The optoelectronic properties of quantum dot (QD) devices are influenced essentially by the size, shape, crystal morphology, chemical composition and strain of the QDs. During the capping process the strain, of necessity, changes because of the different atomic sizes of the species in the QD and the capping layer. The capping process also changes the QD shape and composition. In the present study, single layer InAs/GaAs (001) QD structures with different GaAs capping layer thicknesses have been fabricated by MBE growth. The different techniques of transmission electron microscopy (TEM), including diffraction-contrast and high-resolution lattice imaging (HRTEM) as well as analytical electron microscopy (AEM), have been used to characterise both the microstructure and microchemistry down to the atomic scale. The atomic scale details obtained by HRTEM are combined with the results of field-emission gun scanning transmission electron microscopy (FEG-STEM) to provide information on the structure and composition change of the QDs during their encapsulation by GaAs.

We show that the morphology of the initial monolayers of InP on Al0.48In0.52As grown by metalorganic vapor-phase epitaxy does not follow the expected layer-by-layer growth mode of lattice-matched systems, but instead develops a number of low-dimensional structures, e.g., quantum dots and wires. We discuss how the macroscopically strain-free heteroepitaxy might be strongly affected by local phase separation/alloying-induced strain and that the preferred aggregation of adatom species on the substrate surface and reduced wettability of InP on AlInAs surfaces might be the cause of the unusual (step) organization and morphology

In this letter we report on the formation of long-range surface disorder features during the growth by molecular beam epitaxy (MBE) of homoepitaxial GaAs (001) films having the β2(2×4) reconstruction. Observations were made in real-time at the growth temperature using reflection high energy electron diffraction (RHEED) and analyzed kinematically. We show that kinks (cooperative shifts of whole columns of 2×4units along the [110] direction) form rapidly as growth commences and that the antiphase domain structure present on the substrate prior to growth as a result of the arrangement of As–As dimers persists. This produces a surface with two types of long-range disorder. We speculate on the role of incident Ga atoms on this process.

The homoepitaxial growth of GaAs by hydrogen-assisted molecular beam epitaxy (H-MBE) on (110) substrates vicinal to (111)A has been studied by reflection high energy electron diffraction (RHEED) and atomic force microscopy (AFM) for different kinetic regimes. When the GaAs growth rate is limited by the kinetics of adatom incorporation to steps, the presence of chemisorbed H on the surface after oxide removal promotes the incorporation of adatoms to steps from the lower terraces, leading to the formation of multiatomic step arrays or ridge patterns by a combination of step propagation and two-dimensional layer-by-layer growth. Supply of atomic H during epitaxy favours three-dimensional growth, leading to Ga-induced surface roughening or mound formation. At high temperatures, the Ga–As interactions at step edges are faster and stable growth of GaAs occurs by step propagation, leading to a faceted surface when H is used both during oxide removal and/or MBE growth.

Molecular beam epitaxy (MBE) is an ideal technique for in situ studies of the growth mechanisms of thin films and other low-dimensional structures. There is an ultra-high-vacuum (UHV) environment, so it is compatible with many surface-science techniques, and the incident fluxes can be very accurately controlled. In this paper, I will summarize the application of reflection high-energy electron diffraction (RHEED) and scanning tunneling microscopy (STM) to the homoepitaxial growth of GaAs (0 0 1) films and the growth of InAs on low-index orientation GaAs substrates. I will emphasize the importance of establishing an interpretation that is consistent with results from both methods and also demonstrate the necessity of combining the experimental approach with theoretical treatments, both simulation and analytical.

