Ce document porte sur le développement d’un nouveau banc de mesure pour la caractérisation de l’impédance thermique des HEMTs GaN. Le banc développé repose sur la méthode dite « 3ω » qui consiste à mesurer l’harmonique 3 d’un signal électrique véritable image des variations thermiques du composant. Un balayage en fonction de la fréquence d’excitation conduit à l’extraction de l’impédance thermique. Les résultats de mesures ont été validés par les simulations électriques. Des études complémentaires ont été réalisées pour l’identification des effets de pièges en utilisant différentes méthodes permettant l’extraction de la signature des pièges. La réalisation des modèles non-linéaires est présentée pour les transistors HEMT AlGaN/GaN et InAIN/GaN pour des applications d’amplificateur de puissance dans les bandes de fréquences X et K. / This report is devoted to the development of a new measurement bench for thermal impedance characterization of GaN HEMTs. This measurement test set uses the so-called « 3ω » technique, which consists to measure the electrical signal at third harmonic real image of the thermal magnitude variations of the device. A sweep in function of the excitation frequency allows extracting of the thermal impedance. The measurement results have been validated by electrical simulation. Other complementary studies were performed to identify the trapping effects using different methods to extract the traps signature. The realization of nonlinear models is presented for AlGaN HEMT / GaN and InAIN / GaN to the power amplification applications in frequency bands X and K.

With the growth of smart phones, the demand for more broadband, data centric technologies are being driven higher. As mobile operators worldwide plan and deploy 4th generation (4G) networks such as LTE to support the relentless growth in mobile data demand, the need for strategically positioned pico-sized cellular base stations known as ‘pico-cells’ are gaining traction. In addition to having to design a transceiver in a much compact footprint, pico-cells must still face the technical challenges presented by the new 4G systems, such as reduced power consumptions and linear amplification of the signals. The RF power amplifier (PA) that amplifies the output signals of 4G pico-cell systems face challenges to minimize size, achieve high average efficiencies and broader bandwidths while maintaining linearity and operating at higher frequencies. 4G standards as LTE use non-constant envelope modulation techniques with high peak to average ratios. Power amplifiers implemented in such applications are forced to operate at a backed off region from saturation. Therefore, in order to reduce power consumption, a design of a high efficiency PA that can maintain the efficiency for a wider range of radio frequency signals is required. The primary focus of this thesis is to enhance the efficiency of a compact RF amplifier suitable for a 4G pico-cell base station. For this aim, an integrated two way Doherty amplifier design in a compact 10mm x 11.5mm monolithic microwave integrated circuit using GaN device technology is presented. Using non-linear GaN HFETs models, the design achieves high effi-ciencies of over 50% at both back-off and peak power regions without compromising on the stringent linearity requirements of 4G LTE standards. This demonstrates a 17% increase in power added efficiency at 6 dB back off from peak power compared to conventional Class AB amplifier performance. Performance optimization techniques to select between high efficiency and high linearity operation are also presented. Overall, this thesis demonstrates the feasibility of an integrated HFET Doherty amplifier for LTE band 7 which entails the frequencies from 2.62-2.69GHz. The realization of the layout and various issues related to the PA design is discussed and attempted to be solved.

No description available.

