ar X iv : c on d - m at / 0 30 65 47 v 2 2 9 Se p 20 03 Structure and stability of the Si ( 105 ) surface

Recent experimental studies have shown that well-annealed, unstrained Si(105) surfaces appear disordered and atomically rough when imaged using scanning tunnelling microscopy (STM). We construct new models for the Si(105) surface that are based on single-and double-height steps separated by Si(001) terraces, and propose that the observed surface disorder of Si(105) originates from the presence of several structural models with different atomic-scale features but similar energies. This degeneracy can be removed by applying compressive strains, a result that is consistent with recent observations of the structure of the Ge/Si(105) surface. 1 The self-organized growth of Ge/Si quantum dots has been investigated extensively for more than a decade, driven by their potential applications as optoelectronic devices and nanoscale memories. In the early stages of growth, the quantum dots that have pyramidal shapes bounded by {105} facets evolve from stepped mounds without encountering any energetic barriers for their nucleation [1, 2]. The absence of nucleation barriers has been explained by a competition between the strain-dependent, negative step formation energy and repulsive step-step interactions that have a weak dependence on strain [3, 4]. The atomic configuration of the (105) facets has also been elucidated; a rebonded step model for the Ge(105) surface under mismatch strain was found to play a crucial role for the stability of this surface [3, 5, 6, 7]. While the structure of Ge/Si(105) surface has been recently elucidated, the Si(105) surface shows intriguing features that are not well understood. Experimental work by Tomitori et al. [8], Fujikawa et al. [5] and Zhao et al. [9] reveal that Si(105) is atomically rough even after careful annealing, and its STM image does not display two-dimensional periodicity. The analysis of Zhao and coworkers [9] suggests the presence of a structure for the Si(105) surface with large (001) facets and double-height steps. While a model for this surface with double-height steps was presented in [9], a study of its stability has not been attempted. Motivated by the recent STM investigations [5, 9], we search for reconstructions of Si(105) based on (001) terraces separated by single-and double-height steps oriented along 100 directions. In addition to the currently accepted models of Si(105) (shown, for example, in Ref. [5]), we have found a few other possible structures based on double-height steps. We investigate the stability of these novel reconstructions, and propose that the roughness of the Si(105) surface is due to the coexistence of different structures with …