To keep downsizing transistors, new materials must be explored since traditional 3D materials begin to experience tunneling and other problematic physical phenomena at small sizes. 2D materials are appealing due to their thinness and bandgap. The relatively weak van der Waals forces between layers in 2D materials allow easy exfoliation and device fabrication but they also result in poor heat transfer to the substrate, which is the main path for heat removal. The impaired thermal coupling is exacerbated in few-layer devices where heat dissipated in the layers further from the substrate encounters additional interlayer thermal resistance before reaching the substrate, which results in self-heating and degradation of mobility. This study explores the electro-thermal properties of five materials (MoS2, MoSe2, WS2, WSe2, and 2D black phosphorous) which have been identified as possible replacements for Si in future sub-5-nm channel-length devices. We have developed a coupled electro-thermal model to calculate device mobility. The carrier wavefunctions and distribution are obtained from solving the coupled Schrodinger and Poisson equations in the cross-plane direction. The screening length is then calculated from the screening wavenumber. We calculate TBC for each layer in the stack into the substrate from a model based on first-principles phonon dispersion. We determine the local temperature in each layer from a ratio of its dissipated energy and its TBC. We simulate various devices with self-heating (Delta T does not equal 0, where Delta T is the temperature rise of the few-layer device) under several parameters and examined the effects on mobility and change in device temperature. The effects are compared to the isothermal case (Delta T = 0). We observe that self-heating has a significant effect on temperature rise, layer-wise drain current, and effective mobility. Black phosphorous performs the best electrothermally and WS2 performs the worst overall. This thesis will inform future thermally aware designs of nanoelectronic devices based on 2D materials.

Pour augmenter les performances des MOSFET, il est indispensable de comprendre les différents phénomènes physiques qui dégradent la mobilité apparente des électrons et trous traversant le canal et qui limitent l’amélioration obtenue par réduction de sa longueur. Pour cela, une étude précise du transport par des simulations Monte-Carlo a été effectuée. Cette méthode de simulation semi-classique permet de résoudre l’équation de transport de Boltzmann en prenant en compte à la fois le régime quasi-balistique, les interactions avec les phonons, les impuretés ionisées, la rugosité de surface, et le confinement quantique, par génération aléatoire des électrons et de leurs interactions, décrites selon les lois de la mécanique quantique.Un modèle simple de mobilité a alors pu être établi et validé par les simulations. Il est basé sur trois paramètres importants : la mobilité à canal long, la résistance d’accès et la résistance balistique. Ce modèle de mobilité s’est avéré compatible avec des résultats expérimentaux, ce qui suggère que la résistance d’accès est déterminante dans la réduction de mobilité apparente.Par ailleurs, la contribution du transport balistique dans la mobilité a été calculée en tenant compte précisément du confinement quantique et des fonctions de distribution des différentes sous-bandes, ce qui a ainsi permis d’améliorer le modèle de mobilité apparente de Shur qui sous-estime (d’environ 50 Ω.µm) la résistance balistique. Cette résistance balistique est inférieure à la résistance d’accès mais elle pourrait avoir une incidence sur les dispositifs ultimes. / To improve the MOSFET performances, it is necessary to understand the physical phenomena contributing to the apparent mobility of electrons and holes crossing the channel, and limiting the improvement obtained by reducing the channel length. Therefore, a precise study of transport using Monte Carlo simulations was performed. This semi-classical simulation method allows for solving the Boltzmann transport equation, taking into account the quasi-ballistic regime, phonon and Coulomb scattering, surface roughness, as well as the quantum confinement, by randomly generating electrons and their scattering events described by the laws of quantum mechanics.A simple mobility model has been established and validated by the simulations. It is based upon three important parameters: the long channel mobility, the access resistance, and ballistic resistance. This mobility model proved compatible with experimental results, suggesting that the access resistance is determining in the apparent mobility reduction.By the way, the ballistic transport contribution in the mobility was calculated by taking into account the quantum confinement accurately and the distribution functions of the different subbands, allowing for an improvement of Shur’s apparent mobility model, which underestimates (of about 50 Ω.µm) the ballistic resistance. The latter is lower than the access resistance but it could have an incidence on the ultimate devices.Keywords: MOSFET, FDSOI, mobility degradation, analytical model, contact resistance, ballistic, multi-subband Monte Carlo, simulation.