Transcript flcc_021207

FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
1
Collaborative Platform, Tool-Kit, and
Physical Models for DfM
FLCC Pre-Presentation Feb 12th,
SPIE March 1, 2007
Andy Neureuther, Wojtek Poppe,
Juliet Holwill, Eric Chin, Lynn Wang, Jae-Seok Yang,
Marshal Miller, Dan Ceperley, Chris Clifford
Jihong Choi, Dave Dornfeld
UC Berkeley,
Koji Kikuchi,
Sony Visiting Industrial Fellow
John Hoang, Jane Chang
U.C. Los Angeles
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
2
Need for Design for Manufacturability (DfM):
Parametric Yield Loss Increasing to 25%
Yield σ: ±5%
Yield σ: +5% to -50%
Random
100
90
Process-Related
Systematic
Reticle-Related
(RET impact)
80
Nominal Yields (%)
Systematic
70
60
50
Leakage Power,
Delay & Cross-Talk
Design-Related
(leakage,
performance,
power)
40
30
20
10
0
0.35µm
0.25µm
0.18µm
0.13µm
Feature Dimension
90nm
Source: IBS
Discussion courtesy of Nickhil Jakatdar
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
DfM Requirements: Context-based,
Adaptive Model Resolution

In the Design

Increased model resolution
for optimizing critical and
sensitive paths
In the Flow

Adaptive model resolution
and speed-accuracy tradeoff to match abstraction
level
RTL
Synthesis
Prototyping
Physical
Synthesis
Nets/Paths
Routing
Regions
Optimization
Sign-off
Slide courtesy of Nickhil Jakatdar, Cadence
Copyright 2007, Regents of University of California
Model Resolution

3
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
4
Technology Advances to Integrate into DfM
Slide courtesy Martin van den Brink, Photomask 06
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
5
Opportunities in DfM
 Platforms



Broad Reach to encompass all contributions to complexity
New Collaborations (Process, Device, Design) supporting
viewpoint of each with the intuitive terminology of each
New Functionalities for Visualization and Assessment
 Physical


Models
Very Fast and first-cut accurate
Complete set of processes:


to Integrate Complexity
Litho, CMP, Etch, Device, Metrology
Complete set of interactions between processes:

Litho-CMP
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
6
Key Leverage Point for DfM:
Express nonidealities at the mask plane
/NA)
/NA)
Defocus
Spherical
Mask phases
• yellow = 0°
• green = 90°
• red = 180°
HO Spherical
/NA)
Coma

