Catalysis Science & Technology

Transcription

1 Catalysis Science & Technology PAPER Cite this: DOI: /c6cy00078a Received 12th January 2016, Accepted 22nd February 2016 DOI: /c6cy00078a 1. Introduction Propylene is one of the important feedstocks in the chemical industry to produce different valuable products. Approximately two thirds of the propylene produced worldwide is consumed in the production of thermoplastic polypropylene, a commonly used compound in the fabrication of household appliances, plastic films and many other applications. The present worldwide propylene sales/demand reaches over ninety billion dollars. 1 There are three major commercial processes available for propylene production, including steam cracking, catalytic cracking (FCC) and catalytic dehydrogenation. 2 The steam cracking process consumes a large amount of energy, which accounts for 70% of the overall production cost. The coke formation during the cracking of heavy hydrocarbon molecules is another drawback of the catalytic cracking process. This causes severe process operational problems, especially fouling, which requires frequent plant shut-downs for cleaning. a Department of Chemical Engineering, Dhahran, Saudi Arabia. mhossain@kfupm.edu.sa; Fax: ; Tel: b Center of Research Excellence in Nanotechnology, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia c Research & Development Center, Saudi Aramco Oil Company, Dhahran, Saudi Arabia View Journal Oxidative dehydrogenation of propane to propylene over VO x /CaO γ-al 2 O 3 using lattice oxygen Afees A. Ayandiran, a Idris A. Bakare, b Housam Binous, a Sameer Al-Ghamdi, c Shaikh A. Razzak a and Mohammad M. Hossain* ab Oxidative dehydrogenation (ODH) of propane to propylene is studied using a new vanadium catalyst supported on CaO γ-al 2 O 3 under a gas phase oxygen free atmosphere. The catalysts are synthesized with different CaO/γ-Al 2 O 3 ratios, keeping the vanadium loading at 10 percent. The prepared catalysts are characterized using various physicochemical techniques. Raman spectroscopy reveals that the catalysts have monovanadate and polyvanadate surface species (VO x ) with minute crystal particles of V 2 O 5. FTIR spectroscopy and XRD analysis confirm the presence of V 2 O 5, CaO and γ-al 2 O 3 in the catalyst. The catalysts show stable reduction and re-oxidation behavior in repeated TPR and TPO cycles, respectively. NH 3 -TPD shows that catalyst acidity decreases with increasing CaO content. The NH 3 -TPD kinetics analysis reveals that the activation energy of desorption increases with higher CaO, indicating stronger active site support interactions. The ODH of propane experiments are conducted in a fluidized CREC Riser Simulator under gas phase oxygen free conditions. Among the studied catalysts, VO x /CaO γ-al 2 O 3 (1 : 1) displays the highest propane conversion (65%) and propylene selectivity (85%) and a low CO x due to its excellent oxygen carrying capacity, balanced acidity and moderate active site support interactions. In the FCC process, the coke generation is deliberate. The formed coke is combusted in a catalyst regenerator producing heat and is supplied back to the catalytic cracking unit using the catalyst as an energy carrier. This energy is essential for the FCC reactor to carry out the endothermic cracking reactions. In the FCC process the propylene is obtained as a by-product, in addition to the lighter gasoline and other fuels. The yield of propylene can be increased by manipulating the FCC operating conditions and using catalyst additives. Recent research shows that the FCC catalysts can also improve the propylene yield about 4.5% to 10%. However, the propylene production cost in the FCC process is still high due to the energy requirement by the endothermic cracking reactions. This high energy demand and the continuous catalyst regeneration make the FCC process capital intensive. Consequently, the building of a FCC system for the sole aim of producing propylene is not economical. The third available technology, catalytic dehydrogenation, also suffers from the problem of coke formation and high energy requirement as a result of the endothermic nature of the reaction. 3 Contrary to the above discussed commercial processes, the oxidative dehydrogenation (ODH) of propane to propylene is more attractive due to its low operational cost and minimal environmental impacts. The abundant availability of propane in natural gas and refinery off gases can make the propylene production by ODH propane even more This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

2 sustainable compared to the conventional processes. 2 Its most important advantage is the exothermic nature of the reaction, which requires no additional energy to accelerate the reaction. The formation of water as a by-product of the ODH makes it possible to avoid thermodynamic constraints as observed in the non-oxidative routes. 4 However, the appropriate selection of the reactor is a key factor to achieve the above mentioned advantages of the ODH technology. 5 Fixed bed reactors are simple but difficult to operate under isothermal conditions. Catalyst deactivation is another problem associated with fixed bed systems. The fluidized bed reactors have numerous advantages over the conventional fixed reactor systems. These include controlled operating conditions at constant temperature, which assist in circumventing the issues with hot spots in fixed bed reactors. The absence of limitations of mass transfer and uniform residence time distributions (RTD) is also a merit of fluidized bed reactors. Moreover, transportation of reduced catalytic species from the oxidative dehydrogenation unit to the regeneration unit is also one of the merits of fluidized bed reactors that allows periodic catalyst re-oxidation. This enables a twin reactor set up, one for oxidative dehydrogenation and the other for regeneration of catalyst, which makes this route attractive for commercial scale production. 6 8 In addition to the appropriate reactor selection, the development of suitable catalysts is another key aspect for the commercial implementation of ODH processes. High yield of propylene can be obtained by employing an efficient catalyst. 4 It is reported in the literature that vanadium-based catalysts give the highest alkane conversion and alkene selectivity for ODH of lighter hydrocarbons The wisdom behind this is the provision of lattice oxygen for dehydrogenation of alkanes by vanadium catalysts The reactions involved in the oxidation of propane include the desired propane oxidative dehydrogenation to propylene as well as combustion of propane and produced propylene to carbon oxides. High selectivity for propylene is only favorable at low propane conversions due to lower propane reactivity when compared to propylene. Thus, there is a need to design a catalyst that will provide the optimum level of lattice oxygen to selectively produce propylene with minimum carbon oxides via primary/secondary combustions. 