Propane Oxidative Dehydrogenation on Vanadium-Based Catalysts under Oxygen-Free Atmospheres

Catalytic propane oxidative dehydrogenation (PODH) in the absence of gas phase oxygen is a promising approach for propylene manufacturing. PODH can overcome the issues of over-oxidation, which lower propylene selectivity. PODH has a reduced environmental footprint when compared with conventional oxidative dehydrogenation, which uses molecular oxygen and/or carbon dioxide. This review discusses both the stoichiometry and the thermodynamics of PODH under both oxygen-rich and oxygen-free atmospheres. This article provides a critical review of the promising PODH approach, while also considering vanadium-based catalysts, with lattice oxygen being the only oxygen source. Furthermore, this critical review focuses on the advances that were made in the 2010–2018 period, while considering vanadium-based catalysts, their reaction mechanisms and performances and their postulated kinetics. The resulting kinetic parameters at selected PODH conditions are also addressed.


Introduction
Propylene is one of the most important building blocks in the petrochemical industry [1][2][3]. It is industrially employed to produce polypropylene, which is used extensively to make packaging and labeling, textile products, laboratory equipment, loudspeakers, and automotive components. Propylene is also used for the manufacturing of acrylonitrile, propylene oxide derivatives, and other substances [4,5]. The current global propylene demand is the range of 90 million metric tons (MMT) per year, according to the IHS Chemical World Analysis. It is estimated that this demand will rise up to 130 MMT per year by 2023 [6,7].
The traditional olefin production involves steam cracking, fluid catalytic cracking (FCC), and catalytic dehydrogenation (DH) [1,3,[8][9][10]. Typical feedstocks for ethane steam cracking are LPG (light petroleum gas) and naphtha. These olefin production processes involve homogeneous reactions where the hydrocarbon species are steam cracked into smaller olefins [3]. For FCC, commonly used feedstocks are vacuum gas oil, refinery hydrocarbon residues, and de-asphalted oil, which are converted into light and higher value products, such as gasoline. Cracking processes are endothermic and, thus, consume large amounts of heat. To accomplish this, these processes require reactor designs that can be operated at high temperatures [9]. Additionally, significant undesirable amounts of coke are formed, imposing severe operating constraints with frequent plant shutdowns [1,10,11].
Catalytic dehydrogenation (CDH) is an economical route for upgrading low-cost saturated alkanes, such as ethane and propane, into the more valuable olefin feedstocks (e.g., ethylene, propylene) [12]. Furthermore, and given the recent increase of shale gas availability [13][14][15], there is a renewed interest in efficient and economical routes to convert alkanes into olefins. One should note that CDH is At chemical equilibrium, a reacting system achieves a condition where the total Gibbs Free Energy is minimized [69]. Two approaches are possible for assessing chemical equilibrium: (a) stoichiometric and (b) non-stoichiometric [70,71]. The "stoichiometric" approach requires a defined stoichiometry, involving all reactants and products with their molecular formulae. This stoichiometric approach is the one considered for PODH. Table 1 reports both the ∆G R and the ∆H R , at the reference condition of 298K, for the various reactions that are involved in propane ODH, under both O 2 -rich and O 2 -free atmospheres. Thus, one can observe that the PODH reactions with gaseous oxygen are more exothermic than the catalytic PODH reactions using the lattice oxygen. Thus, there is in principle, a valuable strategy if PODH reactions are driven by lattice oxygen. These based lattice oxygen reactions can likely provide conditions for the better management of temperature runaways and better control of undesired reactions.
Propane can be converted under the absence of oxygen in the gas phase (e.g., cracking and dehydrogenation relations), as described in Equations (5)- (9). However, these reactions are negligible in the 500-550 • C range and under short reaction times (<20 s). Thus, one can ignore them in equilibrium calculations [59].
However, and if oxygen is co-fed with propane, several reactions can take place as described in Equations (10)- (19). If one examines the negative ∆G R298 for all these reactions, one can observe that chemical equilibrium favors various postulated reactions. The equilibrium constant K r for each of these reactions can be related to the ∆G r , according to the following equation: where, R is the universal gas constant with a 8.314 J/(mol K) value.
As an example, when considering Equation (10), at 550 • C and with pressures close to atmospheric, one can see that the activities of the chemical species involved in oxy-dehydrogenation can be approximated with their respective partial pressures, as follows: Chemical species partial pressures can be further expressed in terms of the system total pressure P and the gases mole fractions, as follows: In the case of Equation (10) in Table 1, and given that all the other K 823 equilibrium constants for oxy-dehydrogenation with oxygen being co-fed (Equations (11)- (19) in Table 1) exceed the value of 1 by far, one can see that these gas phase reactions are not limited by chemical equilibrium. The anticipated product molar fractions for gas phase PODH led to limited propylene selectivity with significant yields of the undesirable CO 2 and ethylene [29,30].
Thus, given these facts, the discovery of alternative catalytic pathways for PODH to improve the homogenous PODH is a requirement. One possible promising avenue is the one of using V 2 O 5 supported on γ-alumina.
One can see that the PODH, as shown in Table 1 for a V 2 O 5 catalyst, might occur via a network of possible reactions (Equations (20)- (24)). For Equation (20) of Table 1, the following relation is anticipated under chemical equilibrium: Equation (20) (refer to Table 1) displays a ∆G R,298 = −41.5 KJ/mol and a K 823K = 5.88×10 8 , as reported in Figure 1. This shows that the PODH is unconstrained by chemical equilibrium at the anticipated selected process conditions. While this is favorable, one can also see that there is a host of other undesirable secondary reactions with even larger ∆G R,298 competing with Equation (20). Thus, thermodynamic equilibrium analysis for catalytic PODH oxy-dehydrogenation is valuable. It shows the promise of PODH via Equation (20) while pointing to the need of new catalysts to achieve high propylene selectivities.
Finally, and regarding the ODH process under an O2-free atmosphere, the catalyst regeneration reaction (2V2O4 + O2 = 2V2O5), which can be calculated by Equation (25) in Table 1, can be effected in a separate regenerator unit. This catalyst regeneration is both spontaneous and exothermic. Therefore, the energy for the PODH under an O2-free atmosphere with a high selectivity for Thus, thermodynamic equilibrium analysis for catalytic PODH oxy-dehydrogenation is valuable. It shows the promise of PODH via Equation (20) while pointing to the need of new catalysts to achieve high propylene selectivities.
Finally, and regarding the ODH process under an O 2 -free atmosphere, the catalyst regeneration reaction (2V 2 O 4 + O 2 = 2V 2 O 5 ), which can be calculated by Equation (25) in Table 1, can be effected in a separate regenerator unit. This catalyst regeneration is both spontaneous and exothermic. Therefore, the energy for the PODH under an O 2 -free atmosphere with a high selectivity for propylene being endothermic can be recovered from the catalyst regeneration reaction.

