Nano

Resume : Chemical etch processes for GaN materials and devices are significantly underdeveloped due to its apparent inertness to common wet etchants. For the processing and fabrication of optical structures and devices, wet anisotropic (crystallographic) chemical etching is a key process technology for the fabrication of semiconductor devices. Thus, to fully realize the potential of the III-nitrides in new opto-electronic devices, such as in optical nano- and microcavity lasers, more complete knowledge and development of techniques with anisotropic etching are needed. Here, we explore the etch characteristics of GaN using the general geometric principles of crystallographic dissolution processes to enable the prediction of facet-determined etch structures. Significant etch rates albeit with extremely high crystallographic anisotropy are observed in KOH-based etchants, reaching ratios of > 2000:1 (compared to Si of up to ~160:1 and GaAs of ~6:1). Fabricating novel GaN based devices requires understanding the consequences of this extremely high etch rate anisotropy for the resulting etch geometries. For silicon, full knowledge of orientation-dependent etch rates and geometry evolution has enabled design of silicon MEMS structures. Here, we use the same framework (Wulff-Jaccodine method) for understanding and predicting the geometry of the facet evolution and final structure, and apply it to GaN. We perform the first complete orientation-dependent etch rate measurements, clearly showing fast a-plane etch rate relative to m-plane, and use them to predict the faceting of pillar structures with good agreement with experiment. Finally, we use these developments in GaN wet etching to fabricate functional optical nanowire and microcavity structures, where faceting enables smooth sidewalls for a high quality cavity structure. Control over the lasing properties of nanowires fabricated using this method, including single-mode lasing, polarization control, wavelength tuning, and beam shaping, will also be highlighted. This work was performed, in part, at the Center for Integrated Nanotechnologies, a U.S. Department of Energy, Office of Basic Energy Sciences user facility. Sandia National Laboratories is a multi-mission laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.

Resume : GaN on Si substrate is an attractive material because it allows large diameter, low cost, and integration with Si devices. However, high-density dislocations generate in GaN layer due to the difference in lattice constants with Si. In this study, we have developed the method of dislocation reduction by using in-situ dry etching of GaN within MOCVD reactor. AlN buffer and GaN layers were grown on Si(111) substrates. A SiNx layer was deposited on the surface of GaN layer by using Tetramethylsilane and NH3, and then GaN was etched in H2 and NH3 ambient. By etching, many pits appeared on GaN surface, which corresponds to threading dislocations in GaN layer. The SiN layer was needed to fabricate deeper pits. Finally additional GaN was re-grown on the base GaN layer. The deeper pits could prevent the propagation of dislocations to the second GaN layer. In the second GaN growth, the density of GaN nuclei became lower with the decrease of the V/III ratio. As a result we could achieve low dislocation density because the frequency of the coalescence between GaN crystal islands became low. For the further reduction of dislocation density, it is desirable to deposit second SiNx layer on pits after the etching process. The second SiNx prevent GaN nuclei from growing on the edge of the pits. By using this process, the dislocation density of top GaN layer is reduced to about 6.3E7/cm2 and this is a cost effective method completing all the necessary processes in an MOCVD reactor.

Resume : Bright solid-state single-photon sources based on quantum dots (QDs) have much attention for a key component of quantum information technology. Fabricating single photon emitter with high collection efficiency from QDs to optical waveguide is important task for quantum information devices. However, conventional QDs embedded in a high index planar substrate have low collection efficiency less than 5 %. Although various photonic structures such as pillar and nano-wire have been suggested to enhance the collection efficiency, few studies have focused on controlling the emission direction of single photon. In this study, we propose a QD embedded in an inverted pyramid structure for high collection efficiency as well as controllability of emission direction. Silver-coated InGaN/GaN nano-pyramid structure were grown on sapphire substrate. Using an ultraviolet-curable optical adhesive material, the nano-pyramid structures can be detached from a sapphire substrate, which formed inverted pyramid structure. Highly unidirectional single photon emission from a QD in the detached inverted-pyramid structure was observed, which radiated with the Gaussian shaped radiation profile. We emphasize that this highly directional emission from the nano-pyramid structure can easily be integrated with a predesigned photonic waveguide system with high coupling efficiency.

