Two-source optical quenching spectroscopy demonstrates that the four site-selective Er3+ photoluminescence (PL) spectra observed in Er-implanted GaN contribute to the above-gap excited Er3+ PL spectrum, with relative efficiencies determined by the carrier capture cross sections and concentrations of the defects or traps which mediate the excitation of each Er site. The above-gap pumped PL spectrum is dominated by two of the trap-mediated Er3+ PL spectra, while the highest concentration Er site, which is efficiently pumped only by direct 4f absorption, contributes only weakly. These experiments indicate that the same defects and impurities are involved in the trapmediated processes responsible for both the above- and the below-gap excitations of the Er3+ PL.

We report on results of low-temperature photoluminescence measurements performed on GaN films, grown by molecular beam epitaxy (MBE) on sapphire substrates. The GaN films are either Mg doped (p-type) or consist of a Mg-doped layer on top of a Si doped GaN layer (n-type). In the p-doped samples, the sharpness of the donor-acceptor-pair transition is striking, three phonon replicas are clearly resolved. A transition band occurs around 3.4 eV, which becomes dominant for samples with an np-layer structure. The position and the composition of the near band edge transitions are influenced by the growth of the buffer layers. Depending on the growth conditions a transition at 3.51 eV can be observed.

We investigate the optical properties of n-type Gallium Nitride (GaN) with concentrations ranging from 5×1016 to 7×1018 cm−3. The near-band edge ultraviolet (UV) transition increases monotonically with the doping concentration. The photoluminescence linewidth of the near-bandgap optical transition increases from 47 to 78 meV as the doping concentration is increased. The broadening is modeled by taking into account potential fluctuations caused by the random distribution of donor impurities. Excellent agreement is found between experimental and theoretical results. We also investigate the origin of the yellow luminescence in GaN. At low excitation densities the experimental ratio of the UV-to-yellow photoluminescence does not change significantly as the doping concentration is increased by two orders of magnitude. Analysis of the luminescence in terms of a theoretical model indicates that the yellow luminescence is due to compensating impurities or defects.

We report results of high temperature photoluminescence and photoluminescence excitation studies of Er doped GaN. Er was doped into GaN either by ion-implantation or during growth by metalorganic molecular beam epitaxy (MOMBE). Using above gap excitation (λex=325nm), GaN:Er showed strong 1.54 μm Er3+ luminescence up to 550K which indicates the potential of this material for high temperature opto-electronic applications. In addition, we performed timeresolved photoluminescence excitation (PLE) measurements over the wavelength range 420 to 680 nm using a Nd:YAG pumped Optical Parametric Oscillator (OPO). Similar to our previous PLE results of Er doped AIN, we observed that Er3+ ions in GaN can be excited either through resonant pumping of Er3+ energy levels or through energy-transfer processes involving defect states. The relative contribution of direct and indirect Er3+ excitation, however, seems to depend on the material preparation method.

We report the results of a photoluminescence excitation (PLE) study on excitonic transitions in GaN. PLE measurements were carried out as a function of temperature to investigate exciton formation, thermalization, and annihilation processes in GaN. The PLE spectra detected at free-exciton resonance were found to exhibit a large number of oscillatory emission structures with an energy spacing of LO-phonon (92 meV), and such emission persisted up to 400K. The observation suggests that a phonon-assisted indirect exciton formation process, which simultaneously generates a free-exciton and an LO-phonon, is very significant and efficient in GaN, and the lifetime of the free-exciton is longer than the relaxation time of LO-phonon emission but much shorter than that of acoustic-phonon emission. The appearance of the broad spectral features at the PLE spectrum detected at free-exciton resonance, similar to those measured at the boundexciton resonance, is attributed to the higher order transition processes which involve acousticphonon scattering and thermal repopulation of free-excitons. In addition, we observed up to the 12th LO-phonon emission in PLE spectrum around 120 K, indicating the existence of the excitons with a large kinetic energy in GaN.

We propose a new approach to explain the unusually low pressure coefficients of the luminescence peaks observed in single-quantum-well InGaN-based light emitting diodes manufactured by Nichia Chemical Industries. In view of the most recent first principles band structure calculations for InN under hydrostatic pressure, we find that it is possible to reproduce the measured low (12–16meV/GPa) pressure coefficients of the luminescence by assuming the formation of zincblende InN inclusions in the InGaN quantum well layers. These cubic inclusions, surrounded by the usual wurtzite material, should act like quantum dots giving rise to enhanced electron localization. Obtained this way, the pressure shift of the luminescence peaks in blue and green InGaN-based emitters should be close to 14meV/GPa, in good agreement with our experimental results. This explanation of the observed low pressure coefficients in these devices is consistent with recent independent evidence for InN inclusions in InGaN epilayers.

Cathodoluminescence (CL) in the scanning electron microscope has been used to investigate the variation along the growth axis direction of luminescence emission from epitaxial GaN films. CL spectra recorded at different positions of the sample cross-section as well as monochromatic CL images show strong spatial variations of the different luminescence emissions along the growth axis. At the buffer layer-substrate interface and at the top part of the sample, which corresponds to a Si doped epilayer, enhanced CL emission is observed as compared with the relatively low emission in the central region of the cross-section. The nature of the defects responsible for the observed CL distribution is discussed.

