Post on 02-Jun-2020
Current status, challenges, and outlook of EUV
Lithography for High Volume Manufacturing
Britt Turkot
Intel Corporation
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Milestone Progress
• Exposure Tool
• Reticle
• Pellicle
• Infrastructure
• HVM Considerations
• Looking Ahead
• Materials
• High NA
• Conclusion
2
Outline
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
NXE:33x0 combined scanner/source availability
• Improvement from
NXE:33X0 to
NXE:3400 platform
• Top contributor is
exposure source
• Need continued
focus on availability
3
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
NXE:3400 combined scanner/source availability
4
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Best data on 3400
comes from
dedicated effort on
small number of
systems – little bit of
luck and lots of focus
• Need to scale to
install fleet
5
Exposure source power meeting roadmap
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
0
50
100
150
200
250
PREDICTED IMPLEMENTATION DATE FOR POWER LEVEL
Sep'08
Mar'09
May'10
Sep'10
Nov'10
Feb'11
Apr'11
Jun'11
Sep'11
Dec'11
Feb'12
Apr'12
Jul'12
Dec'12
Dec'11
Feb'12
Jul'12
Dec'12
Field Actual 3100
ASML Achieved
Field Actual 33x0/3400
EARLYTECHNOLOGY ROADMAPS
NXE:3100ROADMAPS
(105W MAX)
NXE:3300ROADMAPS
(250W MAX)
2014 NXE:3xx0
ROADMAPS
TODAY
2014 NXE:3xx0
ROADMAPS
2016 NXE:3400125WPH(no pellicle)
PR
ED
ICT
ED
PO
WE
R L
EV
EL
(P
ER
RO
AD
MA
P)
(w) • Meeting 250W
exposure source
power established
for NXE:3400
• Proliferation to field
systems
• Continued emphasis
ensuring sufficient
power overhead for
predictable quality
and output
Collector lifetime improvement continues
• Collector degradation
follows continuous,
roughly linear trend –
predictable lifetime
• Recent breakthrough
advances in
reflectivity as f(GP)
• Bottom Line: expect
significant correlation
to system availability
and OpEx
6
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Slide courtesy ASML June 2018
7
Britt Turkot/ Intel
Intel’s Pilot Line: CD trend
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Unfiltered data
• Timeline > 1 year
• Multiple masks
• Multiple features
• Multiple tools
• CD control within tight
distribution
• Stable CD
performance trend
Delta to
Targ
et
CD
(nm
)
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1Delta from Target CD (nm) vs. Process Date
Target
Scanner cleanliness: Intel reticle defectivity
8
• Variability in defect level
after ‘burn-in’; many
tools showing no
adders/reticle load for
several weeks
• Every tool has shown
adders after many
weeks with no adders
• NXE:3400 cleaner
overall
• Unpredictability of adder
events drives need for
pellicle
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Printable reticle defect adders per reticle load
Months in Service
Adders
per
Reticle
Load
NXE:3400
NXE:33X0
NXE:33X0
NXE:33X0
NXE:33X0
0
0.5
1
1.5
3
2
2.5
3.5
1 2 3 4 5 6
Scanner cleanliness: reticle defectivity
9
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
ASML two-fold approach:
one element is to improve
cleanliness → avoid particle
generation in scanner
• Investigation
continues into origin
of defects
• Improved
understanding of
nature of defects
introduced by
scanner
Pellicle membrane progress and infrastructure
10
Britt Turkot/ Intel
• Steady progress in
pellicle membrane
defect levels since
Q3’16
• Multiple
membranes with
zero defects >10um
• Continued focus
expected to deliver
volumes for HVM
SPIE 2018 ASML / Roderik van Es
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
EUV mask pilot line activities
Intel EUV mask manufacturing is capable of volume production with full specification product requirements.