Conversion, Selectivity and Yield Calculations
Paraffin conversion in PODH can be described while using two possible approaches, as described in Table 2: (a) Propane conversion based on Equation (26), using as the basis, the propane fed to the reactor unit. This equation might involve inaccuracies given the potential gas leaks throughout the reactor unit [3,9,56,[74][75][76].
(b) Propane conversion calculated via Equation (27). This propane conversion provides a more accurate propane conversion than Equation (26), as it is dependent on the outlet product measured while using the outlet FID products [10,12,77,78].
Furthermore, propylene selectivity is also a very important reaction parameter. It is subject to similar potential errors than propane conversion, given it is based on input feed conditions (Equation (28)) [3,76]. As an alternative, Equation (29) provides, in principle, a more adequate selectivity definition, where all products need to be considered, including water [9,56,75]. The use of Equation (29) is frequently not recommended, given that water measurements offers analytical challenges and is often inaccurate. As an alternative, Equation (30) using carbon containing species established using output product FID analysis [10,12,74,77,78] excluding water is strongly recommended. Furthermore, and regarding product yields, Equation (31) accounts for the product between propane conversion and propylene selectivity [12,[75][76][77][78]. Consequently, this equation relies on the accuracy of the calculated conversion and selectivity.

Conversion
where, n C3H8,in = moles of propane in; n C3H8,out = moles of propane out; ν i = number of carbon atoms in gaseous carbon containing product i; n i = moles of gaseous carbon containing product i.
where, n ii = moles of product i; n T = total products moles.

Vanadium-Based Alkane ODH Catalysts
Catalyst development for ODH reactions has focused on a variety of metal oxides alone or combined with additives (alkali metals and halides). These oxides can be deposited on a variety of supports. While the first studies used metal oxides as catalysts directly, supported catalysts show advantages over unsupported catalysts [79], as follows: a) they provide a better control of metal loading and metal dispersion, and b) they offer an added flexibility to adjust physicochemical properties. For example, TiO 2 supported VOPO 4 catalysts give higher ethylene selectivity than the unsupported VO x and (VO) 2 P 2 O 7 catalysts [80]. Furthermore, researchers [28,50] also focused on promoter addition, which can be used to improve the catalyst performance. Promoters isolate active surface species, forming secondary metal oxides on the support surface.
Research on catalysts for the ODH process usually falls into two operating thermal levels: above or below the temperatures at which significant gas phase reactions take place. High temperature catalysts mainly contain alkali-metal and/or alkaline-earth-metal oxides supported on transition-metal oxides, rare-earth-metal oxides, and other catalytic materials. Low temperature catalysts, on the other hand, usually consist of reducible transition metal oxides [12,59].
As alkali and alkaline earth-based catalysts contain non-reducible ions and oxides, temperatures in excess of 600 • C are needed for adequate ODH reactivity. However, higher reaction temperatures are less favorable for high olefin selectivity. Therefore, reducible transitional metal oxide catalysts are considered instead. This catalyst group can activate the paraffins at low temperatures. However, while catalytic activity is usually higher with this class of catalysts, lower alkene selectivities are frequently found. The oxides of these metals contain removable oxygen (lattice oxygen), which participates in the ODH reactions under oxygen-free atmosphere conditions. However, lattice oxygen is also involved in the un-selective pathways of ODH, forming carbon oxides. Therefore, different catalyst groups can be studied to find the optimum catalyst that will provide high selectivity of olefins [10,12,27,54,55,59,63,81].
Vanadium is the most frequently considered element for ethane and propane ODH. In particular, the VO x shows promising results for olefin production under an oxygen-free environment [12,54,59,61]. VO x has a suitable atomic geometry and electronic structure. Furthermore, the V 4+ and V 5+ valence states make vanadium valuable for many catalytic reactions [82]. Moreover, supported vanadium oxides have attracted significant attention due to their higher performances, better thermal stabilities, and large specific surface areas.
In this respect, the stoichiometric equation for the conversion of propane ODH to propylene over vanadium-based catalysts is as follows: This reduced vanadium species is regenerated via molecular oxygen following the stoichiometry described below: The redox behavior of supported vanadia catalysts in ODH is generally controlled by three main factors: (i) the VO x surface structure; (ii) the acid-base character of the metal and the support; and, (iii) the redox properties of the VO x species. These three properties are influenced by the support type and vanadium loading [83,84].

VO x Surface Coverage
The molecular structures of surface vanadia species on metal oxide supports have been reported in the technical literature [85][86][87][88]. These studies suggest that, depending on the vanadium loading on the support, four kinds of VO x surface species can be present on the catalyst surface: (a) isolated VO 4 species (monovanadate); (b) polymeric VO 4 species (polyvanadates); (c) a mix of both isolated and polymeric VO 4 surface species; and, (d) V 2 O 5 crystals. Some studies [83,[88][89][90][91] showed that, at low vanadium loading, a highly dispersed isolated VO 4 surface species (monovanadates) is formed. As the VO x surface density increases with vanadium loading, the surface configuration evolves from isolated monovanadates to polymeric polyvanadates. In this respect, it appears that polymeric polyvanadate dominates until a vanadia monolayer surface coverage is reached. However, at high vanadium loadings, crystalline V 2 O 5 nanoparticles form on top of the vanadia monolayer. It is generally accepted that isolated tetrahedral VO x species (which are obtained at low vanadium loading) are more ODH selective but less active than polymeric VO x species [88][89][90].

