Densities and Sizes of Self-assembled Quantum Dots Grown by MOVPE

Abstract: This thesis is based on results concerning the formation of semiconductor self-assembled quantum dots. The quantum dots have been grown by metal organic vapour phase epitaxy in the Stranski-Krastanow growth mode. This growth mode is based on a strain-induced transition from two-dimensional layer-by-layer growth towards three-dimensional island growth. The resulting morphology consists of coherent, size homogene, single-crystalline islands (self-assembled quantum dots) on a thin remaining wetting layer. The samples have been extensively characterized by atomic force microscopy (AFM), photoluminescence (PL) spectroscopy, and transmission electron microscopy (TEM). The investigated materials systems are InP on GaInP(GaAs), InAs on InP, and GaInP on GaP. Experimentally it has been verified that InP island formation on GaInP(GaAs) is nucleation controlled. The areal island density increases with decreasing deposition temperature and increasing deposition rate. For a constant amount of island material, there is an inverse relationship between island density and size. A post-growth annealing period, in the order of minutes, can make density independent control of island sizes possible. By annealing, the islands decrease in height and show a blueshift in PL emission. Investigations in the materials system InAs/InP have shown that the island formation can be very sensitive to substrates and surface preparation. Detailed investigations of GaInP islands on GaP have shown that these islands have an uncommon shape with faced top islands situated on flat base regions. Energy dispersive X-ray spectroscopy (EDS) has shown that the top islands are In rich (In/Ga=3/2) and that the base regions are slightly Ga rich. Theoretical models for the island formation are presented. These are based on rate equations and have been used to calculate InP/GaInP(GaAs) island densities at varying deposition conditions. Average island sizes in the same materials system have been calculated with a materials balancing model. With this model it is also possible to estimate the deposition conditions that give rise to bimodal island size distributions.