder at 450 °C attributed to the reduction of MoO3 into MoO2 [30]. However, the small shoulder centred at 700 °C would correspond to the reduction of MoO2 into metallic molybdenum [30]. It is worth to note that the reduction of the zeolite dehydroxylation products as well as the evaporation of MoO2 would take place at temperatures above 700 °C. As a matter of fact, the interpretation of the high-temperature region is very complicated and its discussion is not of interest in the present work.

When compared with Mo-SUB, the TPR profile of Mo-CVD solid shows an increase in the intensity of the peak ascribed to the reduction of MoO3 with the subsequent shift of the temperature towards the right (from 450 to 470 °C). However, for Mo-IMP and Mo-SSIE solids, the reduction of MoO3 occurs, respectively, at 495 and 515 °C, while tetrahedral Mo6+ ions were reduced at 595 °C. Apparently, the reduction temperature of MoO3 oxide increases with the increase in the interactions developed with the support as depicted in the following sequence: Mo-SUB < Mo-CVD < Mo-Imp < Mo-SSIE.

NH3-TPD and DRIFTS studies

The NH3-TPD profiles of Mo-P solids and the corresponding support are shown in Fig. 6.

Fig. 6
figure 6

NH3-TPD profiles of Mo-containing solids and the corresponding support

Full size image

The NH3-TPD profile of H+-ZSM-5 zeolite and Mo-P solids exhibited a broad desorption peak between 100 and ~ 300 °C ascribed to weakly physisorbed ammonia molecules [42]. The intensity of this peak (denoted as low-temperature peak) depended on the operatory conditions [42], and hence, its contribution in the total acidity will be omitted. The NH3-TPD profile of H+-ZSM-5 zeolite showed a second peak centred at 430 °C which corresponds to the ammonia desorption from strong acid sites [9].

Sarv et al. [43] reported that the framework of ZSM-5 zeolite contains two strong Brønsted acid sites denoted as b and b′. However, in our TPD study, it is not possible to decompose the TPD profile of H+-ZSM-5 into two peaks between 300 and 550 °C since the desorption process depends on kinetics, re-adsorption and diffusion phenomena. As a matter of fact, the deconvolution of the different TPD profiles into Gaussian curves should be performed by taking into consideration the presence of only one high-temperature desorption peak. Figure 7 represents the deconvolution of TPD profiles of Mo-P solids, while Table 6 summarizes the quantitative study results.

Fig. 7
figure 7

Deconvolution of the NH3-TPD profiles of Mo-containing solids

Full size image
Table 6 NH3-TPD results
Full size table

According to Figs. 6 and 7 as well as Table 6, the introduction of molybdenum modified the shape of the NH3-TPD profile of the support due to the consumption of strong acidic sites and the generation of a moderate acidity. Specifically, Mo-IMP solid exhibited the lowest percentage of medium acidic site (46%), while the rest of the solids exhibited quasi-similar percentages of medium and strong acid sites.

The DRIFT spectra of the prepared solids and the zeolite support are presented in Fig. 8.

Fig. 8
figure 8

DRIFT spectra of Mo-P solids and the corresponding support

Full size image

The DRIFT spectrum of the zeolite support exhibits two characteristic bands at 3595 and 3731 cm−1, respectively, assigned to the vibration of Brønsted acid sites (Si–O+H–Al) and the terminal silanol groups (Si–OH) [9]. The exchange of Mo into zeolite induces 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-IMP solid revealed the significant decrease in the 3731 cm−1 band’s intensity which could be explained by the grafting of Mo with silanol groups.

Optical properties: DRS study

Using the optical spectroscopy data, we explored the absorption band gap using the Schuster–Kubelka–Munk function (Eq. 11) and Eqs. (12) and (13) [14, 44]:

$$F(R_{\infty } ) = \frac{{(1 - R_{\infty } )^{2} }}{{2R_{\infty } }}$$
(11)
$$F(R_{\infty } ) = \frac{{(hv - E_{\text{g}} )^{n} }}{hv}$$
(12)
$$[F(R_{\infty } ) \times hv]^{{\frac{1}{n}}} = hv - E_{\text{g}}$$
(13)

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, hν is the photon energy and Eg is the optical energy gap of the material [14, 44]. Plotting [F(R∞) × hv]1/n (n = 2 [14]) versus hν and extrapolating to [F(R∞) × hv]0.5 = 0 yield the Eg value.

