Catalysis Science & Technology

2 downloads 0 Views 1MB Size Report
Feb 6, 2014 - DOI: 10.1039/c3cy01083j www.rsc.org/catalysis. Effect of support on selectivity and on-stream stability of surface VOx species in non-oxidative.

Catalysis Science & Technology View Article Online

PAPER

Cite this: Catal. Sci. Technol., 2014, 4, 1323

View Journal | View Issue

Effect of support on selectivity and on-stream stability of surface VOx species in non-oxidative propane dehydrogenation†

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke and E. V. Kondratenko* Al2O3, SiO2(MCM-41), and Al2O3–SiO2 (Siral®) with a SiO2 content varying from 1 to 70 wt.% were used to prepare supported catalysts with a V loading below one monolayer. Their activity, selectivity and on-stream stability were tested in non-oxidative propane dehydrogenation (DH) at 550 °C. The highest space–time yield of propene was only 25% lower than that over industrially relevant Pt–Sn/Al2O3 under the same reaction conditions. All catalysts deactivated with time on stream, but restored their initial performance after oxidative regeneration as proven in a sequence of 10 DH/regeneration cycles lasting in Received 20th December 2013, Accepted 5th February 2014 DOI: 10.1039/c3cy01083j

total over 60 h. The deactivation was related to propene-derived carbon deposits covering active VOx sites. However, depending on the catalyst, such deposits formed on bare support sites can also participate in propane dehydrogenation. Their DH activity is, however, significantly lower compared to VOx species. Acidic properties of the support were found to be crucial for the generation of such catalytically active

www.rsc.org/catalysis

carbon species.

Introduction Propene is the second most important petrochemical after ethylene with an annual production of around 79 million metric tons in 2011.1 Its demand has been steadily growing for the past twenty years thanks to rising production of various polymers. Today, propene is mostly produced as a by-product of the steam cracking of naphtha and liquid petroleum gas or fluid catalytic cracking (FCC) of heavy fractions of crude oils. Since the existing production capacities cannot completely fulfil the propene demand, two alternative processes were commercialized: non-oxidative propane dehydrogenation (DH)2 and metathesis of ethylene and 2-butene.3 The former becomes increasingly important given the availability of low cost propane from shale gas and co-production of hydrogen. Moreover, the dehydrogenation route is independent of the state of steam crackers or FCC plants, unlike the metathesis process.

Leibniz-Institut für Katalyse e. V. an der Universität Rostock, Albert-Einstein-Str. 29 A D-18059 Rostock, Germany. E-mail: [email protected]; Fax: +49 381 1281 51290; Tel: +49 381 1281 290 † Electronic supplementary information (ESI) available: Further results of catalytic tests. See DOI: 10.1039/c3cy01083j

This journal is © The Royal Society of Chemistry 2014

There are five commercial DH technologies differing in the reactor type, mode of operation and catalysts.2 As far as the catalysts are concerned, CATOFIN (licensed to Lummus–Houdry) and FBD-3 (Snamprogetti–Yarsintez) utilize the chromia– alumina catalyst, while alumina-supported Pt catalysts are used in Oleflex (UOP), PDH (Linde–BASF–Statoil), and STAR (Krupp Uhde) processes. Irrespective of the process, DH activity of the applied catalysts drops with time on stream due to the blockage of active sites by the formed coke.4 As a consequence, catalysts must be oxidatively regenerated to restore their activity. However, this treatment results in a gradual loss of initial activity and reduction of the on-stream stability in successive DH cycles owing to the restructuring of the catalytically active Pt or CrOx species.5 Recently, we have demonstrated that the VOx/MCM-41 catalyst with highly dispersed, near to isolated VOx species showed similar selectivity and, most importantly, stable catalytic performance in several DH/regeneration cycles surpassing in quality the industrially relevant CrOx/MCM-41 and Pt–Sn/Al2O3.6 This was ascribed to the stability of VOx species against agglomeration in the course of DH and regeneration cycles. The aim of the current work was to further explore the potential of VOx-based catalysts for the DH reaction. To this end, we selected commercial Al2O3–SiO2 materials (Siral®), pure Al2O3 and SiO2 as supports to prepare catalysts free

Catal. Sci. Technol., 2014, 4, 1323–1332 | 1323

View Article Online

Paper

Catalysis Science & Technology

of crystalline V2O5 particles. In particular, we focused on (i) elucidating factors governing catalyst on-stream stability, (ii) analysing the role of pristine supports and carbon deposits in propene production and (iii) optimizing reactor operation to achieve industrially relevant propene production. The catalysts' long-term stability was verified in ten optimized DH/regeneration cycles with a total duration of ca. 60 h.

