Upload
others
View
14
Download
0
Embed Size (px)
Citation preview
Tutorial on Design For Manufacturability
for Physical DesignAndrzej J. Strojwas
PDF Solutions Inc., San Jose, CA& Carnegie Mellon University, Pittsburgh, PA
2005 ISPD, San Francisco, CAApril 5, 2005
Overview of Presentation
Yield Loss Mechanism Evolution
Classification of DFM Approaches
True DFM: Defining Proactive DFM
Necessary Conditions for Proactive DFM
Process Characterization
Design Flows that Provide Proactive DFM
DFM Results
Looking into the Future: Extreme Layout Regularity
Yield, Performance, Profitability
The Evolution of Product Yields
Random defects are no longer the dominant yield loss mechanism• Yields are limited by design features, systematic and parametric effects
Yield Limiters by Technology Node
40%
50%
60%
70%
80%
90%
100%
0.8um
0.5um
0.35u
m0.2
5um
0.18u
m0.1
3um
90nm
Technology
Yiel
d
Random Defect Limited YieldDesign Feature Limited YieldTotal Yield
Yield, Performance, Profitability
Random Yield Loss Mechanisms – Al Interconnect
Contact and via opens due to formation defectivity
Active, poly and metal shorts and opens due to particle defects
Random
Yield Loss MechanismsType
Material opens
Material shorts
Yield, Performance, Profitability
Systematic Yield Loss Mechanisms - Cu Interconnect
Misalignment, line-ends/borders
Contact/via opens due to local neighborhood effects (e.g. pitch/hole size)
Leakage from STI related stress
Impact of micro/macro loading design rule marginalitiesSystematicYield Loss MechanismsType
Failure Rate
020406080
100120140160
0.4 1.8 4.2 9
Pitch (um)
Via
Fai
lure
Rat
e (fp
b)
Sparse neighborhood
Yield, Performance, Profitability
Systematic Yield Loss: Printability – Nanometer Era
Material opens
Poor contact coverage due to misalignment and defocus/pull backSystematicYield Loss MechanismsType
Poly/Metal shorts
Yield, Performance, Profitability
Parametric Yield Loss Mechanisms – Nanometer Era
Non-physical corner modelingDevice mismatchDummy fill parasitic effects
Performance variation from lithography effectse.g., contact/via coverage, active/poly flaring, CD variation
Parametric
Yield Loss MechanismsType
ACLV / CD VariationEnvironment dependent poly CD variation
-10.50-10.00-9.50-9.00-8.50-8.00-7.50-7.00-6.50-6.00
300.00 400.00 500.00 600.00 700.00 800.00Idrive
Ioff
Env I Env II Env III
Yield, Performance, Profitability
Technology Challenges: Implications for Manufacturability
65nmLitho:
OPC/PSM integration
issues w/photo window
(DOF)
Front-end/Transistor
Layout dependent
performance
Product ramp issues
Parametric variations -
> yield loss
65nmLitho:
OPC/PSM integration
issues w/photo window
(DOF)
Front-end/Transistor
Layout dependent
performance
Product ramp issues
Parametric variations -
> yield loss
45nmLitho:
Layout pattern
dependence, Scanner NA,
Immersion litho, OPC/PSM
integration, issues w/photo
window (DOF)
Front end/Transistor
New transistor
architectures (UTB, DG SOI)
Product ramp issues
Reliability assurance
45nmLitho:
Layout pattern
dependence, Scanner NA,
Immersion litho, OPC/PSM
integration, issues w/photo
window (DOF)
Front end/Transistor
New transistor
architectures (UTB, DG SOI)
Product ramp issues
Reliability assurance
90nmBack-end integration
issues
Low k: stress and
reliability
CMP - multi-layer
topography issues
Product ramp issues
Variability
Yield-performance
tradeoffs
90nmBack-end integration
issues
Low k: stress and
reliability
CMP - multi-layer
topography issues
Product ramp issues
Variability
Yield-performance
tradeoffs
Yield, Performance, Profitability
DFM is a Business Opportunity"Seamless" DFM Can Contribute 5% More Good Die
What's 5% more good die worth?
