Quantum Dots: Electronic Coupling and Structural Ordering

Authors

Glenn S. Solomon Department of Electrical Engineering, Stanford University

Publication Date

4/13/04

Read full article online

Full Article

Abstract

Current epitaxial crystal growth techniques, with their precise monolayer (ML) control, have led to abrupt heterointerfaces in III–V and group IV semiconductors. This remarkable heterointerface control is responsible for 1-D carrier confinement in the growth direction: When a thin layer is formed from a more narrow bandgap material in the larger bandgap host, a quantum well (QW) is formed. In the III–V material system, this QW has had a dramatical impact in both semiconductor research and mainstream semiconductor technology. Although significant effort has been concentrated toward extending this control to 2-D and 3-D confinement with quantum wires and quantum dots, the results have been encouraging but not resounding. This is generally because lithography techniques typically used to provide increased lateral confinement do not have the monolayer resolution that is available through epitaxial growth techniques. Furthermore, because of interfacial damage, it is difficult to directly pattern active regions using processing. Although useful structures can be fabricated using surface patterning, these techniques are not suitable for all structures. In the 1990s, a purely epitaxial technique was developed to produce quantum dots in the InAs/GaAs and Ge/Si semiconductor systems. This technique utilizes the strain-induced islanding of the Stranski–Krastanow (SK) growth mode, in which the growth surface islands compensate for the increase in energy caused by extra interface surface with a decrease in accumulated strain energy.

The formation of quantum dots by strain-induced islanding has provided a simple, lithography-free method to produce dense ensembles of quantum dots. Unlike the classical self-assembly processes in nonepitaxial systems, in this system, the energies associated with the epitaxial growth process still dominate those that drive the nanostructure formation process. The result is nanostructure features that are not identical but still similar. Strain-induced island formation is perhaps more akin to other surface and interface phenomena such as surface spinodal decomposition, surface reconstruction, and ledge-and-step formation. In fact, our general observation is that the ensemble uniformity and spatial periodicity of InAs islands follow more closely these processes than the self-organized formation of more classical structures such as carbon nanotubes or self-assembled protein structures. Unfortunately, the dominance of the epitaxial process and the large surface migration processes common in this growth lead to large inhomogeneous island size distributions: The spectral features are broadened with respect to QWs, and the narrow, atomic-like transitions are lost in the ensemble broadening. Nevertheless, the association of self-organization with strain-induced islanding, specifically with respect to direct bandgap semiconductors, has done much to focus attention on the possible utility of this system. Although the phenomenological process of strain-induced islanding was observed 60 years ago, the association of spectral features with these islands was made by Tabuchi et al., whereas the general possibilities of this system have been noted, developed, and championed by Leonard et al. This research is encouraging and lively, and with further fine-tuning of the growth processes, as was needed in the development of the successful QW technology, inhomogeneous broadening can be reduced.