Experimental

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Preparation of catalytic materials Aluminosilicates (Siral® 1, 10, 40 and 70) provided by Sasol, pure Al2O3 (Saint-Gobain NorPro), and SiO2 (MCM-41) were used as supports. Siral index stands for the weight percentage of SiO2 in the material. MCM-41 was synthesized following the procedure reported in ref. 7. Supported catalysts were prepared according to the procedure described in ref. 8. Briefly, 3.33 g of vanadyl acetylacetonate (VO(acac)2, Sigma-Aldrich, 99%) were stirred in 1 L dried toluene (Roth, >99.5%, less than 50 ppm H2O) for 3 h and the undissolved substance was separated by filtration. 10 g of the support were added to the filtered dark-blue solution and the mixture was stirred at room temperature for 24 h during which the colour intensity visibly faded. Hereafter, the impregnated support was isolated, washed twice in 400–500 mL of pure toluene to remove the physically adsorbed metal precursor, dried at 80 °C for 20 h and calcined in a muffle furnace at 550 °C for 12 h. To increase the metal loading, the procedure was repeated on each calcined material. The Siral®-based catalysts will be further abbreviated as VOx/S1, VOx/S10, VOx/S40 and VOx/S70; VOx supported on MCM-41 and Al2O3 are called VOx/MCM-41 and VOx/Al2O3, respectively. Prior to the catalytic tests, the catalysts were pelleted, crushed and sieved into 315–710 μm fraction. Catalyst characterisation The specific surface areas of the calcined catalysts (Table 1) were calculated from N2 adsorption–desorption isotherms collected at 77 K on a BELSORP-mini II (BEL Japan). The

Brunauer, Emmett and Teller (BET) equation was applied for a N2 relative pressure range of 0.05 < P/P0 < 0.30. The metal content of the calcined catalysts (Table 1) was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES, Varian 715-ES). The surface acidity of the pristine supports was evaluated via infrared (IR) measurements of adsorbed pyridine using a Tensor 27 spectrometer (Bruker). The instrument was equipped with an in-house developed cell with CaF2 windows connected to a gas dosing system. The powders were pressed into self-supporting wafers (50 mg, 20 mm in diameter). Before pyridine adsorption, the samples were pretreated at 309 °C for 10 min in synthetic air followed by cooling to room temperature and evacuation. Pyridine was adsorbed at 25 °C until saturation was reached. Then, the reaction cell was evacuated to remove physisorbed pyridine. After heating the sample in vacuum to 150 °C, the IR spectra of the adsorbed pyridine were recorded.

Continuous flow catalytic tests DH tests were carried out in a multi-channel set-up consisting of 15 plug flow fixed-bed quartz tube reactors of 3.8 mm inner diameter. All catalytic experiments were carried out at a gas hourly space velocity (GHSV) of 1200 L h−1 kg−1 and an overall pressure of 1.2 bar. Each reactor was filled with 300 mg of fresh catalyst. The catalysts were initially heated in parallel at 5 K min−1 to 550 °C in air (10 mL min−1 per reactor), held in air flow for 1 h and then flushed with N2 for 20 min. Hereafter, a mixture of 40 vol.% C3H8 in N2 was fed at a rate of 6 mL min−1 per reactor. The experiment was run for either 6 or 27 h followed by cooling the reactors to room temperature in N2. For the experiment with reductive catalyst pre-treatment, after heating the catalysts in air up to 550 °C they were flushed with N2 for 20 min and then treated in a mixture of 30 vol.% H2 in N2 passed at 6 mL min−1 for 1 h. Before C3H8 was admitted, the reduced catalysts were flushed again for 20 min with N2 at 550 °C. To derive insights into the catalysts' stability, we performed 10 DH/regeneration cycles under the above

Table 1 Vanadium content, specific surface areas of fresh catalysts, their bulk densities (ρ) and surface-related acidity, the amounts of carbon removed and the temperature maxima of CO2 evolution during TPO of the spent catalysts and pristine supports

Carbon/gC gcat−1 2

Catalysts

V/wt.%

SBET/m g

VOx/Al2O3 VOx/S1 VOx/S10 VOx/S40 VOx/S70 VOx/MCM-41 Al2O3 Siral 1 Siral 10 Siral 40 Siral 70 MCM-41

2.7 4.3 4.2 5.1 4.6 4.9 — — — — — —

74 235 280 353 266 810 89 255 350 475 360 901

−1

1324 | Catal. Sci. Technol., 2014, 4, 1323–1332

−3

ρ/g cm 0.64 0.59 0.52 0.35 0.31 0.30

Acidity/a.u.

2.4 × 2.3 × 1.7 × 1.2 × 5.3 × 0

10−2 10−2 10−2 10−2 10−3

6h 0.07 0.13 0.14 0.19 0.07 0.05 — — — — — —

(89%) (52%) (47%) (86%) (44%) (45%)

Tmax(CO2)/°C 27 h

6h

27 h

0.08 0.26 0.30 0.22 0.15 0.11 3.1 × 7.0 × 4.0 × 1.6 × 4.5 × 0

425 445, 460 470 445, 460, 475 450 380 — — — — — —

440 460 485 490 470 405 505 520 522 521 ca. 530 —

10−3 10−3 10−3 10−3 10−4

This journal is © The Royal Society of Chemistry 2014

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Catalysis Science & Technology

