Upload
buimien
View
287
Download
1
Embed Size (px)
Citation preview
CMOS Transistor Scaling Past 32nmCMOS Transistor Scaling Past 32nm and
I li i V i iImplications on Variation
Kelin J. KuhnIntel Fellow
Director of Advanced Device TechnologyPortland Technology Development
Intel CorporationIntel Corporation
Kelin Kuhn / ASMC / SFO 2010 1
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 2
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 3
VARIATION SOURCESSOURCES
ContactContact
GateSpacer
Epi RSD
Kelin Kuhn / ASMC / SFO 2010 4
VARIATION SOURCES
STI PolishLER/LWR
OPC
SOURCES Poly
Kuhn ICCV 2009Kuhn ITJ 2009
Kuhn ICCV 2009
Gate granularityGate polish
Gate granularity
GateSpacer
Ranade IEDM 2005
StrainSteigerwald IEDM 2008
Oxide2 1 "Clean" CV @ 1MHz
RDF + implants
1.3
1.5
1.7
1.9
2.1
S C
apac
itanc
e [p
F]
Clean CV @ 1MHz
CV with Traps @ 1MHz
Kelin Kuhn / ASMC / SFO 2010 6Auth, VLSI 2008
0.7
0.9
1.1
-2 -1.5 -1 -0.5 0 0.5 1 1.5PMOS V-Gate [V]
PMO
S
C. PrasadKuhn iEDM 2007
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 7
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 8C. KenyonTOK conf.Dec. 2008
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 9C. KenyonTOK conf.Dec. 2008
Layout Restrictions 65nm to 32nm65 nm Layout Style 32 nm Layout Style
• Bi directional features • Uni directional features• Bi-directional features• Varied gate dimensions• Varied pitches
• Uni-directional features• Uniform gate dimension• Gridded layout
Kelin Kuhn / ASMC / SFO 2010 10M. Bohr, ISCC, 2009
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 11C. KenyonTOK conf.Dec. 2008
Optical Proximity Correction (OPC)As a Resolution Enhancement TechniqueAs a Resolution Enhancement Technique
Contour prediction – no OPC Contour prediction – with OPC
SEM Image – no OPC SEM Image – with OPC
Kelin Kuhn / ASMC / SFO 2010 12K. Wells-Kilpatrick: 2007
45nm: OPC as a Variation Management Technique
Top-down resist CD meets spec, but poor contrast leads to poor resist profile which gets transferred to metal pattern after etch, resulting in shorting marginality
Kelin Kuhn / ASMC / SFO 2010 13Computational lithography solutionK. Kuhn, IEDM 2007
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 14C. KenyonTOK conf.Dec. 2008
MEEF Mask Error Enhancement Factor
• MEEF is a scaling factor that causes certain layout• MEEF is a scaling factor that causes certain layout geometries to exhibit a greater sensitivity to mask dimension tolerances.
• Any dimensional error in the mask is magnified on the wafer by the MEEF value.
Wwafer= MEEF * Wmask
• Depending on the value of the mask error and the lithography exposure/focus conditions the final printedlithography exposure/focus conditions the final printed pattern can be either larger or smaller.
