Intrinsic degradation mechanism of nearly lattice-matched InAlN layers grown on GaN substrates

Thanks to its high refractive index contrast, band gap, and polarization mismatch compared to GaN, In0.17Al0.83N layers lattice-matched to GaN are an attractive solution for applications such as distributed Bragg reflectors, ultraviolet light-emitting diodes, or high electron mobility transistors. In order to study the structural degradation mechanism of InAlN layers with increasing thickness, we performed metalorganic vapor phase epitaxy of InAlN layers of thicknesses ranging from 2 to 500 nm, on free-standing (0001) GaN substrates with a low density of threading dislocations, for In compositions of 13.5% (layers under tensile strain), and 19.7% (layers under compressive strain). In both cases, a surface morphology with hillocks is initially observed, followed by the appearance of V-defects. We propose that those hillocks arise due to kinetic roughening, and that V-defects subsequently appear beyond a critical hillock size. It is seen that the critical thickness for the appearance of V-defects increases together with the surface diffusion length either by increasing the temperature or the In flux because of a surfactant effect. In thick InAlN layers, a better (worse) In incorporation occurring on the concave (convex) shape surfaces of the V-defects is observed leading to a top phase-separated InAlN layer lying on the initial homogeneous InAlN layer after V-defects coalescence. It is suggested that similar mechanisms could be responsible for the degradation of thick InGaN layers. (C) 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4790424]

Lattice-matched AlInN alloys for nitride-based optoelectronic devices

Julien Dorsaz

Gallium Nitride (GaN) and its ternary alloys with aluminium and indium have met a growing interest in the last decade. These semiconductors have a large direct bandgap and can be doped with either silicon (Si) for n-type and magnesium (Mg) for p-type layers. Consequently, they are attractive targets for the fabrication of electrically driven light-emitting diodes, laser diodes and photodetectors, operating in the UV and visible range. Nitride semiconductors are also robust ans stable at high temperatures, suggesting their use in high frequency – high power electronic devices. The research on the III-nitride semiconductors is still in its infancy and, although several devices have been demonstrated and commercialized, there are still many issues to be solved. In particular, the III-nitride materials suffer from the lack of an adequate lattice-matched substrate for epitaxial growth, the high density of defects and dislocations, the poor quality of the p-type layers (low hole concentration and mobility), localization effects in indium containing layers, chemical inertness, etc… Also, due to the material quality issues, some electrical and optical parameters are only approximatively known. AlInN alloys have not attracted much attention compared to the InGaN and AlGaN ternary alloys, due to their unstable growth conditions. High-quality AlInN layers were deposited by metalorganic vapor-phase epitaxy (MOVPE). AlInN with an indium content of 18% is lattice-matched to GaN. It has a bandgap energy of about 4.35 eV and a refractive index contrast to GaN of 7% at 450 nm. It is shown that the optical, electrical and structural characteristic of epitaxial AlInN layers start degrading above a critical thickness of 200 nm, due to an ongoing pollution of the MOVPE reactor by products of the reaction. The distributed Bragg reflector (DBR) is an essential building block for advanced optoelectronic devices such as vertical-cavity surface emitting lasers (VCSELs). The introduction of the lattice-matched AlInN quarter-layers allows to grow thick DBR structures, with limited strain issues. It is used with GaN quarter-layers for visible reflectors and AlGaN quarter-layers for UV reflectors Optimizing the growth conditions of AlInN layers leads to the realization of DBRs with reflectivity above 99 % in the blue wavelength range (410-450 nm) as well as in the UV wavelength range (345 nm). The novel AlInN/GaN DBR is introduced in various microcavity LED (MCLED) structures. The basic features of the MCLED devices are presented, as well as optical simulations obtained using the transfer matrix representation (TMR) formalism, coupled with a dipole approximation for source terms. Electrical characterizations on these devices show that an intra-cavity contact (ICC) scheme should be used to achieve a sufficient electron injection in the active region. The external quantum efficiency of MCLED devices with thick AlInN/GaN DBRs (12 pairs) were lower than expected, due to the degradation of the crystalline quality of the active layers caused by the pollution of the reactor. Another requirement for the realization of a nitride VCSEL is the optimization of the electrical injection of carriers in the optical cavity. Assuming that circular metallic contacts are used, the carriers should spread laterally and be injected in the active region under the top DBR. This is generally achieved by the combination of a current spreading cap layer, which helps for the lateral injection of the carriers, and a current confinement layer, which inhibit the injection of the carriers in the active region outside of the vertical cavity. For the III-nitride structures, several methods are available for the lateral injection: n+/p+ tunnel junctions, doped-AlGa(In)N/GaN superlattices, semi-transparent electrodes, etc… A novel technique is developed to oxidize selectively surface or buried AlInN layers and form a stable, semi-insulating compound, with a refractive index of about 1.8. A detailed description of the technique is presented. It can be applied to form current confinement layers in VCSELs, but also layers for high reflectivity distributed Bragg reflectors (DBRs), layers for improved light confinement in planar waveguides or insulating buffer layers for field effect transistors (FETs). Also, oxidized AlInN could be used as a sacrificial layer for the preparation of free-standing GaN substrates. In this work, the technique was applied to form a current confinement aperture in a LED device. A buried AlInN layer was oxidized laterally, leaving a small 5 µm aperture in a 28 µm × 28 µm mesa structure. Optical and electrical confinement are demonstrated. With the availability of a high-reflectivity lattice-matched nitride DBR and a technique to produce small current apertures, both based on the new AlInN material, the realization of an electrically injected VCSEL becomes a realistic goal. An ideal structure is realized and discussed.

