1
1/14/04
1
UCB-TCAD
Optical Projection Printing: Methtodologies, Modeling and Monitoring
UCB TCAD Overview Talk Spring 2004Andy Neureuther, Yunfei Deng, Lei Yuan, Frank Gennari,
Garth Robins, Michael Lam, Scott Hafeman, Greg McIntyre, Dan Ceperley, and Jacob Poppe
UC BerkeleySupported by the Lithography Network DARPA/SRCby industry through the State of California UC-SMART and UC-DiscoveryAdvanced Energy, ASML, Atmel Corp., Advanced Micro Devices, Applied Materials, Asyst Technologies Inc., Cadence, Canon, Cymer, DuPont, Ebara, EVG, Intel Corporation, KLA-TENCOR, Mentor Graphics, Mykrolis Corp., Nikon Research Corp., Novellus Systems Inc., Panoramic Technologies, Photronics, Synopsys, and Tokyo Electron,by Intel EUV Mask Simulation, andby JPL Terrestrial Planet Finding
1/14/04
2
UCB-TCAD
Simulation Discovered Phenomena: Intensity Imbalance in Phase Shifting Masks
Discovered by simulation and verified experimentally, IEDM, 1992 Wong
∆
2
1/14/04
3
UCB-TCAD
Simulation Discovered Theory: InterferometricPattern-and-Probe Monitors
Defocus target
0 Aberration 0.05 waves Defocus 0.05 waves Spherical
Spherical targetsensitivityorthogonality
sensitivityorthogonality
(λ/NA)
(λ/NA)
• 0.4 λ/NA x 0.4 λ/NA probe; red = 180º, yellow = 0º, green = 90º
1/14/04
4
UCB-TCAD
UCB TCAD Research Themes
• Fast EM Analysis methods to attain speeds required for EUV, OPC and die-to-database inspection
• Photomasks as precision instruments for monitoring projection printing
• Linking Process and EDA through multi-student test structure design, pattern-matching and experiment
• Physical models and algorithms for emerging tools and treatments
• Actively apply simulation to guide innovation and characterization
3
1/14/04
5
UCB-TCAD
Outline
• Methodologies (Slides 5-29)– Basic optical physical principles– Resolution enhancement
• Modeling (Slides 30-40)– Immersion as a two-for in a sea of waves– Resist modeling as the proverbial bottleneck– Rigorous EM as the sawmill for EUV and PSM– Fast EM analysis methods for OPC
• Monitoring (Slides 41-57)– Phase-shifting masks as precision instruments– Linking Process and Electronic Design
1/14/04
6
UCB-TCAD
Optical Projection Printing Parameters
Wavelength λ = 248 nm
Partial Coherence Factor σ = (NAc/NAo) = 0.3
Numerical Aperture NA = sin (θ) = 0.5
4
1/14/04
7
UCB-TCAD
Resolution in Projection Printing
NAdf
df λ
λλ 61.0
2
61.022.1 =
=
Minimum separation of a star to be visible.
f = focal distanced = lens diameter
Point spread function
Null position
PDG Fig. Ch 5
1/14/04
8
UCB-TCAD
Resolution ~ Transverse Variation
λ = 248 nm
λTRANS = λ/sinφ= 3.22λ = 800nm
φ
Wave graphic by OngiEnglander and Kien Lam
5
1/14/04
9
UCB-TCAD
Depth of Focus: Phase change on vertical axis
Plane of Best Focus
4.75λ
5.0λ
Plane of Rayleigh l/4 Defocus
Observe phase along a vertical line
Wave graphic by OngiEnglander and Kien Lam
1/14/04
10
UCB-TCAD
Normalized Parameters
NAkLLINEWIDTH
λ1=
( )22 2 NAkDOF λ
=
λ = 365, 248,193, 157, 13.4 nmNA = 0.167, 0.38, 0.5, 0.63, 0.7, 0.75, 0.80
Instead of recalculation for every new combination of λ and NA a universal catalog of image behavior can be utilized if we first determine the k1 and k2factors in the actual system for the linewidth and defocus and look up results in a data based based on λ = 0.5µm and NA =0.5.
mkmkNA
kLLINEWIDTH µµλ
111 5.05.0
===
( ) ( )mkmk
NAkDOF µ
µλ22222 5.02
5.02
===
For any wavelength λ and numerical aperture NA.
