Ab initio simulations of novel 2D materials and interfaces for advanced ICT applications

To continue the development and miniaturization of transistors, new materials and device concepts are required to be developed. This thesis provides a study of two new classes of materials, superlattices and two-dimensional (2D) materials, to assess their potential as a replacement for silicon in the transistor channel. To perform this analysis, modern modeling techniques are used to provide an atomistic description (molecular dynamics, first-principles simulations and quantum transport calculations) of these systems. New device concepts to replace the Metal-Oxide-Semiconductor Field-Effect Transistors (MOSFETs) are also investigated, exploiting the benefits of these materials.

The scaling of the physical dimensions of the transistors is reaching a state where a degradation of the performances and an increase in power consumption are encountered due to effects related to quantum mechanics. Silicon, the current building block of transistors, is approaching its limits in terms of scaling capabilities. As the channel length approaches 10 nm and its body becomes very thin, source-to-drain tunneling currents prevent the device from being
properly turned off, while a quantum quantization effects occur and result in degraded carrier mobilities. In addition to this phenomenon, the thinning of the gate oxide leads to an increased gate leakage current. 

Therefore, to keep on shrinking the transistors and to reduce their power consumption, new materials with high carrier mobilities have emerged as potential replacements for silicon. This work concentrates on the one hand, on the investigation of superlattices, quasi-two-dimensional materials made of an alternation of a semiconductor layer (silicon or germanium) and of an interspersing species (oxygen, nitrogen or carbon). This approach is a promising alternative that allows tailoring the electronic properties of the transistor channel material. On the other hand, I focus on the fundamental properties of two-dimensional materials. Recently, this class of materials has emerged as potential a contender for future transistor designs, thanks to their self-passivated atomic configuration, high carrier mobilities and excellent electrostatic control that atomically thin materials offer. The physics of these materials is not yet well understood, especially in terms of the electronic properties and how the interactions with their surroundings impact on them.

In parallel to the search for new materials, I also studied the applicability  of 2D materials to allow the development of an alternative device concept, the Tunnel Field-Effect Transistor (TFET). This device offers the potential to overcome the intrinsic limits of the MOSFET, whose power consumption can no longer be sufficiently reduced. The TFET is a low-power device that relies on a filtering of the energy distribution of the injected carriers, leading to a very steep switching from the OFF- to the ON-state. As a consequence, the supply voltage can be lowered, which contributes to the reduction of the heat dissipation.

The main objective of this thesis consists in providing a fundamental understanding of the properties of superlattices and 2D materials to evaluate their potential as channel materials for field-effect transistors (both MOSFETs and TFETs). It aims at formulating guidelines for the design and the selection of materials for future electronic devices. Modern modeling techniques are used to evaluate at an atomistic level their structural, electronic and transport properties. 

The results show that by selecting a proper atomic configuration, superlattices can improve the transport properties in the direction of the drive current (lower effective masses), while degrading them in the perpendicular direction. 2D materials are also shown to be promising alternatives for  MOSFETs, provided that a cautious choice of the material is performed. The thickness (or number of 2D layers) has to be carefully controlled to avoid a dielectric screening of the gate bias. Stacked into heterostructures, they also open the door towards the development of an alternative method of doping based on electrostatics, that allows to achieve high doping concentrations (more than 1e13 e/cm2). Incorporated in TFET channels, they enable the development of for low-power applications. However, the weak interaction forces acting between the layers lead to a variability of the performances in the device behavior.