Abstract By carrying out Monte Carlo simulations based on the two-species atomicscale kinetic growth model of GaAs(001) homoepitaxy and comparing the results with scanning tunneling microscope images, we show that initial growing islands undergo the structural transformation before adopting the proper β2(2 × 4) reconstruction. In developing optoelectronic devices, GaAs(001) is often used as the basis substrate for fabrication 1). To study the atomic-scale growth kinetics of this surface, extensive use has been made of ab initio calculations with the energy-minimization procedure 2). With this method, however, it is actually difficult to elucidate the true growth kinetics right at the growth temperatures, because growth proceeds in non-equilibrium conditions there. For example, by investigating the stable sites for Ga adatoms and assuming arsenic species to stick onto them, it is deduced in Ref. 2) that the nucleation of islands is initiated by the sticking of Ga adatoms at the trench sites of the GaAs(001)-β2(2×4) structure, which is depicted in Fig.1. If this growth

Scanning tunnelling microscopy has been used to investigate dislocation-induced surface morphological changes during strain relaxation in the two-dimensional (2D) growth of InAs on GaAs(110) by molecular beam epitaxy. Two distinct classes of dislocation are required owing to the crystallographic anisotropy in the (110) plane: ideal edge dislocations (∼3 ML) and dislocation half-loop slip (>5 ML) systems. Specific emphasis is on the nucleation of the edge dislocations out of the preceding pseudomorphic layer and the manner in which the slip steps accommodate the continuing growth of the epilayer and influence the surface morphology. Between 1 and 2 ML InAs thickness, a substantial redistribution of the surface material occurs, leading to highly uniform "mosaic" structures, which are either close-packed arrays of tiny islands at ∼420°C, or a linear array at ∼480°C. The closure of these fractured morphologies directly incorporates edge misfit dislocations beneath the original boundaries between the surface islands. Since a slip mechanism cannot operate for [11̄0] strain relief, the dislocations must be located directly beneath the surface, a layer or so above the InAs–GaAs interface. Distinctive step signatures due to the slip of surface-nucleated half-loop dislocations, in terms of the screw terminations and their associated wave like topological profiles, are observed after 5 ML InAs deposition. The growth mode for the InAs layer beyond 5 ML is by propagation of the slip steps, with discrete 2D island nuclei never observed. The slip steps can therefore move across the surface in the [001] direction away from their original position, as well as increasing in length as the dislocation half-loops expand along the [11̄0] direction. The linear density of slip steps along [001] decreases with increasing film thickness due to interaction of the steps during growth. The length of each slip step and of each misfit segment is relatively short (≤1000 Å) in comparison with growth on (001) substrates. A related interaction between the half-loop and preceding edge dislocations at the interface is also resolved.

Reflectance anisotropy (RA) oscillations observed during the molecular beam epitaxial growth of AlAs and AlGaAs/AlAs heterostructures on singular GaAs(001) substrates have been investigated. A temperature dependence study of the AlAs oscillations was carried out in the temperature range 500-580 degrees C and oscillations of large amplitude were readily detected. The effect of alloy stoichiometry on the oscillation amplitude was investigated and the amplitude was found to decrease as the Ga content of the alloy was increased. The origin of these oscillations is discussed in terms of the microscopic nature of the step edges associated with the islands formed during growth. (C) 1997 Elsevier Science B.V.

Dynamic reflectance anisotropy and reflectance data recorded using a photon energy of 1.96 eV is presented for the deposition of atomic Si on singular GaAs(001)-c(4 × 4) surfaces under molecular beam epitaxy conditions at 400, 450 and 500°C. Changes in reflectance anisotropy were detected at the 0.1% of a monolayer level underlining the sensitivity of the method. Dynamic changes were found in the reflectance anisotropy response upon the interruption of dosing providing a caveat that static reflectance difference spectroscopy does not provide true dynamical information on the surface processes and that temporal information on surface migration is available. Evidence for dramatic changes in reflectance was found during deposition, these changes being particularly pronounced at higher temperatures and are an indicator of surface roughening processes. Subsequent growth of GaAs returns the surface to GaAs like behaviour after ∼ 15 monolayers and provides real time evidence for Si segregation at the surface.

Reflection high energy electron diffraction (RHEED) intensity oscillations have been used to obtain the arsenic incorporation coefficient for the growth of GaAs on GaAs(110) by molecular beam epitaxy (MBE) using either As-2 or As-4 as the source of arsenic. In both cases, the incorporation coefficients decrease with increasing growth temperature, with the values obtained for As-2 being approximately twice those of As-4. The results are modelled within a kinetic scheme based on the assumption that the incorporation processes are precursor mediated and in both cases involve a molecularly adsorbed As-2 intermediate. The difference in the activation energies for the desorption and incorporation of the As-2 intermediate is the same when using either As-2 or As-4. The implication is that the final incorporation step in the growth of GaAs(110) is independent of the arsenic species used.

