Bismuth molybdate propylene reactions

Another industrially important reaction of propylene, related to the one above, is its partial oxidation in the presence of ammonia, resulting in acrylonitrile, H2C=CHCN. This ammoxidation reaction is also catalyzed by mixed metal oxide catalysts, such as bismuth-molybdate or iron antimonate, to which a large number of promoters is added (Fig. 9.19). Being strongly exothermic, ammoxidation is carried out in a fluidized-bed reactor to enable sufficient heat transfer and temperature control (400-500 °C). 

In the case of selective oxidation catalysis, the use of spectroscopy has provided critical Information about surface and solid state mechanisms. As Is well known( ), some of the most effective catalysts for selective oxidation of olefins are those based on bismuth molybdates. The Industrial significance of these catalysts stems from their unique ability to oxidize propylene and ammonia to acrylonitrile at high selectivity. Several key features of the surface mechanism of this catalytic process have recently been descrlbed(3-A). However, an understanding of the solid state transformations which occur on the catalyst surface or within the catalyst bulk under reaction conditions can only be deduced Indirectly by traditional probe molecule approaches. Direct Insights Into catalyst dynamics require the use of techniques which can probe the solid directly, preferably under reaction conditions. We have, therefore, examined several catalytlcally Important surface and solid state processes of bismuth molybdate based catalysts using multiple spectroscopic techniques Including Raman and Infrared spectroscopies, x-ray and neutron diffraction, and photoelectron spectroscopy. 

Figure 11.26 Plot of the position sensitivity of the degree of conversion for a set of 48 bismuth-molybdate catalysts (same batch) in propylene to acrolein conversion in a Stage II 48-fold-screening reactor (reaction conditions 2% hydrocarbon in air at GHSV of 3000 h-1, column no. 8 contains only inert carrier material).

Three well known examples of processes employing fluidised-bed operations are the oxidations of naphthalene and xylene to phthalic anhydride using a supported V2O5 catalyst and ammoxidation of propylene utilising a mixed oxide composition containing bismuth molybdate. Typically, this latter reaction is executed by passing a mixture of ammonia, air and propylene to a fluidised bed operating at about 0.2 MPa pressure, 400—500°C and a few seconds contact time between gas and fluidised catalyst peirticles. 

Acrolein Production. Adams et al. [/. Catalysis, 3,379 (1964)] studied the catalytic oxidation of propylene on bismuth molybdate catalyst to form acrolein. With a feed of propylene and oxygen and reaction at 460°C, the following three reactions occur. 

Mechanisms There is a derth of knowledge about the mechanisms operative in selective oxidation reactions. The only exceptions are the reactions of ethylene to ethylene oxide on supported silver catalysts and of propylene to acrolein on bismuth molybdate type catalysts. For the latter, it is well established through isotopic labeling experiments that a symmetric allyl radical is an intermediate in the reaction and that its formation is rate-determining. Many studies simply extrapolate the results substantiated for this case to other reactions. New ideas on mechanisms are presented by Oyama, et oL, Parmaliana, et aL, and Laszlo. 

These results suggest that the (101) superstructure observed on the (001) -phase at the catalyst s operating temperature is closely related to Bi2M02O9. A quantification of the microanalysis of the jS-preparation shows a Bi-deficiency. Similar results are observed in the reaction of the a-phase in propylene. In a C3 H6-O2 mixture under working conditions both phases show the presence of this superstructure similar to the jS-structure. The ETEM results are consistent with XPS and Raman data which show that the surface structure of the active bismuth molybdate is close to the jS-phase and that the jS-phase is more active (Matsurra et al 1980, Burrington et al 1983). In these studies dramatic increases in the activity... 

At lower temperatures the Mars-van Krevelen mechanism no longer applies. Sancier et al. (440) studied propylene oxidation in the presence of 1802 over bismuth molybdate and found that the acrolein product contained 180 and not exclusively leO from the oxide lattice in contrast with results obtained by Keulks and co-workers (441, 442) at higher temperatures. This lower-temperature oxidation must involve adsorbed oxygen in some form but the nature is not clear. It is now accepted that not all these oxidation reactions do involve lattice oxygen (442,443). 

