PM6 (Parameterization Method 6)
PM6 is a recently developed semiempirical method that follows in the tradition of other “Dewar-style” NDDO methods such as MNDO, AM1, PM3, SAM1, and PM5. While slightly emphasizing biochemical systems in parameterization, PM6 is a general purpose model of good quality. (Journal of Molecular Modeling 13, 1173–1213, 2007.) Dr. James JP Stewart Stewart is the developer.
The capability to carry out calculations on molecules containing transition metals on elements up through Bismuth (excluding the rare earths) has been added in PM6. A very wide range of systems can be treated efficiently when PM6's parameters are coupled with AMPAC's excellent CI capabilities.
The “PM” of the method’s name stands for “parameterization method”, and this is the primary difference between PM6 and previous NDDO methods. Stewart set out to evaluate and systematize the accreted approximations and changes to recent semiempirical method development. Some of the most important aspects of PM6 that differ from previous work include:
- Careful examination and pruning of the experimental data to ensure quality and consistency
- Use of ab initio and DFT results where experimental data is lacking
- Inclusion of new experimental data (such as dimerization energies) in parameterization
- Addition of many pairwise interaction terms that adjust the core repulsion function
- New method for computing 1-center 2-electron integrals for TMs
- Molecular mechanics correction terms for certain difficult cases.
The general performance of PM6 is better than that for AM1, PM3, or RM1 for traditional systems. Results of similar quality are obtained for ionization potentials, dipole moments and geometries. Check here for a comparison.