Resume : Selective area sublimation can be used to form 1-dimensional (D) or 0D nanostructures from well controlled 2D GaN or InGaN/GaN layers. The principle is to partially mask 2D epitaxial layers by a thermally resistant mask such as SixNy, and then to sublimate under vacuum the regions left uncovered. Due to strong differences in the evaporation rates of the different crystalline planes of GaN, specific shapes can be achieved. In particular, vertical m-planes can be revealed and then nanowires can be formed. The obtained shape depends on the mask geometry that can be made ex situ using standard photolithography or e-beam lithography. Also, in situ SixNy nano-masking is possible using molecular beam epitaxy by exposing the GaN surface to Si. Depending on the Si dose and the sublimation time and temperature, the density and the size of the obtained nanostructures can be strongly varied. Such approach can be used to obtain for example diluted nanopyramids with a density below 108 cm-2 or highly dense arrays of nanowires (mean diameter of 7 nm) with a density above 1011 cm-2. Applying this process to layers incorporating InGaN quantum wells allows forming InGaN quantum discs showing photoluminescence and cathodoluminescence spectra characterized by lateral confinement, strain effects and narrow linewidth for isolated emitters.

Resume : Single self-assembled GaN nanowires (NWs) grown by molecular beam epitaxy are essentially free of extended defects. However, GaN NWs typically nucleate in such a high density that most NWs coalesce, a phenomenon that results in the formation of dislocations. Beside this problem, their high density is also an insurmountable obstacle for the fabrication of homogeneous core-shell heterostructures because the impinging elements cannot directly reach the bottom part of the NW sidewalls. Here, we present two growth approaches to reduce the density while preserving the structural perfection of self-assembled GaN NWs. The first approach consists of taking advantage of the fact that GaN NWs do not form on uniform cation-polar substrates. By depositing Si on a Ga-polar GaN layer, we are able to locally reverse the polarity and thus to induce the formation of N-polar GaN NWs. We demonstrate that the NW density can be tuned by varying the amount of pre-deposited Si. The second approach exploits the high diffusivity of Ga adatoms on TiN layers. On TiN, we find that the NW density is determined by diffusional repulsion among GaN nuclei, which facilitates the growth of ensembles of uncoalesced GaN NWs with small diameters and a homogeneous length of more than 1 µm. Both growth approaches enable a higher degree of control on the properties of self-assembled GaN NWs and pave the way for the fabrication of core-shell heterostructures without relying on ex situ pre-patterned substrates.

Resume : 3-dimensonal core-shell InGaN/GaN microLED structures are promising platforms for next generation LEDs. Such novel structures are known to embrace several merits such as increased active area, lower threading dislocation density and almost strain-free topology, even if a high Si doping is used to promote the aspect ratio during growth. Additionally, the light emitting area is mainly constructed on m-plane sidewalls, where the QCSE is absent. By X-ray diffraction measurement, different strain relaxation behaviors were observed on Ga- and N-polar GaN microrods fabricated by selective area MOVPE growth on patterned SiOx/GaN/sapphire and SiOx/sapphire templates, respectively. The strain state and defect density dependence on the rod’s aspect ratio as well as on the intentional Si doping concentration has also been studied. TEM characterization of single GaN microrods depicts the defect development inside these 3D structures. The indium distribution gradient within a core-shell microLED structure is drastically improved via optimized MOVPE growth experiments, which has been evidenced by probing of cathodoluminescence spectra along the rod’s sidewall. Room-temperature time-resolved photoluminescence (TRPL) measurement performed on InGaN/GaN core-shell structures shows a mono-exponential decay behavior with decay times of a few hundred picoseconds. Further temperature-dependent TRPL measurements will also be presented giving insights into the recombination dynamics and IQE.

