Modeling ballistic double gate MOSFETs using neural networks approach

This paper presents modeling ballistic double gate MOSFETs by a neural network approach. A complete neural network structure is proposed to model the double gate characteristics. To confirm the accuracy of the proposed network, the drain current characteristics are compared to the nanoMOS device simulator data. The comparison shows excellent agreements with percentage errors lower than 1% over a range of drain and gate voltages as well as channel lengths and silicon thicknesses.