Method next up previous
Next: The C Defect Up: Dynamic Properties of Previous: Introduction

Method

The calculations presented here were performed using an ab initio local density functional cluster method on large H-terminated clusters (AIMPRO). A C atom was inserted into the central region of a 71 atom tetrahedral cluster Si tex2html_wrap_inline542 H tex2html_wrap_inline544 to investigate the structure of C tex2html_wrap_inline334 . The migration energy of C tex2html_wrap_inline334 was found by inserting a C atom into a trigonal 86 atom cluster Si tex2html_wrap_inline550 H tex2html_wrap_inline552 , oriented along tex2html_wrap_inline402 111 tex2html_wrap_inline404 . A central Si atom was then replaced by C to investigate the di-carbon defect. Norm-conserving pseudo-potentials for C and Si were taken from [20].

The electronic wave functions were expanded in s- and p- Gaussian orbitals, centered on nuclei and the center of all bonds. The basis used here for the structural and local mode calculations was as following: eight Gaussian fitting functions of different exponents were used on the carbon atom(s) and inner silicon atoms, and a fixed linear combination of eight Gaussians were placed on the remaining Si atoms. A linear combination of three Gaussians were used for the terminating hydrogens. Additionally, three s- and p- Gaussian orbitals were placed at the center of every C-Si and Si-Si bond.

The electron charge density was fitted in exactly the same way as the wavefunctions. The self-consistent energy was then found, and the forces on each atom calculated. The whole structure is then relaxed using a conjugate gradient algorithm, until the equilibrium structure is determined. To explore the saddle points needed for migration energies, the relative bond lengths of the carbon atom were constrained during the relaxation as discussed previously [21]. The second derivatives of the energy with respect to atomic positions were calculated for the carbon atom(s) and their neighboring Si atoms. Energy second derivatives for the remaining atoms were taken from the Musgrave-Pople potential found previously [22], and the dynamical matrix of the cluster then constructed. From the dynamical matrix, the vibrational modes, and their normal coordinates tex2html_wrap_inline566 can be calculated.

We have previously shown that the dipole moments for small molecules can be found accurately using this method [23], and this can be used to evaluate the effective charge tex2html_wrap_inline568 associated with the vibrational modes of the defect. However, a larger basis is required for the accurate determination of tex2html_wrap_inline568 so in this calculation the number of Gaussians used to fit the wavefunction and charge density of C and Si was increased to ten. It is the effective charge which controls the intensity ratio of the two IR absorption lines. In this work, tex2html_wrap_inline568 is calculated for the two highest local vibrational modes of C tex2html_wrap_inline334 , having tex2html_wrap_inline398 and tex2html_wrap_inline400 symmetries. These modes largely involve motion of the carbon atom along [001] and [110] respectively together with movements of the Si neighbors. tex2html_wrap_inline568 is found by calculating the change in the adiabatic dipole moment of the cluster, when the atoms of the cluster are displaced by tex2html_wrap_inline582 / tex2html_wrap_inline584 . Here tex2html_wrap_inline586 is the mass of the displaced atom.


next up previous
Next: The C Defect Up: Dynamic Properties of Previous: Introduction

Antonio Resende
Wed Jan 15 12:41:08 GMT 1997