2. Systematic equipment health
monitoring (fingerprint) addresses…
◄ Unpredictable downtime of equipment disrupts
factory logistics forcing factory to continuously
modify product and resource allocation
◄ Inability of process tools to provide useful data
to factory information and control system (FIDS)
limits opportunity to monitor and predict
equipment health
◄ Long wait times during equipment maintenance
reduces the amount of usable resources in the
factory increasing product queues and creating
bottlenecks
15 October 2012
3. How these industry detractors are
being addressed by ISMI?
◄ Creating industry guidance and methodology for
fingerprinting
◄ Creating guidance for practical creation and use of
independent fingerprints and fault detection
application models
◄ Demonstrating requirements and factory level
management of fingerprint specifications to
promote and improve equipment traceability,
reliability and health assessment
◄ Enhancing data availability and definition for
multiple applications (e.g., FDC, APC, PHM, EPT,
SPC, WTW, etc.)
15 October 2012
4. What is Fingerprinting?
◄ A fingerprint is a unique representation of the functional
performance of a semiconductor equipment or one of its
components
◄ The goal of fingerprinting is to return the equipment to its
previous condition or to match it to an optimally performing
equipment by understanding its differences in relation to a
“golden” fingerprint
15 October 2012
5. In a manufacturing context fingerprint
is…
◄ A set of data variables associated with the equipment
component being fingerprinted
◄ Sampled at some rate over some period of time
◄ Transformed and analyzed using a set of mathematical
techniques
◄ To generate a result that represents the state of that
equipment component at that time
5
6. Differentiation Between Data
Mining and Fingerprinting
◄ Data mining uses mathematical
techniques that are applied to data stored
in the factory database searching for
indicators of problems or potential issues
with the equipment or processing.
◄ Fingerprinting is a systematic approach
that uses equipment component data to
calculate related parameters that help in
the determination of equipment health
and performance.
15 October 2012
7. Fingerprinting landscape scope
areas of interest
• Fingerprinting can be applied to:
◄ Single isolated components
◄ Support sub-systems
◄ Entire process delivery systems
◄ And even abstract items (e.g.,
key process trace variable)
7
8. Fingerprint stakeholders
• Stakeholders span over multiple organizations and roles
within the equipment supplier…
Test engineering, manufacturing QA
Site installation and equipment qualification
support
Field service
Application support
… and the fab customer
Equipment and maintenance engineering
Production operations
Process engineering
8
9. Start
Perform
Fingerprint Use Cases
baseline test
Being Considered
Determine
critical
parameters
◄ Equipment Acceptance and Qualification
Critical
◄ Functional capability
parameters
identified? ◄ Process capability
Perform
◄ Chamber-to-Chamber Matching
Chamber
adjustments
◄ Machine-to-Machine Matching
◄ Equipment Qualification After Maintenance
Retest
◄ Maintenance & part replacement
Matching
accopmplished?
Document and
store results
End
15 October 2012
10. Fingerprinting benefits throughout the
equipment life cycle
◄ Benefits range from
◄ Ensuring multiple tools are shipped in an identical state
◄ Comparing tools from one delivery to the next
◄ Speeding tool acceptance (and “time to money”)
◄ Avoiding incipient failures
◄ Reducing scheduled maintenance time
◄ Improving field service performance (time, hit rate)
◄ Speeding process excursion problem resolution
◄ Support equipment continuous improvement programs
10
11. Fingerprinting approach and method
◄ The fingerprinting process includes
◄ Calculation of basic statistics on historical equipment data
associated with individual equipment components
◄ Execution of an operation sequence to put the equipment in a
known state, running a fingerprint “recipe” to exercise the object of
interest in a near-production context
◄ Development of multivariate control models to determine relative
contribution of equipment parameters to fingerprint results
(perhaps at various operating points)
TOOL 0.7
EFEM 0.9 PM1 0.9 PM2 0.4 PM3 0.65
GAS 0.9 VAC 0.8
BOX PUMP 0.7 RF 0.3 CTRL
GEN 0.3 MATCH 0.7
11
12. Fingerprinting Application
Characteristics
◄ Build-time
and run-time
environment
◄ Data selection and
transformation
◄ Range of analysis methods
◄ Artifactmanagement
(recipes, data sets, results)
◄ Stratificationof capabilities
(Basic, Plus, Pro, etc.)
12
13. Fingerprinting Data Flow: Extraction,
Transformation and Loading Process
CMP Fingerprint Visualization and
Transformation Application Usage
GEM ("Equipment View",
CMP Models
Custom "Wafer View“, etc.)
Implanter
EDA
Extract Transform Load
EDA
Furnace
Data
Data Storage
Custom Raw Transformation
Furnace data
reports
GEM
Etch
EDA Fingerprint MES,
Execution RMS,
Etch Other DB
Fingerprint
CVD Custom
Definition Fingerprint
GEM Analysis
Recipes
CVD Execute
Raw
Data Fingerprint Application System
15 October 2012
14. EHM Promotes Collaborative
Visualization
From Data to Information to Knowledge to Action
• Shared view by IDM and OEM
– Same data
– Same methodologies
– Same conclusions
Fingerprint Benefit
Better equipment matching
Raw Processing
Time Lot 1 (RPT1) TRPT = RPT1 + RPT2
The third lot starts a new cascade and benefits
Acceptance time reduction
Improved performance monitoring
Raw Processing from overlapping of initial events up to setup.
Time Lot 2 (RPT2) A new recipe requires the full setup time
Raw Processing
Initial Lot
Setup Events
Parameter
Checking only
Total Move In to Move Out time for two lots (TTOT)
Time Lot 3 (RPT3)
Full Setup time required
For a new recipe
Equipment reliability Improvement
Engineering resource efficiency
and productivity
15 October 2012
15. Status of the Project
• Fingerprint application developed
– Can be used by suppliers as well as IDMs
– Available through 3rd party software supplier
• Fingerprint demonstration using manufacturing data
– ~ 6 months of data from 4 tools of same type (CVD)
– Results to determine golden fingerprint models and apply
them to chamber matching
• After maintenance fingerprint application
– Determination of fingerprint models based on parts
replacement
• Multiple recipe fingerprint management
– Determination of equipment health monitoring based on
multiple recipe usage in equipment
15 October 2012
16. Related Documents
• Enhanced Equipment Quality Assurance (EEQA) Data Usage Investigation
Work
– TTID: 09095031A-ENG Date Published: 30-Sep-2009
• ISMI Enhanced Equipment Quality Assurance (EEQA) General Guidelines
– TTID: 09125066B-ENG Date Published: 12-Apr-2010
• Equipment Component Common Templates
– TTID: 35255TD Date Published: 28-May-2010
• Enhanced Equipment Quality Assurance (EEQA) Supplier Implementation
Assessment Report
– TTID: 35546TD Date Published: 30-Jul-2010
• Fingerprint Methodology V0.0
– 11085156A-ENG Date Published: 31-Aug-2011
• Fingerprint Factory Level Requirements
– 11105163A-ENG Date Published: 31-Oct-2011
• Fingerprint Data Model and Key Parameters
– 11115168A-ENG Date Published: 30-Nov 2011
• EHM Fingerprint Lessons Learned
– 37813TD Data Published: 19-Dec-2011
15 October 2012
17. Questions?
Contact
Gino Crispieri (PMP)
gino.crispieri@ismi.sematech.org
Tel 518-649-1185
15 October 2012