Resume : III-nitride nanowires (NWs) are of significant interest for optoelectronic and energy harvesting applications, which practically require tight growth control as their electronic and optical properties are polarity-dependent. Also a reliable method to characterize doping is a prerequisite. In this work, we demonstrate the capabilities of Kelvin probe force microscopy (KPFM) as a non-destructive technique for the determination of the polarity and assessment of doping in GaN NWs. This has been implemented on MBE grown GaN self-assembled NWs with mixed polarity and NWs with p-n junction dispersed on atomically flat highly oriented pyrolytic graphite (HOPG). For GaN NW assembly, this technique allows the assessment of the polarity of individual NWs from large images (tens of µm2), and provides the statistics on the polarity of the ensemble hardly measurable by other methods [1]. Our results show that the majority of NWs exhibiting N-polarity show CPD values much lower than that of Ga-polar NWs. These observations are quantitatively consistent with the same measurements on reference samples. For doping assessment in p-n junctions NWs, in p- and n-type single GaN NWs dispersed on HOPG, KPFM allowed spatial localization of the junction and surface built-in potential. KPFM measured in dark and under UV-illumination for photovoltage measurement show that the shift of the Fermi levels with respect to that of HOPG are the most important effects under above band-gap illumination, while changes in surface band bending are minor. The Fermi level shift was measured opposite for n-type and p-type NWs, explainable in terms of charge transfer at the metal-semiconductor junction in non-equilibrium. References [1] A. Minj et al. Nano Lett. 15, 6770-6776 (2015)

Resume : III-nitride quantum dots (QDs) have significant potential for single-photon sources or gain media for low threshold and high efficiency visible and UV lasers, among others. “Bottom-up” Stranski− Krastanov growth is widely used, however both the size distribution and densities are difficult to precisely control. Here, we show that a top-down fabrication process that itself can be controlled by the properties of the nanostructures being fabricated. This process, called quantum size controlled photoelectrochemical (QSC-PEC) etching, uses laser excitation at a selected and narrowband wavelength to control the final sizes of the QDs through an etch that self-terminates when the QD band gaps increase due to quantum confinement effects until they exceed the energy of the incident photons. Beginning with epitaxially grown InGaN films, we examine the etch process from large to quantum-scale nanostructures with AFM, TEM, and photoluminescence measurements. Quantitative analysis of size and density of the ensemble are made after image-post processing techniques and deconvolution of the AFM tip and QD. We further investigate the passivation of the QDs through regrowth of AlGaN and GaN capping layers, as well as fabrication of multilayers of QDs from multiple-quantum well structures. These results show the potential for a combination of unprecedented uniformity and density towards InGaN QD devices. Sandia National Laboratories is a multi-mission laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-AC04–94AL85000.

Resume : Electrical characterizations of basic GaN nanostructures, such as p-n junctions, are becoming indispensable and crucial for devices realization. However, it is challenging to characterize GaN NW p-n junctions, due to difficulties to control Mg dopants incorporation and issues to make contacts on p-type NWs. Only a few techniques [1,2,3] have been proposed so far, nevertheless, a quantitative understanding of the transport properties remains incomplete. Recently, electron beam based techniques have been widely employed to study semiconductor nanostructures [4,5] in our group. In this study, we have performed electron beam induced current (EBIC) on single axial GaN p-n junction NWs grown by plasma-assisted molecular beam epitaxy. Thanks to an improved contact process, both the electric field at the p-n junction (under reverse bias) and at the contact on the p-side (under positive bias) have been located and delineated by EBIC signal. Analyzing EBIC profile in the vicinity of the p-n junction [5], we can deduce a depletion width in the range of 170-180 nm, and also local minority carrier diffusion lengths on p-side and n-side. Following our previous work [6], donor Nd and acceptor Na doping levels were estimated to be Nd = 2-3 × 1e18 at/cm3 and Na = 2-3 × 1e17 at/cm3. A model describing the carrier diffusion processes in the NW will be proposed and discussed. [1] A. Imtiaz et al. Appl. Phys. Lett. 2014, 104, 263107. [2] M. Brubaker et al. J. Electron. Mater. 2013, 42, 868. [3] Y. Lu et al. ACS Nano 2013, 7, 7640. [4] F. Donatini et al. Nano Lett. 2016,16, 2938. [5] P. Tchoulfian et al. Nano Lett. 2014, 14, 3491. [6] Z. Fang et al. Nano Lett. 2015, 15, 6794.

