Phosphorous modified V-MCM-41 catalysts for propane dehydrogenation

WANG Xiao-sheng ，YANG Tao ，LI Qin ，LIU Yu-xiang ，DING Yong-chuan

（1. College of New Energy and Materials, China University of Petroleum-Beijing, Beijing 102249, China；2. College of Chemical Engineering, Qingdao University of Science & Technology, Qingdao 266042, China；3. Shandong Shida Shenghua Chemical Group Co. Ltd., Dongying 257061, China）

Abstract: The dehydrogenation performance of vanadyl catalysts was closely related to the form of surface vanadyl species.To enhance the vanadium dispersion, phosphorus was adopted to modify V-MCM-41 catalysts by using organic vanadium and phosphorus precursors. The influence of phosphorus introduction to the mesoporous structure and vanadyl species were investigated by various characterization techniques. The results showed that the catalysts could maintain ordered hexagonal mesoporous structures though the specific surface area slowly decreased along with the increase of phosphorus content. Both the reducibility and dispersion of the surface vanadyl species were improved. The proportion of polymerized vanadyl species obviously decreased due to the presence of phosphorus species. The propane dehydrogenation reaction results showed that both the catalytic performance and the catalyst stability were improved. Both the maximum surface vanadyl site density and optimum propane dehydrogenation performance were obtained over the sample with Si/P molar ratio of 30.

Key words: phosphorous modification；MCM-41；vanadyl species；propane；dehydrogenation

Considering a wide application of vanadyl catalysts in the oxidative dehydrogenation (ODH) and non-oxidative dehydrogenation (i.e., direct dehydrogenation) of light alkanes[1?5], it has been regarded as a promising alternative to the commercialized Pt- or Cr-based catalysts. At present,the ODH process over vanadium-based catalysts still suffered from low activity or difficult control of overoxidation reaction[6?8], thus leading to more attention being paid to the direct dehydrogenation process.

Metal elements were widely used for modifying vanadium-based catalysts to enhance the stability and selectivity. Harlin et al.[9]found that Mg could effectively enhance theiso-butene selectivity and suppress the coke formation during the dehydrogenation reaction over VOx/Al2O3catalysts.Liu et al.[10]also showed that Mg over VOx/SBA-15 catalysts could enhance the selectivity and yield of olefins by improving the reducibility of VOxspecies and decreasing the surface acidity. Reddy et al.[11]increased the dispersion of VOxspecies by adding Ce and Zr to V2O5/SiO2catalysts for alkane dehydrogenation reaction. They found that the deactivation process was effectively prevented by CexZr1?xO2species on the surface. Raju et al.[12]also tested the dehydrogenation performance of V/CeO2-ZrO2and concluded that the CeO2-ZrO2support played a vital role in maintaining stable catalytic activity.Zhou et al.[13]and Sasikala et al.[14]reported that the dehydrogenation performance of the VOx/HMS(hexagonal mesoporous silica) catalyst was significantly promoted by rare earth elements (La and Y) due to the formation of new active phase YVO4and LaVO4. Yang et al.[15]demonstrated that the incorporation of CrOxinto Al2O3led to more reactive VOxover the surface, thus obtaining higher dehydrogenation rate and selectivity. Ajayi et al.[16]also stated that the dehydrogenation performance over Cr modified V/MCM-41 was largely promoted due to the enhanced reducibility and surface acidity which were correlated to uniformly dispersed Cr-V-O species.Ascoop et al.[17]observed that the WOxspecies on the surface could prevent the aggregation of VOxspecies.

Compared to the metal elements mentioned above,non-metallic elements might also be effective for vanadyl catalysts. Previously we proved that boron could largely enhance the dispersion of surface vanadyl species by the formation of B?O?V bonds[18]. However,the boron precursor was relatively expensive. Similar to boron, phosphorus was also widely used in the modification of silicate-based catalysts (such as SAPO type zeolites) due to the advantage of being less expensive for the phosphorus precursors. Besides,phosphorus oxides could form tetrahedral units through P?O bonds similar to tetrahedral silicon oxide tetrahedron and interact with SiO2perfectly by sharing oxygen atoms, which made it suitable for the modification of silica-based catalysts. In this work, the effect of phosphorus modification on the MCM-41 supported vanadyl catalysts was studied. The influence of phosphorus introduction on the structural properties and surface vanadyl species were systematically examined and the dehydrogenation performances were tested.