Active Lattice Oxygen Species
Other than the ODH catalyst structure, the binding strength of the surface lattice oxygen in the VO x surface species is a main parameter that governs the activities and selectivities of alumina-supported vanadia catalysts. In extensive structural studies [87,92,93] of supported vanadium oxide catalysts, three types of lattice oxygen bonds were identified ( The studies were aimed at determining which type of lattice oxygen bond is responsible for the oxidation activity, which occurs in various catalytic oxidation reactions [92,94]. It was determined that the oxygen in the V-O-support bond, rather than the terminal V=O or the V-O-V bonds, is the one that is involved in this catalytic oxidation reaction.

Effect of Support (Acid/Base Properties)
Metal oxides are composed of redox metal cations and lattice oxygen anions. These are Lewis acid and basic sites, respectively. Thus, the acid-base support characteristics may contribute with several effects, namely: (a) the dispersion of the active species; (b) the specific structure of the active species; (c) the reactant molecule activation; (d) the rates of competitive pathways of transformation; and, (e) the rates of adsorption and desorption of reactants and products.
Furthermore, the acid-base support can influence the vanadium-based catalyst reactivity or olefin selectivity [83,[93][94][95][96]. For instance, the interaction between the acidic V2O5 species and a basic support might be strong (e.g., MgO, La2O3, Sm2O3). This might lead to the formation of highly dispersed VOx species, which is responsible for the high alkene selectivities. In contrast, the interaction between the acidic V2O5 and an acidic support (e.g., SiO2, Al2O3) may be weak. This might lead to a less dispersed vanadium species, which favors the formation of a less active V2O5 crystalline phase [95,97,98].
Moreover, on acidic catalysts, basic reactant adsorption and acidic product desorption are favored. Thereby, this protects the chemical species from further oxidation to carbon oxides. For instance, higher selectivities to ethylene have been obtained using acidic based catalysts, like the VOx/γ-Al2O3 catalysts. These findings were justified, given that higher support acidity decreases the interaction between the ethylene product and the catalyst [55,97]. In addition, dominant Brønsted acid sites are considered to be desirable to facilitate rapid olefin desorption and limit the complete alkane oxidation to COx. Thus, catalyst acidity and acid site type have to be kept at acceptable levels. This is required, so as to not interfere with the overall catalyst activity and catalyst olefin selectivity [96].

Redox Properties of Supported VOx Catalysts
The reducibility of vanadium oxides is considered to be one of the main factors influencing their activity in partial oxidation reactions and, in particular, in the ODH reactions of alkanes. There is, in fact, a close relationship between catalyst reducibility and VOx surface structure on a given metal oxide support. In this respect, the reducibility of the surface VOx species increases with surface VOx coverage. Thus, the following trend for the reducibility of the different supported vanadia species can be considered as follows: polymeric surface VOx > isolated surface VOx > crystalline V2O5 nanoparticles [83,96,99]. Moreover, the type of support used affects the extent of the reducibility of supported vanadium oxide catalysts.
The catalyst support acid-base character also affects redox properties. One can observe that there is decreasing reducibility of the vanadium species when using more basic support oxides [100]. Moreover, the propane ODH reaction runs conducted at 450-550 °C showed the following: (a) the V2O5/TiO2 catalyst, which is less basic and easier to reduce, is the most active catalyst, (b) the

Effect of Support (Acid/Base Properties)
Metal oxides are composed of redox metal cations and lattice oxygen anions. These are Lewis acid and basic sites, respectively. Thus, the acid-base support characteristics may contribute with several effects, namely: (a) the dispersion of the active species; (b) the specific structure of the active species; (c) the reactant molecule activation; (d) the rates of competitive pathways of transformation; and, (e) the rates of adsorption and desorption of reactants and products.
Furthermore, the acid-base support can influence the vanadium-based catalyst reactivity or olefin selectivity [83,[93][94][95][96]. For instance, the interaction between the acidic V 2 O 5 species and a basic support might be strong (e.g., MgO, La 2 O 3 , Sm 2 O 3 ). This might lead to the formation of highly dispersed VO x species, which is responsible for the high alkene selectivities. In contrast, the interaction between the acidic V 2 O 5 and an acidic support (e.g., SiO 2 , Al 2 O 3 ) may be weak. This might lead to a less dispersed vanadium species, which favors the formation of a less active V 2 O 5 crystalline phase [95,97,98].
Moreover, on acidic catalysts, basic reactant adsorption and acidic product desorption are favored. Thereby, this protects the chemical species from further oxidation to carbon oxides. For instance, higher selectivities to ethylene have been obtained using acidic based catalysts, like the VO x /γ-Al 2 O 3 catalysts. These findings were justified, given that higher support acidity decreases the interaction between the ethylene product and the catalyst [55,97]. In addition, dominant Brønsted acid sites are considered to be desirable to facilitate rapid olefin desorption and limit the complete alkane oxidation to CO x . Thus, catalyst acidity and acid site type have to be kept at acceptable levels. This is required, so as to not interfere with the overall catalyst activity and catalyst olefin selectivity [96].