27 In vanadium-based catalysts, the activity and propylene selectivity are directly related to the structure of VO x surface species. The performance of the catalyst can be enhanced by controlling its redox properties, morphology of surface species of VO x and acid base characteristics The surface density of VO x increases with vanadium loading, which is the lowest for monovanadate isolated VO x species and the highest for monolayer coverage species. Catalyst activity and reducibility increases as surface density of VO x increases while its selectivity decreases as the surface density of VO x increases Adjustments of the coordination and environment of the species of VO x can influence the catalytic behavior. The acid base character of VO x catalyst supports has been found to have influence on propylene selectivity in ODH. 4 Propane adsorption and propylene desorption are functions of the acid base properties of the support. Adsorption of a basic reactant and desorption of an acidic product are functions of the acidity of the catalyst. The acidity of the catalyst determines the protection of these chemical species from oxidizing to carbon oxides. 36 The acidic character of alkanes and their corresponding olefins diminishes with increased carbon number and degree of molecule saturation; one can hypothesize that a higher selectivity in ODH could be achieved by designing catalysts with a controlled acidic character. 37,38 There are usually strong interactions between the support (carrier) and the active site (VO x ). Gamma aluminum oxide is not inert towards VO x. Its interaction towards the VO x phase is not weak, hence it results in a very high dispersion of V 2 O 5 on its surface. High vanadium loading can be achieved on γ-al 2 O 3, but resulting samples will show lower surface areas unlike CaO, which has a higher surface area. The use of CaO will improve a resulting catalyst's superficial area and also gives it the desired moderate level of acidity, that will maximize propane adsorption and propylene desorption and also minimize propylene and propane combustion. Hence, the synthesis of mixed γ-al 2 O 3 /CaO supports is an interesting route to achieve catalyst samples with high dispersion of the surface active species, and a surface area that is higher than that of γ-al 2 O Based on the above discussion, the present work is aimed at developing fluidizable vanadium oxide catalysts using a mixed γ-al 2 O 3 CaO with different γ-al 2 O 3 /CaO ratios as support. The prepared catalysts were characterized using various physicochemical characterization techniques (XRD, FTIR, Raman spectroscopy, TPD, TRP/TPO, SEM-EDXS) to determine the desired catalyst properties. The performance of the developed catalysts on the ODH of propane was evaluated in a fluidized CREC Riser Simulator under gas phase oxygen free conditions. In this approach, the oxidative dehydrogenation of propane is accomplished with the lattice oxygen of the catalysts. The major advantage of this process is the possibility of achieving higher propylene selectivity by appropriate control of the lattice oxygen of the catalysts. 2. Materials and methods 2.1. Catalyst synthesis The catalyst samples were prepared by an impregnation method through soaking with excess ethanol as solvent. The support material γ-al 2 O 3 (surface area of 141 m 2 g 1 ) was received from Inframat Advanced Materials, Manchester, UK, while CaO (surface area of 4 m 2 g 1 ) was received from Loba Chemie, India. Before metal loading, the supports were calcined under pure N 2 flow at 500 C for 4 hours to remove moisture and volatile compounds. The calcined γ-al 2 O 3 sample was placed in a beaker and ethanol was added. Desired amounts of vanadyl acetyl acetonate and CaO were then added to the beaker, and the mixture was left under stirring for 12 h. The mixture was then placed for sonication for 10 Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016

3 min. The mixture was filtered and dried at room temperature. Following natural drying, the sample was placed in an oven at 100 C for 24 hours in order to slowly remove any remaining solvent. The dried sample was then reduced with hydrogen (10% H 2 and 90% Ar) at 500 C in an especially designed fluidized bed reactor. Finally, the reduced sample was calcined under air at 500 C for 4 h to obtain the oxide of the catalyst. After this treatment, the catalyst color became yellow indicating the presence of V 2 O 5 on the support surface. Following the above approach, two catalyst samples were prepared with CaO to γ-al 2 O 3 weight ratios of 1 : 4 and 1 : 1, respectively, while keeping the same 10 wt% vanadium loading (VO x /CaO γ-al 2 O 3 (1 : 4), VO x /CaO γ-al 2 O 3 (1 : 1)). The third sample was prepared using pure CaO as support and 10 wt% vanadium (VO x /CaO). The measured BET surface areas of the prepared catalysts range from 14 to 25 m 2 g Catalyst characterization X-ray diffraction (XRD). The XRD patterns of all the samples were recorded on a Rigaku Miniflex diffractometer with monochromatic Cu Kα radiation of nm wavelength, an electrical current of 50 ma, an electrical voltage of 10 kv and a scan rate of 2 per minute (normal scan rate) within the 2θ range from with a 0.02 step size Laser Raman spectroscopy. The molecular structures of various metal oxide species supported on CaO γ-al 2 O 3 and CaO were analyzed using a Horiba Raman spectrometer. For each experiment, 0.5 g of sample was dehydrated under dry air for an hour at 500 C and then cooled to ambient temperature. Then each sample was analyzed using a Raman spectrometer with a thermoelectrically cooled CCD detector ( 73 C). An argon ion laser line of 532 nm wavelength was used to excite the catalyst samples. The Raman spectrometer was used for measuring and recording the spectra produced from the excitation with a resolution of 1 cm 1 at room temperature FTIR spectroscopy. A Nicolet 6700 Thermo Fischer Scientific instrument was used to record the FTIR spectra of the synthesized catalyst samples and the bare support γ-al 2 O 3 and CaO samples. For analysis, 3 mg of sample was uniformly mixed with 0.4 g of potassium bromide. The infrared spectra of pelletized samples were later recorded in the range of cm SEM-EDXS analyses. For SEM analysis, the catalyst samples were dispersed on a stub that is taped with copper tape. Each of the samples were coated with