Resume : Light-emitting diodes (LEDs) based on three-dimensional core-shell microrods (µRods) have emerged as a promising approach for high efficiency solid-state lighting. Due to the nonpolar orientation of their side facets, high aspect-ratio µRods spontaneously emit linearly polarized light. Therefore, to understand the light emission and propagation within and out of µRod LEDs, which both depend on the optical polarization, it is essential to characterize the polarization state of their emission. We investigate the optical emission properties of the active InGaN shell of high aspect-ratio InGaN/GaN cores-shell µRods via confocal resonant polarization-resolved, and excitation density dependent micro-photoluminescence. We observe a homogeneous emission intensity along the whole µRods with a red-shift of 30 nm from the base to the tip. A high degree of linear polarization (DLP) of 0.6-0.66 is measured on the m-plane facets. Assuming an internal quantum efficiency (IQE) of 1 at 10 K, the IQE varies at 300 K from 67±6 % at the base, to 73±7 % in the center and 55±11 % near the tip of the µRods. Simultaneously fitting the IQE and the DLP drop vs. the excitation density with a rate equation model, we are able to extract the recombination rate coefficients. The obtained high values of the radiative rate coefficient as well as the high IQE strongly support the µRod LED approach for highly efficient solid-state lighting. Reference: C. Mounir et al., J. Appl. Phys. 121, 025701 (2017)

Resume : Although III-nitride micro-wires (MWs) benefit of the absence of extended defects, their nonpolar surface orientation, and high surface-to-volume ratios, the frequent presence of alloy and layer thickness gradient in the core-shell geometry cause non-uniform current injection, uncontrolled emission spectra and point defects. To understand the influence of inhomogeneities on the carrier dynamics, we investigate here by time-resolved cathodoluminescence single core-shell MWs integrating an InGaN/GaN m-plane quantum well (QW). At 4 K, a significant redshift of the QW emission from the bottom to the top of the MW is observed, which results from both the In-content and the QW thickness gradient.[1] At 300 K, the drop of the internal quantum efficiency from 50% at the bottom to 10% at the top can be related to a change of the effective lifetime (τeff ) from 220 to 120 ps. This enables us to differentiate the radiative (τr) and non-radiative (τnr) contributions to the lifetime. τnr decreases from 450 to 140 ps from the bottom to the top, which’s related to the increased In-content and the larger QW width in the upper part of the MW that cause a higher density of point defects. Moreover, thanks to the absence of polarization fields, τr remains below 1.4 ns up to room temperature. Such short radiative lifetimes even with thick active layers suggest that in core-shell MWs the efficiency droop could be pushed to higher current densities. [1] Müller, M. et al. Nano Lett. 16, 5340 (2016).

Resume : Unlike GaN nanowires (NWs) which have been thoroughly studied in the recent decades, there have been only few reports on their AlN NW counterparts.1 AlN NWs are desired for ultra-violet optoelectronic applications, but their fabrication is complicated for several reasons. One of them is strong wetting that AlN exhibits on a variety of foreign crystal substrates, preventing thus development of columnar morphology. In this work we report on growth, structural and optical characterization of self-assembled AlN NWs achieved on ~500 nm thick amorphous SiO2 (obtained via thermal oxidation of Si(111)). The NWs were grown by molecular beam epitaxy at ~940 ˚C, under highly nitrogen-rich conditions (Al/N~0.5). Scanning electron microscopy reveals formation of columnar structures, which grow preferentially vertical on the underlying SiO2, with ~600 nm heights, 50 – 150 nm diameters and show a certain degree of coalescence. X-ray diffraction confirms that the NWs grow perpendicular to the underlying SiO2 (with ±3˚ average tilt), but without epitaxial relation to it (i.e. with an arbitrary twist). It also shows that NWs grow nearly free of strain (<0.1%). Their Raman spectra exhibit intense and relatively narrow E2h modes, the mode position further confirming very low level of biaxial strain. Transmission electron microscopy with atomic resolution performed on single NWs, reveals that they have wurtzite structure with Al polarity and excellent crystal quality in the observed region. [1] O. Landré et al, “Molecular beam epitaxy growth and optical properties of AlN nanowires”, Appl. Phys. Lett. 96, 061912 (2010).

