Efficiency droop in zincblende InGaN/GaN quantum wells.

The decrease in emission efficiency with increasing drive current density, known as 'droop', of c-plane wurtzite InGaN/GaN quantum wells presently limits the use of light-emitting diodes based on them for high brightness lighting applications. InGaN/GaN quantum wells grown in the alternative zincblende phase are free of the strong polarisation fields that exacerbate droop and so were investigated by excitation-dependent photoluminescence and photoreflectance studies. Polarisation-resolved measurements revealed that for all excitation densities studied the emission from such samples largely originates from similar microstructures or combinations of microstructures that form within the quantum well layers. Emission efficiency varies significantly with excitation at 10 K showing that non-radiative recombination processes are important even at low temperature. The onset of efficiency droop, as determined by photomodulated reflection measurements, occurred at a carrier density of around 1.2 × 1020 cm-3 - an order of magnitude greater than the value reported for a reference wurtzite quantum well sample using the same method. The high carrier density droop onset combined with the much shorter carrier lifetime within zincblende InGaN/GaN quantum wells indicate they have the potential to effectively delay efficiency droop when used in GaN based light-emitting diodes. However, the material quality of the quantum well layers need to be improved by preventing the formation of microstructures within these layers, and the importance of the role played by non-radiative centres in the QW layer needs to be elucidated, to fully realise the material's potential.

Disentangling the Impact of Point Defect Density and Carrier Localization-Enhanced Auger Recombination on Efficiency Droop in (In,Ga)N/GaN Quantum Wells.

The internal quantum efficiency of (In,Ga)N/GaN quantum wells can surpass 90% for blue-emitting structures at moderate drive current densities but decreases significantly for longer emission wavelengths and at higher excitation rates. This latter effect is known as efficiency "droop" and limits the brightness of light-emitting diodes (LEDs) based on such quantum wells. Several mechanisms have been proposed to explain efficiency droop including Auger recombination, both intrinsic and defect-assisted, carrier escape, and the saturation of localized states. However, it remains unclear which of these mechanisms is most important because it has proven difficult to reconcile theoretical calculations of droop with measurements. Here, we first present experimental photoluminescence measurements extending over three orders of magnitude of excitation for three samples grown at different temperatures that indicate that droop behavior is not dependent on the point defect density in the quantum wells studied. Second, we use an atomistic tight-binding electronic structure model to calculate localization-enhanced radiative and Auger rates and show that both the corresponding carrier density-dependent internal quantum efficiency and the carrier density decay dynamics are in excellent agreement with our experimental measurements. Moreover, we show that point defect density, Auger recombination, and the effect of the polarization field on recombination rates only limit the peak internal quantum efficiency to about 70% in the resonantly excited green-emitting quantum wells studied. This suggests that factors external to the quantum wells, such as carrier injection efficiency and homogeneity, contribute appreciably to the significantly lower peak external quantum efficiency of green LEDs.

Quantitative analysis of carbon impurity concentrations in GaN epilayers by cathodoluminescence.

In this work, a technique for quantifying carbon doping concentrations in GaN:C/AlGaN buffer structures using cathodoluminescence (CL) is presented. The method stems from the knowledge that the blue and yellow luminescence intensity in CL spectra of GaN varies with the carbon doping concentration. By calculating the blue and yellow luminescence peak intensities normalised to the peak GaN near-band-edge intensity for GaN layers of known carbon concentrations, calibration curves that show the change in normalised blue and yellow luminescence intensity with carbon concentration in the 1016 - 1019 cm-3 range were derived at both room temperature and 10 K. The utility of such calibration curves was then examined by testing against an unknown sample containing multiple carbon-doped GaN layers. The results obtained from CL using the normalised blue luminescence calibration curves are in close agreement with those from secondary-ion mass spectroscopy (SIMS). However,the method fails when applying calibration curves obtained from the normalised yellow luminescence likely due to the influence of native VGa defects acting in this luminescence region. Although this work shows that indeed CL can be used as a quantitative tool to measure carbon doping concentrations in GaN:C, it is noted that the intrinsic broadening effects innate to CL can make it difficult to differentiate between the intensity variations in thin ( < 500 nm) multilayered GaN:C structures such as the ones studied in this work.

Characterization of buried interfaces using Ga Kα hard X-ray photoelectron spectroscopy (HAXPES).

