A mobiblity model for TCAD simulation of current variation by random discrete dopant

Abstract

In order to investigate the influence of drive current fluctuation by random discrete dopant (RDD) in the source/drain region using drift-diffusion (DD) solver, a new mobility model with mobility doping profile is proposed considering the nonlocal effect of the Coulomb scattering. The similar approach proposed by Sano is used to create a new impurity profile and introduce the charge smoothing parameter (rcs) to match with the experimental values for the mobility vs. doping concentration. Smoothed doping profile with rcs is used only for the doping dependent mobility calculation and the carrier localization by RDD is resolved using density gradient (DG) method. In summary, two input doping profiles is used to TCAD simulation

one for the impurity mobility and the other is the real RDD profile for the Poisson equation. It is interesting to notice that our mobility model for the RDD effect may capture some of the screening related physics even though rcs is not exactly same as the screening length in the Brooks-Herring model. In addition to the Coulomb mobility due to the RDD effect, mobility degradation by the normal field to the gate and parallel field should be considered. In particular, degradation of the mobility due to the normal and parallel fields in the lightly doped (RDD) in the source and drain regions gives additional limitation in the driving current of the DRAM cell transistors. A strategy for the field dependent mobility models is rather empirical. The Lombardi model for normal field dependence and the extended Canali model for high field dependence with fitting parameters in the models were employed. The proposed model has been applied to the DRAM cell transistor of the 20 nm technology generation. The RDD effect in the drain region of the cell transistor alone gives relative standard variation in the driving current of ~3.1%. In this thesis, the simple and efficient doping dependent mobility model is proposed. This model is expected to provide a clue of the variation reduction strategy together with the mobility boosting technique based on the material and device structure.