$50M over the life of a cell phone
$80M over the life of a game chip
$100M per year per fab at 90nm
Yield, Performance, Profitability
A Brief History of DFM
Design rules guarantee yield!…well, not really……then recommended rules …and opportunistic design data base post-processing to enforce them
Functional Yield means Rules
The corners represent the process…The corners don’t represent the process but they are conservative…Within chip variations are important so..
• Restrict transistor layouts?• Statistical timing simulation?
Performance Yield is Covered by CornersIdsat Distribution
-500
-450
-400
-350
-300
-250
-200600 650 700 750 800 850 900 950 1000 1050
NIdsatPI
dsat
ASIC Corners
Realistic Corners
Yield, Performance, Profitability
Post-GDS Yield Opt.
OPC/ RET
DesignDesign ManufacturingManufacturingMDP and
Mask Making
Yield Ramp
IP lib. Design
pre-MDP
Timing and SI
AnalysesPhysical
Verification
Physical SP&R
Verification and Yield Opt.Verification and Yield Opt.Dummy Fill
and Cheesing
Volume Production
Post-GDS Yield Opt.
OPC/ RET
DesignDesign ManufacturingManufacturingMDP and
Mask Making
Yield Ramp
IP lib. Design
pre-MDP
Timing and SI
AnalysesPhysical
Verification
Physical SP&R
Verification and Yield Opt.Verification and Yield Opt.
Dummy Fill and
CheesingVolume
Production
Process Characterization + Yield Modeling
DFM needs to be ProactiveOccurring early in the design flowUp-front accurate process characterizationYield modeling to characterize IP and drive EDA tools
Proactive DFMDesigner access to the process is limitedMost DFM today is Reactive
•Increase in design cycle time•Misaligned mask GDSII and design database
•Risky design feature changes
Yield, Performance, Profitability
Necessary Conditions for Pro-Active DFM
Accurate characterization of design-interactions at target fab(s)
Effects modeled across the whole process window
Quantification of alternatives that allow EDA tools to make millions of DFM trade-offs
Integration early in the design flow where there are more degrees of freedom
• Floor planning• SP&R
Modifications made prior to verification
Yield, Performance, Profitability
Yield Simulation is Core to Proactive DFM
A yield model for DFM:• Model of failure rate of a design element (e.g., transistor, contact, via)
as a function of the layout design• Example:
• What is the failure rate of a single via vs. double?• What is the probability of a short in two metal lines if there are lots of vias
underneath them
To do this, we need process characterization
Yield Simulation allows for better understanding of the DFM universe
Yield, Performance, Profitability
Process Yield Loss Mechanisms
ActivePoly Poly
CC
M1V1
M1V1
M2M2V2V2
Mx
Yield Loss MechanismsYield Loss Mechanisms•Layer defect densities•Attribute dependent failure rates•Lithography / CMP driven interactions•…
Yield Loss Mechanisms (Yield Loss Mechanisms (YLM’sYLM’s))• Root causes of process related yield loss• Each YLM must be characterized in the process with a specific test
structure – Characterization Vehicle (CV)
Yield, Performance, Profitability
Typical Die Yield Estimation
Model implies that for a given die size, yield is based only upon process maturity – this is not correct
• Yield is different for different IP content• Process defectivity is different for each module
)(cm area dieA
)(1/cm defects meanDyielddieY
whereeY
2
20
AD
=
=
=
= −
,
0Die Yield Model
0%
20%
40%
60%
80%
100%
0 0.5 1 1.5 2Mean Defectivity (1/cm^2)
Die
Yie
ld
Area = 2 cm^2Area = 1.5 cm^2Area = 1 cm^2Area = 0.5 cm^2
Yield, Performance, Profitability
Process-Design Interaction
Chip Yield is a strong function of design content• Physical design features interact with specific module weakness