Kelin Kuhn / ASMC / SFO 2010 15
MEEF Impact on Ze Errorp
Ze error can be eitherpositive or negative 65nm Simulation Notch width 120nmpositive or negative 65nm Simulation Notch width = 120nm
Notch height = 250nm
MEEF = 8.4
YellowYellow: DCCD contour after OPCGreen: with -3.375 nm mask making errorR d ith 3 375 k ki
Kelin Kuhn / ASMC / SFO 2010 16
Red: with 3.375 nm mask making error
MEEF and Historical gate CD vs. pitch1262 CD vs. Pitch
90
100
110
120
nm)
nm)
90nm node 1264 PSM CD vs. Pitch
70
80
90
nm)
nm)
65nm node1260 CD vs. Pitch
140
150
160
170
nm)
(nm
)
130nm node
50
60
70
80
100 200 300 400 500 600 700 800 900 1000
CD
(nC
D (n
30
40
50
60
100 200 300 400 500 600
CD
(nC
D (n
100
110
120
130
200 300 400 500 600 700 800 900 1000
CD
(C
D (
Pitch (nm)100 200 300 400 500 600
Pitch (nm)
45nm node
Pitch (nm)
1268 PSM CD vs Pitch32nm node
193nm; APSM; model-based OPC
248nm; OAI; model-based OPC
193nm; OAI; model-based OPC
(nm
)D
(nm
)
45nm node
Contacted
1268 PSM CD vs. Pitch
55
60
65
70
(nm
)
32nm node
D (n
m)
50 100 150 200 250 300 350 400
CD
CD gate pitch
40
45
50
55
80 100 120 140 160 180
CD
C
D
C Kenyon Pitch (nm) Pitch (nm)
193nm; APSM; model- based OPCdouble patterning
193nm; immersion; APSM; model- based OPC; double patterning; polarization
C. KenyonTOK conf.Dec. 2008
Kelin Kuhn / ASMC / SFO 2010 17
Low MEEF requires targeting in the “flat” portion of CD vs. pitch Process innovations continue this trend in the 32nm node
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 18C. KenyonTOK conf.Dec. 2008
FLARE• Flare is unwanted scattered
light arriving at the wafer• Flare is caused by
interactions that force theinteractions that force the light to travel in a "non-ray trace" direction.
• Flare is both a function of local environment around a feature (short range flare)feature (short range flare) and the total amount of energy going through the gy g g glens (long range flare).
Kelin Kuhn / ASMC / SFO 2010 19
Impact of flare on gate CDs All structures have identical reticle CD and pitch
73
74High chrome density
DC
CD
69
70
71
72
m)
1x09
0a D
66
67
68
69
CD
(nm
Moderate chrome density
63
64
65
DV Near Etest E-Test FREA Nested DV Nested Metro
1nm
• During 65nm process development, large CD d i ti b d f t t h i
chrome density
Low chrome density
Structure
deviations were observed for structures having identical pitch and reticle CD due to flare
• Gates only 500m away from one another could C. KenyonTOK conf.Dec. 2008
Kelin Kuhn / ASMC / SFO 2010 20
be >5nm different in CD
Flare Variation Improvement with OPCwith OPC
56.0 – 63.0
49 0 56 0
Color Code
Chrome Fraction56.0 – 63.0
49 0 56 0
Color Code
Chrome Fraction
49.0 - 56.0
42.0 – 49.0
35.0 – 42.0
28.0 – 35.0
49.0 - 56.0
42.0 – 49.0
35.0 – 42.0
28.0 – 35.0
21.0 – 28.0
14.0 – 21.0
7.0 – 14.0
21.0 – 28.0
14.0 – 21.0
7.0 – 14.0
0.0 – 7.00.0 – 7.0
Development effort produced an algorithm capable of scanning designs and binning regions by local chrome fraction
Binning algorithm is combined with flare-calibrated OPC model
Kelin Kuhn / ASMC / SFO 2010 21
g a go t s co b ed t a e ca b ated O C odeC. Kenyon, TOK conf., Dec. 2008
Phase mask Phase mask data
Design
OPC/RET Trim mask Reticle
manufacturingTrim mask data
ExposurePutting it all
together for the gate layer of a
65nm MPU
Etch(magnified 25,000X)
