Theoretical study of the Cu$_{\rm PL}$ defect in Si

Copper is a common contaminant in Si processing. When in supersaturation, a fraction of 1\%\ of the Cu in the sample forms an electrically-active defect easily seen by photoluminescence. This Cu$_{\rm PL}$ defect in Si has a no-phonon line at 1014 meV. It has long been believed to consist of an interstitial copper (Cu$_i$) weakly bound to a substitutional copper (Cu$_s$) : The \{Cu$_s$Cu$_i$\} pair. However, PL studies in isotopically pure $^{28}$Si crystals have shown that the defect contains not two but four copper atoms [1]. We examine the possibility that the core of the defect consists of not one but two adjacent substitutional Cu atoms. This core traps two Cu$_i$ atoms, resulting in defect with $D_{3d}$ symmetry. We will discuss its formation mechanism and stability, and show that they are consistent with the conditions at which Cu$_{\rm PL}$ is observed. If this model is correct, then then DLTS lines associated with Cu$_s$ should be re-assigned to \{Cu$_s$Cu$_s$\}. \\[4pt] [1] M. Steger {\it et al}., Phys. Rev. Lett. 100, 177402 (2008)