G. Zhang1 2 3 4 5 6 7 8 9 10 11 12 13 142004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Act
ivit
ies
(Pla
tes)
Year
6” wafer
1999: wafer
based EUV
mask process
developed
2001: World 1st
6” EUV reticle
fabricated at
Intel after SEMI
std P37
established
2004: 6700 sq ft
class 1 clean
room built for
EUV mask pilot
2010: EUV mask
process
infrastructure
established for
device
demonstration
2014: Full
field EUV
pellicle
demonstrated
2017: Product
reticles on volume
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, CaliforniaBritt Turkot/ Intel
11
EUV infrastructure has 8 key programs
7 are ready or near-ready now; 1 has significant gaps
E-beam Mask Inspection: In use for low volume production. Need TPT increase.
Actinic Blank Inspection (ABI): Ready for qualification of HVM quality blanks
AIMS Mask Inspection: Systems installed in field; NXE:3400 illumination emulation underway
Pellicle: ASML commercializing
EUV blank quality: Process and yield improvements continue
Blank multi-layer deposition tool: Improving defect results
EUV resist QC: RMQC center at IMEC online
Actinic Patterned Mask Inspection (APMI): High resolution PWI for fab. Still need actinic inspection in mask shop.
EUV infrastructure readiness snapshotFrom 2017 SPIE
Significant progress in EUV infrastructure12
Britt Turkot / Intel
Judged as of Today
2018 International Workshop on EUV Lithography, 12 June, Berkeley, CaliforniaBritt Turkot/ Intel
Overall milestone progress messages
• Combined scanner/source availability improving
• Exposure source remains largest contributor to tool downtime
• NXE:3400 availability encouraging; need to scale to install fleet
• Exposure source power meeting 250W roadmap, field upgrades in process
• Scanner defectivity levels improving with introduction of NXE:3400
• Every tool has shown defects after weeks of clean performance
• Underscores need for pellicle and associated infrastructure / support
• Significant progress in pellicle program over past year
• Pellicle membranes manufactured with zero defects >10um; lifetime and power resiliency continue to increase
• Progress has been made in pellicle membrane material development, but continued improvement necessary for
increasing transmission, withstanding increased source power, and extending lifetime (OpEx)
• Pellicle membrane power resiliency needs to keep pace with increasing source power (300W, 500W, …)
• Manufacturing increasing number of defect-free 7nm EUV masks
• Inspection of pelliclized reticles is needed to ensure predictable yield. APMI is not a show-stopper, but without it
yield and cost may be an issue – no change
13
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Milestone Progress
• Exposure Tool
• Reticle
• Pellicle
• Infrastructure
• HVM Considerations
• Looking Ahead
• Materials
• High NA
• Conclusion
14
Outline
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
HVM insertion considerations → Predictability
• Capability demonstrated:
✓ Exposure source power meeting HVM roadmap
✓ Pilot line imaging performance
• What is impact on fleet predictability of availability vs. power?
• Simulate HVM conditions – how do these parameters affect reliable TPT?
