Impact of STI Effect on Flicker Noise in 0.13-$\mu \hbox{m}$ RF nMOSFETs

Chih-Yuan Chan, Yu‐Syuan Lin, Yen-Chun Huang, Shawn S. H. Hsu, Ying-Zong Juang

IEEE Transactions on Electron Devices · 2007 · 17 citations · 19 references

Concepts

Abstract

This paper reports on the impact of shallow-trench isolation (STI) on flicker noise characteristics in 0.13-mum RF nMOSFETs. The drain noise current spectral density was measured in both triode and saturation regions for a more complete study. The devices with a relatively small finger width and a large finger number (W=1 mum/N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">finger</sub> =40 and W=5 mum/N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">finger</sub> =8) presented more pronounced generation-recombination (G-R) noise characteristics compared to those with W=10 mum/N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">finger</sub> =4. In addition, a wide noise level variation of more than one order of magnitude was associated with the more obvious G-R noise components. The observed trends can be explained by the nonuniform stress effect of STI and also the associated traps at the edge of the gate finger between STI and the active region. To further study the noise mechanism, the single-linger devices with different STI-to-gate distances [SA(SB)=0.6,1.2, and 10 mum] were investigated. The measured results provided a direct evidence of STI effect on flicker noise characteristics. The activation energy of the traps was extracted at various temperatures in a range from E <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</sub> -0.397 to E <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</sub> -0.54 eV. Moreover, the calculated standard deviation sigma <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dB</sub> showed a strong dependence of noise variation on device geometry (sigma <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dB</sub> =2.95 dB for W=1 mum/N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">finger</sub> =40 and sigma <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dB</sub> =1.54 dB for W=10 mum/N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">finger</sub> =4). The analysis suggests that the carrier number fluctuation model with the correlated mobility scattering is more suitable for the noise characteristics in these devices.

References

19