Upload
others
View
9
Download
1
Embed Size (px)
Citation preview
KCUM/ Slide 1<file name><version 00><author>
EUV Defects – Kevin Cummings ASML - Brian Niekrewicz, Youri van Dommelen, Thomas Laursen, Robert Watso, John Zimmerman, Aiqin Jiang, Bart Kessels, Sander Bouten, Robert Routh
SEMATECH - Liping Ren
KLA-Tencor - Sumanth Kini, Nithin Yathapu
AMTC - Christian Holfeld, Jan Hendrik Peters
Toppan Photomasks - Brid Connolly
KCUM/ Slide 2
Introduction
PurposeTo get an early understanding of EUVL defectivity
ProcessUtilize the equipment and processes of today to understand the consequences for pilot and production tomorrowBaseline and minimize manual reticle handlingBaseline Alpha Demo Tool (ADT) contributions
NOTE: NXE designs remove ADT limitations
Initial study – reticle particles using wafer inspectionThe “Defect Mask”Results to date
KCUM/ Slide 3
Details of this study
Uncoated quartz substrates are used in baseliningASML Alpha Demo Tool [AD1] is located in the CNSE Albany Nanotech Fab North [NFN]SEMATECH’s Lasertec M1350 is in Nanotech Fab South
53 nm PSL capability
Minimized particles from manual reticle handling with optimized procedures “Partitioned” particle tests run to determine and minimize contributions from AD1Continuing to lower these background particles to enable further studies of EUV defectivity
KCUM/ Slide 4
Partition test Manual transfer from Lasertec to internal position within AD1
0
5
10
15
20
25
Transferto load
lock
Transferto elevator
Transferto robot
Transferto dock
Transferto clamp
Transferto
exposure(no light)
Part
icle
s
1 exchange10 exchanges 25 exchanges
40 exchanges
10
1
5 55 4 4
15 15
21
Particle size >53nmTo
tal P
artic
le c
ount
KCUM/ Slide 5
Inspection results show that added particles are concentrated in the corners
40 cycles21 particles (bin 4+ is > 53nm)0.525 particles per passPrint area 102 x 132 mm
Quality area 142 x 142 mm
No particles appearin print area
KCUM/ Slide 6
0
5
10
15
20
25
Transferto load
lock
Transferto elevator
Transferto robot
Transferto dock
Transferto clamp
Transferto
exposure(no light)
Part
icle
s
1 exchange25 exchanges
40 exchanges
10
1
0
3 2
0
Particle size >53nm
<0.031 particles / exchange
Background contribution acceptable for testing
Tota
l Par
ticle
cou
nt
Partition test Manual transfer from Lasertec to internal position within AD1
Print area only
KCUM/ Slide 7
45nm grating H/V
45nm grating H/V
45nm grating H/V
38nm grating H/V
38nm grating H/V
38nm grating H/V
32nm grating H/V
32nm grating H/V
32nm grating H/V
Initial study – The Defect Mask
Pos defects Neg defects
BridgesGaps
Optimize inspectionon program defects
Resistimages
Reticleimages
KCUM/ Slide 8
Optimization sequence for KLA 2800*
Identify programmed defects for S/N
optimization
Optimize S/N for programmed defects of
interest
10x run on all programmed defects
and determine capture rate
Check nuisance level, acceptable y/n
Investigated High throughput, High sensitivity, GHI-Line, Blueband DUV & Broadband DUV
High Sensitivity Blueband DUV & Broadband DUV gave best signal to noise* Note: 2800 wafer inspector spec’d for 65nm node. Results from current-gen tool may differ
KCUM/ Slide 9
Capture rate on programmed defects
• Measure 5 fields on KLA2800, repeat 10x• 4 sets of programmed defects per field (bridge, gap, neg & pos defect)
• Each repeater should show up 10x, if capture rate = 100%
Example for 45nm features
1 2
34
5
KCUM/ Slide 10
0.010.020.030.040.050.060.070.080.090.0
100.0
avg
capt
ure
rate
[%]
Max capt rate [%]
Min capt rate [%]
AVG capt rate [%]
Capture rate on programmed defects, 32nm features Bridges, avg capture rate (5 flds, 10rpts/fld)
KCUM/ Slide 11
0
10
20
30
40
50
60
70
80
90
100
avg
capt
ure
rate
[%] Max c apt rate [% ]
Min c apt rate [% ]
A V G c apt rate [% ]
Capture rate on programmed defects, 32nm features Neg. defects, avg capture rate (5 flds, 10rpts/fld)
KCUM/ Slide 12
Capture rate not related to feature size Only plotting capture rates < 100% Positive defects, avg capture rate (5 flds, 10rpts/fld)
45nm feature 38nm feature 32nm feature
0
10
20
30
40
50
60
70
80
90
100
avg
capt
ure
rate
[%]
Max capt rate [% ]
Min capt rate [% ]
A V G capt rate [% ]
45nm feature 38nm feature
KCUM/ Slide 13
Capture rate not related to feature size Only plotting capture rates < 100% Neg. defects, avg capture rate (5 flds, 10rpts/fld)
0102030405060708090
100
avg
capt
ure
rate
[%]
Ma x ca p t ra te [% ]
Min ca p t ra te [% ]
AVG ca p t ra te [% ]
45nm feature 32nm feature
32nm feature 32nm feature
32nm feature
32nm feature
38nm feature
38nm feature
38nm feature
38nm feature
38nm feature
45nm feature
38nm feature
KCUM/ Slide 14
1st defect data measured – no optimization or tweaking
Random defects from 23 fields exposed on EUV AD1 and measured on KLA2800Remove reticle and programmed defectsSEM review
Random defects - raw data
Count : 3134
KCUM/ Slide 15
Defect pareto Random defects
NV = 27%
0
5
10
15
20
25
30
35
40
45
50
PartialBridge
FullBridge
PartialBridgeEOL
FullBridgeEOL
Particle Other Open NotV isual
Part
icle
s
1 field 100% review
KCUM/ Slide 16
Summary Manual transfer of reticles is possible at rates as low as 0.03 particles (per reticle exchange/exposure)
Allows meaningful defect work using today’s limitationsWork continues to reduce particle number further
Printed capture rate vs defect size is different for feature typesCannot “simply” state capture rate vs defect sizeHere capture rate seems to be independent of feature size
Caveat: - KLA2800 is spec’d for 65nm node. Future studies using current-generation wafer inspector may reveal different trends
Determining reticle defects from wafers is complexIndustry coordination/cooperation would be beneficial
Initial random defect data set shows = ~14 defects /cm2
First look and “as is” – no optimizationsMajority of these defects are bridgesNuisance level 27%
Continuing reticle-wafer defect printability studies