Vanadium Oxide Supported on MSU-1 as a Highly Active Catalyst for Dehydrogenation of Isobutane with CO 2

Vanadium oxide supported on MSU-1, with VOx loading ranging from 2.5 to 17.5 wt. %, was developed as a highly active catalyst in dehydrogenation of isobutane with CO2. The obtained catalysts of VOx/MSU-1 were characterized by X-ray diffraction (XRD), N2 adsorption-desorption, and H2-temperature programmed reduction (H2-TPR) methods and the results showed that the large surface area of MSU-1 was favorable for the dispersion of VOx species and the optimal loading of VOx was 12.0 wt. %. Meanwhile, the catalytic activity of VOx/MSU-1 was investigated, and VOx/MSU-1 with 12.0 wt. % VOx content was found to be the best one, with the conversion of isobutane (58.8%) and the selectivity of isobutene (78.5%) under the optimal reaction conditions. In contrast with the reaction in the absence of CO2, the presence of CO2 in the reaction stream could obviously enhance the isobutane dehydrogenation, which raised the conversion of reaction and the stability of VOx/MSU-1.


Introduction
Dehydrogenation of lower alkanes to alkenes is always attractive research around the world [1][2][3][4].With the use of downstream products of isobutene increasing [5,6], the demand of isobutene is rapidly growing, so the dehydrogenation of isobutane to isobutene is becoming one of the important ways to produce isobutene.The process of dehydrogenation can be classified into two types: direct dehydrogenation and oxidative dehydrogenation.In fact, the direct dehydrogenation of isobutane to isobutene has been achieved in industrialization [7], but there are still some serious technical problems, such as high production cost, complex processing, and deactivation of the catalyst by carbon deposition.As a result, the oxidative dehydrogenation of isobutane (ODB) has received more and more attention, recently [8,9].ODB will not be limited by thermodynamic equilibrium, so the conversion of isobutane can obviously be raised and the energy consumption can be greatly reduced.Due to the deep oxidation of product, there is lower isobutene selectivity using oxygen (O 2 ) as an oxidant [10,11].Recently, CO 2 has received much attention as a co-feed gas for dehydrogenation, because CO 2 can act as a mild oxidant [12][13][14][15].Unlike oxygen, CO 2 will not be able to fully oxidize the catalyst to its original state due to its weak oxidation capacity.It is a thermodynamically stable and a kinetically inert molecule, and it is also cheap and abundant.Moreover, because of its high heat capacity, using CO 2 as a co-feed can alleviate the effects of the exothermal ODB reaction, avoiding hot spots that can lead to cracking of the alkane [14,16].It has been confirmed that CO 2 had the ability to partially refresh active species on the catalysts and participated in the oxidative dehydrogenation [17].When CO 2 was present in the feed, the reaction followed the oxidative dehydrogenation route but was also accompanied by the reverse water gas shift (RWGS) reaction in combination with the direct dehydrogenation route [17,18].Moreover, CO 2 can eliminate coke deposited over the catalysts, to some extent, and improve the stability of catalysts [19].Therefore, in recent years, the research of dehydrogenation of isobutane with CO 2 has been actively studied, which mainly focused on the development of different catalysts.
The catalysts containing vanadium were found to have excellent catalytic activity for the dehydrogenation reaction [20][21][22][23][24][25].Ogonowski et al. [24] loaded vanadium oxides on active carbon (AC), SiO 2 , Al 2 O 3 , and ZnO, and found that the key factor influencing catalytic activity was the alkalinity/acidity of catalyst surface and oxidation-reduction potential of vanadium species.Research of Ma et al. [25] on V 2 O 5 /Al 2 O 3 and V 2 O 5 /SiO 2 catalyst showed that vanadium species with higher dispersion had better catalytic activity.However, the traditional carriers, such as Al 2 O 3 and SiO 2 , possessed low specific surface area, small pore size and wide pore size distribution, which restricted the activity of catalysts.In view of these, mesoporous materials used as the carrier were widely studied.MSU-x is an important family of mesoporous material, with 3D worm-like holes, which possess many advantages of pore structure and is in favor of the diffusion of molecular objects [26].Furthermore, the non-ionic surfactants used as templates in the synthesis of MSU-x materials are low-cost, non-toxic, and biodegradable; thus, it is more suitable for the carriers of catalysts.In the early work of our group, the catalysts had been prepared by loading chromium on MSU-x for catalytic dehydrogenation of propane and ethane [27][28][29][30], which had showed high catalytic activity.
In this paper, MSU-1, was prepared at room temperature using sodium silicate as the source of silicon.Then, vanadium oxide, VO x , was loaded over the carrier of MSU-1 and a series of vanadium-based catalysts, VO x /MSU-1, were prepared by wet impregnation method.The obtained catalysts were characterized with X-ray diffraction (XRD), N 2 adsorption-desorption, and H 2 -temperature programmed reduction (H 2 -TPR) techniques.The catalytic performance of VO x /MSU-1 was investigated for the dehydrogenation of isobutane with or without CO 2 .