Kelin Kuhn / ASMC / SFO 2010 22C. KenyonTOK conf.Dec. 2008
45nm highlights role of lithography/etch in resolving LER/LWRresolving LER/LWR
Improvement A
Original Final after improvements A,B,C
Improvements B,C
Kelin Kuhn / ASMC / SFO 2010 23K. Kuhn, ITJ, 2008
Technology Trend Systematic Gate CD Lithography Variation
Gate CD variation improvements10
30nm
Systematic Gate CD Lithography Variation
Gate CD variation improvements with technology scaling
lized
to 1
3
0 7X1
on n
orm
a
WID-total0.7X
G (V
aria
tio
WIW-total TOTAL
0.1130nm
33690nm260
65nm220
45nm160
32nm112 5
LOG
GENERATIONGATE PITCH336 260 220 160 112.5
Critical to management of variation is the ability to deliver a 0.7X gate CD variation improvement in each generation
GATE PITCH
Kelin Kuhn / ASMC / SFO 2010 24enabled by continuous process technology improvements
Lithography Pipeline1 1000
W l th
N
Wavelength248nm
193nm
TER
0.1
MIC
RO
N
100OPCPhase shiftImmersion A
NO
MET
32nm22nm
15nmFeature Size EUV
13.5nm
NA
0.011980 1990 2000 2010 2020
1015nm
Extend 193nm Optical Lithography as far as possibleDeploy EUV Lithography when available/affordable
Kelin Kuhn / ASMC / SFO 2010 25
p y g p y
Non-EUV Lithography Beyond 32 nmPitch Doubling 2-D Features
Double PatterningDouble Patterning• Pitch doubling
• Improved 2-D features• Improved 2-D features
Spacer Gate Patterning• Pitch doubling
• Improved variation
Kelin Kuhn / ASMC / SFO 2010 26
M. Bohr, ISCC, 2009Bencher et al, Proc. of SPIE Vol. 6924 69244E-7
Pitch doubling and gate CD control
Pattern transfer layer 2nd pattern transfer layer1st pattern transfer layer Neither Resist Freeze
Gate layerPattern transfer layer
Gate layer2 pattern transfer layer
nor Double Pattern Transfer achieve full benefit of patterningbenefit of patterning at ½ pitch
Both techniques still i l tirequire resolution
of a very small space (MEEF LWR etc )
Resist freeze
(MEEF, LWR etc.)
Double PatternT f
Kelin Kuhn / ASMC / SFO 2010 27
TransferC. Kenyon, TOK conf., Dec. 2008
Disadvantages of Double-patterningMisalignment
DSA
TA
LZED
IDN
OR
MA
Print 1 Print 2Registration (nm)
Misalignment between the 2 exposures is a crucial liability for this technique and can limit its usability
Transistor parameters can be affected by asymmetry between the source and drain regions
Kelin Kuhn / ASMC / SFO 2010 28
Pitch doubling and gate CD matching
in
A AA1266
12641262
860
M V
ccm
i
B B BB1266
SRA
M
Gate CD mismatch
Pitch doubling eliminates the close correlation which currently exists between the CDs of adjacent gates
Thi h i li ti f ll d th i itThis has implications for memory cells and other circuits which depend upon this CD matching
Kelin Kuhn / ASMC / SFO 2010 29C. Kenyon, TOK conf., Dec. 2008
Pitch doubling and gate CD matching
Pitch doublingadjacent gate CDadjacent gate CD
mismatches
Total gate CD di t ib tidistribution
Single patterning adjacent gate CDadjacent gate CD mismatches
Single patterning: the distribution of CD mismatches between dj t t i ll f ti f t t l t CD i ti
-4 -3 -2 -1 0 1 2 3 4
adjacent gates is a very small fraction of total gate CD variation
Pitch doubling: the distribution of CD mismatches is GREATER than the total gate CD variation
Kelin Kuhn / ASMC / SFO 2010 30
GREATER than the total gate CD variationC. Kenyon, TOK conf., Dec. 2008