Yellow luminescence (YL) from GaN was systematically investigated through the intentional introduction of carbon, from propane, and excess H2 during growth by the halide vapor phase epitaxy technique. All GaN films were studied by photoluminescence, X-ray diffraction and Hall measurements. The unintentionally doped GaN showed undetectable or very weak YL signal, while both C-doping and H2 addition resulted in a significant enhancement of YL. The blue- and red-shift of the YL band of the C-doped and ‘H2-grown’ GaN with the increasing temperature indicated that different mechanisms existed in these two cases. The temperature dependence of the integrated intensity of the YL band of both groups implicated that shallow donors, not ‘shallow’ acceptors participated the YL transition and that there were more than one radiative recombination channel within the YL band.

Cathodoluminescence spectroscopy is used to depth profile the indium mole fraction of a 0.4 μm InGaN epilayer grown by metalorganic chemical vapour deposition, and to probe an underlying 1.0 μm GaN epilayer and the sapphire substrate beneath. Spectral information is obtained by using a variable energy electron beam as the excitation source. Calibration of beam penetration is achieved using Monte Carlo simulations of electron beam trajectories. The indium mole fraction is found to decrease from a mean value of 27% at the surface of the InGaN layer to ≈ 24% at its interface with the GaN layer.

Photoluminescence spectra of hexagonal GaN were measured in the temperature range T= 2 – 1200 K. We identify the Free Exciton (FX) as the dominant recombination process in our high quality samples for temperatures above 200 K. From the line shape fit of the FX we determine the excitonic band gap shift with temperature. An analysis according to the empirical Varshni equation gives Eg (T)-Eg(0 K) = (-α T2)/(T + β), with α = (7.3 ± 0.3)·10−4 eV/K and β = (594 ± 54) K. We have detected significant differences in the band gap energy at low and higher temperatures for GaN layers grown on different substrate materials. Heating GaN above 1200 K leads to irreversible changes in the near band gap photoluminescence spectra.

It is shown that heteroepitaxial GaN layers grown on NdGaO3, in spite of a very high conductivity (˜ 10 Ω−1cm−1) have very efficient luminescence properties. It is shown that a high electrical conductivity is caused by contamination of GaN layers with oxygen. Efficient emission due to donor bound excitons (at hv = 3.475 eV), free excitons and free electron – hole recombination have been identified. The total PL emission in the exciton region exceeds the intensity from the homoepitaxial GaN layers. It is argued that a high oxygen concentration eliminates nonradiative channels connected with point defects, leading to efficient radiative recombination.

This paper describes the characteristics of the luminescence centers observed in the various photoluminescence (PL) and photoluminescence excitation (PLE) spectra of hexagonal InxGa1−xN microcrystals, synthesized by the nitridation of sulfide in the In concentration range of 2%<x<6%. The grown crystals showed macroscopic and microscopic inhomogeneity. A series of component bands expressed by Gaussian functions with discrete peak energies selected from the frequently observed peaks in the PL spectra and full width at half maximum obtained from the narrowest PL bands were fitted well to observed broad emission bands. As some of the peak energies of the component bands coincide with those of the bands that are attributed to the decay of excitons at the localized states caused by the In content fluctuation in the literature, we concluded the luminescence center for the component bands also originates from the fluctuation of the In contents. Although the In contents of the samples determined by the X ray diffraction method were compatible the gap energies from the PLE measurements were different. The PLE spectra showed that the Stokes shifts were large and the gap energies were irrelevant to the emission band peaks. The excitation bands for the emission bands in the range of 2.8 to 2.48 eV were located at 3.36, 3.26, and 3.21 eV. These results indicate that the states of the luminescence center are localized and the energy levels are discrete. To obtain the microscopic structure of the localized states in InxGa1−xN, we propose the isoelectronic atom cluster model comparable with Te clusters in ZnSeTe.

We demonstrate photoluminescence from self- assembling InGaN quantum dots (QDs), which are artificially fabricated on AlGaN surfaces via metal- organic chemical vapor deposition. InGaN QDs are successfully fabricated by the growth mode transition from step- flow to three dimensional island formation by using anti-surfactant silicon on AlGaN surface. The diameter and height of the fabricated InGaN QDs are estimated to be ˜10nm and ˜5nm, respectively, by an atomic- force- microscope (AFM). Indium mole fraction of InxGal−x N QDs is controlled from x=˜0.22 to ˜0.52 by varying the growth temperature of QDs. Intense photoluminescence is observed even at room temperature from InGaN QDs embedded with the GaN capping layers. In addition, the temperature- dependent energy shift of the photoluminescence peak- energy shows a localization behavior.

Elastic recoil detection (ERD) using energetic heavy ion beams (e.g. 170 MeV 127I) is a suitable method to measure depth profiles of light and medium heavy elements in thin films. The main advantages of ERD, which makes it favorable over many other techniques for elemental analysis, is the possibility to obtain reliable and quantitative results, a sensitivity in the ppm region or a depth resolution even better than 1 nm.