Catalyst Characterization
Figure 1 showed XRD patterns of the samples prepared, including undoped and VO x -modified MSU-1 (VO x /MSU-1) in two different regions of 2θ angle, where they were 0.5 ˝-10 ˝and 10 ˝-85 ˝.
In the 2θ range of 0.5 ˝-10 ˝, one well-resolved diffraction peak was observed (Figure 1A), which corresponded to d 100 reflection [31].This result clearly indicated that they all had the typically less long-range-ordered, worm-like mesoporous structure.Compared with undoped MSU-1, the intensity of diffraction peak of VO x -containing MSU-1 gradually decreased with VO x content increasing.However, even in the case of MSU-1 with the highest VO x loading, 17.5 wt.%, the d 100 diffraction peak were still obviously observed, indicating that the mesoporous structure of the catalysts, VO x /MSU-1, remained well during impregnation and subsequent calcination.
According to the large-angle XRD patterns of 10 ˝-85 ˝(Figure 1B), it could be seen that the diffraction peaks corresponding to crystal phase of V 2 O 5 were 15.5 ˝, 20.2 ˝, 26.2 ˝, and 31.1 ˝.However, these peaks were absent for the samples with VO x loading lower than 12.0 wt.%.This could be attributed to high dispersion of crystal phase of bulk V 2 O 5 on the surface of MSU-1.When VO x loading reached 17.5 wt.%, the peaks became very clear, which indicated that more and larger crystal phase of bulk V 2 O 5 appeared over MSU-1 with the increase of VO x loading.The catalytic activity and selectivity of vanadium oxide for oxidative dehydrogenation depended on its structure, dispersion, and the characteristics of its supporting material.The most active catalytic form for this type of reaction was the monomeric, isolated VO x species [32].The presence of bulk V 2 O 5 on the other hand decreased the selectivity and favored total oxidation.So, the fit VO x loading for VO x /MSU-1 catalysts should be 12.0 wt.% or so.and the characteristics of its supporting material.The most active catalytic form for this type of reaction was the monomeric, isolated VOx species [32].The presence of bulk V2O5 on the other hand decreased the selectivity and favored total oxidation.So, the fit VOx loading for VOx/MSU-1 catalysts should be 12.0 wt.% or so.Table 1 listed Brunauer-Emmett-Teller (BET) surface area and pore volume of undoped and VOx-modified MSU-1.From Table 1, it could be seen that the carrier of MSU-1 possessed high-surface area (1021.3m 2 /g) and moderate pore volume (0.54 cm 3 /g), which was favorable for the dispersion of active components.As for VOx/MSU-1 samples, both surface area and pore volume decreased gradually with VOx content increasing in comparison to MSU-1.This could be due to that more VOx loading led to the formation of larger VOx crystals and as a result, a certain amount of MSU-1 pores was blocked by the bulk crystals, which was also confirmed by XRD patterns of VOx/MSU-1 samples (Figure 1).When VOx loading was 17.5 wt.%, the surface area and pore volume of VOx/MSU-1 decreased to 297.9 m 2 /g and 0.30 cm 3 /g, respectively, which was not advantageous to the dehydrogenation of isobutane.The pore diameter distribution of different samples was shown in Figure 2, which was determined by Barrett-Joyner-Halenda (BJH) method.As shown in Figure 2, the peak at 2.2 nm was attributed to the characteristic behavior of mesoporous materials MSU-1, and the pore size distribution was narrow and normal, indicating that the carrier of MSU-1 had a uniform pore structure.Moreover, with VOx content increasing, the peak shape gradually moved to the left, as a whole, and the width of peak also increased in the meantime.When VOx loading reached 17.5 wt.%, the peak had moved to 1.8 nm from 2.2 nm, which suggested that some pores of MSU-1 might be partly occupied by vanadium oxides, resulting in smaller pore size.As a result, the pore size distribution of VOx/MSU-1 catalysts were broadened and the surface area and pore volume were decreased.Table 1 listed Brunauer-Emmett-Teller (BET) surface area and pore volume of undoped and VO x -modified MSU-1.From Table 1, it could be seen that the carrier of MSU-1 possessed high-surface area (1021.3m 2 /g) and moderate pore volume (0.54 cm 3 /g), which was favorable for the dispersion of active components.As for VO x /MSU-1 samples, both surface area and pore volume decreased gradually with VO x content increasing in comparison to MSU-1.This could be due to that more VO x loading led to the formation of larger VO x crystals and as a result, a certain amount of MSU-1 pores was blocked by the bulk crystals, which was also confirmed by XRD patterns of VO x /MSU-1 samples (Figure 1).When VO x loading was 17.5 wt.%, the surface area and pore volume of VO x /MSU-1 decreased to 297.9 m 2 /g and 0.30 cm 3 /g, respectively, which was not advantageous to the dehydrogenation of isobutane.The pore diameter distribution of different samples was shown in Figure 2, which was determined by Barrett-Joyner-Halenda (BJH) method.As shown in Figure 2, the peak at 2.2 nm was attributed to the characteristic behavior of mesoporous materials MSU-1, and the pore size distribution was narrow and normal, indicating that the carrier of MSU-1 had a uniform pore structure.Moreover, with VO x content increasing, the peak shape gradually moved to the left, as a whole, and the width of peak also increased in the meantime.When VO x loading reached 17.5 wt.