Fundamental constants are taken from the CODATA report [71]. A good introduction to MOPAC can be found in Tim Clark's book [131].
Results
The SCF M.O.s, which diagonalize the Fock matrix, can be localized [38] to give M.O.s which can be identified with the conventional picture of two-electron bonds and lone pairs. The localization scheme is faster at the semiempirical level than the Edmiston-Ruedenberg [62] or Boys [61] methods. Associated with each conventional M.O. is a bond-index [23], which represents the contribution to the bond-order matrix due to each M.O. Bond orders and valencies can be displayed by use of BONDS [15]. Other phenomena relating to bonding can also be calculated [24,25,26]. An alternative to the normal Coulson density matrix is the Mulliken [132], the center of mass is used. Higher terms, e.g., polarizability, and first and second hyperpolarizability, can be calculated [47] by POLAR.
Ionization potentials [59] can be corrected using Green's Functions [34,35,36,33,37,32]
Solvent and Electrostatics
Solvent phenomena can be studied. The COSMO technique [30], unlike the self-consistent reaction fields [74], allows geometries to be optimized. Although the Miertus-Scrocco-Tomasi model [75,76] cannot optimize geometries, is more sophisticated in that it allows cavitation effects. This model has been modified [77,78,79,80,81] to allow NDDO methods to be used. In this, optimized VdW radii [83,84] are used to construct [82] a cavity.
The free energy of hydration is computed as the addition of three contributions:
1. The electrostatic term, which is computed from the linear free energy response theory [75,76,77,78,79,80,81].
2. The cavitation contribution, which is computed from Pierotti's scaled particle theory [85].
3. The van der Waals terms, which is computed using a linear relation with the solute accessible surface, and optimized "hardness" parameters [83,84].
In addition to the free energy of hydration a "solvent-adapted" wavefunction is obtained. Such a wavefunction can be used to determine changes in solute properties due to the solvent [86,87,88,89].
Electrostatic potentials can be used with the MST method both by deorthogonalizing the wavefunction [90,91,40] and by keeping the wavefunction orthogonal [92,93]
Other ESP methods available are the Merz-Bessler technique [31] and the Ford-Wang procedure [45,46]. The Ford-Wang is much faster and more accurate than the Merz-Bessler method, but is limited to AM1 calculations on systems containing H, C, N, O, F, and Cl, only.