DFTB simulation of carbon nanotube field-effect transistors

Abstract

Carbon nanotubes (cnts) are considered as alternatives to traditional materials in nanoelectronics because of their special electronic properties. The use of cnts in nanoelectronics in devices such as field-effect transistors (fets) has opened paths to numerous applications. The understanding and predictions of factors affecting the feasibility and performance of such devices, especially in increasingly miniaturized scales, can be aided by simulations where the relevant factors can be easily modified. Simulations are used to study several experiment-inspired designs of car- bon nanotube field-effect transistors (cnt-fets) where semiconducting cnts are used as the channels by looking into their transfer characteristics obtained from steady-state calculations, using the method of density-functional tight- binding with non-equilibrium Green’s function (dftb-negf) for open sys- tems, on model devices where several material and structural parameters are varied. The different models include the use of different lead materials, including metals and graphenes, another carbon structure. The structural parameters varied include the diameters of semiconducting zigzag cnts (of about 1nm) as well as channel lengths which are in the range of nanometre scale, from about 10nm down to about 5nm. In contrast to the cnts and the lead electrodes which are simulated explicitly with their atomic structures, the gates in the model devices are simulated implicitly, allowing different gate geometries to be explored: cases where the gate is applied on one side of the cnt channel (top gate) or all around the cnt channel (gate all around or gaa) and cases with different gate lengths that may be shorter or longer than the channel lengths. The ambipolarity of cnts, as a result of their almost symmetric valence and conduction bands, the appearance of discrete states in cnts, partially as a result of the short gate lengths, and their im- plications are also discussed with reference to the simulated results of the cnt devices as well as model systems. It is noted that the cnt devices can perform reasonably well at short channel lengths while at the same time ne- cessitating gates of considerable lengths or leads with appropriate energies, implying that while it is possible to compactly put multiple devices in series sharing the same cnts, such designs may require the same gates controlling several consecutive devices or special treatments for the leads. The results calculated and the trends observed from these simulations point to considerations concerning the performance of cnt-fets, the design of integrated systems containing such devices, and how they may work as possible alternatives to traditional silicon-based nanoelectronics.