gold in order to eliminate charge build-up, obtain better contrast and enhance visibility at magnification of one million times Temperature programmed reduction (TPR). A Micromeritics AutoChem II 2920 analyzer was used to conduct H 2 - TPR experiments at kpa. For TPR analysis, 0.05 g of catalyst sample was loaded in a U-shaped quartz tube using glass wool to hold the catalyst particles inside. Before analysis, the sample was pretreated under Ar flow at 500 C to remove any volatile component. After pretreatment, the sample was completely oxidized by circulating a gas mixture of 5% O 2 and He balance, at 500 C with a heating rate of 10 C min 1. The sample was then cooled down to ambient temperature under argon flow to ensure flushing out any gas phase O 2 that might be trapped in the catalyst bed. The temperature programmed reduction experiment was carried out by circulating a gas stream of 10% H 2 /Ar at 50 cm 3 min 1. Under these conditions, the sample temperature was raised from the room temperature to 850 C at a heating rate of 10 C min 1. A calibrated thermal conductivity detector (TCD) detected the variation of the hydrogen concentration due to the reduction of the catalyst samples Temperature programmed desorption (TPD). The acidity and acid strength of the catalyst were investigated using ammonia temperature programmed desorption analysis. The NH 3 -TPD desorption kinetics analysis also helps evaluating the metal support interactions of the supported catalyst. The NH 3 -TPD experiments were conducted using an AutoChem II 2029 Analyzer received from Micromeritics, USA. Similar to the TPR experiments, 0.5 g catalyst sample was first loaded into the U-shaped quartz container and degassed for 2 h at 500 C under Ar flow at 30 ml min 1. The sample was then cooled to 120 C and brought to saturation with ammonia using a NH 3 /He gas mixture (5% NH 3 /He) at a rate of 50 ml min 1. Following the ammonia saturation, the system was purged with helium at 100 C at50cm 3 min 1 to remove any gas phase ammonia in the system and unadsorbed ammonia trapped in the catalyst bed. For desorption analysis, the catalyst bed temperature was raised from room temperature to 750 C at10 C min 1. The ammonia chemisorbed was desorbed as the temperature was elevated to 750 C. The ammonia concentration of the effluent gas was monitored by the thermal conductivity detector Fluidized ODH of propane The gas phase oxygen free ODH propane experiments were conducted in a fluidized CREC Riser Simulator (CREC: Chemical Reactor Engineering Centre). The CREC Riser Simulator, a bench scale fluidized reactor (53 cm 3 ), is very useful for catalyst evaluation and kinetic studies. It has several outstanding advantages including the simulation of fluidized conditions of a riser/downer reactor even with a small amount of catalyst, minimization of mass transfer limitations by using small sized catalyst particles, constant residence time distributions and controlled isothermal conditions. The CREC Riser Simulator reactor operates alongside different accessories, which include temperature controllers, a gas chromatograph, a vacuum box, a main power switch, a water pressure indicator and a push button selector. The details of the CREC Riser Simulator can be found in Al-Ghamdi et al. (2012). 6 Propane ODH runs were carried out at different temperatures ranging from 550 C to 640 C while the contact time was varied between s. The reaction temperatures were selected within the reduction temperature range of the This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

4 catalysts as determined from the TPR study, given that the solid catalysts are the only source of oxygen. The ODH of propane experiments were conducted using 0.5 g of catalyst. The oxidized catalyst sample was loaded into the catalyst basket located in the lower shell of the main reactor body of the CREC Riser Simulator. Following the catalyst loading, the system was pressurized up to 30 psi at room temperature to perform a leak test. A stable pressure reading at closed conditions confirmed the absence of any leak. Now the reactor is ready to be heated to the desired temperature. During the heating period, the system was maintained under argon flow to keep the reactor from any air interference. Once the reactor reached the desired temperature level, the argon flow was stopped. Consequently, the reactor pressure started to decrease sharply. The four-port valve was closed, as the reactor pressure approached one atm (14.7 psi). Following the isolation of the reactor, the vacuum pump was turned on to evacuate the vacuum box down to 20.7 kpa (3.75 psi). A preloaded syringe was used to inject 1.2 ml of feed (propane) into the reactor after setting the impeller in motion. The pressure transducer was used to record the pressure profile of the reactor. At the end of the reaction period, the reactor contents were evacuated into the vacuum box. The gas product contained in the vacuum box was analyzed by gas chromatography (GC) using both a thermal conductivity detector (TCD) and a flame ionization detector (FID). Catalyst performance was studied based on propane conversion, selectivity and yield given below. where z j and n j are the number of atoms of carbon and moles of gaseous carbon containing product j, respectively. n propane is the mole of unconverted propane in the product stream. 3. Results and discussion 3.1. XRD analysis Fig. 1 shows the XRD patterns of the three VO x /CaO γ-al 2 O 3 catalyst samples with different amounts of CaO and γ-al 2 O 3 and with the same amount of VO x. For the VO x /CaO γ-al 2 O 3 (1 : 1) and VO x /CaO γ-al 2 O 3 (1 : 4) samples, the γ-al 2 O 3 peaks appeared at 2θ angles of 48 and 67. The peaks which appeared at 32, 38 and 55 can be attributed to CaO. 27,40 All three samples confirmed these peaks and expectedly the intensity of the CaO peaks decreased when the content of CaO in the sample decreased. The 19.5 peak on the VO x / CaO γ-al 2 O 3 (1 : 1) and VO x /CaO catalyst samples can be ascribed to V 2 O 5 crystals. Small V 2 O 5 peaks were also detected in the CaO sample at 2θ angles above 60. This observation indicates that the VO x species in the catalyst samples mainly (1) (2) Fig. 1 XRD patterns of VO x /CaO and VO x /CaO γ-al 2 O 3 catalyst samples. appeared as a highly dispersed amorphous phase on the support samples. 18 The other probable phases, AlV 2 O 9 and CaV 2 O 6, were also not detected in any of the catalyst samples. One can infer from this observation that the reaction between vanadium and the support materials γ-al 2 O 3 and/or CaO is negligible during the treatment even at 750 C Laser Raman spectroscopy Fig. 2 presents the Raman spectra of the catalyst samples, which were obtained at ambient temperature. The Raman spectra analysis suggests that all the three catalyst samples VO x /CaO, VO x /CaO γ-al 2 O 3 (1 : 1), VO x /CaO γ-al 2 O 3 (1 : 4) contain both monovanadate and polyvanadate with minute crystal particles of V 2 O 5. The band at 870 cm 1 is attributed to the stretching mode of the polyvanadate species (V O V). The 1069 cm 1 broad band is ascribed to the stretching mode of the V O bond in isolated monovanadate surface Fig. 2 Raman spectra of VO x /CaO and VO x /CaO γ-al 2 O 3 catalyst samples. Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016

