Kinetic and Strain-Induced Self-Organization on Si

Instabilities during epitaxial growth are a general phenomenon that has been intensively studied an various metal surfaces. Despite the much larger application relevance of semiconductor surfaces, systematic studies have only begun in the last years an such materials. So far, strain-driven growth phenomena found most attention, and they are meanwhile widely exploited for the selforganization of quantum dots and other low-dimensional structures. Recently, we found a stepbunching instability an the technically most relevant Si(001) surface, which occurs under frequently employed homoepitaxial growth conditions and is of purely kinetic origin. lt is the purpose of this project to separate kinetic and strain-driven mechanisms to identify the microscopic origin of this instability, and to exploit the kinetic effect and its interplay with straininduced effects for new schemas of self-organized growth. To achieve these goals, it is first necessary to map out the behavior of the kinetic growth instability over the experimentally relevant range of the multi-dimensional growth parameter space, and to study its behavior an the atomic scale by scanning probe techniques. This experimental input will be used for deriving a sufficiently accurate diffusion potential for two-dimensional kinetic Monte Carlo simulations. This should lead to a quantitative atomic- scale model of the kinetic step bunching instability, which would be most useful for predicting the growth behavior under modified deposition conditions. To exploit, the kinetic step bunching phenomenon, routes toward a combination with straininduced dot Formation in the Si/SiGe system are investigated. By kinetic step bunching it becomes possible to generate a template with adjustable, quasi-periodic corrugations that can be combined with the Stranski-Krastanov island growth mode of Ge and Ge-rich SiGe alloys an Si. This could provide a new level of self-organized nanostructures, with strain- induced dot-Formation and kinetically adjusted long-range order. These systematic studies will also be applied to an hitherto unresolved question of great application relevance: Strained SiGe quantum wells an Si are known to exhibit carrier mobilities that are far below theoretical estimates. It is not yet clean whether interface roughness scattering or alloy scattering in the quantum well are the limiting factor. By using kinetically roughened interfaces with systematically varying roughness parameters, it will be possible to distinguish between these two mechanisms. The results are essential for future hetero-Field-effect transistors.

The Si(001) surface is the technically most important semiconductor surface, because it is used in all digital integrated circuits. Epitaxial growth on such a surface is a standard process during device fabrication. Nevertheless, it was not known until a publication from our group in 1999 that epitaxial growth on this surface is intrinsically unstable against the kinetic formation of surface corrugations (step bunches) under certain epitaxial growth conditions. FWF project P16223N08 was based on this experimental finding, aiming toward a more concise understanding of this phenomenon, and on a possible exploitation as a mechanism of self-organization. By means of Kinetic Monte Carlo simulations, we could identify the dominating step-bunching mechanism as the interplay between the atomic reconstruction of the Si(001) surface with its pronounce diffusion anisotropy, and the adsorption/desorption kinetics at atomic-height step edges. In subsequent experiments, it could be demonstrated that the surface corrugations caused by step-bunching can indeed be used as preferential nucleation sites for Ge and SiGe islands grown in a completely different, strain-driven self-organization regime. This leads to a fair, but not perfect, ordering of the islands. Perfect ordering could be demonstrated on Si(001) substrates with lithographically defined pit-arrays. In that case, the SiGe or Ge islands nucleate at the bottom of the pits and thus form an perfect island array with the geometrical properties of the template. Such heterostructure templates are of great interest for applications that require the individual addressability of the Ge islands, e.g. in applications where the carrier confinement in or near an island provides a means for information storage or information processing. Based on the initial experiments that combined kinetic step bunching with strain-driven 3D island growth, we derived a qualitative model for the preferential island nucleation on pit-patterned substrates. This model was corroborated by subsequent simulations, which revealed that the bottom of a pit is both kinetically and energetically the most favorable nucleation site for 3D island growth in a strained heterosystem. These result have led to a new level of understanding regarding self-organization in the SiGe heterosystem. They are expected to become relevant for new applications that combine top-down (lithography) and bottom-up (self- organization) mechanisms for the creation and ordering of semiconductor nanostructures. Also, the results of P16223N08 formed the basis for project P2 in the FWF-funded SFB 025 "IRoN".