%, the peak had moved to 1.8 nm from 2.2 nm, which suggested that some pores of MSU-1 might be partly occupied by vanadium oxides, resulting in smaller pore size.As a result, the pore size distribution of VO x /MSU-1 catalysts were broadened and the surface area and pore volume were decreased.H2-TPR patterns of undoped MSU-1 and VOx/MSU-1 catalysts with different VOx loading were comparatively shown in Figure 3.There were no H2 consumption peaks to be found for undoped MSU-1 in Figure 3.For the samples containing 2.5-12.0wt.% VOx, a weak reduction peak at 680-720 K were observed, in which vanadium species existed mainly in the form of surface oligo-species with the octahedral coordination [33].TPR curves of all VOx-supported samples, VOx loading from 2.5 wt.% to 17.0 wt.%, exhibited one strong reduction peak in the temperature range of 820-850 K, which was found to shift to higher temperatures as the VOx loading was increased, suggesting increased particle size of microcrystalline vanadia with loading.This has been confirmed that as the vanadia species became more bulk-like, i.e., the particle size increased with an increase in loading, the vanadia became more difficult to reduce due to bulk diffusion limitations resulting in a shift in TPR peaks to higher temperatures [33,34].The samples, containing more than 7.0 wt.% VOx were characterized by new peaks appearing already between 920 and 970 K, which gradually increased with VOx loading, showing that more bulk-like VOx crystals had been shaped.This had been concluded that there was a correlation between the reducibility of surface vanadium species and their catalytic properties.High activity was ascribed to vanadate species, either isolated or polymerized up to the formation of a VOx monolayer, whereas lower activity was usually associated with the presence of bulk V2O5 [34,35].H 2 -TPR patterns of undoped MSU-1 and VO x /MSU-1 catalysts with different VO x loading were comparatively shown in Figure 3.There were no H 2 consumption peaks to be found for undoped MSU-1 in Figure 3.For the samples containing 2.5-12.0wt.% VO x , a weak reduction peak at 680-720 K were observed, in which vanadium species existed mainly in the form of surface oligo-species with the octahedral coordination [33].TPR curves of all VO x -supported samples, VO x loading from 2.5 wt.% to 17.0 wt.%, exhibited one strong reduction peak in the temperature range of 820-850 K, which was found to shift to higher temperatures as the VO x loading was increased, suggesting increased particle size of microcrystalline vanadia with loading.This has been confirmed that as the vanadia species became more bulk-like, i.e., the particle size increased with an increase in loading, the vanadia became more difficult to reduce due to bulk diffusion limitations resulting in a shift in TPR peaks to higher temperatures [33,34].The samples, containing more than 7.0 wt.% VO x were characterized by new peaks appearing already between 920 and 970 K, which gradually increased with VO x loading, showing that more bulk-like VO x crystals had been shaped.This had been concluded that there was a correlation between the reducibility of surface vanadium species and their catalytic properties.High activity was ascribed to vanadate species, either isolated or polymerized up to the formation of a VO x monolayer, whereas lower activity was usually associated with the presence of bulk V 2 O 5 [34,35].H2-TPR patterns of undoped MSU-1 and VOx/MSU-1 catalysts with different VOx loading were comparatively shown in Figure 3.There were no H2 consumption peaks to be found for undoped MSU-1 in Figure 3.For the samples containing 2.5-12.0wt.% VOx, a weak reduction peak at 680-720 K were observed, in which vanadium species existed mainly in the form of surface oligo-species with the octahedral coordination [33].TPR curves of all VOx-supported samples, VOx loading from 2.5 wt.% to 17.0 wt.%, exhibited one strong reduction peak in the temperature range of 820-850 K, which was found to shift to higher temperatures as the VOx loading was increased, suggesting increased particle size of microcrystalline vanadia with loading.This has been confirmed that as the vanadia species became more bulk-like, i.e., the particle size increased with an increase in loading, the vanadia became more difficult to reduce due to bulk diffusion limitations resulting in a shift in TPR peaks to higher temperatures [33,34].The samples, containing more than 7.0 wt.% VOx were characterized by new peaks appearing already between 920 and 970 K, which gradually increased with VOx loading, showing that more bulk-like VOx crystals had been shaped.This had been concluded that there was a correlation between the reducibility of surface vanadium species and their catalytic properties.High activity was ascribed to vanadate species, either isolated or polymerized up to the formation of a VOx monolayer, whereas lower activity was usually associated with the presence of bulk V2O5 [34,35].