Paper

conditions in the same 15-channel reactor. The 6 h DH stages were followed by 2 h oxidative regeneration also performed at 550 °C in a flow of O2 (20 vol.%) in N2. DH and regeneration stages were always interlaid by flushing with N2 (10 mL min−1 per reactor) lasting for 20 min. After the last DH cycle, the catalysts were cooled to room temperature in N2. The feed components and the reaction products were analysed using an on-line gas chromatograph (Agilent 6890) equipped with PLOT/Q (for CO2), HP-PLOT Al2O3 “KCl” (for hydrocarbons) and Molsieve 5 (for H2, O2, N2, and CO) columns and flame ionization and thermal conductivity detectors. The GC analyses were carried out successively, i.e. reactor by reactor. The propane conversion was calculated from the inlet and outlet propane molar flows (eqn (1)). The propene yield and selectivity to propene and to carbon were calculated according to eqn (2), (3) and (4), respectively: X (C3 H 8 ) 

nCinlet  nCoutlet 3 H8 3 H8 nCinlet 3 H8

Y (C3 H 6 ) 

S (C3 H 6 ) 

Scarbon 

nCoutlet 3 H6

(2)

nCinlet 3 H8 nCoutlet 3 H6

n

inlet C3 H 8

(1)

 nCoutlet 3 H8

X (C3H8 )   Yi i

X (C3 H 8 )

(3)

(4)

fibre were threaded through the furnace wall to face quartz reactors.9 All spectra are presented as the Kubelka–Munk function F(R) calculated according to eqn (5): F ( R) 

(1  R ) 2 2 R

(5)

where R is the reflectance. Quantification of carbon deposits The amount of carbon deposits formed during a DH cycle was determined from the amounts of CO and CO2 evolved during temperature-programmed oxidation (TPO) of the catalysts used in the DH reaction run at 550 °C for 6 or 27 h. The TPO measurements were performed in a setup equipped with eight individually heated plug-flow fixed-bed quartz tube reactors of 6.9 mm inner diameter. The used catalysts were heated to 900 °C at 10 K min−1 in a flow (10 mL min−1) of 10 vol.% O2 in Ar. The measurements were performed successively, i.e. during heating (TPO) of the analysed sample, all other samples were kept at 25 °C under Ar. Oxygen consumption and formation of the reaction products were monitored using a quadrupole mass spectrometer (Pfeiffer Vacuum OmniStar 200). The following atomic mass units (AMUs) were obtained: 44 (CO2), 40 (Ar), 32 (O2), 28 (CO, CO2), and 18 (H2O). The concentrations of O2, CO and CO2 were determined from the respective AMUs using standard fragmentation patterns and sensitivity factors determined by analysing calibration gas mixtures.

Results Characterisation of catalytic materials

where ni with superscripts “inlet” and “outlet” stands for the molar flow of gas-phase components at the reactor inlet and outlet respectively and Yi is the yield of the gaseous product i. In situ UV/Vis studies The experiment designed to monitor the state of VOx species and carbon accumulation under DH conditions was carried out in an in-house developed catalytic set-up enabling in situ UV/Vis catalyst characterization with simultaneous analysis of gas-phase components using an on-line gas-chromatograph (Agilent 7890).9 150 mg of fresh VOx/MCM-41, VOx/S10 and VOx/S40 were loaded and heated at 5 K min−1 to 550 °C in 20 vol.% O2 in Ne flow (7 mL min−1 per reactor), held at 550 °C for 1 h and flushed with Ne for 20 min. Hereafter, a mixture of 40 vol.% C3H8 in Ne was fed at a flow rate of 3 mL min−1 per reactor (GHSV 1200 L h−1 kg−1) for 5.5 h followed by cooling the catalyst to room temperature in Ne flow. UV/Vis spectra were recorded every 5 min. The spectra were collected using an AVASPEC UV/Vis spectrometer (Avantes) equipped with a DH-2000 deuterium-halogen light source and a CCD array detector. BaSO4 was used as a white reference material. High-temperature reflection UV/Vis probes consisting of 6 radiating optical fibres and 1 reading

This journal is © The Royal Society of Chemistry 2014

Catalytic materials used in this study and their selected physicochemical characteristics are presented in Table 1. The specific surface areas (SBET) of the calcined catalysts were lower than those of the corresponding pristine supports with the highest loss of 26% observed on VOx/S40 and VOx/S70. The distribution of VOx species on the surface of the freshly calcined catalysts was studied by UV/Vis spectroscopy. The UV/Vis spectra of the catalysts at 550 °C in a flow of O2 (20 vol.%) in N2 are shown in Fig. 1. The spectrum of VOx/MCM-41 is dominated by the charge transfer (CT) transition from O2− to V5+ with a broad absorption band at ca. 300 nm characteristic of highly-dispersed dehydrated VO4 surface species.8,10 As the surface density of V increases, the red shift of the CT onset occurs for VOx/Al2O3 and VOx/Siral (Fig. 1) due to the appearance of polymerized VO4 entities linked via oxygen bridges.10 Catalytic performance In order to gain insights into the catalysts' activity, selectivity and on-stream stability, we initially performed a 27 h DH test at 550 °C. Fig. 2 shows propane conversion and selectivity to propene and to carbon as a function of time on stream. Other gas-phase side products were methane, ethylene and ethane. As evident from the individual plots, VOx/S1, VOx/S10

