The Impact of Hot Electrons and Self-Heating During Hard-Switching in AlGaN/GaN HEMTs (original) (raw)
Related papers
—Distance functions are at the core of important data analysis and processing tools, e.g., PCA, classification, vector median filter, and mathematical morphology. Despite its key role, a distance function is often used without careful consideration of its underlying assumptions and mathematical construction. With the objective of identifying a suitable distance function for hyperspec-tral images so as to maintain the accuracy of hyperspectral image processing results, we compare existing distance functions and define a suitable set of selection criteria. Bearing in mind that the selection of distance functions is highly related to the actual definition of the spectrum, we also classify the existing distance functions based on how they inherently define a spectrum. Theoretical constraints and behavior, as well as numerical tests are proposed for the evaluation of distance functions. With regards to the evaluation criteria, Euclidean distance of cumulative spectrum (ECS) was found to be the most suitable distance function.
Hyperspectral Sensor Characteristics
Fundamentals, Sensor Systems, Spectral Libraries, and Data Mining for Vegetation, 2018
This chapter sets forth the fundamental concepts of hyperspectral (HS) imaging-a powerful tool to collect the precise remote sensing data involved in detection, recognition, and examination of the different objects of scientific, economic, and military character. The development of the relevant equipment is also discussed as well as current and future applications of HS technology. Hyperspectral sensor (HSS) is one of the devices in use in the remote sensing activity, aimed at observing the Earth's surface with ground-based, airborne, or spaceborne imaging gear [1-7]. In outline, remote sensing consists in photo-or electronic recording of the spatial and spectral distributions of electromagnetic radiation, as emitted by any object under monitoring either on the Earth's surface or underground. Present-day technologies enable such a recording throughout the entire spectrum, from the radio waveband to the x-ray and gamma regions. The respective receivers differ widely, for each of them is based on specific physical principles. Basically, three distinct spectral regions are used in remote sensing, namely, visible light with very-near infrared (IR), thermal, and radio. In each of them, an image of any object is unique; it is defined by the object's specifics as to emission, reflection, absorption, and scattering of the electromagnetic waves. In the visible and very-near IR regions, it is the object's capacity to reflect the solar energy in the first place, determined by the chemical composition of its surface.
Applied Physics Letters, 2006
Deep levels were observed using capacitance deep level optical spectroscopy (DLOS) in an AlGaN /GaN heterostructure equivalent to that of a heterojunction field effect transistor. Band gap states were assigned to either the AlGaN or GaN regions by comparing the DLOS spectra in accumulation and pinch-off modes, where the former reflects both AlGaN- and GaN-related defects, and the latter emphasizes defects residing in the GaN. A band gap state at Ec-3.85eV was unambiguously identified with the AlGaN region, and deep levels at Ec-2.64eV and Ec-3.30eV were associated with the GaN layers. Both the AlGaN and GaN layers exhibited additional deep levels with large lattice relaxation. The influence of deep levels on the two-dimensional electron gas sheet charge was estimated using a lighted capacitance-voltage method.
Journal of Applied Physics, 2022
High electron mobility transistors (HEMTs) based on gallium nitride (GaN) with a wide range of application potentials need to be rigorously examined for reliability to take advantage of their intrinsically extraordinary properties. The most vital parameter of the reliability, the hotspot, or Tmax, resides in the two-dimensional electron gas (2DEG) temperature profile inside the device where optical access is often restricted. The device surface temperature can be measured by widespread IR thermography with the limitation of diffraction-based IR transmission losses. However, Tmax on the sub-surface cannot be reached thermographically. Although finite element analysis (FEA)-based thermal simulations can easily reveal the 2DEG temperature profile, accuracy is tightly dependent on the realistic modeling of material/structure parameters. Because these parameters are rather sensitive to fabrication and processing, it is quite difficult to specify them accurately. To overcome these drawbac...
International Journal of Electrical and Computer Engineering (IJECE), 2018
This paper is mainly dedicated to understand the phenomena governing the formation of two-dimensional electron gas (2DEG) confined in the quantum well which hold the role of the channel in the high electron density transistors (HEMT) based on AlGaN / GaN heterojunction. The theory takes into account: the crystal structure, the spontaneous and piezoelectric polarization concept, the formation mechanism of two-dimensional electron gas at the AlGaN / GaN interface, the approximate resolution of the Poisson and Schrödinger equations to determine the density of Two-dimensional electron gas after the analytical formula of the current-voltage characteristic is established. Our study is also concerned with the dependence of the two-dimensional electron gas density on the following technological parameters: Aluminum molare fraction, AlGaN layer thickness and AlGaN layer doping, In order to control the influence of these parameters on the device performance. Finally, the current-voltage characteristic which reflects the variation of the drain-source current as a function of the modulation of the gate voltage has been discussed. Keyword: 2DEG AlGaN/GaN Current-voltage characteristic HEMTS Polarizations 1. INTRODUCTION The area of microelectronics has been marked by an increase in the demand for components which can operate in frequency increasingly higher [1-3]. Emerging needs in this area are not only related to the telecommunications market, but in other areas of application such as embedded electronics at bound for the automotive and aeronautic. The largest concerns in semiconductor integrated circuits are high-speed operation and low power consumption [4]. Since the 90s, a new generation of semiconductor components (HEMT) is under study and evaluation thanks to the advent of wide band gap materials such as GaN [5], GaN based devices are very useful for high frequency high temperature microwave applications such as radar systems [6]. Transistors AlGaN / GaN HEMTs results are very promising for power electronics and high frequency due to their two-dimensional electron gas 2DEG high density and high mobility as well as to their high breakdown field [7]. To develop a reliable model of HEMT, an accurate estimate of the two-dimensional electron gas density at AlGaN / GaN interface is of considerable importance. A number of charge control models for HEMTs have been developed to characterize the 2-DEG concentration. These models are useful as they provide an insight into the physical operation of the device, but they generally require simplifying assumptions for the 2DEG density. Some models assume a linear dependency of 2DEG density at the gate bias, but the values of 2DEG density are underestimated near the threshold [8], [9]. Other models offer a
Remote Sensing, 2020
The critical issue facing hyperspectral image (HSI) classification is the imbalance between dimensionality and the number of available training samples. This study attempted to solve the issue by proposing an integrating method using minimum noise fractions (MNF) and Hilbert–Huang transform (HHT) transformations into artificial neural networks (ANNs) for HSI classification tasks. MNF and HHT function as a feature extractor and image decomposer, respectively, to minimize influences of noises and dimensionality and to maximize training sample efficiency. Experimental results using two benchmark datasets, Indian Pine (IP) and Pavia University (PaviaU) hyperspectral images, are presented. With the intention of optimizing the number of essential neurons and training samples in the ANN, 1 to 1000 neurons and four proportions of training sample were tested, and the associated classification accuracies were evaluated. For the IP dataset, the results showed a remarkable classification accura...