Modeling double gate FinFETs by using artificial neural network

The minimum feature size of the transistors will be decreases in the future years as predicted by the international technology roadmap for semiconductors. Multi-gate FETs such as FinFETs have emerged as the most promising candidates to extend the CMOS scaling into the sub-25nm regime when considering the low scale effects is important for decreasing the scale. Solving and simulating the equations of these devices are so complicated and time consuming. In this paper we use RBF network for simulating the I-V characteristics of common symmetric multi gate FinFETs by using some BSIM-CMG data as a database for training. The results show a good agreement between RBF network and BSIM-CMG. The maximum error between BSIM-CMG and RBF is only 1%. The RBF is used for simulating or predicting I-V curve for different inputs without solving the complicated equations.