Catal. Sci. Technol., 2014, 4, 1323–1332 | 1325

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Paper

Fig. 1 UV/Vis spectra of fresh catalysts collected after 1 h in air at 550 °C.

and VOx/S40 had similar initial propane conversion around 30%. VOx/Al2O3 and VOx/MCM-41 were only slightly less active, while VOx/S70 showed the lowest conversion of about 18%. The latter catalyst was the second most stable in terms of sustaining X(C3H8), while VOx/Al2O3 and VOx/MCM-41 demonstrated the worst and the best stability, respectively. Among other catalysts, VOx/S10 showed higher stability in the first 3–4 h on stream and remained slightly more active through the rest of the experiment. With time on stream, an increase in selectivity to propene concomitant with a decrease in propane conversion took place over VOx/S1, VOx/S10, VOx/S40, and, to a lower extent,

Catalysis Science & Technology

over VOx/Al2O3. Such selectivity trends arise from the fact that the side products, including carbon, are formed from propene, the concentration of which drops with time on stream. Consecutively, carbon selectivity over these catalysts decreased to zero after 12–20 h on stream. On the other hand, the on-stream selectivity to propene practically did not change over VOx/MCM-41 and VOx/S70, remaining through the entire run in the 88–90% and 93–94% range, respectively. The corresponding carbon selectivity was in the range of 1–3% and 2–5% being the lowest among all catalysts when compared at similar degrees of propane conversion. Further insight into coke formation and its role in the DH reaction is given in the sections “Analysis of formation and removal of carbon deposits” and “Discussion”. In the next step, long-term stability of the catalysts was studied in a series of 10 successive cycles comprising a DH stage followed by an oxidative regeneration stage. The duration of each DH cycle was fixed to 6 h because during this period of time no dramatic losses in propane conversion and propene yield were observed over all catalysts with the exception of VOx/Al2O3 (Fig. 2 and S1 in the ESI†). Fig. 3 and S2† show time on stream propene yield and propane conversion obtained in the first three and last two DH cycles (others are omitted for the purpose of clarity). By integrating the yield profiles, we calculated the amount of propene formed over each catalyst in different cycles. The results are summarized in Fig. 4. Among the VOx-based catalysts, VOx/S70 exhibited the strongest loss of 25% in propene production over 10 DH cycles, while VOx/S10 had identical catalytic performance from cycle to cycle. All other catalysts lost some of their activity in the first 7–8 cycles and remained at the same level in the last 2–3 cycles. VOx/S10 produced the highest amount of propene

Fig. 2 Conversion of propane (○) and selectivity to propene (△) and to carbon (▲) with time on stream in a 27 h propane DH test: (a) VOx/S1, (b) VOx/S10, (c) VOx/S40, (d) VOx/S70, (e) VOx/MCM-41, (f) VOx/Al2O3. Reaction conditions: 550 °C, C3H8/N2 = 40/60, GHSV = 1200 L h−1 kg−1.

1326 | Catal. Sci. Technol., 2014, 4, 1323–1332

This journal is © The Royal Society of Chemistry 2014

View Article Online

Catalysis Science & Technology

Paper

indicate that i) the activity of supported VOx species increased with time on stream or ii) some other active species were additionally formed under DH conditions. The difference in catalytic behaviour reflected in the catalysts' activity and on-stream stability as well as the origin of the maxima in their Y(C3H6) and X(C3H8) profiles are thoroughly treated in the “Discussion” section together with the catalytic and characterisation results.

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Analysis of formation and removal of carbon deposits

Fig. 3 Yield of propene in the first three and the last two DH/ regeneration cycles from a series of 10 cycles on: VOx/MCM-41 (○), VOx/S1 (■), VOx/S10 (□), VOx/S40 (▲), VOx/S70 (▽) and VOx/Al2O3 (●). Reaction conditions are given in the caption of Fig. 2.

Fig. 4 Amount of propene produced in each 6 h DH stage on VOx/ MCM-41 (○),VOx/S1 (■),VOx/S10 (□), VOx/S40 (▲), VOx/S70 (▽), VOx/ Al2O3 (●) and Pt–Sn/Al2O3 (♦). The data for Pt–Sn/Al2O3 are from ref. 6. Reaction conditions are given in the caption of Fig. 2.

in each DH cycle among the catalysts prepared in this study, but less than that of reduced Pt–Sn/Al2O3 tested in our previous work.6 However, the Pt-based catalyst lost its X(C3H8) from cycle to cycle whereas VOx/S10 showed stable performance. Another important observation derived from Fig. 3 is the presence of a maximum in the time on stream profiles of propene yield obtained over all catalysts based on mixed Al2O3–SiO2 supports. These maxima could not be detected in the 27 h test since the time intervals between the GC measurements were longer than in the 6 h stages (2 h 15 min vs. 50 min). Measurements at yet higher periodicity (27 min) did not reveal such a maximum for VOx/MCM-41 (Fig. S3†). Such maxima are present in all cycles thus indicating a reversibility of the process(es) responsible for their appearance. Paralleling the trend observed for the yield curves, their propane conversion curves also contain maxima (Fig. S2†). Such maxima

