A. M. Dabiran

Structure and composition of GaN(0001) A and B surfaces

Homoepitaxial, GaN films on both c-plane surfaces of bulk GaN crystals were examined using reflection high-energy electron diffraction (RHEED). Differences in the RHEED pattern, time development of the RHEED intensity, and surface reconstructions were observed. The substrate surfaces were prepared either by mechanical polishing [GaN(0001)A] or by chemo-mechanically polishing [GaN(0001)B]. Then films were grown by molecular beam epitaxy; Ga was provide by a Knudsen cell and nitrogen from NH3. On the B surface, the Ga rich reconstructions reported by Smith and co-workers [Phys. Rev. Lett. 79, 3934 (1997)] were observed. On the A surface, a (2×2) reconstruction was observed. Both reconstructions were much sharper than those seen on GaN films grown on sapphire. RHEED measurements of the specular intensity vs time showed that two different surface terminations could be maintained on the B surface, one of which is a stable, gallided surface, while the other is a nitrided surface, which is unstable in vacuum. I...

N-Limited Versus Ga-Limited Growth on

GaN was grown on by MBE using NH3 and a Ga Knudsen cell. The growth kinetics on samples of this polarity were investigated with desorption mass spectroscopy (DMS) and reflection high energy electron diffraction (RHEED). Both techniques were used to observe and control surface termination, Ga condensation and surface temperature. GaN growth and decomposition rates were measured by DMS. Two stable surface terminations were found to exist — N-terminated and Ga-terminated . The N-terminated surface also contained hydrogen which desorbed during growth at a rate proportional to the growth rate. Low temperature reconstructions were only observed by adding weakly adsorbed Ga on top of the Ga-terminated surface. During growth two distinct growth regimes were identified: growth under excess NH3 and growth under excess Ga. Growth is limited in both regimes by GaN decomposition at high temperatures with an activation energy of 3.4 eV. Growth in the excess Ga regime ceased below the Ga condensation temperature. Under conditions of excess NH3, strong but damped oscillations in the specular RHEED intensity were observed on smooth surfaces. Contrary to previous suggestions, the period of these oscillations did not correspond exactly to integral layer deposition and was not characteristic of a narrow growth front. Further, the growth mode changed from island nucleation to step flow with an activation energy of 1.2 eV. Under conditions of excess Ga, the diffraction was 2-D but RHEED intensity oscillations were not observed, indicating a step flow growth mode. In this latter regime RHEED measurements were very sensitive to termination changes on the surface, and the growth rate was found to decrease linearly with increasing Ga flux. This reduction is explained by a model in which weakly adsorbed Ga blocks reaction at strongly bound Ga. A map is presented to provide a framework for categorizing the overall growth.

{\rm GaN}(000\bar{1})

Abstract The surface reconstructions of AlAs(100) layers grown by molecular beam epitaxy (MBE) on GaAs(100) were mapped as a function of substrate temperature and arsenic flux. Three main reconstructions were observed - a c (4 × 4) at lower temperatures and higher arsenic fluxes, a (2 × 4) at middle temperatures, and a (3 × 2) at higher temperatures and lower arsenic fluxes. Growth of AlAs on AlAs(100) is layer-by-layer for the high temperature and low temperature reconstructions. In the mid-temperature region, AlAs grows rough on (2 × 4) reconstructed AlAs(100) as indicated by rapidly damped reflection high-energy electron diffraction (RHEED) intensity oscillations and the appearance of three-dimensional (3D) features. The addition of fractional layers of Ga enhances the smooth growth of AlAs. A metastable (5 × 2) reconstruction was observed when a fraction of a layer of Ga was present on the surface. The results indicate that Ga segregates during the growth of AlAs on GaAs(100) at temperatures at least as low as 500°C, and that annealing at temperatures above 700°C removes most of the Ga from the surface.

by MBE Using NH3

We have studied the effect of Sn coverage on the growth of GaAs and strained InGaAs on GaAs(001) by reflection high‐energy electron diffraction (RHEED). Upon deposition of fractional monolayer (ML) amounts of Sn on the GaAs(001) surface, the (2×4) reconstruction disappeared and the diffracted beam intensities decreased. However, there was no change in the width of the beams nor was there an increase in the diffuse scattering—indicating an epitaxial Sn or tin—arsenide formation. This submonolayer coverage was found to enhance the layer‐by‐layer growth of GaAs and was found to remain at the surface longer than previously predicted. Most dramatically, strong beats in the RHEED intensity oscillations during GaAs growth were observed at a Sn coverage of ∼0.3 ML. The growth of GaAs on a vicinal (001) surface after the deposition of Sn, was found to order the step terrace lengths and to reduce the meandering of the steps. The vicinal (001) surface oriented toward [110] ultimately ordered more than the one oriented toward [110]. For InxGa1−xAs with x=0.36 or 0.54, the transition between the two‐dimensional and three‐dimensional growth modes was found to be independent of the Sn coverage. This last result indicates that at these mole fractions and under the growth conditions used here, Sn does not appear to act as a surfactant for the growth of InGaAs.

