83,712 research outputs found
AlN/GaN-based MOS-HEMT technology: processing and device results
No full textProcess development of AlN/GaN MOS-HEMTs is presented, along with issues and problems concerning the fabrication processes. The developed technology uses thermally grown Al<sub>2</sub>O<sub>3</sub> as a gate dielectric and surface passivation for devices. Significant improvement in device performance was observed using the following techniques: (1) Ohmic contact optimisation using Al wet etch prior to Ohmic metal deposition and (2) mesa sidewall passivation. DC and RF performance of the fabricated devices will be presented and discussed in this paper
LCT-GAN - Improving the efficiency of tabular data synthesis via latent embeddings
No full textIn the past decade data-driven approaches have been at the core of many business and research models. In critical domains such as healthcare and banking, data privacy issues are very stringent. Synthetic tabular data is an emerging solution to privacy guarantee concerns. Generative Adversarial Networks (GANs) are one of the emerging solutions for synthesizing data. However in order to capture all relevant relationships between columns, tabular data needs to be numerically encoded. As columns might be of different types, this is a challenging task as proven by recent approaches. Throughout this paper, we focus on the dimensionality explosion problem, which leads to high-dimensional datasets alongside computational overhead and increase in training time. We introduce a novel synthesis pipeline - LCT-GAN - an improvement to the current state-of-the-art in tabular data synthesis CTAB-GAN. Our approach addresses the dimensionality explosion problem by introducing a low-dimensional embedding step via an autoencoder prior to training. It is then combined with a novel conditional GAN architecture, operating in latent space. After thorough evaluation, we observe that our solution achieves more than 30\% improvement in certain statistical metrics in comparsion to CTAB-GAN, accompanied by 5 fold decrease in size and 150 times speedup in training time for a single epoch. We successfully show that it is possible to embed data using autoencoders, and that GANs are able to learn complex relationships in latent space in the context of tabular data.CSE3000 Research ProjectComputer Science and Engineerin
FCT-GAN: Fourier Neural Operator for Global Relation Enhancement in Tabular Data Synthesizing using Generative Adversarial Networks
No full textSince the regularization of data privacy (e.g., GDPR), the effectiveness of data sharing has decreased. A promising technique to circumvent this problem is tabular data synthesis (i.e., the generation of fake tabular data that statistically resembles the original data). However, the state-of-the-art tabular data synthesis model, CTAB-GAN, fails at robustly imitating global data dependencies and underperforms when column orders get permuted. CTAB-GAN internally uses Convolutional Neural Networks (CNN) which limits the model’s performance due to a strictly non-global data perspective during iterative training phases. To address this limitation, this paper proposes FCT-GAN which leverages the Fourier Neural Operator to learn global dependencies in the frequency domain. Specifically, it enhances CTAB-GAN by replacing the CNN of the discriminator with a four-layered two-dimensional Fourier Neural Operator. As a consequence of FCT-GAN’s global nature and cross-column relation robustness, it outperforms CTAB-GAN and additionally offers the column permutation invariant property. The evaluation of FCT-GAN on five datasets shows that the generated data, remarkably resembles the real data and reveals an increase in accuracy, by up to 19% for five machine learning algorithms independent of data column order, compared to CTAB-GAN.CSE3000 Research ProjectComputer Science and Engineerin
Coincident electron channeling and cathodoluminescence studies of threading dislocations in GaN
No full textWe combine two scanning electron microscopy techniques to investigate the influence of dislocations on the light emission from nitride semiconductors. Combining electron channeling contrast imaging and cathodoluminescence imaging enables both the structural and luminescence properties of a sample to be investigated without structural damage to the sample. The electron channeling contrast image is very sensitive to distortions of the crystal lattice, resulting in individual threading dislocations appearing as spots with black–white contrast. Dislocations giving rise to nonradiative recombination are observed as black spots in the cathodoluminescence image. Comparison of the images from exactly the same micron-scale region of a sample demonstrates a one-to-one correlation between the presence of single threading dislocations and resolved dark spots in the cathodoluminescence image. In addition, we have also obtained an atomic force microscopy image from the same region of the sample, which confirms that both pure edge dislocations and those with a screw component (i.e., screw and mixed dislocations) act as nonradiative recombination centers for the Si-doped c-plane GaN thin film investigated
Photoluminescence study of InGaN/GaN quantum dots grown on passivated GaN surface
No full textWe have measured photoluminescence (PL) and time-resolve photoluminescence (TRPL) from InGaN/GaN quantum dots (QDs) grown on passivated GaN surfaces by metalorganic chemical vapor deposition (MOCVD). Strong PL emission was observed from the QDs structure even at room temperature. By comparing the PL and TRPL dependence on temperature, a significant difference between the QD and wetting layer emissions was revealed. The QD emission is characterized by a strong exciton localization effect, which leads to a larger thermal activation energy, a nearly constant radiative lifetime independent of temperature and an unusual temperature behavior of the PL peak energy. (C) 2003 Elsevier B.V. All rights reserved
