We have studied the low-temperature (T=6K) optical properties of a series of InGaN∕GaN single-quantum-well structures with varying indium fractions. With increasing indium fraction the peak emission moves to lower energy and the strength of the exciton–longitudinal-optical (LO)-phonon coupling increases. The Huang–Rhys factor extracted from the Fabry–Pérot interference-free photoluminescence spectra has been compared with the results of a model calculation, yielding a value of approximately 2nm for the in-plane localization length scale of carriers. We have found reasonable agreement between this length scale and the in-plane extent of well-width fluctuations observed in scanning transmission electron microscopy high-angle annular dark-field images. High-resolution transmission electron microscopy images taken with a short exposure time and a low electron flux have not revealed any evidence of gross indium fluctuations within our InGaN quantum wells. These images could not, however, rule out the possible existence of small-scale indium fluctuations, of the order of a few at. %.

Low-pressure MOCVD has been used to grow layers of inP, InGaAs, GalnAsP and quantum weii material on planar substrates patterned with silica masks. The thicknesses and, where relevant, the compositions of these selectively grown layers were investigated by optical and scanning electron microscopy, surface profiling, energy dispersive x-ray analysis, secondary-ion mass spectroscopy and spatially resolved photoluminescence. The epitaxial layers were found to be both thicker and richer in indium in the vicinity of a mask. The perturbations in the thickness and composition of material grown around a given mask pattern were independent of the orientation of the pattern with respect to the gas flow and the crystallographic axes of the substrate. Lateral movement of material from the masked regions to the surrounding areas was found to take place in the gas above the wafer surface. A gas-phase diffusion model based on Laplace's equation was used to analyse the thickness and compositional variations caused by selective growth. The emission wavelength of selectively grown InGaAsl GalnAsP MOW material was shifted by over 100 nm without degradation in emission efficiency. The lasing wavelength of Fabry-Pbrot lasers fabricated on such material was increased by a similar amount without degradation of threshold current.

Direct evidence for In-segregation in InGaN/GaN quantum-well structures is given via highly spatially resolved energy dispersive x-ray analysis performed in a dedicated scanning transmission electron microscope. The In fluctuations become increasingly pronounced in the vicinity of dislocations. The latter assist In diffusion and cause severe Ga/In intermixing.

We have determined the indium content and the layer thicknesses in an InGaN epilayer and InGaN/GaN quantum well structures by high-resolution x-ray diffraction (XRD) using the (002) reflection. The thickness of the total repeat (an InGaN well plus a GaN barrier) in the superlattice is easily determined from the spacing between the satellite peaks in an omega/2theta scan. Measurement of the individual thickness of InGaN and GaN layers and the indium content is less straightforward, since for multilayer structures the peak positions are influenced by both the indium content and the thickness ratio of the GaN to the InGaN layer. Thus, several different models may give reasonable fits to data collected over a limited range (about 1° omega/2theta either side of the (002)) showing only lower-order (−3 to +3) satellite peaks. Whenever possible, we have collected data over a wide range (about 4° omega/2theta) and determined the thickness ratio by examination of the relative intensities of weak higher-order satellite peaks (−7 to +5). An alternative but less sensitive method is to use relative intensities from x-ray reflectivity measurements to give the thickness ratio. Once the thickness of both the InGaN and GaN layers has been established, the InGaN composition can be determined from the peak positions. If the quality of the samples is poor, because of inhomogeneities or wide diffuse interfaces, there are insufficient data to characterize the structures. There is good agreement between the composition of the epilayer as determined by XRD and secondary ion mass spectroscopy and good agreement between x-ray and electron microscopy results for the quantum well structures. We find no variation from Vegard’s rule for In contents less than 0.20. This article shows that structural parameters of high-quality InGaN/GaN superlattices with 10 and 5 repeats can be determined reliably by x-ray techniques: The InGaN and GaN thicknesses to ±1 Å and the In content to ±0.01.

Photoluminescence and electroluminescence measurements on InGaN/GaN quantum well (QW) structures and light emitting diodes suggest that QWs with gross fluctuations in width (formed when, during growth, the InGaN is exposed unprotected to high temperatures) give higher room temperature quantum efficiencies than continuous QWs. The efficiency does not depend on the growth temperature of the GaN barriers. Temperature-dependent electroluminescence measurements suggest that the higher efficiency results from higher activation energies for defect-related non-radiative recombination in QW samples with gaps. At high currents the maximum quantum efficiency is similar for all samples, indicating the droop term is not dependent on QW morphology.