There has been interest recently in constructing reaction schemes to allow the simulation of propene autoxidation.[3,42,43] However, of the seven peroxyl radicals that have been identified as contributing to the epoxidation of the alkene, rate constants for only two have been measured, with the rest being estimated on the basis of ad hoc assumptions. Consequently, the relationship with the highest correlation coefficient, that between E_c, the charge transfer energy decrease, and the activation energy for the reaction has been used to estimate activation energies for the unstudied reactions (epoxidation of propene by methylperoxyl (CH₃O₂), acrylperoxyl (C₂H₃C(O)O₂), 1-hydroxy-2propylperoxyl (HOC₃H₆O₂), hydroxymethylperoxyl (HOCH₂O₂) and allylperoxyl (C₃H₅O₂)). Electron affinities and ionisation energies were estimated using Mopac 6.0, and are given in table 2. , while the charge transfer energy decrease, dE_c, and the activation energies calculated using equation 6 are given in table 3. These values, along with the average A factor of 1.28x10⁸ dm³ mol^-1 s^-1 can be used in subsequent computer models of propene oxidation.

An activation energy for methylperoxyl of 57.1±5.2 kJ/mol was also calculated using only data for the correlation between activation energy for epoxidation by methylperoxyl radicals and the alkene ionisation energy ( figure 1. ). This value is within one standard error of that calculated using equation 6 (61.3±5.3 kJ/mol). Experimental values for epoxidation by acetylperoxyl and hydroperoxyl radicals are also given in table 3. for comparison; the calculated values are higher by 4.5 and 7.2 kJ/mol respectively, and fall within the combined standard errors of 6.5 and 7.2 kJ/mol.

The correlations of epoxidation activation energy with both N_c and (I_alkene-A_peroxyl) (equations 5 and 7) were also used to calculate activation energies for the epoxidation of propene; however, they are not significantly different from values calculated using equation 6. Those determined using equation 5 were lower by an average of 0.9 kJ/mol (with a standard deviation of 1.6 kJ/mol), whereas values found using equation 7 were lower by an average of 1.8 kJ/mol (standard deviation of 3.7 kJ/mol), within the estimated standard errors of the calculated values of 5 - 7 kJ/mol.

As described in the previous section, using MOPAC 6.0 with the AM1 Hamiltonian probably overestimates the electron affinities of the more branched radicals t-C₄H₉O₂ and i-C₃H₇O₂, so to examine the effect of these values on predicted activation energies, the correlation between (I_alkene-A_peroxyl) and epoxidation activation energy was recalculated without data for t-C₄H₉O₂ and i-C₃H₇O₂. The resulting predicted activation energies for the species given in table 3. were lower than values found with equation 7 by 3.6 kJ/mol (with a standard deviation of 0.3 kJ/mol), which is less than the standard error on the estimated values found using equation 7 of 7.4 kJ/mol.

Of the five previously unmeasured activation energies for the epoxidation of propene, four are within one standard error (±5.3 kJ/mol) of the previous estimates, giving credence to the ad hoc assumptions used (that C₂H₃C(O)O₂ can be assumed analogous to CH₃C(O)O₂, HOC₃H₆O₂ behaves like i-C₃H₇O₂, etc.).[3] However, the activation energy for epoxidation by HOCH₂O₂ calculated here is significantly higher, by 15 kJ/mol, than the previous estimate used in computer simulation of propene autoxidation.[3] Preliminary modelling studies suggest that the higher activation energy gives a rate constant low enough for epoxidation by HOCH₂O₂ to be insignificant under all conditions. Previously, in the modelling of fuel-rich propene autoxidation at 500-550 K and relatively high pressures (4-50 bar), this route was the main source of formic acid (formed by the oxidation of the HOCH2O radical resulting from epoxidation by HOCH₂O₂).4 Formic acid can be a significant product, particularly at the higher pressures; the mechanism of formic acid formation will therefore need to be re-examined.