1 Experimental

1.1 Synthesis of V-P-MCM-41

The phosphorus modified V-MCM-41 catalysts were prepared as follows: Firstly, 3.6 g vanadium acetylacetonate (99%, Energy Chemical), triethyl phosphate (A.R., Sinopharm) and 60.0 g TEOS (A.R.,Merck) were dissolved in 300.0 g methanol (A.R.,Sinopharm) under mild stirring. At the same time,another solution was prepared via the following steps:12.5 g cetyl trimethyl ammonium bromide (A.R.,Sinopharm) was dissolved in 200.0 g deionized water under stirring followed by 75.0 g aqueous ammonia solution (25 %, A.R., Sinopharm). The methanol mixture obtained in the first step was added dropwise to the water solution and stirred vigorously for 30 min.Then the mixture was sealed in autoclaves at 120 °C for 48 h. The obtained solids were filtered, washed with deionized water and dried overnight. After calcination(550 °C, 6 h), the powder was pressed, crushed and sieved to small particles in 30?40 mesh. The obtained catalysts were marked as P-x(xwas the Si/P ratio in the synthesis process). A reference catalyst was also prepared without addition of phosphorus precursors and was denoted as P-0.

1.2 Characterization

The X-ray diffraction measurements of the V-PMCM-41 samples were performed on a Bruker D8 advance XRD system using CuKα radiation in the 2θrange of 1.5°?35°.

The N2physisorption isotherms were measured and analyzed on an ASAP 2020 (Micromeritics)apparatus. The samples were first vacuum-degassed at 350 °C for 5 h before N2adsorption. The specific surface area and pore distribution were determined by BET and BJH method, respectively.

The vanadium content of the catalysts was measured by ICP-AES (Perkin Elmer Optima 7300 V).The catalysts were fully dissolved in the mixture of concentrated HF and HNO3solution for testing.

TPR experiments were carried out in the range of 100?800 °C on an AutoChem 2920 instrument equipped with a MS detector. Typically, 100 mg particles (30?40 mesh) were used for each run.After treating in dry air (500 °C, 1 h) and cooling down, the samples were reduced in a flow of 10%H2/Ar. The heating ramp was kept at 10 °C/min.

NH3-TPD profiles were also obtained on the same AutoChem 2920 equipment. After dehydration in He at 600 °C for 2 h, the sample (0.1 g) was cooled down to room temperature before NH3adsorption. Then NH3desorption was conducted in the temperature range of 40?600 °C under He flow (30 mL/min). The heating ramp was kept as 10 °C/min throughout desorption.

XPS spectra were obtained from a Thermo FisherK-Alpha instrument. The charge calibration was based on the C 1speak at 284.6 eV.

O2pulsed adsorption was used to determine the number of vanadium sites on the surface by assuming Vsurf=Oads[18]. The specific experimental steps and conditions were described in detail in the previous studies[18,19].

The catalysts were characterized by Raman spectroscopy under dehydrated conditions on a Raman spectrometer system (Horiba Lab RAM HR800) at room temperature. The samples were excited with a 325-nm laser and the spectral resolution was 1 cm?1.

The meso-structure of the catalysts was directly observed by TEM (FEI Tecnai F20) operated at 200 kV.

1.3 Propane dehydrogenation reaction

The propane dehydrogenation reactions were carried out in a fixed-bed reactor under atmospheric pressure and packed with 1.0 g catalyst particles in the isothermal zone by quartz sand at both ends. The catalysts were first dehydrated by pure N2at 600 °C(kept for 2 h). The effluent gas was monitored by using an on-line gas chromatography. The reaction temperature was 600 °C and the total gas flow was 50 mL/min (20% propane in N2). The propane conversion and product selectivity were calculated based on carbon atom balance method (coke already ignored).