III-nitride mainly GaN semiconductors are the most important materials for a wide range of applications, in particular high-power devices, due to the tunable direct bandgap, their chemical, and thermal stability. However, their growth on suitable substrates is still problematic, and low UV GaN efficiency hinders the efforts aimed at improving the performance of emitting devices. This dissertation presents novel growth and device fabrication methods capable of overcoming these issues using different novel strategies. The work reported in this dissertation comprises five parts. The first two parts demonstrate a new low-cost pulsed laser deposition (PLD)-based strategy for large-scale applications. This was developed to grow high-quality dislocation-free GaN NWs epitaxially on any bulk, flexible, or two-dimensional (2D) substrates without a catalyst, irrespective of the lattice mismatch or type of the substrate. As part of the work reported here, Si, p-GaN, Ga2O3, sapphire, graphene, MXene, and transition-metal dichalcogenide (TMD) substrates were utilized. Also, the adopted growth mechanisms are discussed, along with the advanced structural and optical characterizations. Advanced structural and optical characterizations further confirm the growth mechanism and demonstrate the superior optical and structural quality of GaN NWs. In the third part, a novel multiple quantum wells (MQWs)-based structure grown on the NWs is described, indicating that these NWs can be used as a template to grow III-nitride-based devices. In the fourth part of the work, the significance of these GaN NWs is further demonstrated by reporting on the fabrication of a high-performance self-powered broadband photodetector incorporating these NWs hybridized by two perovskite types: organic/inorganic as well as all-inorganic perovskites (CH3NH3PbI3 and CsPbBr3), revealing two different self-powered photodetector characteristics with high photo-responsivity at 0V. In the last part of this work, the focus is given to a new environmentally friendly strategy to enhance the device UV emission efficiency by functionalizing GaN NWs with solution-processed p-MnO quantum dots (QDs) characterized by much wider bandgap energy than that of GaN. The energy transfer mechanism from QDs to NWs is also discussed using different structural and optical characterizations. This novel strategy is based on drop-casting QDs on NWs, which is simple, cost-effective, and applicable for large-scale applications.

QC 20100623

In this work, we propose a model to address the challenge of droop in internal quantum efficiency in InGaN/GaN Multiple Quantum Well LEDs. Efficiency droop limits the performance of high brightness LEDs as they operate at currents greater than 350mA. The efficiency droop is a multi-physics problem posed by various entities such as (1) dislocation recombination, (2) Auger recombination in active region, (3) non-radiative recombination, and (4) current overflow in the active region. This work aims at reducing the droop associated with non-radiative recombination by engineering the quantum well barrier thickness and materials. The goals are three-fold, namely: (1) To explore the role of barriers in determining the droop in internal quantum efficiency and to justify the use of multiple barriers to increase the carrier density and reduce the leakage current thereby increase the radiative recombination at higher current densities ; (2) Propose optimum barrier specifications such as number, material combination, and thickness for downscaling the efficiency droop, and thereby improving the device efficiency; and (3) Finally, obtain improved efficiency by engineering the barrier in a realistically-sized device by considering the effects of long-range strain fields in the device.

GaN is a III-V semiconductor that is a promising material used in production of light emitting devices and high power/high frequency electronics. The electronic and optical properties of GaN are subdued by defects that occur during the growth processes of this material. The emitted photoluminescence (PL) from optically excited GaN gives insight into the origins and effects of point defects within the crystal lattice structure of GaN. In this study, PL spectroscopy is used to examine and analyze the point defects that occur in Zn-doped GaN. The blue luminescence band seen in undoped and Zn-doped GaN have identical fine structure and properties. This band is attributed to a ZnGa acceptor. In Zn-doped, the PL intensity quenches abruptly at certain temperatures, which increase with increasing excitation intensity. This behavior is different from the PL quenching in undoped GaN. The PL behavior was simulated with a phenomenological model based on rate equations. A program created with mathematical modeling software, in conjunction with the basic rate equations, was used to explain the unusual behavior of the abrupt thermal quenching observed in Zn-doped GaN.

Photoluminescence (PL) has been studied from different types of bulk GaN samples grown by hydride vapor phase epitaxy technique at Kyma Technologies. Point defects in bulk and at the surface affect the electrical and optical properties of GaN and could be analyzed by PL. The surface of the samples was polished with different techniques: one is chemical mechanical polish (CMP) and another is mechanical polish (MP). PL data from MP and CMP surfaces show that PL intensity from the CMP-treated surface is much higher than that from the MP-treated surface. This can be explained by defects formed during the process of MP polish. However, after the MP-treated surface is etched with RIE method, the optical quality of the MP-treated surface improves. In particular, as the depth of etching increases from 50 nm to 700 nm, the PL intensity increases by a factor of 1000. PL from the CMP surfaces of undoped bulk GaN samples contains a broad red luminescence (RL) band and a broad green luminescence (GL) band. However, PL from the CMP surfaces of Fe-doped GaN samples contained a blue luminescence band (labeled as BL2 in literature) and the yellow luminescence (YL) band. PL from MP-treated surfaces (both undoped and Fe-doped) was very weak and it contained relatively narrow red and green bands. These bands, labeled RL2 and GL2, respectively, are quenched at relatively low temperatures, in contrast to the RL and GL bands which are almost independent of temperature in the range from 15 to 300 K.

