Charge transport modelling in silicon ultra-scaled devices with native oxide (SINOXI)

点击翻译按钮获取中文摘要

In approaching the ultimate silicon technology scaling limits, the interface between the active channel and the oxide affects critically the charge transport properties: it determines the wave function confinement, thus the electronic and dielectric properties of the active channel; it introduc-es elastic scattering by means of interface roughness and dielectric disorder and it contributes to the screening of remote Coulomb scattering. Nevertheless a proper inclusion of native oxide in device modeling is an open challenge. The Empirical Tight Binding method, a very popular choice for ultra-scaled atomistic device modeling, cannot easily account for the chemical disorder at the interface and for size effects in dielectric properties of the channel. On the other hand, fully ab-initio methods are computationally too expensive and therefore their application in full device modeling is so far limited to proofs of concept. In all cases, the transport formalism is developed in the framework of Non Equilibrium Green Function (NEGF) method. The aim of this proposal is to advance the quantum atomistic modeling of ultra-scaled silicon based field effect transistors towards a quantitatively accurate prediction of current-voltage (I-V) characteristics by including explicitly the gate oxide and electron-phonon interactions. The goal of the project is dual: technological and methodological. Technological, as the inclusion of the oxide interface and the elucidation of the role of different scattering mechanism will be of utmost im-portance in the design and engineering of such systems. Methodological, as in order to achieve this goal we will further advance the NEGF simulation scheme coupled with a Density Functional based Tight Binding (DFTB) Hamiltonian; in doing so we will try overcome some of the limitations of Empirical Tight Binding methods maintaining at the same time a reasonable computational cost. In fact, we recently demonstrated that the NEGF-DFTB method is computationally feasible and allows for an accurate description of the silicon-oxide interface. The methodological improvements will be general enough to be combined with ab-initio methods and to be applied to different materi-als and technologies.