• Simulation methodology assumptions:
• Acceptable level of collector degradation (50%-100% RR)
• Fixed exposure source power (300W)
• Vary availability 65%-95%
• 10,000 runs with comparable results for tool count N=25 and N=100
15
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Impact of Availability is critical
65% availablity
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
16
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Fleet Productivity
• Wide spread in
productivity
65% availablity
75% availablity
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
17
Fleet Productivity
Impact of Availability is critical
• Fleet productivity
distribution
improves with
increasing
availablity
65% availablity
75% availablity
85% availablity
Britt Turkot/ Intel
18
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Fleet Productivity
Impact of Availability is critical
65% availablity
75% availablity
85% availablity
95% availablity
Britt Turkot/ Intel
19
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Fleet Productivity
→ Best-case
productivity of fleet
with 65% average
system availability is
less than the worst-
case productivity of a
fleet with 95%
average availability
Impact of Availability is critical
Availability vs. Source Power
Britt Turkot/ Intel
20
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
10% availability equivalent to
~50W power in terms of fleet
productivity
Fle
et
Pro
du
cti
vit
y
Increasing Availability / Power →
Cannot overlook importance of availability
AVAILABILITY
POWER
• Milestone Progress
• Exposure Tool
• Reticle
• Pellicle
• Infrastructure
• HVM Considerations
• Looking Ahead
• Materials
• High NA
• Conclusion
21
Outline
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
EUV Materials
22
Britt Turkot/ Intel
http://www.australiangeographic.com.au/blogs/wild-journey/2016/07/black-swan-the-impossible-bird
https://geoffpark.files.wordpress.com/2015/04/swans3.jpg
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
23
EUV materials and resolution
Test pattern – Post Litho Test pattern – Post Etch
A.Lio
• EUV enables 2D design features, e.g. corner segments
• Need materials that can take advantage of improved EUV resolution
• Adequate for EUV introduction
• Need materials that are tunable for desired properties
• Materials development constrained by photon availability (BL, MET, NXE)
For continued material development, suppliers need an understanding of fundamental properties of materials
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, CaliforniaBritt Turkot/ Intel
24
Looking ahead: More than photon shot noise
>2.5X dose vs resist 1 and 2
~ 10% CDU improvement
CAR 3
<20 mJ/cm2
CAR 1
<20 mJ/cm2
CAR 2
NXE3300, 28 nm hole
72K measurements • 2.5x higher dose provides <10%
LCDU improvement
• Not consistent with photon shot noise
alone
• There must be a chemical effect
• We must gain a deeper understanding
of how EUV radiation interacts with
resist and design resist for stochastics
• Next generation EUV requires
materials innovation
Britt Turkot/ Intel
Anna Lio
Materials suppliers must have the means to study fundamental properties of materials
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
25
EUV materials
• EUV material interaction processes are complex‒ Photon absorption
‒ Photoelectron gen.
‒ Secondary electron gen.
‒ Radical generation
• Electron-stimulated process induce chemical reactions
• Dissolution mechanism varies for different material systems
• Not one-size-fits-all
J. Van Schoot, ASML, February 2018
Must consider multiple options / material systems
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, CaliforniaBritt Turkot/ Intel
26
High NA EUV: The next step
Britt Turkot/ Intel
Slide courtesy ASML February 2018
• Next logical step
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Milestone Progress
• Exposure Tool
• Reticle
• Pellicle
• Infrastructure
• HVM Considerations
• Looking Ahead
• Materials
• High NA
• Conclusion
27
Outline
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
• Exposure source → First field system meeting source power roadmap: power improvements
proliferating to install fleet
• Availability → NXE:3400 platform demonstrating improved capability – keep focus
• Pellicle → Needed to ensure EOL yield; pellicle program continues to make significant
progress
• Infrastructure → demonstrating increased maturity; single gap remains
• HVM requires predictability
− Output impact of 1% improvement in availability > 4% improvement in source power
− OpEx (mostly source consumables) – Collector lifetime improvements encouraging – need to translate to field
systems
• Materials
− Materials performance – Won’t gate introduction of EUV, but need to consider stochastics for decreasing
feature sizes and high NA: need to understand the interaction of EUV radiation with resist and design resist
materials for stochastics
28
Conclusion: Preparation for HVM and beyondJudged as of today
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
29
AcknowledgementsSteve Carson
Anna Lio
Mark Phillips
Brian McCool
Eric Stenehjem
Tim Crimmins
Ying Zhou
Markus Kuhn
Curt Ward
Sam Sivakumar
Guojing Zhang
Ted Liang
Jeff Farnsworth
Sang Lee
Florian Gstrein
Frank Abboud
Britt Turkot/ Intel 2018 International Workshop on EUV Lithography, 12 June, Berkeley, California
Backup
2018 International Workshop on EUV Lithography, 12 June, Berkeley, California