Name:Grid_MC
Description:In modern-day submicron silicon devices, high electric field and high field-gradient conditions are routinely encountered during operation, and the carrier dynamics are far from thermal equilibrium.
Hot electrons can heat the device with important consequences for long-term reliability, or can be injected into the oxide creating an instability in the device performance. Commercial simulation tools based on the Drift-diffusion model cannot capture these off-equilibrium phenomena, and for these reasons the TCAD designer needs accurate simulation tools to predict unwanted phenomena in the device. The Direct Simulation Monte Carlo (DSMC) find a stochastic solution of the Boltzmann transport equation, which replaces the distribution function with a representative set of particles. In this way, a more accurate description of carrier transport even under such highly non-equilibrium conditions is obtained, because the various scattering mechanisms and band structure models are taken into account explicitly. However, in exchange for this advantage, significantly longer computer execution times are required. Moreover the DSMC contains in addition stochastic and systematic errors to be estimated to ascertain the quality of the results. Hence it is not possible to evaluate the efficiency of a DSMC without a detailed analysis of these errors, because the efficiency of a MC algorithm is inversely proportional to the CPU time required.
During the PI2S2 project we have studied the DSMC systematic error with respect to the particle number and time step, as well as the efficiency of the algorithm.
^
Abstract:In modern-day submicron silicon devices, high electric field and high field-gradient conditions are routinely encountered during operation, and the carrier dynamics are far from thermal equilibrium.
Hot electrons can heat the device with important consequences for long-term reliability, or can be injected into the oxide creating an instability in the device performance. Commercial simulation tools based on the Drift-diffusion model cannot capture these off-equilibrium phenomena, and for these reasons the TCAD designer needs accurate simulation tools to predict unwanted phenomena in the device. The Direct Simulation Monte Carlo (DSMC) find a stochastic solution of the Boltzmann transport equation, which replaces the distribution function with a representative set of particles. In this way, a more accurate description of carrier transport even under such highly non-equilibrium conditions is obtained, because the various scattering mechanisms and band structure models are taken into account explicitly. However, in exchange for this advantage, significantly longer computer execution times are required. Moreover the DSMC contains in addition stochastic and systematic errors to be estimated to ascertain the quality of the results. Hence it is not possible to evaluate the efficiency of a DSMC without a detailed analysis of these errors, because the efficiency of a MC algorithm is inversely proportional to the CPU time required.
During the PI2S2 project we have studied the DSMC systematic error with respect to the particle number and time step, as well as the efficiency of the algorithm.
^
Created:2010-05-01
Last updated:2010-05-01
Software
Virtual Organizations
Software Appliances
Container Catalog
Sites/Resource Providers
Applications Database