5 species.27 All other bands appearing around 180, 285, and 345 cm 1 are ascribed to bulk V2O5 crystals FTIR analysis Fig. 3 displays the FTIR spectra of VOx/CaO γ-al2o3 (1 : 4), VOx/CaO γ-al2o3 (1 : 1) and VOx/CaO catalysts. The infrared band at 880 cm 1 wavenumber represents γ-al2o3. CaO has bands at 450 cm 1, 1410 cm 1, and 3650 cm 1. Vanadium oxide has an infrared band at 1014 cm The absorption peak at 450 cm 1, 1410 cm 1, and 3650 cm 1 confirms the presence of CaO in all the samples. The band at 1014 cm 1 confirms the presence of V2O5 in the catalyst samples and the band at 880 cm 1 confirms the presence of γ-al2o3 in the VOx/CaO γ-al2o3 (1 : 4) and VOx/CaO γ-al2o3 (1 : 1) catalyst samples. The peak at 1014 cm 1 corresponds to a strong terminal oxygen bond (V5+ O) SEM-EDX analysis Scanning electron microscopy (SEM) together with energy dispersive X-ray analysis (EDX) was carried out to determine the metal dispersion. A representative field emission scanning electron microscope image of one of the catalyst samples VOx/CaO γ-al2o3 (1 : 1) is presented in Fig. 4a. The images of the elemental distribution can be used to envisage the quality of the dispersion. The dispersion of the element vanadium over the oxygen carrier samples is shown in Fig. 4b. It is evident that vanadium particles are well dispersed on the CaO γ-al2o3 support Reduction and oxygen carrying capacity TPR/TPO is an important technique for the characterization of gas phase oxygen free ODH catalysts. It gives information about the reducibility and regeneration ability of the catalyst as expected during the actual ODH reaction with propane (eqn (3) and (4)). TPR: V2O5 + 2H2 V2O3 + 2H2O (3) ODH of propane: C3H8 + ½V2O5 C3H6 + H2O + ½V2O3 (4) One can see that both the TPR (eqn (3)) and ODH of propane (eqn (4)) reduce V2O5 to V2O3. On the other hand, the TPO cycle (eqn (5)) represents the catalyst regeneration cycle following the reduction in TPR. TPO: V 2 O3 + O2 V 2 O5 Fig. 3 FTIR absorption spectra of VOx/CaO and VOx/CaO γ-al2o3 catalyst samples. (5) In addition, the TPR/TPO data can be further processed to determine the oxygen carrying capacity of the catalysts for Fig. 4 a. SEM images of VOx/CaO γ-al2o3 (1 : 1). b. Vanadium elemental mapping in the VOx/CaO γ-al2o3 (1 : 1). This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

6 the oxidative dehydrogenation of propane without any additional gas phase oxygen (catalyst reduction cycle). Therefore, TPR analysis indicates the temperature range of catalyst activation and amount of available lattice oxygen for ODH of propane. Fig. 5 presents the TPR profiles of the VO x /CaO γ-al 2 O 3 catalyst with different CaO to γ-al 2 O 3 ratios. The TPR profiles show that the VO x /CaO sample has two humps between C and C while VO x /CaO γ-al 2 O 3 (1 : 1) shows only one hump between 260 and 450 C due to the highly reducible VO x species that appeared on the support surfaces. The low temperature reduction hump with the VO x /CaO γ- Al 2 O 3 (1 : 4) sample was less pronounced than the other two samples. In addition to the initial reduction humps, all the three catalyst samples exhibit a major reduction peak between C. While the initial reduction hump can be attributed to the reduction of bulk V 2 O 5 -like surface species, the major peak confirmed the presence of monomeric and polymeric VO x species. For all the catalyst samples, there was no peak attributed to CaO or Al 2 O 3. This is due to the fact that calcium and aluminum are higher in the electrochemical series compared to vanadium and hydrogen. Indeed, the temperatures that will be required for reduction of CaO and Al 2 O 3 with hydrogen are higher than the temperature range considered in the TPR experiment. However, the reduction peak temperatures of the samples significantly varied with the variation of the CaO content in the catalyst formulation. The peak temperature of the lowest CaO containing VO x / CaO γ-al 2 O 3 (1 : 4) sample was 515 C. With the increase in the CaO content, for the VO x /CaO γ-al 2 O 3 (1 : 1) sample, the peak temperature shifted to 560 C. The CaO supported VO x / CaO sample shows the highest peak temperature at 583 C. This shift of reduction temperature is possibly due to the increased active site support interactions introduced by the addition of CaO. 44,45 The TPR data were further processed to evaluate the degree of reduction for the three catalyst samples. The degree Fig. 5 Temperature programmed reduction profiles of VO x /CaO and VO x /CaO γ-al 2 O 3 catalyst samples. of reduction can be defined as the percentage of VO x reduced to the actual quantity of vanadium oxide available in the catalyst. The exposed reducible VO x was calculated from the amount of hydrogen uptake evaluated by numerical integration of the resulting temperature programmed reduction peak area. The mass of reducible vanadium oxide in the catalyst sample was evaluated using the molar volume of gas at STP, volume of hydrogen uptake, molecular weight of vanadium oxide and stoichiometric number of hydrogen in the gas solid reaction involved in reduction. The percentage of vanadium oxide reduction can be calculated using the following relations: where (1) W V is the amount of reduced vanadium (g), (2) MW v is the molecular weight of vanadium (g mol 1 ), (3) V H2 is the volume of reacted hydrogen (cm 3 at STP), (4) V g is the molar volume of gas (cm 3 mol 1 at STP), (5) W o is the initial weight of vanadium (g) and (6) v is the stoichiometric number of hydrogen based on the following reaction stoichiometry. Assuming that V 2 O 5 is the initial reducible catalyst species present on the support, then the following reduction equation applies: (6) (7) V 2 O 5 +2H 2 V 2 O 3 +2H 2 O (8) Table 1 shows the hydrogen uptake of the catalyst samples. One can see from this table that the hydrogen uptake increased at higher CaO contents in the catalyst samples. The higher hydrogen uptake is possibly due to the higher dispersion of vanadium species as observed in the low temperature reduction humps of the VO x /CaO γ-al 2 