The extension of X-ray photoelectron spectroscopy (XPS) to measure layers and interfaces below the uppermost surface requires higher X-ray energies and electron energy analysers capable of measuring higher electron kinetic energies. This has been enabled at synchrotron radiation facilities and by using lab-based instruments which are now available with sufficient sensitivity for measurements to be performed on reasonable timescales. Here, we detail measurements on buried interfaces using a Ga Kα (9.25 keV) metal jet X-ray source and an EW4000 energy analyser (ScientaOmicron GmbH) in the Henry Royce Institute at the University of Manchester. Development of the technique has required the calculation of relative sensitivity factors (RSFs) to enable quantification analogous to Al Kα XPS, and here we provide further substantiation of the Ga Kα RSF library. Examples of buried interfaces include layers of memory and energy materials below top electrode layers, semiconductor heterostructures, ions implanted in graphite, oxide layers at metallic surfaces, and core-shell nanoparticles. The use of an angle-resolved mode enables depth profiling from the surface into the bulk, and is complemented with surface-sensitive XPS. Inelastic background modelling allows the extraction of information about buried layers at depths up to 20 times the photoelectron inelastic mean free path.

Directly correlated microscopy of trench defects in InGaN quantum wells.

Directly correlated measurements of the surface morphology, light emission and subsurface structure and composition were carried out on the exact same nanoscale trench defects in InGaN quantum well (QW) structures. Multiple scanning probe, scanning electron and transmission electron microscopy techniques were used to explain the origin of their unusual emission behaviour and the relationship between surface morphology and cathodoluminescence (CL) redshift. Trench defects comprise of an open trench partially or fully enclosing material in InGaN QWs with different CL emission properties to their surroundings. The CL redshift was shown to typically vary with the width of the trench and the prominence of the material enclosed by the trench above its surroundings. Three defects, encompassing typical and atypical features, were prepared into lamellae for transmission electron microscopy (TEM). A cross marker technique was used in the focused ion beam-scanning electron microscope (FIB-SEM) to centre the previously characterised defects in each lamella for further analysis. The defects with wider trenches and strong redshifts in CL emission had their initiating basal-plane stacking fault (BSF) towards the bottom of the QW stack, while the BSF formed near the top of the QW stack for a defect with a narrow trench and minimal redshift. The raised-centre, prominent defect showed a slight increase in QW thickness moving up the QW stack while QW widths in the level-centred defect remained broadly constant. The indium content of the enclosed QWs increased above the BSF positions up to a maximum, with an increase of approximately 4% relative to the surroundings seen for one defect examined. Gross fluctuations in QW width (GWWFs) were present in the surrounding material in this sample but were not seen in QWs enclosed by the defect volumes. These GWWFs have been linked with indium loss from surface step edges two or more monolayers high, and many surface step edges appear pinned by the open trenches, suggesting another reason for the higher indium content seen in QWs enclosed by trench defects.

Publisher Correction: Impact of alloy fluctuations and Coulomb effects on the electronic and optical properties of c-plane GaN/AlGaN quantum wells.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Cross-shaped markers for the preparation of site-specific transmission electron microscopy lamellae using focused ion beam techniques.

We describe the use of a cross-shaped platinum marker deposited using electron-beam-induced deposition (EBID) in a focused ion beam - scanning electron microscope (FIB-SEM) system to facilitate site-specific preparation of a TEM foil containing a trench defect in an InGaN/GaN multiple quantum well structure. The defect feature is less than 100 nm wide at the surface. The marker is deposited prior to the deposition of a protective platinum strap (also by EBID) with the centre of the cross indicating the location of the feature of interest, while the arms of the square cross make an acute angle of 45° with the strap's long axis. During the ion-beam thinning process, the marker may be viewed in cross-section from both sides of the sample alternately, and the coming together of the features relating to the arms of the cross indicates increasing proximity to the feature of interest. Although this approach does allow increased precision in locating the region of interest during thinning, it also increases the time required to complete the sample preparation. Hence, this method is particularly well suited to directly correlated multi-microscopy investigations in previously characterised material where high yield and the precise location are more important than preparation time. In addition to TEM lamella preparation, this method could equally be useful for preparing site-specific atom probe tomography (APT) samples.

Impact of alloy fluctuations and Coulomb effects on the electronic and optical properties of c-plane GaN/AlGaN quantum wells.

