Motivation:
Significant challenges for various CD measurement matching procedures are reaching a comparable complexity as result of negative effects of roughness on the features. Due to the constant trend of integrated circuit in features reduction, impact of roughness start to be more destructive for various sets of measurement algorithms. Commonly used attempts to increase magnification for pattern recognition in addressing mode could in turn detect higher deviation from predefined patterns and thus initiate shift in placement of measurement gate.
Description of the approach:
The purpose of this paper is to discuss how to reduce measurement gate placement variation
impact and filter acquired data using edge correlation approach – creation of width correlation
function represents particular feature under test and it’s comparison to “golden” one as a mean of
detection of uncorrelated scans, which in turn should be excluded from overall computation of
matching results.
We describe general approach for algorithm stepping and various techniques for judgment of measurement comparison validity. Presented approach also has particular interest in determination of specified tool performance for predefined pattern recognition feature as well as for pattern recognition algorithm robustness study - direct interest for manufacturer.
Evaluation of results:
Precise matching estimation as part of Round Robin routines creating possibility to work with
restricted amount of data and perform quick reliable qualification procedures.
This paper concentrated on practical approach and used both simulation data and actual
measurement data before and after proposed optimization taken by various generation tools by
Hitachi (S-8840, S-9300, S-9380) in production environment.
3. Scaling Matching procedures.
Qual matching procedures
CLV and CLH patterns – pitch measurement 239.9 nm
Just rough calculation of matching – high dependency of results on the
X-Y placement. Statistically based, pitch measurement (not line or space).
Inter site matching procedure.
Back to back 2 runs of 4-6 different critical layers with 2 to 6 different
measurements per layer. Everything should be matched in 1 nm to 5 nm
(per layer).
PM matching procedures.
Daily monitor/Weekly monitor – impact of charging. See resolution monitor
for results.
Advanced Matching for CDSEM
4. Line/Space/Pitch dilemma
CDD qual require matching up to 1 nm for most critical layers (95% CI)
How we are measuring ?
Why it is problematic ?
No. Chip No. D No.1 D No.2 D No.3 D No.4 D No.5 D No.6
VSP VL VPTCH HS HL HPTCH
Data P No. Data P No. Data P No. Data P No. Data P No. Data P No.
1 03,04 105.8 1 171.9 2 280.7 3 114.1 4 167.3 5 282.0 6
2 03,05 111.8 1 168.8 2 282.1 3 113.9 4 166.2 5 280.3 6
3 04,04 107.1 1 171.3 2 278.7 3 112.4 4 167.1 5 279.9 6
4 04,05 108.2 1 169.4 2 277.8 3 115.6 4 165.8 5 281.6 6
5 05,04 117.1 1 164.5 2 281.6 3 118.8 4 164.4 5 283.0 6
6 05,05 102.5 1 172.2 2 275.9 3 110.7 4 167.6 5 279.7 6
7
8
9
Maximum 117.1 172.2 282.1 118.8 167.6 283.0
Minimum 102.5 164.5 275.9 110.7 164.4 279.7
Mean 108.7 169.7 279.5 114.3 166.4 281.1
Max-Min 14.7 7.7 6.2 8.1 3.2 3.4
3 Sigma 15.3 8.6 7.3 8.3 3.6 4.0
Advanced Matching for CDSEM
5. Line/Space/Pitch dilemma – Normalization issue as result of 2
measurement gates application for pitch measurement
On the specimen
Line Space
Pitch
SE intensity profile
0
50
100
150
200
250
1
Pixel
Brightness
Line
Threshold
60%
SE intensity profile
0
50
100
150
200
250
1
Pixel
Brightness
Space
Threshold
60%
SE intensity profile
0
50
100
150
200
250
1
Pixel
Brightness
Pitch
Threshold
60%
Advanced Matching for CDSEM
6. Placement dilemma – Are we measuring on the same place?
Line/Space/Pitch dilemma – Why Line + Space≠ Pitch ?
