Modeling of temperature effects on nano-CMOS devices with the predictive technologies

Abstract

Predictive technology model (PTM) enables early device and circuit characterization by providing the research community the expected profiles of devices in the future technologies. The parameters that need particularly accurate estimation are the speed and the power consumption. Fluctuations in the die temperature affect the device characteristics thereby altering the performance of integrated circuits. Furthermore, increase in the doping concentration and the enhanced electric fields in scaled devices tend to affect the rate of change of device parameter variations when the temperature fluctuates. For precise characterization of the MOSFETs at various temperatures, the parameters that model the temperature effects have to be accurately extracted. With the PTMs, however, the model parameters that characterize the temperature effects are typically transferred from an older technology generation without any change. This inappropriate transfer of parameters across different generations of predictive CMOS technologies may lead to gross errors during the characterization of CMOS circuits at various die temperatures. In this paper, the principle factors that determine the device characteristics when the temperature fluctuates are presented. The impact of technology scaling on the temperature fluctuation induced variation of the device parameters is illustrated. The parameters that model the temperature effects of the predictive technologies are compared. The study indicates that the device and circuit characterization with the PTMs may not be reliable at temperatures other than a single nominal temperature where the critical model parameters are extracted.