We report on a combined theoretical and experimental study of the impact of alloy fluctuations and Coulomb effects on the electronic and optical properties of [Formula: see text]-plane GaN/AlGaN multi-quantum well systems. The presence of carrier localization effects in this system was demonstrated by experimental observations, such as the "S-shape" temperature dependence of the photoluminescence (PL) peak energy, and non-exponential PL decay curves that varied across the PL spectra at 10 K. A three-dimensional modified continuum model, coupled with a self-consistent Hartree scheme, was employed to gain insight into the electronic and optical properties of the experimentally studied [Formula: see text]-plane GaN/AlGaN quantum wells. This model confirmed the existence of strong hole localization arising from the combined effects of the built-in polarization field along the growth direction and the alloy fluctuations at the quantum well/barrier interface. However, for electrons these localization effects are less pronounced in comparison to the holes. Furthermore, our calculations show that the attractive Coulomb interaction between electron and hole results in exciton localization. This behavior is in contrast to the picture of independently localized electrons and holes, often used to explain the radiative recombination process in [Formula: see text]-plane InGaN/GaN quantum well systems.

Resonant photoluminescence studies of carrier localisation in c-plane InGaN/GaN quantum well structures.

In this paper we report on changes in the form of the low temperature (12 K) photoluminescence spectra of an InGaN/GaN quantum well structure as a function of excitation photon energy. As the photon energy is progressively reduced we observe at a critical energy a change in the form of the spectra from one which is determined by the occupation of the complete distribution of hole localisation centres to one which is determined by the resonant excitation of specific localisation sites. This change is governed by an effective mobility edge whereby the photo-excited holes remain localised at their initial energy and are prevented from scattering to other localisation sites. This assignment is confirmed by the results of atomistic tight binding calculations which show that the wave function overlap of the lowest lying localised holes with other hole states is low compared with the overlap of higher lying hole states with other higher lying hole states.

Lattice Location of Mg in GaN: A Fresh Look at Doping Limitations.

Radioactive ^{27}Mg (t_{1/2}=9.5 min) was implanted into GaN of different doping types at CERN's ISOLDE facility and its lattice site determined via ß^{-} emission channeling. Following implantations between room temperature and 800 °C, the majority of ^{27}Mg occupies the substitutional Ga sites; however, below 350 °C significant fractions were also found on interstitial positions ∼0.6 Å from ideal octahedral sites. The interstitial fraction of Mg was correlated with the GaN doping character, being highest (up to 31%) in samples doped p type with 2×10^{19} cm^{-3} stable Mg during epilayer growth, and lowest in Si-doped n-GaN, thus giving direct evidence for the amphoteric character of Mg. Implanting above 350 °C converts interstitial ^{27}Mg to substitutional Ga sites, which allows estimating the activation energy for migration of interstitial Mg as between 1.3 and 2.0 eV.

The atomic structure of polar and non-polar InGaN quantum wells and the green gap problem.

We have used high resolution transmission electron microscopy (HRTEM), aberration-corrected quantitative scanning transmission electron microscopy (Q-STEM), atom probe tomography (APT) and X-ray diffraction (XRD) to study the atomic structure of (0001) polar and (11-20) non-polar InGaN quantum wells (QWs). This paper provides an overview of the results. Polar (0001) InGaN in QWs is a random alloy, with In replacing Ga randomly. The InGaN QWs have atomic height interface steps, resulting in QW width fluctuations. The electrons are localised at the top QW interface by the built-in electric field and the well-width fluctuations, with a localisation energy of typically 20meV. The holes are localised near the bottom QW interface, by indium fluctuations in the random alloy, with a localisation energy of typically 60meV. On the other hand, the non-polar (11-20) InGaN QWs contain nanometre-scale indium-rich clusters which we suggest localise the carriers and produce longer wavelength (lower energy) emission than from random alloy non-polar InGaN QWs of the same average composition. The reason for the indium-rich clusters in non-polar (11-20) InGaN QWs is not yet clear, but may be connected to the lower QW growth temperature for the (11-20) InGaN QWs compared to the (0001) polar InGaN QWs.

Hysteretic photochromic switching of Eu-Mg defects in GaN links the shallow transient and deep ground states of the Mg acceptor.