This journal is © The Royal Society of Chemistry 2014

To quantify carbon deposits formed during the DH reaction, we performed TPO of spent catalysts after the 27 h experiment and after the 10th cycle of the 6 h test. The amounts of carbon formed in the corresponding DH tests were determined by integrating the CO and CO2 profiles recorded in the TPO tests and are given in Table 1. Relative to the total amount of carbon deposited on the catalyst during the 27 h test, VOx/S70 accumulated 44% (lowest) and VOx/Al2O3 89% (highest) during the 6 h run. Taking these data into account, it can be concluded that the rate of carbon deposition decreases with increasing time on stream, which is in accordance with the S(C) shown for all catalysts in Fig. 2. Irrespective of the DH test duration, carbon deposition was intense on the catalysts showing high propene yield (VOx/S1, VOx/S10, VOx/S40) thus supporting the hypothesis that propene serves as a precursor of deposited carbonaceous species. For all catalysts, the temperatures of maximal CO2 evolution (Tmax) in the TPO tests after the 6 h run are lower than those after the 27 h run (Table 1). This suggests that the degree of carbon condensation increases with time on stream. The CO2 profiles of VOx/MCM-41 after the 6 and 27 h tests had the lowest Tmax at 380 and 405 °C, respectively. The second lowest corresponding Tmax of 425 and 440 °C were determined on VOx/Al2O3. For the Siral-based catalysts, the Tmax values were higher and reached the maximum on VOx/S40 (Table 1). In addition, two and three distinguishable Tmax were observed in the CO2 profiles of VOx/S1 and VOx/S40, respectively, after the 6 h test. Thus, our TPO tests suggest that the support and the duration of the DH run influence the degree of condensation of carbon deposits. Further insight into the nature of carbon species and the rate of their formation was derived from the in situ UV/Vis characterization of VOx/MCM-41, VOx/S40 and VOx/S10. Upper panels in Fig. 5 display the UV/Vis spectra of these catalysts in their fresh (fully oxidized) state and after 0.5 and 5 h on stream. After 0.5 h on stream, the spectra of each catalyst conformed to a unique shape that persisted through the rest of the experiment. For VOx/MCM-41, the Kubelka–Munk function (F(R)) exhibited monotonous growth with an increasing wavelength, while in the case of VOx/S10 it showed a maximum at ca. 425 nm and then descended. The spectra of VOx/S40 appeared to be a combination of the above two. Their complex nature is in agreement with the multiple Tmax featured by the TPO profile of the same catalyst after 6 h on stream.

Catal. Sci. Technol., 2014, 4, 1323–1332 | 1327

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Paper

Catalysis Science & Technology

Fig. 5 The UV/Vis spectra collected on VOx/S10, VOx/S40 and VOx/MCM-41 in the fresh state, after 0.5 and 5 h of propane DH reaction. Lower panels: temporal evolution of the Kubelka–Munk function at 700 nm in the 1st (■), 2nd (○), 3rd (△) and 4th (▼) DH cycles in the endurance test. Reaction conditions are given in the caption of Fig. 2.

The lower panels in Fig. 5 show temporal evolution of F(R) at 700 nm recorded in four consecutive 6 h DH cycles. This wavelength was selected since it falls into the 570–800 nm range characteristic for polyaromatic graphitic species.11 For VOx/MCM-41 and VOx/S10, F(R) continuously increases with time on stream, while for VOx/S40 it reaches a plateau after 2 h. Flattening of the signal on the latter catalyst is related to a decrease in the rate of carbon deposition. This statement is in good agreement with the results of TPO tests; the amount of carbon deposited over VOx/S40 after 6 h on DH stream corresponds to 86% of total carbon formed after 27 h, while the corresponding values for VOx/MCM-41 and VOx/S10 are 45 and 47%, respectively. Comparing the slopes of the F(R) growth and the time to attain a plateau, we put forward that the rate of carbon deposition follows this order: VOx/MCM-41 < VOx/S10 < VOx/S40.

Discussion In the present study, we have demonstrated the importance of SiO2 content in SiO2–Al2O3 supports for activity, selectivity and on-stream stability of supported VOx species in nonoxidative propane DH (Fig. 2 and 3). It is worth mentioning that dispersed VOx species on low-silica aluminosilicates afford the industrially attractive space–time yield (STY) of propene,

1328 | Catal. Sci. Technol., 2014, 4, 1323–1332

which is a crucial process characteristic. Fig. 6 shows the maximal and the average STY values of all catalysts prepared in the current study and of reduced Pt–Sn/Al2O3 used in our previous work as the industrially relevant catalyst.6 The highest STY was measured on VOx/S1 and VOx/S10 and comprised, respectively, 75 and 65% of 212 kgC3H6 h−1 mcatalyst−3 achieved on Pt–Sn/Al2O3. The reason behind the lower STY obtained on VOx/S10 compared to that on VOx/S1 is merely the lower catalyst powder density of the former stemming from a higher SiO2 content. Besides the effect on propene productivity, the supports with different Al2O3/SiO2 ratios also exerted a strong influence on the on-stream behaviour of the supported catalysts. The following discussion is aimed to elucidate the fundamental origins of this influence. As shown in Fig. 3 and S2,† Y(C3H6) and X(C3H8) profiles of all catalysts based on Al2O3–SiO2 supports are characterized by maxima appearing after the first 2–3 h on-stream, with those of VOx/S1 and VOx/S10 being the most pronounced. Such temporal behaviour indicates that the catalysts initially improved their activity and then continuously lost it. The origins of this activating phenomenon may be: (i) the higher catalytic activity of the reduced VOx species that were formed at the beginning of the DH stage or (ii) participation of carbon species formed on bare supports in DH. In situ reduction of V5+ under DH conditions