During the history of a half century from the first discovery of the reaction (/) and 35 years after the industrialization (2-4), these catalytic reactions, so-called allylic oxidations of lower olefins (Table I), have been improved year by year. Drastic changes have been introduced to the catalyst composition and preparation as well as to the reaction process. As a result, the total yield of acrylic acid from propylene reaches more than 90% under industrial conditions and the single pass yield of acrylonitrile also exceeds 80% in the commercial plants. The practical catalysts employed in the commercial plants consist of complicated multicomponent metal oxide systems including bismuth molybdate or iron antimonate as the main component. These modern catalyst systems show much higher activity and selectivity... 

In spite of the accumulated mechanistic investigations, it still seems difficult to explain why multicomponent bismuth molybdate catalysts show much better performances in both the oxidation and the ammoxidation of propylene and isobutylene. The catalytic activity has been increased almost 100 times compared to the simple binary oxide catalysts to result in the lowering of the reaction temperatures 60 80°C. The selectivities to the partially oxidized products have been also improved remarkably, corresponding to the improvements of the catalyst composition and reaction conditions. The reaction mechanism shown in Figs. 1 and 2 have been partly examined on the multicomponent bismuth molybdate catalysts. However, there has been no evidence to suggest different mechanisms on the multicomponent bismuth molybdate catalysts. 

Catalytic oxidation of propylene to acrolein was first discovered by the Shell group in 1948 on Cu20 catalyst (/). Both oxidation and ammoxidation were industrialized by the epoch-making discovery of bismuth molybdate catalyst by SOHIO (2-4). The bismuth molybdate catalyst was first reported in the form of a heteropoly compound supported on Si02, Bi P,Mo,2052/Si02 having Keggin structure but it was not the sole active species for the reactions. Several kinds of binary oxides between molybdenum trioxide and bismuth oxide have been known, as shown in the phase... 

In the 1960s, a number of binary oxides, including molybdenum, tellurium, and antimony, were found to be active for the reactions and some of them were actually used in commercial reactors. Typical commercial catalysts are Fe-Sb-O by Nitto Chemical Ind. Co. (62 -64) and U-Sb-O by SOHIO (65-67), and the former is still industrially used for the ammoxidation of propylene after repeated improvements. Several investigations were reported for the iron-antimony (68-72) and antimony-uranium oxide catalysts (73-75), but more investigations were directed at the bismuth molybdate catalysts. The accumulated investigations for these simple binary oxide catalysts are summarized in the preceding reviews (5-8). 

Typical Reaction Conditions for the Oxidation and Ammoxidation of Propylene on the Simple and Multicomponent Bismuth Molybdate Catalyst°... 

Investigations into the scheelite-type catalyst gave much valuable information on the reaction mechanisms of the allylic oxidations of olefin and catalyst design. However, in spite of their high specific activity and selectivity, catalyst systems with scheelite structure have disappeared from the commercial plants for the oxidation and ammoxidation of propylene. This may be attributable to their moderate catalytic activity owing to lower specific surface area compared to the multicomponent bismuth molybdate catalyst having multiphase structure. 

Oxidation of propylene to form acrolein depends on the first order of propylene and is independent of oxygen on multicomponent bismuth molybdate catalysts under the usual reaction conditions. The observed kinetics is the same with simple bismuth molybdates and suggests that the oxidation of propylene proceeds via the similar reaction scheme reported for simple molybdates, the slow step being the abstraction of allylic hydrogen (9-15, 19, 20). However, the reaction sometimes depends on the partial pressure of oxygen under lower temperature and lower oxygen pressure (41, 42). 

