Laplace deep-level transient spectroscopy and photoluminescence have been used to demonstrate that the FeGa (0/−) acceptor level in dilute AlxGa1−xN (x ≤ 0.063) can be considered as a common reference level as expected for energy levels of transition metals in isovalent semiconductor compounds. Furthermore, the conduction and valence band offsets (ΔEC and ΔEV, respectively) in GaN/AlxGa1−xN heterojunctions, as a function of Al content for samples grown by the metalorganic vapor-phase epitaxy technique on native Ammono-GaN substrates, have been found. The band-offsets determined in this study are ΔEC = 1.17x eV and ΔEV = −0.95x eV over the range of x studied and are in good agreement with other experimental results reported for actual GaN/AlxGa1−xN heterojunctions as well as with the recent theoretical calculations based on hybrid density functional theory. Moreover, we confirmed that the band bowing effect related to compositional dependence in AlxGa1−xN is accommodated practically only in the conduction band as suggested by theoretical calculations.

In this letter, we demonstrate the successful development of graphene Schottky barrier diodes (Gr SBDs), which act as an efficient semi-transparent contact to n-GaN. We show that Gr SBDs can be used for deep trap investigations in n-GaN by capacitance-based spectroscopic techniques. To demonstrate its functionality, Gr Schottky barrier diodes on an n-GaN sample grown by the metal organic vapor phase technique were fabricated and then used in the measurements of steady-state photo-capacitance (SSPC) and deep level optical spectroscopy (DLOS). It is shown that the SSPC and DLOS spectra obtained for Gr SBDs are in excellent agreement with Ni-based semi-transparent contacts to n-GaN used in this study for comparison. Deep trap levels located at Ec-3.3 eV, Ec-2.6 eV, and Ec-1.84 eV for Ni SBD and Ec-3.3 eV and Ec-2.6 eV for Gr SBD were found, respectively. The presence of a trap level with Ec-1.84 eV observed only in the Ni SBD samples suggests that this trap originates from different sample treatments prior to Schottky contact deposition. Additionally, the optical capture cross-section data (σ0) derived from DLOS were fitted using the Lucovsky model under the assumption of no lattice relaxation for all deep traps observed in this study. Discrepancies in trap concentration derived from SSPC measurements among different diodes for a trap with EC-3.3 eV were attributed to reduced light transmission through the Gr contact in the UV spectral range and the presence of some macroscopic defects related to Gr transfer to the n-GaN film.

AlxGa1−xN epilayers are used as the basis of ultraviolet LEDs and detectors. The trap states produced by defects and impurities can play a key role in the device performance. In this work, conventional deep-level transient spectroscopy, photoluminescence (PL), and secondary ion mass spectrometry have been used to characterize a deep-level trap termed as E3 in dilute AlxGa1−xN (x < 0.063) epilayers grown by metal-organic vapor phase epitaxy (MOVPE) on highly conductive ammono-GaN substrates. The AlxGa1−xN epilayers were doped with silicon to about 3 × 1016 cm−3. The electrical and the optical measurements were conducted on Ni/Au Schottky barrier diodes and virgin samples, respectively. First, we observed a general trend that the E3 (FeGa) electron trap concentration significantly changes along the wafers in AlxGa1−xN layers that is fully consistent with previously reported results for GaN materials grown by the MOVPE technique. Second, we report that the activation energies for electron emission for the E1 and E3 traps in dilute AlxGa1−xN exhibit linear variations with Al content. Moreover, low-temperature PL results show a proportional relation between the intensity of the line with its maximum at 1.299 eV and concentration of residual Fe impurity. Finally, we discuss how the presence of defects resulting from Fe contamination may result in degradation of AlxGa1−xN-based devices.