EPFL2006

Physics and applications of intersubband transitions in nitride based heterostructures

Sylvain Nicolay

In the past decades, III-nitride semiconductor compounds have attracted an increasing amount of interest due to their applications to blue-violet laser diodes and white light emitting diodes, or for their use as ultraviolet emitting devices for biomedical applications. In addition, these materials can also be used to achieve high power, high frequency electronic devices, such as high electron mobility transistors. Furthermore, thanks to the large conduction band offset of 1.75 eV in the AlN/GaN system, it is also possible to develop active optoelectronic devices working in the telecom range using intersubband transitions (ISBT). This is quite interesting as in nitride compounds these transitions are characterized by fast recombination dynamics, which should open new prospects for optoelectronic devices operating in the THz regime. Another property of interest of nitride compounds is their mechanical hardness and good thermal behavior, which enables foreseeing applications in harsh environment (space). Up to this work, the only reported results of ISBT in the telecom range were achieved using molecular beam epitaxy (MBE) growth technique. On the other hand, ISBT energies reported on samples grown with metal-organic vapor phase epitaxy (MOVPE) were far below the telecom limit. Nevertheless, MOVPE being the technique of choice for the production of nitride based optoelectronic devices, it is worth from a technological point of view to investigate whether or not it is possible to reach the telecom range with this technique. In this work, the origins of the discrepancies between results obtained with the two techniques will be investigated. First the effect of different growth parameters over ISBT will be studied for MOVPE grown samples. This will lead to the conclusion that with this growth technique, strain plays a critical role on the AlN/GaN heterostructure properties. In fact, it will be shown that strain induced interface instability occurring at the well/barrier interface is detrimental to reach the telecom range. It will be proven that, for MOVPE grown samples, such wavelengths can be reached if the active region is deposited on AlN templates. Following investigations carried out on MOVPE structures, MBE grown samples will also be studied and it will be shown that for complex heterostructures ( e.g. with coupled QWs), this growth technique is better suited. This is notably due to its lower growth temperature leading to a decrease of interdiffusion/segregation effects, also detrimental to ISBT energy. In addition, AlInN lattice-matched to GaN will be investigated in order to achieve ISBT in the mid-infrared range with low defect density heterostructures. It will also be shown that thick LM AlInN waveguide layers operating in the telecom range with low propagation losses can be grown by MOVPE. Finally, the development of ISB based nitride devices will be studied and a first electro-optical modulator working up to a frequency of 2.5 GHz will be demonstrated.

EPFL2008

Impact of Strain and Alloy Disorder on the Electronic Properties of III-Nitride Based Two-dimensional Electron Gases

Pirouz Sohi

Over the past 20 years, III-nitrides (GaN, AlN, InN and their alloys) have proven to be an excellent material group for electronic devices, in particular, for high electron mobility transistors (HEMTs) operating at high frequency and high power. This is mainly thanks to the wide band gap of III-nitrides and the spontaneous polarization along the technologically relevant (0001) direction. Heteroepitaxy of III-nitride heterostructures enables the formation of two-dimensional electron gases (2DEGs) with high carrier densities (> 10^13 cm^-2) and high electron mobilities (2000 cm^2V^-1s^-1). The heterostructures usually consist of AlGaN, InAlN or AlN barriers grown epitaxially on GaN buffers. Devices based on these structures exhibit excellent performance and have reached technological maturity, enabling notably the commercialization of AlGaN/GaN HEMTs. Despite the success of III-nitride HEMTs thus far, the full potential of the material group for electronics has not yet been unleashed. The aim of this thesis is to investigate two novel heterostructure designs and to determine the physical mechanisms limiting their electronic properties. This was done by epitaxial growth followed by characterization of material and electronic properties. The first studied heterostructure was AlN/GaN/AlN where the GaN channel is fully strained to AlN. In order to achieve pseudomorphic growth of GaN on AlN, the onset of strain relaxation, i.e., the critical thickness, is of paramount importance. Furthermore, the impact of growth parameters such as temperature and initial dislocation density need to be considered. Over the past years several, research groups have reported on the electronic properties of ultra-thin GaN channels on AlN. Interestingly, all reported 2DEGs suffer from a low electron mobility with values usually below 600 cm^2V^-1s^-1. In this thesis, the strain relaxation and critical thickness of GaN on AlN were determined. Furthermore, the origin of low electron mobility for AlN/GaN/AlN heterostructures and in general thin-GaN channels grown on AlN were systematically studied. The second heterostructure that was investigated during this thesis was based on InGaN channels. In this case, the 2DEG is formed in an In-rich InGaN layer. The low electron effective mass of InN (0.05 m0) compared to GaN (0.2 m0) should theoretically give rise to a higher electron mobility. However, for In-rich channels alloy disorder scattering is particularity strong in III-nitrides. This has been shown both by optical measurements and theoretical calculations based on the localization landscape theory. Surprisingly, high electron mobilities have been reported in literature for high In content InGaN channels. In these cases, the electron mobility appears not to be limited by alloy disorder scattering. During this thesis, the electron mobility of InGaN channels was determined as a function of In content. Furthermore, the unintentional growth of GaN interlayers is proposed as the possible origin for the reported high electron mobilities in literature.

EPFL2019