Facile method for synthesis of nanosized β–MoO₃ and their catalytic behavior for selective oxidation of methanol to formaldehyde

Molybdic acid (H₂MoO₄, powder) and α-MoO₃ was supplied by Sigma-Aldrich; 37% HCl solution, 68% HNO₃ solution and high purity methanol (99.99%) were purchased from Merck. The entire chemicals were used as received without any further purification.

4 g of commercial molybdic acid was dissolved in 36 ml of 37% HCl solution to prepare a yellowish transparent solution. 3 ml of 68% HNO₃ solution was then added and the received solution was aged at 90 °C for 30 min before evacuated at this temperature for 30 min and 150 °C for 1 h to eliminate HCl and water. The as-received powder was then transferred into a quartz reactor and subsequently calcined at 350 °C with a heating rate of 10 °C min⁻¹ for 2 h in an air flow of 25 min ml⁻¹ to get β-MoO₃.

The samples were characterized with differential scanning calorimetry (DSC), x-ray powder diffraction (powder XRD), field emission scanning electron microscopy (FE-SEM), infrared (IR) and Raman spectroscopy. DSC measurement was carried out on a Setaram Labsys Evo in 20 ml min⁻¹ of air flow with heating rate of 10 °C min⁻¹ from room temperature to 500 °C.

Powder XRD patterns were obtained by a powder x-ray diffractometer (ULTIMA IV, Rigaku) with Cu-Kα radiation operating at 30 kV and 20 mA from 10° to 70° with a scanning speed of 2° min⁻¹ and scanning step of 0.03°. The morphologies and microstructures of the synthesized products were characterized using a FE-SEM (JSM-5700F, JEOL, Japan) operating at 20 kV.

Fourier transform infrared spectra were recorded on a Bruker Tensor 27 spectrometer (Germany) using KBr pellets in the 4000–400 cm⁻¹ region. Raman measurements were carried out at ambient condition using a Labram-HR (Horiba Jobin Yvon) spectrometer. Raman scattering was excited at 632.8 nm by means of a HeNe laser. The laser beam was focused using an objective lens (50×) with laser power less than 5 mW and 0.5 cm⁻¹ of resolution. The system was calibrated with SiO₂/Si wafer with peak at 520 ± 0.5 cm⁻¹.

The oxidation of methanol was carried out using a fixed bed quartz reactor (500 mm length and 10 mm inner diameter) at atmospheric pressure. Feed mixture of 6.2%(v/v) methanol in air was prepared by bubbling air into methanol at 5 °C. An amount of 0.15 g catalyst (0.25–0.5 mm) was mixed with quartz of the same size at ratio of 1/3 (v/v) and used for each run. The catalyst was activated at 350 °C for 1 h in an air flow of 25 ml min⁻¹ and then evaluated at varying temperature and feed flow rate. The reaction was left to reach the steady state for 2 h before analyzing all products. CO and CO₂ were continuously recorded by a gas sensor (Senko); methanol, formaldehyde, methylformate (MF), dimethylether (DME) and dimethoxymethane were analyzed on a gas chromatography HP 5890 Series II equipped with a TCD detector and HP-Plot U capillary column (30 m × 0.32 mm × 10 μm) by ChemStation Software.

The DSC curve of the prepared β-MoO₃ was shown in figure 1. An exothermic peak at around 433 °C shows the phase transformation from metastable phase to stable α-MoO₃. This phase transformation temperature lies in the range of reported results. The highest known phase transition temperature from β to α-MoO₃ is 450 °C [17], and the lowest one is 387 °C [23]. Moreover, the existence of only one exothermic peak may be considered as a proof of pure metastable MoO₃ phase presented in our sample.