ERD analysis was employed to obtain quantitative information about the aluminium content x in MBE grown AlxGa1−xN layers on Al2O3 substrates. Using this stoichiometry information and the lattice constants obtained from high resolution X-ray diffraction, Vegard's law could be confirmed with high accuracy. Secondly, nitridation of heated Al2O3 substrates in NH3 atmosphere was investigated using high resolution ERD. A substantial nitrogen content on the surface of the substrate was detected which means a nearly complete AIN layer grown on the Al2O3 surface by a heat treatment only. Such a nitridation layer can be the base for further growth of nitrides on Al2O3 surfaces. As a third, the impurity content of group III nitrides was investigated in dependence on deposition conditions for both, MBE and MOCVD grown samples. In all samples investigated an oxygen concentration larger than 100 ppm was detected which is much higher than the intrinsic charge carrier density of these samples. In addition it is shown that the efficiency of p-doping by Mg may not only be hindered by hydrogen but also by carbon impurities.

Positron annihilation experiments have been performed to identify native point defects in GaN bulk crystals as well as in epitaxial layers. The results show that Ga vacancies are present at concentrations of 1017 – 1018 cm−3 in undoped GaN bulk crystals and layers, whereas the Mgdoped samples are free of Ga vacancies. The Ga vacancies are negatively charged and their concentration correlates with the intensity of the yellow luminescence. We conclude that the Ga vacancies contribute to the electrical compensation of n-type GaN and their acceptor levels are involved in the yellow luminescence transition.

The optical properties and their modification by crystal defects of wurtzite GaN are investigated using spatially resolved electron energy-loss spectroscopy (EELS) in a dedicated ultra-high vacuum field emission gun scanning transmission electron microscope. The calculated density of states of the bulk crystal reproduces well the features of the measured spectra. The profound effect of a prismatic stacking fault on the local electronic structure is shown by the spatial variation of the optical properties derived from low-loss spectra. It is found that a defect state at the fault appears to bind 1.5 electrons per atom.

Heteroepitaxial GaN films grown on sapphire are usually in a compressive residual biaxial stress state that is typically 100 – 500 MPa but can be as high as 1 GPa. Spatially resolved measurements of GaN film stresses obtained via micro-Raman spectroscopy using shifts in the frequency of the E2 phonon are presented. The spatial resolution of this stress measurement technique is ca. 2 μm laterally and 10 μm in the axial (growth) direction. Two interesting stress gradients are found near cleaved edges. (1) On the surface of the cleaved edge the compressive film stress is a maximum at the GaN/sapphire interface and relaxes to a nearly stress-free state towards the free edge. (2) In the lateral direction, the stress increases away from the edge, reaching a steady-state value at approximately 10x the film thickness. It is shown that the effects are consistent with existing finite element models used to treat strained growth of surface structures in other semiconductor systems. An axial stress gradient that is not associated with edge effects is found in some thick (>20 μm) HVPEgrown films. The interface region is found to be up to 500 MPa in compression, but this stress declines toward the film surface, which is found to be nearly stress free.

Although GaN has been extensively studied for applications in both light emitting and high power devices, the AlN/GaN valence band offset remains an area of contention. Values quoted in the literature range from 0.8eV (Martin)[1] to 1.36eV (Waldrop)[2]. This paper details an investigation of the AIN/AlxGa1-xN band offset as a function of alloy composition. We find an AlN/AlxGa1-xN valence band offset that is nearly linear with Al content and an end point offset for AlN/GaN of 1.36 ± 0.1 eV. Samples were grown using radio frequency plasma assisted molecular beam epitaxy and characterized with x-ray photoelectron spectroscopy(XPS). Core-level and valence-band XPS data for AIN (0001) and AlxGa1-xN (0001) samples were analyzed to determine core-level to valence band maximum (VBM) energy differences. In addition, oxygen contamination effects were tracked in an effort to improve accuracy. Energy separations of core levels were obtained from AlN/AlxGa1-xN(0001) heterojunctions. From this and the core-level to valence band maximum separations of the bulk materials, valence band offsets were calculated.

GaN/(0001)Sapphire grown by low pressure MOVPE is studied by high resolution Z-contrast imaging using STEM. First direct observation of the threading dislocation with edge character shows the atomic core structure, which appears to have a similar configuration to the {10–10} surface. The surfaces of the nanopipe walls are on {10–10} with the terminating layer between the atoms with one bond per pair. In addition, the high resolution Z-contrast image of the prismatic stacking fault confirms the results by conventional HRTEM.

The surface electronic structure of wurtzite GaN (0001) (1 × 1) has been investigated using angle-resolved photoemission spectroscopy. Surfaces were cleaned by repeated cycles of N2 ion bombardment and annealing in ultra-high vacuum. A well-defined surface state below the top of the valence band is clearly observed. This state is sensitive to the adsorption of both activated H2 and O2, and exists in a projected bulk band gap, below the valence band maximum. The state shows no dispersion perpendicular or parallel to the surface. The symmetry of this surface state is even with respect to the mirror planes of the surface and polarization measurements indicate that it is of spz character, consistent with a dangling bond state.