Propene oxide is currently produced either by the catalytic oxidation of propene with a complex (and relatively expensive) hydroperoxide, or via the chlorohydrin process which produces large quantities of halogenated waste (Kahlich et al, 1993, Sheldon, 1991). Ideally it would be manufactured using molecular oxygen as the oxidant, as with ethylene oxide which is produced with high selectivity (70-80%) via the oxidation of ethylene with oxygen over a silver catalyst, with a per pass conversion of the hydrocarbon of 7 - 15% (Rebsdat and Mayer, 1987, Sheldon, 1991). However, the analogous reaction using propene is unsuccessful, producing a complex mixture of unsaturated products, resulting from the cleavage of the relatively weak allylic-H bond. Consequently, the direct non-catalytic reaction of propene and molecular oxygen has been examined for many years as an alternative route to the epoxide (Lenher, 1932, 1935, Lemon et al, 1964). Generally though, only low epoxide selectivities have been reported (less than 30%) which are not high enough for the process to be commercially viable, although at relatively low temperatures and high pressures in fuel-rich mixtures, propene oxide selectivities of ca 50% can be obtained (Aprahamian, 1976, Pennington, 1992, Stark and Waddington, 1995).

It has been proposed that the epoxide is formed from propene by reaction with several different peroxy radicals, most significantly hydroperoxy, hydroxypropylperoxy and acetylperoxy radicals (Stark and Waddington, 1995). In spite of selectivities of ca 50%, primary reactions of this type do not seem adequate on their own of giving a high enough propene oxide selectivity to be commercially viable, due to the formation of large quantities of aldehydes as byproducts of the reactions leading to the epoxide. The experiments reported here were performed to increase the selectivity to, and rate of production of, the epoxide, closer to the point where the direct non-catalytic process would be commercially viable.

It has been suggested that additional epoxide may be formed by the further oxidation of these aldehydes (Ray et al, 1973, Ray and Waddington, 1973). Indeed if acetaldehyde is in large excess over the alkene, very high selectivities (ca 100%) can be achieved (Ruiz Diaz et al., 1975), although it would not be commercially viable to add the aldehyde as a feedstock to a propene:oxygen mixture as its cost per mole is comparable with the resulting epoxide. In this paper, it is demonstrated that the selectivity of propene oxide, produced by propene autoxidation, can be increased by recycling acetaldehyde to ensure its complete consumption (shown schematically in figure 1), but only if the temperature is lowered sufficiently to ensure that the resultant acetyl radicals add oxygen to give the highly efficient epoxidising agent, the acetylperoxy radical.

Two sets of experiments have been conducted, one using a static system and pressures of 0.92 bar, and the second, extending the pressure range up to 55 bar, using a flow system. Although the static, low pressure experiments have a fairly high selectivity to the epoxide, they produce it at a very low absolute rate and practical reactors would have to generate the epoxide at a much faster rate. As a step toward this goal, the high pressure pilot plant experiments reported here have a production rate of epoxide of 0.2 kg h^-1, four orders of magnitude larger than from the low pressure experiments.

A computer model, developed to describe propene oxidation with conditions of 505-549K and up to 4 bar in a static reactor (Stark and Waddington, 1995), was used to simulate these acetaldehyde:propene co-oxidation experiments with the comparison between model and experiment used to highlight the strengths and weaknesses of the model.