The Ballistic Nanotransistor: A Simulation Study

Future MOS transistors may operate near their ballistic limits [1], so it is important to understand ballistic device physics and the prospects for achieving quasi-ballistic operation. In this paper, we explore the device design and physics issues of MOSFETs at the scaling limits using semiclassical and full quantum simulations. The device we presume is a double-gate (DG) MOSFET with symmetrical (n + /n +), asymmetrical (n + /p +) or symmetric midgap gates (depending on the body thickness) to achieve suitable V T [2]. Simulations treat either classical or quantum transport, either ballistic or with scattering, and use methods that range from Matlab scripts which run in a few seconds to large-scale codes that require a parallel supercomputer. This work establishes: 1) the physics of charge control in 10 nm scale MOSFETs, 2) the maximum source velocity achievable under ballistic conditions, 3) the influence of multi-subband transport, 4) the separate roles of quantum effects in the normal and longitudinal directions, and 5) the possibility of quasi-ballistic operation given the expected values of µ eff. We then use these quantum-scale simulations to critically examine scaling limits for thin and ultra-thin body devices considering issues such as parasitic source/drain resistance, gate overlap with the source/drain, DIBL, V T roll-off, and subthreshold slope. The paper provides a clear description of the physics of MOSFETs at the scaling limits, of the device design issues that arise for such devices, as well as a description of a suite of simulation tools that can be used to explore nanotransistors more generally. Two classes of simulation tools have been developed for this work. The first (2D/1D) couples a 2D Poisson-solver to a 1D transport model to accurately treat MOS electrostatics and efficiently treat transport. The 1D transport model is a good approximation for DG MOSFETs (exact in the 1-subband limit). We have also implemented both semiclassical and quantum ballistic transport models, as well as conventional drift-diffusion and energy-transport models. A simple treatment of scattering has also been implemented. This work extends recent pioneering work [3] by: 1) treating quantum transport along the channel with a Green's function approach that eliminates the WKB approximation and provides a way to include scattering [4], 2) treating quantum confinement normal to the channel with self-consistent, Schrödinger-Poisson simulations, 3) including multiple subbands, 4) comparing quantum and classical transport models, and 5) examining the effects of scattering and the possibilities of quasi-ballistic transport. …