O 3 (1 : 1) and VO x / CaO catalysts. The increased basicity with higher quantity of CaO also contributes to the increased hydrogen consumption, given the acidic nature of hydrogen gas, which has higher affinity to catalysts with higher basicity. In order to assess the oxygen carrying capacity and stability, the catalyst samples were subjected to repeated TPR and TPO cycles. The hydrogen consumption in each TPR cycle was measured using the calibrated TCD signals. The percentage of reduction of the catalysts in each TPR cycle calculated from the hydrogen uptake data is presented in Table 1. It was observed that in the repeated TPR/TPO cycles, the hydrogen uptake for the samples was within 2.5% error range (Table 1). This observation indicates that the oxygen carrying capacity of the catalyst remains stable over the repeated redox (TPR/TPO) cycles NH 3 -TPD The acid sites of the three catalyst samples were characterized by TPD using NH 3 as the basic probe molecule. The area Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016

7 Table 1 TPR data comparing hydrogen consumption of VO x /CaO and VO x /CaO γ-al 2 O 3 catalyst samples H 2 uptake (mmol g 1 ) (% reduction) % Sample 1st 2nd 3rd 4th Average Error VO x /CaO γ-al 2 O 3 (1 : 4) 1.9 (48%) 1.8 (46%) 1.8 (45%) 1.8 (45%) VO x /CaO γ-al 2 O 3 (1 : 1) 2.6 (65%) 2.5 (63%) 2.4 (62%) 2.4 (62%) VO x /CaO 2.9 (72%) 2.8 (70%) 2.8 (70%) 2.7 (69%) of the TPD curve peak gives the acid amount while the position of the peak indicates the acid distribution in the catalyst samples. Ammonia TPD can distinguish sites only by sorption strength, hence its shortcoming lies in its inability to differentiate between Lewis and Bronsted acid sites. Ammonia was used in this research work to make comparison of the total acidity and acid strength for the catalyst samples with different CaO/Al 2 O 3 ratios. Fig. 6 shows the relationship between the desorption volume as a function of the temperature. One can easily see that all three samples show similar TPD profiles although the peak intensity and desorption peaks shifted with the variation of the CaO/Al 2 O 3 ratios. The NH 3 -TPD profile for the VO x /CaO γ-al 2 O 3 (1 : 4), VO x /CaO γ- Al 2 O 3 (1 : 1) and VO x /CaO samples showed an initial desorption peak at 183, 300, and 302 C followed by a high temperature desorption peak at 676, 636 and 620 C, respectively. Clearly, the intensities of the high temperature desorption peaks were significantly higher than that of the low temperature peaks. This indicates that the percentage of the strong acid sites is much higher than the weak acid sites. The total acidity of each sample was calculated by integrating the calibrated TPD profiles. Table 2 shows the uptake of NH 3 by the three catalyst samples and their respective temperature of desorption. Expectedly, the total acidity of the catalyst samples decreased when increasing the CaO content, which is due to the basic nature of CaO NH 3 -TPD kinetics study An ammonia desorption kinetics study was conducted to determine active site support interactions in the catalyst samples. The activation energy of ammonia desorption and the preexponential factors were estimated by modeling the NH 3 -TPD experimental data of each catalyst sample. Cvetanovic and Amenomiya described the desorption rate as a function of temperature, which is based upon the following assumptions: 6,8 (i) Temperature (T) of desorption has a linear relationship with time (t). (ii) The rate of desorption is of first order in coverage. (iii) The concentration of ammonia gas through the catalyst bed is uniform. (iv) Desorbed ammonia has zero feasibility for re-adsorption. (v) The catalyst's surface is homogenous for the NH 3 adsorption, which means that for the desorption constant, k d = k do expij E/RT). The desorption constant is independent of the surface coverage. Suitable experimental conditions were selected in order to satisfy the assumptions in (i) and (iii). A high flow of ammonia gas through the catalyst bed was maintained in order to satisfy the assumption in (iv). Unimolecular desorption of ammonia was assumed in order to consider the assumption in (ii). The ammonia desorption rate at a uniform first order energy of desorption can be evaluated using a component balance of desorbing NH 3. where T m is the centering temperature in C, V m is the volume of NH 3 adsorbed at saturated conditions in ml g 1, V d is the volume of ammonia desorbed at different temperatures in ml g 1, θ is the surface coverage of the adsorbed species, E is the energy of ammonia desorption in kj mol 1, and k do is the pre-exponential factor in ml g 1 min 1. Temperature (T) in a TPD experiment has a linear relationship with time (t). (9) T = T o + αt (10) where T is the desorption temperature at time (t). (11) Fig. 6 NH 3 -temperature programmed desorption profiles for the catalyst samples. (12) This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

8 Table 2 Catalyst acidity as measured by NH 3 -TPD Catalyst sample Low temp High temp Low temp High temp Total VO x /CaO γ-al 2 O 3 (1 : 4) (17%) 2.69 (83%) 3.24 VO x /CaO γ-al 2 O 3 (1 : 1) (21%) 2.04 (79%) 2.58 VO x /CaO (26%) 1.58 (74%) 2.13 (13) (14) (15) The first order ordinary differential equation was solved using the separation of variable method to obtain the following equation: (16) V o and T o are the initial volume desorbed in ml g 1 and the initial desorption temperature. R is the universal gas constant in kj mol 1 K 1 ; the heating rate, α, was taken as 10 C min 1. The TPD data obtained from the experiment and the proposed model have good agreement for all the catalyst samples as shown in Fig. 7. This proves the validity of the proposed desorption model. The TPD data was correlated to the resulting equation (eqn (16)) using the non-linear regression analysis tool of MATLAB. Hence, the desorption energies and pre-exponential factors of each catalyst sample were obtained. The norm of the residuals and the coefficient of correlation were calculated for each catalyst sample using MATLAB and MINITAB software at 95% confidence limit. The energy of desorption for the three synthesized catalysts is reported in Table 3. Statistical properties such as the correlation coefficient R 2, norm of residuals and 95% confidence intervals were considered in the analysis. The values of R 2 and residual norms for all the three catalysts are close to 1 and 0, respectively, which shows that the proposed desorption model is applicable. Peak temperatures C NH 3 uptake (mmol g 1 ) The values in the table show that as the loading of CaO is increased and that of γ-al 2 O 3 is decreased, the energy of desorption increases. This can be explained