We systematically investigate the effects of basis set quality on the prediction of a representative amorphous tetrahedral (Formula presented) carbon structure. In fully self-consistent first-principles calculations, variations in the quality of the basis set do result in significant variations in predicted structure. Substantial differences between the calculations, ranging from a minimum basis to a high-quality double-zeta plus polarization (DZP) basis, for amorphous carbon stem from two sources. We discover that to properly relax candidate (Formula presented) structures requires a high-quality basis set in the calculation. A minimum basis set is inadequate to negotiate the transition geometries involved in the making and breaking of bonds while relaxing a structure. Relaxation of an (Formula presented) structure proposed by Drabold, Fedders, and Stumm [Phys. Rev. B 49, 16 415 (1994)] using a minimum, and a DZP basis demonstrates this point. The minimum basis set calculation leaves the bonding topology essentially unchanged, while relaxation using a DZP basis removes most of the small rings and triples the number of threefold bonded atoms. In addition, we find that to accurately represent the energetics of highly defected local structures, as found in (Formula presented) also requires a high-quality basis set. We demonstrate this point by using molecular analogs of local structures found in (Formula presented) Notable is how little rebonding and energy separates (Formula presented) structures that have qualitatively different densities of threefold atoms.
|Original language||English (US)|
|Number of pages||10|
|Journal||Physical Review B - Condensed Matter and Materials Physics|
|State||Published - Jan 1 1998|
ASJC Scopus subject areas
- Electronic, Optical and Magnetic Materials
- Condensed Matter Physics