Upload
len-mei
View
98
Download
1
Embed Size (px)
Citation preview
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
ITRS Conference December 1 & 2, 2003
HsinChu, Taiwan
2003 ITRS Yield Enhancement (YE) Update
Presented by Dr. Len Mei
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Yield Enhancement Presentation Outline
• ITWG Co-chairs and Contributors
• Chapter Outline
• Definition and Scope
• Difficult Challenges
• Technology Requirements and Potential Solutions
• Summary
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
2003 ITRS YE ITWG Co-chairs • Taiwan
– Tings Wang
• Promos Technologies
– Len Mei
• Promos Technologies
• United States
– Fred Lakhani
• International Sematech
– Christopher Long
• IBM
• Europe
– Ines Thurner
• Infineon
– Dick Verkleij
• Philips
• Japan
– Masahiko Ikeno
– Renesas
– Hiroshi Kitajima
– Selete
• Korea
– TBD
Note: Contributions of all TWG members around the world are
gratefully acknowledged.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
2003 YE ITWG Contributors • Europe
– Ines Thurner (Infineon)
– Dick Verkleij (fei)
– Lothar Pfitzner (FhG-IISB)
– Andreas Nutsch (FhG-IISB)
– Andreas Neuber (M+W Zander)
– Hans-Martin Dudenhausen (isiltec)
• Japan
– Masahiko Ikeno (Renesas)
– Hiroshi Kitajima (Selete)
– Toshihiko Osada (Fujitsu)
• Taiwan
– Steven Ma (Mxic)
– Jimmy Tseng (PSC)
– CH Chang (SIS)
– Chan-Yuan Chen (TSMC)
– Jim Huang (UMC)
– CS Yang (Winbond)
– Tings Wang (Promos Tech)
– Len Mei (Promos Tech)
• United States
– Fred Lakhani (ISMT)
– Christopher Long (IBM)
– Mike Patterson (Intel)
– Kevin Pate (Intel)
– Mike Retersdorf (AMD)
– Ron Remke (ISMT)
– Mike McIntyre (AMD)
– Rick Jarvis (AMD)
– Ken Tobin (ORNL)
– Hank Walker (Texas A&M)
– Charles Weber (Portland State Univ.)
– Ralph Richardson (Air Products)
– Mark Camenzind (Air Liquide)
– Joe O’Sullivan (Intel)
– John DeGenova (TI)
– Jeff Chapman (IBM)
– Val Stradzs (Intel)
– Keith Kerwin (TI)
– James McAndrew (Air Liquide)
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Yield Enhancement Chapter Outline
• Definition and Scope
• Difficult Challenges
• Technology Requirements and Potential Solutions
– Yield Model and Defect Budget (YMDB)
– Defect Detection and Characterization (DDC)
– Yield Learning (YL)
– Wafer Environment Contamination Control
(WECC)
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Yield Enhancement Definition and Scope
• Definition:
– To improve the baseline yield for a given technology
node from R&D yield level to mature yield.
– The definition assumes a functional baseline process
for a given process technology and it’s compatibility
with the design of the product being fabricated.
– The definition reinforces the chapter focus on the
yield ramp portion of the yield learning curve.
• Scope
– Scope of YE chapter has been limited to wafer sort
yield.
– Fab line yield, final test yield, assembly/packaging
yield are not included in the scope of the YE chapter.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
YE Difficult Challenges
• High-Aspect-Ratio Inspection.
– High-speed, cost-effective tools are needed to rapidly detect defects
at 1/2 X ground rule (GR) associated with high-aspect-ratio contacts,
vias, and trenches and especially defects near or at the bottoms of
these features.
• Non-visual Defect Detection
– In-line and end-of-line tools and techniques are needed to detect
non-visual defects.