/NA)
/NA)
HO Coma
Move nonidealities to the mask plane for
visualization and quantification early in Design
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
7
Lithography Network: Tools and Concepts
defocus
Aberration
Monitors
TEMPEST
•90°
•0°
•180°
•270°
Pattern Matching
Pattern
(coma)
IFT
Mask
Layout
Collaborative Platform for DFM
Pattern
Matcher
Aerial
Image
Simulator
Module 1
Processing
BSIM model
SPLAT
Match
Location(s)
Module 3
Circuit
Standardized
BSIM Model
Output
Copyright 2007, Regents of University of California
Module 2
Device
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
8
Feature Level Compensation and Control:
Industry and UC Discovery Collaboration
SVTC
Copyright 2007, Regents of University of California
17 Supporters
2 Contributors
3 Former
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
9
FLCC: Modeling and Characterization
Litho
Etch
Metrology
and Control
oxide
poly-Si
CMP
oxide
Device
Collaborative Experiments
Toppan & Photronics
Cypress & ASML
AMD
CMP
Drain 1
Cypress
Cypress Poly
Cypress Poly
Block Fab.
Wafer
Block
ON
Metrology
Contact
Metal Active Center
Poly
DuPont
Cypress DDLI
DDLI
Cypress
Block
Mask
Block
Drain 2
ON
ON
ON
Quasar
OPC
Poly
Annular
OPC
Poly
Corner
Poly
ON
Source 1
Copyright 2007, Regents of University of California
Source 2
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
10
Process Aware EDA Toolkit
Lateral Image Effects
Robustness
Metrics
Tool Kit
Interactions & Placement
Crosstalk
Interconnect
Delay
Move polygon
with mouse
Easy to find
optimal
solution
Module 1
Processing
BSIM model
Module 3
Circuit
Standardized
BSIM Model
Output
Module 2
Device
Original
proximity
effect
As polygon moves, hotspots disappear and appear
Collaborative Platform for DfM
Drag-and-Drop Hot Spot Fixer
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
11
Collaborative Platform for DfM:Concept
Wojtek Poppe
Parametric Yield
Simulator
Transistor
Modeling
Simulation
Process Simulation
Collaborative
Platform
for DFM
Experiment
Circuit Simulation
65nm Testchips
SolutionsCopyright
across2007,
disciplines
rather
than within disciplines
Regents of University
of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
12
Collaborative Platform for DfM: Implementation
Module 1
Processing
Viewpoint Rosetta Stone
BSIM model
Module 3
Circuit
Standardized
BSIM Model
Input
Module 2
Device
Intuitive Parameters
Circuit Simulation across
characterized process
window
Copyright 2007, Regents of University of California
Non-rectangular
transistors
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
13
Collaborative Platform for DfM: Implementation
SPLAT
Module 1
Processing
Viewpoint BSIM
Rosetta Stone
BSIM model
Module 3
Circuit
Standardized
BSIM Model
Input
HSPICE
Module 2
Device
Intuitive Parameters
Circuit Simulation across
characterized process
window
Copyright 2007, Regents of University of California
Non-rectangular
transistors
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
14
Collaborative Platform for DfM: Use
Circuits
Design Evaluation
Implementation
on Design Side
Processing
Incremental
Improvements
Module 1
Processing
BSIM model
Module 3
Circuit
Standardized
BSIM Model
Output
Module 2
Device
CAD
Create robustness
metrics for process
aware timing and
power analysis
Copyright 2007, Regents of University of California
Device
Use standard
BSIM model
Modeling
non-rectangular
transistors
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
15
Collaborative Platform for DfM: Validation
Six contributors
from five
research areas
FLCC
Enhanced Transistor
Electrical CD Metrology
Made
possible
Feature Level
Compensation
Copyright
2007,
Regentsby
of University
of California
and Control (FLCC)
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
Process Aware EDA Toolkit
Wojtek Poppe Lynn Wang
Eric Chin Juliet Holwill Jae-Seok Yang
Robustness Interconnect Lateral Image
Metric
Delay
Interactions & Placement Crosstalk
• quantifying the circuit performance robustness of
layout snippits with an indexing metric
• control of leakage through maximizing optical image
quality of drivers/buffers,
• mitigating optical spillover effects and optimizing
robustness metrics in placement,
• visualizing chip level effects on delay variation, and
• checking robustness closure through estimating
variations in interconnect.
Copyright 2007, Regents of University of California
16
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
17
Drag-and-Drop Hot Spot Fixer
Move polygon
with mouse
Easy to find
optimal
solution
Original
proximity
effect
As polygon moves, hotspots disappear and appear
•
•
•
Problem hotspot identified in layout on left.
Drag and Drop Hotspot Fixer on right shows how a designer can drag
a polygon with a mouse and have real-time hotspot re-evaluation as
the polygon moves.
Hotspot severity is color coded, green means hotspot free.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
18
Visualization of Focus Effects at Layout
Juliet Holwill
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
19
Image Behavior with Focus and Coma
Intensity Vs Distance at a range of coma levels for 0.193 NA = 0.5, defocus = 0.01
Intensity Vs Distance at a range of coma levels for 0.193 NA = 0.5, defocus =
-0.01 0.9
Intensity versus Distance for a range of focus values with Lambda
= 0.193 NA = 0.5
0.8
0.7
1.6
0.6
Intensity
1.4
1.4
1.2
0.5
0.4
0.3
1.2
0.2
1
0.1
-0.04 Defocus
1
-0.04 coma
0
0 Defocus
0.02 Defocus
0.6
0.04 Defocus
Intensity
Intensity
1
-0.02 Defocus
0.8
1.05
1.1
1.15
1.2
1.25
Distance (um )
1.3
-0.02 Coma
0 Coma
0.8
0.02 Coma
0.04 Coma
0.6
0.4
0.4
0.2
0.2
0
0
0
0.5
1
1.5
2
2.5
3
3.5
0
0.5
1
2
2.5
3
3.5
Distance (um)
Distance
Cutline
1.5
The intensity change with focus
generally increases regardless of the
sign, whereas the intensity of coma
changes sign with the sign of coma.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
Pattern Matching Accuracy:
Line End Shortening (LES)
20
Line End Shortening
0.06

