Transcript Reticle Haze Detection

http://www.TEAsystems.com
Haze Aberration Detection using Weir PW
 Haze aberration has been shown to result in side-lobe formation next to
the immediate edge.
 Side-lobe formation results not from the mathematical Gibbs
Phenomenon description but rather from the physical introduction of
Spherical Aberration and other perturbations into the optical train
 The reticle is an intimate part of the optical system of the scanner.
 Haze introduces scatter and aberrations that:




Result in image perturbations that reach way beyond the immediate deposition
Influence adjacent features
Result in perturbations even when they do not touch the feature.
First degrade the Bossung Response of the reticle image and eventually the
dose required to create an on-size image.
Reticle Haze
Haze formation on
feature edges does
directly influence the
edge, however the effects
are more far-reaching.
Critical Reticle Haze
Directly influences this features definition
through side-lobe ringing artifacts (Gibbs
Phenomenon) that distort the square wave of
the chrome or PSM edge http://www.TEAsystems.com - 2
Haze
TEA Systems
Yield Enhancement thru Modeling
Feature Edge Effects
Three graphic instances of reticle-face haze formation are shown in this illustration.
• Feature Edge formations
• Most likely to form since they represent they form on high-energy initiation seed sites
where edge scatter and depositions will first adhere to the mask. Previous
presentation illustrated the influence based on the Gibbs effect that results in squarewave degradation by image “ringing” resulting in edge-intensity over/under shoot
and in side-lobe generation
On Obscuration Feature Haze Formation
• Clear area Haze
Influence on Feature profile?
• Not addressed as a factor
• On-Feature Haze formation
• Not addressed as a factor
Critical Reticle Haze
Directly influences this features
definition through side-lobe ringing
artifacts (Gibbs Phenomenon) that
distort the square wave of the chrome
or PSM edge
http://www.TEAsystems.com - 3
Haze
Influence on Feature Response?
TEA Systems
Yield Enhancement thru Modeling
Seed sites for Haze
 Haze deposition first forms on high-energy areas known as seedsites
 Seed sites are not singularities forming at one or two isolated
points
 Haze initiation will form across an extended area of the mask
surface




Formation is a function of the interaction of the feature loading
localized optical wavefront characteristics (lens edge verses center)
Wavelength of illumination
Local surface contaminants on the mask
 From manufacture
 Cleaning
 Also tend to located on high-energy feature edges
 Areas where edges are undercut or chrome is thinned from etch.
 Areas of unequal etch or PSM feature thickness that result in nonoptimum wave extinction during phase shifting.
http://www.TEAsystems.com - 4
TEA Systems
Yield Enhancement thru Modeling
Knife-edge optical effects
To see the effects of haze formation at a feature edge consider the opticians knife-edge.
The thin chrome obscuration acts as a knife edge discontinuity. Knife edge analyses have
been used for years in optics development because they allow the aberrations of the lens
to be accurately measured.
The Profile at the edge of the knife edge is NOT a pure Dirac step function as assumed in
the Gibbs Phenomenon. It is a complex Intensity gradient that incorporates the basic
“Gibbs Phenomenon” artifact plus a stronger variance caused by optics- limited
distortions, scatter and localized changes in the effective Numeric Aperture caused by the
finite edge. This results in an intensity profile that behaves like a Gibb’s function but is
actually stronger in intensity than described in the original presentation.
The Chrome is not a true knife edge in that it’s thickness is actually many wavelengths
across. The thickness therefore directly compounds profile changes by polarization and
coherence perturbations.
Intensity Profile
Chrome
http://www.TEAsystems.com - 5
TEA Systems
Yield Enhancement thru Modeling
Scatter effects
The chrome feature image is further complicated in that it is supported by a quartz
substrate. The wavefront at the feature edge therefore encounters a change in the index
of refraction (Quartz-to-Air) at the same time that it encounters the chrome feature
obscuration.
The index change at the edge results in scatter and this in turn reduces the edge resolution.
The effect also interacts with the image wavefront to induce localized aberrations.
AT&T, in 1982, was issued a series of patents for glass photomask coverplates to protect the
chrome photomask elements. The coverplate interface to the mask incorporated an index
matching fluid to prevent this scatter and reflection interference. This patent also noted
the improvement in image resolution and depth-of-focus that resulted because the
chrome was now encapsulated in a continuous index of refraction and scatter was
eliminated.
scatter
Intensity Profile
Air
Chrome
Quartz
Wavefront http://www.TEAsystems.com - 6
TEA Systems
Yield Enhancement thru Modeling
Open-area haze formation
Haze does not form randomly. It needs a high-energy seed-site.
Seed-sites therefore start areas containing:
• localized damaged from repair
• Undercut
• Impurities or localized stress in the quartz substrate
The resulting wavefront will be a convolution of the intensity profile across the hazed area
PLUS the chrome edge profiles from nearby features as far as 2 microns away PLUS the
scatter added by the chrome edge, haze edge and internal haze phase boundaries from
acrylic crystalline transitions.
The translucent haze area also behaves as a micro-lens and introduced refractive aberrations
that further interfere with the wavefront.