The effect of ambient on photoluminescence (PL) from GaN was studied. We found that the PL intensity in vacuum was nearly four times higher than in air. The PL intensity also increased after etching the sample in Aqua Regia and BOE to remove the native oxide layer. After etching, the PL intensity was very stable in vacuum, but substantially degraded in air ambient. In HCl vapor (low pH), the PL intensity increased as compared to air ambient, while in NH3 vapor (high pH) it decreased. The quantum efficiency of the exciton and blue luminescence bands increased significantly with increasing excitation power density. This increase could not be explained by reduction of the depletion region width (field effect mechanism), but could be explained by changes in the nonradiative recombination rate at the surface (recombination mechanism). We therefore assume that in vacuum and acid vapor some surface species are desorbed or passivated, resulting in a decreased nonradiative recombination rate and increased PL intensity.

Ce manuscrit présente les résultats d’une analyse exhaustive des mécanismes physiques qui limitent les performances et la fiabilité des transistors à haute mobilité d’électrons (HEMT) sur nitrure de gallium (GaN). En particulier : • Les phénomènes de dégradation à fort champ électrique des HEMT sur GaN sont analysés en comparant les données expérimentales avec les résultats de simulations physiques. Des stresses DC de 150 heures ont été effectués en conditions de canal ouvert et de pincement. Les effets des dégradations qui ont caractérisé ces deux types de stresses sont les suivants: une chute de courant DC de drain, une amplification des effets de gate-lag, et une diminution du courant inverse de grille. Les simulations physiques indiquent que la génération simultanée de piéges de surface (et/ou barrière) et de volume peut expliquer tous les modes de dégradation décrits plus haut. Les mesures expérimentales ont également montré que le stress en canal ouvert a causé une chute de la transconductance seulement pour de fortes valeurs de la tension VGS, alors que le stress au pincement a provoqué une chute de transconductance uniforme pour toutes les valeurs de VGS. Ce comportement peut être reproduit par la simulation physique pourvu que, dans le cas de stress a canal ouvert, on considère que les piéges s’accumulent au long d’une vaste région qui s’étend latéralement du bord de la grille vers le contact de drain, tandis que, dans le cas du stress au pincement, on considère que la génération des pièges ait lieu dans une portion plus petite de la zone d’accès à proximité de la grille et qu’elle soit accompagnée par une grande dégradation des paramètres de transport du canal. Enfin on propose que les électrons chauds et l’augmentation de la contrainte par le champ électrique soient à l’origine des dégradations observées après les stresses a canal ouvert et au pincement respectivement. • Les piéges dans les HEMT sur GaN ont été caractérisés en utilisant les techniques de DLTS et leur comportement associé de charge/décharge est interprété à l’aide des simulations physiques. Sous certaines conditions de polarisation, les piéges du buffer peuvent produire de faux signaux de piéges de surface, c'est-à-dire, le même type de signaux I-DLTS et ICTS attribués généralement aux piéges de surface. Clarifier cet aspect est très important à la fois pour les tests de fiabilité et pour l’optimisation des dispositifs, car il peut provoquer une identification erronée du mécanisme de dégradation, et par conséquent induire une mauvaise correction des procédés technologiques. • Les mécanismes physiques qui provoquent l’effondrement du courant RF dans les HEMT sur GaN sont analysés par le biais de mesures expérimentales et de simulations physiques. Ce travail propose les conditions suivantes : i) les piéges du buffer aussi bien que ceux de surface peuvent contribuer à l’effondrement du courant RF à travers un mécanisme identique qui impliquerait la capture et l’émission des électrons provenant de la grille; ii) la passivation de la surface diminue considérablement l’effondrement du courant RF par la réduction du champ électrique en surface et la diminution qui