I
L 
dL dI
LES
Line End Shortening (um)
0.05
0.04
Coma
0.03
0.02
0.01
0
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.04
0.05
Coma Amount (waves RMS)
Aberration Level
Line End Shortening
0.085
Defocus
LES

LES can be modeled using
the product of the match
factor times the aberration
level.
For Coma, LES is linear
For Defocus, LES is
parabolic
Line edge shortening (um)

0.08
0.075
0.07
0.065
0.06
0.055
0.05
0.045
Juliet Make
0.04
LES = Line End Shortening
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
Defocus Amount (waves RMS)
Aberration Level
Copyright 2007, Regents of University of California
0.03
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
21
Phase-Shifting Mask Spillover Trends
Mask
Type
Slope
Coma
MF
ΔL
Splat
ΔL PM Focus
(0.02 Coma)
MF
(0.02 Coma)
ΔL
Splat
(0.04 Focus)
Binary 6.95
0.075
4nm
0.309
38nm
Att.
PSM
0.094
2nm
0.318
27nm
7.933
Alt.
9.083 0.150 18nm
0.342 30nm
PSM
The Pattern Match Factors and spillover
light increases with the additional light
Cutline
through the mask but the impact on edge
placement is partially mitigated by the
increase in image slope.
Copyright 2007, Regents of University of California
ΔL PM
(0.04 Focus)
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
22
Off-Axis Illumination Trends:
Mutual Coherence Mask Functions
Tophat Illumination
Quadrupole Illumination
Annular Illumination
Dipole Illumination
Mutual coherence functions taken from “Resolution Enhancement Techniques in Optical Lithography”, Alfred Wong.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
23
Polarization Trends
Radial periodic grating focuses high-NA effects into center