• Design for Manufacture & Test (DFM & DFT) and
Systematic Mechanisms Limited Yield (SMLY)
– IC designs must be optimized for a given process capability and
must be testable and diagnosable. Understanding SMLY is
mandatory for achieving historic yield ramps in the future.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Void
caused by
residue left
behind
after via
etch
process
step
0
5
10
15
20
180
nm
150
nm
130
nm
100
nm
70
nm
Aspect Ratios at Various ITRS Nodes
A/B
C/D
Combined
Lower Dielectric
Upper Dielectric
Via Dielectric
Via Etch
Barrier
Hard Mask
Canal Etch
Barrier
CMP Barrier
Canal
High-Aspect Ratio Inspection (HARI)
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
YE Difficult Challenges (continue) • Data Management for Rapid Yield Learning.
– Automated, intelligent analysis and reduction algorithms that
correlate facility, design, process, test, and work-in-process (WIP)
data must be developed to enable the rapid root-cause analysis of
yield-limiting conditions.
• Yield Models
– Random, systematic, parametric, and memory redundancy models
must be developed and validated to correlate process-induced defects
(PID), particle counts per wafer pass (PWP), and in-situ tool/process
measurements to yield.
• Correlation of Impurity Level to Yield.
– Data, test structures and methods are needed for correlating fluid/gas
contamination types and levels to yield.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Yield Management Today Increasing Data Complexity
fabrication
process
A
fabrication
process
B bin, bit, and
parametric electrical
test data
in-line
optical defect
data
assembly
&
final test
tool state, ISPM,
moisture,
etc.
product data management
system
process data management
system
WIP data management
system
SSA
Databases
optical
confocal
SEM
FIB
ADC SPC
tool state, ISPM,
moisture,
etc.
Yie
ld
Ma
na
ge
me
n
t
...
AEA
(knowledge discovery)
engineering analysis
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
YE Difficult Challenges (continue)
• Line edge roughness, ACLV, subtle process
variation. Where does process variation stop and
defect start? Need to improve signal to noise to
delineate defect from process variation.
• Contamination transferred from wafer edge and
backside.
• Unified definition of defects based on yield impact.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Technology Requirements and Potential Solutions
• Yield Model and Defect Budgets
– With systematic mechanisms limited yield (SMLY) dominating the rate of
yield learning, a concerted effort is required to understand, model and
eliminate SMLY detractors.
– The impact of line edge roughness (LER) on yield needs to be understood,
modeled and controlled to achieve acceptable yields for current and future
technology nodes.
– Continuous improvement in tool cleanliness will be required to achieve
acceptable yields.
• Defect Detection and Characterization
– Cost effective high throughput high aspect ratio inspection (HARI) tools
are needed urgently to achieve acceptable yields for current and future
process technology nodes.
– Signal to noise improvements are required for defect metrology tools to
detect ever shrinking critical defects of interest.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Technology Requirements and Potential Solutions…continue
• Yield Learning
– With increasing process complexity and longer cycle times, tools and
methods are needed to increase the number of yield learning cycles for
each technology node.
– With move to smaller features, longer processes, 300mm wafers and new
materials (low k, high k, etc.), numerous tools and methods are required
to understand all the yield detracting interactions. Use of SOI and SiGe
will further challenge yield learning.
• Wafer Environment Contamination Control
– Data, test structures and methods are needed to identify and control yield
detracting contaminants in the wafer environment, airborne and process
critical materials and ultra pure water.
2 December 2003 – ITRS Public Conference — Hsin Chu, Taiwan
Summary
• The YE chapter revision for the 2003 ITRS roadmap has been
completed with contribution from the global yield community. Several
regions led the revision of various sections of the chapter:
– Yield Model and Defect Budget Japan
– Defect Detection and Characterization Europe
– Yield Learning USA
– Wafer Environment Contamination Control USA
• Ever shrinking yield critical defects, high aspect ratio defects, non-
visual defects and systematic mechanisms limited yield (SMLY) top the
list of challenges for Yield Enhancement.
• The yield enhancement community is constantly challenged to achieve
acceptable yield ramp and mature yields due to increasing process
complexity and fewer yield learning cycles with each subsequent
technology node.