Figure 9 represents the deconvolution of the optical absorption spectra of the different solids (for the spectrum of Mo-SSIE solid, see Ref. [14]). We used Origin 8.0 (Microcal Software Inc., USA) in order to decompose the spectra into Gaussian peaks (for more details, see Ref. [14]).

Fig. 9
figure 9

Optical absorption spectra of Mo-SUB, Mo-IMP and Mo-CVD solids

Full size image

The spectra of Mo-SUB and Mo-IMP solids revealed the presence of four O2− → Mo6+ charge transfer bands assigned to MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2− (respectively, grey-, navy-, wine- and purple-coloured curve) [14, 45]. Nevertheless, the spectrum of Mo-CVD solid does not reveal the presence of the band ascribed to [MoO4]2−, while Mo-SSIE solid [14] contains only MoO3 and [Mo2O7]2−.

The concentration of MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2− species in each sample was determined by using the formula given in Eq. 14:

$${\text{C}}_{\text{Mo}} = A_{\text{Mo}} \times k_{\text{Mo}}^{{\prime }}$$
(14)

here AMo is the band’s area of each Mo specie (obtained by the deconvolution in Fig. 9), while k′ stands for the absorption coefficient. In our previous work [14], we determined the absorption coefficient of each Mo specie (see page 627 in Ref. [14]) to be 8.45 × 10−5, 0.31 × 10−5, 3.22 × 10−5 and 2.71 × 10−3 mol g−1 (a.u.)−1, respectively, for MoO3, [Mo7O24]6−, [Mo2O7]2− and [MoO4]2−.

Table 7 summarizes the corresponding quantitative study results, while Table 8 represents the amount and the molar fraction of each Mo moiety.

Table 7 Edge energy (Eg) values relative to each Mo specie and the area of the corresponding band
Full size table
Table 8 Amount of each Mo moiety expressed in μmol g−1 (value between brackets: the corresponding molar fraction in %). Last column: the total amount of Mo moieties (μmol g−1)
Full size table

According to the UV/Vis study results, Mo-P solids contain very low amounts of polymolybdates. However, Mo-SUB and Mo-IMP solids contain essentially monomeric species (77 and 92%, respectively), while Mo-SSIE and Mo-CVD samples contain only MoO3 and [Mo2O7]2− moieties.

Catalytic results

Table 9 illustrates the catalytic behaviour of the prepared materials in ethane ammoxidation into acetonitrile (Eq. 1) at 500 °C. It is worth to mention that NH4+-ZSM-5 zeolite support does not exhibit any activity in the studied reaction (and therefore the results are not included in Table 9), while the rest of the catalysts are stable under the reaction conditions even after several hours on stream. On the other hand, the reproducibility tests do not reveal any significant change in activity.

Table 9 Catalytic behaviour of the prepared catalysts in C2H6 ammoxidation at 500 °C
Full size table

The data compiled in Table 9 indicate a distinguishable difference in catalytic activity. For example, Mo-SSIE and Mo-CVD exhibited similar low activity towards ethylene formation (0.30 μmol s−1 g−1), while the highest value was recorded over Mo-SUB catalyst (\({\text{Ac}}_{{{\text{C}}_{ 2} {\text{H}}_{ 4} }}\) = 0.62 μmol s−1 g−1). Interestingly, Mo-SUB catalyst exhibited the highest selectivity towards ethylene (32%), while Mo-CVD and Mo-IMP solids led to the highest \({\text{S}}_{{{\text{CO}}_{ 2} }}\) values (45 and 20%, respectively).

To compare the activity of Mo species in different catalysts, turnover frequency values (TOF, ethane molecule converted to ethylene or to acetonitrile per one Mo moiety and per second) were calculated using Eq. 6. The results are compiled in Table 10.

Table 10 Ethane ammoxidation: TOF values at 500 °C
Full size table

We believe that the intrinsic activity of a given Mo specie (either MoO3 or [Mo2O7]2− or [MoO4]2−) should be the same in all catalysts. Nevertheless, the data compiled in Table 10 clearly evidenced that Mo species exhibit quite different activity with respect not only to different preparation methods but also to the different Mo moieties present in each catalyst. Such a discrepancy has been previously reported by Wichterlová and co-workers [46, 47]. In our recent work [14], we reported that 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 during ammoxidation. In other words, if the active and/or inactive sites are too juxtaposed, the interactions between reactants and/or intermediates take place, leading to a discrepancy in individual TOF values.

