Combining quantum mechanics and interatomic potential functions in ab initio studies of extended systems

The errors made when large chemical systems are replaced by small models are discussed: interrupted charge transfer, missing structure constraints, neglected long‐range interactions. A combined quantum mechanics (QM)–interatomic potential function (Pot) approach is described. Characteristic features of the QM‐Pot approach include: (1) periodic boundary conditions, (2) consistent definition of forces in the presence of link atoms that terminate the QM cluster, (3) interatomic potential functions parametrized on ab initio data and accounting for polarization effects, (4) use of reaction force fields (EVB potentials) in combination with QM methods for efficient localization of transition structures in large systems, (5) implementation as a loose coupling of existing QM and Pot engines. Comparison is made with some other hybrid QM/MM methods. Applications of the combined QM‐Pot method for ab initio modeling of the structure and reactivity of zeolite catalysts are reviewed with both protons and transition metal cations as active species. Potential functions of the ion‐pair shell‐model type available for such studies are compiled. The reliability of the method is checked by comparison with periodic ab initio studies and by examining the convergence of the results with increasing size of the QM cluster. The problems tackled are: different types of Cu+ sites in the CuZSM‐5 catalyst and their properties, acidity differences between active sites in different zeolite framework structures (energies of deprotonation, NH3 adsorption energies), and proton mobility in acidic zeolites. The combined QM‐Pot approach made possible a full ab initio prediction of reaction rates for an elementary process on the surface of solid catalysts and of how these rates differ between different catalysts with the same active site. © 2000 John Wiley & Sons, Inc. J Comput Chem 21: 1470–1493, 2000