Study for Mechanical Stress Impact in Scaled...
Transcript of Study for Mechanical Stress Impact in Scaled...
Study for Mechanical Stress Impact in Scaled MOSFETs
Nuo Xu, Lynn Wang, Xin Sun, Prof. Andrew Neureuther, Prof. Tsu-Jae King Liu
Electrical Engineering and Computer Sciences Department
University of California at Berkeley
03/10/2010
Introduction
*L.-T. Pang et al., 2006 Symp. VLSI Circuits
• Systematic Transistor Variations
S. E. Thompson, TED, 25, 191 (2004)
Intel 90nm HP Tech.
C.-H. Jan et al, IEDM, (2008)
NMOSFET PMOSFET
Intel 45nm LP Tech.
SiGe
CESL CESL
• Uniaxial Stress Technology is
already in production since…
A. Khakifirooz et al, IEDM, (2007)D. Chanem et al,2005 Symp. VLSI Tech.
• MOSFET Performance is Largely Improved
Outline
• Introduction
• Fast-models for Stress-induced Variations
• Injection Velocity Model for Strained MOSFET
• 3D Stress Design for Multigate FET
Stress-Layout Dependence The Impact of Process-induced Stress Variations on MOSFET:
SPICE/Compact models: Adjust mobility and threshold voltage empirically
by fitting a few layout parameters:
Models are too simplistic, do not capture all layout parameter effects
TCAD simulations: numerical (FEM) analysis to determine full stress
profiles in the transistor
Computationally expensive
Need a fast, accurate method for determining channel stress profile, for
accurate estimation of transistor performance
BSIM Stress Models
NMOS PMOS
TCAD Simulation from Synopsis
Flamant Solution
• Analytical solution to concentrated force problems in an infinite
large planeY
X
rr1 r2
Active
Channel
Point Source
(0,0)
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.0610
2
103
104
Str
es
s X
X (
Pa
)
Distance from channel centre (um)
Lg = 40nm
Tox = 1nm
spacing (to channel centre): 90nm
Initial stress: Tensile 2GPa
stressor thickness: 60nm
stressor area: 2nmx2nm
numerical TCAD
analytical model
Gate Edge
Actual:
Stress contoursσx and σy distributions
Classical:
Simulation Approach
point-source stressor
1. Perform 3-D TCAD
simulations to determine the
channel stress profiles (σXX,
σYY) induced by a point
source in the source/drain
region
2. Fit analytical models to the
simulated stress profiles
3. Integrate point-source
contributions to obtain the full
channel stress profiles
Point-source Stress Function
-0.02 -0.01 0.00 0.01 0.02
500
1000
1500
2000
2500
3000
3500
4000
S
tre
ss
XX
(P
a)
Position along Channel Length (um)
Y=0nm
Y=10nm
Y=50nm
Model
-0.2 -0.1 0.0 0.1 0.2
0
300
600
900
1200
1500
1800
2100
Position along Channel Width (um)
Str
es
s X
X (
Pa
)
X=15nm
X=10nm
X=0nm
Model
𝑆1 =𝑆𝑥𝑥 ,0
𝑟𝛼 sin3 𝜃
𝑆2 = 𝑆𝑥𝑥′
1
𝑟1𝛽𝑒−𝑥𝐿0 +
1
𝑟2𝛽𝑒𝑥𝐿0
F.S. Term:
Correc. Term:
Stress XX
Y
X
0 10 15
50
10
0
Y
X
-18 0 18
50
0
Stress YY
𝑆1 =𝑆𝑦𝑦 ,0
𝑟𝛼 cos3 θ −
𝑆𝑦𝑦′
𝑟𝛽 sin3 𝜃
𝑆2 =
𝑆𝑦𝑦
"
𝑟0 𝑒
𝑥𝐿0 𝑟𝑖𝑔𝑡 𝑠𝑖𝑑𝑒
𝑆𝑦𝑦"
𝑟0 𝑒
−𝑥𝐿0 𝑙𝑒𝑓𝑡 𝑠𝑖𝑑𝑒
F.S. Term:
Correc. Term:
-0.2 -0.1 0.0 0.1 0.2
-200
-100
0
100
200
Position along Channel Width (um)
Str
es
s Y
Y (
Pa
)
X=18.5 nm
X=-18 nm
X=0 nm
Model
-0.02 -0.01 0.00 0.01 0.02
-200
-100
0
100
200
300
400
500
Position along Channel Length (um)
Str
es
s Y
Y (
Pa
)
Y=0nm
Y=50nm