In order to classify the catalysts in terms of catalytic activity, the TOF relative to ([Mo2O7]2− + [MoO4]2−) is calculated in Table 10. In effect, we discarded MoO3 and [Mo7O24]6− from the TOF calculation as the first specie may be either active (crystalline) or inactive (amorphous) [9, 14], while the second one exists at low concentrations (Table 8).

According to the TOF values compiled in Table 10, the specific activity of Mo-P catalysts towards ethylene formation (Eq. 15) increases in the following sequence: Mo-IMP < Mo-SSIE and Mo-CVD < Mo-SUB. However, the activity in ethane ammoxidation (Eq. 1 = ∑ Eqs. 15–17) increases in the following sequence: Mo-IMP < Mo-CVD < Mo-SSIE < Mo-SUB:

$${\text{C}}_{2} {\text{H}}_{6} + {\text{O}}_{2} \to {\text{C}}_{2} {\text{H}}_{4} + {\text{H}}_{2} {\text{O}}$$
(15)
$${\text{C}}_{2} {\text{H}}_{4} + {\text{NH}}_{3} \to {\text{CH}}_{3} {\text{CH}}_{2} {\text{NH}}_{2}$$
(16)
$${\text{CH}}_{3} {\text{CH}}_{2} {\text{NH}}_{2} + {\text{O}}_{2} \to {\text{CH}}_{3} {\text{CN}} + 2{\text{H}}_{2} {\text{O}}$$
(17)

Discussion

Theoretically, the diffusion of the trigonal-bipyramidal MoCl5 molecule (kinetic diameter is 5.9 Å [22]) throughout the channels of the microporous NH4+-ZSM-5 zeolite (channel dimensionality: 5.3 × 5.6 and 5.1 × 5.5 Å [48]) requires a driving force, i.e. a specific thermal activation. Nevertheless, in reality, during the thermal treatment of the fresh zeolite under inert gas, water molecules evolve (Eq. 18), which modify the structure and the geometry of MoCl5:

$$\left[ {{\text{NH}}_{4}^{ + } - {\text{ZSM}}-5} \right]\;x\;{\text{H}}_{2} {\text{O}} = {\text{NH}}_{4}^{ + } - {\text{ZSM}}-5 + x\;{\text{H}}_{2} {\text{O}}$$
(18)

In fact, it has been demonstrated (see the graphical abstract of Ref. [22]) that the thermal treatment of pure MoCl5 under helium stream led to the formation of MoOCl4 and MoOCl3 decomposition intermediates, while, in the presence of NH4+-ZSM-5 zeolite, pure MoCl5 was decomposed into MoOCl4 and MoO2(OH)2. Consequently, in this study, the hydration state of the used zeolite was the focus of our attention.

Starting from anhydrous MoCl5, we used a fresh (humid) zeolite sample for exchanging molybdenum in the solid–solid interface, i.e. during solid-state ion exchange. However, the exchange in the solid–liquid interface by impregnation was carried out with a dehydrated zeolite sample and anhydrous MoCl5 dissolved in CCl4. Dehydrated CCl4 was chosen in order to form the same decomposition intermediate as obtained during the solid-state ion exchange (i.e. MoOCl4). In fact, electron-rich CCl4 molecule oxidizes MoVCl5 into MoVIOCl4 as revealed by our TA/MS results (not shown). Nevertheless, the exchange at the solid–gas interface using sublimation of anhydrous MoCl5 was performed also with dehydrated zeolite sample. In these conditions, MoCl5 would be transformed in the absence of water into the volatile MoOCl4 (between 75 and 200 °C [22]) and then into gaseous MoOCl3 (at 313 °C [22]). The chemical vapour deposition is a combined preparation method which consists of reacting the hydrated zeolite and the volatile Mo(CO)6, firstly in the solid–gas and then in the solid–solid interface. In the Teflon liner, the evaporation of Mo(CO)6 occurs (Eq. 19), while during the thermal post-treatment, adsorbed Mo(CO)6 (Eq. 20) losses C≡O molecules (Eq. 21) as revealed by the evolution of the signal intensity of CO+ MS fragment (see Fig. S2). It is important to note that Mo(CO)6 molecule has a significant size (the distance between opposite O atoms is 6.4 Å [49], see Scheme S1) and is therefore adsorbed at the zeolite surface instead of diffusing throughout the channels. On the other hand, the use of hydrated zeolite is necessary in order to avoid the formation of Mo(0) (Eq. 22) at the detriment of MoOx species (Eq. 23):