Scanning tunnelling microscopy (STM) and reflection high energy electron diffraction (RHEED) have been used to study the deposition of Si below 400‡C onto GaAs (001) surfaces grown in situ by molecular beam epitaxy (MBE). The emphasis is on the island formation and growth, as well as surface ordering, for submonolayer quantities of Si (up to 0.2 ML) deposited on two different As-rich reconstructions of GaAs (0 0 1) (2 × 4) and c(4 × 4). For deposition on the c(4 × 4) surface, an asymmetric (3 × 1) RHEED pattern is formed, a consequence of anisotropic “needle-like” islands, which grow adjacent to each other along the [1 1 0] direction and produce a three-fold periodic superstructure. Individual islands grow by a site exchange process in which the additional As layer of the c(4 × 4) structure acts as a surfactant and enables the Si atoms to occupy Ga sites in the GaAs lattice. In contrast, Si deposition on the (2 × 4) surface does not lead to any new surface periodicities as monitored by RHEED. The Si atoms form poorly ordered clusters distributed randomly across the surface. The site exchange process cannot occur in this case as the (2 × 4) surface is terminated with only one layer of arsenic. Instead, the Si atoms occupy sites on top of the outer arsenic layer.

Silicon-germanium/silicon (Si1 − xGe x /Si, x < 0.50) multiple quantum wells (MQWs) have been grown on (001) Si substrates by gas source molecular beam epitaxy (GSMBE) using disilane (Si2H6) and germane (GeH4) as source gases. Their structural properties have been evaluated by X-ray diffraction (XRD), rocking curve techniques and transmission electron microscopy (TEM). For the substrate temperatures used in this work (450 ‡C to 520 ‡C) the Si growth rate is limited by hydrogen desorption kinetics, whereas the growth of SiGe is limited primarily by the arrival rate of the source gases onto the Si substrates. XRD analysis of the structures indicates a significant well plus barrier period variation of approximately 5–10%, attributed to fluctuations in the substrate temperature during growth, since these cause significant variations in the growth rate of the Si barriers. For x < 0.30 we find nearly ideal Si/SiGe interfaces as determined from a comparison of the XRD data with dynamical simulations of the 004 X-ray reflectivity, although TEM micrographs indicate that the x = 0.30 samples exhibit undulations in the first SiGe/Si interface of the structures. For x = 0.50 such undulations occur throughout the MQW structure; the undulation amplitude decreases with decreasing growth temperature but the period remains unchanged. The observed improvement in the SiGe/Si interface planarity at lower growth temperatures is attributed to a reduction in the surface diffusion of Si and Ge with decreasing growth temperature.

The growth dynamics of GaAs, AlAs and (Al, Ga)As films grown by molecular beam epitaxy (MBE) on GaAs(110) and (111)A substrates have been studied using reflection high energy electron diffraction (RHEED) intensity oscillations and scanning tunnelling microscopy (STM). In contrast to growth on (001) oriented substrates, the period of the RHEED intensity oscillation does not in general provide a measure of the growth rate. This is explained by the very different surface chemistry involved, since the short lifetime of arsenic molecules (As2 or As4) on non-(001) surfaces results in cation-stable surface conditions, which generate arsenic (anion)- induced intensity oscillations, whereas on (001) surfaces they are cationinduced under all normal growth conditions. The effects of this behaviour on surface morphology are illustrated, as are the relative influences of Ga and Al. STM images obtained during the first few monolayers of growth provide a detailed indication of the growth mode and in particular explain in a simple manner the origin of bilayer period RHEED intensity oscillations obtained during growth on GaAs (110).

We have made a systematic study of the effect of growth conditions on Si incorporation in GaAs layers grown on GaAs (111)A substrates. We show that it is dominated by the low incorporation coefficient of As4 on the GaAs (111)A surface. The site occupancy of the Si has been shown by local vibrational mode spectroscopy to be on Ga and As lattice sites to provide donor and acceptor character, respectively. The doping behaviour of Si for specific growth conditions may be predicted from the product of the As4:Ga flux ratio and the As4 incorporation coefficient at the growth temperature. The growth conditions for which Si acts as an acceptor produce films with poor surface morphology.

