In this article we report quantum scattering calculations on the reactive system H + O2 OH + O for total angular momentum J=0, 1, 2, and 5. To facilitate our calculations we performed them on a parallel computer using the Coriolis coupled method of Goldfield and Gray.
From the shape of the J=0 reaction probability we infer that the reaction proceeds through two different reaction mechanisms. In the first reaction mechanism, at low energies (E < 1.25 eV), the reaction probably goes through an intermediate HO2-complex. This gives rise to very sharp resonances. These resonances take a long time to resolve ( 150 000 a.u.). At higher energies (E > 1.25 eV), a second non-complex forming (direct) reaction path opens up, leading to enhanced reaction probability. The broader resonances in this energy range are caused by the reactants that do form a complex. This is consistent with earlier findings in the literature[16,25,26,27,32,33,34]. The exact reason for the broadening of the resonances can not be obtained from the present work.
Comparing the J=0 to the J=1 results we see a big decrease in the reaction probability at higher energies. Examining the J=1 reaction probabilities, we note that the difference is primarily caused by the reaction probabilities for which the wave function starts in the or in the substate. The difference is tentatively explained by realizing that when the wave function starts substate, the complex is initially in a ``planar'' configuration, in which the O2 molecule rotates in the plane of the 3 atoms. On the other hand, when the wave function starts in the or substate, the complex is initially in a ``perpendicular'' configuration (see Sec. IVB). In the first configuration the reaction probability is enhanced, because it is possible for the H atom to ``hit'' one of the O atoms directly, opening up the ``direct'' pathway. This explanation is consistent with the findings of Bronikowski et al.. The differences between the calculation and the calculation we tentatively explain by centrifugal barrier effects (i.e., change in the J(J+1) term in the Hamiltonian).
Going from J=1 to J=2 to J=5, the above mentioned reaction mechanism and explanations for the reactivity of the wave packet in different initial states are still valid. This leads us to believe that there might be a way to devise an approximate method to calculate cross sections based on only a small number of calculations with J>0. Furthermore, we see that the reaction probability is strongly dependent on the initial state of the wave packet. This means that , which is not a good quantum number, does not scramble very quickly. However, looking at the distributions for J=5, it is clear that no states can be left out of the calculation. This might change, when we choose a different embedding for the calculations, i.e., when we take the O2 bond to be the BF z-axis or when we perform the calculation in product coordinates. However, it is not clear beforehand, whether this will make a difference and the fact that the resonance features for higher J are less pronounced than at J=0 suggests that it may not. We are currently investigating this issue.
Currently, we are performing calculations for J>5 to check whether the models and explanations given in this article still hold at higher J. Ultimately, we want to use our calculations to get cross sections for this reaction. Furthermore, we want to devise an accurate approximate method for performing calculations for J>0 based on only a few J>0 calculations.
The authors wish to thank Dr. R. T Pack for sending us his results and for useful discussions. We wish to thank Dr. J. Z. H. Zhang for useful discussions. Further, we wish to thank Dr. G. C. Groenenboom for critically reading the manuscript and useful comments. Partial support for this research comes from NSF grant CHE-9526658. We also acknowledge generous grants of computer time from the Cornell Theory Center, the Argonne High Performance Computing Research Facility and the High Performance Computing Facility at Wayne State University.