Relation Between the Mobility, $\hbox {1}/f$ Noise, and Channel Direction in MOSFETs Fabricated on (100) and (110) Silicon-Oriented Wafers

A study of the impact of the channel direction over the effective mobility and the 1/f noise in MOSFETs fabricated on (100) and (110) silicon-oriented wafers finding its outcome in the fabrication of future nonplanar device structures has been done. We found that, apart from a slight enhancement of the effective mobility maximum for the p-MOSFETs with a channel along the 100 direction, the channel direction had no effect on the noise level and performances of the transistors when they are fabricated on (100) silicon-oriented wafers. This suggests that a slight but effective enhancement of the already existing CMOS technology is possible by fabricating the MOSFETs with a channel along the 100 direction. Regarding the (110) silicon-oriented wafers, the modification of the channel direction resulted in a change of effective mobility, with the enhancement being maximum for a p-channel along the 110 direction and an n-channel along the 100 direction. Whereas this came with a noise level apparently unchanged for the p-MOSFETs, a perceptible change has been noticed for the n-MOSFETs, with the devices with a channel along the 100 direction showing the highest noise level. Finally, for the Si(110) n-MOSFETs, the earlier limitation of the effective mobility coming from the surface roughness scattering led to a negative transconductance; for the first time, it was clearly shown experimentally that two different uncorrelated noise sources generate the 1/f noise, with one becoming predominant over the other. The first noise source, attributed to the fluctuation of the insulator charge inducing fluctuations in both flatband voltage and mobility, initially contributes by itself to the total 1/f noise, then decreases and lets the second noise source, i.e., the fundamental mobility fluctuations, come out. This strongly suggests that, in the MOSFET, both noise sources, the fluctuation of the insulator charge and the fundamental mobility fluctuations subsist together, with one coverin- - g up the other according to the physical characteristics of the device and to the bias conditions.