Non-EUV Lithography Beyond 32 nmPitch Doubling 2-D Features
Double PatterningDouble Patterning• Pitch doubling
• Improved 2-D features• Improved 2-D features
Spacer Gate Patterning• Pitch doubling
• Improved variation
Kelin Kuhn / ASMC / SFO 2010 31M. Bohr, ISCC, 2009Bencher et al, Proc. of SPIE Vol. 6924 69244E-7
Alternative: Spacer patterningp p g
Spacer patterning retains correlation between doubled features
Kelin Kuhn / ASMC / SFO 2010 32Bencher et al, Patterning by CVD Spacer Self Alignment DoublePatterning (SADP), Proc. of SPIE Vol. 6924 69244E-7
Alternative: Spacer patterningp p g
Potential asymmetriesPotential asymmetries
Need for trim mask
Cross-section Top-down
However spacer patterning comes with challenges of its own
Kelin Kuhn / ASMC / SFO 2010 33Bencher et al, Patterning by CVD Spacer Self Alignment DoublePatterning (SADP), Proc. of SPIE Vol. 6924 69244E-7
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 34
HiK-MG: Gate First vs Gate LastGate-First
Dep Hi-k & Met 1
Patt Met 1 & Dep Met 2
Patt Met 2 & Etch Gates
S/D formation & Contacts
Dep & PattHik+Gate
S/D formation &ILD dep /polish
Rem Gate & Patt Met 1
Dep Met 2+Fill &Polish
Hik-First, Gate-Last
Advantages of gate last flow• High Thermal budget available for Midsection
– Better Activation of S/D ImplantsBetter Activation of S/D Implants• Low thermal budget for Metal Gate
– Large range of Gate Materials available• Significant enhancement of strain
Kelin Kuhn / ASMC / SFO 2010
Significant enhancement of strain– Both NMOS and PMOS benefit
Auth – Intel – VLSI 2008 35
CMP Integration at 45 nm – HiK Metal Gate
STI deposition and polish STI STI Wells and VT implants
ALD deposition of high-k gate dielectric
Polysilicon deposition and gate patterning
CMPCMP
POP POP CMPCMPPolysilicon deposition and gate patterning
S/D extensions, spacer, Si recess and SiGe deposition
S/D formation, Ni silicidation, ILD0 deposition
CMPCMP
Poly Opening Polish, Poly removal
PMOS workfunction metal deposition
Metal gate patterning NMOS WF metal deposition
K.Mistry et al., IEDM (2007)C A th t l VLSI S (2008)
Metal gate patterning, NMOS WF metal deposition
Metal gate fill and polish, ESL deposition MGD MGD CMPCMP
First Generation HiK Replacement Metal Gate
C.Auth et al. VLSI Symp, (2008)J. Steigerwald, IEDM (2008)J. Steigerwald, IEDM (2008)
Kelin Kuhn / ASMC / SFO 2010 36
First Generation HiK – Replacement Metal GateThree critical CMP operations in the FE
CMP Integration at 45 nm – HiK Metal Gate
STI deposition and polish STI STI Wells and VT implants
ALD deposition of high-k gate dielectric
Polysilicon deposition and gate patterningPOP POP CMPCMP
CMPCMP
Polysilicon deposition and gate patterning
S/D extensions, spacer, Si recess and SiGe deposition
S/D formation, Ni silicidation, ILD0 deposition
CMPCMP
Poly Opening Polish, Poly removal
PMOS workfunction metal deposition
Metal gate patterning NMOS WF metal deposition
K.Mistry et al., IEDM (2007)C A th t l VLSI S (2008)
Metal gate patterning, NMOS WF metal deposition
Metal gate fill and polish, ESL deposition MGD MGD CMPCMP
C.Auth et al. VLSI Symp, (2008)J. Steigerwald, IEDM (2008)J. Steigerwald, IEDM (2008)
First Generation HiK Replacement Metal Gate
Kelin Kuhn / ASMC / SFO 2010 37
First Generation HiK – Replacement Metal GateThree critical CMP operations in the FE
