Slide 1: Reducing Statistical Variability in Conventional "Bulk" CMOS
Reducing Statistical Variability in Conventional "Bulk" CMOS Asen Asenov University of Glasgow Synopsys Semiwise 1
Slide 2: Statistical MOSFET variability
Statistical MOSFET variability 2
Slide 3: Numerical simulations of statistical MOSFET variability
Numerical simulations of statistical MOSFET variability 1998 3
Slide 4: Dopants closely to the interface determine the statistical MOSFET variability
Dopants closely to the interface determine the statistical MOSFET variability A. Asenov, IEEE TED, 98 4
Slide 5: Both number and positions of dopants matter
Both number and positions of dopants matter VT = 0.55 V VT = 0.78 V C2 5
Slide 6: Variability trends
Variability trends 40 CMOS matching T1010 10/1 10/0.25 0.2/4 210.18 5% 5 20 0.4/10 2/1 0.4/1 S waft 0.5 um NMOS A aioe (mV) § | 0.18 pm NMOS 5 90 nm NMOS 519 af Gorn Intel, IEDM 2014 3 ~ 0 pt a a Sr = inmV 13dnm 90nm 65nm 45nm 32nm 22nm 14nm 0 : Figure 7: Random Varation Trend (oVt) 00 02 04 06 08 1
Slide 7: Reduction of statistical variability in epitaxial delta doped MOSFETS
Reduction of statistical variability in epitaxial delta doped MOSFETS 1999 A. Asenov, IEEE TED, 99 7
Slide 8: Reduction of statistical variability in epitaxial delta doped MOSFETS
Reduction of statistical variability in epitaxial delta doped MOSFETS A. Asenov, IEEE TED, 99 8
Slide 9: Other sources of statistical variability
Other sources of statistical variability Random dopants Metal gate or Polysilicon Line edge roughness Granularity A. Asenov, IEEE TED, 03 9
Slide 10: The 3D 'atomistic' simulations are predictive – 40 nm
The 3D ‘atomistic’ simulations are predictive – 40 nm Collaboration with ST Microelectronics 10
Slide 11: The 3D 'atomistic' simulations are predictive – 40 nm
The 3D ‘atomistic’ simulations are predictive – 40 nm AVT=2.8 Collaboration with ST Microelectronics 11
Slide 12: Variability at the ultimate bulk CMOS scaling (20 nm)
Variability at the ultimate bulk CMOS scaling (20 nm) 12
Slide 13: Epitaxial channel transistors – 20 nm CMOS
Epitaxial channel transistors – 20 nm CMOS Conventional Channel first Channel last 13
Slide 14: Epitaxial channel transistors – 20 nm CMOS
Epitaxial channel transistors – 20 nm CMOS Conventional Channel first Channel last C2 2.22 1.91 1.06 14
Slide 15: Epitaxial channel transistors – CL + back doping
Epitaxial channel transistors – CL + back doping C2 15
Slide 16: Strong interplay between statistical variability and statistical reliability
Strong interplay between statistical variability and statistical reliability
Slide 17: Mobility improvement
Mobility improvement 1 . 1 . L L L ‘ 800 — CL Template a S Ss T l Electron Mobility [om’/Vs| s =) Ng=7.7x10"7 cm? y =) i=) T N4=8.4x10"7 cm | 02 04 06 08 1.0 ph Vos {Vv} Ege [Mv/em]
Slide 18: MST: Mechanism and Benefits
MST®: Mechanism and Benefits • Both NMOS and PMOS Dopant Engineering Mobility Enhancement 1019 350 • Idsat and Gm enhanced Boron concentration (cm-3) Post-STI after 1010C 10s MST Electron Mobility (cm2/V•s) 300 • Gate Leakage reduced Precision Doping Profiles control 1018 250 • 200 • Variability red