Dynamical behavior of radiative recombination has been assessed in the In0.20Ga0.80N (3nm)/In0.05Ga0.95N (6 nm) multiple quantum well (MQW) structure by means of transmittance (TR), electroreflectance (ER), photoluminescence excitation (PLE) and time-resolved photoluminescence (TRPL) spectroscopy. The PL at 20 K was mainly composed of two emission bands whose peaks are located at 2.920 eV and 3.155 eV. The ER and PLE revealed that the transition at 3.155 eV is due to the excitons at quantized level between n=1 conduction and n=1 A(Γ9υ) valence bands, while the main PL peak at 2.920 eV is attributed to the excitons localized at the trap centers within the well. The TRPL features were well understood as the effect of localization where photo-generated excitons are transferred from the n=1 band to the localized centers, and then are localized further to the tail state. The origin of the localized centers were attributed to the In-rich region in the wells acting as quantum dots which could be observed by transmission electron microscopy (TEM) and energy-dispersive X-ray microanalysis (EDX).

We have studied optical transitions (absorption and luminescence) in nominally undoped and Mg-doped GaN deposited by MOCVD and MBE. In the range between 3.0 and 3.4 eV, a variety of well known low-intensity luminescence lines are observed, whose origin is discussed. In particular, by comparing excitation with subgap versus above-gap laser lines as well as by combining optical subgap absorption with spectrally resolved photoconductivity, we identify localized optical transitions occuring in isolated cubic inclusions in the otherwise hexagonal GaN epitaxial layers. Implications of these strucural defects for photocurrent transients are also presented.

We present results of spatially-resolved photoluminescence and Raman measurements on a 200 μm thick GaN layer grown on sapphire by hydride vapor phase epitaxy. Our microphotoluminescence measurements reveal that the peak position of the excitonic and donoracceptor-pair transitions strongly depends on the distance to the substrate interface. We observed a strong blue shift near the interface and discuss the influence of strain, which we quantified by micro-Raman experiments.

The 364 nm PL-system in GaN is attributed to the formation of dislocation excitons and charged dislocation excitons on c-axis screw dislocations. The binding energy for the dislocation exciton, charged dislocation exciton and a hole on the screw dislocation were determined as 35 meV, 7 meV and 65 meV respectively, in accord with experiment.

Character of the metal-insulator transition which occurs at about 23 GPa in bulk GaN crystals has been studied by means of high pressure Raman spectroscopy. The related freeze-out of electrons is caused by the localized donor state formed by most likely oxygen and emerging at high pressures to the band gap of GaN. As a result, the electron concentration drops from its initial value of 5.1019 cm-3 to about 3. 1018 cm-3. These remaining electrons originate likely from another donor center with effective mass character, probably carbon. The obtained results raise a question whether the nitrogen vacancy is abundant enough to be observed in bulk GaN crystals.

We report a comparative study by luminescence and reflectivity of GaN grown on sapphire by MOVPE, HVPE and GSMBE. Whatever the growth technique, undoped GaN shows well resolved reflectivity spectra, allowing hetero-epitaxial strain to be taken into account in the interpretation of luminescence spectra. MOVPE provides the highest purity samples so far. From the luminescence study of n- and p-type doped samples, activation energies of 265 ± 10 meV and 34 ± 2 meV are deduced for Mg acceptors and Si donors respectively. From the luminescence spectra of HVPE grown samples, the presence of a perturbed interfacial n+ layer is evidenced.

We present a detailed photoluminescence excitation study of the optical transitions in GaN. This technique is employed to distinguish between band-to-band excitation and exciton contribution to the formation of the free exciton, bound exciton, violet and yellow photoluminescence bands. We show the dominant role of the Fröhlich polar intraband scattering in the formation of the free exciton states. We demonstrate that bound exciton states in a large extent are created by the capture of the free excitons by shallow impurities as well as by phononassisted resonant excitation of the bound exciton states. The capture of the free carriers excited in the band continuum is a main excitation source for the violet and yellow bands. However, distinct A- and C-exciton resonances are detected in the excitation spectra of the violet and yellow emission bands.