STR Pattern Density Variation Impacty p
High Pattern Density Low Pattern DensityHigh Pattern Density Low Pattern Density
OxideSiliNitride
OxideSilicon
Slower Polish Rate Faster Polish Rate
Kelin Kuhn / ASMC / SFO 2010 38
STI Step Height VariationHigh PatternDensity Area
Low PatternDensity Area
STI topography
STI STItopography
impacts transistorLe and Ze
Positive Step Height Zero Step Height
Kelin Kuhn / ASMC / SFO 2010 39
STI Step Height VariationHigh PatternDensity Area
Low PatternDensity Area
STI t hSTI STI
STI topographyimpacts transistorLe and Ze
Poly Poly
Zero Step Height
Kelin Kuhn / ASMC / SFO 2010 40
STI Step Height Impact on Gate CD
NegativeSTI
Polyg
Step Height GATE
Diffusion
“Dogbone” Lg is longer at the diffusion boundary
PositiveSTI
PolyStep Height GATE
Diffusion
“Icicle” Gate CD is shorter at the diffusion boundary
Kelin Kuhn / ASMC / SFO 2010 41
Gate CD is shorter at the diffusion boundary
CMP Integration at 45 nm – HiK Metal Gate
STI deposition and polish STI STI Wells and VT implants
ALD deposition of high-k gate dielectric
Polysilicon deposition and gate patterning
CMPCMP
POP POP CMPCMPPolysilicon deposition and gate patterning
S/D extensions, spacer, Si recess and SiGe deposition
S/D formation, Ni silicidation, ILD0 deposition
CMPCMP
Poly Opening Polish, Poly removal
PMOS workfunction metal deposition
Metal gate patterning NMOS WF metal deposition
K.Mistry et al., IEDM (2007)C A th t l VLSI S (2008)
Metal gate patterning, NMOS WF metal deposition
Metal gate fill and polish, ESL deposition MGD MGD CMPCMP
First Generation HiK Replacement Metal Gate
C.Auth et al. VLSI Symp, (2008)J. Steigerwald, IEDM (2008)J. Steigerwald, IEDM (2008)
Kelin Kuhn / ASMC / SFO 2010 42
First Generation HiK – Replacement Metal GateThree critical CMP operations in the FE
Variation Challenges of RMG CMP Steps• Gate height control critical to reducing variation• PMOS/NMOS differences complicate CMPp
nWFM pWFM
NiSi
HiK
NMOS PMOS
Epi S/DHiK
NMOS PMOS
C.Auth et al. VLSI Symp, (2008)
Kelin Kuhn / ASMC / SFO 2010 43J. Steigerwald, IEDM 2008
Variation Challenges of RMG CMP StepsOVERPOLISHExposes raised S/D
UNDERPOLISHUnderetched contact
NMOS S/D
Rext/mobility impact Rext impact
Gate region
region contact
S/D region – attacked during poly etch
S/D region – marginal contact
Kelin Kuhn / ASMC / SFO 2010 44
during poly etchJ. Steigerwald, IEDM 2008
contact
45 nm: POP CMP ImprovementOverscaling Topography ImprovementOverscaling Topography Improvement
1ra
phy 0.7X
improvement
Topo
g
0.1 45nm: 2X greater than
standard
CM
P
0 01
standard technology
scaleJ. Steigerwald, IEDM 2008
Technology node (nm)
0.01350 250 180 130 90 65 45
Improvements in polish enabled dramatic impro ements in topograph ariation
Technology node (nm)
Kelin Kuhn / ASMC / SFO 2010 45
improvements in topography variation
Generational ImprovementsPatterning and Polish
45nm – WIDE32nm – WIDE
0 171 m2
22nm – WIDE0.092 m2
90nm – TALL1.0 m2
65nm – WIDE - 0.57 m245nm WIDE
0.346 m2 0.171 m2
65nm to 22nm: Patterning and polish enhancements
• Improved CD uniformity across STI boundaries• Square corners (eliminate “dogbone” and “icicle” corners)
Kelin Kuhn / ASMC / SFO 2010 46
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 47
MOSFET Ch llChallenges
Capacitance (Increased fringe to
Gate control(SCE limitations ith ll L ff)