Molecular beam epitaxy (MBE) growth of GaAs buffer layer and subsequent deposition of InAs layers on GaAs (110) misoriented by 1.5° towards (111) A are studied using transmission electron microscopy. The observations show that step bunching occurred during the buffer layer growth. InAs is found to nucleate and grow on the bunched step edges.

We have realized two dimensional electron gases (2DEGs) in tensile strained silicon (Si) channels between strain relaxed silicon germanium (Si0.70Ge0.30) barriers grown on Si(100) substrates by Gas Source Molecular Beam Epitaxy (GSMBE). Disilane (Si2H6), germane (GeH4), and arsine (AsH3) are used as the source gases. Compositionally graded buffer layers with a linear gradient of 30%; Ge/1 Μm relax the strain of the Si0.70Ge0.30 barrier layers by an amount greater than 95%; as determined from X-ray diffraction (XRD) rocking curves. Dislocation densities in the vicinity of the active strained Si channels are below 107 cm−2 as determined from transmission electron microscopy (TEM) measurements. These structures have low n-type background impurity concentrations ( < 1016 cm−3) and the Si0.70Ge0.30 barriers can be successfully doped with a unity activation ratio in the 1017 to 1020 cm−3 range. At present, we obtain 300 K (0.4 K) electron mobilities and sheet densities in our 2DEGs of 103 (5.3 × 104) cm2/Vs and 3 × 1012 (5.2 × 1011) cm−2, respectively. A discussion of the requirements for growing these structures by GSMBE and the modifications needed to improve the transport properties of the 2DEGs is presented.

This paper reviews our present understanding of particular aspects of the surface processes involved in the growth of epitaxial semiconductor films by molecular beam epitaxy. Emphasis is placed on adatom migration and incorporation on GaAs (001) substrates during the growth of GaAs, a comparison with equivalent growth effects on (110) and (111)A oriented substrates, and the influence of mismatch and substrate orientation on growth mode and strain relaxation in the InAs/GaAs system. A brief indication of surface segregation behaviour is also included.

Behavior of Sn as donor species in the MBE growth of GaAs on (111)A substrates has been investigated by varying the growth temperature from 460 to 620°C, As4:Ga flux ratio from 4 to 25, and Sn concentration from 1016 to 1020 atoms cm-3. Secondary ion mass microscopy measurements show that Sn does not surface segregate on (111)A substrates under this growth condition, in contrast to that on (001) substrates. Sn is uniformly incorporated throughout the bulk of the grown layer for all samples, apart from the most highly doped ones. To increase the Sn carrier concentration on the (111)A substrates, the measured carrier concentration shows that doping should be carried out at a low growth temperature and/or high As4:Ga flux ratio.

We have utilised the technique of reflectance anisotropy (RA) to study the growth of GaAs, AlAs, and InAs on (001) GaAs substrates by molecular beam epitaxy and simultaneously recorded both the RA response and the surface electron diffraction pattern obtained using (RHEED). For macroscopic coverage variations the change in the RA response in all cases appears to correlate exactly with the appearance of a new RHEED pattern. In this respect RA growth "oscillations" are observed which correlate exactly in period with the usual RHEED oscillations observed during growth and can be made to be in-phase with the RHEED oscillations by varying the RHEED collection system. In the case of InAs growth, much more subtle (ca. 0.1 monolayers) coverage variations are easily detectable by RA while the macroscopic (2×4) to (4×2) phase transition which occurs for this system at InAs coverages of ca. one monolayer produces a marked change in both the RA and RHEED response. Since all three systems were studied using the same optical excitation energy (ca. 1.92 eV) and since in all cases the change in the RA response occurred concurrently with the change in the RHEED pattern we question the currently popular theory regarding the nature of the RA response in relation to a supposed local surface dimer state resonant absorption and suggest that geometric factors or longer range surface state resonances may be of equal if not more importance.