An attempt was made to demonstrate, for the first time, the possible existence of planar point defect Si self-interstitials on the {311} plane. It was recalled that the conventional procedure for minimizing dangling bonds in interstitial chains along <110> was to insert 2 interstitials into the center of the 6-membered ring. However, the current study also proposed a number of interesting possibilities for building chains in the <110> direction. The propagation of such interstitial chains, using the present interstitials, did not have to involve so-called cooperative atomic processes which required the simultaneous motion of more than one atom at a time. Furthermore, the (332) interstitials could perhaps explain features such as the steps and diffuse streaks which were observed in {311} clusters. The possibility of a large number of configurations for the self-interstitial, together with perturbations which extended over several sites, could explain the puzzle concerning the high entropy of self-diffusion in Si (approximately 9k). Previous calculations of the split <110> interstitial, with its extended perturbations, had also yielded a similar entropy value. It was concluded that, in minimum-energy configurations, the defect and the 4 neighbours to which it was bonded lay predominantly in {311} planes. On the other hand, a configuration which tried to attain tetrahedral bonding yielded a far higher formation energy under the same conditions. Upon using the Ackland potential, an interstitial which was split in the [110] direction was found to have a higher formation energy, by 1eV. The interstitials which were proposed here could provide a plausible reason for the formation of the commonly observed {311} clusters, following ion implantation. These interstitials did not leave any dangling bonds, and could be considered to be extended due to spread-out perturbations.

Planar Self-Interstitial in Silicon M.M.De Souza, C.K.Ngw, M.Shishkin, E.M.S.Narayanan: Physical Review Letters, 1999, 83[9], 1799-801