James J. P. Stewart

MOPAC: A semiempirical molecular orbital program

Before we start, we need a working definition for MOPAC. The following description has been used many times to describe MOPAC: MOPAC is a general-purpose, semiempirical molecular orbital program for the study of chemical reactions involving molecules, ions, and linear polymers. It implements the semiempirical Hamiltonians MNDO, AM 1, MINDO/3, and MNDOPM3, and combir_es the calculations of vibrational spectra, thermodynamic quantities, isotopic substitution effects, and force constants in a fully integrated program. Elements parameterized at the MNDO level include H, Li, Be, B, C, N, O, F, A1, Si, P, S, C1, Ge, Br, Sn, Hg, Pb, and I; at the PM3 level the elements H, C, N, O, F, A1, Si, P, S, C1, Br, and I are available. Within the electronic part of the calculation, molecular and localized orbitals, excited states up to sextets, chemical bond indices, charges, etc. are computed. Both intrinsic and dynamic reaction coordinates can be calculated. A transition-state location routine and two transition-state optimizing routines are available for studying chemical reactions.

Application of localized molecular orbitals to the solution of semiempirical self‐consistent field equations

When conventional matrix algebra is used to solve the semiempirical self-consistent field equations for large systems, the time required rises as the third power of the size of the system. A consequence of this is that self-consistent calculations of large systems such as enzymes are impractical. By using localized molecular orbitals instead of matrix methods, the time required for these systems can be made almost proportional to the size of the system. In partial geometry optimizations, the time required depends only upon the size of the fragment being optimized and is almost independent of the size of the whole system.

Calculation of the geometry of a small protein using semiempirical methods

Abstract When the geometry of large systems is optimized using semiempirical methods, problems are encountered that are not found in smaller systems. In particular, the geometry optimizers conventionally used are all unsuitable for the study of large systems. By modifying the popular BFGS procedure for updating the inverse Hessian matrix, and by modifying the calculation of the step, a method that appears to allow the geometries of large systems to be optimized has been developed. The new method has been applied to the calculation of crambin using PM3, in which the geometry was defined using Cartesian coordinates for the molecular backbone and internal coordinates for the side chains.

Calculation of vibrational frequencies using molecular trajectories

Abstract Vibrational frequencies calculated using the mass-weighted Hessian matrix assume simple harmonic motion in a parabolic potential. By explicit calculation of the time-dependent normal coordinates, the non-parabolic nature of the potential can be considered. For small systems, taking into account the non-parabolic nature of the potential lowers the higher vibrational frequencies by approximately 40 cm−1.

AB INITIO AND SEMI-EMPIRICAL, CALCULATIONS ON THE OXYHALIDES OF SILICON

Abstract In 1982 the heats of formation of the silicon oxyhalides were determined by experiment to be -267.9 Kcal/mole (SiOF2), -167.8 (SiOCl2), -137.4 (SiOBr2), and -99.4 (SiOI2). MNDO-PM3 semiempirical calculations indicate that the heats of formation are nearer to -229.4, -121.2, -94.1, and -50.2 Kcal/mole. respectively. Ab-initio calculations using a 6-31G(d) basis set confirm the semiempirical results for the fluoride and chloride systems.