Ez(center) = (- j 2 Ex NA) 2 cos()
0
= Ez(null, L=S)
2 sin()
(unpolarized light)
Intensity (CF)
Maximum signal at  = 90
1.0
0.66
0.33
0.4
0.6
Copyright 2007, Regents of University of California
0.8 Make
Andy
NA 1
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
24
Assessment of Gate Layout Effects: Approach
The pattern matcher is a fast tool for searching layouts for locations
with the highest similarity to a given image.
The input patterns are chosen to be the most sensitive to a particular
aberration or illumination error.
The snippets from these locations can then be simulated, rather than
the whole layout
 The
match factor is a measure of
similarity of a layout geometry to a
pattern at a particular location
 It is calculated as the 2D discrete
Juliet Make
convolution
Copyright 2007, Regents
of University
California
 Range
= of[-1,+1]
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
25
Assessment of Gate Layout Effects: Results
Match Factor Vs Intensity Change with Coma
MF
I 
2
Intensity Change
0.15
The match factors returned
for a given layout may be
used to predict the expected
intensity change in the
presence of coma
-0.05
-0.10
Match Factor
0.25
The match factors returned for a given
layout may be used to predict the expected
intensity change in the presence of coma
Changing pitch and
additional ‘kickers’ give
this layout snippet a high
match factor
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
26
DRC Compatible Monitors and Calibration
Many modifications are possible, and each will
be tested for sensitivity. These are some
examples of constructed patterns that might be
produced.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
27
Going where “Design Rules Do Not Reach”
Coma
proximity
effect
• Function is about 5 feature sizes in diameter and easily
reaches across cell or compaction boundaries.
• By computing influence functions for diffraction limited
proximity Z1, Defocus Z4 and Coma Z7 it is possible to
quickly assess image changes through the process window
and along a scanner slit.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
28
Standard Cell Interactions: Approach
Boundary
Lynn Wang
Adjacent cells
increase Match
Factors and hence
variation through
focus
MF = 0.3
Cell i
Copyright 2007, Regents of University of California
Cell j
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
29
Standard Cell Interactions: Accuracy
Delta L
delta L vs. PMF
90 nm Defocus
-0.3
-0.2
0.03
0.025
0.02
0.015
0.01
0.005
0
-0.1 -0.005 0
-0.01
-0.015
0.1
R2 = 0.7131
0.2
PMF
Copyright 2007, Regents of University of California
0.3
0.4
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
30
Standard Cell Interactions: Results
Cell Interaction Chracterization
90 nm Library: Metal Layer 1 Defocus
Cell 1 Cell 3
Cell 2 Cell 3
Cell 3 Cell 3
Cell 1 Cell 2
Cell 2 Cell 2
Cell 3 Cell 2
Cell 1 Cell 1
Cell 2 Cell 1
Cell 3 Cell 1
0.3
0.28
0.24
0.22
0.2
0.18
0
1
2
3
4
5
6
7
8
9
10
Cell Distance Characterization
90 nm: Metal Layer 1 Defocus
Cell #
0.35
0.3
Greatest Range= 0.1
Smallest Range= 0.01
0.25
PMF
PMF
0.26
0.2
0.15
0.1
0.05
0
0
0.2
0.4
0.6
Distance (um)
Copyright 2007, Regents of University of California
0.8
1
1.2
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
31
CMP Variation Assessment: Physical Model
STI process simulation: HDPCVD and CMP
HDPCVD
Jihong Choi Dave Dornfeld
Characterization:
Cell 1
Cell 2
HDPCVD
CMP
Cell 3
Topography map
z
z2
z1
CMP
α

