Stress engineering is widely used in the microelectronics industry to improve the oncurrent (I on) performance of the metal-oxide-semiconductor (MOS) transistors through the strain-induced mobility enhancement. However, there are still debates regarding the relevance of the low-field mobility in the saturation drain current of the nanoscale MOS transistors. Based on velocity saturation model, the high-field velocity is independent of the low-field mobility. In the other words, velocity saturation model predicts that mobility enhancement techniques will not improve I on of the nanoscale MOS transistors. Ballistic transport model considers an ideal situation where the channel carriers do not experience any scattering when they transit from the source to the drain. Since mobility is a concept that involves channel scattering, ballistic transport regards mobility as irrelevant in the nanoscale MOS transistors. In quasi-ballistic transport model, channel carriers will undergo a number of channel scatterings before reaching the drain. Hence, quasi-ballistic transport model is able to account for the strain-induced I on improvement in nanoscale MOS transistors. However, the saturation drain current equation of a transistor in the quasi-ballistic model comprises parameters that are not properly defined. Furthermore, some researchers managed to use velocity saturation model to fit the saturation current of the nanoscale MOS transistor. By improvising Lundstrom's 1997 theory on the quasiballistic transport and unifying the merits of existing transport models, we arrive at a simplified saturation drain current equation for nanoscale MOS transistors. Most research in stress engineering focus on the strain-induced I on improvement, but disregard the effects of the mechanical stress on the off-current (I off). Using externally applied mechanical stress, we can isolate the effects of process variations from the straininduced effects from the strain-induced effects on I off. We studied the physics behind the strain-induced increase in the subthreshold I off and found that the strain-induced change in the quantum mechanical confinement decreases the subthreshold swing (S ts) of NMOS transistor but increases S ts of p-channel MOS (PMOS) transistor. It is well-known that uniaxial tensile stress leads to electron mobility enhancement and a reduction in the threshold voltage of NMOS transistor. Since our experimental results show that uniaxial tensile stress can increase the subthreshold I off of NMOS transistor, the effects of straininduced reduction in threshold voltage and the strain-induced mobility enhancement will dominate over the strain-induced improvement in the subthreshold swing. To extend the study to the process-induced stress, we need to reduce the effects of process variation on I off. In order to reduce the effects of wafer-to-wafer variation on I off , we intentionally use iii consecutive wafers with different amount of process-induced stress. In addition, we also identify NMOS transistors whose I off is less sensiti...