This journal is © The Royal Society of Chemistry 2014

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Catalysis Science & Technology

Fig. 6 Maximum (white bars) and average (black bars) STY of propene obtained within 6 h on stream. Reaction conditions are given in the caption of Fig. 2.

has been previously reported in the literature6,12 and was also verified in this work by detecting COx in the products at the beginning of the DH reaction over initially oxidized catalysts, while their pre-reduced counterparts did not form carbon oxides (Fig. S4†) However, taking into account much greater amounts of hydrogen produced alongside with COx (Fig. S4†), the contribution of oxidative propane dehydrogenation by a lattice oxygen of VOx to propene formation is deemed low compared to that of the non-oxidative DH route. Nevertheless, the hypothesis which holds that reduced VOx species are more active than oxidized ones can be excluded since pre-reduced VOx/S1 and VOx/S10 showed lower initial propane conversion than their oxidized counterparts (Fig. S5†) with the difference vanishing with time on stream. In addition, propane conversion over both reduced and oxidized catalysts passed over a maximum in the first 3 h on stream. It is worth reiterating that carbon deposits are formed over all catalysts from the beginning of a DH run as evidenced by non-zero coke selectivity in Fig. 2 and by on-stream changes in the UV/Vis spectra of the catalysts in Fig. 5. However, some tested catalysts lost their activity continuously, while the others activated within the first hours on stream. This variance leads us to propose that if carbon deposits contributed to propane dehydrogenation at all, their contribution would depend on the catalysts where they are formed. The role of carbon deposits is explored in Fig. 7, where propane conversion obtained directly after starting the reaction (no coke on the catalyst surface) as well as after 6 and 27 h on stream, is plotted versus the amount of carbon formed after these times. For VOx/Al2O3, VOx/S70 and VOx/MCM-41, the plots reveal a clear decrease in the conversion with rising amount of deposits, i.e. formation of carbon is the reason for activity loss. The other catalysts based on Al2O3–SiO2 supports show a complex dependence of propane

This journal is © The Royal Society of Chemistry 2014

Paper

conversion on the amount of deposited carbon. For instance, the conversion of propane over VOx/S1 and VOx/S10 after 6 h on stream was very close to the initial one, despite the fact that they deposited a higher amount of carbon than VOx/Al2O3 and VOx/MCM-41 even after 27 h. The highest amount of carbon after 6 h on stream was formed over VOx/S40, which, however, exhibited more stable performance than the two latter catalysts but deactivated faster than VOx/S1 and VOx/S10. On the other hand, VOx/S1, VOx/S10 and VOx/S40, being the most stable in the 6 h test, showed significantly lower propane conversion after 27 h on stream than VOx/MCM-41. They also deposited the highest amount of carbon. Based on the above considerations, it can be concluded that deactivation of all catalysts tested in the current work should be related to coke deposition, which is in agreement with several previous studies on propane dehydrogenation over supported catalysts having Pt,13,14 CrOx11,15–17 or VOx6 as the catalytically active phase. However, coke can also play a positive role as suggested by Weckhuysen and co-workers, who attributed the initial increase in propane conversion over

Fig. 7 Propane conversion after different times on stream versus the amount of carbon deposited per gram of the catalyst: (a) VOx/Al2O3, VOx/S70 and VOx/MCM-41 and (b) VOx/S1, VOx/S10 and VOx/S40. Open, grey and black symbols represent propane conversion: initial, after 6, and 27 h on stream respectively.

Catal. Sci. Technol., 2014, 4, 1323–1332 | 1329

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Paper

CrOx/Al2O3 to participation of carbon species in dehydrogenation.11,18,19 The authors assumed that such species deposited in the vicinity of CrOx sites favour propane adsorption, while dehydrogenation of adsorbed propane takes place over CrOx species. Yet, if this hypothesis was of a general character and equally applicable to VOx catalysts, what would be the reason for the catalyst effect on the time on-stream profiles of propene yield (Fig. 3) and propane conversion (Fig. S2†) observed in our study? Searching for an answer, we performed propane DH over the pristine supports under the same conditions as for their VOx-containing counterparts. The initial propene yield on all supports was below 1% (Fig. 8) and very close to the value observed in a reactor filled with catalytically inert silicon carbide (dashed line in the same figure). First, it should be stressed that such a low yield should be viewed as further evidence of the importance of VOx species for achieving high propene productivity. Siral 70 and MCM-41 retained low activity for the entire 27 h test, while propene yield over Al2O3, Siral 1, Siral 10 and Siral 40 increased with time on stream approaching 5% over Siral 1 by the end of the experiment. Such an increase suggests that DH active species were formed in situ under reaction conditions. To check if carbon was also formed during the DH reaction over the pristine supports, we examined all used supports by TPO. Trace amounts of carbon were deposited on MCM-41 and Siral 70. Other supports ordered according to the amount of carbon formed as Siral 1 > Siral 10 ~ Al2O3 > Siral 40 (Table 1). Compared to the supports, the respective VOx-containing catalysts deposited significantly greater amount of carbon species. This is due to the higher propene partial pressure caused by the significantly higher DH activity of VOx species, and propene is known to be a precursor of coke.5,14,18,20,21 In addition, the catalysts possess acidic VOx species, which adsorb propene, too. In the case of VOx/MCM-41,

