Performance Optimization of Broadwell-Y Shaped Transistor Using Artificial Neural Network and Moth-Flame Optimization Technique

Finfets are the emerging 3d-transistor structures due to strong electrostatic control of active channel by gate from more than one side which was not possible in conventional transistor. finfet structures with rectangular and trapezoidal shape have been excessively analyzed in literature. the main purpose of this work is to present a finfet structure with such a compact fin shape that the gate has high controllability over it; and thus reduced short channel effects in comparison to existing structures. here, finfet with broadwell-y shape, proposed by intel has been designed and its short channel effects were analysed. simulations of the designed finfet have been performed in technology computer aided design (tcad) tool. performance of broadwell-y shaped finfet was compared with the existing rectangular and trapezoidal structures for the same input design parameters and it was noticed that broadwell-y shaped finfet outperformed the last two structures in terms of short channel effects. then the performance of the designed device was optimized using moth flame optimization (mfo) after the network was trained through artificial neural network (ann). results obtained from matlab were in close agreement with those obtained from tcad simulations. output parameters like leakage current (ioff) of 2.407e-12a, on-off current ratio (ion/ioff) of 4.5e06, subthreshold swing (ss) of 65.4mv/dec and drain induced barrier lowering (dibl) of 37.9mv/v were obtained after optimization. short channel effects are improved for 20nm gate length as ss is close to ideal value 60mv/dec and dibl is below 100mv/v which makes this designed structure a good option for applications at nanoscale.