based on the amount of ammonia uptake for each of the catalyst. The catalyst with the highest desorbed ammonia has the lowest desorption energy while the one with the lowest desorbed ammonia has the highest desorption energy. A similar observation was described by Al-Ghamdi et al. 27 on γ-al 2 O 3 supported VO x catalysts where a higher desorption energy corresponds to a lower amount of NH 3 adsorbed from the catalysts. The increase in the activation energy can also be linked to the heterogeneity of the catalyst samples. The interaction between the mixed support and the active site (VO x ) would also play a significant role in the value of energy required during the gas solid reactions involved during the oxidative dehydrogenation of propane under the gas phase oxygen free conditions. Weak active site support interactions would allow an easy reaction between VO x and propane/ propylene as opposed to strong active site support interactions. However, these surface oxygens are less selective to the formation of propylene. Hence, the moderate active site support interactions as shown by the VO x /CaO γ-al 2 O 3 (1 : 1) catalyst can be favorable to achieve higher propylene selectivity Catalyst evaluation The gas phase oxygen free oxidative dehydrogenation (ODH) of propane experiments were conducted in a fluidized CREC Riser Simulator using pure propane (99.95% purity) as feed. Before performing the actual catalytic ODH runs, thermal Fig. 7 Experimental data and fitted model of ammonia desorption during NH 3 -TPD for different catalyst samples. Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016

9 Table 3 Estimated parameters for ammonia-tpd kinetics at 10 C min 1 Sample E (kj mol 1 ) k do (ml g 1 min 1 ) 10 5 Norm of residuals 10 4 (mmol g 1 ) VO x /CaO Al 2 O 3 (1 : 4) VO x /CaO Al 2 O 3 (1 : 1) VO x /CaO experiments (without any catalyst) were conducted to confirm contribution of any thermal conversion. The highest reaction temperature (640 C) was selected for the thermal experiments. The GC analysis of the thermal run products showed mainly unconverted propane and a trace amount of ethane and methane most likely due thermal cracking of propane in the absence of a catalyst. In the catalytic experiments, the reaction temperature was varied between 550 and 640 C, while the reaction was attuned from 10 to 31 s. The product analysis of the preliminary experimental runs shows that the product is composed of unreacted propane, propylene and carbon dioxide. Under the studied reaction conditions, no hydrogen was detected, indicating the absence of cracking and/or dehydrogenation. The propane conversion and product selectivity in the experimental repeats are found to be within 2.5% error limits (Fig. 8). Mass balances were established for each of the three repeats of each individual run. The mass balance closed consistently in excess of 95%. From the product analysis, one can consider the following possible reaction steps during the fluidized ODH of propane runs in the absence of gas phase oxygen: (17) Complete oxidation of propane: C 3 H 8 +5V 2 O 5 3CO 2 +4H 2 O+5V 2 O 3 (18) Fig. 8 Propane (C 3 H 8 ) conversion and propylene (C 3 H 6 ) and CO 2 selectivity and their error limits in repeated runs at T: 640 C; cat.: 0.5 g; propane injected: 1.2 ml. (19) therefore, it is very important to identify the best reaction conditions in order to achieve the highest possible propylene yield and suppress the complete combustion reactions which produce CO 2. Keeping the above in mind the following experiments were conducted under different conditions to demonstrate the effects on the propane conversion and product selectivity of (i) the consecutive propane injection without catalyst regeneration, (ii) reaction temperatures and (iii) contact times Successive propane injections. The successive oxidative dehydrogenation of propane without catalyst regeneration experiments were conducted to demonstrate the effects of the degree of catalyst reduction on the propane conversion and product distribution. To ensure the same reaction conditions, the reactor was loaded with 0.5 g of catalyst and the temperature was maintained at 640 C. Further, in each run, the same amount (1.2 ml) of propane was injected and the reactions were allowed to proceed for 17 s. Fig. 9 plots the propane conversion and propylene and carbon oxide selectivity over the successive injection of propane runs. One can see in Fig. 9(a) that all three VO x /CaO Al 2 O 3 catalysts give the highest propane conversion in the first injection, which gradually decreased in the following successive propane injections. The availability of the oxygen in the catalyst surface mainly contributed to the high propane conversion in the first injection. The appreciable levels of catalyst activity after all the four successive injections can be attributed to the lattice oxygen availability in the catalyst matrix. On the other hand, the diminishing trend of the propane conversion is due to the progressive consumption of lattice oxygen in the catalysts. Among the three catalysts, VO x /CaO γ-al 2 O 3 (1 : 1) displays the highest propane conversion and propylene selectivity. This is consistent with its moderate acidity, moderate active site support interactions and balanced oxygen carrying capacity compared to the other two catalysts as observed in the TPR analysis. The selectivity of both the desired propylene and undesired carbon dioxide also significantly varies during the successive propane injection runs as seen in Fig. 9(b). Unlike propane conversion, the first injection gives the lowest propylene selectivity and the highest carbon dioxide selectivity. This indicates that the surface oxygen favors the complete oxidation of propane/propylene producing carbon dioxide. This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