Fig. 8 Yield of propene with time on stream over pristine MCM-41 (○), Siral 1 (■), Siral 10 (□), Siral 40 (▲), Siral 70 (▽) and Al2O3 (●) and in a reactor filled with silicon carbide particles (dashed line) in the 27 h DH test. Reaction conditions are given in the caption of Fig. 2.

1330 | Catal. Sci. Technol., 2014, 4, 1323–1332

Catalysis Science & Technology

propene adsorption takes place exclusively on such sites because pristine MCM-41 is inert for carbon formation (Table 1). Importantly, the supports that deposited carbon demonstrated an increase in propane conversion with time on stream, and the higher was the amount of carbon deposits, the higher propene yield was achieved (Fig. 9). Moreover, Al2O3 reached stable activity with a propene yield of 2.5% after 15 h on stream. This value is actually very close to that measured over VOx/Al2O3 after 27 h on stream (Fig. 2). For the VOx/Al2O3 catalyst, it can be hypothesised that after 27 h on stream all VOx sites are blocked by carbon deposits and only the coke sites are still active in dehydrogenation thus giving a similar propane conversion as over the coked pristine Al2O3. The plateau was reached on Al2O3 earlier than on other supports because it had the lowest surface area. Thus, compared to propane dehydrogenation over CrOx/Al2O3,11 we suggest that carbon deposits participate in propane dehydrogenation over V-based catalysts. This hypothesis is supported by the fact that propene yield increases with the amount of carbon species formed over the pristine supports (Fig. 9). Based on the fact that the pristine supports develop DH activity with time on stream (Fig. 8), we explain the increase in propene yield and propane conversion within the first 3 h on stream over VOx/Siral (Fig. 3 and S2†) by the formation of carbon species from propene, which actively participate in propane DH. This statement is indirectly supported by the fact that the best performing VOx/S1 and VOx/S10 catalysts showed the lowest initial propene production at the highest carbon selectivity (Fig. 2). Owing to the low surface density of VOx species, it is suggested that carbon deposits are initially formed on support sites. Thus, the overall number of catalytically active sites increases resulting in an increase in propene production. As the reaction progresses, more intrinsically active VOx species become blocked by carbon deposits resulting in a drop in the overall catalytic activity. The presence and duration of the activation phase should depend on the acidic properties

Fig. 9 Yield of propene versus the amount of carbon formed over the pristine supports in the 27 h DH test at 550 °C.

This journal is © The Royal Society of Chemistry 2014

View Article Online

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

Catalysis Science & Technology

Fig. 10 Amount of carbon deposited on the surface of the pristine supports versus their acidity measured by IR pyridine adsorption.

Paper

relevant Pt–Sn/Al2O3 and are approximately two times higher than that delivered by the VOx/MCM-41 catalyst with a similar vanadium loading. With the exception of the catalyst containing 70 wt.% of SiO2 in the support, all other catalysts showed excellent ability to restore their initial activity as proven in a series of 10 DH/oxidative regeneration cycles. The optimal duration of the DH cycle without a significant loss of initial propene yield over Siral®-based catalysts was determined to be 6 h. This was explained by the fact that carbon deposits are initially formed on the support sites not carrying VOx and that such deposits are active in propane dehydrogenation. Since their activity is significantly lower than that of VOx species, the catalysts start to deactivate when the latter also become covered by coke. The tendency of the catalysts to form coke increases with their surface acidity.

Acknowledgements of the support and its available free surface area. The catalyst acidity favours propene adsorption22 followed by its oligomerisation and finally formation of carbonaceous deposits.5,14,18,20,21 To compare the pristine supports in terms of their acidity, we studied pyridine adsorption by IR spectroscopy. As shown in Table 1, Al2O3 possesses the highest surface total acidity (Lewis and Brønsted), while no acidic sites were identified on MCM-41. When the amount of carbon deposited per area of the pristine support is plotted versus the acidity values as shown in Fig. 10, the promoting effect of surface acidity on carbon formation becomes evident. Since non-acidic MCM-41 formed virtually no carbon on its surface, coke formation on VOx/MCM-41 must have started directly on the acidic VOx species thus blocking the DH active sites. As a consequence, we observe no maximum in catalytic activity but rather deactivation occurring from the beginning of the catalytic run. This assumption is supported by the fact that the carbon deposits formed on VOx/MCM-41 strongly differ from those formed on the activating catalysts when compared in terms of their Tmax of CO2 evolution (Table 1). In summary, in order to profit from the dehydrogenating activity of in situ formed carbon species, catalytic materials must possess the largest possible surface area and an optimal ratio of the support acidic sites to VOx species. The coverage with such species must be definitely lower than a monolayer. Moreover, the duration of a DH cycle must be also optimized.