How we could proceed in order to reduce impact of placement shift
and normalization error
Advanced Matching for CDSEM
7. Placement and Line/Space/Pitch dilemma – Solution by SW
Compare just correlated parts of the images taken for matching.
Correlated picture
Shifted picture
Shifted picture
Advanced Matching for CDSEM
8. Impact of Placement dilemma – testing results
Verticalsize1
0.992
0.993
0.994
0.995
0.996
0.997
0.998
0.999
1
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.994599
0.994521
0.994358
0.995913
Mean
0.001199
0.000896
0.002050
0.002302
Std Dev
0.00054
0.00040
0.00092
0.00115
Std Err Mean
0.99311
0.99341
0.99181
0.99225
Low er 95%
0.99609
0.99563
0.99690
0.99958
Upper 95%
Means and Std Deviations
Oneway Analysis of Vertical size 1 By ID
Horizontalsize1
0.9875
0.99
0.9925
0.995
0.9975
1
1.0025
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.995852
0.995347
0.995519
0.994763
Mean
0.002431
0.000126
0.003779
0.000952
Std Dev
0.00109
0.00006
0.00169
0.00048
Std Err Mean
0.99283
0.99519
0.99083
0.99325
Low er 95%
0.9989
0.9955
1.0002
0.9963
Upper 95%
Means and Std Deviations
Oneway Analysis of Horizontal size 1 By ID
Verticalsize2
0.99
0.992
0.994
0.996
0.998
1
1.002
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.994476
0.994563
0.994676
0.996406
Mean
0.001527
0.000808
0.003688
0.003328
Std Dev
0.00068
0.00036
0.00165
0.00166
Std Err Mean
0.99258
0.99356
0.99010
0.99111
Low er 95%
0.9964
0.9956
0.9993
1.0017
Upper 95%
Means and Std Deviations
Oneway Analysis of Vertical size 2 By ID
Horizontalsize2
0.992
0.993
0.994
0.995
0.996
0.997
0.998
0.999
1
1.001
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.996068
0.995563
0.995640
0.994824
Mean
0.001704
0.000530
0.002962
0.000979
Std Dev
0.00076
0.00024
0.00132
0.00049
Std Err Mean
0.99395
0.99490
0.99196
0.99327
Low er 95%
0.99818
0.99622
0.99932
0.99638
Upper 95%
Means and Std Deviations
Oneway Analysis of Horizontal size 2 By ID
Verticalsize3
0.99
0.9925
0.995
0.9975
1
1.0025
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.999648
0.994572
0.995677
0.993940
Mean
0.000996
0.001492
0.003655
0.003117
Std Dev
0.00045
0.00067
0.00163
0.00156
Std Err Mean
0.99841
0.99272
0.99114
0.98898
Low er 95%
1.0009
0.9964
1.0002
0.9989
Upper 95%
Means and Std Deviations
Oneway Analysis of Vertical size 3 By ID
Horizontalsize3
0.99
0.9925
0.995
0.9975
1
1.0025
After 1 After 2 Before 1 Before 2
ID
Excluded Row s 1
After 1
After 2
Before 1
Before 2
Level
5
5
5
4
Number
0.995766
0.995904
0.995919
0.996379
Mean
0.001444
0.001715
0.002354
0.004105
Std Dev
0.00065
0.00077
0.00105
0.00205
Std Err Mean
0.99397
0.99378
0.99300
0.98985
Low er 95%
0.9976
0.9980
0.9988
1.0029
Upper 95%
Means and Std Deviations
Oneway Analysis of Horizontal size 3 By ID
Advanced Matching for CDSEM
9. Gain in implemented.
1. Qual speed up.
2. Reliable “of the fly” measure of matching.
3. As vendor application for tool calibration (image shift
parameter).
4. In connection with resolution monitor powerful application for
tool stability tracing.
5. PM elimination (from time based activity to performance based).
Advanced Matching for CDSEM