Directed Self-Assembly of 3D Ge/Si Quantum Dot Epitaxial Crystals for Thermal and Electronic Applications

(click to enlarge)

Generously supported by the II-VI Foundation and now the NSF

PI: Jerry Floro
Graduate Student: Christopher Duska
Project Status: transitioning to NSF funding

Goals: This research employs directed self-assembly to create spatially-ordered arrays of epitaxial quantum dots that are periodic in all three dimensions, with lateral periodicities of 10-100 nm. These quantum dot mesocrystals (QDMC) could represent a new class of material with tailored bandstructure and pre-imposed symmetry that will extend our ability to control and manipulate the transport of both electricity and heat in semiconductor devices.

Technical Approach: Heteroepitaxial growth of Ge on Si (001) by molecular beam epitaxy (MBE) leads to strain-induced self-assembly of quantum dots. Formation of the quantum dots is either spatially random, or in some cases weakly periodic, see Figure 1a. Growth of Si/Ge multilayers can lead to a quantum dot superlattice, where dots vertically align in chains (guided by the propagating strain fields of the underlying quantum dot layer), but still have only weak in-plane alignment (Figs. 1b and 1c). In principle, 3D crystals can be grown if the initial layer has imposed 2D ordering. We employ focused ion beam (FIB) patterning to pre-template the substrate surface, thereby dictating preferred nucleation sites. In this way, subsequent layers should replicate this order, leading to QDMCs (Fig. 1d), where we can tailor the size and shape of the unit cell through the FIB pattern and the interlayer spacing. If we are successful, we will examine the thermal and electronic transport in the QDMCs using photoluminescence and time domain thermoreflectance (with the Hopkins group).

Key Results: We readily obtain 2D patterned seed crystals with 50 nm pitch (Fig. 2a), and have had some success down to 35 nm interdot spacing. We find that Ge QDs are dome-like clusters; after selectively etching the QDs away, it is clear that they formed on the raised “crown” region at the center of a 2D square unit cell of pits (Figs. 2b and 2c). This runs counter to most established theories, and many experiments and may result from the fact that the pattern spacing is approaching the intrinsic size for Ge QDs established by the balance between strain and surface energy. Alternatively, we are currently examining the residual defectivity at the FIB sites, which might reduce Ge sticking. We have made double-dot layers with good pattern fidelity (Fig. 3), but the pattern rapidly degrades above 2 layers, and we are working to understand how to better optimize the process.