$${\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{s}})}} = {\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{g}})}}$$
(19)
$${\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{g}})}} = {\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{ads}})}}$$
(20)
$${\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{ads}})}} \to {\text{Mo}}\left( 0 \right) + 6{\text{CO}}_{{({\text{g}})}}$$
(21)
$${\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{ads}})}} + {\text{dehydrated}}\;{\text{zeolite}} \to {\text{Mo}}\left( 0 \right) + {\text{dehydrated}}\;{\text{zeolite}}$$
(22)
$${\text{Mo}}\left( {\text{CO}} \right)_{{6({\text{ads}})}} + {\text{hydrated}}\;{\text{zeolite}} \to {\text{MoO}}_{x} \cdots {\text{zeolite}}$$
(23)

Prior to the SSIE, there is a partial transfer of H2O molecules from the hydrated NH4+-ZSM-5 zeolite to the anhydrous MoCl5 in the agate mortar. However, during the exchange at the solid–solid interface, the hydrated salt (MoCl5·xH2O) is transformed into MoOCl4(g) and then into MoO3 [22]. This later oxide reacts with H2O issued from the zeolite hydration (Eq. 18) and forms the volatile MoO2(OH)2(g) species (Eq. 7) leading to a significant Mo weight loss (49%, Table 1).

Based on the optical properties studied by UV/Vis DRS (Table 8), residual MoO3 over Mo-SSIE solid represents 67% of the total Mo, while the remaining 33% of Mo exists in the dimeric form (78.9 μmol g−1) reducible at 595 °C (Fig. 5). As a matter of fact, this solid does not contain monomeric nor heptameric species. However, the acidity measurement revealed that the formation of dimeric species was accompanied by the consumption of Brønsted acid sites (see the NH3-TPD results in Fig. 6 and Table 6 as well as the attenuation of the DRIFTS band’s intensity at 3595 cm−1 in Fig. 8). The residual band at 3595 cm−1 in Fig. 8 would belong to the vibrations of residual Brønsted acid sites (peak at 356 °C, Table 6). Apparently, the consumption of a fraction of Brønsted acid sites was accompanied by the formation of a medium acidic site (63%, Table 6) which evolves ammonia at 286 °C.

According to XPS results (Table 3), the surface of Mo-SSIE solid contains low amounts of [Mo2O7]2− species (18.2%). However, residual MoO3 (reduced under H2 at 515 °C, Fig. 5) which does not transform into MoO2(OH)2 (in Eq. 7) remained at the surface (81.8%, Table 3). It is worth to note that residual MoO3 oxide exists in a crystalline phase since the corresponding Eg value (2.85 eV, Table 7) is close to the one obtained for pure crystalline α-MoO3 (2.88 eV, Table 7).

Following the impregnation of the dehydrated zeolite with MoCl5/CCl4 solution at 70 °C, MoCl5 is transformed into MoOCl4(s) in the presence of certain degree of moisture (see the evolvement of H2O between room temperature and 50 °C, red curve in Fig. 1). However, MoOCl4(s) evaporates during the thermal post-treatment (precisely between 70 and 200 °C based on the TA/MS results obtained with pure MoCl5 [22]) and reacts with the zeolite. Due to the absence of H2O in the medium, the transformation of MoOCl4(g) into MoO3 (see page 275 in Ref. [22]) during the thermal activation is less-extended and the percentage of MoO3 in Mo-IMP solid is therefore low (6%, Table 8). Additionally, the transformation of MoO3 and H2O into MoO2(OH)2 (Eq. 7) is excluded and therefore the Mo weight loss is low (10%, Table 1). The exchange of Mo by impregnation led to the expansion of the zeolite unit-cell (27Al MAS NMR and XRD results, Tables 4 and 5) upon the departure of CCl4 guest molecules. However, following the exchange, very low amounts (8.4 μmol g−1, Table 8) of dimeric species were stabilized over Mo-IMP solid which corroborates the moderate decrease (from 100 to 54%) in the percentage of strong acid sites (Fig. 7 and Table 6). Theoretically, the condensation of a Mo monomer with another one to form Mo dimer is disfavoured due to the unit-cell expansion, and therefore, Mo-IMP solid stabilized very high amounts of monomeric Mo (504 μmol g−1, 92%, Table 8) reduced under H2 at 595 °C (Fig. 5). It is important to note that the formation of monomeric species over Mo-IMP solid led to the consumption of Si–OH groups (see the attenuation of the DRIFTS band’s intensity at 3731 cm−1, Fig. 8) following a mechanism previously described [9].