Model
Channel Length (um)
Ch
an
nel W
idth
(um
)
a) b)
Superposition
in the channel
Stress XX Stress YY
Channel Stress Profiles
Impact from STI
-0.10 -0.05 0.00 0.05 0.10
100M
200M
300M
400M
500M
Stress YY
Str
es
s V
alu
es
(P
a)
Position along Channel Width (um)
STIV = 0nm
STIV = 50nm
STIV = 100nm
STIV = 200nm
Stress XX
-0.10 -0.05 0.00 0.05 0.10
50.0M
100.0M
150.0M
200.0M
250.0M
300.0M
350.0M
400.0M
450.0M
500.0M
Str
es
s V
alu
es
(P
a)
Position along Channel Width (um)
0nm
50nm
100nm
150nm Stress YY
Stress XX
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15
360M
390M
420M
450M
480M
510M
Model
STIP=100nm, STIV=100nm
STIP=50nm, STIV=100nm
STIP=100nm, STIV=50nm
Str
es
s X
X (
Pa
)
Distance along Channel Width (um)
-0.02 -0.01 0.00 0.01 0.02300M
400M
500M
600M
700M
800M
900M
1G
Str
es
s X
X (
Pa
)
Distance along Channel Length (um)
Model
STIP=100nm, STIV=100nm
STIP=50nm, STIV=100nm
STIP=100nm, STIV=50nm
Channel Length (um)
Ch
an
nel
Wid
th (
um
)
a) b)
LOD = 0.5um; W = 0.3um; Lg = 40nm
STIP = 0.1um; STIV = 0.1um
CESL 2GPa Tensile
Stress XX Stress YY
Model Verification with TCAD-1
• LOD Impact
-0.02 -0.01 0.00 0.01 0.02
300M
400M
500M
600M
700M
800M
900M
1G
Str
es
s X
X (
Pa
)
Position along Channel Length (um)
LOD = 700nm
LOD = 500nm
LOD = 300nm
Model
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.1570M
80M
90M
100M
110M
120M
130M
Str
es
s Y
Y (
Pa
)
Position along Channel Width (um)
LOD = 300nm
LOD = 500nm
LOD = 700nm
Model
-0.02 -0.01 0.00 0.01 0.0250M
100M
150M
200M
250M
300M
350M
400M
450M
Str
es
s Y
Y (
Pa
)
Position along Channel Length (um)
LOD = 300nm
LOD = 500nm
LOD = 700nm
Model
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15
280M
320M
360M
400M
440M
480M
520M
Str
es
s X
X (
Pa
)
Position along Channel Width (um)
LOD = 700nm
LOD = 500nm
LOD = 300nm
Model
Model Verification with TCAD-2
• Channel Width Impact
-0.02 -0.01 0.00 0.01 0.02
300M
400M
500M
600M
700M
800M
900M
1G
Str
es
s X
X (
Pa
)
Position along Channel Length (um)
W = 300nm
W = 200nm
W = 400nm
Model
-1.0 -0.5 0.0 0.5 1.0
300M
330M
360M
390M
420M
450M
480M
510M
Str
es
s X
X (
Pa
)
Normalized Position along Channnel Width (W/2)
W = 300nm
W = 200nm
W = 400nm
Model
-0.02 -0.01 0.00 0.01 0.0250M
100M
150M
200M
250M
300M
350M
Str
es
s Y
Y (
Pa
)
Position along Channel Length (um)
W = 300nm
W = 200nm
W = 400nm
Model
-1.0 -0.5 0.0 0.5 1.0
60M
75M
90M
105M
120M
135M
Str
es
s Y
Y (
Pa
)
Normalized Position along Channel Width (W/2)
W = 300nm
W = 200nm
W = 400nm
Model
Verification with Silicon Data
Intrinsic Transistor Delay
Mobility %
Injection Velocity %
Threshold Voltage %
Effective Current %
RO Frequency %
13-stage Ring Oscillators Control Logic
0.2 0.3 0.4 0.5 0.6 0.7 0.80
2
4
6
8
10
12
% C
ha
ng
e i
n R
O F
req
ue
nc
y
LOD Length (um)
Conventional Model
Injection Velocity Model
Experimental Data
45nm RO Test Structure
with CESL
Channel Length (um)
Ch
an
nel W
idth
(u
m)
a) b)
x
x
x
x
x
x
LOD
Test Chip information:E. Josse et al, IEDM (2006)
Test RO
Mobility Distribution
Predictive Device Modeling
What limits the ON-state current for modern transistors ?
