Comparative analysis of static and switching performance of 1.2 kV commercial SiC transistors for high power density applications

This paper presents a critical comparison of static and switching performance of commercially available 1.2 kV SiC BJTs, MOSFETs and JFETs with 1.2 kV Si IGBTs. The experiments conducted are mainly focussed on investigating the temperature dependence of device performance. As an emerging commercial device, special emphasis is placed on SiC BJTs. The experimental data indicate that the SiC BJTs have relatively smaller conduction, off-state and turn-off switching losses, in comparison to the other devices. Furthermore, SiC BJTs have demonstrated much higher static current gain values in comparison to their silicon counterparts, thereby minimising driver losses. Based on the results, the suitability of SiC devices for high power density applications has been discussed. © 2013 IEEE.

The Random Forest Kernel and other kernels for big data from random partitions

We present Random Partition Kernels, a new class of kernels derived by demonstrating a natural connection between random partitions of objects and kernels between those objects. We show how the construction can be used to create kernels from methods that would not normally be viewed as random partitions, such as Random Forest. To demonstrate the potential of this method, we propose two new kernels, the Random Forest Kernel and the Fast Cluster Kernel, and show that these kernels consistently outperform standard kernels on problems involving real-world datasets. Finally, we show how the form of these kernels lend themselves to a natural approximation that is appropriate for certain big data problems, allowing $O(N)$ inference in methods such as Gaussian Processes, Support Vector Machines and Kernel PCA.

A flame surface density approach to large-eddy simulation of premixed turbulent combustion

The flame surface density approach to the modeling of premixed turbulent combustion is well established in the context of Reynolds-averaged simulations. For the future, it is necessary to consider large-eddy simulation (LES), which is likely to offer major advantages in terms of physical accuracy, particularly for unsteady combustion problems. LES relies on spatial filtering for the removal of unresolved phenomena whose characteristic length scales are smaller than the computational grid scale. Thus, there is a need for soundly based physical modeling at the subgrid scales. The aim of this paper is to explore the usefulness of the flame surface density concept as a basis for LES modeling of premixed turbulent combustion. A transport equation for the filtered flame surface density is presented, and models are proposed for unclosed terms. Comparison with Reynolds-averaged modeling is shown to reveal some interesting similarities and differences. These were exploited together with known physics and statistical results from experiment and from direct numerical stimulation in order to gain insight and refine the modeling. The model has been implemented in a combustion LES code together with standard models for scalar and momentum transport. Computational results were obtained for a simple three-dimensional flame propagation test problem, and the relative importance of contributing terms in the modeled equation for flame surface density was assessed. Straining and curvature are shown to have a major influence at both the resolved and subgrid levels.

Increased carbon nanotube area density after catalyst generation from cobalt disilicide using a cyclic reactive ion etching approach

We used a cyclic reactive ion etching (RIE) process to increase the Co catalyst density on a cobalt disilicide (CoSi2) substrate for carbon nanotube (CNT) growth. Each cycle of catalyst formation consists of a room temperature RIE step and an annealing step at 450 °C. The RIE step transfers the top-surface of CoSi2 into cobalt fluoride; while the annealing reduces the fluoride into metallic Co nanoparticles. We have optimized this cyclic RIE process and determined that the catalyst density can be doubled in three cycles, resulting in a final CNT shell density of 6.6 × 10 11 walls·cm-2. This work demonstrates a very effective approach to increase the CNT density grown directly on silicides. © 2014 AIP Publishing LLC.

Fourier spectrum based extraction of an equivalent trap state density in indium gallium zinc oxide transistors

Segregating the dynamics of gate bias induced threshold voltage shift, and in particular, charge trapping in thin film transistors (TFTs) based on time constants provides insight into the different mechanisms underlying TFTs instability. In this Letter we develop a representation of the time constants and model the magnitude of charge trapped in the form of an equivalent density of created trap states. This representation is extracted from the Fourier spectrum of the dynamics of charge trapping. Using amorphous In-Ga-Zn-O TFTs as an example, the charge trapping was modeled within an energy range of ΔEt 0.3 eV and with a density of state distribution as Dt(Et-j)=Dt0exp(-ΔEt/ kT)with Dt0 = 5.02 × 1011 cm-2 eV-1. Such a model is useful for developing simulation tools for circuit design. © 2014 AIP Publishing LLC.