Looking at Self-Consistent-Charge Density Functional Tight Binding from a Semiempirical Perspective

Abstract

The self-consistent-charge density functional tight binding (SCC-DFTB) method is compared with other semiempirical methods (MNDO, AM1, PM3, OM1, OM2, OM3). Despite the differences in the underlying philosophy and derivation, these methods share many common features. Systematic evaluations of their performance are reported for standard test sets that are in common use. The overall accuracy of SCC-DFTB and the other methods is in the same range, with the overall tendency AM1<SCC-DFTB<OM2, but any such ranking depends on the properties and compound classes considered. SCC-DFTB is excellent for geometries and performs well for biological systems. It seems less suitable for the energetics of radicals and electronically excited states, and suffers from occasional outliers (e.g., for compounds with NO bonds). In an overall assessment, SCC-DFTB is a viable semiempirical method with specific strengths and weaknesses which may be an attractive choice especially for biomolecular applications.