1. Kostas Adam, Andrew Neureuther
Characterization of Phase Defects in Phase Shift Masks and a New Model for Mask Edges
SFR Workshop
November 8, 2000
Kostas Adam, Andrew Neureuther
Berkeley, CA
2001 GOAL: complete simulations for polarization masks and multi-parameter test structures by 9/30/2001
11/8/2000

2. Motivation
Phase Shift Masks are finding their way into the production of ICs. However, the impact of phase defects needs to be carefully characterized:
The inspection of phase defects is a difficult task. New approaches in locating defects in the quartz need to be explored
Ideal mask simulations although speedier are known to suffer in accuracy. Taking into account edge diffraction effects the accuracy can be significantly improved
11/8/2000

3. An Isolated Quartz Post is Different than a Void
l=248nm x y
Observation plane
200nm x y
Observation plane 1 2
Normalized |Ey|
1.0
1.2
1.4
1.6
1.8
-120
-80
-40 40
Mask Position (mm)
Phase (deg)
Mask Position (mm)
Phase (deg)
100
120
140
160
180
200
220
1.0
1.2
1.4
1.6
1.8
11/8/2000

4. E-field Phase and Average Intensity vs. Defect Size
11/8/2000

5. Example: Alternating Phase Shift Mask
defect-free APSM
post defect
phase defects
void defect
defect height
underetch 0o
180o d
size 0o
180o
CDmask
Illumination: Linearly polarized plane wave (TE/TM), l=248nm
Imaging: Demagnification=4X, s=0.3, NA=0.68
160x160nm (4X) post
|E|
Phase of E
Glass posts and voids have different near-field characteristics
160 x 160nm (4X) void
|E|
Phase of E
11/8/2000

6. CD Variations Caused by Phase Defects
%DCDaerial=(CDdef-CDref)/CDref
11/8/2000

7. Inspection of Phase Defects: Approach
The defect proximity to the feature affects both its inspectability and its printability
Isolated quartz defect: Spatial variations of EM field (modes) 0o
180o
Approach (similar to Socha’s contact hole inspection):
- Excite a spatial mode within the defect and find the EM field outside
- Time reversal: Use the calculated EM field as new illumination
resulting EM field
excite defect mode
11/8/2000

8. New Model for Mask Edges: Approach
Goal: Model the bias between ideal and rigorous simulations to allow use of the thin mask approximation in the design of OPC etc.
geometrical position of edge
Illumination: E
Quartz
clear level (1)
Stack
middle level
Air
gray level (.265)
New edge location
for ideal simulation
New Model:
Modify mask geometry that is input in an ideal SPLAT simulation. This can be
done by extracting information from the near-field diffraction pattern of 2D
rigorous simulations (TEMPEST).
50nm
So far:
New Model:
In SPLAT
Mask
In SPLAT
Mask
a+Da a a a b b b
b +Db
11/8/2000

9. Conclusions and Goals
Isolated quartz voids are seen to behave more like Cr defects isolated quartz posts behave more like phase defects
Voids (of equal size) are less than 50% as harmful to the aerial image than posts (first reported by Friedrich et al., SPIE2000, see also Adam et al., EIPBN2000 )
Vertical propagation programs (SPLAT) underestimate the effect of a defect by 25%-50%, when the insufficient 1/180o model for phase defects is used
Establish guidelines for 3D defect sizing and feature interactions (Goal for 9/2002)
Propose alternative inspection approach for phase defects and demonstrate its limits (Goal for 9/2002)
11/8/2000