10 Fig. 9 a. Conversion of propane in successive propane injections without catalyst regeneration. b. C 3 H 6 and CO 2 selectivity in successive propane injections without catalyst regeneration (T: 640 C; cat.: 0.5 g; propane injected: 1.2 ml, time: 17 s; experimental repeats: ±2.5% standard deviation). The propylene selectivity significantly increased in the second injection; after that the increment became minimal in the remaining runs although the increasing trend is still evident. This variation in selectivity indicates that an optimum level of lattice oxygen is required to maximize selectivity to propylene and minimize selectivity to carbon dioxide. The above observations are in line with the fact that selectivity to propylene in oxidative dehydrogenation of propane over VO x -based catalysts is affected positively by the binding energy between the lattice oxygen and the catalyst. 46 At higher oxidation state of the catalyst, the binding energy of the lattice oxygen is low (low active site support interaction), which eventually leads to combustion of propane/propylene to carbon oxides. Furthermore, the surface oxygen atoms on the fresh or regenerated catalyst are loosely bonded with the catalysts, which easily react with propane/propylene to produce carbon dioxide. In this case, a selective catalyst surface would be obtained only after consumption of the bulk V 2 O 5 - like surface species in the first propane injection. When compared with the other two catalysts, VO x /CaO γ- Al 2 O 3 (1 : 1) shows significantly higher propylene selectivity. It also shows lower carbon dioxide selectivity than that of the VO x /CaO catalyst. This catalyst shows up to 96% propylene Fig. 10 a. Conversion of propane at different temperatures. b. C 3 H 6 and CO 2 selectivity at different temperatures (cat.: 0.5 g; propane injected: 1.2 ml, time: 17 s; experimental repeats: ±2.5% standard deviation). selectivity while the higher CaO containing catalysts produces up to 83% propylene. This can be attributed to the moderate level of acidity of VO x /CaO γ-al 2 O 3 (1 : 1) as depicted in the NH 3 -TPD results. This observation is also consistent with the XRD and TPR results. The appropriate balance of CaO/Al 2 O 3 influences the VO x dispersion forming more isolated noncrystalline VO x species, which favors the propylene formation and suppresses the complete oxidation to CO 2. Furthermore, the increased V support interaction with the CaO promoted sample (revealed by the TPD kinetics analysis) also favors propylene production by a controlled ODH reaction between propane and lattice oxygen of the catalyst. There are studies in the open literature which showed propylene selectivity as a function of the oxidation state of vanadium-based catalysts. 6,34,43,46 49 Lopez-Nieto et al. 34 found that the selectivity to propylene and butylene, with usage of propane and butane as the feed, respectively, could be strongly influenced by the reducibility of the vanadium-based catalyst. Balcaen et al. 47 also observed the same trend for ODH of propane over vanadium-based catalysts. Ethane ODH over a γ-alumina supported vanadium catalyst in the absence of oxygen, as shown by Al-Ghamdi et al., 46 also confirms that the absence of gas-phase oxygen is important for the selective conversion Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016

11 of alkane to alkene with the binding energy of lattice oxygen as the main driver of the reaction Effect of reaction temperature. Fig. 10(a and b) present propane conversion and the desired product propylene and undesired carbon dioxide selectivities at different reaction temperatures and a constant 17 s contact time. These experiments are conducted using oxidized catalysts. After each run the catalyst was re-oxidized by circulating air through the catalyst bed. One can see that the VO x /CaO γ-al 2 O 3 (1 : 4) sample shows very low conversion at 550 C, which is consistent with its higher initial reduction temperature as observed in the TPR analysis (Fig. 5). The VO x /CaO γ-al 2 O 3 (1 : 4) catalyst sample is mainly reduced between 520 and 580 C. Therefore, there is only a small fraction of lattice oxygen available for reaction at 550 C, leading to a low propane conversion. In contrast, both the VO x /CaO γ-al 2 O 3 (1 : 1) and VO x /CaO samples show some reduction at low temperatures, which might have contributed to the higher propane conversions at 550 C when using these two samples. The propane conversion increased with the increasing reaction temperature as the lattice oxygen of the catalyst is activated at higher temperature (Fig. 5, TPR analysis). Interestingly, when increasing the reaction temperature, all the catalyst showed increased propylene selectivity and decreased carbon dioxide selectivity (Fig. 10(b)). The variation in the degree of reduction of the catalyst with the reaction temperatures was responsible for the rise in the selectivity of propylene. At higher temperatures, the degree of catalyst reduction increases (Fig. 5, TPR analysis) as a result of the lower binding energy of lattice oxygen. At such higher degrees of reduction of the catalysts, the selective pathway toward ODH is preferred over that for combustion as observed in the successive propane injection experiments. The good selectivity to propylene can also be attributed to the non-formation of larger molecules due to the interaction of the mixed support and the active site of each catalyst. Among the three studied catalysts, VO x /CaO γ-al 2 O 3 (1 : 1) shows the highest propylene selectivity. The carbon dioxide selectivity with this catalyst is also lower than that of the VO x /CaO catalyst while slightly higher than the VO x / CaO γal 2 O 3 (1 : 4) catalyst. The superior propylene selectivity of the VO x /CaO γ-al 2 O 3 (1 : 1) can be attributed to the moderate level of acidity of VO x /CaO γ-al 2 O 3 (1 : 1) as depicted in the NH 3 -TPD results Effect of contact time. Propane ODH experiments were carried out at 10, 17, 24 and 31 s in order to study the effect of contact time on propane conversion, propylene selectivity and carbon oxide selectivity at the optimal temperature, 640 C. It is evident that propane conversion for all catalysts increases with the contact time as shown in Fig. 11(a). The propylene selectivity slightly increases from 10 to 17 s and after that it decreases with contact time (above 17 s). Conversely, the carbon dioxide selectivity slightly increases from 17 to 31 s (Fig. 11(b)). The decrease of propylene selectivity and increase of carbon dioxide selectivity at higher contact time are mainly due to the consecutive oxidation of propylene and/or complete oxidation of propane to carbon dioxide. 27 Fig. 11 a. Conversion of propane at different contact times. b. C 3 H 6 and CO 2 selectivities at different contact times (cat.: 0.5 g; propane injected: 1.2 ml, T: 640 C; experimental repeats: ±2.5% standard deviation). The above observation suggests that, although the higher contact time favors high propane conversions, the optimum or moderate contact time favors high propylene selectivity and low carbon dioxide selectivity. The good selectivity to propylene obtained from the three catalysts can be attributed to the high proportion of monovanadate VO x species which Fig. 12 C 3 H 6 and CO 2 selectivities as a function of C 3 H 8 conversion at constant temperature (cat.: 0.5 g; propane injected: 1.2 ml, T: 640 C; experimental repeats: ±2.5% standard deviation). This journal is The Royal Society of Chemistry 2016 Catal. Sci. Technol.