Surface reconstructions and growth mode transitions of AlAs(100)

Abstract Reflection high energy electron diffraction (RHEED) measurements were performed during the molecular beam epitaxial (MBE) growth of GaAs and InGaAs on GaAs(111) A and (111) B surfaces. Under a fixed Ga flux the period of these intensity oscillations was observed to increase with increasing As 4 flux on the 2 × 2 reconstructed GaAs(111) B surfaces. Layer thickness measurements, using cross-sectional transmission electron micrographs of AlAs/GaAs superlattices, indicated that the real growth rate did not correspond to the measured period of the intensity oscillations. The results are explained in terms of a reduction in Ga incorporation and an enhancement of Ga surface diffusion as the arsenic coverage of the 2 × 2 reconstructed (111) B surfaces is increased. The reduced Ga incorporation, on GaAs(111) B, promotes the formation of facets, commonly observed as three-dimensional islands or hillocks, which rob a portion of the Ga flux. The MBE growth and relaxation of strained InGaAs layers on GaAs(111) B were also studied by RHEED intensity oscillations and in situ surface lattice constant measurements. It is shown that by tuning the MBE parameters, during the growth of GaAs buffers and InGaAs layers on GaAs(111) B, premature strain relaxation due to the formation of twin defects can be prevented. Unlike the growth of InGaAs on GaAs(100) no two-dimensional to three-dimensional transition was observed even at high strains.

The effect of submonolayer Sn δ‐doping layers on the growth of InGaAs and GaAs

GaN(0001) films were grown by molecular beam epitaxy using ammonia and elemental Ga. The surface reactivity and growth kinetics of GaN(0001) were investigated as a function of growth parameters using desorption mass spectroscopy. Growth proceeds either by island nucleation or by step flow, depending on the steady state surface coverage of Ga. Three Ga adsorption states were found on the surface, one chemisorption and two weak states. One of the weak states corresponds to Ga adsorbed on a gallided surface, while the other corresponded to an intrinsic physisorption state on a hydrogen-passivated, nitrided surface. An abrupt growth mode transition between excess Ga and excess nitrogen was found as a function of growth parameters. The transition was modeled by rate equations based on growth at step edges and the three types of adsorption states.

Reflection high energy electron diffraction measurements of molecular beam epitaxially grown GaAs and InGaAs on GaAs(111)

The growth of AlGaAs has been known to exhibit a region of substrate temperatures where rough growth morphologies are observed. A number of mechanisms had been proposed to explain this forbidden window of growth temperatures. Here, these mechanisms are examined, with emphasis on the growth of the binary, AlAs, in order to rule out effects due to differences in diffusion and desorption of Ga and Al. Reflection high energy electron diffraction is used to monitor the growth of AlAs in both solid source and gas source MBE as a function of substrate temperatures. Even without Ga present, 3D features are observed at intermediate substrate temperatures. Above and below these temperatures, RHEED intensity oscillations are observed, indicating layer-by-layer growth. This reentrant behavior is observed when using As4 or when using As2 produced from arsine, but not using As2 produced from elemental arsenic. Oxygen is not observed to incorporate differently in AlAs layers grown in the different temperature regimes. Atomic hydrogen is observed to smoothen the growth morphology. To explain the results we consider potential barriers to adatom diffusion at step edges. The temperature boundaries of the reentrant behavior were measured versus the incident Al flux. The Al flux dependence of the boundaries is consistent with the existence of diffusion barriers at step edges.

A rate equation model for the growth of GaN on GaN(0001̄) by molecular beam epitaxy

During the growth of GaAs and AlAs on vicinal GaAs(100) by molecular‐beam epitaxy, reflection high energy electron diffraction was used to measure the transition temperature between two‐dimensional nucleation and pure step propagation when submonolayer amounts of Sn were present on the surface. On samples misoriented by 0.5° to either the [011] or the [011] direction, the transition temperature decreased by approximately 100u2009°C after the deposition of 0.6 monolayers of Sn, indicating that the Ga mobility increased. The presence of Sn also increased the surface mobility of the Al adatoms on AlAs(100) surfaces as indicated by the annealing behavior of the AlAs surface at 600u2009°C.

Al and Ga diffusion barriers in molecular beam epitaxy

The evolution of nanoscale pores or dimples during ion etching of GaN was used to measure the magnitude of the curvature-dependent roughening. GaN(0001) surfaces were ion etched with glancing-incident, 300 eV Ar and nitrogen ions using a beam flux of 3.6×1014ionscm−2s−1. The samples were rotated during the etching, and the sample temperatures maintained between room temperature and 600u2009°C. This etch process smoothened the surface but left nanoscale dimples or pores with diameters between 30 and 800 nm. The density of these dimples remained constant during the etch process but the dimples were observed to grow larger in size until coalescence occurred. The formation of these ion-induced, nanoscale features was analyzed in terms of a continuum model that included a curvature-dependent roughening term and a smoothening term. The integral of the removed material was measured in order to directly determine the curvature dependence of the sputter yield. From the evolution of the dimple dimensions, we measured t...

Enhanced surface cation mobility on Sn delta-doped (Ga,Al)As

Ion-beam-assisted molecular beam epitaxy was used to grow GaN on sapphire by reacting Ga from an effusion cell with ammonia. With the ion beam at low glancing angle and energy between 60–500eV, periodic, nanoscale ripple and dots were observed both with and without growth. By changing the growth parameters, the dimensions could be tuned from 40to800nm. The ripple wavelength was analyzed in terms of continuum models. However, the time constant for pattern formation was several orders of magnitude less than that predicted by linear or nonlinear theories. The mobile adatom concentration was calculated and found to be quantitatively reasonable, increasing with increasing net growth rate. The ripple wavelength was observed to be well developed for nitrogen ion beams, but not with Ar ions within our energy and flux range. Adding growth reduced the measured wavelength rather than increasing it.