Catalytic Performance
The catalytic performance of VO x /MSU-1 with different VO x loading for the isobutane dehydrogenation were shown in Table 2, where the main byproducts of reaction were also listed out, including propylene (C 3 H 6 ) and methane (CH 4 ).Undoped MSU-1 exhibited low catalytic activity in comparison to VO x -doped samples, with 11.8% conversion of isobutane (i-C 4 H 10 ) and 52.5% selectivity of isobutene (i-C 4 H 8 ) (Entry 1).The conversion of isobutane gradually increased with VO x loading, and reached maximum value, 58.8% (Entry 4), when VO x loading was 12 wt.%.While VO x loading further increased to 17.5 wt.%, the conversion of isobutane begun to decrease to 51.0% (Entry 6).This could be due to that more VO x loading resulted in the poor dispersion of VO x species and formation of more bulk-like VO x crystals, which had been demonstrated by XRD (Figure 1) and H 2 -TPR (Figure 3) characterization.However, the selectivity of isobutene did not exhibit obvious change in a certain extent with the increase of VO x loading, basically keeping at 79% or so.The selectivity of main byproducts, propylene and methane, also remained stable in principle, approximately 10% and 5%, respectively.Moreover, contrasting Entries 4 and 5 in Table 2, it could be found that the presence of CO 2 significantly enhanced the efficiency of isobutane dehydrogenation, which was generally attributed to that CO 2 as a weak oxidant could eliminate hydrogen produced during the dehydrogenation through RWGS reaction [18,36].Coupling of RWGS reaction with the dehydrogenation reaction therefore improved the equilibrium conversion of reaction toward isobutene production.In Table 2, with CO 2 conversion increasing, the conversion of isobutane gradually increased, which further confirmed the role of CO 2 coupling reaction.Reaction conditions: reaction temperature 873 K, total flow rate 24 mL/min, catalyst 0.2 g, V(CO 2 ):V(i-C 4 H 10 ) = 3:1; * V(Ar):V(i-C 4 H 10 ) = 3:1.
The effect of temperature on the dehydrogenation of isobutane with CO 2 were shown in Figure 4, using 12 wt.% VO x /MSU-1 as catalyst.The conversion of isobutane gradually increased as the reaction temperature was raised, whereas the selectivity of isobutene, the object product, decreased on the contrary.Although high temperature could accelerate the dehydrogenation rate of isobutane, meanwhile the elevation of temperature favored cracking of isobutane to produce propylene and methane.So, considering all reaction factors, 873 K was chosen as the optimal reaction temperature, with the conversion of isobutane 58.8% and the selectivity of isobutene 78.5%.
Figure 5 exhibited the catalytic activity of 12 wt.% VO x /MSU-1 as a function of reaction time.As the reaction proceeded, the isobutane conversion of dehydrogenation gave a visible drop, either in the presence of or in the absence of CO 2 , suggesting that the catalyst was deactivated to some degree.In contrast with the reaction in the absence of CO 2 , the presence of CO 2 in the reaction stream here could obviously enhance the stability of VO x /MSU-1, and the drop rate of isobutane conversion with CO 2 was lower than that without CO 2 .For example, the conversion of isobutane decreased from primal 58.8% to 44.2% with CO 2 and 40.5% to 21.9% without CO 2 , respectively, after reacting for 3 h.This was due to that CO 2 in the reaction stream could suppress the formation of coke deposited over VO x /MSU-1 [17,37,38].Moreover, from Figure 5, it could also be seen that the selectivity to isobutene gave a slight increase with proceeding reaction, especially with CO 2 .