Redox Properties of Supported VO x Catalysts
The reducibility of vanadium oxides is considered to be one of the main factors influencing their activity in partial oxidation reactions and, in particular, in the ODH reactions of alkanes. There is, in fact, a close relationship between catalyst reducibility and VO x surface structure on a given metal oxide support. In this respect, the reducibility of the surface VO x species increases with surface VO x coverage. Thus, the following trend for the reducibility of the different supported vanadia species can be considered as follows: polymeric surface VO x > isolated surface VO x > crystalline V 2 O 5 nanoparticles [83,96,99]. Moreover, the type of support used affects the extent of the reducibility of supported vanadium oxide catalysts.
The catalyst support acid-base character also affects redox properties. One can observe that there is decreasing reducibility of the vanadium species when using more basic support oxides [100]. Moreover, the propane ODH reaction runs conducted at 450-550 • C showed the following: (a) the V 2 O 5 /TiO 2 catalyst, which is less basic and easier to reduce, is the most active catalyst, (b) the V 2 O 5 /Al 2 O 3 catalyst, which is more acidic and difficult to reduce, is the most selective in propylene production. These variations of vanadia oxide reducibilities on different metal oxide supports could be related to the different reducibilities of the various V-O-support bonds existing on different support types [83,97,100].
Thus, one can conclude that the catalytic activities and selectivities of the supported vanadium oxide catalysts are significantly affected by the properties of the support oxide material, the interaction of the surface VO x species with the oxide support, and the vanadium loading. The redox sites are in charge of transferring the lattice oxygen to the adsorbed propane to form propylene. The acid sites catalyze the condensation of the intermediates. Therefore, it is necessary to investigate the synergistic effects of the redox and acidic properties of a catalyst as well as their dependence on the catalyst composition and reaction conditions to enhance the efficiency of such a process.

Propane ODH Catalysts
To date, several vanadium-based catalysts have been studied for PODH. In these cases, the only oxygen source was the catalyst lattice oxygen. This research has been led in the last few years by Prof. de Lasa's research team at the Chemical Reactor Engineering Centre (CREC), University of Western Ontario Canada [12,55,101,102].
Rostom et al. [59,60] established a mixture of ZrO 2 and γAl 2 O 3 (1:1 weight ratio) as a support for a vanadium-based catalyst. These authors loaded different percentages of vanadium (5, 7.5, and 10 wt. % V) on this support and achieved the best result with 7.5 wt. % vanadium loading. So far, this is the best-reported propane conversion (25%) and propylene selectivity (94%) combination with a negligible amount of CO x selectivity (2.1%) at 550 • C. The same support was utilized by researchers [54,56,75] for ethane ODH in the temperature range of 525-600 • C and with 20-50 s contact times. They achieved 82% ethylene selectivity at 8.5% ethane conversion.
Hossain et al. [10,61] developed a vanadium-based CaO-γAl 2 O 3 support for propane ODH. They prepared CaO to γAl 2 O 3 weight ratios of 1:4 and 1:1 and achieved the best results with the 1:1 weight ratio. They performed ODH at a 550-640 • C temperature range and achieved 25.5% propane conversion, 94.2% propylene selectivity, and 5.8% CO x selectivity at 640 • C. This high temperature had a negative impact on the ODH reaction.
Ghamdi et al. [12] reported different PODH catalysts prepared with various vanadium loadings on γAl 2 O 3 (5, 7, and 10 wt. % V) for propane ODH at 475-550 • C and reaction times of 5-20 sec. The re-adsorption of the propylene product was assigned to methane, ethane, and ethylene cracking, as γ-Al 2 O 3 is acidic in character. Therefore, the maximum selectivity of propylene that these authors achieved was 85.94% at 11.73% propane conversion. The rest of the products were CO x , CH 4 , C 2 H 4 , and C 2 H 6 .
Fukudome et al. [62,63] incorporated VO x species into a SiO 2 -frame to obtain a higher concentration of isolated VO x species. This catalyst was synthesized by an alkoxy exchange between a metal alkoxide and polyethylene glycol. Following this, PODH using a VO x lattice oxygen was performed in a fixed-bed reactor at a 450 • C temperature under atmospheric pressure. These researchers found that VO x that was incorporated into the SiO 2 showed higher propylene selectivity than VO x loaded onto SiO 2 . This could be ascribed to the isolated VO 4 3species in the silica framework. Table 3 reports the various reaction conditions and experimental results when using vanadium-based catalysts on different supports, where catalysts lattice oxygen is the only oxygen source.

Experimental Laboratory Reactors for PODH
The CREC Riser Simulator [103,104], as described in Figure 3, is a bench-scale mini-fluidized bed reactor (53 cm 3 ). It provides conditions that are equivalent to those of a twin circulating fluidized reactor process (reactor-regenerator). CREC researchers have already demonstrated promising ethane conversions and ethylene selectivities using this reactor. Here, a VOx/γ-Al2O3 catalyst is utilized in the absence of molecular gas oxygen. In the CREC Riser Simulator, VOx/c-Al2O3 [84], VOx-MoOx/γ-Al2O3 [3], VOx-Nb/La-γ-Al2O3 [9], and VOx/γ-Al2O3-ZrO2 [54,56,76] catalysts were used for ethane ODH. VOx/γ- It provides conditions that are equivalent to those of a twin circulating fluidized reactor process (reactor-regenerator). CREC researchers have already demonstrated promising ethane conversions and ethylene selectivities using this reactor. Here, a VO x /γ-Al 2 O 3 catalyst is utilized in the absence of molecular gas oxygen. In the CREC Riser Simulator, VO x /c-Al 2 O 3 [84], VO x -MoO x /γ-Al 2 O 3 [3], VO x -Nb/La-γ-Al 2 O 3 [9], and VO x /γ-Al 2 O 3 -ZrO 2 [54,56,76] catalysts were used for ethane ODH. VO x /γ-Al 2 O 3 [12], VO x /CaO-γ-Al 2 O 3 [10,61], and VO x /γ-Al 2 O 3 -ZrO 2 [59,60] catalysts were used in propane ODH in this reactor in the absence of gas phase oxygen. This reactor operates under batch conditions and it is designed for catalyst evaluation and kinetic studies under fluidized bed (riser/downer) reactor conditions. One of the main advantages of this unit is its capability to simulate fluidized bed reactions conditions by using a very small amount of catalyst. Details of this reactor are found in literature [12]. Figure 4a-c describe typical PODH pressure profiles in the CREC Riser Simulator during and after the reaction. Initially, as propane is being fed via a syringe into the reactor at atmospheric pressure, the reactor pressure increases significantly. Following this, the reactor pressure is further augmented, while the PODH reaction is progressing. Finally, and a preset time, the unreacted feed and the products are transferred almost instantaneously to a vacuum box, with the reactor pressure decreasing abruptly. Regarding the studies in the CREC Riser Simulator, two modes of operation were considered: (a) single propane injections with catalyst regeneration in between, and (b) ten consecutive propane injections without catalyst regeneration in between. These types of possible PODH operations are discussed in the upcoming sections.