6
1/14/04
11
UCB-TCAD
Bragg Condition for Off-Axis Illumination
PP=L+S
L S
Chrome
Quartz
λ
φn Ray of Light
Wavefronts
φINC_R
λ
d2
d1
d1 + d2 = nλ
Psin(φINC_R) + Psin(φn) = nλ
sin(φn) = nλ/P − sin(φINC_R) for angles from the right
sin(φn) = nλ/P + sin(φINC_L) for angles from the left
1/14/04
12
UCB-TCAD
(Sinθx, Sinθx) wave accounting system
SinθX
SinθY
1.0
Location (Sinθx, Sinθx)corresponds to a ray making angles θx and θY with the downward z-axis
NA
Lens PupilSinθMAX = NA
σNA
Lens IlluminationSinθMAX = σNA
7
1/14/04
13
UCB-TCAD
Pupil Wave Traffic: Partial Coherence
SinθX
SinθY
1.0
NA
Lens PupilSinθMAX = NA
Cone of Incident Light
Diffracted Orders from a mask with period P
+20-2 -1 +1
Some misses pupil
Potential for entering the pupil
1/14/04
14
UCB-TCAD
Electric Field and Intensity: Plane Waves
( ) ( ) ∑∑ ⋅−+− ==n
xkjn
n
znkxkjnTOTAL
nn eEeEE)()cossin( 00 θθ
zzxxx ˆˆ +=Spatial position vector
( ) ( )zkxkzkxkk nnzxn ˆcosˆsinˆˆ 00 θθ +=+=Propagation (k) vector
where k0 = (2π/P)
Genralizes to y-direction
( ) ( ) ( ) )cos22(1
)cos220(0
)cos22(1
101 zxPjzx
PjzxPj
TOTAL eEeEeEE +− +−+
+⋅−+−−− ++=
θλππ
θλππ
θλππ
Three wave case for on-axis illumination of mask with period P
( )2
1cos
−=Pn
nλ
θ( )n
nλ
θ =sin
8
1/14/04
15
UCB-TCAD
Illumination Controls Proximity EffectsTwo Pinholes in a mask
• Wafer• Tails of electric field overlap (spillover)• Relative phases depend on phase of illumination
2221
21 2 EEEEITOTAL ++= 2
22121 2 EEEEITOTAL +−=
0180Normal Off-axis
1/14/04
16
UCB-TCAD
Mutual Coherence Function: Top Hat
9
1/14/04
17
UCB-TCAD
Aerial Image Intensity for Knife Edge
σ = ∞
σ = 0.6
Toe
Slope
Overshoot
1/14/04
18
UCB-TCAD
Feature Type Effect with Defocus
Contrast for Dense = 0.75
Dense L = SSpace
Line
0.99
0.11 ( )( ) 80.0
11.098.011.098.0
=+−
=DENSEC
DOF = k2 = 1 µm
Isofocal points
10
1/14/04
19
UCB-TCAD
Superposition Failure for Partially Coherent Images! (add Electric-Field instead)
0.3λ/NA 0.8λ/NA0.5λ/NA
Mask linewidth percentage errors are twice as large on the wafer. Thus the mask enhancement factor (MEF) is due to partial coherence!
Much taller and wider.
Peak intensity initially increases as the square of linewidth.
1/14/04
20
UCB-TCAD
High-Fidelity Audio System Off-Axis Analogy
• The lateral spatial variation across a wafer of an off-axis light ray is analogous the temporal variation of a note in a Hi-Fi audio system.
• More rapid variations (spatial or temporal) from higher frequencies (spatial or temporal) allow sharper artifacts(spatial {lithography feature} or temporal {drum beat}) to be produced.
• Just as it is difficult to improve upon the pulse width times bandwidth product it is difficult to improve upon the feature size times NA product.
• BUT IN RESOLUTION ENHANCEMENT WE TRY TO GET A FACTOR OF TWO INCREASE ANYWAY.
