Multiscale modeling of formation and structure of oxide embedded silicon and germanium nanocrystals
This thesis research involves the development of theoretical foundations for studying the synthesis and structure of oxide-embedded silicon and germanium nanocrystals, by integrating various state-of-the-art theoretical techniques at different time and length scales. The primary focus was placed on (1) investigation of mechanisms underlying the formation of silicon and germanium nanocrystals in an oxide matrix and (2) development of kinetic models capable of predicting the structural properties of the silicon-germanium-oxide nanosystem under various processing conditions. The discovery of efficient room temperature luminescence has generated significant interest in silicon and germanium nanocrystals embedded in an oxide matrix because of their potential applications in electronic, optoelectronic, and optical devices in Si-compatible technology. Earlier experimental investigations have suggested the absorption and luminescence properties of the embedded nanocrystal systems would be governed by a complex combination of: nanocrystal sizes, shapes, and size distributions; crystal-matrix interface structures, bonding, and defects; and matrix structure and composition. This may imply that atomic-level control of such structural properties would offer great opportunities in the development of silicon/germanium nanocrystal based novel devices. However, even the fundamental mechanics of the growth and structure of embedded silicon and germanium nanocrystals are still unclear. Using multiscale modeling and simulation, we have identified mechanisms underlying the formation of silicon and germanium nanocrystals in a silicon-rich oxide matrix. Our multiscale approach combines: first principles quantum mechanical calculations of fundamental processes; Metropolis Monte Carlo simulations of amorphous structures; and kinetic Monte Carlo simulations of long-time scale growth. We find that the formation of oxide embedded silicon clusters is primarily attributed to a chemical phase separation to silicon and silicon dioxide, which is mainly driven by suboxide penalty, with a minor contribution of strain. The phase separation turns out to be primarily controlled by oxygen out-diffusion from silicon-rich regions, rather than excess silicon diffusion and agglomeration. From kinetic Monte Carlo simulations based on these fundamental findings we identify two growth characteristics: “coalescence-like” and “pseudo Ostwald ripening”. The simulation results agree well with experimental observations of strong dependence of the cluster size on the initial Si supersaturation and rapid formation of Si clusters at the early stages of annealing with very slow ripening. On the other hand, we find that the formation of gemanium nanocrystals in a silicon dioxide matrix is attributed to the diffusion and agglomeration of germanium precipitates. We have also determined the atomic structure and stability of embedded silicon nanocrystals and the mechanisms of silicon oxidation. While current experimental techniques are still limited to providing complementary atomic-level, real-space information, our comprehensive multiscale modeling based on first-principles quantum mechanics contributes greatly to understanding the fundamental behavior and properties of the silicon-germanium-oxide nanosystems.