Catalyst Preparation
The carrier of MSU-1 was prepared at room temperature using sodium silicate as the source of silicon by the method reported before [22,23].All the catalysts of VOx/MSU-1 were prepared by wet impregnation method.Typically, 2 g of dry MSU-1 was treated with 20 mL of aqueous solution containing the desired amount of NH4VO3 (analytically pure) overnight.The impregnated samples were evaporated and dried at 353 K for 7 h and finally calcined under air at 873 K for 4 h.These catalysts are designated as x wt.% VOx/MSU-1, where x expresses the total VOx loading.

Catalyst Characterization
XRD patterns were recorded in two ranges of 0.5°-10.0°and 10°-85° (2θ) at room temperature, using X'pert Pro MPD X-ray diffractometer from PANalytical (Almelo, Holland) operated at 40 kV and 30 mA, equipped with a Cu Kα X-ray source.

Catalyst Preparation
The carrier of MSU-1 was prepared at room temperature using sodium silicate as the source of silicon by the method reported before [22,23].All the catalysts of VOx/MSU-1 were prepared by wet impregnation method.Typically, 2 g of dry MSU-1 was treated with 20 mL of aqueous solution containing the desired amount of NH4VO3 (analytically pure) overnight.The impregnated samples were evaporated and dried at 353 K for 7 h and finally calcined under air at 873 K for 4 h.These catalysts are designated as x wt.% VOx/MSU-1, where x expresses the total VOx loading.

Catalyst Characterization
XRD patterns were recorded in two ranges of 0.5°-10.0°and 10°-85° (2θ) at room temperature, using X'pert Pro MPD X-ray diffractometer from PANalytical (Almelo, Holland) operated at 40 kV and 30 mA, equipped with a Cu Kα X-ray source.