2 Results and discussion

2.1 Structure and morphology

The XRD patterns of the obtained catalysts were given in Figure1. All samples clearly exhibited three well-resolved diffraction peaks assigned to [100], [110]and [200] of MCM-41, indicating that the mesostructure was not dramatically changed after the introduction of vanadium and phosphorus[20]. No bulk V2O5clusters were detected as there was no distinct diffraction peak of V2O5both in the small and large angle XRD profiles[21]. The peak intensity decreased gradually with the phosphorus content, indicating the decrease of the crystallinity. The diffraction peak position at around 2° slowly shifted towards larger angle in samples containing high phosphorus amount,reflecting that the crystalline interplanar spacing increased after the introduction of phosphorus (shown in Table1). Besides, no diffraction peaks of VOPO4,(VO)2P2O7or other vanadyl-phosphorus species was detected, meaning the absence of bulk vanadylphosphorus species. This also suggested that the vanadyl species were not aggregated or polymerized after the introduction of phosphorus and the possible vanadyl-phosphorus species were also highly dispersed on the surface.

Table1 Physio-chemical properties of V-P-MCM-41 catalysts

The specific surface areas and pore distribution were measured by N2physisorption, as show in Figure2.All samples showed a clear hysteresis loop at aroundp/p0=0.3, suggesting the existence of ordered mesoporous structure. Besides, all samples showed a narrow pore width distribution concentrated in the range of 22?35 ?, which was typical for MCM-41 materials[22]. More details about the specific surface and pore structure parameters were listed in Table1. The specific surface area of the V-P-MCM-41 samples showed an obvious reductive trend alongside with the phosphorus introduction amount while the pore volume slightly increased. Besides, the interplanar spacing and cell parameter of the V-P-MCM-41 also slightly increased with the phosphorus content, suggesting that the introduction of phosphorus enlarged the crystal cell to some extents[23].

The TEM images of V-P-MCM-41 catalysts were given in Figure S1. Clearly could be seen were the long ordered mesopores in all the catalysts. Compared with the V-MCM-41 catalyst (Figure S1 f), there was no significant difference in the TEM images of the V-PMCM-41 samples. This directly proved that mesostructure was not greatly influenced by the introduction of phosphorus.

2.2 Surface vanadyl species

The reducibility of the V-P-MCM-41 catalysts was examined by H2-TPR and the results were given in Figure3. There was a wide main reduction peak in the range of 350?550 °C for all the samples, which should be ascribed to the direct reduction of V5+to V3+of highly dispersed vanadyl species with tetrahedral coordination. Besides, a small shoulder peak (around 600 °C) also could be observed. As the reduction temperature of VOPO4or other vanadyl-phosphorous species was at around 700 °C or higher and there was almost no H2consumption at above 700 °C, the shoulder peak should be ascribed to polymerized vanadyl species with octahedral coordination[24?26]. This indicated that highly-dispersed vanadyl species were dominant on the surface while a small amount of aggregated vanadyl species were still present[27,28].Hence, the reduction peaks were deconvoluted to distinguish different vanadyl species. According to the results in Table2, the proportion of polymerized vanadyl species decreased along with increasing phosphorous content. This suggested that the introduction of phosphorus could decrease the number of octahedrally coordinated vanadyl species, thus overall improving the reducibility of the vanadyl species.

Table2 TPR, TPD and O2 chemisorption results of V-P-MCM-41 catalysts

The acid properties of the V-P-MCM-41 catalysts were investigated by NH3-TPD and the results were given in Figure4. Only one broad peak at below 200 °C was observed for the P-0 sample, suggesting the adsorption of NH3molecules on the surface vanadyl species[29,30]. After the introduction of phosphorus, a peak at lower temperature (around 90 °C) appeared and its integration area increased gradually alongside the phosphorus content. The weaker acid sites should be ascribed to the vanadyl species connected with phosphorus species[31,32]. For all the samples, the acid sites with medium/high strength were rare as the ammonia desorption profile was almost flat in the temperature range of 250?600 °C. This suggested that the introduction of phosphorus only change the weak acid sites by forming highly dispersed vanadylphosphorus species (Table2).