Zoom In Zoom Out Reset image size

In order to study the crystalline structure of the prepared samples, an XRD measurement was carried out. Figure 2 shows the XRD patterns of commercial molybdic acid, commercial α-MoO₃ powder and the as-prepared molybdenum oxide powder. For comparative purposes, the XRD pattern of β-MoO₃ produced by Mizushima's method [23] was also included. None of typical peaks of H₂MoO₄ precursor as well as stable phase was detected in XRD profile of the prepared sample. As can be seen in curve d of figure 2, the prepared β-MoO₃ exhibits two sharp peaks located at 2θ = 22.9° and 24.9° which were assigned to the characteristic reflections from (011) and (200) planes of β-MoO₃ (JCPDS 47-1081), respectively. These peaks are also excellently fixed with the XRD patterns of the β-MoO₃ produced by Mizushima's method (curve c in figure 2). The results shown that similar product which contains pure β-MoO₃ can be easily obtained by using a more simple and effective method.

Zoom In Zoom Out Reset image size

Additional information regarding structure change during transformations of MoO₃ phases was obtained using Raman spectroscopy. Figure 3 presents the Raman spectra of the commercial H₂MoO₄, commercial α-MoO₃, β-MoO₃ produced by Mizushima's method [23] and the prepared MoO₃ sample. As cited in literature, there are three types of stretching vibrations that can be used to identify α-MoO₃: (1) a sharp peak at 996 cm⁻¹ due to stretching vibration of Mo = O; (2) a very strong and well-defined peak at 820 cm⁻¹ originated from stretching vibrations of Mo–O–Mo groups; and (3) a couple of weak peaks at 666 and 472 cm⁻¹ is characteristic of the vibrations of Mo₂O₂ units formed by edge shared MoO₆ octahedra [30–34]. Three well-defined peaks centered at 991, 816, and 663 cm⁻¹ were detected in the case of commercial α-MoO₃ powders as can be seen in curve b of figure 3; however, no Raman peak at around 990 cm⁻¹ and 400–700 cm⁻¹ was observed in the case of prepared β-MoO₃ (curve d in figure 3) due to the absence of Mo = O and Mo₂O₂ entities in this metastable phase. Moreover, the Raman peaks of the prepared β-MoO₃ which are located at 773, 848 and 905 cm⁻¹ are nearly similar to those of the β-MoO₃ produced by Mizushima's method [23] (curve c in figure 3) and other reported results [30, 35]. It can be concluded that a pure β-MoO₃ powder was successfully produced by a more simple method than those reported in the literature.

Zoom In Zoom Out Reset image size

Raman spectra of the prepared molybdenum trioxides after catalytic activity study at different reaction temperatures are demonstrated in figure 4. The reaction of methanol oxidation to formaldehyde is strongly exothermic; therefore, the catalyst was deactivated more rapidly with an increase in reaction temperature. Similarity of spectrum of the as-prepared β-MoO₃ (curve d of figure 3) and that of the used one after catalytic study at 320 °C (curve a in figure 4) indicates that the prepared catalyst was still stable in this reaction condition. The spectra of sample after catalytic test at 350 °C (curves b, c and d in figure 4) exhibit the presence of alpha MoO₃. Since the prepared sample only transfers into alpha MoO₃ at temperature above 420 °C as pointed out by DSC measurements, the formation of alpha MoO₃ during catalytic test at 350 °C may originate from the local overheating due to the considerable amount of heat released from reaction. It can be seen obviously that the longer time on stream, the more amount of alpha phase transformed during catalytic test.

Zoom In Zoom Out Reset image size

IR spectroscopy was employed to additionally identify MoO₃ phases. Figure 5 presents the IR spectra of the prepared MoO₃ powders and commercial one. There are three main absorption patterns in the region of the stretching vibrations, between 1050 and 400 cm⁻¹ that can be used to identify α-MoO₃ (curve a in figure 5): (1) a narrow IR band at 990 cm⁻¹ due to the vibration of the Mo = O terminal bond; (2) an absorption pattern with two maxima at 876 and 818 cm⁻¹ coming from the vibration of Mo–O–Mo bridging bonds; and (3) a broad and complex band centered at about 600 cm⁻¹ is characteristic of the vibration of Mo₂O₂ units formed by edge shared MoO₆ octahedra [30, 36]. Therefore, the main difference between alpha and beta phases was found in the absorption regions around 600 cm⁻¹ and at 990 cm⁻¹. As can be seen in curve b of figure 5, no absorption bands at 990 cm⁻¹ and bands around 600 cm⁻¹ is observed due to the absence of Mo = O and Mo₂O₂ entities in this polymorph. Insteed, the bands at 900 cm⁻¹ and 790 cm⁻¹ can be observed in the case of monoclinic MoO₃ [37, 38].

