This dissertation demonstrates the growth and optical characterization of ordered InAs/InP quantum dot (QD) arrays grown by chemical-beam epitaxy (CBE). The creation of InAs/InP QD arrays is governed by self-organized anisotropic strain engineering of InAs/InGaAsP superlattice (SL) templates leading to the formation of linear ordered one-dimensional (1-D) InAs QD arrays on InP (100) substrates and a periodic square lattice of two-dimensional (2-D) InAs QD arrays on InP (311)B substrates. The photoluminescence (PL) emission of the InAs QD arrays reveals excellent optical quality up to room temperature (RT). The emission wavelength is tuned into the technological important 1.55-µm region for telecom applications through the insertion of ultrathin GaAs interlayers beneath the QDs. Electronic coupling in linear InAs QD arrays on InP (100) is revealed by temperature-dependent PL and polarization-dependent PL measurements. Identical emission wavelength of multilayer-stacked linear QD arrays is achieved by increasing the GaAs interlayer thickness in successive layers. Self-assembled semiconductor QDs have led to numerous device applications ranging from nanophotonics, nanoelectronics to quantum information processing due to their three-dimensional carrier confinement. Owing to the modern crystal growth fabrication technologies, e.g. molecular-beam epitaxy (MBE), metal-organic vapor phase epitaxy (MOVPE), and CBE, QDs and related quantum structures can be prepared precisely at the atomic scale leading to the ultimate control of the QD shape, size, composition, and emission wavelength. In addition, control of the QD nucleation sites and QD ordering is the prerequisite for exploitation of new functionalities for novel quantum devices. The most common technique to position the QDs is by using artificially patterned substrates where the pattern geometry is defined by electron-beam lithography. The main disadvantages are, however, the degradation of the structural and optical properties due to lithographic imperfections and etching defects. To tackle this obstacle, a novel technique for the lateral alignment of QDs was realized based on self-organized anisotropic strain engineering creating ordered strain-modulated SL template structures. In this study the concept is further developed and transferred from the previously investigated GaAs-based system by MBE to the InP-based system by CBE where the QD emission wavelength is more suitable for telecom device applications. For the InAs/InP system, usually the formation of various nanostructures such as elongated QDs, called "quantum dashes" (QDashes) or even quantum wires (QWRs) is observed at practically identical growth conditions. Most probably this is due to the low lattice mismatch between InAs and InP of 3 % as compared to InAs and GaAs of 7 % and the strong As/P exchange reaction. In this study, we identify the surface morphology of the buffer layer as key parameter for the formation of InAs QDs or QDashes on lattice-matched InGaAsP on InP (100) substrates. Growth conditions leading to the formation of QDashes are always accompanied by a rough buffer layer morphology. Although other growth parameters such as higher growth temperature, larger As flux, and compressive buffer layer strain favor the formation of QDs, once, the buffer layer has a rough morphology, QDashes are formed during InAs growth. On the contrary, well-shaped and symmetric QDs are reproducibly formed on smooth buffer layers. Hence, we conclude that not the growth conditions during InAs depositions, but rather the related surface morphology of the buffer layer determines the formation of either QDs or QDashes, which both exhibit high optical quality. On smooth buffer layers, laterally ordered linear 1-D InAs QD arrays and a periodic square lattice of 2-D InAs QD arrays are formed by self-organized anisotropic strain engineering of InAs/InGaAsP SL templates on InP (100) and (311)B substrates, respectively. The SL template formation comprises InAs QD growth, thin InGaAsP capping, annealing at higher temperature, InGaAsP overgrowth, and stacking. This produces wirelike InAs nanostructures along  on InP (100) and spot-like InAs nanostructures oriented along ¿45° off [-233] on InP (311)B substrates due to anisotropic adatom surface migration during annealing and lateral/vertical strain correlation during stacking. The orientation of the linear InAs QD arrays is determined by the elastically soft directions of the InP crystal and the substrate miscut. InAs QD ordering is governed by local recognition of the lateral strain field modulations on the SL template surfaces. The growth parameters for obtaining straight, well-ordered, and uniform QD arrays are optimized, such as InGaAsP cap layer thickness, annealing temperature, InAs amount and growth rate, and number of SL periods. The InAs QD arrays reveal strong PL emission up to RT. This is the advantage of self-organized QD ordering compared to methods based on artificial substrate patterning which often degrade the optical quality. The emission of InAs QD arrays is tuned into the important 1.55-µm telecommunication wavelength region by the insertion of ultrathin, 0.8 – 2.0 monolayers (ML), GaAs interlayers beneath the QDs which suppress As/P exchange during InAs QD formation. For the linear InAs QD arrays on InP (100) substrates, lateral electronic coupling of the QDs along the chains is observed as indicated by temperature dependent PL measurements and the linear PL polarization. The concept of self-organized anisotropic strain engineering for QD ordering has been extended for formation of more complex architectures of lateral QD arrays by combining it with step engineering on artificially patterned InP (100) and (311)B substrates. On shallow- and deep stripe-patterned InP (100) substrates, depending on the stripe orientation, the linear 1-D InAs QD arrays are rotated away from their natural direction due to the presence of vicinal stepped sidewall planes modifying the self-organization process, coexisting with QD free steep side facets on deep-patterned substrates. On shallow- and deep-patterned InP (311)B substrates only QD free side facets form with flat top and bottom areas, not affecting the natural ordering of the 2-D InAs QD arrays. On the deep-patterned substrates a row of dense QDs forms on top along the side facets due to their slow-growing behavior. In the last chapter, multilayer-stacked linear InAs QD arrays on InAs/InGaAsP SL templates on InP (100) substrates are realized. Identical emission wavelength of the stacked QD arrays in the 1.55-µm region at RT is achieved by increasing the thickness of the GaAs interlayer beneath the QDs in successive layers. The sub-ML increment of the GaAs interlayer thickness compensates the QD size/wavelength increase during vertical strain correlated stacking. This is the demonstration of a three-dimensionally self-ordered QD crystal with fully controlled structural and optical properties. Finally, further investigations of the physical and optical properties of the 1-D and 2-D InAs/InP QD arrays are envisioned for future research. Cross-sectional scanning tunneling microscopy (X-STM) or cross-sectional transmission electron microscopy (X-TEM) is useful to obtain more information about material intermixing, geometry of the capped QD layers in the SL template structure to provide deeper insight into the growth mechanism and give input for possible theoretical models. Advanced optical measurement techniques, e.g., (time-resolved) single QD PL spectroscopy or magneto PL spectroscopy could further indicate the electronic coupling in the 1-D QD arrays. Moreover, selective-area epitaxy could lead to the formation of localized QD arrays on microscopic mesa patterns with complex arrangements. The growth rate enhancement in selective-area epitaxy by CBE on dielectric masked substrates, e.g., Silicon Nitride (Si3N4) or Silicon Dioxide (SiO2), is negligible compared to that in MOCVD due to desorption of group-III molecules (TMI, TEG) on masked areas. This allows the growth parameters to be controlled similar to those for unmasked planar substrates. The localized QD arrays could be useful for future applications such as quantum information processing and quantum computing.
|Qualification||Doctor of Philosophy|
|Award date||15 Jun 2010|
|Place of Publication||Eindhoven|
|Publication status||Published - 2010|