( gcontact/facet)Resistance
(Decreased S/D opening) Contact with smaller Leff)Contact
GateSpacer
Epi RSD
Mobility(Reduced strain with
decreased pitch)
Kelin Kuhn / ASMC / SFO 2010 48
decreased pitch)
NextG ti
MOSFET Ch llGeneration
Capacitance (Increased fringe to
Challenges
Gate control(SCE limitations ith ll L ff)
( gcontact/facet)Resistance
(Decreased S/D opening) Contact with smaller Leff)Contact
GateSpacer
Epi RSD
Mobility(Reduced strain with
decreased pitch)
Kelin Kuhn / ASMC / SFO 2010 49
decreased pitch)
Ultra-thin bodyith RSDwith RSD
Contact
SpacerGate
Epi RSD
Ultra-thinbody (UTB)
Kelin Kuhn / ASMC / SFO 2010 50
MuGFET
Contact
SpacerGate
Vertical thin body
Kelin Kuhn / ASMC / SFO 2010 51
MuGFET VARIANTS FINFET TRIGATE PI GATEFINFET TRIGATE PI GATEVARIANTS
Nearly ideal sub-th h ld lha
nnel
gate
FINFET TRIGATE PI-GATE
hann
el
gatehann
el
gate
FINFET TRIGATE PI-GATE
threshold slope (gates tied together)
BOX ch
Silicon
-GATE GAA (GATE-ALL-AROUND)
BOX ch
Silicon
BOX ch
Silicon
-GATE GAA (GATE-ALL-AROUND)
Kelin Kuhn / ASMC / SFO 2010 52
Nanowire
Contact
SpacerGate
Nanowire
Kelin Kuhn / ASMC / SFO 2010 53
Nanowire
ContactLooking at all these i d t il
SpacerGate
in more detail
Nanowire
Kelin Kuhn / ASMC / SFO 2010 54
Ultra-thin bodyith RSD
Benefitswith RSD
Extension of l t h lplanar technology
(less disruptive to manufacturing)
Improved RDF (low doped
channel)
C tibl
channel)
Compatible with RSD
technology
Excellent channel
Potential for body bias
Kelin Kuhn / ASMC / SFO 2010
controly
55
ChallengesUltra-thin bodyith RSD
V i ti
with RSDCapacitance
(Increased fringe to Variation:(film thickness changes affects
VT and DIBL)
(Increased fringe to contact/facet)
Rext: (Xj/Tsi
limitations)
VT and DIBL)
limitations)
Strain:(strain transfer from
Manufacturing:
(strain transfer from S/D into the channel) Performance:
(transport challenges with thin Tsi)
Kelin Kuhn / ASMC / SFO 2010
g(requires both thin Tsi and thin BOX)
with thin Tsi)
56
Ultra-thin bodyBarral – CEA-LETI– IEDM 2007
Ch IBM VLSI 2009Cheng – IBM – VLSI 2009
Lg=25nmTsi=6nm
Kelin Kuhn / ASMC / SFO 2010 57
MuGFET Benefits
Nearly ideal sub-th h ld l
Double-gate relaxes Tsi requirementsFin Wsi > UTB Tsi
threshold slope (gates tied together)
(less scattering, improved VT shift)
Improved RDF (low doped
channel)channel)
Excellent channel controlCan be on
Kelin Kuhn / ASMC / SFO 2010
controlCan be on bulk or SOI
58
MuGFET with RSD
Benefitswith RSD
Nearly ideal sub-th h ld l
Double-gate relaxes Tsi requirementsFin Wsi > UTB Tsi
threshold slope (gates tied together)
(less scattering, improved VT shift)
Improved RDF (low doped
channel)channel)
Compatible with RSD
technology
Excellent channel control
Kelin Kuhn / ASMC / SFO 2010
gy control
59
MuGFET Benefits
Possibility for i d d t t
Double-gate relaxes Tsi requirementsFin Wsi > UTB Tsi
independent gate operation
(less scattering, improved VT shift)
Improved RDF (low doped
channel)channel)
Excellent channel control
Kelin Kuhn / ASMC / SFO 2010
control
60
MuGFET ChallengesVariation
(Mitigating RDF but acquiring Gate wraparound
(Endcap coverage)Capacitance Hsi/Wsi/epi) (Endcap coverage)Capacitance (fringe to contact/facet)(Plus, additional “dead
space” elements)
Small fin pitch
space elements)
Small fin pitch (2 generation scale?)