We report a photoluminescence (PL) and photoluminescence excitation (PLE) investigation of the deep iron acceptor in hexagonal GaN. Codoped samples give rise to a new zero-phonon line (ZPL) at 1.268 eV. The spectral shape of its phonon side band is very similar to that of the Fe3+ -spectrum. A lifetime of 2.6 ms was measured for this ZPL which indicates a spinforbidden transition. In contrast to isolated Fe3+, no Zeeman-splitting is observed. PLE spectra of the Fe3+ (4T1-6A1) zero-phonon line at 1.299 eV in semi-insulating GaN samples reveal intracenter excitation processes via excited states of the Fe3+ center. Three excited Fe3+ crystal field states at 1.299 eV (4T1), 2.01 eV (4T2), and 2.731 eV (4E) above the 6A1, ground state were identified. PLE spectroscopy for the 1.268 eV zero-phonon line reveals resonances at 2.3 eV and 2.65 eV. The 1.268 eV ZPL is tentatively attributed to a Fe-complex with a nearby donor.

We report several new aspects of the excitonic properties of heteroepitaxial GaN grown on sapphire or 6H-SiC. In particular, we observed the n = 2 free exciton associated with both A and B excitons (which are distinct from the n = 1 C exciton) using reflectance and 1.7 K photoluminescence. We also studied the behavior of the n = 2 A-exciton using magnetoluminescence in fields up to 12 T. The large diamagnetic shift and splitting positively confirm the identification, yielding an exciton binding energy of about 26.4 meV. Several previous identifications of the n = 2 free exciton yielding a smaller exciton binding energy are probably in error, based on our results. We have also detected the two-electron replica of the neutral donor-bound exciton for the first time in GaN and observed its splitting pattern in magnetic fields up to 12 T. This feature is 22 meV below the principal neutral donor-bound exciton peak, independently of strain shifts in the overall spectrum. It yields a precise donor binding energy of 29 meV for the shallow residual donor in material grown by metalorganic chemical vapor deposition and gas-source molecular beam epitaxy, considerably smaller than that of the residual donor reported earlier in hydride vapor phase epitaxial material (about 35.5 meV).

Cathodoluminescence (CL) spectroscopy and microscopy were used to study the luminescent properties of a variety of GaN films, both Si-doped and unintentionally-doped, grown on sapphire substrates. A narrow and intense near band-edge emission was found in the CL spectrum of each film examined, and deep-level emission was also observed for some of the films. The luminescence efficiency of near band-edge emission increased with a faster rate than that of deep-level emission when the pumping current was increased. Spatial nonuniformities of luminescence were observed in monochromatic CL microscopy, and microstructures were observed in scanning electron microscopy. No correlations between luminescence features and microstructural features were seen. Degradation of near band-edge luminescence was observed, accompanied by growth of deep-level emission.

Raman analysis of film quality, phonon-plasmon coupling, and phonon-exciton interaction in GaN films with varying Si doping levels is presented. The films exhibit a small stress component ∽0.1 GPa, calculated from the frequency shift of the E2 mode. No correlation between the stress and the doping concentration was found. No forbidden Raman lines were detected in the spectra, implying high quality oriented films. The Raman lineshape of the E2 mode is a Lorentzian with similar linewidths at room temperature and at 10K indicating a homogeneous lifetime broadening mechanism which is not significantly affected by the change in temperature. The linewidth is also independent of Si concentration. The phonon-plasmon mixed frequency modes were calculated to be at ω−.=86 cm−1 and at ω+=741 cm−1 The modes are not present in the spectra and the only effect of the plasmons is a change of the Al(LO) lineshape. Analysis of the A1(LO) line indicated a uniform spatial doping distribution. A resonance effect was observed for the symmetry-allowed A1(LO) mode at T=10K with sub bandgap excitation light. The resonance interaction is consistent with the free exciton model.