Memory
Logic
Memory
Logic
Ratio << 1.0Ratio << 1.0 Ratio >> 1.0Ratio >> 1.0
LOGICMEMORYYYY =
Die Content Impact on Yield (~40mm^2 Die Size)
5%
15%
25%
35%
45%
55%
0.0 1.0 2.0 3.0 4.0Memory/Logic Ratio
Yiel
d
Total Yield (130nm)Total Yield (90nm)
Yield, Performance, Profitability
Accurate Process Characterization
Defect Size Distribution by Process Module
Defect Size [um]D
efec
t Den
sity
[D
efec
ts/c
m^2
]
ondistributitsize/coundefectDSD(x) =
Metal Opens/Shorts Metal Opens/Shorts Characterization VehicleCharacterization Vehicle
Poly and Active Opens/Shorts Poly and Active Opens/Shorts Characterization VehicleCharacterization Vehicle
Yield, Performance, Profitability
Product Design Attributes
SRAM
ASIC Logic
ROM
Analog
I/O
Custom Logic
Design Attributes:Design Attributes:• Physical design properties that interact with specific module marginalities• Each attribute can be extracted from physical layout• Design attribute extraction (DAE) enables quantification of design content specific
YLM models
Design AttributesDesign Attributes• Widths, lengths and spacing• Counts• Densities• Overlaps/enclosures• …
Yield, Performance, Profitability
Design Attribute Extraction
Contact/Via Count Design Attribute Extraction
Contacts and Vias
Cou
nt
Critical Area Shorts Design Attribute Extraction
Defect Diameter [um]
Crit
ical
Are
a [u
m^2
]
Contacts/Via Counts•N/P Active•Poly•V1-Vx
Critical Area Shorts Sensitivities•AA•Poly•M1-Mx
xtoty sensitiviattributedesign(x)AC =Yield, Performance, Profitability
Accurate Process Characterization
Process Margin
0
0.2
0.4
0.6
0.8
1
1.2
0
0.02
0.04
0.06
0.08 0.
1
0.12
0.14
0.16
0.18 0.
2
Spacing
Yiel
d
0
0.05
0.1
0.15
0.2
0.25
0.3
p(sp
acin
g)
Layo
utMe
tric
Misalignment
Mask Error
DOE on litho parameters RSM of layout metric
INPUT: Characterization of litho process statistics
Defocus
Exposure
Yield Loss
Defocus
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Misalignment
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
−3.0 σ
−2.5 σ
−2.0 σ
−1.5 σ
−1.0 σ
−0.5 σ
0.0 σ
0.5 σ
1.0 σ
1.5 σ
2.0 σ
2.5 σ
3.0 σExposure
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Yield, Performance, Profitability
PSD: condition 1
PSD: condition2
PSD Analysis
Process Margin data from CV
Implicit YLM model: process window effects
Yield, Performance, Profitability
Yield Simulation Results: Yield Impact Matrix
SRAM
ASIC Logic
ROM
Analog
I/O
Custom Logic
Physical design properties that interact with specific module defectivitiesEach attribute can be extracted from GDSIIDesign attribute extraction (DAE) enables quantification of die content specific yield models
Design AttributesDesign Attributes• Widths, lengths and spacing• Counts• Densities• Overlaps/enclosures• …
YIMP Matrix
Active Random 98% 99% 99%Pattern Dependent 99% 99% 100%
Total 97% 98% 99%Poly Random 97% 98% 99%
Pattern dependent 94% 95% 99%Total 91% 95% 96%
Metal Random 97% 98% 99%Pattern dependent 97% 98% 99%
Total 94% 95% 99%Holes 97% 98% 99%
Total 81% 82% 99%
Metal Islands 87% 89% 90%Pattern Density 91% 93% 94%Narrow Space Wide Neighbor 97% 99% 100%Via induced Metal Shorts 77% 78% 79%
Total 62% 64% 64%
Ran
dom
Def
ects
Sys
tem
atic
LOGIC
Limited Yield
Failure Mode Full Chip SRAM
Yield, Performance, Profitability
Yield Simulation Accuracy
Yield modeling accuracy is excellent when you characterize right• Error represents unidentified yield loss mechanisms
or lack of yield model
Accuracy of Yield Prediction
0%
20%
40%
60%
80%
100%
A B C D E FProduct
Yiel
d
-10%
-8%
-6%
-4%
-2%
0%
2%
4%
Abs
olut
e Er
ror
PredictedActualAbs. Error
Yield, Performance, Profitability
Overview of Layout Design for Manufacturability
Critical Area Based Wire Spreading • Most useful in Al interconnect era (random metal shorts – dominant