en découle de l’injection d’ électrons de la grille vers les pièges ; iii) pour des densités de piéges de surface inférieures à 9 × 1012 cm-2 , des barrières de potentiel superficiels dans l’ordre de 1-2 eV peuvent coexister avec des piéges de surface ayant des énergies plus faibles et qui causent l’effondrement du courant RF caractérisé par des constantes de temps relativement courtes. • Les effets de l’effondrement du courant dans les HEMT sur GaN sont étudiés en utilisant les résultats de mesures expérimentales et de simulations physiques. D’après les mesures pulsées, les dispositifs employés montrent un gate-lag considérable et un drain-lag négligeable qui peuvent être attribués à la présence de piéges de surface et de buffer respectivement. / This thesis reports the results of an extensive analysis of the physical mechanisms that limit the performance and reliability of gallium nitride (GaN) based High Electron Mobility Transistors (HEMT). In particular: • High electric field degradation phenomena are investigated in GaN-capped AlGaN/GaN HEMTs by comparing experimental data with numerical device simulations. Under power- and OFF-state conditions, 150-h DC stresses were carried out. Degradation effects characterizing both stress experiments were as follows: a drop in the dc drain current, the amplification of gate-lag effects, and a decrease in the reverse gate leakage current. Numerical simulations indicate that the simultaneous generation of surface (and/or barrier) and buffer traps can account for all of the aforementioned degradation modes. Experiments also showed that the power-state stress induced a drop in the transconductance at high gate–source voltages only, whereas the OFF-state stress led to a uniform transconductance drop over the entire gate-source-voltage range. This behavior can be reproduced by simulations provided that, under the power-state stress, traps are assumed to accumulate over a wide region extending laterally from the gate edge toward the drain contact, whereas, under the OFF-state stress, trap generation is supposed to take place in a narrower portion of the drain-access region close to the gate edge and to be accompanied by a significant degradation of the channel transport parameters. Channel hot electrons and electric-field-induced strain-enhancement are finally suggested to play major roles in power-state and off-state degradation, respectively. • Traps are characterized in AlGaN-GaN HEMTs by means of DLTS techniques and the associated charge/discharge behavior is interpreted with the aid of numerical device simulations. Under specific bias conditions, buffer traps can produce ‘‘false’’ surface-trap signals, i.e. the same type of current-mode DLTS (I DLTS) or gate-lag signals that are generally attributed to surface traps. Clarifying this aspect is important for both reliability testing and device optimization, as it can lead to erroneous identification of the degradation mechanism, thus resulting in wrong correction actions on the technological process. • The physical mechanisms underlying RF current collapse effects in AlGaN-GaN high electron mobility transistors are studied by means of measurements and numerical device simulations. This work suggests the following conclusions: i) both surface and buffer traps can contribute to RF current collapse through a similar physical mechanism involving capture and emission of electrons tunneling from the gate; ii) surface passivation strongly mitigates RF current collapse by reducing the surface electric field and inhibiting electron injection into traps; iii) for surface-trap densities lower than 9 × 1012 cm-2, surface-potential barriers in the 1–2 eV range can coexist with surface traps having much a shallower energy and, therefore, inducing RF current-collapse effects characterized by relatively short time constants. • Current collapse effects are investigated in AlGaN/GaN HEMTs by means of measurements and numerical device simulations. According to pulsed measurements, the adopted devices exhibit a significant gate-lag and a negligible drain-lag ascribed to the presence of surface and buffer traps, respectively. Furthermore, illumination of the devices with