The same synergy effect between bismuth molybdates and mixed iron and cobalt molybdates on the mechanical mixture of both particles was reported by Millet et al. (98). However, it was also found that the surface of mixed iron and cobalt molybdate particle was changed during catalysis and a thin layer of bismuth molybdate was formed on the surface of mixed iron and cobalt molybdates after the reaction. It is doubtful that pure mechanical mixture shows the synergy effect for propylene oxidation, and it seems likely that propylene was mainly oxidized on the thin layer of bismuth molybdates formed on the mixed iron and cobalt molybdate in the experiment reported by Millet et al. (98). 

Molybdenum and bismuth are indispensable elements, forming the a-phase of bismuth molybdate, which is located mainly on the surface of the catalyst particle and constitutes the reaction site of propylene. 

Kinetics of reaction must be considered when attempting to postulate mechanisms, but kinetic equations alone are unreliable in fixing mechanism. For example, in the oxidation of propylene to acrolein, cuprous oxide and bismuth molybdate have very different kinetics, yet the studies of Voge, Wagner, and Stevenson (18), and especially of Adams and Jennings (1, 2) show that in both cases the mechanism is removal of an H atom from the CH3 group to form an allylic intermediate, from which a second H atom is removed before the O atom is added. The orders of the reactions and the apparent optimum catalysts (16) are as follows ... 

The use of isotopic tracers has demonstrated that the selective oxidation of propylene proceeds via the formation of a symmetrical allyl species. Probably the most convincing evidence is presented by the isotopic tracer studies utilizing, 4C-labeled propylene and deuterated propylene. Adams and Jennings 14, 15) studied the oxidation of propylene at 450°C over bismuth molybdate and cuprous oxide catalysts. The reactant propylene was labeled with deuterium in various positions. They analyzed their results in terms of a kinetic isotope effect, which is defined by the probability of a deuterium atom being abstracted relative to that of a hydrogen atom. Letting z = kD/kH represent this relative discrimination probability, the reaction paths shown in Fig. 1 were found to be applicable to the oxidation of 1—C3He—3d and 1—QH —1 d. 

In comparison to the bismuth molybdate and cuprous oxide catalyst systems, data on other catalyst systems are much more sparse. However, by the use of similar labeling techniques, the allylic species has been identified as an intermediate in the selective oxidation of propylene over uranium antimonate catalysts (20), tin oxide-antimony oxide catalysts (21), and supported rhodium, ruthenium (22), and gold (23) catalysts. A direct observation of the allylic species has been made on zinc oxide by means of infrared spectroscopy (24-26). In this system, however, only adsorbed acrolein is detected because the temperature cannot be raised sufficiently to cause desorption of acrolein without initiating reactions which yield primarily oxides of carbon and water. 

Sancier et al. (43) used oxygen-18 to examine the relative role of adsorbed versus lattice oxygen in propylene oxidation over a silica-supported bismuth molybdate catalyst as a function of temperature. At 400°C they observed the formation of predominantly acrolein[I60] rather than acrolein[I80], indicating significant participation of lattice oxygen. However, as the reaction temperature was decreased, the authors concluded that the role of adsorbed oxygen became more important. 

These observations suggest a reaction scheme for bismuth molybdate catalysts where the allylic species is formed initially at a bismuth center and then reacts further at a molybdenum site to produce acrolein. Thus, once the allylic complex is formed, the MoO polyhedra are highly active and selective for acrolein formation. This hypothesis was tested by investigating the oxidation of bromoallyl (C3HjsBr) over molybdenum oxide 116). Since the C—Br bond in bromoallyl is much weaker than the C—H bond in propylene, the ease of formation of the allylic species should be significantly enhanced with bromoallyl compared with propylene. If the initial propylene activation occurs on bismuth, then the reaction of bromoallyl over molybdenum oxide should approach the activity and selectivity of propylene over bismuth molybdate. This was the observed result, and the authors concluded that the bismuth site was responsible for the formation of the allylic intermediate. 

Example 4.2 Acrolein is being produced by catalytic oxidation of propylene on a bismuth molybdate catalyst. The following reactions are taking place in the reactor ... 

---