11
1/14/04
21
UCB-TCAD
Frequency0 Frequency0
Frequency0
Frequency0
Frequency0
Frequency0
ModifiedIllumination
PhaseMask
In-LensFilter
ConventionalIllumination
BinaryMask
LensCapture
Resolution Enhancement Emphasizes High Frequencies
Resolution Enhancement Techniques
Bokor, Neureuther, Oldham, Circuits and Devices, 1996
1/14/04
22
UCB-TCAD
Two Ray Infinite DOF
θ1 θ2
Ray # 1 Ray
# 2
TransversekPitchPeriod
∆==
π2
( )( )
NAPitch NA
2sin2sin λ
θλ θ →= =
( )θsin2 0kkTransverse =∆
When θ1 =θ2 the contributions from Ray #1 and Ray #2 track each exactly with axial distance and an INFINITE depth of focus is produced.
kx
ky
Doubled Resolution! With infinite DOF
12
1/14/04
23
UCB-TCAD
Strategy to improve both resolution and DOF
• Since the small features or edges are are most important emphasize the high-frequency off-axis ray to improve resolution.
• Since the change in the image with focus comes from the relative phase change among the rays with axial distance, utilize rays at similar azimuthal angles that track in phase with focus to improve DOF.
NAPitchSPOT ⋅=
λσ
1/14/04
24
UCB-TCAD
Illumination Schemes
• The k1 factor is inversely proportional to the lateral separation of the illumination k1 = 1/(2 x separation)
Top Hat – General Shapesk1 = 0.67k2 = 1.3
Annular – DOF, Contactsk1 = 0.55k2 = 1.7
Dipole – V lines, DOFk1 = 0.35k2 = 3.0
Quadruple – H,V lines, DOFk1 = 0.45k2 = 2.0 H lines and
contacts formed via a double exposure
1-1
1-1 1-1
1-1
Top Hat – General Shapesk1 = 0.67k2 = 1.3
Annular – DOF, Contactsk1 = 0.55k2 = 1.7
Dipole – V lines, DOFk1 = 0.35k2 = 3.0
Quadruple – H,V lines, DOFk1 = 0.45k2 = 2.0 H lines and
contacts formed via a double exposure
Pupil
Pupil Pupil
Pupil
σIN = 0.55
σOUT = 0.85
13
1/14/04
25
UCB-TCAD
Phase-Shifting Mask
Sheats and Smith
P P
P P/2
1/14/04
26
UCB-TCAD
Phase-Shifting Mask
Sheats and Smith
14
1/14/04
27
UCB-TCAD
Phase Defects May Print Worse Out of Focus
Wantanabe, et al.
Severity Factor [1-Mcos(φ)] becomes [2] when phase of defocus is included.
M = E field transmission
amplitude
1/14/04
28
UCB-TCAD
Resolution Enhancement: In-Lens Filter
Fukuda JVST B Nov/Dec 91
( ) rjrj eer22 222 5.05.02cos πβπβπβ −+=
Defocus away and toward the lens.
• The cos(2πβr2) filter creates dual defocused images that are very effective in increasing the total focal range of contact patterns.
15
1/14/04
29
UCB-TCAD
Resolution and Focus Trends
k1 = 0.61
k2 = 1.0
Rayleighcriteria
Trend with modified illumination and resists
Trend with resolution enhancement techniques
1/14/04
30
UCB-TCAD
Vector Addition at High NAParallel Orientation Perpendicular Orientation
EtotalEtotal
M. Lam
16
1/14/04
31
UCB-TCAD
SPLAT 6.0 for High-NA and Immersion• Uses vector or
scalar TCC• Models high-NA
and immersion• Exploits some
symmetry in vector TCC calculations
• In agreement with high-NA models by Flagello and Adam
t
Projection Optics
Mask Plane (m)
Entr. Pupil Plane (p)
Wafer Plane (w)
Image Plane in Resist (R)
Efield x, y orientations
Scattered Orders
Radial,SagittalPupil position
Location(subscript)
Variables
Aberrations
t||, ts, Ray tiltObliquity factor
StandingwaveExR, EyR, EzR
STE(Ow), STM(Ow)
E xR=
E yR=
E zR=
||tE rp
||tE rp
||tE rp
)( wTMS θ
)( wTMS θ
)( wTMS θ
Rθcos
Rθcos
Rθsin
pφcos
pφsin
)( wTES θ
)( wTES θ
pφsin
pφcos
⊥tE pφ
⊥tE pφ
-
+TM polarization problemLower maximumHigher minimum S. Hafeman
1/14/04
32
UCB-TCAD
B.J. Lin Sept. 02
17
1/14/04
33
UCB-TCAD
B.J. Lin, 157 Workshop, Sept. 02
1/14/04
34
UCB-TCAD
Immersion Lithography
• Promise – Improve resolution of 193 to that of 157 using a
lower NA (0.9 => 0.77) and a slight increase (1.08) in DOF.