The residual MoO3 over Mo-IMP solid (31.3 μmol g−1, 6%, Table 8), reduced under H2 at 495 °C (Fig. 5), occupied the zeolite channels (micropore blocking effect is 0.74 vs. 0.82 for Mo-SSIE, Table 2). On the other hand, this oxide phase was stabilized in amorphous state due to the discrepancy between Eg values obtained with Mo-IMP solid and crystalline α-MoO3 (2.97 and 2.88 eV, respectively, Table 7).

During the sublimation, MoCl5 is transformed into MoOCl4(s) in the presence of some moisture (probably retained during tarring). Nevertheless, the evaporation of MoOCl4(s) occurred between 70 and 550 °C and extended for 12 h, leading to a very high metal loss (76%, Table 1). The exchange of MoOCl4(g) in the solid–gas interface led to the consumption of 64% of Brønsted acid sites (Table 6) and to the formation of 15.1 μmol g−1 of [Mo2O7]2− species (i.e. 10%, Table 8). Besides dimeric species, 77% of the total Mo content over Mo-SUB solid exists in the form of [MoO4]2− though the corresponding reduction feature is not distinguishable (Fig. 5) due to the strong interactions established with the support (Eg ([MoO4]2−) = 5.90 vs. 5.33 eV for Mo-IMP solid, Table 7) and its low concentration (122.0 vs. 504.0 μmol g−1 for Mo-IMP solid, Table 8).

Owing to the absence of H2O in the atmosphere, Mo-SUB solid stabilized low amounts of crystalline MoO3 (18.6 μmol g−1, Table 8) reduced under H2 at 450 °C (Fig. 5).

In the case of Mo-CVD solid, Mo(CO)6 losses carbonyl ligands (Eq. 21) during the thermal treatment, but the presence of H2O issued from the zeolite led to the formation of significant amounts of MoO3 (126 μmol g−1, Table 8) at the near-surface (71.2%, Table 3). These results would explain the significant micropore blocking effect over Mo-CVD solid (0.63, Table 2) when compared with Mo-SSIE (0.82, Table 2) which loaded 81.2% of MoO3 at the surface. Additionally, MoO3 oxide exists in amorphous state (Mo-CVD and Mo-IMP solids exhibit the same Eg value, Table 6) and was reduced under H2 at 470 °C (Fig. 5).

Apart from the MoO3 oxide phase, Mo-CVD solid stabilized 75 μmol g−1 (36%) of dimeric species (Table 8) which consumed 61% of the Brønsted acidic sites (Fig. 7 and Table 6). Nevertheless, the reduction of these species under H2 is not detected (Fig. 5) as they established very strong interactions with the support (Eg ([Mo2O7]2−) = 4.72 vs. 4.20 eV for Mo-SSIE solid, Table 7).

The differences between the catalytic properties of Mo-P catalysts are noticeable. However, in this section we will only discuss the TOF and some specific selectivity values obtained at 500 °C. In effect, ethane ammoxidation (Eq. 1) is a very complex process in which three distinct steps overlap (Eqs. 15–17). Firstly, Mo-CVD and Mo-SSIE catalysts exhibited quasi-similar percentages of Mo species as well as quasi-similar TOF values towards C2H4 formation (0.40 and 0.38 s−1, Table 10). On the other hand, the selectivity towards C2H4 does not differ significantly (13 and 9%, Table 9), while the selectivity towards CO2 is different. Apparently, for these two catalysts, the reaction expressed by Eq. 15 was successfully catalysed over dimeric Mo. Nevertheless, C2H4 molecules issued from Eq. 15 were successfully transformed into ethylamine (Eq. 16) and then into acetonitrile (Eq. 17) over Mo-SSIE, while amorphous MoO3 oxide catalyses the secondary reaction over Mo-CVD catalyst, leading to higher selectivity towards CO2 (45% vs. 4% over Mo-SSIE, Table 9).

