Introduction

The oxidative dehydrogenation (ODH) of n-butane to butenes and butadiene offers an attractive route for the production of olefins. The ODH reaction is thermodynamically feasible and is not limited by equilibrium [1]. It possesses many advantages: it is exothermic, irreversible, and can proceed at temperatures lower (by ca. 200 °C) than that required for the non-oxidative pyrolysis process. The ODH approach can limit coke formation and thermal cracking of products, and hence can prolong the life span of the catalysts [2]. The ODH of n-butane has been extensively studied since 1967 and discussed in several review papers [e.g., 3–5]. However, the ODH of n-butane is still far from being attractive for industrial applications because the overall combustion of n-butane, butenes and butadiene leads to low selectivity, especially at high n-butane conversion. Indeed the design of a catalyst that gives a high C4-olefins yield for this reaction is still a challenge.

It has been reported that V2O5 supported on SiO2 [6–9], Al2O3 [10,11], and MgO [12–14] are catalytically active for the ODH of light alkanes for alkene generation. It is generally accepted that the dispersion of V2O5 and the nature of support have great influence on both alkane conversion and alkene selectivity. Kung et al. [6] studied the effect of vanadia loading on silica for the ODH of n-butane. The results showed that the well-dispersed vanadia (0.58 wt% vanadium loading) catalyst was more selective than the one (6.4 wt% vanadium loading) with crystalline V2O5 formation.

Mesoporous molecular sieves M41S can provide more room for the dispersion of vanadia and hence generating more active sites [15–18]. Solsona et al. [16] reported that compared with amorphous silica, the use of MCM-41 as support resulted in higher concentration of isolated tetrahedral vanadium species and greater density of active sites. As a result, the VO x /MCM-41 catalysts showed higher olefins yield in the ODH of propane and ethane.

Using a triblock copolymer as template, Zhao et al. [19] generated the mesoporous SBA-15, which possess structures more regular and channel walls much thicker than those of MCM-41. The pore size of SBA-15 can be controlled and the pores are well-defined. The SBA-15 materials are thermally stable and of high surface areas. With these unique properties, SBA-15 has been investigated as catalyst support for reaction such as n-butane oxidation to maleic anhydride [20], selective photo-assisted oxidation of methane to formaldehyde [21], and partial oxidation of methane to formaldehyde [22]. Hess et al. [23] prepared VO x /SBA-15 using grafting/anion-exchange procedure and characterized the catalysts by Raman and UV–vis spectroscopy. The results showed that at vanadium loading of up to 7.2 wt%, the vanadia species are isolated whereas at higher loadings V2O5 crystallites are formed in addition to the monomeric and polymeric vanadia species. Fan and coworkers used VO x /SBA-15 as catalysts for the ODH of propane [24, 25]. They found that catalytic activity over VO x /SBA-15 is higher than that of V2O5/MCM-41 and attributed the enhanced performance to the particularly larger pore size and lower surface acidity of the former. In another work, Hess et al. [26] adopted a preparation procedure similar to that proposed in Ref. [23] to obtain the nanostructured SBA-15-supported vanadia catalysts and found that these materials exhibited high selectivity to acrylic acid in the partial oxidation of propane.

With large pore size, the SBA-15 materials permit easy diffusion of gaseous molecules inside the pore channels. With the anchoring of VO x species onto the inner walls of the mesopores, there is good accessibility of reactant gas to the active sites and rapid departure of the desired products. In this work, we prepared VO x /SBA-15 of different VO x loadings and investigated for the first time the catalytic activity of VO x /SBA-15 catalysts for the ODH of n-butane.

Experimental

Catalyst preparation

The SBA-15 was synthesized according to the procedure described by Zhao et al. [19]. In a typical synthesis, 2.0 g triblock poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) (EO20PO70EO20, P123, Aldrich) was dissolved in 60 g deionized water, and 11.8 g HCl (37%, Aldrich) was added with stirring. Then 4.4 g tetraethyl orthosilicate (TEOS, 98%, aldrich) was added to the solution at 40 °C. After being stirred for 24 h, the mixture was transferred to an autoclave and placed in an oven at 90 °C for 48 h. The precipitate was filtered and then washed with deionized water and acetone, respectively, before being dried at room temperature (RT) and calcined in air at 600 °C for 5 h, with a ramp rate of 1 °C/min.

