Consideration of Random Dopant Fluctuation Models for Accurate Prediction of Threshold Voltage Variation of Metal-Oxide-Semiconductor Field-Effect Transistors in 45 nm Technology and Beyond

Two well-known dopant models for the Coulomb potential, the atomistic model and the long-range model, have been compared using device parameters in the 45 nm technology node and beyond. We have found that the atomistic model has unacceptable dependences of the average threshold voltage on mesh spacing and substrate dopant concentration, while the long-range model has minimum dependences on these factors. Consequently, the atomistic model severely overestimates the Takeuchi coefficient BVT, which is one of the most important parameters for random threshold voltage variation. We conclude that the long-range model is more suitable for the prediction of random variation in future aggressively scaled metal–oxide–semiconductor field-effect transistors (MOSFETs).