Single Propane Injection
A single propane injection in the CREC Riser Simulator allows for one to understand the interaction between the alkane feed and a fully oxidized catalyst. Thus, in these experiments, the catalyst is repeatedly reduced when reacting with the alkane and then re-oxidized by air at various reaction temperatures and contact times. Following every reaction injection, the catalyst is regenerated with air. Therefore, every time the catalyst is regenerated, the PODH is mainly driven by non-selective oxygen species on the catalyst surface. This could involve loosely bound lattice oxygen or weakly adsorbed oxygen species [59,61]. This loosely bound lattice oxygen is considered to be more reactive and, thus, more likely to cause carbon-carbon bond breakage, promoting propane total oxidation. As a result, both types of oxygen species may contribute to the total oxidation of propane and lead to low propylene selectivity. Therefore, it can be concluded that fully oxidized (fresh) catalysts are active, but not selective, in propane ODH reactions [12]. In addition, single injections followed by catalyst re-oxidation show the value of using a catalyst with an optimized oxygen state, where the density of the oxygen species on the catalyst surface is controlled. This is significant in achieving higher propylene selectivity. Regarding the studies in the CREC Riser Simulator, two modes of operation were considered: (a) single propane injections with catalyst regeneration in between, and (b) ten consecutive propane injections without catalyst regeneration in between. These types of possible PODH operations are discussed in the upcoming sections.

Single Propane Injection
A single propane injection in the CREC Riser Simulator allows for one to understand the interaction between the alkane feed and a fully oxidized catalyst. Thus, in these experiments, the catalyst is repeatedly reduced when reacting with the alkane and then re-oxidized by air at various reaction temperatures and contact times. Following every reaction injection, the catalyst is regenerated with air. Therefore, every time the catalyst is regenerated, the PODH is mainly driven by non-selective oxygen species on the catalyst surface. This could involve loosely bound lattice oxygen or weakly adsorbed oxygen species [59,61]. This loosely bound lattice oxygen is considered to be more reactive and, thus, more likely to cause carbon-carbon bond breakage, promoting propane total oxidation. As a result, both types of oxygen species may contribute to the total oxidation of propane and lead to low propylene selectivity. Therefore, it can be concluded that fully oxidized (fresh) catalysts are active, but not selective, in propane ODH reactions [12]. In addition, single injections followed by catalyst re-oxidation show the value of using a catalyst with an optimized oxygen state, where the density of the oxygen species on the catalyst surface is controlled. This is significant in achieving higher propylene selectivity.

Successive Propane Injections
Successive propane injections can be implemented to overcome the problem of single injection experiments where low propylene selectivity is an issue, as described in Figure 4a-c. Here, the PODH catalyst is progressively reduced via the consecutive alkane injections and, therefore, propylene selectivity gradually increases. There is, as mentioned, no catalyst regeneration in between propane injections.
With these data and for each of the injections, the instantaneous conversions and selectivities for the main products can be calculated. Furthermore, the degree of reduction of the catalyst can be defined as the ratio of the remaining oxygen in the catalyst after each injection to the original oxygen content of the catalyst. The former can be determined by analyzing the various oxygen-containing products resulting from each alkane injection. The latter can be calculated from the oxygen uptake of the O 2 -chemisorption characterization technique.
In this respect, researchers reported high propylene selectivity and good propane conversion while following the second and third successive injection [12,59,61]. This is the case, given that successive injections deplete quickly weakly adsorbed oxygen species. Furthermore, only then, the catalyst lattice oxygen drives the PODH reaction. For instance, a VO x /ZrO 2 -γAl 2 O 3 catalyst that was developed at CREC-UWO and used with multiple injections yielded 25% propane conversion with 94% selectivities [60]. Figures 5 and 6 report propane conversion, and propylene and carbon oxide selectivity, respectively, with every injection. Initially, due to the presence of surface oxygen, propane conversion and CO x selectivity are high with a very low propylene selectivity. After the first injection, surface oxygen decreases and lattice oxygen starts contributing to PODH, as shown in Appendix A. There is a drop in propane conversion and CO x selectivity, and a rise in propylene selectivity. From the third to the tenth injections, propane conversion, propylene, and CO x selectivities are almost constant due to the effect of the lattice oxygen.
Catalysts 2020, 10, x FOR PEER REVIEW 13 of 24 oxygen state, where the density of the oxygen species on the catalyst surface is controlled. This is significant in achieving higher propylene selectivity.

Successive Propane Injections
Successive propane injections can be implemented to overcome the problem of single injection experiments where low propylene selectivity is an issue, as described in Figure 4a-c. Here, the PODH catalyst is progressively reduced via the consecutive alkane injections and, therefore, propylene selectivity gradually increases. There is, as mentioned, no catalyst regeneration in between propane injections.
With these data and for each of the injections, the instantaneous conversions and selectivities for the main products can be calculated. Furthermore, the degree of reduction of the catalyst can be defined as the ratio of the remaining oxygen in the catalyst after each injection to the original oxygen content of the catalyst. The former can be determined by analyzing the various oxygen-containing products resulting from each alkane injection. The latter can be calculated from the oxygen uptake of the O2-chemisorption characterization technique.
In this respect, researchers reported high propylene selectivity and good propane conversion while following the second and third successive injection [12,59,61]. This is the case, given that successive injections deplete quickly weakly adsorbed oxygen species. Furthermore, only then, the catalyst lattice oxygen drives the PODH reaction. For instance, a VOx/ZrO2-γAl2O3 catalyst that was developed at CREC-UWO and used with multiple injections yielded 25% propane conversion with 94% selectivities [60]. Figures 5 and 6 report propane conversion, and propylene and carbon oxide selectivity, respectively, with every injection. Initially, due to the presence of surface oxygen, propane conversion and COx selectivity are high with a very low propylene selectivity. After the first injection, surface oxygen decreases and lattice oxygen starts contributing to PODH, as shown in Appendix A. There is a drop in propane conversion and COx selectivity, and a rise in propylene selectivity. From the third to the tenth injections, propane conversion, propylene, and COx selectivities are almost constant due to the effect of the lattice oxygen.