The loading of VO x (2.24–28.0 wt% V content) on SBA-15 was by means of incipient wetness impregnation of SBA-15 with a solution of ammonium metavanadate and oxalic acid (molar ratio 1:2) as described by Dai et al. [11]. First, the required amount of ammonium metavanadate (99%, Aldrich) and corresponding amount of oxalic acid dihydrate (99%, Aldrich) were dissolved in 4 mL deionized water. Then 1.0 g of SBA-15 powder was added and the mixture was dried at 120 °C overnight. The dried precursor was heated from RT to 500 °C at a rate of 1 °C/min and kept at this temperature for 6 h in air. The samples are denoted as xV-SBA-15, where x represents the V weight percentage in the calcined catalysts. For comparison, a V2O5/Aerosil catalyst with 8.96 wt% V content was also prepared. This catalyst is named as 8.96V-SiO2. In the synthesis, the fumed silica Aerosil 200 (Degussa, surface area = 187 m2/g) was first treated with deionized water and dried at 120 °C overnight before being subject to similar treatment for VO x impregnation. The weight percent of vanadium for all the catalysts are calculated relative to the weight of the support.

Catalyst characterization

The BET surface areas and pore size distributions of the samples were measured by means of a Quantachrome NOVA-1200 instrument. The samples were degassed at 300 °C for 3 h before the adsorption of N2 at −196 °C. Powder X-ray diffraction investigation was conducted using a Rigaku diffractometer (Rigaku D-MAX) with Ni filtered Cu Kα radiation. The morphology of the catalyst was examined via the use of a FEI TECNAI Field Emission HRTEM. The XPS investigation was conducted using a spectrometer equipped with a VG CLAM 4 MCD Analyzer and Mg-Kα (1253.6 eV) radiation. The binding energies (BEs) were calibrated against the C1s signal (284.6 eV) of contaminant carbon. The surface concentrations of elements were estimated on the basis of the corresponding peak areas being normalized using the Wagner Factor database.

H2 temperature-programmed reduction (H2-TPR) was carried out in the 100–900 °C range. Samples (50 mg) were treated in dry air at 500 °C for 1 h and then cooled down to RT in N2 flow (50 mL/min) before being reduced in a flow of 5% H2/Ar (50 mL/min) at a ramping rate of 10 °C/min. The effluent was monitored with a TCD detector. NH3 temperature-programmed desorption (NH3-TPD) was performed on the same instrument. The sample was pretreated in a flow of dry air (30 mL/min) at 500 °C for 1 h. Then the sample was exposed to NH3 at RT for 30 min. After N2 purging of the sample (to remove gas-phase NH3) for 1 h, TPD was performed in N2 flow (50 mL/min) at a heating rate of 10 °C/min, and the desorbed NH3 was monitored by a TCD detector.

Samples after various treatments were transferred to a JEOL JES-TE100 ESR spectrometer operating in X-band without being exposed to air. The EPR spectra were recorded at RT; microwave power was 1.17 mW with a resonance frequency of ca. 9.44 GHz at 100 kHz modulation. The tube for EPR measurement is shown in Scheme 1. The sample was first treated in Treatment Tube. After treatment, the sample was cooled to room temperature in the treatment gas. With the sealing of gas inlet and outlet and the turning (90° anticlockwise) of the tube assembly, the treated sample was transferred to the Measurement Tube for EPR analysis.

Scheme 1.
scheme 1

Sample tube for EPR measurement.

Full size image

Catalytic test

The ODH of n-butane was carried out in a quartz microreactor at atmospheric pressure at a feed gas (C4H10/O2/N2 = 4:8:88) flow rate of 100 ml/min. Unless specified otherwise, the amount of catalyst for each test was 0.2 g. A thermocouple was placed in the middle of the catalyst bed to measure the reaction temperature. The effluent was analyzed on-line by using a GC system (Shimadzu 8A). One GC equipped with FID detector and 0.19% picric acid/Carbograph column was used to separate hydrocarbons. The other one equipped with TCD and columns of Porapak Q and molecular sieve 5A was used to detect CO, CO2 and O2. Conversion of n-butane and selectivity to products were calculated based on the balance of carbon.