Drift + Diffusion Quasi-Ballistic
Low-field Mobility(Along whole channel)
High-field Saturation Velocity(Drain side)
Injection Velocity(Source edge)
?
Lg >> Lg ~ (20nm for Si)
Source Drain
CESL
Lg
l
Channel Potential
Compact Modeling Flow
Low-field Mobility
Backscattering Rates
Critical length
Average stress in critical
length region
Band shift
Effective Mass Variation
Charge density
Surface Fermi Level
Subband Energy
Calculate current
Drain bias
Stress
profiles
Gate bias
Self-consistentsolution
Quasi-ballistic current:
“Critical length”:
Backscattering rate:0.0 0.2 0.4 0.6 0.8 1.0 1.2
5
10
15
20
25
30
35
40
Av
era
ge
stre
ss
_x
x* (P
a)
Cri
tic
al
Le
ng
th (
nm
)
Drain bias (V)
Ave. Stress_xx along
whole channel
Vds = 1.2V
Lg = 40nm
240M
250M
260M
270M
280M
Energy-band
diagram:
𝐼 = 𝑞𝑊 2𝑚𝑦
22
𝑖𝑥 ,𝑦 ,𝑧
𝑘𝑇
𝜋
32
(1− 𝑅) ℱ12
𝜂𝑖1 − ℱ12
𝜂𝑖2
𝑙 = 𝐿𝑔 ×
𝑘𝐵𝑇
𝑞𝑉𝑑× 𝑡𝑎𝑛−1
𝑞𝑉𝑑𝑘𝐵𝑇
𝑅 =𝑙
𝑙 + 0
𝜆 =
2𝜇
𝜈𝑇
𝑘𝐵𝑇
𝑞
ℑ02 (𝐸𝑓 )
ℑ1′ (𝐸𝑓 ) × ℑ1
2(𝐸𝑓)
𝐼 =𝑞𝑊 2𝑚𝑦
𝑘𝐵𝑇𝜋
32
2ℏ2 1 − 𝑅 [ℑ1
2
𝐸𝑓1 − ℑ12
𝐸𝑓2 ]
Mean Free Path
Backscattering Rates
Critical Length
Injection Velocity Model
A. Rahman et al. TED 49, 481 (2002).
Model Verification
0.0 0.2 0.4 0.6 0.8 1.0 1.21E-8
1E-7
1E-6
1E-5
1E-4
1E-3
Vds = 0.05 V
Id
(A)
Vgs (V)
Exp. Data in [11]
Model
Vds = 1.0 V
Fitting Parameters:
Gate Control Para. 0.85
Drain Control Para. 0.06
Low-field Mobility 250 cm2/V.s
Series Resistance 180 Ohm.um
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.40
400
800
1200
1600
2000
Exp. Data in [11]
Model
Id
(uA
/um
)
Vds (V)
1.2V
1.1V
1.0V
0.9V
0.8V
0.7V
0.6V
0.5V
0.4V
S. Mayuzumi et al., IEDM Tech. Dig. p. 293, 2007
Experiment data from:
• Only 4 fitting parameters
• Physically-based, Predictive
Model Prediction
15 20 25 30 35 40 45 50
12
15
18
Poly gate
Metal gate
Work in [11]
Channel Length (nm)
Inje
cti
on
Ve
loc
ity
(1
06
cm
/s)
Ballistic velocity w/o strain
Injection velocity w/o strain
Injection velocity with strain
15 20 25 30 35 40 45 500.4
0.5
0.6
0.7
0.8
0.9
1.0
Work in [11]
Poly gate
Metal gate
Channel Length (nm)In
trin
sic
De
lay
(p
s)
W/O Strain
With Strain
ITRS
HP Tech Node 90nm 68nm 52nm 40nm
Lg (nm) 32 25 20 16
Tinv(nm) 1.93 1.84 1.04 0.82
Body Doping
(cm-3) 3.3e18 4.8e18 4.1e18 6.6e18
S/D Series
Resistance
(Ohm.um)
180 200 200 180
Vdd (V) 1.1 1.1 1 1
𝜏 =
𝑉𝑑𝑑 − 𝑉𝑡 +𝐶𝑓𝑉𝑑𝑑𝐶𝑖𝑛𝑣𝐿𝑔
3− 𝛿 𝑉𝑑𝑑
4− 𝑉𝑡
𝐿𝑔
𝑣𝑖𝑛𝑗
Technology Parameters from ITRS 2007
Modern bulk MOSFETs are still far from
ballistic limits
Stress benefits decrease as Lg scales
Planar & 3D MOSFET Design
