Carbon nanotube transistors (CNTFETs) are particularly attractive due to the possibility of near ballistic channel transport, easy application of high-K gate insulators, and novel device physics. It is important, therefore, to understand device physics, optimize transistor designs, assess ultimate performance limits, and to identify appropriate applications. Device simulation can be a powerful tool for such studies. In this presentation, we describe a self-consistent, atomistic scale simulation tool for ballistic carbon nanotube transistors and use the tools to explore CNTFETs device physics and design.
CNTFETs were simulated by solving the Schrödinger equation using the non-equilibrium Green's function (NEGF) formalism self-consistently with the Poisson equation. Ballistic transport was assumed. An atomistic description of the nanotube using a tight binding Hamiltonian with an atomistic (pz orbital) basis was used. Significant computational savings were achieved by using a mode space approach. Open boundary conditions with a phenomenological description of the metal-nanotube contact were used. Both Schottky barrier and conventional MOSFET-like CNTFETs were examined.
The electrostatics of carbon nanotube devices differ significantly from conventional silicon devices due to the1D geometry of the nanotube. For an intrinsic carbon nanotube attached to the bulk contacts, charge transfer doping is significant if the metal-nanotube barrier height is low and the insulator dielectric constant is high. The contact geometry plays an important role. If the contacts are metal wires rather than bulk contacts, the charge density of the nanotube channel is essentially determined by the electrostatics environment rather than the contact properties. The penetration length of the source/drain field can be engineered by the gate oxide thickness and the contact size, which provides methods to suppress the electrostatic short channel effects.
The scaling behavior of CNTFETs was also studied. We restrict our attention to Schottky barrier carbon nanotube FETs whose metal source/drain is attached to an intrinsic carbon nanotube channel. Ambipolar conduction is found to be an important factor that must be carefully considered in device design, especially when the gate oxide is thin. The channel length scaling limit imposed by source-drain tunneling is found to be between 5nm and 10nm, depending on the off-current specification. Using a large diameter tube increases the on-current, but it also increases the leakage current.
Although most carbon nanotube transistors behave as Schottky barrier transistors so far, the ultimate transistor performance is achieved when the transistor at the ballistic MOSFET-like operation limit when 1) the current is limited by barrier inside the channel rather than at the M/S contacts and 2) the channel is ballistic. To explore the possible role of CNTFETs in future electronic systems, it is important to compare the upper limit performance of a ballistic CNTFET to that of a ballistic silicon MOSFET. The results show that at the same transistor intrinsic delay, the on-off ratio of CNTFETs is hundreds of times larger than Si MOSFETs.