Study on surface morphology of GaN growth by MOCVD on GaN/Si(111) template
No full textThe surface morphology of GaN grown by MOCVD on GaN/Si template was studied. Rough morphology and deep pinhole defects on some surface areas of the samples were observed and studied. The formation of rough morphology is possibly related to Ga-Si alloy produced due to poor thermal stability of template at high temperature. The deep pinhole defects generated are deep down to the surface of MBE-grown GaN/Si template. The stress originated from the large thermal expansion coefficient difference between GaN and Si may be related to the formation of the pinhole defects. The surface morphology of the GaN can be improved by optimizing the GaN/Si template and decreasing the growth temperature
CTAB-GAN: Effective Tabular Data Synthesizing
No full textWhile data sharing is crucial for knowledge development, privacy concerns and strict regulation (e.g., European General Data Protection Regulation (GDPR)) unfortunately limit its full effectiveness. Synthetic tabular data emerges as an alternative to enable data sharing while fulfilling regulatory and privacy constraints. The state-of-the-art tabular data synthesizers draw methodologies from Generative Adversarial Networks (GAN). In this thesis, we develop CTAB-GAN, a novel conditional table GAN architecture that can effectively model diverse data types with complex distributions. CTAB-GAN is extensively evaluated with the state of the art GANs that generate synthetic tables, in terms of data similarity and analysis utility. The results on five datasets show that the synthetic data of CTAB-GAN remarkably resembles the real data for all three types of variables and results into higher accuracy for five machine learning algorithms, by up to 17%. Additionally, to ensure greater security for training tabular GANs against malicious privacy attacks, differential privacy (DP) is studied and used to train CTAB-GAN with strict privacy guarantees. DP-CTAB-GAN is rigorously evaluated using state-of-the-art DP-tabular GANs in terms of data utility and privacy robustness against membership and attribute inference attacks. Our results on three datasets indicate that strict theoretical differential privacy guarantees come only after severely affecting data utility. However, it is shown empirically that these guarantees help provide a stronger defence against privacy attacks. Overall, it is found that DP-CTABGAN is capable of being robust to privacy attacks while maintaining the highest data utility as compared to prior work, by up to 18% in terms of the average precision score
The influence of AlN/GaN superlattice intermediate layer on the properties of GaN grown on Si(111) substrates
No full textAlN/GaN superlattice buffer is inserted between GaN epitaxial layer and Si substrate before epitaxial growth of GaN layer. High-quality and crack-free GaN epitaxial layers can be obtained by inserting AlN/GaN superlattice buffer layer. The influence of AlN/GaN superlattice buffer layer on the properties of GaN films are investigated in this paper. One of the important roles of the superlattice is to release tensile strain between Si substrate and epilayer. Raman spectra show a substantial decrease of in-plane tensile strain in GaN layers by using AlN/GaN superlattice buffer layer. Moreover, TEM cross-sectional images show that the densities of both screw and edge dislocations are significantly reduced. The GaN films grown on Si with the superlattice buffer also have better surface morphology and optical properties
Recommended from our members
Multi-scale modelling of III-nitrides: from dislocations to the electronic structure
No full textGallium nitride and its alloys are direct band gap semiconductors with a wide variety of applications. Of particular importance are light emitting diodes and laser diodes. Due to the lack of suitable lattice-matched substrates, epitaxial layers contain a high density of defects such as dislocations. To reduce their number and to design a device with desired specifications, multilayered systems with varying composition (and thus material properties) are grown. Theoretical modelling is a useful tool for gaining understanding of various phenomena and materials properties.
The scope of the present work is wide. It ranges from a continuum theory of dislocations treated within the linear elasticity theory, connects the continuum and atomistic level modelling for the case of the critical thickness of thin epitaxial layers, and covers some issues of simulating the electronic structure of III-nitride alloys by means of the first principle methods.
The first part of this work discusses several topics involving dislocation theory. The objectives were: (i) to apply general elasticity approaches known from the literature to the specific case of wurtzite materials, (ii) to extend and summarise theoretical studies of the critical thickness in heteroepitaxy. Subsequently, (iii) to develop an improved geometrical model for threading dislocation density reduction during the growth of thick GaN films.
The second part of this thesis employs first principles techniques (iv) to investigate the electronic structure of binary compounds (GaN, AlN, InN) and correlate these with experimentally available N K-edge electron energy loss near edge structure (ELNES) data, (v) to apply the special quasi-random structures method to ternary III-nitride wurtzite alloys aiming to develop a methodology for modelling wurtzite alloys and to get quantitative agreement with experimental N K-edge ELNES structures, and (vi) to theoretically study strain effects on ELNES spectra
- …