*J.G. Fossum, IEDM 2004
MuGFET Design Rule
*X. Sun, to be published
Performance Benefits Reduced Variations
*X. Sun, EDL 29,p493(2008)
3D Stress Design for MuGFETs
• Various Stress Configurations
a) Uniaxial Stress
(CESL)b) Biaxial Stress c) Transverse Stress
(Metal Gate)
PolyCESL
Si
Poly
SOI
Si
Metal Gate
Si
Poly
y
Z(100)
x <110> or <100>
Sxx : Szz = 1 : -1 Sxx : Syy = 1 : 1 Syy : Szz = 1 : 1
Only transverse direction
Compressive
NMOS: Tensile
PMOS: Compressive
NMOS: Tensile
PMOS: Tens. & Comp.
• Experiments
0 5 10 15 20 25 30 350
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18strain vs micrometer knob
micrometer knob
str
ain
bi-axile tensile
bi-axile compressive
0.11%
Micrometer Knob
Str
ain
[%
]
TEM Cross-section - Test chip from TI
<110> NMOS
Lg=5um
Gate Voltage [V]
1E-14
1E-12
1E-10
1E-08
1E-06
1E-04
-0.5 0 0.5 1 1.5
0E+00
1E-05
2E-05
0.11%
0%
1E-14
1E-12
1E-10
1E-08
1E-06
1E-04
-1.5 -1 -0.5 0 0.5
0E+00
1E-05
2E-05
3E-05
4E-05
5E-05
0.11%
0%
DR
AIN
CU
RR
EN
T [
A]
TR
AN
SC
ON
DU
CT
AN
CE
[S]
<110> PMOS
Lg=1um
• Simulation (Fully Quantum Mechanical)
N-MuGFET 2D Poisson-EMA Schrodinger
Analytical model for strain impact
non-parabolic E-k for mobility calculation
P-MuGFET2D Poisson-6x6 K.P
Luttiger-Kohn Hamiltonian, with strain
impact coupled
Detailed E-k for mobility calculation
Parameter set (calibrated with experiments):NMOS:Dac = 14.7eVDk_LO=11x1010 eV/mDk_TO=2x1010 eV/mDk_TA=2x1010 eV/mPMOS:Dac = 9.6eVDk = 12.9x1010 eV/mhw= 61.2meVSurface Roughness:
(100) Top surface: =3.30A =20A
(100) Sidewall: =4.05A =20A
(110) Sidewall: =6.90A =26A
hw_LO=64meVhw_TO=64meVhw_TA=48meV
MuGFET Universal Mobility
• Simulation & Measurements
0.0 3.0x1012
6.0x1012
9.0x1012
1.2x1013
1.5x1013
100
150
200
250
300
350
400
450
500
N-MuGFET
HSi
= 58nm
TSi
= 20nm
(100) WaferEle
ctr
on
Mo
bil
ity
(c
m2/V
s)
Ninv
(cm-2)
<100> Unstrained
<100> Metal Gate
<100> Biaxial Tensile
<110> Unstrained
<110> Metal Gate
<110> Biaxial Tensile
Simulation
0.0 3.0x1012
6.0x1012
9.0x1012
1.2x1013
100
150
200
250
300
350
<110> Unstrained
<110> Metal Gate
<110> Biaxial Tensile
<100> Unstrained
<100> Metal Gate
<100> Biaxial Tensile
Simulation
P-MuGFET
HSi
= 58nm
TSi
= 20nm
(100) Wafer
Ho
le M
ob
ilit
y (
cm
2/V
s)
Ninv
(cm-2)
Electron Hole
Mobility vs. Different Types of Stress
• Simulation Results
0 200 400 600 800 1000 1200 14000
20
40
60
80
100
120
140
160
180
Ninv
= 9x1012
cm-2
HSi
= 58nm, TSi
= 20nm
(100) wafer
<110> CESL
<110> Bi. Tens.