The results of junction spectroscopy measurements on deep-level defects in MOVPE (Metal-Organic Vapor-Phase Epitaxy) n-GaN samples grown on highly doped Ammono-GaN and subjected to 1.5 MeV electron irradiation are compared with the published results for epi-GaN materials grown by other techniques. It is found that in addition to the commonly observed deep-level traps in n-type GaN, such as E1 (0.25 eV) and E3 (0.59 eV), 1.5-MeV electron irradiation introduces two other electron traps, EE1 and EE2, with electronic levels at about 0.14 and 0.98 eV below the conduction band edge (EC), respectively. In the case of the EE1 level, a strong influence of the electric field (E) on the electron emission rate (eem) is observed. This suggests a donor type character of this trap level. Further, we have observed that strong electric field, as high as 2 x 105 V/cm, results in lowering the activation energy of electron emission from the EE1 level down to the value of 0.095 eV. The strong eem(E) dependence for the EE1 trap can explain the wide variation in electronic signatures of this trap reported in previous publications. The analysis of the EE1 trap concentrations in the electron irradiated samples allowed us to estimate the average production rate of this trap by 1.5 MeV electrons as 0.125 cm -1 for n-GaN material grown on Ammono-GaN substrate. A series of DLTS measurements with different filling pulse lengths revealed a complexity of the EE1 trap level structure, since at least three emission signals with different capture and emission rates have been separated within the broad EE1 emission signal. Possible origins of the detected deep -level traps are discussed.

The results of conventional deep-level transient spectroscopy (DLTS) and high-resolution Laplace DLTS measurements of the FeGa(0/−) acceptor level in dilute AlxGa1−xN layers (x ≤ 0.05) grown by MOVPE technique (metal–organic vapor phase epitaxy) on native ammono-GaN substrates are analyzed and discussed. It is shown that the electron emission signal related to the FeGa acceptor level in AlxGa1−xN splits into individual components due to aluminum fluctuations in the second-nearest neighbor (2NN) shell around the FeGa impurity atoms. The calculations of the probability of finding a given number of aluminum atoms in the 2NN shell of the FeGa defect agree well with the experimental concentrations determined from Laplace DLTS peak intensities. This finding shows that in dilute AlxGa1−xN layers grown by MOVPE, aluminum and iron atoms are randomly distributed in the material. Finally, we demonstrate that the energy level of the FeGa acceptor with no Al atoms in the 2NN shell in the AlxGa1−xN samples shifts linearly with the aluminum content and the shifts are 28 and 55 meV relative to that in GaN for the samples with x = 0.025 and 0.05, respectively.

In this paper, we report on the electrical properties of two commonly observed deep levels in n-GaN known as E1 (E-c = 0.25 eV) and E3 center (E-c = 0.55-0.6 eV). The defects analysis has been carried out with the use of deep level transient spectroscopy and Laplace deep level transient spectroscopy techniques, respectively. We have found that both E1 and E3 trap concentration strongly depends on the distance to the original wafer edge, which is called here the radial effect, and decreases by over two orders of magnitudes as the distance increases. Moreover, we speculate that both deep levels can be related to iron or its complexes.

The work shows a successful fabrication of AlGaN/GaN high electron mobility transistor (HEMT) structures on the bulk GaN substrate grown by ammonothermal method providing an ultralow dislocation density of 10(4) cm(-2) and wafers of size up to 2 inches in diameter. The AlGaN layers grown by metalorganic chemical vapor phase epitaxy method demonstrate atomically smooth surface, flat interfaces with reproduced low dislocation density as in the substrate. The test electronic devices-Schottky diodes and transistors-were designed without surface passivation and were successfully fabricated using mask-less laser-based photolithography procedures. The Schottky barrier devices demonstrate exceptionally low reverse currents smaller by a few orders of magnitude in comparison to the Schottky diodes made of AlGaN/GaN HEMT on sapphire substrate.