Summary: Isolated haze introduces wavefront distortions and aberrations that
influence nearby features.
Haze profile
Chrome edge profile
haze
http://www.TEAsystems.com - 7
chrome
TEA Systems
Yield Enhancement thru Modeling
Chrome-obscured haze
Chrome is not a complete blocker of the wavefront. It’s complex index of refraction results in a
portion of the electromagnetic wave that penetrates and thin film and interacts with the
overall image formation. In short, chrome is translucent even at deep-UV illumination.
The wavefront intensity and phase immediately above the chrome surface is not zero. A haze
element will react with the chrome causing thinning, cracking and other localized physical
reactions.
Scatter from other parts of the imaging layer will interact with this wavefront and also be
gathered by the lenticular behavior of the translucent haze. This results in localized
aberrations of near edge images not directly involved with the haze seed.
Summary: On
chrome haze has a smaller but still finite perturbing influence
on the wavefront that introduces aberrations and scatter.
http://www.TEAsystems.com - 8
TEA Systems
Yield Enhancement thru Modeling
Effects of Haze Seed Formation: Summary
 Haze does not form on isolated singularities
 Haze formation is a high-energy area effect.
 Haze does not have to be intimately associated with a feature edge
to influence image formation
 Early-formation isolated segments act as micro-lens elements
 On-feature surface haze influences overall scatter and dark-image
formation.
 The image of the photomask is converted to a frequency spectrum
at the entrance pupil of the lens. Scatter and aberrations from
haze change this spectrum and also change the influence of the
inherent lens aberrations on the image that results in large-area
image degradation
 All lenses retain finite coma and spherical aberration as balanced
aberrations tuned to the ideal photomask image.
http://www.TEAsystems.com - 9
TEA Systems
Yield Enhancement thru Modeling
Consider: Thirty years of process windows
Collapsed line
BCD
M
CD  
m 0
n
 ES  m
a
n0 nm 1  E  F
N
• J. Bossung, SPIE 1977 vol. 100
• This is a well established technique
Next few slides are from TEA Systems Class: “Lithograph Control and Characterization”
http://www.TEAsystems.com - 10
TEA Systems
Yield Enhancement thru Modeling
Review: Bossung curve analysis
 A = Isofocal Dose
 Dose at which feature size is
independent of Focus
B
Feature Size
 B = Locus of Best Focus
 “Best Focus” is located at the maxima
or minima of each dose curve
 The greater the curvature, the greater
will be the aberrations of the system
A
 UCL/LCL
UCL
 Upper and Lower Control Limits for
the process
 EL
EL
LCL
DoF
One curve for each dose
 Exposure Latitude or the dose range
over which the feature size lies between
the UCL and LCL
 DoF
 Depth of Focus or the focus range over
which the feature size lies between the
Focus
Zernike Analyses are a quantitative method of lens aberration analysis. UCL and LCL
Bossung curve characteristics can show the presence and effect of aberrations.
More strongly than dose reduction,
HAZE CAUSES
ABERRATIONS TEA
http://www.TEAsystems.com
- 11
Systems
Yield Enhancement thru Modeling
Typical Bossung Focus analysis for center-site
Field Layout with
measured-site shown in red.
J. W. Bossung,
SPIE (1977) Vol. 100
http://www.TEAsystems.com - 12
TEA Systems
Yield Enhancement thru Modeling
Sites from field center & 4 corners
Plotted sites in red
 An aberration free lens would result in exactly duplicated feature response.
 The scatter and aberrations caused by localized haze result in this phenomenon.
 More haze = more scatter + Image perturbations
http://www.TEAsystems.com - 13
TEA Systems
Yield Enhancement thru Modeling
Feature Size
Ideal focus/aberration response of features
 Aberration free features result in
 Linear feature size response to dose (blue line)
 An unchanging Best Focus response (flat) of the features for
changes in dose as shown here for the near-zero change in focus
with dose
http://www.TEAsystems.com - 14
TEA Systems
Yield Enhancement thru Modeling
Aberration influence on Feature response
Feature Size
Cd vs size curves
Dense Best Focus
Isolated Best Focus
 In this example of Contact response and proximity we see:
 Un-influenced (flat) Best Focus response for widely separated contacts (blue & black
lines)
 Aberrated Best Focus response (red line) for small, dense vias
 The onset of Haze introduces aberrations that can be seen much sooner than
simple dose change from neutral density obscuration effects.
http://www.TEAsystems.com - 15
TEA Systems
Yield Enhancement thru Modeling
Side Lobe
Cause: Aberration and scatter NOT dose-change
•In experimental SEMs, side lobes are seen inside a line (left) and
outside a trench (right )
•Figure shows experimental SEMs of side lobes for a line and trench
for 8% attPSM. Because the Gibb's phenomenon takes place on
either side of the discontinuity, the side lobes can be seen inside the
line and outside the trench.
http://www.TEAsystems.com - 16
TEA Systems
Yield Enhancement thru Modeling
Side Lobe
Notice the
perturbation of
CD and feature
profiles as a
result of the sidelobes resulting
from induced
aberrations
http://www.TEAsystems.com - 17
TEA Systems
Yield Enhancement thru Modeling
Influence of Spherical Aberration
Calculated with 3rd & 5th Orders
i-line rim PSM 0.35 um contact
hole SEM (overexposed)
10% attenuation PSM, 0.35 um hole
NA=0.5, DUV, s=0.3
From TEA Systems Class: “Lithograph Control and Characterization”
http://www.TEAsystems.com - 18
TEA Systems
Yield Enhancement thru Modeling
Side Lobe Formation
 The intensity of a side lobe increases with higher
transmission.