Zoom In Zoom Out Reset image size

In order to further understand structure of the commercial and prepared MoO₃ before and after catalytic test at different conditions, SEM images of them were recorded and depicted in figure 6. It can be seen that the prepared β-MoO₃ was composed of small primary particles in the range 50–100 nm (figure 6(a)) and its morphology was dramatically changed after methanol oxidation at 350 °C for 15 h (figure 6(b)). The morphology similarity of the used β-MoO₃ and commercial α-MoO₃ (figure 6(c)) was observed as a proof of metastable phase transformation during reaction. Both of them were composed of micro blocks as can be seen in their SEM images.

Zoom In Zoom Out Reset image size

The major product was formaldehyde for all catalytic tests. Small amounts of MF and DME were observed as byproducts. CO_x was not detected for all tested conditions. The carbon balance was within 99.5 ± 2% for all measurements.

The activity and the selecticity of the prepared MoO₃ samples were evaluated as a function of temperature. The reaction was left for 2 h before analyzing all products. As can be seen in figure 7, the conversion and selectivity increase with rising temperature. In the case of using the prepared β-MoO₃ as a catalyst, formaldehyde selectivity can reach up to 98.8% at about 99% methanol conversion. It is obviously more active and selective than the commercial α-MoO₃. According to some authors [23, 27, 29], this significant difference is mainly due to the differences between their structures. Both of them are built from the same MoO₆ octahedra but in different ways. A layer structure of α-MoO₃ is formed from stacking many double-layered sheets composed of corner- and edge-shared MoO₆ octahedra along the (010) direction by week van der Waals interactions [39]. As a result, cleavage along the (010) direction without breaking any Mo–O bond is easier than other direction making coordinatively saturated Mo atoms become predominant in case of the bulk α-MoO₃ terminated. However, dissociative chemisorption of methanol was observed to be only occurred on the undercoordinated surface Mo sites [8, 40]. Therefore, an anisotropic structure of α-MoO₃ was considered as an important factor that limits its activity as a catalyst for methanol oxidation [29, 41]. In case of β-MoO₃, its structure is composed of corner-shared MoO₆ octahedra. Accordingly, its symmetric structure can be cleaved in any direction with breaking of the Mo–O bonds, resulting in a large amount of coordinatively unsaturated Mo atoms on the surface. It can attribute to a higher reactivity of β-MoO₃ compared to α-MoO₃.

Zoom In Zoom Out Reset image size

The stability test of the prepared β-MoO₃ were performed at 320 °C and 350 °C for approximate 15 operating hours. Figure 8(a) shows that the methanol conversion and HCHO selectivity catalyzed by the as-synthesized material at 320 °C do not change with time on stream. They are stabilized at around 84% and over 95% respectively for the first 15 h test. Unfortunately, as can be seen in figure 8(b), activity of the prepared β-MoO₃ was stable for only 2.5 h at 350 °C, then decreased with reaction time. After 150 min on stream, methanol conversion decreased sharply from 99% to 18% and then stayed constant for the rest of the reaction time. This deactivation is caused by transformation of β-MoO₃ to the stable phase as confirmed by Raman spectra of the spent catalyst (figure 4). The SEM images of the used β-MoO₃ and commercial α-MoO₃ showed a similarity in their morphology as can be seen in figures 6(b) and (c). As a result, their activity on methanol oxidation was found to be the same. However, the selectivity of the prepared sample changed insignificantly during deactivation whereas both activity and HCHO selectivity decreased considerably in the case of deactivation of iron molybdate [42, 43]. This can be considered as an advantage of the pure MoO₃ catalyst in comparison with the industrial one.

Zoom In Zoom Out Reset image size