12 Table 4 Comparison of performance of VO x /CaO γ-al 2 O 3 with other ODH catalysts in the literature Catalyst Reactivity temperature C was detected from the laser Raman spectroscopy results. Again, VO x /CaO γ-al 2 O 3 (1 : 1) shows the highest propane conversion and propylene selectivity. This can be attributed to the moderate level of acidity of VO x /CaO γ-al 2 O 3 (1 : 1) as depicted in the NH 3 -TPD results. Thus, one can conclude that the performance of the VO x / CaO γ-al 2 O 3 (1 : 4), VO x /CaO γ-al 2 O 3 (1 : 1) and VO x /CaO catalyst samples is strongly influenced by both contact time and temperatures as well as catalyst regeneration. It can be inferred that successive feed injections are the best for the ODH reaction, since it is only on completion of the successive reaction cycles that catalysts could be regenerated. This can be applied industrially using a fluidized bed reactor that has reactor regenerator compartments and has the ability to transfer a small percentage of the catalyst to the regenerator. Fig. 12 presents propylene and carbon dioxide selectivities as a function of propane conversion at constant temperature. One can see that with increasing propane conversion, propylene selectivity decreases which is compensated by increasing CO 2 selectivity. All the three catalysts show similar trends. This suggests that propylene is the primary reaction product of propane while CO 2 comes from deep oxidation of propane as well as consecutive oxidations of propane and propylene. The performance of the catalyst of our present study, VO x / CaO γ-al 2 O 3 (1 : 1), is compared with the ODH catalyst performance from past literature (Table 4). There is an appreciable comparison of the propylene selectivity of the VO x /CaO γ- Al 2 O 3 (1 : 1) catalyst with the other catalyst as shown in the table. This suggests that the ODH of propane with lattice oxygen (gas phase oxygen free conditions) is promising to enhance the propylene selectivity even at higher propane conversion. 4. Conclusions The following are the conclusions of the present study: i. By using FTIR, Raman spectroscopy and XRD, V 2 O 5, CaO and γ-al 2 O 3 species were detected in the synthesized catalysts. The XRD showed a very small amount of crystalline VO x phases. The remaining VO x appeared as an amorphous phase. ii. SEM images and elemental mapping showed good vanadium oxide dispersion on the mixed CaO γ-al 2 O 3 support. iii. The repeated TPR/TPO experiments showed consistent reduction and re-oxidation behavior of the prepared catalysts. The oxygen carrying capacity of the catalysts was increased with increasing CaO content. Propane conversion% (at the highest selectivity) iv. The acidity of the catalysts progressively decreased with increasing CaO content, as revealed by NH 3 -TPD analysis. v. The activation energy of ammonia desorption decreased due to the increased amount of CaO reflecting higher active site support interactions. The increased active site support interaction controls the reaction of the lattice oxygen and favors propylene as the selective product. vi. Gas phase oxygen free conditions favored the formation of selective product propylene and minimized complete oxidations to CO x. A higher degree of catalyst reduction favored the propylene selectivity. vii. The catalyst with intermediate acidity and moderate active site support interactions (VO x /CaO γ-al 2 O 3 (1 : 1)) displayed the highest propylene selectivity (85%) at higher propane conversion (65%). viii. The isolated VO x phases, as observed in FTIR spectroscopy and XRD, are favorable towards higher propylene selectivity. Acknowledgements The author(s) would like to acknowledge financial support provided by the King Abdul Aziz City for Science and Technology (KACST) to this project under research grant number AT References Highest propylene selectivity % 17.5% MoO/γ-Al 2 O % MoO/γ-Al 2 O % VO x /γ-al 2 O VO x /CaO γ-al 2 O 3 (1 : 1) The present study Ref. 1 Market Study: Propylene, Ceresana Research, February 2011, ceresana.com. Retrieved R. D. Ashford, Ashford's Dictionary of Industrial Chemicals, 2011, 3rd edn., ISBN , pp S.A.Al-Ghamdi,Oxygen-freepropane oxidative dehydrogenation over vanadium oxide catalysts: reactivity and kinetic modeling, Ph.D. dissertation monograph, E. Heracleous, M. Machli, A. A. Lemonidou and I. A. Vasalos, Oxidative dehydrogenation of dehydrogenation of ethane and propane over vanadia and molybdena supported catalysts, J. Mol. Catal. A: Chem., 2005, 232, L. Chalakov, L. K. Rihko-Struckmann, B. Munder and K. Sundmacher, Oxidative dehydrogenation of ethane in an electrochemical packed-bed membrane reactor: Model and experimental validation, Chem. Eng. J., 2009, 145, S. A. Al-Ghamdi, M. Volpe, M. M. Hossain and H. I. de Lasa, VO x /c-al 2 O 3 catalyst for oxidative dehydrogenation of ethane to ethylene: desorption kinetics and catalytic activity, Appl. Catal., A, 2013, 450, A. W. H. Elbadawi, M. S. Ba-Shammakh, S. A. Al-Ghamdi, S. A. Razzak and M. M. Hossain, Reduction kinetics and Catal. Sci. Technol. This journal is The Royal Society of Chemistry 2016