Computational Study of the Mechanism and Product Yields in the Reaction Systems C2H3 + CH3 ⇄ C3H6 ⇄ H + C3H5 and C2H3 + CH3 → CH4 + C2H2

Calculational and experimental investigations of the pressure effects on radical-radical cross combination reactions: C2H5+C2H3

Pressure-dependent product yields have been experimentally determined for the cross-radical reaction C2H5 + C2H3. These results have been extended by calculations. It is shown that the chemically activated combination adduct, 1-C4H8*, is either stabilized by bimolecular collisions or subject to a variety of unimolecular reactions including cyclizations and decompositions. Therefore the "apparent" combination/disproportionation ratio exhibits a complex pressure dependence. The experimental studies were performed at 298 K and at selected pressures between about 4 Torr (0.5 kPa) and 760 Torr (101 kPa). Ethyl and vinyl radicals were simultaneously produced by 193 nm excimer laser photolysis of C2H5COC2H3 or photolysis of C2H3Br and C2H5COC2H5. Gas chromatograph/mass spectrometry/flame ionization detection (GC/MS/FID) were used to identify and quantify the final reaction products. The major combination reactions at pressures between 500 (66.5 kPa) and 760 Torr are (1c) C2H5+C2H3-->1-butene, (2c) C2H5 + C2H5-->n-butane, and (3c) C2H3+C2H3-->1,3-butadiene. The major products of the disproportionation reactions are ethane, ethylene, and acetylene. At moderate and lower pressures, secondary products, including propene, propane, isobutene, 2-butene (cis and trans), 1-pentene, 1,4-pentadiene, and 1,5-hexadiene are also observed. Two isomers of C4H6, cyclobutene and/or 1,2-butadiene, were also among the likely products. The pressure-dependent yield of the cross-combination product, 1-butene, was compared to the yield of n-butane, the combination product of reaction (2c), which was found to be independent of pressure over the range of this study. The [1-C4H8]/[C4H10] ratio was reduced from approximately 1.2 at 760 Torr (101 kPa) to approximately 0.5 at 100 Torr (13.3 kPa) and approximately 0.1 at pressures lower than about 5 Torr (approximately 0.7 kPa). Electronic structure and RRKM calculations were used to simulate both unimolecular and bimolecular processes. The relative importance of C-C and C-H bond ruptures, cyclization, decyclization, and complex decompositions are discussed in terms of energetics and structural properties. The pressure dependence of the product yields were computed and dominant reaction paths in this chemically activated system were determined. Both modeling and experiment suggest that the observed pressure dependence of [1-C4H8]/[C4H10] is due to decomposition of the chemically activated combination adduct 1-C4H8* in which the weaker allylic C-C bond is broken: H2C=CHCH2CH3-->C3H5+CH3. This reaction occurs even at moderate pressures of approximately 200 Torr (26 kPa) and becomes more significant at lower pressures. The additional products detected at lower pressures are formed from secondary radical-radical reactions involving allyl, methyl, ethyl, and vinyl radicals. The modeling studies have extended the predictions of product distributions to different temperatures (200-700 K) and a wider range of pressures (10(-3)-10(5) Torr). These calculations indicate that the high-pressure [1-C4H8]/[C4H10] yield ratio is 1.3+/-0.1.

Predictive theory for the combination kinetics of two alkyl radicals

An ab initio transition state theory based procedure for accurately predicting the combination kinetics of two alkyl radicals is described. This procedure employs direct evaluations of the orientation dependent interaction energies at the CASPT2/cc-pvdz level within variable reaction coordinate transition state theory (VRC-TST). One-dimensional corrections to these energies are obtained from CAS+1+2/aug-cc-pvtz calculations for CH3 + CH3 along its combination reaction path. Direct CAS+1+2/aug-cc-pvtz calculations demonstrate that, at least for the purpose of predicting the kinetics, the corrected CASPT2/cc-pvdz potential energy surface is an accurate approximation to the CAS+1+2/aug-cc-pvtz surface. Furthermore, direct trajectory simulations, performed at the B3LYP/6-31G* level, indicate that there is little local recrossing of the optimal VRC transition state dividing surface. The corrected CASPT2/cc-pvdz potential is employed in obtaining direct VRC-TST kinetic predictions for the self and cross combinations of methyl, ethyl, iso-propyl, and tert-butyl radicals. Comparisons with experiment suggest that the present dynamically corrected VRC-TST approach provides quantitatively accurate predictions for the capture rate. Each additional methyl substituent adjacent to a radical site is found to reduce the rate coefficient by about a factor of two. In each instance, the rate coefficients are predicted to decrease quite substantially with increasing temperature, with the more sterically hindered reactants having a more rapid decrease. The simple geometric mean rule, relating the capture rate for the cross reaction to those for the self-reactions, is in remarkably good agreement with the more detailed predictions. With suitable generalizations the present approach should be applicable to a wide array of radical-radical combination reactions.

Molecular Dynamics of Complex Gas-Phase Reactive Systems by Time-Dependent Groups

A novel way of assembling the total potential for performing molecular dynamical studies of complex gas-phase reactive chemical systems is introduced. The method breaks the calculation of the total potential and gradients of the potential into time-dependent groups that are governed by spatial cutoffs. These groups evolve during the course of the simulation and their number may increase or diminish as the dynamics of the system determine. In an effort to extend the simulation time of these complex reactive processes and to use high levels of theory when necessary, multiple levels of theory may be used over the groups for the calculation of both the intragroup and intergroup interactions. Representative simulations are performed to illustrate the method and a computationally facile method of obtaining the groups of a simulation are also discussed.