yield loss mechanism)
Contact/via Doubling• Effective in reduction yield losses due to random hole opens (Al & Cu)
Systematic Yield Model Based Local and Global P&R Layout Modifications
• Essential for 3-D topography effects due to Cu CMP and low-k interconnect
Printability and Performance Variability Driven Layout Generation
• Absolute must in the Nanometer Era• Pro-active (not afterthought in RET)
Yield, Performance, Profitability
Limited Yield Optimization
Product yields must be optimized by improving LY’s• LYGAIN ratio of improvement for any specific optimization
Prioritizing which LY’s to improve requires an understanding of the specific Design Manufacturability ObjectivesDesign Manufacturability Objectives
LY(SRAM)
LY(ASIC)
LY(ROM)
LY(Analog)
LY(I/O)
LY(Custom)
( )( )
variant originalvvariant optimizedv
Parametric c, SystematiRandomjwhere
Y
YLY
2
1
j vj
vjGAIN
==
=
= ∏
,,
2
1
Yield, Performance, Profitability
0
10
20
30
40
50
60
40% 60% 80% 100%Yield
# of
lots
Design Manufacturability Objectives
Best Yield Variability
Best Average
Yield
Tailored specifically for a process and product
Custom
MicroprocessorsPerformance Variability
Networking, Graphics, DSP
Yield Variability
DVD, STB, Cell Phone, Digital Cameras
Average Yield
ProductsObjectives
Customization of DFM objectives• High volume parts in mature process• Lower volume parts during ramp
Yield, Performance, Profitability
Enabling Proactive DFM For DesignersDesign Flow
RTL Design
Hierarchical Floorplan
Physical Synthesis
Chip Assembly
Sign-off
VERI
FICA
TION Yield Models
Yield View (.pdfm)
DFM Optimized
LibraryModule
Standard IP Platform
DFM Software
Yield Simulation
and optimization
Three components to enable Proactive DFM• DFM library module• Layout attribute dependent yield models (DFM library view)• Yield simulation and optimization software
Yield, Performance, Profitability
DFM Variant Generation Flow
Cell LayoutsCell Layouts
Manufacturability Hot Spot LocalizationManufacturability
Hot Spot LocalizationConstraint generation
Constraint generation
Layout Design
DFMLibraryDFM
Library
Hot Spots Minimized?Hot Spots
Minimized?No Yes
MFG Characterization
MFG Characterization
PruningPruning
DFM Architecture SpecificationsDFM Architecture Specifications
Average Yield
Average Yield
Yield Variability
Yield Variability
Performance Variability
Performance Variability
NetlistNetlist
Yield, Performance, Profitability
Yield Variants ExampleHigh Density Random/Systematic Yield Variant
Litho hot spots
M1 shorts
M1 opens
Yield/Speed Consistency Variant
+20% area
-50% FR (ppb)
+40% area
-50% FR (ppb)
-85% YV (std.dev)
Yield, Performance, Profitability
Standard Cell Optimization Flow
Litho Environment
Hot Spot Analysis
Litho Simulated Layout
0.14u
Optimized Layout
0.18u
Synthesized Layout
0.16u
Optimized Layout
0.2u
MOS1 d g s b nmos L=0.13u W=1uMOS2 d g s b pmos L=0.13u W=0.8uMOS3 d g s b nmos L=0.13u W=0.7uMOS4 d g s b pmos L=0.13u W=1.2uMOS5 d g s b nmos L=0.13u W=1.3uMOS6 d g s b pmos L=0.13u W=1u
Original Netlist
Contact redundancy definitionSpacing rules definitionTransitions rules definitionMetal islands definitionTaps…
Cell Architecture
DFM Library View for EDA
.pdfm view.pdfm view includes relevant data to enable EDA tools to make yield tradeoffs
• Process Defectivities• Design Attributes• Yield and Manufacturability
Models
.pdfm data also enables BEOL routing optimzation
Characteristic Library View
Layout GDSII
Schematic SPICE Netlist
Logic Function Verilog
Manufacturability .pdfm
Performance .lib
P&R Footprint LEF
DFM Library Module
Core Libraries
Process Characterization
Design Attribute Extractions
Defectivities
Compiler