Kinetics and Reaction Mechanisms of Propane ODH over Vanadium-Based Catalysts
Baerns et al. reported [105] the mechanism of the initial activation of the free radicals in the alkane ODH reaction over transition metal oxides (i.e., VOx, MoOx etc.). According to this, oxygen from the metal oxide extracts hydrogen from the alkane. The OH groups formed are then removed from the catalyst surface by dehydration. Thus, the catalyst surface is reduced in the formation of propylene and the products of total oxidation (COx). Thus, and as a result, the catalyst has to be subsequently re-oxidized by gas phase oxygen. This type of mechanism is known as the ODH redoxmechanism, as described by Equation (36 PODH has been extensively studied using vanadium-based catalysts. This has been done in order to understand the selective pathways for propylene production [101,[106][107][108][109][110]. All of these studies revealed that PODH consists of a set of consecutive and parallel reactions: (a) the ODH of propane; (b) the undesired combustion of propane to COx; and, (c) the secondary combustion of propylene to COx (Figure 7). The two latter reactions can limit the propylene selectivity and the propylene yield during PODH. Furthermore, it is generally agreed that, as propane conversion increases, selectivity towards the desired propylene decreases. Therefore, a significant fraction of the propane and/or propylene is unselectively converted into carbon oxides.
Thus, the major challenge of PODH is to control the propylene selectivity, as the alkenes formed during ODH tend to be combusted via a secondary reaction.

Kinetics and Reaction Mechanisms of Propane ODH over Vanadium-Based Catalysts
Baerns et al. reported [105] the mechanism of the initial activation of the free radicals in the alkane ODH reaction over transition metal oxides (i.e., VO x , MoO x etc.). According to this, oxygen from the metal oxide extracts hydrogen from the alkane. The OH groups formed are then removed from the catalyst surface by dehydration. Thus, the catalyst surface is reduced in the formation of propylene and the products of total oxidation (COx). Thus, and as a result, the catalyst has to be subsequently re-oxidized by gas phase oxygen. This type of mechanism is known as the ODH redox-mechanism, as described by Equation (34).
PODH has been extensively studied using vanadium-based catalysts. This has been done in order to understand the selective pathways for propylene production [101,[106][107][108][109][110]. All of these studies revealed that PODH consists of a set of consecutive and parallel reactions: (a) the ODH of propane; (b) the undesired combustion of propane to CO x ; and, (c) the secondary combustion of propylene to CO x (Figure 7). The two latter reactions can limit the propylene selectivity and the propylene yield during PODH. Furthermore, it is generally agreed that, as propane conversion increases, selectivity towards the desired propylene decreases. Therefore, a significant fraction of the propane and/or propylene is unselectively converted into carbon oxides.

Kinetics and Reaction Mechanisms of Propane ODH over Vanadium-Based Catalysts
Baerns et al. reported [105] the mechanism of the initial activation of the free radicals in the alkane ODH reaction over transition metal oxides (i.e., VOx, MoOx etc.). According to this, oxygen from the metal oxide extracts hydrogen from the alkane. The OH groups formed are then removed from the catalyst surface by dehydration. Thus, the catalyst surface is reduced in the formation of propylene and the products of total oxidation (COx). Thus, and as a result, the catalyst has to be subsequently re-oxidized by gas phase oxygen. This type of mechanism is known as the ODH redoxmechanism, as described by Equation (36).
PODH has been extensively studied using vanadium-based catalysts. This has been done in order to understand the selective pathways for propylene production [101,[106][107][108][109][110]. All of these studies revealed that PODH consists of a set of consecutive and parallel reactions: (a) the ODH of propane; (b) the undesired combustion of propane to COx; and, (c) the secondary combustion of propylene to COx (Figure 7). The two latter reactions can limit the propylene selectivity and the propylene yield during PODH. Furthermore, it is generally agreed that, as propane conversion increases, selectivity towards the desired propylene decreases. Therefore, a significant fraction of the propane and/or propylene is unselectively converted into carbon oxides.
Thus, the major challenge of PODH is to control the propylene selectivity, as the alkenes formed during ODH tend to be combusted via a secondary reaction.  Thus, the major challenge of PODH is to control the propylene selectivity, as the alkenes formed during ODH tend to be combusted via a secondary reaction.
Recent literature has reported the feasibility of the Mars van Krevelen [101,111] and Langmuir Hinshelwood [60,61] mechanisms for gas phase oxygen-free PODH. It was hypothesized that the dominant reaction mechanism for catalytic alkane ODH involves transition metal oxides. The reaction steps include catalyst lattice oxide being reduced by adsorbed alkanes. In this case, olefins are formed through several possible intermediate species. Gas phase molecular oxygen can then re-oxidize the reduced catalyst sites.
Rostom et al. [60] reported the feasibility of the Langmuir-Hinshelwood mechanism for PODH kinetic modeling under oxygen-free conditions. The adsorption constants were independently defined through experiments in the CREC Riser Simulator. These authors found that the high propylene selectivity can be justified given the much larger 2.82 × 10 −5 mol.gcat −1 s −1 frequency factor for propylene formation versus the 1.65 × 10 −6 mol.gcat −1 s −1 frequency factor for propane combustion. However, calculated energies of activation (55.7 kJ/mol for propylene formation and 33.3 kJ/mol for propane combustion) appear to moderate this favorable effect on propylene selectivity, with the influence of frequency factors prevailing. Furthermore, propylene conversion into CO x oxidation appears as a non-favored reaction step given the 98.5 kJ/mol activation energy and 4.80 × 10 −6 mol.gcat −1 s −1 frequency factor. This kinetic model is considered of special value for the further development of a scaled-up twin fluidized bed reactor configuration for PODH.
Hossain et al. [61] compared the kinetics of the oxidative dehydrogenation (ODH) of propane to those of propylene over VO x /CaO and VO x /CaO-γAl 2 O 3 catalysts in the absence of gas phase oxygen through the Langmuir−Hinshelwood mechanism. These researchers found that propylene formation using a VO x /CaO-γAl 2 O 3 catalyst has a lower activation energy (120.3 kJ/mol) than the one when using the VO x /CaO (126.7 kJ/mol) catalyst. In contrast, the VO x /CaO-γAl 2 O 3 catalyst requires higher activation energies (55.2 kJ/mol) than the VO x /CaO catalyst (32.8 kJ/mol), in order to prevent the undesired CO 2 formation. These values are consistent with the product selectivity, as observed in the catalyst evaluation experiments.
Ghamdi et al. [101] reported the kinetic modeling of propane oxidative dehydrogenation (ODH) via the Mars van Krevelen mechanism. The calculated pre-exponential factors (k 1 0 , k 2 0 , and k 3 0 ) increased as the vanadium loading was augmented. In addition, the activation energies for CO x formation (E 2 and E 3 ) were consistently smaller than the ones for propylene formation (E 1 ). Table 4 displays the activation energy values for propane ODH reactions under gas phase oxygen-free conditions, as reported in the technical literature.