 ( x, y )
d
z pad
z3
Pad Asperity
z ( x, y )
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
32
CMP Variation Assessment: Chip Model
Chip Layout
Pattern density
Line width
Line space
HDP-CVD Deposition Model
CMP Input Thickness
CMP model
Evolution
Copyright 2007, Regents of University of California
Nitride thinning
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
33
Etch Variation Assessment:
Physical Model
UCLA
Jane Chang John Hoang
High density plasma, with O2 in Cl2
Low DC ratio
High DC ratio
161
264
275
164
86.9º
136
189
70.5º
247
208
82.5º
293
137
226
85.6º
270
130
72.9º
74.1º
164
High density plasma, O2 in Cl2, low DC ratio
Low substrate
High substrate
bias
bias
89
363
161
264
300
86.9º
136
189
33.7º
181
69.8º
247
195
250
82.5º
293
87.7º
320
130
74.1º
71.4º
137
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
34
Etch Variation Assessment: Layout
Chip Field Layout
Etcher
Iouter
Iinner
Ws
Coil Power
Ws
Cl2
N2
O2
Pressure
Substrate Bias
 Couple
W
feature-profile simulations with tool-scale
models and plasma ion energy models (Graves
and Lieberman at UCB)
 Identify factors in profile model that have feature
level and pattern density dependencies
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
35
Interconnect Variation Assessment: Concept
Major Physical Contributors to Variation:
- Lithography (Focus, Overlay, Aberrations, …)
- CMP (Density, …)
- Etch (Sidewall Angle, …)
Delay Variation = f(local layout,
layout in layers above and below, die
location, wafer position)
Key idea: Predict interconnect delay
variations by tracing Pattern
Matches through circuits and
adjusting extracted RCs.
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
36
Interconnect Variation Assessment: System
Pattern Matcher
Parasitic Extraction
Predict Geometrical Variations
Netlist Backannotation
Estimate Changes in R, C
Timing Analysis
Copyright 2007, Regents of University of California
Fast-CAD Techniques:
1) Focus solely on
critical paths to
improve runtime
2) Pre-characterize
libraries to model
geometrical
variations for
different match
factors.
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
37
Interconnect Variation Assessment: Result
Structure: Array Above Ground Plane
Resistance Calculation:
L
R
WH
 eff  3  cm
110nm
110nm
M1
175nm
Capacitance Extraction:
450nm
ILD0 κ=3.4
Synopsys Raphael: 2D Field Solver
C total Cbottomgp  2Ccoupling
Nominal
CMP Erosion
Linewidth (nm)
110
98
122
110
98
122
Values from ITRS 90nm Technology Node
GP
Delay of a 1mm M1 interconnect (array) Pitch=220nm
Space (nm) Height (nm) Ct (fF/mm) R (Ohm/mm) Delay (ps) Error
110
175
180.724
1558.44
97.17 0.00%
122
175
166.61
1749.27
100.55 3.48%
98
175
197.65
1405.15
95.82 -1.39%
110
158
171.39
1726.12
102.06 5.04%
122
158
158.185
1937.48
105.74 8.82%
98
158
187.181
1556.34
100.50 3.43%
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
38
Cross-Talks Variation Assessment: Concept
Particle
AGGRESSOR
AGGRESSOR
VICTIM
VICTIM
AGGRESSOR
AGGRESSOR
< Designed Layout >
Jae-Seok Yang
Driver variation
- Non-linear rectangle poly(Litho)
- Spatial correlation (Drivers are close)
Focus correlation
Interconnect variation
-Width/Spacing(Litho, Etch)
-Height(CMP)
CD(Victim Driver)
distance
Dose correlation
- CD(poly)
+ CD(Poly)
Target
window
1
Crosstalk verification
over the process window
focus
1
focus
Correlation factor(β) Correlation factor (α)
< Manufactured Layout >
- CD
+ CD
Target
window
CD(Aggressor Driver) is bound
dose
distance
Copyright 2007, Regents of University of California
dose
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
39
Cross-Talk Variation Assessment: Result
65nm
Vdd=1.2V
Peak Noise over the process windows (mV)
AGGRESSOR
100nm
VICTIM
AGGRESSOR
Defocus: -40nm
Defocus: 0nm
Defocus: 40nm
Defocus: 100nm
Dose: 10%
243.1
237.2
238.3
234.9
Dose: 0%
242.3
237.5
234.4
231.2
Dose: -10%
234.8
233.5
232.7
230.8
100um
5.3% noise variation
over the process window
Proposed flow for variation aware SI verification
Layout (Full-chip)
OPC / Arerial image sim.
RC extraction
Xtalk analysis
critical condidate nets
in terms of Xtalk noise
Real critical nets
for DOF/dose margin
Repeat for the next ED condition
P&R
Arerial image sim.
( poly )
ED constraints considering
spatial correlation
for poly layer
Arerial image sim.
( metal )
ED constraints for metal
RC extraction
Xtalk analysis
Failure criterion
No
Failure
Save as a critical net
Yes
Copyright 2007, Regents of University of California
Repeat for the next
critical condidate nets
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
40
Photomask Edge Effect: Physics & Impact
Form of
Cherenkov
radiation
=193nm
Ey (TE)
n=1
Difference field =
actual – vertical
propagation field