Although p-type activation of GaN by Mg underpins a mature commercial technology, the nature of the Mg acceptor in GaN is still controversial. Here, we use implanted Eu as a 'spectator ion' to probe the lattice location of Mg in doubly doped GaN(Mg):Eu. Photoluminescence spectroscopy of this material exemplifies hysteretic photochromic switching (HPS) between two configurations, Eu0 and Eu1(Mg), of the same Eu-Mg defect, with a hyperbolic time dependence on 'switchdown' from Eu0 to Eu1(Mg). The sample temperature and the incident light intensity at 355 nm tune the characteristic switching time over several orders of magnitude, from less than a second at 12.5 K, ~100 mW/cm2 to (an estimated) several hours at 50 K, 1 mW/cm2. Linking the distinct Eu-Mg defect configurations with the shallow transient and deep ground states of the Mg acceptor in the Lany-Zunger model, we determine the energy barrier between the states to be 27.7(4) meV, in good agreement with the predictions of theory. The experimental results further suggest that at low temperatures holes in deep ground states are localized on N atoms axially bonded to Mg acceptors.

Mg doping affects dislocation core structures in GaN.

Aberration-corrected scanning transmission electron microscopy was used to investigate the core structures of threading dislocations in undoped GaN films with both high and low dislocation densities, and in a comparable high dislocation density Mg-doped GaN film. All a-type dislocations in all samples have a 5/7-atom core structure. In contrast, most (a+c)-type dislocations in undoped GaN dissociate due to local strain variations from nearby dislocations. In contrast, Mg doping prevents (a+c)-type dislocation dissociation. Our data indicate that Mg affects dislocation cores in GaN significantly.

Atom probe tomography and transmission electron microscopy of a Mg-doped AlGaN/GaN superlattice.

The electronic characteristics of semiconductor-based devices are greatly affected by the local dopant atom distribution. In Mg-doped GaN, the clustering of dopants at structural defects has been widely reported, and can significantly affect p-type conductivity. We have studied a Mg-doped AlGaN/GaN superlattice using transmission electron microscopy (TEM) and atom probe tomography (APT). Pyramidal inversion domains were observed in the TEM and the compositional variations of the dopant atoms associated with those defects have been studied using APT. Rarely has APT been used to assess the compositional variations present due to structural defects in semiconductors. Here, TEM and APT are used in a complementary fashion, and the strengths and weaknesses of the two techniques are compared.

Imaging dislocations in gallium nitride across broad areas using atomic force microscopy.

We have employed an atomic force microscope with a high sampling rate to image GaN samples grown using an epitaxial layer overgrowth technique and treated with silane and ammonia to enlarge the surface pits associated with threading dislocations (TDs). This allows TDs to be identified in high pixel density images tens of microns in size providing detailed information about the spatial distribution of the TDs. An automated software tool has been developed, which identifies the coordinates of the TDs in the image. Additionally, we have imaged the same sample using Kelvin probe force microscopy, again at high pixel density, providing data about the local changes in surface potential associated with hundreds of dislocations.

Lattice distortions in GaN on sapphire using the CBED-HOLZ technique.

The convergent beam electron diffraction (CBED) methodology was developed to investigate the lattice distortions in wurtzite gallium nitride (GaN) from a single zone-axis pattern. The methodology enabled quantitative measurements of lattice distortions (alpha, beta, gamma and c) in transmission electron microscope (TEM) specimens of a GaN film grown on (0,0,0,1) sapphire by metal-organic vapour-phase epitaxy. The CBED patterns were obtained at different distances from the GaN/sapphire interface. The results show that GaN is triclinic above the interface with an increased lattice parameter c. At 0.85 microm from the interface, alpha=90 degrees , beta=8905 degrees and gamma=11966 degrees . The GaN lattice relaxes steadily back to hexagonal further away from the sapphire substrate. The GaN distortions are mainly confined to the initial stages of growth involving the growth and the coalescence of 3D GaN islands.

Electron energy-loss near edge structure (ELNES) of InGaN quantum wells.

Lasers and light-emitting diodes (LEDs) that emit in the blue to green region are often based on InxGa1-xN quantum well structures. Ionization edges in the electron energy-loss spectrum contain fine structures (called the energy-loss near edge structure (ELNES)) and provide information about the electronic structure. In this paper we compare the experimental and calculated ELNES for the N-K ionization edge of InxGa1-xN quantum wells. When the effects of the core-hole are included in the calculations, agreement between experimental and calculated spectra is very good. Strain has been shown to accentuate the effects of In on the ELNES and moves the ionization edge onset down in energy, relative to the other features. These results suggest that ELNES may provide an alternative method to lattice imaging to determine the presence of strain in this system.