Device and Circuit Performance Issues with Deeply Scaled High-K MOS Transistors

In this paper we look at the effect of Fringe-Enhanced-Barrier-lowering (FEBL) for high-K dielectric MOSFETs and the dependence of FEBL on various technological parameters (spacer dielectrics, overlap length, dielectric stack, S/D junction depth and dielectric thickness). We show that FEBL needs to be contained in order to maintain the performance advantage with scaled high-K dielectric MOSFETs. The degradation in high-K dielectric MOSFETs is also identified as due to the additional coupling between the drain-to-source that occurs through the gate insulator, when the gate dielectric constant is significantly higher than the silicon dielectric constant. The technology parameters required to minimize the coupling through the high-K dielectric are identified. It is also shown that gate dielectric stack with a low-K material as bottom layer (very thin $SiO_2$ or oxy-nitride) will be helpful in minimizing FEBL. The circuit performance issues with high-K MOS transistors are also analyzed in this paper. An optimum range of values for the dielectric constant has been identified from the delay and the energy dissipation point of view. The dependence of the optimum K for different technology generations has been discussed. Circuit models for the parasitic capacitances in high-K transistors, by incorporating the fringing effects, have been presented.