• Industry View – Gives 193 the punch of 157 without the
complications of 157 and cheaper to explore
• Issues– Liquid (optics), liquid (resist), liquid (machine)
18
1/14/04
35
UCB-TCAD
Comparison of Imaging for 193 nm Immersion and 157nm air
• Cut lines taken at 50 nm into the resist stack• 193 nm immersion shows better field coupling into resist
0.7 lambda/NA Contact,NAair=0.85, 300 nm resist film n=1.72
Air @157nm
0.00
0.20
0.40
0.60
0.80
-1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1Position (lambda/NA)
Inte
nsity
(A.U
.)
Scalar Vector
Water @193nm
0.00
0.20
0.40
0.60
0.80
1.00
-1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1Position (lambda/NA)
Inte
nsity
(A.U
.)
Scalar Vector
S. Hafeman
1/14/04
36
UCB-TCAD
Simulation of Inhomogeneous Liquid Effects
• Several models for n(x,y,z) are algebraically tractable• PACIFIC: Add-on to SPLAT 6.0 that determines an
equivalent pupil map describing the fluid• Used to develop goals for liquid purity and fluid flow
CDF Simulation (Wisconsin) PACIFIC (Image Cone)
Pupil Map
S. Hafeman
19
1/14/04
37
UCB-TCAD
Simulation Extraction of Resist Parameters in 2D
Reaction enhanced diffusion
Reaction reduced diffusion
The sequentially double exposed corner shape determines the type of acid diffusion and enables 2D pattern prediction. STORM simulation
Sequentially double exposed cross
L. Yuan
1/14/04
38
UCB-TCAD
Fickean diffusion Enhanced non-Fickean
Reduced non-Fickean SEM of APEX-E
Deprotection level of one quarter double exposed cross (STORM simulation and SEM)
L. Yuan
20
1/14/04
39
UCB-TCAD
Reaction Reduced Diffusion is best 2D Predictor for APEX-E and UV210
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0.2 0.3 0.4 0.5 0.6 0.7
Designed trench width
Diag
onal
cro
ss s
pace
(u
m)
ExperimentalFickeanReducedEnhanced
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.3
0.4
0.4
0.4
0.5
0.5
0.5
0.5
0.5
0.5
0.1
0.1
0.1
0.2
0.2
0.2
0.3
0.3
0.3
0.4
0.4
0.4
0.5
Fickean
Reaction Reduced Diffusion
• Only Reaction Reduced Diffusion predicts corner-to-corner spacing correctly
• Only Reaction Reduced Diffusion predicts standing wave removal correctly
• Does Reaction Reduced Diffusion reduce line-edge roughness? L. Yuan
1/14/04
40
UCB-TCAD
Electric Field Enhanced PEB vs. Standard:Improved Focus Latitude
5.3mJ, +0.6um defocus, 500/500nm L/S
Standard PEB, none of the trenches cleared
EFE-PEB, downward E-field of ~200,000 V/cm
•Apex-E resist with Shipley’s RTC top coat, an ASML KrF stepper, NA 0.5, σ 0.2, dose 4.5~5.5mJ/cm2, spin 3000rpm, SB 100oC, 60sec, PEB 90oC, 60sec, resist thickness 900nm.