Resume : The use of quantum dots (QDs) is a proven way to improve the internal quantum efficiency of LEDs by preventing exciton nonradiative recombination on threading dislocations. GaN QDs are generally obtained via a Stranski–Krastanov (SK) growth mode resulting in a self-assembled organisation of 3D islands connected by a 2D wetting layer. However, if QDs nucleate close to dislocations, at sites with a local extension of the lattice parameter, a nonradiative channel might be still available for the excitons confined within these QDs. We study the nucleation, growth mechanism and optical properties of GaN/AlGaN QDs fabricated by ammonia-assisted molecular beam epitaxy. The QD morphology at different stages of the QD evolution was carefully investigated by scanning transmission electron microscopy (STEM). From initially flat disks the QDs take a perfect pyramidal shape with {1-103} lateral facets. Subsequent evaporation of GaN results in a complete vanishing of the GaN wetting layer and therefore leads to the formation of an array of disconnected QDs. GaN/AlGsN QD sample gives a strong macroscopic response in photoluminescence, despite the fact that QDs nucleate preferentially in a close vicinity of threading dislocations. A cathodoluminescence of individual QDs was studied as a function of the distance to the nearest dislocation in a dedicated scanning transmission microscope having a 10 nm spatial resolution. This work is supported by the “NANOGANUV” project ANR-14-CE26-0025-01.

Resume : Solar cells are one of the major source of renewable energy available today. InGaN alloys are promising candidates for future solar cells due to their direct tunable bandgap between 0.7 eV and 3.4 eV, covering major absorption range of solar spectrum. However, the conversion efficiency of InGaN-based solar cells remains limited because of the poor crystalline quality of thick InGaN layers, particularly at high indium composition. In this work, to overcome this limit, we propose two structures based on InGaN core?shell and axial InGaN nanowires. We performed the hydride vapor phase epitaxy (HVPE) to directly grow n-type GaN wires and InGaN nanowires on p-type silicon substrate. The metalorganic vapor phase epitaxy (MOCVD) was applied to grow the InGaN core?shell structure. The energy dispersive X-ray spectroscopy (EDS) and cathodoluminescence measurements revealed that nanowires grown on silicon exhibit an uniform indium distribution of ~ 40?50 %, with an ideal bandgap configuration of InGaN 1.8 eV/Si 1.12 eV for high conversion efficiency. Then, the photovoltaic properties of the fabricated solar cells were assessed. As a consequence, short-circuit current density and open-circuit voltage were typically ~20 µA/cm2 and 0.1 V under a 1 sun (1.5 AMG) illumination, respectively. Even though, the efficiency measured was still low, this work suggests the potential of directly integrated InGaN nanowires on silicon for solar cell applications.

Resume : The control and characterization of nanowires (NWs) doping is a prerequisite for the realization of devices. The literature on the topics is, however, scarce due to the NW geometry itself, which makes the use of standard methods, such as SIMS or Hall effect measurements, difficult or even impossible in most cases. In particular, the issue of dopant segregation is a crucial one, specific of NWs, and needs to be clarified in order to fully understand/control electrical transport properties. Using a combination of SEM, quantitative EDX and Atom Probe Tomography, we will show that even though Si solubility limit is higher in GaN NWs than in thin layers, Si atoms are segregating at the periphery above a critical concentration [1]. Furthermore we find evidences that Mg is strongly accumulating at the periphery of GaN NWs, with a marked incorporation gradient along the radius. This suggests that Mg is preferentially incorporated in the m-plane, possibly assisted, at least partially, by the presence of residual hydrogen. Our data also put in evidence that the lateral diffusion of Mg in the m-plane is limited by the presence of edges, likely due to the Ehrlich-Schwöbel barrier prohibiting the diffusion from one m-plane facet to the adjacent ones [2]. These results demonstrate that radial segregation of Si and Mg is a general feature in nitride NWs. It is expected that this lack of volume homogeneity of the doping may drastically affect the electrical transport properties of such NWs and should be taken into account to fully control NW-based devices. [1] Z. Fang et al. Nano Lett. 2015, 15, 6794-6801 [2] A-M Siladie, L. Amichi, et al, to be published