Abstract
This dissertation describes investigations of structural ordering in the Si-Ge material system from nano-to-meso scale in order to modify electronic and thermal transport properties for potential improvement in thermoelectric or nanoelectronics applications.
We first present the synthesis and characterization of mono-layer scale chemically ordered Si1-xGex alloys. Here, metastable ordering to an L11-like structure, occurring during heteroepitaxial growth of Si1-xGex thin film alloys on Si(001) and Ge(001) substrates, is investigated. A parametric study was performed to study how strain, surface roughness, and growth parameters affect the order parameter during the alloy growth. The order parameter for the alloy films was quantified carefully using x-ray diffraction, taking into account an often-overlooked issue associated with the presence of multiple spatial variants associated with ordering along equivalent <111> directions. The sometimes contradictory roles of strain, extended surface roughness and surface steps in dictating the observed order parameter is discussed.
For the second part, we present our comprehensive investigation on the directed self-assembly of Si-Ge alloy quantum dots (QD) on patterned Si surfaces with variable morphology. Coherently strained Si1-xGex QD self-assemble during epitaxial growth on Si(001) substrate via the Stranski-Krastanov growth mode, typically with a broad size distribution and spatial disorder. For device applications control over the QD position and size distribution can be realized by templated self-assembly, where a lithographic technique is used to precisely define the nucleation sites and improve the size homogeneity. The templated QDs can then be used as 2D seed “crystals” to propagate the dots into third dimension through subsequent multilayer growth to form quantum dot mesocrystals (QDMC). QDMCs have their own synthetic properties related to the mesocrystal structure, such that by manipulating the interdot spacing and the size of the dots of a QDMC structure, we can potentially tune the transport properties. This work develops both an enabling synthesis technology for QDMCs, and generates better fundamental understanding of self-assembly directed by pre-existing modulation of the substrate surface morphology. The multilayer growth is not covered here; however, the research serves as a core foundation to the ongoing efforts in our group for QDMC synthesis along with characterization of the resultant electronic and thermal transport measurements.
For the successful growth of a QDMC structure, the initial underlying 2D template or the substrate pattern morphology plays a critical role. Therefore, the primary challenge was to develop a process and methodology to create highly uniform pit-patterned 2D templates on Si(001) substrates. Specifically, the challenges of patterning with focused-ion-beam (FIB) and electron beam lithography (EBL) based approach will be discussed. Significant effort was made to avoid extrinsic biases to surface diffusion by residual defectivity and contamination during the pit-patterning of Si substrates. Growth of Si1-xGex QDs on these pit-patterned Si(001) substrates was investigated. Growth of QDs on such patterned substrates is reported to yield precise positioning and improved size homogeneity. However, contradictory results are found regarding QD site selection on patterned substrates. A large body of both theoretical and experimental work has been published to study QD evolution on patterned substrates, but a unified analysis is still lacking regarding the site-selection of QDs on modified substrates. Here, we investigate QD site-selection on a patterned Si(001) substrate as a function of underlying substrate pattern morphology from pit-in-terrace to quasi-sinusoidal. The pit-in-terrace morphology leads to well-ordered QDs centered in the pits over a wide range of pattern wavelengths. However, for a quasi-sinusoidal morphology, when the pattern wavelength is twice the intrinsic wavelength, QDs suddenly bifurcate and shift to form in every saddle point, with high uniformity in size and site occupancy. Comparison of our results with existing models of QD formation on patterned surfaces will be discussed. Further, we use a relatively unusual approach to growing the Si-Ge QDs, in which a conformal layer is grown at low temperature, and then annealed in situ to promote surface diffusion and self-assembly. This provides additional control over the size of the QDs, and turns out to lend additional insights into QD growth kinetics, Ostwald ripening, and morphological transitions in the presence of an underlying surface modulation.