Effect of Reaction Temperature
Ghamdi et al. reported the temperature effect on ODH experiments over partially reduced catalysts [12]. During successive injection ODH experiments, propylene selectivity showed a gradual increase with increasing reaction temperature. This is attributed to the variation of the catalyst degree of reduction with both reaction temperature and the number of propane injections. At higher degrees of reduction of the catalysts, the ODH selective pathway is preferred over that for propane combustion. CO x formation is low due to the lower availability of lattice oxygen. Rostom et al. also reported similar behavior [60].
Temperature increases beyond 650 • C may result in a significant reduction of propylene yield due to the higher influence of unselective thermal reactions [112]. Thus, thermal levels above 650 • C for PODH are not recommended. Furthermore, a certain extent of surface reduction occurs at temperatures below 650 • C, which is recommended for high propylene selectivity.

Effect of Reaction Time
Increasing reaction time augments propane conversion, enhances propylene selectivity, and decreases CO x selectivity [12,60]. In this respect, longer contact times yield higher lattice oxygen utilization, favoring propylene formation and decreasing CO x . However, very long reaction times are not favorable, as they may allow for the products (i.e., propylene) to re-adsorb on the catalyst surface, lowering propylene selectivity and increasing CO x selectivity. Furthermore, long reaction times (more than 20 s) are not adequate for industrial use, given that they would require additional lengths of downflow fluidized beds to be included in the reactor system.

Reactor Concepts for PODH
Adsorption and desorption during PODH reactions are governed by phase residence times, chemical species interaction with each other, and with the catalyst. The reactor type and the operation mode, in turn, determine these factors. Efficient heat removal from the PODH process is a major concern for reactor operation, given that PODH is an exothermic reaction.

Fixed-Bed Reactors
Most of the ODH literature references consider fixed-bed type reactors, mainly due to their perceived operational simplicity [8,40,74,[113][114][115]. For instance, temperature gradients can hardly be eliminated in a traditional fixed-bed reactor unless the catalyst bed is diluted with inert particles while using a large inert particle/catalyst particle ratio. This greatly increases the fixed cost and it is a significant challenge for the economical industrial scale production of propylene that is based on the PODH process.
Researchers analyzed multi-tubular reactors with periodic air injection [116], membrane reactors [117], and wall-cooled catalytic reactors [118] to overcome the exothermicity of ODH fixed-bed reactors. The aim was to efficiently remove the generated heat from the catalyst bed. A micro-channel reactor was utilized to reduce the fixed-bed reactor cost [21]. It is reported that the micro-channel reactor can achieve the same reactor productivity as a traditional fixed-bed reactor with less than 20% of the fixed-bed reactor volume. This is the case, given the inherent suitability of micro-channel reactors for highly exothermic reactions, due to their excellent heat transfer and heat removal capabilities.
Recently, the performances of phase-pure M1 MoVNbTeO x catalysts in the ODHE reaction in both a micro-channel reactor and a small-sized fixed-bed reactor, under same conditions, were investigated [21]. This comparison was carried out to demonstrate the advantages of the micro-channel reactors for improved heat management. XRD, SEM, and ICP characterization indicated that the M1-PVA catalyst plate has a high stability in the micro-channel system.