Raman scattering experiments on silicon-doped GaN show that donor impurities quench the Al(LO) Raman line at 735 cm−1. This is due to interaction between lattice vibrations and the free carrier plasma. The spatial variation of the Al(LO) signal has been imaged directly using newly developed instrumentation. Features with dimension under on micron are observed in faceted GaN crystallites. The variation in free carrier concentration is attributed to preferential incorporation of donor impurities during growth.

Non-equilibrium electron distributions as well as phonon dynamics in wurtzite GaN have been measured by subpicosecond time-resolved Raman spectroscopy. Our experimental results have demonstrated that for electron densities n ≥ 5 × l017cm−3, the non-equilibrium electron distributions in wurtzite GaN can be very well described by Fermi-Dirac distribution functions with the temperature of electrons substantially higher than that of the lattice. The population relaxation time of longitudinal optical phonons was directly measured to be τ ≅ 5 ± 1 ps at T = 25 K. The experimental results on the temperature dependence of the lifetime of longitudinal optical phonons suggest that the primary decay channels for these phonons are the decay into (1) one transverse optical phonon and one high energy, longitudinal or transverse acoustical phonons; and (2) one transverse optical phonon and one E2 phonon.

The electronic structure of a set of several tens of a few micrometer-thick GaN epilayers grown by MOVPE on (000l)-oriented sapphire is investigated by means of photoluminescence and reflectance spectroscopy. The strain-fields effects are quantitatively interpreted using a group theory analysis which predicts seven pertinent parameters: the crystal field splitting, two spin-orbit interaction parameters and four deformation potentials. From a theoretical point of view, we also examine the exciton problem. We establish the selection rules which control the strength of optical transitions for Г5 and Г1 excitons, and estimate the electron hole exchange interaction and the longitudinal transverse splitting of A, B, C excitons. Raman spectroscopy measurements are performed at various temperatures to observe the strain-induced shift of Raman frequencies. We determine four phonon deformation potentials in w-GaN.

We report electrically-detected magnetic resonance (EDMR) and electroluminescence-detected magnetic resonance (ELDMR) results on InGaN/AJGaN single-quantum-well light emitting diodes. The dominant feature detected by either technique is a broad resonance (ΔB ≈ 13 mT) at g ≈ 2.01 which is enhanced by high current stressing. Our ELDMR measurements show that, depending on bias, this defect is predominately associated with either an increase or a decrease in electroluminescence at resonance while our EDMR measurements show that this resonance is associated with an increase in current at resonance before stressing and a decrease after stressing. We suggest that this is associated with a nonradiative recombination path, in parallel with the radiative recombination path and with recombination in the depletion region of a contact. A second resonance, more prominent before stressing, with g ≈ 1.99 and ΔB ∽ 7 mT is very similar to the deep donor trap, previously observed in double heterostructure diodes and is associated with a decrease in both the current and electroluminescence at resonance.

AlGaN/GaN single quantum wells (QW) have been grown on 2” sapphire substrates (c-plane) by metal-organic chemical vapor deposition (MOCVD). The well width was varied between 20 and 40 Å for barriers containing 4 % and 16 % of aluminium. Cathodoluminescence (CL) and Photoluminescence (PL) spectra of the samples show, as expected, a shift of the quantum well emission to higher energies with decreasing well width, whereas the barrier luminescence stays at constant energy. Examination of the QWs by resonant Raman spectroscopy tuned to the gap of the well, clearly shows the GaN A1(LO) phonon besides the AlGaN A1(LO) phonon from the barrier. For a well width of 20 Å we observe a shift of the A1(LO) GaN phonon indicating a certain degree of intermixing at the GaN/AlGaN interface. Atomic Force Microscopy (AFM) reveals that the layers are growing in a 2-dimensional step flow growth mode with step heights of 3 and 6 Å corresponding to mono- and biatomic steps. High Resolution Transmission Electron Microscopy (HRTEM) micrographs of the 40 Å well show a very low interface roughness of 1–2 atomic layers.