Results and discussion

Characterization

Textural properties

Figure 1 gives the pore size distributions of SBA-15 and VO x /SBA-15. The related data are summarized in table 1. As depicted in table 1, SBA-15 has high surface area (881 m2/g), large average pore diameter (51.3 Å), and large pore volume (1.13 mL/g). The loading of VO x on SBA-15 resulted in decreases in surface area and pore volume. It is obvious that the surface area of 8.96V-SBA-15 was much larger than that of 8.96 V-SiO2. The surface density for a nominal monolayer is about 6.6 V/nm2. Except for the 8.96V-SiO2 and 28.0 V-SBA-15 samples, the surface concentration of vanadium remained in the sub-monolayer region.

Figure 1.
figure 1

Pore size distributions of xV-SBA-15 catalysts.

Full size image
Table 1 Physical properties of SBA-15, xV-SBA-15, and V2O5/Aerosil 200
Full size table

SBA-15 and all the VO x /SBA-15 catalysts showed type IV isotherms with a sharp increase in gas-uptake volume in the P/P0 range of 0.5–0.8, typical of mesoporous structures with well-defined pores. Compared with SBA-15, the VO x /SBA-15 catalysts are narrower in pore size distribution (figure 1). At V content higher than 4.48 wt%, the pore size of the catalysts deviates from that of SBA-15 and shifts to lower values. At 2.24 wt% V content, the amount of VO x in the pores is low enough not to cause obvious change in pore size. At 4.48 wt% V content, with more VO x species anchored on the walls of the SBA-15 mesopores, there was a reduction in pore size (from 81.6 to 76.5 Å). At 8.96 wt% V content, the mesopore size was further reduced to 73.0 Å. From 8.96 to 28.0 wt% V content, there was little change in pore size. The result suggests that addition of vanadium within this range does not lead to a growth in the thickness of V clusters within the mesopores but the further deposition would cause the formation of larger particle outside the mesopores. In general, we observed that with an increase in VO x loading, there is a gradual increase in average pore diameter. The phenomenon is attributed to the blocking of micropores that branch out from the SBA-15 mesopores. The very fact that the average pore diameter (51.3 Å) is smaller than the mesopore diameter (81.6 Å) is a proof of the existence of micropores in SBA-15 [20,27,28].

The HRTEM images of SBA-15, VO x /SBA-15 with 2.24 and 8.96 V content are shown in figure 2. The highly ordered array of mesoporous channels of these samples can be seen clearly. The TEM images reveal that the ordered mesoporous structure of SBA-15 is preserved after it is loaded with vanadium oxide.

Figure 2.
figure 2

TEM images of (a, b) parent SBA-15, (c) 2.24V-SBA-15, and (d) 8.96V-SBA-15.

Full size image

XRD

The small-angle XRD patterns of SBA-15 and VO x /SBA-15 with V content below 13.4 wt% show three well-resolved Bragg peaks at 2θ =  0.90, 1.75, and 2.08°, attributable to (100), (110), and (200) diffraction, respectively. These peaks are characteristics of the hexagonal ordered structure of SBA-15 [19]. With further rise in VO x loading, the (100) peak weakens in intensity and gradually disappeared, a result of the partial blocking of SBA-15 mesopores (by VO x ) and the decline in long-range order of hexagonally arranged porosity. The (110), and (200) diffraction peak can still be detected at 28.0 wt% V content. In other words, the SBA-15 pore structure is retained in the prepared VO x /SBA-15 catalysts.

