The new ONIOM (our own n-layered integrated molecular orbital and molecular mechanics) approach has been proposed and shown to be successful in reproducing benchmark calculations and experimental results. ONIOM3, a three-layered version, divides a system into an active part treated at a very high level of ab initio molecular orbital theory like CCSD(T), a semiactive part that includes important electronic contributions and is treated at the HF or MP2 level, and a nonactive part that is handled using force field approaches. The three-layered scheme allows us to study a larger system more accurately than the previously proposed two-layered schemes IMOMO, which can treat a medium size system very accurately, and IMOMM, which can handle a very large system with modest accuracy. This three-layered scheme has been applied to activation barriers for the Diels−Alder reaction of acrolein + isoprene, acrolein + 2-tert-butyl-1,3-butadiene, and ethylene + 1,4-di-tert-butyl-1,3-butadiene. In general, the results for both geometry optimizations and single point energy calculations agree well with benchmark predictions and experimental results. The scheme has also been applied to the transition state for the oxidative addition of H2 to Pt(P(t-Bu)3)2. The activation energy of this 83-atom reaction is predicted to be 14.2 kcal/mol with the ONIOM3(CCSD(T):MP2:MM3) method.

For reactions of gas phase, ground state, neutral transition metal atoms from the 4d series with alkanes and alkenes, we combine 300 K kinetics measurements with ab initio electronic structure calculations to infer mechanisms in some detail. The theoretical method PCI-80 with zero-point energy corrections to the bare potential surface apparently produces bond energies, reaction exothermicities, and even saddle point energies accurate to within 2-3 kcal/mol, provided that the correct ground state has been located, which is sometimes difficult. The reactions fall into two general categories: termolecular stabilization of long-lived M(hydrocarbon) complexes and bimolecular elimination of Hz. By using the ab initio energies and vibrational frequencies in a statistical unimolecular rate theory (RRKM theory), we can model the lifetimes of M(hydrocarbon) complexes to assess the plausibility of a saturated termolecular mechanism at 1 Torr He. Termolecular examples include the reactions of Pd with alkanes to form long-range v2 complexes; the reactions of Rh and Pd with alkenes to form n complexes; and probably the reactions of Y, Zr, Nb, Rh, and Pd with cyclopropane to form CH or CC insertion complexes. In other reactions, all of the evidence indicates a bimolecular H2 elimination mechanism. Rhodium is unique among the 4d metal atoms in effecting HZ elimination from ethane and larger alkanes. Yttrium, zirconium, and niobium almost surely insert in CH bonds of ethylene and larger alkenes, ultimately eliminating H2. We discuss the general requirements on the pattern of atomic electronic states that pennit efficient CH bond activation and H2 elimination. The good agreement between the observed reaction rates and the PCI-80 calculations lends confidence to future efforts to apply ab initio techniques to more complicated catalytic systems, including condensed phase reactions involving ligated metal centers.

The integrated MO+MO (IMOMO) method, recently proposed for geometry optimization, is tested for accurate single point calculations. The principle idea of the IMOMO method is to reproduce results of a high level MO calculation for a large ‘‘real’’ system by dividing it into a small ‘‘model’’ system and the rest and applying different levels of MO theory for the two parts. Test examples are the activation barrier of the SN2 reaction of Cl−+alkyl chlorides, the C=C double bond dissociation of olefins and the energy of reaction for epoxidation of benzene. The effects of basis set and method in the lower level calculation as well as the effects of the choice of model system are investigated in detail. The IMOMO method gives an approximation to the high level MO energetics on the real system, in most cases with very small errors, with a small additional cost over the low level calculation. For instance, when the MP2 (Mo/ller–Plesset second-order perturbation) method is used as the lower level method, the IMOMO method reproduces the results of very high level MO method within 2 kcal/mol, with less than 50% of additional computer time, for the first two test examples. When the HF (Hartree–Fock) method is used as the lower level method, it is less accurate and depends more on the choice of model system, though the improvement over the HF energy is still very significant. Thus the IMOMO single point calculation provides a method for obtaining reliable local energetics such as bond energies and activation barriers for a large molecular system.