Conclusions The content of SiO2 (0 to 100 wt.%) in the studied Al2O3–SiO2 supports was demonstrated to have a strong effect on the activity and particularly the on-stream stability of the supported VOx species in propane DH. The highest space–time yields of 146 and 158 kgC3H6 h−1 mcatalyst−3 at a propane conversion of 30% were achieved over the catalysts based on the Siral® supports with 10 and 1 wt.% of SiO2, respectively. These values approach the 212 kgC3H6 h−1 mcatalyst−3 measured on the industrially

This journal is © The Royal Society of Chemistry 2014

We would like to thank Ursula Bentrup for in situ IR pyridine adsorption experiments and express our gratitude to SaintGobain NorPro and Sasol for supplying alumina and Siral® carriers, respectively.

Notes and references 1 B. Hyde, Light Olefins Market Review, Information Handling Services, 2012. 2 K. J. Caspary, H. Gehrke, M. Heinritz-Adrian and M. Schwefer, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, Wiley-VCH Verlag GmbH & Co. KGaA, 2008, vol. 7, pp. 3206–3229. 3 J. C. Mol and P. W. N. M. van Leeuwen, in Handbook of Heterogeneous Catalysis, ed. G. Ertl, H. Knözinger, F. Schüth and J. Weitkamp, Wiley-VCH Verlag GmbH & Co. KGaA, 2008, vol. 7, pp. 3240–3256. 4 E. Daniel, E. Resasco and G. L. Haller, in Catalysis, ed. J. J. Spivey and S. K. Agarwal, Royal Society of Chemistry, Cambridge, 1994, vol. 11, pp. 379–411. 5 A. Iglesias-Juez, A. M. Beale, K. Maaijen, T. C. Weng, P. Glatzel and B. M. Weckhuysen, J. Catal., 2010, 276, 268–279. 6 S. Sokolov, M. Stoyanova, U. Rodemerck, D. Linke and E. V. Kondratenko, J. Catal., 2012, 293, 67–75. 7 G. Grubert, J. Rathousky, G. Schulz-Ekloff, M. Wark and A. Zukal, Microporous Mesoporous Mater., 1998, 22, 225. 8 E. V. Kondratenko, M. Cherian, M. Baerns, X. Su, R. Schlögl, X. Wang and I. E. Wachs, J. Catal., 2005, 234, 131–142. 9 O. Ovsitser, M. Cherian, A. Brückner and E. V. Kondratenko, J. Catal., 2009, 265, 8–18. 10 M. Baltes, K. Cassiers, P. Van Der Voort, B. M. Weckhuysen, R. A. Schoonheydt and E. F. Vansant, J. Catal., 2001, 197, 160–171. 11 T. A. Nijhuis, S. J. Tinnemans, T. Visser and B. M. Weckhuysen, Chem. Eng. Sci., 2004, 59, 5487–5489. 12 N. Ballarini, F. Cavani, M. Ferrari, R. Catani and U. Cornaro, J. Catal., 2003, 213, 95–102.

Catal. Sci. Technol., 2014, 4, 1323–1332 | 1331

View Article Online

Paper

18 S. J. Tinnemans, M. H. F. Kox, T. A. Nijhuis, T. Visser and B. M. Weckhuysen, Phys. Chem. Chem. Phys., 2005, 7, 211–216. 19 T. A. Nijhuis, S. J. Tinnemans, T. Visser and B. M. Weckhuysen, Phys. Chem. Chem. Phys., 2003, 5, 4361–4365. 20 D. Shee and A. Sayari, Appl. Catal., A, 2010, 389, 155–164. 21 J. V. Ibarra, C. Royo, A. Monzón and J. Santamaría, Vib. Spectrosc., 1995, 9, 191–196. 22 E. V. Kondratenko, N. Steinfeldt and M. Baerns, Phys. Chem. Chem. Phys., 2006, 8, 1624–1633.

Published on 06 February 2014. Downloaded on 28/04/2014 10:54:00.

13 P. Sun, G. Siddiqi, W. C. Vining, M. Chi and A. T. Bell, J. Catal., 2011, 281, 165–174. 14 M. M. Bhasin, J. H. McCain, B. V. Vora, T. Imai and P. R. Pujadó, Appl. Catal., A, 2001, 221, 397–419. 15 F. Cabrera, D. Ardissone and O. F. Gorriz, Catal. Today, 2008, 133–135, 800–804. 16 S. M. K. Airaksinen, M. A. Bañares and A. O. I. Krause, J. Catal., 2005, 230, 507–513. 17 J. Gascón, C. Téllez, J. Herguido and M. Menéndez, Appl. Catal., A, 2003, 248, 105–116.

Catalysis Science & Technology

1332 | Catal. Sci. Technol., 2014, 4, 1323–1332

This journal is © The Royal Society of Chemistry 2014