We report a self-consistent approach for substrate current simulation by incorporating an energy-dependent impact ionization (II) model in PISCES-2ET. Simulation results are presented for submicron conventional and LDD MOSFETs with comparison to experimental data.

An extension of Newton iteration method which provides an effective way to project an initial guess for a subsequent bias conditions — Newton projection Method (NPM) — is described. In particular, we applied the NPM to a system of semiconductor device equations including energy transport. The computational advantages of initial guess projection via NPM are illustrated through examples.

The carrier transport mechanism in GaAs which includes a negative different mobility in the carrier velocityfield curve when implemented in the 2D device simulator, causes grid-dependent numerical instability and poor convergence behavior. It has well been established in the literature [I, 27 that the conventional drift-diffusion (DD) approximation which assumes a local thermal equilibrium among the same type of carriers (electrons or holes) has significant shortcomings in the description of GaAs device operation. In this paper we report an extension of the energy transport (ET) model [3,4] in which carrier energy conservation equations are solved to GaAs device simulation. A technique whereby the empirical form of the field-dependent mobility is transformed to an energy-dependent one used in the ET model is presented. The empirical formula describing the dependence of mobility on electric field implemented in the 2D device simulator PISCES [6] has the form ~ ( N T , E) = p 0 ( N T ) + y ( ’ 0 ) 4 1+(e14 , where p, is the low-field mobility, E is the