To precisely measure the number of surface vanadium sites, O2chemisorption was used by pulsing a fixed amount of O2through the reduced catalysts. The O2signal curves were illustrated in Figure5 and the calculated surface vanadium sites were listed in Table2.It could be seen that the total O2consumption first increased along with the phosphorus content and then it slightly decreased. Accordingly, the surface vanadyl species also showed a volcano shape with the phosphorus content. The maximum amount of surface vanadium sites (3.35×10?4mol/g) was obtained over the P-30 sample. This indicated that a suitable amount of phosphorus was beneficial for the enhancement of surface vanadium sites. Combined with the results of O2chemisorption and ICP-AES, the decrease of surface vanadium sites might be attributed to the occupation of surface area by the excessive phosphorous species[33,34].Besides, the surface vanadyl species density also exhibited a volcano shape along with the phosphorous content. This further proved that the suitable amount of phosphorous introduction could effectively enhance the vanadyl dispersion.

The chemical state of surface elements over the VP-MCM-41 catalysts was studied by XPS and the results were summarized in Table3. It was clear that the binding energies of O 1sand Si 2pare mainly derived from SiO2[35?37]. According to previous studies[24,38,39], the peaks of V 2p3/2could be deconvoluted into two peaks at 518.4 and 517.0 which should be ascribed to polymerized vanadyl species(V5+) and highly dispersed vanadyl species (V5+). As shown in Figure6 and Table3, a small amount of polymerized vanadyl species could still be identified,which was consistent with the TPR results. After the introduction of phosphorus, the proportion of polymerized vanadyl species first decreased along with the phosphorus content and then slightly increased.This further illustrated that the dispersion of vanadyl species was enhanced and the reducibility of the vanadyl species was improved with suitable amount of phosphorous introduction[40].

Table3 XPS results of V-P-MCM-41

The P2pspectra were shown in Figure S2. Due to the low phosphorus content, it could be seen that the P-50 and P-30 catalysts all exhibited a broad peak, whose specific attribution was difficult to be distinguished.However, the peak of P-10 located at around 134.6 eV was much clearer. As all samples were synthesized in the same way, they should share similar phosphorus species. According to the P2pspectra, most probably,the peak should be ascribed to the P?O bond in the P?Si?O composite oxide (similar to those phosphorus species with tetrahedral coordination inside the SAPO type zeolites) instead of VOPO4(133.8 eV) or P2O5(around 135.0 eV)[41,42], although there might still exist a small amount of V?O?P bonds in the catalysts.According to Table4, the molar percentage of vanadium content slightly decreased due to the introduction of phosphorus while the weight percentage of vanadium increased. The surface vanadium content determined by XPS exhibited the same trend with the results of ICP-AES, suggesting that introducing a suitable amount of phosphorus contributed both to the overall vanadium loading and the surface vanadium sites.

Table4 Surface composition of V-P-MCM-41 catalysts determined by XPS

The form of VOxspecies was further studied by Raman spectroscopy (Figure7). The peak at 1024 cm?1attributed to the terminal V=O bond of highly dispersed tetrahedral VOxwas found on all the samples[43,44].Besides, a shoulder peak at around 990 cm?1which should be ascribed to aggregated VOxspecies that has two- or three-dimensional networks similar to bulk V2O5clusters[45?47]was also found on the P-0 sample.After the phosphorus introduction, the shoulder peak gradually disappeared along with the increasing phosphorus content. This indicated that an appropriate amount of phosphorus could enhance the vanadium dispersion. Compared to the normal V2O5Raman peak which was two- or three-time higher than the peak of the V=O bond, the small peak at 485 cm?1indicated that the highly dispersed vanadyl species were dominant and there was only a small amount of polymerized vanadyl species[7,21,48]. This indicated that the highly dispersed vanadyl species were dominant on the surface of the catalysts, which was consistent with the O2chemisorption and XPS results. Besides, no sharp peaks assigned to vanadyl-phosphorus species(925 and 1035 cm?1) were detected in the spectra,indicating that there were no crystalline vanadylphosphorus species formed on the catalysts.