Fin/gate fidelity on 3’D(Patterning/etch)
Rext: (Xj/Wsi
Topology(Polish / etch challenges)
Fin Strain engr.(Effective strain
t f f fi
Kelin Kuhn / ASMC / SFO 2010
( jlimitations)
challenges)transfer from a fin into the channel)
61
MuGFETKavalieros – Intel – IEDM 2006
Chang – TSMC – IEDM 2009
Kelin Kuhn / ASMC / SFO 2010 62
Nanowire Benefits
Nearly ideal sub-th h ld l
Nanowire further
threshold slope (gates tied together)
Nanowire further relaxes Tsi / Wsi
requirementsImproved RDF
(low doped channel)channel)
Excellent channel control
Kelin Kuhn / ASMC / SFO 2010
control
63
BenefitsNanowire
Nearly ideal sub-th h ld l
Nanowire further
threshold slope (gates tied together)
Nanowire further relaxes Tsi / Wsi
requirementsImproved RDF
(low doped channel)channel)
Compatible with RSD
technology
Excellent channel control
Kelin Kuhn / ASMC / SFO 2010
gy control
64
Nanowire ChallengesGate conformality
Capacitance (fringe to contact/facet)
Gate conformality(dielectric and metal)
(Plus, additional “dead space” elements)
Integrated wire fabrication
Wire stability(bending/warping)
Variation(Mitigating RDF but
wire fabrication (Epitaxy? Other?)
(bending/warping)
acquiring a myriad of new sources)Mobility degradation
(scattering)
Fi / t fid lit 3’D
Rext:
Fin Strain engr.(Effective strain
transfer from wire
Fin/gate fidelity on 3’D(Patterning/etch)
Topology
Kelin Kuhn / ASMC / SFO 2010
(Xj/Wsi limitations)
into the channel)Topology
(Polish / etch challenges) 65
Nanowire FETsDupre – CEA-LETI – IEDM 2008
Bangsaruntip – IBM – IEDM 2009
Kelin Kuhn / ASMC / SFO 2010 66
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 67
Systematic and Random
• Statistician’s viewpoint:
Random
viewpoint:Systematic
• Process engineer’s viewpoint:
FixFixviewpoint:
Random Systematic
• Device engineer’s viewpoint:
VT1
D
VT2 VT1
D
VT2
p
Random Systematic
S D S D S D S D
Kelin Kuhn / ASMC / SFO 2010 68
Random Systematic
Measurement of Random and Systematic VT Variation at the Device LevelVT Variation at the Device Level
Traditional method:VT1 VT2 Traditional method: 1. Measure two identical adjacent devices and
extract the difference (VTA-VTB)2. Measure the entire population of all devices
VT1
S D
VT2
S D2. Measure the entire population of all devices
and extract (VTpop)
Random Variation for a matched pair
)()( DVTVTVTStdDevRandom BAmp
)()( DVTVTVTStdDRandom Variation for a single device 2
)(2
)( DVTVTVTStdDevRandom BAdeviceone
22
2)()(
DVTVTSystematic pop
Systematic Variation for a single device
Kelin Kuhn / ASMC / SFO 2010 69
2
Using Arrays for Variation Measurement(this example is metal resistors)
256 256
(this example is metal resistors)
1024
256structures
In low densityfillers
256structures
in highdensityfillers
Etest structuresin nominal density
fillers fillers
256Nom
256NomNom
densityfillers
Nomdensityfillers
Kelin Kuhn / ASMC / SFO 2010