<110> Gate
<100> CESL
<100> Bi. Tens.
<100> Gate
Mo
bil
ity
Ch
an
ge
(%
)
Stress (MPa)
0 200 400 600 800 1000 1200 1400-40
-20
0
20
40
60
80
100
120
140
160
180
200
Mo
bil
ity
Ch
an
ge
(%
)
Stress (MPa)
<110> CESL
<100> CESL
<110> Bi. Comp.
<100> Bi. Comp.
<100> Gate
<100> Bi. Tens.
<110> Gate
<110> Bi. Tens.
Ninv
= 1x1013
cm-2
HSi
= 58nm, TSi
= 20nm
(100) wafer
<110> channel/fin is more sensitive to stress
CESL stress gives largest mobility enhancement
Electrons Holes
Bench Marking between FinFET & TriGate
• Simulation Results
0 200 400 600 800 1000 1200 1400100
200
300
400
500
open symbols: FinFET
closed symbols: Tri-Gate
Ele
ctr
on
Mo
bil
ity
(c
m2/V
s)
Stress (MPa)
<100> CESL
<110> CESL
<100> Biaxial Tensile
<110> Biaxial Tensile
0 200 400 600 800 1000 1200 1400
100
200
300
400
500
600
open symbols: FinFET
closed symbols: Tri-Gate
Ho
le M
ob
ilit
y (
cm
2/V
s)
Stress (MPa)
<110> CESL
<110> Biaxial Compressive
<110> Biaxial Tensile
Electrons Holes
FinFET-like:
TSi = 1/3 Lg
HSi = Lg
TriGate:
TSi = Lg
HSi = 1/3 Lg
MuGFET Injection Velocity
• Simulation Results
20 40 60 80 100 1200.0
2.0x104
4.0x104
6.0x104
8.0x104
1.0x105
1.2x105
1.4x105
1.6x105
Vds
= 1V
Ninv
= 1x1013
cm-2
open symbols: FinFET
closed symbols: Tri-Gate
Ele
ctr
on
In
jec
tio
n V
elo
cit
y (
m/s
)
Channel Length (nm)
<100> CESL 1400MPa
<110> CESL 1400MPa
<100> unstrained
<110> unstrained
Extracted Data
uniaxial stressed
unstrained
20 40 60 80 100 120140
0.0
2.0x104
4.0x104
6.0x104
8.0x104
1.0x105
1.2x105
open symbols: FinFET
closed symbols: Tri-Gate
unstrained
Ho
le I
nje
cti
on
Ve
loc
ity
(m
/s)
Channel Length (nm)
<110> CESL -1400MPa
<110> unstrained
<100> unstrained
Extracted Data
uniaxial stressed
Vds
= -1V
Ninv
= 1x1013
cm-2
Electrons Holes
<110> channel/fin has large stress enhancement
TriGate & FinFET is similar due to effective mass domination
Simulation compared with published planar CMOS measured data
Summary
• A Fast Way to evaluate stress-induced layout dependent variations
physically-based functions, rely on few parameters
very fast, yet accurate enough for DFM research
• A Compact Model for MOSFET I-V under quasi-ballistic regime
Only 4 fitting parameters
Predictive for future technology node transistor performance
• A Quantum Simulation Approach for MuGFET mobility study
Design the stress strategy for 3D MOSFET structures
Explore new physical phenomena for scaled devices
• … …
Stress will be used … to the end of CMOS!
New problems research opportunities raised for scaled devices
Acknowledgements
• IMPACT program & UC Discovery Grant
• UC Berkeley Graduate Fellowship
• Test chips donated from STMicroelectronics &
Texas Instruments