Catalyst Preparation
The carrier of MSU-1 was prepared at room temperature using sodium silicate as the source of silicon by the method reported before [22,23].All the catalysts of VO x /MSU-1 were prepared by wet impregnation method.Typically, 2 g of dry MSU-1 was treated with 20 mL of aqueous solution containing the desired amount of NH 4 VO 3 (analytically pure) overnight.The impregnated samples were evaporated and dried at 353 K for 7 h and finally calcined under air at 873 K for 4 h.These catalysts are designated as x wt.% VO x /MSU-1, where x expresses the total VO x loading.

Catalyst Characterization
XRD patterns were recorded in two ranges of 0.5 ˝-10.0 ˝and 10 ˝-85 ˝(2θ) at room temperature, using X'pert Pro MPD X-ray diffractometer from PANalytical (Almelo, Holland) operated at 40 kV and 30 mA, equipped with a Cu Kα X-ray source.
N 2 adsorption-desorption measurements were carried out on an Autosorb series ASIMP apparatus from Quantachrome (Boynton Beach, FL, USA).Before measurements, the samples were degassed at 573 K under vacuum for 3 h.Calculation of specific surface area (BET), pore volume and pore size distribution (BJH method) were performed with the software of the apparatus.
H 2 -TPR analysis (self-made experimental instrument) were performed using Ar/H 2 gas mixture (95/5 vol %).The total flow rate of the feed was 30 mL/min.Before experiment the sample (100 mg) was preheated in a stream of dry argon at 873 K for 30 min, and then cooled to room temperature.The samples were heated at 15 K/min to the final temperature of 1073 K. H 2 consumption was measured by a thermal conductivity detector and V 2 O 5 was used as a reference for the calibration of H 2 consumption.

Catalytic Tests
The catalytic tests of isobutane dehydrogenation were carried out in a fixed-bed quartz tubular reactor with an inner diameter of 6.0 mm at 773-973 K and atmospheric pressure.For each test, about 0.2 g of catalyst sample (40-60 mesh) was diluted with quartz grain and loaded in the constant temperature zone of the reactor.The reaction stream was constituted of isobutane and CO 2 or Ar with the mole ratio of n(CO 2 )/n(C 4 H 10 ) or n(Ar)/n(C 4 H 10 ) being 3 and the total flow rate being 24 mL/min.
The reactants and products were analyzed on-line using a gas chromatograph (Shimadzu GC-2014, Kyoto, Japan).One of column was equipped Rt-Al 2 O 3 (30 m ˆ0.53 mm ˆ10.0 µm) and a flame ionization detector.This column was used to analyze hydrocarbons C 1 -C 4 .The second column was equipped GDX-502 (3 m ˆ2 mm) and a thermal conductivity detector for CO and CO 2 analyses, which carrier gas is He.Conversion (C) of isobutane and carbon dioxide and selectivity (S) of isobutene and other byproducts were calculated according to the following equations: where a i was the number of carbon atoms of the compound i and n i is the mole number of the compound i.The calculations did not consider the conversion of isobutane to coke.

Conclusions
In conclusion, the catalysts of VO x /MSU-1, prepared by loading VO x over MSU-1, were active and selective in the oxidative dehydrogenation of isobutane with CO 2 .Their catalytic properties were related to the VO x loading and high surface area of MSU-1.XRD, N 2 adsorption-desorption and H 2 -TPR characterization for VO x /MSU-1 showed that the optimal loading of VO x was 12.0 wt.% and more VO x could led to the formation of larger bulk-like VO x crystals and the decrease of the surface area and pore volume.As a result, 12.0 wt.% VO x /MSU-1 exhibited the highest catalytic activity, with the conversion of isobutane 58.8% and the selectivity of isobutene 78.5% at 873 K.In the feed of reactants, the presence of CO 2 played quite important role on the dehydrogenation of isobutane, which not only could remove H 2 produced in the reaction through RWGS reaction but also suppress the formation of coke deposits and improve the stability of VO x /MSU-1.

Table 1 .
Physical properties of MSU-1 and VOx/MSU-1 with different VOx loading.

Table 1 .
Physical properties of MSU-1 and VO x /MSU-1 with different VO x loading.

Table 2 .
Catalytic performance for isobutane dehydrogenation over VO x /MSU-1 with different VO x loading.