J. Poppe
21
1/14/04
41
UCB-TCAD
TEMPEST FDTD Simulator
Instantaneous Electric Fields
Maxwell’s Equations on a Staggered Grid
15 nodes/λ;
50 bytes/node;
30 cycles
1/14/04
42
UCB-TCAD
n=1 n=1.563
λ=193nm
39.7o
Ey (TE)
ScatterningScatterning from the Phasefrom the Phase--Well CornerWell Corner
Front
Results in crosstalk between trenches. K. Adam
22
1/14/04
43
UCB-TCAD
µm
µm
µm
µm
TE : Ey polarization
TM : Ex polarization
CDSB
x-axis
z-ax
is
Incident radiation
y-axis
CDtarget=130nmMag=4Xλ=193nm
|Ey|
|Ex|
Adam, SPIE 4000-72
With SB
Defocus
Aer
ial I
mag
e C
D
Scatterbarsimprove DOF
Scattering Bar Simulation with TEMPEST
1/14/04
44
UCB-TCAD
TEMPEST EM simulation of EUV multilayer masksMultilayer reflecting bilayers, N: 20Multilayer e-Beam affected bilayers, N: 20Period 1 (0°, R~62%): 6.938nmPeriod 2 (540°, R~4%): 6.312nmEdge transition: slow, sigma=10nm,
sharp, sigma=1nm
e-Beam
Period 2
Period 1Phase Well Depth
Geometry for attPSM (540°, 0°, 540°, 0°, 540°)
X position on Mask (nm)
Z Z
Electrical Field Intensity distribution
X position on Mask (nm)
0 °
540 °0 °
540 °540 °Y. Deng
23
1/14/04
45
UCB-TCAD
EUV Mask with 20,000,000 nodes as largest TEMPEST simulation to date
Coma target: center square at 180° (0.4*M*λ/NA), two rectangle sidebars at 0° & 180° (0.6*M*λ/NA) x (1.2*M*λ/NA)
Wavelength λ = 13.4nm, NA=0.25, M=4, σ=0.3Coma Target: Geometry X-Z cut-plane
X position on Mask (nm)
Z
Cr 100nm
0.6*M*λ/NA
1.2*M*λ/NA
0.4*M*λ/NA
X at wafer (nm)Yunfei Deng SPIE 2002
1/14/04
46
UCB-TCAD
How can you “rigorously” simulate this pattern?
> ~ 10µm
> ~
10µm
157nm < λ < 248nm
4000 λ2 − 1600 λ2
~ 1µm height26,000 λ3 − 6,500 λ3 !!
~ 0.4 – 0.8 billion nodes!
~ 13-26Gb required memory
~ 32bytes/node
K. Adam
24
1/14/04
47
UCB-TCAD
Fast-CAD: Edge-by-Edge Decomposition of Phase-Shifting Masks•Any Manhattan mask geometry can be broken into a sum of its isolated edge diffractions, minus the background.• Libararies of pre-simulated rigorous 2D edge diffractions from all types of edges (phase, underetch, etc.) can be stored and accessed to synthesize the nearfields of an arbitrary 2D or 3D structure.
80nm Cr
180deg50nm
x
z
yEy
x
y
z .
80nm Cr
180deg
50nm
Ey
Lx
Ly
x
yz.
clear polygon
Ey
edge shadow regions
unaffected field through Cr-layer
edges subject to TE(//) polarization
edges subject to TM( ) polarization
When is this accurate enough for imaging?K. Adam and M. Lam
1/14/04
48
UCB-TCAD
Pattern-and-Interferometric-Probe Aberration Monitors
0
0.2
0.4
0.6
0.8
1
1.2
0 50 100 150 200x-position in field (pixels)
Inte
nsity
(100
% C
F)
Probe Position
Discovered through simulation
Becoming a practical technologyLead to a new theory
Defocus targetExperiment on AIMS at low NA looks just like simulation!