In the wide-angle XRD study, we observed no peaks of crystalline V2O5 below 13.4 wt% V content, indicative of good VO x dispersion on SBA-15. At or above 13.4 wt% V content, peaks of crystalline V2O5 appeared. These results are in agreement with the data of surface area and pore size distribution. Above 13.4 wt% V content, the specific surface area decreased rapidly. With the accumulation of VO x on the external surface of SBA-15, VO x aggregates to form V2O5 crystallites, blocking some of the pore mouths leading to both decrease in specific surface area and pore volume. Since SiO2 has much lower surface area than SBA-15, V2O5 crystallites formed at lower loading were detected over 8.96 V-SiO2.

XPS

The XPS results of the catalysts are summarized in table 2. The V2p3/2 BEs of VO x /SBA-15 are in the range of 516.4–516.8 eV, lower than that of V2O5 (517.4 eV) reported in the literature [29]. It is clear that the oxidation state of vanadium in the catalysts is below 5+, i.e., there is certain amount of V4+ in the samples. By deconvolution, two components at 516.1 and 517.1 eV are obtained under the V2p3/2 profile, the former is assigned to V4+ whereas the latter to V5+ (figure 3). The percentage of V5+ and V4+ to total V species are depicted in table 2. One can see that there is a decrease in V4+ but a rise in V5+ content with increasing loading of vanadium. During the preparation of VO x /SBA-15, the precursor of VO x (NH4VO3) first formed VO +2 in the acidic solution. Then the VO +2 species was reduced to VO2+ species by oxalic acid. The good stability of SBA-15 and the accommodation of V4+ species inside the SBA-15 matrix favor the retention of tetravalent vanadium. At or above 13.4 wt% V content, with the aggregation of the vanadium species and the formation of V2O5 crystallites on the surface of SBA-15, there is the conversion of V4+ to V5+.

Table 2 XPS results of SBA-15, V2O5, and xV-SBA-15 catalysts
Full size table
Figure 3.
figure 3

Deconvolution of V2p3/2 signals of xV-SBA-15.

Full size image

The O1s BEs are in the 532.8–533.0 eV range, much higher than that of V2O5 (530.2 eV) but close to that of SBA-15 (533.0 eV). So the O1s signals of the catalysts are mainly from SBA-15. Only very weak O1s signal of V2O5 is detected at high VO x loadings, suggesting that most of the loaded VO x species are inside the pores.

The surface composition and the surface V/Si atomic ratios calculated based on V2p3/2 and Si2p peak areas and atomic sensitivity factors are listed in table 2. As shown in figure 4, the slope of the V/Si ratio against V content plot between 2.24 and 8.96 wt% is larger than that between 13.4 and 28.0 wt%. The results give a clear indication that above 8.96 wt%, there is a decline in VO x dispersion on the external surface of SBA-15 due to VO x aggregation. For the sake of comparison, the V/Si atom ratios of xV-SBA-15 based on bulk composition are also included in table 2. The V/Si atom ratios of bulk composition are much larger than that of XPS results. It is because the vanadium species anchored on the inner walls of the pores are undetectable by the XPS technique, which is surface sensitive.

Figure 4.
figure 4

Plot of V/Si atomic ratio versus V content based on the data in table 2.

Full size image

H 2−TPR

In H2-TPR results, reduction of the supported samples occurred within the 340–650 °C range, much lower than the reduction range of V2O5 which is above 660 °C. There was a reduction peak at around 450 °C on all the V-SBA-15 catalysts. Comparing the results with those of V2O5/SiO2 reported by Arena et al. [30], we assign the 450 °C peak to the reduction of surface V ions with unsaturated coordination. It is clear that due to the promoted dispersion of VO x species in the mesopores of SBA-15, there is enhanced amount of V ions of unsaturated coordination anchored on the pore walls. For the samples with V content ranging from 2.24 to 8.96 wt%, there was a reduction peak at ca. 546 °C. This peak became dominant at 4.48 wt% V content and shifted gradually from 546 to 589 °C within the 8.96–28.0 wt% range, and one can see that there is a gradual development of a shoulder at about 615 °C. The peak at 546 °C can be attributed to the reduction of well dispersed tetrahedral vanadium species [30]. The progressive shift of the major H2 consumption peak to higher temperature with loading rise is due to the gradual formation of polymeric vanadium species [31–34]. The shoulder peaks at 615 °C and 665 °C on catalysts with high V content are due to the step-wise reduction of V2O5 crystallites. The presence of V2O5 crystallites has been proved by the wide-angle XRD results.

