Characterization and physical modeling of degradation in MoS2 transistors

Abstract

Over the last decades Moore's Law was mainly maintained by scaling the dimensions of silicon transistors. This scaling led to gate lenghts corresponding to a few dozen layers of silicon atoms, hence it is approaching a hard limit. Therefore, new device concepts have to be evaluated. One novel concept is based on two-dimensional channel materials. This idea was strongly promoted by the two-dimensional material graphene. However, the application of graphene is limited due to its lack of a band gap. Other two dimensional materials were investigated and while there are promising materials which have a band gap it turned out that many of these transistors suffer from severe reliability issues. To the current understanding, these problems do not arise merely out of immature processing conditions but are inherent to the materials and might determine the success or the failure of new transistor concepts. In order to be able to predict viable transistor technologies, the physical modeling of novel devices and their degradation mechanisms is essential. In this work the theoretical framework enabling a physical description of degradation effects was applied to transistors based on two-dimensional layers of Molybdenum Disulfide (MoS2). The material properties of transition metal dichalcogenides, the material class to which MoS2 belongs, have to be known for modeling the transistor's characteristics using drift diffusion equations. Therefore, these properties are discussed in detail. At the current stage of device development, where the drain current is governed by scattering events, quantum mechanical effects can be neglected. As a main source of degradation the interaction of charge carriers with defects in the underlying Silicon Dioxide (SiO2) is identified. This causes a hysteresis in the transfer characteristics of the devices as well as the observed Bias Temperature Instabilities (BTI). The measurements characterizing these effects can be successfully simulated using the four-state nonradiative multiphonon model which describes charge capture and emission events in the oxide. In the framework of this model a new defect band, situated above the defect band governing the degradation effects in conventional transistors, was extracted. The alignment of this defect band to the band edges of MoS2 leads to the conclusion that nMOSFETs, using a single layer of MoS2 as a channel material and SiO2 as a gate dielectric, will probably always suffer from hysteresis and large negative BTI. Nevertheless, this works proves the applicability of established physics-based degradation models to the new device class. Thereby, it paves the way for future studies targeting at identifying reliable two-dimensional materials.