For Mo-SUB catalyst, the activity towards C2H4 is very high essentially due to the presence of dimeric (TOF ([Mo2O7]2−) = 4.11 s−1, Table 10) and monomeric Mo (TOF ([MoO4]2−) = 0.51 s−1, Table 10). In fact, over this catalyst, the contribution of monomeric Mo in ethane activation (Eq. 15) is more significant than that of crystalline MoO3 (Eg = 2.84 eV, Table 7). This point is confirmed by the fact that Mo-SSIE loaded higher amount of crystalline MoO3 than Mo-SUB (158.0 vs. 18.6 μmol g−1, Eg = 2.85 eV) but exhibited lower ethylene activity (0.30 vs. 0.62 μmol s−1 g−1, Table 9).

Despite the higher activity towards ethylene, Mo-SUB solid exhibited low \({\text{Ac}}_{{{\text{CH}}_{ 3} {\text{CN}}}}\) value (1.18 μmol s−1 g−1, Table 9) evidencing that Eqs. 16 and 17 are less displaced to right side. Indeed, the very high Eg([MoO4]2−) value (5.90 eV, Table 7) indicates that monomeric species are strongly anchored to the support which discourages the ethylene desorption upon ethane oxidative dehydrogenation (Eq. 15). On the other hand, Mo-SUB solid is less acidic (0.07 and 0.04 mmol NH3 g−1, respectively, for medium and strong acidity, Table 6), and therefore, the adsorption of NH3 in Eq. 16 is less extended.

In the case of Mo-IMP catalyst, C2H6 was moderately converted into C2H4 (activity towards ethylene = 0.25 μmol s−1 g−1, Table 9) despite the presence of higher amounts of monomeric species (504 = μmol g−1, Table 8). However, amorphous MoO3 transforms a fraction of C2H6 into CO2 (\(S_{{{\text{CO}}_{ 2} }}\) = 20%, Table 9), and due to the lack of dimeric species (8.4 μmol g−1, Table 8), a fraction of C2H4 issued from Eq. 15 would also be oxidized into CO2. The ethylene molecules which were able to react with ammonia in Eq. 16 transform into AN in Eq. 17. The expansion of the unit cell would inhibit the interactions between ethylamine molecules and improves therefore the AN formation (\(S_{{{\text{CH}}_{ 3} {\text{CN}}}}\) = 72%, Table 9).

Conclusions

In this study, Mo/ZSM-5 zeolite catalysts were prepared, characterized and tested in ethane ammoxidation into acetonitrile at 500 °C and at a very low contact time (80 ms). Different protocols were adopted for the preparation of the different solids by taking into account the dehydration degree of NH4+-ZSM-5 zeolite. In terms of catalytic activity, the prepared materials could be classified as follows: sublimation of MoCl5 (TOF = 8.78 s−1) > solid-state ion exchange (TOF = 3.80 s−1) > sublimation of Mo(CO)6 (TOF = 1.28 s−1) > impregnation (TOF = 0.44 s−1). For sublimation, the use of MoCl5 and dehydrated zeolite sample inhibited the formation of the inefficient MoO3 in amorphous state which catalyses the hydrocarbon(s) combustion into CO2. However, the lack of H2O in the impregnation medium avoided the Mo weight loss, which enhanced the grafting of monomeric species into silanol groups at the detriment of active [Mo2O7]2− species formation. Using hydrated ZSM-5 sample, the sublimation of Mo(CO)6 and the exchange of MoCl5 in the solid state led to quasi-similar percentages of MoO3 and dimeric Mo. Nevertheless, the formation of amorphous MoO3 during the sublimation of Mo(CO)6 deteriorates the catalytic activity. Whatever the used method, the different prepared materials stabilized very low amounts of heptameric Mo which does not contribute to ethane ammoxidation into acetonitrile.