The work shows new results showing advantages and disadvantages of using GaN and off-oriented SiC for high-electron-mobility-transistors fabrication. We will show a difference of defect density what influences the leakage current of the Schottky diodes, as well as thermal conductivity what is of importance in high-power devices. Possible influence on the life-time and noise level will be also discussed. Special attention will be paid to the substrate off-orientation. In the case of SiC substrates, 2-deg off may lead to cheaper technology of high-frequency devices and integrating GaN-based and SiC-based devices. In the case of GaN substrates, the off-orientation must be optimized as it influences the point-defect concentration, cracking, and layer morphology.

The first part of the paper shows the advantages and disadvantages of using GaN and SiC substrates with an off-cut up to 2 degrees. The following experimental observations were made: i) higher efficiency in p-doping with magnesium, ii) higher critical conditions for AlGaN cracking, iii) trigonal deformation of AlGaN and InGaN unit cells. A successful growth of AlGaN/GaN HEMT epi structure on 2 degree-off SiC substrates will be shown, what paves the way to a monolithic integration of GaN-based and SiC based devices on a common SiC substrate. In the second part of the paper, we show the experimental results on AlGaN/GaN degradation upon prolonged annealing at 600oC. In the case of epi structures on Si and sapphire, the parasitic parallel conductivity appeared, whereas in the case of SiC substrates, annealing introduced no distinct changes.

Experimental data on indium incorporation in InGaN and InAlN layers grown by metalorganic chemical vapor epitaxy (MOVPE) on bulk GaN substrates are presented and discussed. For the step-flow growth mode, realized for InGaN layers grown at relatively high temperatures (around 800°C), incorporation of indium increases with growth rate, and similarly, with a decrease in GaN substrate misorientation. Both dependences are explained by a higher velocity of flowing steps incorporating the indium atoms. For InAlN layers, three-dimensional nucleation takes place, and thus no significant changes of indium incorporation versus either growth rate or GaN substrate misorientation were observed.

The pressure dependence of the electronic band structure of InN is investigated with emphasis on the conduction band effective mass and its dependence on free-electron concentration. Transport measurements are performed under hydrostatic pressure on three n-type samples of InN with different electron concentrations. The effective mass extracted from the electron mobility is found to increase with pressure, however with a pressure coefficient, which is lower for the samples with higher electron concentration. Calculations of the InN band structure are performed within the density functional theory by means of the linear muffin-tin-orbital method, including an empirical adjustment to reproduce known band edges at ambient pressure. The calculated variations of the highest occupied conduction band energy and the electron effective mass with free-electron concentration are compared to available experimental information. The results show a pronounced deviation from parabolic behavior of the lowest conduction band, and for k>0 this induces large differences between the values of the optical and curvature masses. Both the fundamental band gap and the electron effective mass increase with pressure, but due to the nonparabolic character of the conduction band, the pressure coefficient of the effective mass decreases with electron concentration, in agreement with the experimental observation.

Calculations of the electronic band structure and effective mass of InN are performed within the density functional theory by means of the linear muffin‐tin‐orbital method. The results show a pronounced nonparabolicity of the conduction band. Calculated variations of the highest occupied conduction band energy and electron effective mass with free electron concentration are presented and compared to available experimental data. Pressure effects are studied. Both the fundamental band gap and the electron effective mass increase with hydrostatic pressure, but due to the nonparabolic character of the conduction band of InN the pressure coefficient of the effective mass decreases with electron concentration. Experimental verification of this behavior has been performed on three n‐type samples of InN with different electron concentrations. The measurements and calculations agree in the description of the dependence of effective mass and its pressure coefficient on electron concentration. (© 2008 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

We discuss the influence of indium segregation-induced disorder effects in InxGa1-xN alloys. Changes of the transport mechanism between InN and InxGa1-xN with x=0.58 were demonstrated by means of temperature-dependent conductivity measurements. Furthermore, an increase of (i) full width at half maximum of photoluminescence (PL) and (ii) the Stokes shift between PL and absorption was seen for samples approaching an In content of 0.5, which can also be attributed to growing disorder. Hydrostatic pressure dependent PL measurements of In-rich InGaN alloys are diacussed. Due to the fact that PL in InGaN originates from regions with higher-than-average In content, the luminescence pressure coefficient dE(E)/dp should not be associated with the average In content, but with the In content which is in accordance with the energy of the photon emission. This correction leads to a reduction of the large bowing of dE(E)/dp (associated with the band gap) which was reported earlier. Furthermore, it is shown that the electron concentration in InN has a significant influence on the measured value of dE(E)/dp.