NH3-TPD

The NH3-TPD results are shown in figure 5. There is a weak peak at 77 °C over SBA-15 which may be related to the NH3 adsorbed at very weak acidic sites. Over the VO x /SBA-15 samples, there are two NH3 desorption peaks partially overlapping in the 70–250 °C range. With rise in VO x loading, there is a rise in peak intensity, and the position of the second component shifts to higher temperatures. The results suggest that there are two kinds of acidic sites and with rise in VO x loading, the catalysts become more acidic. The results are different from those reported by Fan et al. [24]. They observed no NH3-TPD peaks over SBA-15 whereas over VO x /SBA-15 they detected a wide peak at 205–220 °C. Such discrepancies can be due to the difference in ammonia adsorption temperature.

Figure 5.
figure 5

NH3-TPD profiles of SBA-15 and xV-SBA-15.

Full size image

EPR

The EPR characterization of the as-prepared 8.96V-SBA-15 sample and the sample after different treatments were carried out at RT. The as-prepared 8.96V-SBA-15 catalyst shows one axially symmetric signal with hyperfine splitting. The hyperfine splitting is originated from the magnetic interactions between magnetic moments of the unpaired 3d1 electron of vanadium and the magnetic atomic 51V nuclei (I= 7/2, natural abundance, 99.75%). The signal can be attributed to VO2+ species because V5+ is silent to EPR. For the sample treated in air at 500 °C for 1 h, there is a significant reduction in VO2+ intensity due to the conversion of VO2+ to V5+. After reduction in H2 at 500 °C for 1 h, the 8.96V-SBA-15 sample showed well-resolved hyperfine signals similar to those of the as-prepared sample. The enhanced signal intensity of VO2+ species reflects the reduction of V5+ to VO2+ species. The reduction of V5+ to V4+ species during ODH of alkanes has been reported by Brückner et al. [35]. The reduction of V5+ to V4+ during the ODH of n-butane can be seen clearly in the spectrum of the sample treated in reaction gas at 500 °C for 1 h. Due to the increased concentration of VO2+, the signal of VO2+ is strong.

Catalytic activity

The ODH reactions of n-butane with oxygen over the VO x /SBA-15 catalysts yield ethene and ethane (C2), propene and propane (C3), iso-butene and butane (i-C4), 1-butene (1-C4), cis- and trans-2-butene (c−C4 and t−C4), 1,3-butadiene (1,3-C4) and carbon oxides (CO x , i.e., CO and CO2). Pure SBA-15 also shows certain ODH activity with 12.7% selectivity to C4-olefins at 13.7% n-butane conversion at 520 °C and a space velocity of 30,000 cm3h−1g−1 cat . As reported by Owens and Kung [6], V2O5 and 0.53 wt% V/Cabosil silica show a n-butane conversion of 11.0 and 22.8%, and C4-olefins selectivity of 7 and 46%, respectively. The 8.96V-SiO2 studied here with much higher V content showed significantly higher conversion (47.8%) but lower C4-olefins selectivity (15.6%). Compared with V2O5/SiO2,VO x /SBA-15 showed much higher catalytic performance in the ODH of n-butane. From table 3, one can find that higher conversion and selectivity are obtained when the V content of VO x /SBA-15 is below 13.4 wt%. The conversion and selectivity first increases and then decreases with the rise in V content, reaching a maximum value at 4.48 and 8.96 wt%, respectively. At 4.48 and 8.96 wt% V content, the catalysts give a maximum yield of ca. 15% under the adopted reaction conditions. According to Mamedov and Cortés Corberán [5], isolated monovanadate is more selective than polyvanadate (i.e., polymeric vanadate). That is to say, the latter is more active but less selective. But in a reaction of restricted supply of oxygen, there is enhancement in butene selectivity and in n-butane conversion because ODH reaction consumes less oxygen than deep oxidation to CO x . When the vanadia loading is over one monolayer, crystalline V2O5 appears (XRD and TPR results), resulting in decrease in both conversion and selectivity.