Using Arrays for Variation Measurement(this example is metal resistors)(this example is metal resistors)
lowdensity
highdensity
nomdensity
nom nom
Kelin Kuhn / ASMC / SFO 2010
Using Arrays for Variation Measurement(this example is metal resistors)(this example is metal resistors)
lowdensity
highdensity
nomdensity
nom nom
Kelin Kuhn / ASMC / SFO 2010
Important of Comprehending de Biasing in Arraysde-Biasing in Arrays
Kelin Kuhn / ASMC / SFO 2010 73
Random and Systematic Variationfor Matched Ring Oscillatorsfor Matched Ring Oscillators
Random:• Calculate Delta 2
200*F BF AFreqBFreqADelta
Calculate Delta
• Random Variation
2FreqBFreqA
)(DeltaStdDevRand per data unit
)(FreqAStdDevSystematic:• Total Sigma per data unit
2)()( FreqBMeanFreqAMean
• Grand Mean2
22
100* RandSyst
• Systematic Variation per data unit
Total Variation: 100*)(FreqAStdDevTotal per data unit
Kelin Kuhn / ASMC / SFO 2010 74
Total Variation:)(FreqAMean per data unit
45nm Product wafer: Random variation
1 5iatio
n
15cm
1
1.5
ZED
% V
ari
8 cm
0
0.5
NO
RM
ALI
Z
K. Kuhn, ITJ 2008
Kelin Kuhn / ASMC / SFO 2010 75N
Standard oscillator
,
Random and Systematic Variation TrendsNormalized systematic variation
standard deviation per oscillator (%)
4
5
(%)
1
2
3
4
PE
RC
EN
T Systematic WIW variation is comparable from one generation to the next
0
1
130nm 90nm 65nm 45nm 32nm
P generation to the next
Normalized random variationstandard deviation per oscillator (%)
5
%)
Random WIW ariation in
2
34
ER
CE
NT
(% Random WIW variation in 32nm is comparable to 45nm and significantly
HiK-MG
01
130nm 90nm 65nm 45nm 32nm
PE improved over 65nm and
90nm due to HiK-MG
Kelin Kuhn / ASMC / SFO 2010 76
AGENDA
I. Overview – variation sourcesII. Next generation variation – lithographyIII. Next generation variation – polishg pIV. Next generation devicesV Measurements results and interpretationV. Measurements, results and interpretationVI. Closing thoughts
Kelin Kuhn / ASMC / SFO 2010 77
Yield: A pragmatic measure of variation
90 nm 65 nm 45 nm 32 nm350 nm 250 nm 180 nm 130 nm
og s
cale
)D
ensi
ty (l
oD
efec
t D
2002 2003 2004 2005 2006 2007 2008 2009 20101993 1994 1995 1996 1997 1998 1999 2000 2001
Kelin Kuhn / ASMC / SFO 2010 78
Yield: A pragmatic measure of variation
90 nm 65 nm 45 nm 32 nm90 nm 65 nm 45 nm 32 nm350 nm 250 nm 180 nm 130 nm
og s
cale
) 90 nm 65 nm 45 nm 32 nm
(log
scal
e)
Den
sity
(lo
Den
sity
(
Def
ect D
Def
ect
2002 2003 2004 2005 2006 2007 2008 2009 20101993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
2002 2003 2004 2005 2006 2007 2008 2009
Kelin Kuhn / ASMC / SFO 2010 79
Closing ThoughtsClosing Thoughts
Process variation is not anProcess variation is not an insurmountable barrier to Moore’s
L b t i i l thLaw, but is simply another challenge to be overcome.g
Kelin Kuhn / ASMC / SFO 2010 80
Q&AQ&AQ&AQ&A
Kelin Kuhn / ASMC / SFO 2010 81