two specific wavelengths can result in either a recovering of current collapse or a decrease in the gate current. On the other hand, numerical device simulations suggest that the kink effect can be explained by electron trapping into barrier traps and the subsequent electron emission after a critical electric-field value is reached. / Questa tesi riporta i risultati ottenuti da un’ampia analisi dei meccanismi fisici che limitano le prestazioni e l’affidabilità dei transistor ad alta mobilità di elettroni (HEMT) al nitruro di gallio (GaN). In particolare: • I fenomeni di degradazione ad alto campo elettrico nei GaN/AlGaN/GaN HEMT sono analizzati confrontando i dati sperimentali con i risultati delle simulazioni numeriche. Sono stati effettuati stress DC di 150 ore in condizioni di canale aperto e chiuso. Gli effetti di degradazione che hanno caratterizzato entrambi i tipi di stress sono i seguenti: una caduta nella corrente DC di drain, un’amplificazione degli effetti di gate lag, e una diminuzione della corrente inversa di gate. Le simulazioni numeriche indicano che la generazione simultanea di trappole in superficie (e/o barriera) e buffer può spiegare tutti i suddetti modi di degradazione. Le misure sperimentali hanno mostrato inoltre che lo stress a canale aperto ha causato una caduta della tranconduttanza solo ad alte tensioni VGS, mentre lo stress a canale chiuso ha provocato una caduta della transconduttanza uniforme a tutte le tensioni VGS. Questo comportamento può essere riprodotto con le simulazioni se, nel caso di stress a canale aperto, si assume che le trappole si accumulano lungo un’ampia regione che si estende lateralmente dal bordo di gate verso il contatto di drain, mentre, nel caso di stress a canale chiuso, si suppone che la generazione delle trappole abbia luogo in una più stretta porzione della zona di accesso vicino al bordo di gate e che sia accompagnata da una degradazione significativa dei parametri di trasporto del canale. Infine si propone che gli elettroni caldi del canale e l’aumento di strain indotto dal campo elettrico siano alla base delle degradazioni osservate dopo gli stress a canale aperto e chiuso rispettivamente. • Le trappole in AlGaN-GaN HEMTs sono caratterizzate usando le tecniche di DLTS e il relativo comportamento di carica/scarica é interpretato con l’aiuto delle simulazioni numeriche. Sotto particolari condizioni di polarizzazione, le trappole di buffer possono produrre falsi segnali da trappole di superficie, ossia lo stesso tipo di segnali I-DLTS e forma d’onda di gate lag attribuiti generalmente alle trappole di superficie. Chiarire questo aspetto è molto importante sia per le prove di affidabilità che per l’ottimizzazione dei dispositivi, in quanto può provocare una errata identificazione del meccanismo di degradazione, portando ad azioni correttive sbagliate nell’ottimizzazione del processo tecnologico. • I meccanismi fisici che originano il collasso di corrente RF negli HEMT AlGaN-GaN sono analizzati usando misure sperimentali e simulazioni numeriche. Questo lavoro suggerisce le seguenti condizioni: i) sia le trappole di superficie che quelle di buffer possono contribuire al collasso di corrente RF tramite un simile meccanismo fisico che coinvolge la cattura e l’emissione di elettroni provenienti dal gate; ii) la passivazione della superficie diminuisce fortemente il collasso della corrente RF tramite la riduzione del campo elettrico in superficie e la conseguente diminuzione dell’iniezione di elettroni dal gate alle trappole; iii) per densità di trappole di superficie minori di 9 × 1012 cm-2 , barriere di potenziale superficiale di 1-2 eV possono coesistere con trappole di superficie aventi energie relativamente basse e che provocano effetti di collasso di corrente RF caratterizzati da costanti di tempo relativamente corte. • Gli effetti di collasso di corrente negli HEMT AlGaN-GaN sono studiati usando i risultati delle misure sperimentali e delle simulazioni numeriche. Basandosi sulle misure delle caratteristiche d’uscita impulsate, i dispositivi utilizzati mostrano un evidente gate-lag e un trascurabile drain-lag, attribuiti alla presenza di trappole di superficie e buffer rispettivamente.