DFM Library View
“pDfx enabled” P&R
Include yield in the cost functionTake advantage of design slacksTiming convergence properties unchanged
Global/ Detailed
Routing
Logic/Physical Synthesis
Clock routing
Timing Sign-off
Post-Routing
Optimization
.PDFMSTD.CELL YLM models
STD.CELL YLM models
BEOL YLM models
BEOL YLM models
Stat. Gate Models
CMP/Litho Models
And the Silicon Says…
Silicon shows up to 12% GDPW improvement by applying our Silicon shows up to 12% GDPW improvement by applying our Proactive DFM methodologyProactive DFM methodology
0%2%4%6%8%
10%12%
GD
PW
Impr
ovem
ent (
%)
A (0.15um)
B (0.13um)
C (0.13um)
D (0.13um)
E (0.13um)
Chip / Technology
Percentage GDPW Improvement• Fabless• Graphics chip• Early process
lifecycle
• IDM• Controller chip• Mid-process lifecycle
• IDM• Cell phone chip• Mid process lifecycle
• IDM• Set top box• Mid-process lifecycle
• IDM• Graphics• Mature process
Yield, Performance, Profitability
Looking Into the Future: Extreme Layout Regularity
RETs are time-consuming and difficult to optimize for large process windows• Defocus, Illumination conditions, Pattern neighborhood
Conventional design rules are becoming insufficient to guaranteedesign’s adherence to RETs• Increasing need for more geometry regularity by design
Sensitive to growdue to defocus
Sensitive to shrinkdue to defocus
Sensitive to exposure variation
Sensitive to resist effects
Regular Logic Bricks
Arrays offer advantages for silicon characterization and even inventory, but at a high area penaltyRegular Logic Bricks can offer more efficient use of area, and still provide the required regularity
Map a simple set of logic primitives onto regular fabric patterns to form logic bricks
Macro Regularity
Optical proximity:
r ≈ 1um
2-3 INVs on boundaries of
logic bricks
Regular fabric ensures that logic bricks share geometry regularity with all other bricks and registers Fabric “rules” co-optimized for memories, IPs, and bricksCompatible RETs and geometries at brick boundaries
On-gridgeometries
On-gridgeometries
On-gridgeometries
On-gridgeometries r
r
Region of influence
Some Experimental Results
Little area benefit for using more than 10 unique bricks
• Few unique bricks allows for macro regularity and silicon validation analogous to CLBsand bit cells
Trade-off between number of geometry patterns and area/performanceIdeally, regular fabric synthesis tool would determine this for each design in an application-specific manner
Firewire Controller
0
5000
10000
15000
20000
25000
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Number of Bricks
Are
a Es
timat
e
0
1
2
3
4
5
6
7
8
9
10
Avg
. no.
of V
ias
per B
rick
AreaNo. of vias
Koggstone
0
5000
10000
15000
20000
25000
30000
35000
40000
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Number of Bricks
Are
a Es
timat
e
0
1
2
3
4
5
6
Avg
. no.
of V
ias
per B
rick
AreaNo. of vias
Exploiting RegularityLogic regularity provides opportunities for new predictable methodologies
1. Small library of primitives2. Mapping into a set of configurable
pre-defined bricks
Factors f & f’ from FG1 & G2 are positive & negative cofactors with respect to function f
Ultimate Goal
From synthesis we define the set of regular logic bricks that cover the extracted logic modulesFrom the fabrics level, perform shapes-level physical-synthesis to construct configurable bricks• Nano-scale constraints allow us to
simplify this shapes-level physical synthesis problem
• Can be optimized with simulation-based printability modeling
Automated layout
Further ConsiderationsGlobal effects and system-level issues
• Memory and logic compatibility • Regular BEOL – structured routing• Analog components
Cu thickness distribution
WLAN WCDMA GPS