This paper critically investigates the advantages and limitations of the current-transient methods used for the study of the deep levels in GaN-based high-electron mobility transistors (HEMTs), by evaluating how the procedures adopted for measurement and data analysis can influence the results of the investigation. The article is divided in two parts within Part I. 1) We analyze how the choice of the measurement and analysis parameters (such as the voltage levels used to induce the trapping phenomena and monitor the current transients, the duration of the filling pulses, and the method used for the extrapolation of the time constants of the capture/emission processes) can influence the results of the drain current transient investigation and can provide information on the location of the trap levels responsible for current collapse. 2) We present a database of defects described in more than 60 papers on GaN technology, which can be used to extract information on the nature and origin of the trap levels responsible for current collapse in AlGaN/GaN HEMTs. Within Part II, we investigate how self-heating can modify the results of drain current transient measurements on the basis of combined experimental activity and device simulation

Multiple GaInN quantum wells (QWs) were grown on facets with reduced piezoelectric fields (PFs) of selectively grown GaN stripes oriented along the ⟨11¯00⟩ and ⟨112¯0⟩ directions by metalorganic vapor phase epitaxy. We found a higher normalized growth rate for the GaInN QWs on the {11¯01} facets compared to the {112¯2} facets and the planar grown reference sample on unstructured template. The different luminescence wavelengths observed for the QWs on these different facets can partly be explained by the reduced PFs, but additionally indicate that the In incorporation efficiency depends on the facet type. On stripes with trapezoidal cross section, we found strong interfacet migration of In and Ga changing the local thickness and composition significantly.

No abstract

The authors demonstrate the fabrication and evaluation of bright semipolar GaInN∕GaN blue light emitting diodes (LEDs). The structures are realized by growing five GaInN∕GaN quantum wells on the {11¯01} side facets of selectively grown n-GaN stripes with triangular shape running along the ⟨112¯0⟩ direction covered with a Mg-doped GaN top layer. The growth was done by metal organic vapor phase epitaxy using a conventional [0001] sapphire substrate. The devices have circular mesa structures with diameters between 70 and 140μm. Continuous wave on-wafer optical output powers as high as 700μW and 3mW could be achieved under dc conditions for 20 and 110mA, respectively. The current dependent blueshift of the peak emission wavelength caused by screening effects of the piezoelectric field was only 1.5nm for currents between 1 and 50mA. This is less than half the value measured on c-plane LEDs and confirms the reduced piezoelectric field in our LED structures.

Direction and strength of piezoelectric built-in fields of GaInN quantum wells have been experimentally determined. The quantum wells have been grown either on the conventional {0001} crystal plane of GaN or on {11¯01} facets of selectively grown GaN stripes. The emission peak position of the electric-field-dependent photoluminescence can be modeled and yields value and sign of the piezoelectric field dependent on the strain of the quantum wells. On the semipolar {11¯01} facets, the quantum wells show a much weaker field (−0.1MV∕cm) compared to quantum wells grown on polar {0001} planes (−1.9MV∕cm), consistent with theoretic predictions.