Defocus Spherical HO Spherical
(λ/NA) (λ/NA)(λ/NA)
(λ/NA) (λ/NA)
Coma HO Coma
Mask phases• yellow = 0°• green = 90°• red = 180°
G. Robins
25
1/14/04
49
UCB-TCAD
Reading Strategy: Peak vs. Diameter
• Peak is easier to read• Trend across die is clear ~100nm• 25nm focus accuracy ~ 1/8 RU
Across die
∆X = 1800µm ∆f = 200nm ~ 1RU
λ = 248 nm, NA = 0.8
G. Robins
1/14/04
50
UCB-TCAD
Coma Target Results
• Performance issuesDo not see the expected rings0° regions are significantly brighter than the 180° regions
• Possible reasonsMaxwell Demon at mask: electromagnetic intensity imbalanceVector addition Demon at image: high-NAMask making Demon: phase etch depth and tolerancesWafer Demon: wafer thickness variation
λ = 248 nm, NA = 0.70
0o
G. Robins
26
1/14/04
51
UCB-TCAD
PSM as Precision Instruments for Illumination and Mask Making
Concentric spillover for measuring dipole imbalance
0.05
0.10
0.15
-20% -15% -10% -5% 0%
90 probe 0 probe
Transmission error of 180° region
Example Alt-PSM Target:Probe intensity vs. transmission
0° probe90° probeChrome
A(0°)
B(180°)
0.10
0.15
0.20
0 5 10
90 probe 0 probe
Phase error of 180° region (deg)
Example Alt-PSM Target:Probe intensity vs. phase error
Inte
nsity
(cle
ar fi
eld)
Inte
nsity
(cle
ar fi
eld)
Phase TransmissionG. McIntyre
1/14/04
52
UCB-TCAD
Linking Process and EDAthrough Pattern Matchng
Zernike.txt
IFT
PatternMatcher
SPLAT
MaskLayout
Pattern(coma)
MatchLocation(s)
Aerial Image SimulatorRule-based speed at accuracy of
model based approaches Goal: real-time OPC and die-to-
database comparison for
Concept:Given a residual process effect find a test pattern with the maximal lateral impactThen identify worst-case impact locations in a layout using the degree of similarity of the local layout
http://cuervo.eecs.berkeley.edu/Volcano/F. Gennari
27
1/14/04
53
UCB-TCAD
Layout Locations Impacted by Process Residuals
0/180 phase chip layout with top 100 matches for each of 12 patterns
Close-up of coma match location with match factor of 0.374
All locations along edges and corners of layout are searched fordegree of similarity to match patternPattern matching is faster than an OPC iterationMatch time is 1.5 hrs on 1GHz PIII for all edges and corners on417mm2 mask layout with 35 million shapes – 2.6 billion test points
3mm x 3mm400,000 rects1200 results
F. Gennari
1/14/04
54
UCB-TCAD
Examples of Maximal Lateral Impact Functions
Aberrations
PSM
Reflective Notching
Reflection from Slope
Laser Assisted Thermal Processing
Thermal Conduction
Defects
Image
CMP Dishing
Procedural based on distances and density
Large Areas
Alignment
Multilayer
F. Gennari
28
1/14/04
55
UCB-TCAD
Test Mask: Self-Testing Phase and Intensity
Probe size: 1um
(A) 0-180,90 : 0-180 patterns with 90 probeG. McIntyre
1/14/04
56
UCB-TCAD
Multi-Student Process-EDA Test-MaskEdward Hwang CMP Garth Robins AberrationGreg McIntyre Illumination, PSM phase error, plasma etch Jason Cain MetrologyJihong Choi CMPLing Wang CMP
• Point of Contact: Greg McIntyre, EECS, UC Berkeleyoffice (510)642-8897, [email protected]
• Mask info: All dimensions are for mask in um (unless otherwise noted) ; Wavelength = 248nm; 4 phase etches (0,90,180,270); Main field size = 105 x 105 mm (noted by border); Dark field mask; Alignment markers are for ASML PAS 5500/90 (our microlab stepper); Will also be used on ASML: 4x .63na , 4x .7na, & 4x .8na
G. McIntyre
29
1/14/04
57
UCB-TCAD
Future Devices: Mass Manufacturing Needs
Photonic Crystal WaveguideIssues: Performance, Tolerances, Inspection
Air background, εr=3.376, pitch=0.5 µm, r=0.1 µm, λ=1.55 µm, TE
TCAD => nano-CADRefractive Index Map of holes in silicon forming an optical stop-band.
Plane Wave excitationPath into and along which the light propagates.
D. Ceperley
1/14/04
58
UCB-TCAD
Summary
The ability to model lithography tools and treatments is playing a very important role
• in guiding innovation in future lithography systems, • in application of automatic compensation to layout
designs, and • in creation of strategies for manufacturing control.
TCAD is important for many phases of advanced lithography and especially mask issues