Twin Circulating Fluidized Bed Reactors for PODH
Circulating fluidized bed reactors and, more specifically downer reactors for ODH, have received much attention recently due to the several shortcomings of the fixed-bed reactors [3,60,102]. Downer reactors, if operated with fine particles in the 60-100 micron range, provide controlled thermal change, adequate particle concentration, and residence times, allowing an ODH with high olefin selectivity [67].
In ODH, under oxygen-free atmospheres and once the surface lattice oxygen has been exhausted, the rate of water formation decreases. This occurs with a gradual increase in the formation of molecular H 2 , which, in turn, slows down the ODH reaction. Therefore, catalyst re-oxidation (regeneration) is necessary for adequate catalyst activity [45,61,96,116,119,120]. Thus, in ODH, periodic catalyst re-oxidations are required and, as a result, the ODH process can be viewed as a system of twin fluidized reactors: an oxidative dehydrogenation reactor and a re-oxidation reactor [12,59,61] (Figure 8). In these ODH systems, the gas phase oxygen is never allowed to reach the oxidative dehydrogenation unit. This limits the possibility of the complete combustion of the propane feed and propylene product.
Catalysts 2020, 10, x FOR PEER REVIEW 17 of 24 of molecular H2, which, in turn, slows down the ODH reaction. Therefore, catalyst re-oxidation (regeneration) is necessary for adequate catalyst activity [45,61,96,116,119,120]. Thus, in ODH, periodic catalyst re-oxidations are required and, as a result, the ODH process can be viewed as a system of twin fluidized reactors: an oxidative dehydrogenation reactor and a re-oxidation reactor [12,59,61] (Figure 8). In these ODH systems, the gas phase oxygen is never allowed to reach the oxidative dehydrogenation unit. This limits the possibility of the complete combustion of the propane feed and propylene product. Thus, a remaining outstanding challenge for ODH is to have available fluidizable catalysts with high selectivity towards propylene formation and be able to supply oxygen for dehydrogenation. Therefore, the most important characteristics of a successful ODH catalyst are its activity in the ODH reactor and its reactivity in the re-oxidation cycle. In addition, the fluidizable ODH catalysts should: (a) be stable under repeated reduction/oxidation cycles, (b) be resistant to agglomeration, (c) be able to withstand the friction stress that is associated with the high circulation of particles, and (d) be environmentally benign and affordable.

Circulating Fluidized Bed Reactor Models for PODH
The technical literature reports a mathematical model for PODH. The model was based on a two-phase fluidized bed reactor representation [68]. The model used the kinetics for a V-Mg oxide catalyst. Simulations showed that temperature, pressure, feed composition, particle size, and gas velocity were all factors affecting propane conversion and propylene selectivity. The model was validated while using previously published experimental data from a circulating fluidized bed reactor unit.
Recently, Rostom et al. [121] reported a downer fluidized bed reactor simulation using a hybrid CPFD Barracuda software, which represents the catalyst particles as particle clusters. The kinetics established using experimental data were incorporated in the simulation. The data showed good propane conversion and propylene selectivity prospects for an industrial scale unit. Thus, a remaining outstanding challenge for ODH is to have available fluidizable catalysts with high selectivity towards propylene formation and be able to supply oxygen for dehydrogenation. Therefore, the most important characteristics of a successful ODH catalyst are its activity in the ODH reactor and its reactivity in the re-oxidation cycle. In addition, the fluidizable ODH catalysts should: (a) be stable under repeated reduction/oxidation cycles, (b) be resistant to agglomeration, (c) be able to withstand the friction stress that is associated with the high circulation of particles, and (d) be environmentally benign and affordable.

Circulating Fluidized Bed Reactor Models for PODH
The technical literature reports a mathematical model for PODH. The model was based on a two-phase fluidized bed reactor representation [68]. The model used the kinetics for a V-Mg oxide catalyst. Simulations showed that temperature, pressure, feed composition, particle size, and gas velocity were all factors affecting propane conversion and propylene selectivity. The model was validated while using previously published experimental data from a circulating fluidized bed reactor unit.
Recently, Rostom et al. [121] reported a downer fluidized bed reactor simulation using a hybrid CPFD Barracuda software, which represents the catalyst particles as particle clusters. The kinetics established using experimental data were incorporated in the simulation. The data showed good propane conversion and propylene selectivity prospects for an industrial scale unit.

PODH Industrial Prospects
ODH under oxygen-free atmosphere, with the source of oxygen being the catalyst lattice oxygen, offers unique advantages. Catalyst lattice oxygen, instead of gas phase molecular oxygen or other mild oxidants, prevents deep alkane oxidation, thus limiting CO x formation. This catalytic reaction leads to higher olefin selectivity, securing a much more energy efficient route for olefin production. To accomplish this, Rostom et al. [121] recently reported an industrial PODH process, where the PODH catalyst is circulated between a PODH reactor and a catalyst regenerator unit. After leaving the PODH reactor, the catalyst is divided in a splitter. Using this device, a major portion of the catalyst is recirculated back to the PODH reactor with a smaller fraction going to the regenerator (1/10 times) to be reactivated via re-oxidation. In this way, a partially reduced catalyst is always maintained, circulating in between the systems. This special reactor configuration helps to overcome the negative influence of combustion reactions, increasing as a result of overall propylene selectivity. It is envisioned that the industrial scale downer reactor for PODH will involve a 20 m length with 2.8-3.5 m/s particle cluster velocities and contact times in the 5-7 s range.

Conclusions
(a) The oxidative dehydrogenation of propane using vanadium-based catalysts offers valuable prospects for olefin production.
(b) In particular, vanadium-based catalysts supported on variety of fluidizable supports can be utilized to develop a suitable propane oxy-dehydrogenation process. These fluidizable catalysts can supply the lattice oxygen that is required for high olefin selectivity.
(c) A variety of reaction mechanisms and kinetic models can be considered for PODH. However, if one aims to develop a PODH leading to high propylene selectivity, special experimental devices, such as the CREC Riser Simulator, must be used.
(d) This approach provides kinetic models that will represent the PODH, under the conditions of successive propane injections, which are identified as being the most favorable for high propylene selectivity.
(e) An appropriate reactor selection for PODH is an important aspect to consider when establishing PODH at the commercial scale. In this respect, special reactor configurations involving downer reactors with partial catalyst re-oxidation are favored to achieve high propylene selectivities in a continuous process. Application of this methodology showed a progressive increasing O min,RS conversion reaching in injection 10 a 72.9% value as shown in Figure A1. One should notice that under these conditions both high C 3 H 6 and low CO 2 selectivity were observed and adequate O-S availability until the last tenth injection was shown. Application of this methodology showed a progressive increasing Omin,RS conversion reaching in injection 10 a 72.9% value as shown in Figure A1. One should notice that under these conditions both high C3H6 and low CO2 selectivity were observed and adequate O-S availability until the last tenth injection was shown.