According to the characterization results above, it could be concluded that the meso-structure of MCM-41 were in general preserved despite that the cell parameters were slightly increased. The introduced phosphorus was successfully incorporated into the silicate, and it had prohibited the formation of polymerized vanadyl species. Therefore, both the dispersion and reducibility of the vanadyl species were improved by the phosphorus introduction.

2.3 Effect of phosphorus to the catalytic performance

The as-prepared V-P-MCM-41 catalysts were tested for propane dehydrogenation and the results were given in Figures 8 and 9. All samples showed a volcano shape during the dehydrogenation process,which should be ascribed to the introduction process of vanadyl catalysts (V5+was reduced to V4+or V3+species). The propane conversion over the P-0 sample decreased rapidly from 33% to 24% during the dehydrogenation process. However, the propane conversion over the V-P-MCM-41 samples increased and the deactivation rate was largely reduced. The propane conversion over these catalysts was closely related to the surface vanadium sites (determined by O2chemisorption), which was consistent with the conclusion that the surface vanadyl sites were the active center for the dehydrogenation process. The selectivity of propylene and methane were shown in Figure9. The propylene selectivity all exhibited an increasing trend during the dehydrogenation process. It was clear that the propylene selectivity over the samples with phosphorous modification was much higher than the P-0 sample. Besides, the propylene selectivity was closely related to the phosphorous content. The initial propylene selectivity over the P-50 sample was 83.0% and it finally increased to 85.2%.However, it increased to 84.8% and 88.5% over the P-10 sample, respectively. The enhancement of propylene selectivity by phosphorous introduction was obvious.Correspondingly, the selectivity of methane over the phosphorous modified samples also declined remarkably. The methane selectivity over the P-0 sample decreased from 9.5% to 6.9% during the dehydrogenation process. Over the P-10 sample, the initial and final methane selectivity was 7.1% and 5.1%. This indicated that the side reaction of propane cracking was suppressed by the phosphorous introduction. Combined with the data of propane conversion and product selectivity, it could be concluded that both the dehydrogenation performance and stability were improved after the phosphorus introduction.

The catalytic performance of supported vanadium catalysts was closely related to the vanadyl species on the surface[49,50]. After the phosphorous introduction, the vanadium loading, surface vanadyl sites amount and surface vanadyl sites density all slightly increased and reached the maximum over the P-30 sample. However,a further increase of phosphorus content was detrimental for the highly dispersed vanadyl species although the vanadium loading amount still slightly increased. The propane conversion also increased due to the increase of surface vanadium sites. The best propane dehydrogenation performance was also obtained over the P-30 sample. This indicated that there existed a suitable range for the phosphorus introduction. A suitable amount of phosphorus introduction could prohibit the formation of polymerized vanadyl species by forming V?O?P or Si?P?O bond while excessive amount of phosphorus on the surface did not help further increase the dispersion of vanadyl species due to the decrease of specific surface area. The selectivity of propylene and methane were closely related to the structure and chemical environments of vanadyl species. The isolated or highly-dispersed vanadyl species was more favorable to form propylene while the aggregated vanadyl species were less selective. After phosphorus introduction, the reducibility of the vanadyl species was improved and more vanadyl species could easily complete the catalytic cycle of dehydrogenation,resulting in the enhanced stability and selectivity. The similarity of products selectivities over P-30 and P-10 indicated that the surface vanadyl species shared similar chemical environments. The decrease of the vanadyl sites on P-10 should be strongly correlated to the mild aggregation of vanadyl species caused by the decreased specific surface areas.

3 Conclusions

Phosphorus-modified V-MCM-41 catalysts were synthesized and tested for propane dehydrogenation reaction. Although the specific surface area slightly decreased after the introduction of phosphorus, the ordered mesoporous structure remained stable. Both reducibility and dispersion of the surface vanadyl species were improved. The vanadyl species were highly dispersed on the surface, and the content of polymerized vanadyl species decreased due to the presence of phosphorus species. The surface vanadium sites reached a maximum of 3.35×10?4mol/g over the P-30 sample. The propane dehydrogenation reaction results showed that both of the catalytic performance and stability of the catalyst were obviously increased after the phosphorus introduction.

Acknowledgments

The authors thank the Beijing Key Laboratory of Biogas Upgrading Utilization for the support of characterization.