 However the stronger effect is the aberration of
the wavefront emanating from the local hazed
area
 Wavefront will influence both the immediate feature AND other
nearby features through the introduction of spherical aberrations into
the image.
http://www.TEAsystems.com - 19
TEA Systems
Yield Enhancement thru Modeling
Conclusions #1
 Previously Shown:
 Resist erosion is inevitable, however, when using att
PSM with higher transmissions.
 The constructive interference among the secondary
maxima of nearest neighbors increases the intensity
of side lobes.
 The worst situation is when the secondary maxima
of four neighboring contact holes interact at their
diagonal interaction and produce maximum intensity
regions.
 Now Recognize
 The overexposures shown previously do not
illustrate the effect of the haze on overall profile and
process response.
 Process aberrations will extend well beyond the
hazed area
http://www.TEAsystems.com - 20
TEA Systems
Yield Enhancement thru Modeling
Conclusions #2
 Weir PW:
 Aberrations are quickly discovered through the
Weir PW analysis of perturbation and feature
response uniformity across the full-wavefront
process window.
 The influence of the haze-induced aberration
directly upon the process-robustness of the reticle
feature can be directly measured using our
techniques.
 This technique will discover the onset of haze
formation very much earlier than reflectance or
transmission intensity monitors
 The following slides illustrate the Weir PW tools for
detection, identification and location of the influence
of Haze Formation.
http://www.TEAsystems.com - 21
TEA Systems
Yield Enhancement thru Modeling
Process-Window Derived Feature and DoF
Focus Response Analysis
Focus
Uniformity
Depth of Focus
Uniformity
The Feature Derived Best Focus is next calculated for every site on the field. This focuscontour is not the exact focus-wavefront of the lens but it is the response surface
experienced by the measured feature and rapidly degrades with the onset of Haze. Similarly
the features Depth-of-Focus (DoF) can be visualized for every point on the exposure.
TEA Systems
http://www.TEAsystems.com - 22
Yield Enhancement thru Modeling
Dose Response Analysis
Isofocal Analysis
Isofocal deviant curves
Aberration level at each dose
Isofocal Analysis: All-sites on field, BCD features Vertical (black) and Horizontal (red)
1)
Calculate IsoFocal dose for each feature and site
Classic: IsoFocal point is found when the 2nd derivative of the process window = 0
Bossung: IsoFocal deviant = the magnitude of the 3rd Bossung curve coefficient
Isofocal point is at the minimum for the curve
2) Lower curves: Level of aberrations for the dose plotted as magnitude of 2 nd Bossung curve
coefficient
3) IsoFocal performance is highly sensitive to both lens and feature design and so responds to Haze
http://www.TEAsystems.com - 23
TEA Systems
Yield Enhancement thru Modeling
Dose Analysis Derivatives
Exposure Latitude @ Best Focus
Exposure Latitude %
at Best Focus
BCDv
BCDh
Contour Plot
Vector Format Plot
The Exposure Latitude Percentage uniformity can now be plotted
for the field at Best Focus providing an improved characterization
for resist setup.
Any point modeled in the field will see the influence of nearby
haze incidence and EL% will degrade as haze increases
http://www.TEAsystems.com - 24
TEA Systems
Yield Enhancement thru Modeling
Dose Uniformity @ BF for 80 nm
Dose Analysis Derivatives
BCD Horizontal
BCD Vertical
Focus errors removed and having calculated the “Feature v Dose” response for each site, we can now
calculate the optimum dose needed to obtain the feature target value at each site. These contours still
contain reticle non-uniformity, lens aberration and scan-perturbations. Haze onset will quickly degrade
this reticle and scanner specific signature
http://www.TEAsystems.com - 25
TEA Systems
Yield Enhancement thru Modeling
At Best Focus/dose for 80 nm
BCDh
Focus Error at 22 mj/cm2
BCDv
http://www.TEAsystems.com - 26
TEA Systems
Yield Enhancement thru Modeling
Conclusions #3
 Weir PW:
 The Weir PW techniques directly
measure the process degradation of the
reticle to feature profiles, loss of
Exposure Latitude and Depth-of-Focus
reduction through modeling of the
response across the entire image
wavefront.
 Weir PW answers the need for rapid
haze detection and the avoidance of
process yield loss long before the effects
are noticed through the demands of the
haze for increase exposure-dose.
 When first detected, the user can
monitor the gradual degradation in fullfield process window and conveniently
schedule reticle cleaning or process
correction
http://www.TEAsystems.com - 27
TEA Systems
Yield Enhancement thru Modeling