Table 3 Catalytic performance of V2O5, SBA-15, and xV-SBA-15 at 520 °C and a space velocity of 30,000 cm3 h−1 g−1 cat
Full size table

Blasco et al. [36] found that selectivity to dehydrogenation products as well as distribution of C4-olefins can be related to the acid-base character of the catalysts. We observed that the distribution of products depends on the physico-chemical properties of the VO x /SBA-15 catalysts, especially on the vanadium species and the acid–base property. It has been reported that 1-butene and butadiene selectivity decreased when the strength and number of acid sites increased, while selectivity to 2-butene and CO x showed the opposite trend [36]. It is generally accepted that on crystalline V2O5   n-butane is more likely to undergo deep oxidation to produce CO x . Similar trends were observed in our results (table 3). The overall selectivity to 1-butene and 1,3-butadiene over the 28.0V-SBA-15 catalyst was much lower than that over the 2.24V-SBA-15. As shown in table 3, CO x selectivity increased obviously with the rise in V content: CO x selectivity was 64.6% at 2.24 wt% V content and became 85.8% at 28.0 wt% V content.

At a n-butane concentration of 4% and a n-butane to oxygen ratio of 1/2, we investigated the effect of reaction temperature (460–540 °C) on the catalytic activity over the 8.96V-SBA-15 catalyst (figure 6). The conversion of n-butane varied slightly within a narrow range of 51–58%. The decline in n-butane conversion above 480 °C could be related to the reduction in the amount of V5+ as revealed by the EPR results, which showed that during the reaction some of the V5+ species were reduced to V4+ species. Another possible reason is that a certain amount of active sites had been covered by the carbon formed during reaction. The increase-and-then-decrease trend of selectivity is clear, and the best yield was achieved at 520 °C. The significant decreases in selectivity and yield above 520 °C is due to the cracking and deep oxidation of olefins to by-products.

Figure 6.
figure 6

n-Butane conversion, C4-olefin selectivity, and C4-olefin yield over 8.96V-SBA-15 as a function of temperature.

Full size image

At reaction temperature of 520 °C and n-butane feed concentration of 4%, we investigated the effect of contact time on the catalytic activity of 8.96V-SBA-15 (figure 7). One can see that the conversion increased linearly with the rise in contact time, reflecting the irreversible nature of the reaction [2]. The selectivity and yield passed through a maximum at contact time of 18.7 gcat h mol −1 c4 . Above 18.7 g cat  h mol −1 c4 , the decline in selectivity and yield was a result of enhanced olefins conversion to cracking products and carbon oxides.

Figure 7.
figure 7

n-Butane conversion, C4-olefin selectivity, and C4-olefin yield over 8.96V-SBA-15 as a function of contact time.

Full size image

Conclusion

Mesoporous VO x /SBA-15 catalysts with 2.24–28.0 wt% V content have been prepared by incipient wetness impregnation and characterized by a number of techniques for the ODH reaction of n-butane. According to the results of XRD and H2-TPR investigations, at or below 8.96 wt% V content, the VO x species show good dispersion and spread on the wall of pore channels as well as on the external surface of SBA-15. At loadings higher than 8.96 wt% V content, VO x species aggregate and V2O5 crystallites are formed on the external surface of SBA-15.

The catalytic selectivity can be related to the extent of the vanadium deposition and acidic nature of the catalysts. With the rise of acidity at higher loadings, there is a drop in 1-butene and 1,3-butadiene selectivity but a rise in CO x selectivity. Compared to 8.96V-SiO2, the 8.96V-SBA-15 catalyst showed significantly higher n-butane conversion and C4-olefins selectivity. Relating the EPR and XPS results with the performance of the catalysts, it is apparent that V4+ has a role to play. Despite V4+ is less active than V5+, the former can lower the vanadium redox potential and is more selective than the latter for butenes formation. Over VO x /SBA-15 catalyst with 4.48 wt% and 8.96 wt% V content, C4-olefin yield is ca. 15% at 520 °C, much higher than the maximum yield (10.4%) reported over conventional V2O5/SiO2 catalysts.