Volume 3, Issue 2 012686 pp. 211-224
Article
Open Access

Formulation of Macroscopic Transport Models for Numerical Simulation of Semiconductor Devices

Edwin C. Kan

Edwin C. Kan

Center for Integrated Systems Stanford University Stanford, CA 94305, USA , stanford.edu

Search for more papers by this author
Zhiping Yu

Zhiping Yu

Center for Integrated Systems Stanford University Stanford, CA 94305, USA , stanford.edu

Search for more papers by this author
Robert W. Dutton

Robert W. Dutton

Center for Integrated Systems Stanford University Stanford, CA 94305, USA , stanford.edu

Search for more papers by this author
Datong Chen

Datong Chen

Linear Product Division National Semiconductor Corporation Santa Clara, CA 95052-8090, USA , national.com

Search for more papers by this author
Umberto Ravaioli

Umberto Ravaioli

Beckman Institute University of Illinois at Urbana-Champaign Urbana, IL 61801, USA , uillinois.edu

Search for more papers by this author
First published: 01 January 1995
Citations: 3

Abstract

According to different assumptions in deriving carrier and energy flux equations, macroscopic semiconductor transport models from the moments of the Boltzmann transport equation (BTE) can be divided into two main categories: the hydrodynamic (HD) model which basically follows Bløtekjer′s approach [1, 2], and the Energy Transport (ET) model which originates from Strattton′s approximation [3, 4]. The formulation, discretization, parametrization and numerical properties of the HD and ET models are carefully examined and compared. The well-known spurious velocity spike of the HD model in simple nin structures can then be understood from its formulation and parametrization of the thermoelectric current components. Recent progress in treating negative differential resistances with the ET model and extending the model to thermoelectric simulation is summarized. Finally, we propose a new model denoted by DUET (Dual ET)which accounts for all thermoelectric effects in most modern devices and demonstrates very good numerical properties. The new advances in applicability and computational efficiency of the ET model, as well as its easy implementation by modifying the conventional drift-diffusion (DD) model, indicate its attractiveness for numerical simulation of advanced semiconductor devices

The full text of this article hosted at iucr.org is unavailable due to technical difficulties.