Study and simulation of advanced si-based nanodevicesschottky-barrier mosfets and tunnel fets

Resum

The aim of the work presented in the thesis is to deepen the simulation study of devices based of new injection mechanisms, which are currently regarded as potentially interesting to replace conventional MOSFETs and overcome their fundamental subthreshold swing limitation of 60mV/dec. The physical impossibility of breaking this limit fosters most of the ongoing research precisely in the direction of exploring novel devices such as those considered in this work: the Tunneling Field--Effect Transistors (TFETs) and the Schottky Barrier MOSFETs (SB--MOSFETs). For each one of the considered devices, and assuming the existing State--of--the--Art as starting point, we have structured and organized the performed work according to the following roadmap: - Exposition of the physical phenomena and mechanisms to be studied for a precise understanding and description of the devices. - Identification of the existing limitations, incompatibilities and problems that arise during simulation processes. In our case, using Silvaco ATLAS. - Development of simple simulation algorithms that allow to overcome the encountered difficulties and keep using this widely employed commercial simulator. - Presentation of simulation results obtained from the application of these proposed simulating approaches. The scope and orientation of the work in the case of SB--MOSFETs was set by the fact that before performing it, we had experimental results that appropriately aimed our efforts at fitting them. The main problem that we found studying these devices was that barrier lowering processes were not completely implemented in ATLAS when applied to carrier injection mechanisms involving tunneling (field emission and thermionic field emission). Considering the relevant impact that small variations in barrier heights may have on the total current, and taking into account that depending on the bias conditions the relative importance of the different injection mechanisms changes, it becomes essential to suitably include those barrier lowering processes in our simulations. For that purpose, in this work we developed an iterative procedure inside ATLAS to account for barrier lowering (which also applies for tunneling processes), making it vertically dependent on the depth inside the channel. Very accurate fits between experimental results and simulations have been obtained especially for those regions where tunneling processes proved to be dominant. In addition, some short channel effects like the observed current reduction when decreasing the channel length due to the overlap of the potential profiles of the Schottky barriers are also satisfactorily reproduced. In the case of TFETs, as most of the existing research on them still involves semiclassical approaches and considering their progressive reduction in size, we wanted to take a step forward by somehow performing a more complete treatment which needed to include the effect of quantum confinement. The necessity of such an approach was obvious regarding that whenever the presence of confinement is significant, the existence of a discrete sprectum of energy levels replacing the formerly continuous conduction and valence bands should be greatly affecting the so--called band--to--band tunneling injection of carriers. In that context, the work developed in this thesis lies between those semiclassical models not accounting for the effects of confinement, and the more recent approaches involving rigorous quantum mechanical treatments. The inclusion of confinement made us realize that the numerical solvers employed by ATLAS when using the non--local band--to--band tunneling model (to inject carriers), and the self consistent Schrödinger--Poisson model (to account for subband quantization) were not compatible. At that stage, we decided to exploit the capabilities of the simulator by designing an iterative approach that reasonably allowed to account for confinement in a way that offers great possibilities for researchers that may be potentially interested in the study of these devices. Thanks to the development of this approach we have been able to analyze the impact that confinement indeed has over the underlying physics in TFETs, and how it modifies their total current levels or affects the global trends of electrical parameters of utmost importance for their characterization (threshold voltages and subthreshold swings).