Transport studies of as-grown and proton-irradiated n-InN have been performed aiming at verification of the nature of localized donor states resonant with the InN conduction band. These resonant donor states (RDS) show a clear contribution to the electrical conduction in low electron concentration InN epitaxial layers. We used proton irradiation to increase the number of incorporated native point defects of donor character in InN layers. Then, the performed studies of pressure dependence of the Hall electron concentration clearly show no increase in the number of RDS in samples exposed to irradiation in spite of the increase in the conducting electron concentration.

We have studied the pressure and temperature dependences of Hall electron concentration and mobility in degenerated, not intentionally doped InN samples. The results obtained for a whole set of samples with electron concentrations from 5.4 x 10(17) cm' to 3.3 x 10(18) cm(-1) consistently revealed the existence of a localized donor type state resonant with the conduction band. The concentration of this state is of the same order in all studied samples and not exceeds 10(18) cm. Population of this state determines electrical properties of InN samples with low carrier concentration. This donor state is not the main source of conducting electrons in these not intentionally doped samples and can be entirely populated and hidden in samples with high electron concentration exceeding about 3 x 10(18) cm(-3).

The authors studied the hydrostatic pressure dependence of photoluminescence (PL) from In-rich InxGa1−xN with In contents x between 0.58 and 1.00. The observed PL pressure coefficients of 20–25meV∕GPa agree well with previously reported experimental and theoretical values of the band gap pressure coefficient, from which they conclude that band-to-band recombination is responsible for PL emission. This contrasts with earlier reports, where relatively low PL pressure coefficients were interpreted as evidence of the involvement of strongly localized states in the PL emission. The reported observation of band-to-band recombination in In-rich InGaN is encouraging from the point of view of the construction of light emitters, since band-to-band recombination is more efficient than recombination via localized states. Furthermore, significant bowing of the band gap pressure coefficient in In-rich InxGa1−xN, as predicted by theory, is confirmed.

The hydrostatic pressure dependence of the polarization-induced interface charge in an AlGaN∕GaN heterostructure is investigated via capacitance-voltage (C-V) characterization of the two-dimensional electron gas (2DEG) at the heterojunction. The linear pressure coefficient of (3.3±0.4)×1011cm−2GPa−1 of the polarization-induced interface charge as obtained via C-V measurements of the investigated (28nmAl0.26Ga0.74N)∕GaN heterostructure agrees well with existing theoretical and experimental data (obtained via Hall-effect measurements). It is demonstrated that C-V characterization can be applied even when Hall-effect measurements of the 2DEG are inconclusive because of parallel conduction via the underlying GaN layer. In addition, we discuss the influence on the 2DEG concentration of interface trap states, which introduce a dependence of the measured 2DEG concentration on C-V test frequency. It is shown, however, that this effect does not influence the value of the pressure coefficient of the polarization-induced interface charge as determined via C-V characterization. Collaterally, the applicability of capacitance measurements of AlGaN∕GaN heterostructures for pressure sensing applications is confirmed.

We have studied the pressure dependence of Hall electron concentration and mobility in degenerated, not intentionally doped InN samples. The results obtained for a whole set of samples with electron concentrations from 5.4×1017cm−3 to 3.3×1018cm−3 consistently reveal the existence of a localized donor type state, resonant with the conduction band. Its energy position is estimated to be about 80–90 meV above the bottom of the conduction band. This donor state is not the only source of electrons in these not intentionally doped samples and can be entirely populated and hidden in samples with very high electron concentrations.