Search results

An empirical velocity‐field relationship, based on Monte Carlo simulation, is used to modify a drift‐diffusion model for the characterization of short gate GaAs MESFET's. The modified drift‐diffusion model is used to generate both the steady‐state and the small‐signal parameters of submicron GaAs MESFET's. The current, transconductance, and cutoff frequency are compared with two‐dimensional Monte Carlo simulation results on a 0.2 µm gate‐length. The model is also used to predict measured I‐V and s‐parameters of a 0.5 µm gate‐length ion‐implanted GaAs MESFET. The comparison and the analysis made, support the accuracy of the modified drift‐diffusion simulator and makes it computationally efficient for analysis of short‐gate devices.

Gas insulated systems, such as gas insulated lines (GIL), use insulating gas, mostly sulfur hexalfluoride (SF₆), to enable a higher dielectric strength compared to e.g. air. However, under high voltage direct current conditions, charge accumulation and electric field stress may occur, which may lead to partial discharge or system failure. Therefore, numerical simulations are used to design the system and determine the electric field and charge distribution. Although the gas conduction shows a more complex current–voltage characteristic compared to solid insulation, the electric conductivity of the SF₆ gas is set as constant in most works. The purpose of this study is to investigate different approaches to address the conduction in the gas properly for numerical simulations.

In this work, two approaches are investigated to address the conduction in the insulating gas and are compared to each other. One method is an ion-drift-diffusion model, where the conduction in the gas is described by the ion motion in the SF₆ gas. However, this method is computationally expensive. Alternatively, a less complex approach is an electro-thermal model with the application of an electric conductivity model for the SF₆ gas. Measurements show that the electric conductivity in the SF₆ gas has a nonlinear dependency on temperature, electric field and gas pressure. From these measurements, an electric conductivity model was developed. Both methods are compared by simulation results, where different parameters and conditions are considered, to investigate the potential of the electric conductivity model as a computationally less expensive alternative.

The simulation results of both simulation approaches show similar results, proving the electric conductivity for the SF₆ gas as a valid alternative. Using the electro-thermal model approach with the application of the electric conductivity model enables a solution time up to six times faster compared to the ion-drift-diffusion model. The application of the model allows to examine the influence of different parameters such as temperature and gas pressure on the electric field distribution in the GIL, whereas the ion-drift-diffusion model enables to investigate the distribution of homo- and heteropolar charges in the insulation gas.

This work presents numerical simulation models for high voltage direct current GIL, where the conduction in the SF₆ gas is described more precisely compared to a definition of a constant electric conductivity value for the insulation gas. The electric conductivity model for the SF₆ gas allows for consideration of the current–voltage characteristics of the gas, is computationally less expensive compared to an ion-drift diffusion model and needs considerably less solution time.

Develops a finite element analysis and solution strategy for the augmented drift‐diffusion equations in semiconductors device theory using a multilevel scheme. Decouples the drift‐diffusion equations using Gummel iteration with incremental continuation in the applied voltage. Includes augmentation of the carrier mobility to address the issue of non‐local electric field effects (velocity overshoot) within the framework of the drift‐diffusion formulation. Gives comparison results with hydrodynamic and Monte Carlo models and sensitivity studies with respect to the augmentation parameter. Discretizes the equations with a special finite element method using bases of variable polynomial degree. Accomplishes solution of the resulting linear systems with a multilevel method using the basis degree as the grid level. Presents performance results and comparison studies with direct elimination.

A discretization technique is proposed for the multi‐dimensional, steady‐state hydrodynamic model of semiconductor devices, and a derivation of the model's appropriate boundary conditions is given. The model includes the complete balance equations for charge, momentum and energy, coupled with Poisson's equation, thus accounting for both diffusion and convection phenomena. The technique, like the Scharfetter—Gummel scheme for the simpler drift‐diffusion model, provides an efficient method for solving the hydrodynamic equations, allowing for a more detailed investigation of carrier dynamics in semiconductor devices.

This paper is concerned with the analysis of global uniqueness of the solution to the drift—diffusion models, for stationary flow of charges carriers in semiconductor devices. Two uniqueness cases are found. Firstly, small applied voltages with a proof introducing new ‘quasi‐monotony condition’ verified for solutions in W and not necessarily in H. Secondly, large applied voltage to the semiconductor with small 2D domain, and not large doping functions. These uniqueness cases allow the construction of algorithms that yield converging sequences of solutions.

In this paper, we present the results of submicron pseudomorphic AlGaAs/InGaAs/ GaAs HEMT simulations. Our main interest is the study of electronic temperature behavior in the device and improvement of the current‐voltage characteristic curves. Three types of models are being used. The first is the well known drift‐diffusion model. The second is of the hydrodynamic type and the third is a combination of the two preceding models. The numerical treatment is based on the discretization by the Galerkin finite element method for both Poisson and continuity equations with the streamline‐diffusion method being used for the energy equation. A comparison of the different approaches have been realized and a synthesis on the validity of each of these models is being drawn.

In this paper, the authors describe a more efficient and economical method for a splitting scheme for drift‐diffusion models for semiconductors. It enables one to calculate stationary and non‐stationary processes in semiconductor plasma.

The use of energy‐momentum models for carrier transport is considered in the context of coupled device‐circuit simulation. We point out that certain computational methods for uncoupling the computations, successfully employed using drift‐diffusion models, can be slightly modified and made compatible with energy‐momentum models, assuming only a known or computable “momentum relaxation time”. By such means one can readily compare the results of transient simulations using energy‐momentum and drift‐diffusion models.

Electrical potentials in a junction field transistor can be calculated using a simplified model based on a complete depletion assumption. This gives rise to a free boundary problem. We show here how we can approximate this problem with a quasi‐variational inequality technique and the shape optimization method. A detailed analysis of these methods is presented. Using some numerical experiments we compare our results with the solution of the discrete drift‐diffusion system, accomplished with a Gummel‐like algorithm. The numerical results suggest that the methods proposed here work successfully and that the shape optimization technique provides a reasonably free boundary without excessive iterations.

Uses a streamline‐diffusion finite element method, specially designed for semiconductor device models, to simulate silicon MESFET devices in two space dimensions. Considers the full hydrodynamic model, a simplified hydrodynamic model and drift‐diffusion model. The method, which reduces to the well‐known Scharfetter‐Gummel discretization for the conventional drift‐diffusion model in one space dimension, proves to be a robust numerical tool. It performs well also when the solution has layers of rapid variation across junctions which are not aligned with mesh lines. Makes comparisons for the different models. Finds a qualitative discrepancy between the solutions of the hydrodynamic model and the drift diffusion model. Observes a small difference, however, between the full and simplified hydrodynamic models.