A genetic algorithm to optimize the performance of the tunneling field-effect transistor

Abstract

The double-gate tunneling field-effect transistor (DGTFET) is investigated with two channel lengths (50 and 20 nm), along with the effect of the variation of the device design on its overall direct-current (DC) and radiofrequency (RF) performance. The studied design parameters are the drain doping abruptness (or spread) and its shift relative to the gate electrode. Additionally, the gate work function, and the dielectric material and its thickness are investigated. The studied performance parameters are the ON/OFF ratio, the maximum cutoff frequency, the subthreshold swing, and the ambipolar current. The device’s figure of merit (FOM) is expressed as a weighted-sum objective function for optimization by a genetic algorithm (GA). The results are validated against multifactorial experimentation to study the effect of changing each parameter on the FOM. It is shown that the genetic optimization can enhance the performance of the 20-nm-channel device to become comparable to that of a device with a long channel of 50 nm. The GA is run multiple times using parallel processing, MATLAB, and a technology computer-aided design (TCAD) model to validate its efficiency for the optimization of electronic devices when a TCAD model can be built without a great need to define a mathematical model in closed form.