Strain in GaN epitaxial layers at room temperature is measured with three complementary methods: Raman spectroscopy (via shifts of phonon frequencies), low temperature photoluminescence (via shifts of band-edge luminescence), and X-ray diffraction (via shifts in lattice spacings). GaN films grown on the c-plane of sapphire tend to be in compression. Increasing the Si-dopant concentration (up to 1019 cm−3) is observed to add compressive strain to the layer. Axially resolved measurements obtained by micro-Raman in 4 μm thick Si-doped films reveal strain relaxation toward the sample surface at Si concentrations above 1018 cm−3. Mg- and Si-doped GaN films on SiC substrates are found to be in tension. An experimental methodology is presented that separates two contributions to the room temperature residual stress in GaN epilayers: (1) the thermal stress due to differences in the thermal expansion coefficients of the epilayer and substrate and (2) the intrinsic stress, which is influenced by the growth conditions. We measure stress as a function of temperature up to 325 C, about one-third of the growth temperature, by monitoring the frequency of the E2 phonon mode by Raman spectroscopy. A high-quality bulk single crystal of GaN is used as a strain-free standard. Over this temperature range, most layers behave elastically; the observed stress trends are well-fit by a thermal expansion model using previous reported values of the thermal expansion coefficients of GaN and the substrates. The intrinsic stress states at the growth temperature for films grown on sapphire and SiC are predicted to be tensile and compressive, respectively, in agreement with the a-plane lattice coefficient mismatch.

We provide the widest estimate thus far of the range of tensile and compressive stress (−3.8 to 3.5 kbar) that GaN epitaxial material can withstand before relaxation occurs, and an unambiguous determination of the spin-orbit splitting Δso = 17.0 ± 1 meV for the material. These are achieved by analyzing 10K reflectance data for the energy separation of transitions between the uppermost valence bands and the lowest conduction band of wurtzitic GaN as a function of biaxial stress for a series of GaN films grown on both Al2O3 and 6H-SiC substrates. Our data explicitly show the nonlinear behavior of the excitonic energy splittings B-A and C-A vs. the energy position of the A exciton, which stands in contrast to the linear approximations used by previous workers analyzing material grown only on Al2O3 substrates. Further, the lineshape ambiguities present in GaN reflectance spectra that hindered the accurate determination of such excitonic energies have also been resolved by analyzing these data in reciprocal space, where critical point energies are determined by phase effects to an accuracy of ±0.5 meV.

The electronic structure of thin film wurtzite GaN has been studied using a combination of angle resolved photoemission, soft x-ray absorption and soft x-ray emission spectroscopies. We have measured the bulk valence and conduction band partial density of states by recording Ga L- and N K- x-ray emission and absorption spectra. We compare the x-ray spectra to a recent ab initio calculation and find good overall agreement. The x-ray emission spectra reveal that the top of the valence band is dominated by N 2p states, while the x-ray absorption spectra show the bottom of the conduction band as a mixture of Ga 4s and N 2p states, again in good agreement with theory. However, due to strong dipole selection rules we can also identify weak hybridization between Ga 4s- and N 2p-states in the valence band. Furthermore, a component to the N K-emission appears at approximately 19.5 eV below the valence band maximum and can be identified as due to hybridization between N 2p and Ga 3d states. We report preliminary results of a study of the full dispersion of the bulk valence band states along high symmetry directions of the bulk Brillouin zone as measured using angle resolved photoemission. Finally, we tentatively identify a non-dispersive state at the top of the valence band in parts of the Brillouin zone as a surface state.

Two positive ODMR resonances are commonly observed on a luminescence band in GaN at 2.2 eV, one identified as a shallow donor, the other currently unidentified. We here report a study of their dependencies on a variety of experimental parameters, including microwave modulation frequency, microwave power, photoexcitation power and photoexcitation energy. ODMR simulations using two theoretical models are compared to experimental results which are consistent with spin-dependent recombination between the two defects, assuming the donor has a spin-lattice relaxation time shorter than the spin-dependent recombination lifetime. The photoexcitation energy dependence suggests that the spin-dependent recombination associated with the 2.2 eV band is not the same recombination that is responsible for the luminescence. This supports the two stage model put forth by Glaser et al. for the luminescence process.