Towards Directed Self-Assembly of Quantum Dot Mesocrystals of Ge/Si Using Focused Ion Beam Patterning

A quantum dot mesocrystal (QDMC) is a three-dimensional, spatially ordered array of quantum dots (QDs) epitaxially embedded in a matrix material. By manipulating the QD array, e.g. size, periodicity, symmetry and chemical composition, we can potentially tailor the electrical, thermal, optical and magnetic properties of the crystal. While the dots themselves possess size-dependent properties due to electronic confinement and discrete energy states, bringing the quantum dots close enough together on the nanometer scale can result in electron wavefunction overlap and extended state or even miniband formation. Control of these properties can lead to potential applications in optoelectronics, nanoelectronics and thermoelectrics.

The strategy for forming ordered three-dimensional QDMCs of Ge on and in Si (001) is to encapsulate a well-ordered two-dimensional array of Ge QDs. Self-assembly of the ordered 2D array is directed by a surface topographical pattern pre-imposed on the substrate. Embedded Ge QDs create an inhomogeneous strain field in the Si encapsulation layer, onto which additional Ge is deposited. The strain field drives the nucleation to sites directly over the embedded QDs, thereby replicating the layer underneath. Repeating this process can result in a highly ordered QDMC. The challenge is to achieve this at lengthscales near or below the intrinsic lengthscale of the QDs themselves. The latter is dictated by the competition between elastic and surface energy.

The first part of the thesis discusses the directed formation of Stranski-Krastanow Ge QDs by molecular beam epitaxy (MBE). Highly uniform arrays are produced using a Ga+ focused ion beam (FIB), in conjunction with wet chemistry, to create surface morphology conducive to QD growth at ion doses considerably less than previously reported. This enabled successful self-assembly down to at least 50 nm interdot spacing, which is much less than previously achieved using FIB. Localization of the QDs can be achieved by creating topographical features, e.g. pits, that lower the barriers to formation through geometry, strain, or surface energy anisotropy. At the optimal dose, FIB patterned substrates at 50 nm create a template that nucleates QDs with a narrow normalized volume distribution width, with very high occupancy. The formation of the QDs is not in the pits, which is typical, but in the four-fold “crown” region in between the pits. We discuss the competing mechanisms that drive the nucleation of the QDs. The effect of buffers, deposition thickness, and temperature are explored.

The second part of the thesis discusses the formation of three-dimensional QDMCs on the 2D seed layers of Ge QDs. While atomic force microscopy suggests that high-fidelity multilayer replication of the initial, patterned layer can occur, cross-section transmission electron microscopy reveals complex behavior. For example, in several instances, extreme pattern morphology was completely undetected by the AFM. Although these morphologies are not optimal for QDMC formation, they also provide interesting insights into the propensity for self-assembly in this system. The limitations of the FIB for patterning at 50 nm and below are discussed.