Air
39.7o n=1.563
Front
Glass
Intensity imbalance IEDM 1992 Alfred Wong
Cherenkov radiation SPIE 2001 Costas Adam
Domain decomposition (edge sources) SPIE 2002 Costas Adam
PSM only 180 degrees for one pitch BACUS 2006 Gleason
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
41
Photomask Edge Effect: Characterization
Real
Imy
CER
CEI
Marshal
Miller
Sqrt(I2)
0.01



Koji
Kikuchi
Dan
Ceperle
y
Imy
Both Real and Imy Edge Effects
Real Duty Cycle
50%
Plot Sqrt(I) vs Duty Cycle
Adjusting Phase Etch Depth adds Purple to cancel Yellow
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
42
Photomask Edge Effect: Characterization
Sqrt(I2)
0.01
Imy
50%
Real
3
5
5 5 deg
10
Copyright 2007, Regents of University of California
Duty Cycle
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
43
CEI
CER
Photomask Edge Effect: Characterization
TM
Cross Talk
TE
Period in 
Plot CER and CER vs mask period
 Large periods give edge parameters
 Small periods show cross-talk effects

Copyright 2007, Regents of University of California
Period in 
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
44
EUV Defect/Mask Interactions: Modeling
Absorber Pattern
Specifications
Incident
Wave
Absorber Layout
Simulator
Defect and Multilayer
Specifications
FT{
Near
Field
Multilayer
Simulator
}
Plane Wave
Near Field
FT{ Near
Field }
Absorber Layout
Simulator
Final Result
RADICAL:
Rapid Absorber Defect Interaction
Computations for Advanced Lithography
Mask with Buried Defect
Refractive Indices of Simulation Domain
Resulting
Image
Plot of AerialAerial
Image for NA:
0.45
0.7
1
0
Chris Clifford
Defect
No Defect
0.99
50
0.6
0.98
100
Defect
0.5
0.97
Y-Axis (nm)
0.96
200
0.95
250
0.94
Intensity (E2)
150
0.4
0.3
300
0.93
350
0.2
0.92
0.1
400
0.91
450
0
50
100
150
200
250
300
X-axis (nm)
350
400
450
500
0.9
0
160
180
Copyright 2007, Regents of University of California
200
220
Distance (nm)
240
260
280
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
45
EUV Defect/Mask Interactions: Assessment
 Defect


Projector for DFM
Characterize buried defect and mask pattern separately
Use fast methods to determine:


Effect of defect for various layouts
Layout mask blank interactions
Characterize
Defect
Defect Library
Ray Tracing
Multilayer Simulator
Possible Layouts
and Defect
Locations
Fast Interaction
Simulator
Based on RADICAL
Copyright 2007, Regents of University of California
Severity of
Pattern
Variation
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
46
RTL
Synthesis
Prototyping
Physical
Synthesis
Nets/Paths
Routing
Model Resolution
Conclusion: Nature of DfM
Optimization
Regions
Sign-off

DfM requires






complexity management (beyond human comprehension)
new modes and multiple viewpoints for collaboration to integrate
process, device and circuit
very fast first-cut accurate models
complete scope across processes and their interactions
circuit performance assessment (power, delay, cross-talk)
Designers add complexity management skills to process and
device understanding and should be invited to collaborate.
Thanks to DARPA, SRC, Industry and U.C. Discovery
Copyright 2007, Regents of University of California
FLCC Pre-Presentation of SPIE DFM-PI 07 ARN
47
Conclusion: Strategies & Prototypes
Module 1
Processing
BSIM model
Module 3
Circuit



Standardized
BSIM Model
Output
Module 2
Device
Many nonidealities of manufacturing can be moved to the mask
plane and visualized/quantified early.
Pattern Matching and Perturbation Modeling have both
exceptional speed and adequate accuracy
Prototype DfM tools and methodologies were shown for


Parametric Yield Simulation process/device/circuits,
Visualization/quantification at the mask level
Thanks to DARPA, SRC, Industry and U.C. Discovery
Copyright 2007, Regents of University of California