Data Management

1
Data Management

• Gino Crispieri ISMI

Advanced Materials Research Center, AMRC,
International SEMATECH Manufacturing Initiative,
and ISMI are servicemarks of SEMATECH, Inc.
SEMATECH, the SEMATECH logo, Advanced Technology
Development Facility, ATDF, and the ATDF logo are
registered servicemarks of SEMATECH, Inc. All
other servicemarks and trademarks are the
property of their respective owners.
2
Outline

• EDA Evaluations Lessons Learned
• IC Maker Data Quality Issues
• ISMI Time Synchronization Guidelines
• ISMI Data Quality Guidelines
• Data Quality Approach
• Summary

3
EDA Equipment Evaluations

• Purpose
• Accelerate availability of conformant EDA
  interface implementations in process equipment
• Scope
• Functional verification
• Based on ISMIs EDA Evaluation Method V3.0
• Performance
• Verification of data generation rates with
  variables per trace, sampling rate, and group
  size
• Measurement points for analysis and diagnostics
  in a log file (e.g., before un-marshalling, after
  un-marshalling, before tracing, after tracing)
• ISMI EDA Evaluation Report
• Supplier overall feedback and suggestions for
  improvement
• Shared with ISMI member companies

4
Equipment Lessons Learned

• Some OEMs ready to deliver EDA
• Solutions from 3rd party sources as well as from
  internal development
• Implementations vary from mature to beta versions
• Some applications do not have SEMI Objects and
  300 mm state models
• State model and parameter availability is
  sometimes limited
• Data and events availability
• Which ones first and which ones later
• Suppliers and IC Makers
• Data collection plan definition
• Trace should generate even number of reports per
  second to avoid jitter
• Find best performance based on number of
  requested parameters
• Data freshness and precision
• Data is not always refreshed at the requested
  collection rates
• Data does not provide the required precision
• Timestamps are not accurate
• Equipment physical structure versus logical
  modeling
• Critical to get advantage of EDA data collection
  features
• Provides better representation of the equipment
  functional behavior

5
EDA Host Side Evaluations

• Purpose
• Accelerate the availability of commercial EDA
  host solutions
• Scope
• Functional Verification
• Based on ISMIs EDA Evaluation Method V3.0
• Performance
• Verification of data consumption rates with
  variables per trace, sampling rate, and group
  size
• Measurement points for analysis and diagnostics
  in a log file (e.g., before un-marshalling, after
  un-marshalling, before tracing, after tracing)
• Application Maintenance
• Evaluate usability, completeness, etc. from an IT
  and process engineers point of view and provide
  feedback from end users perspective
• Application Connectivity
• Verify the capability to share and store data to
  a factory application and database without delay

6
EDA Host Pilot Results

• Performance results from the client and the
  actual equipment show equipment CPU utilization
  is higher than the client as seen on the right
• When using mid (100 parameters) size data
  collection plans both the client and equipment
  behave normally

• Performance results from the client and the
  actual equipment show equipment behavior is
  tracked by the client
• Spikes were found in the transmission of data
  caused when other processes started while the
  equipment was collecting data as seen at left

7
EDA Client Performance Results

• Client performance for 10 separate connections
  with data collection of 10,000 parameters/sec
• Client was capable of handling up to 50,000
  parameter at 60 CPU usage
• Data source drives client CPU usage
• There are limits on the number of connections a
  single client can handle

8
Factory Side Lessons Learned

• Three levels of commercial support
• Complete application
• Partial functionality
• Interface drivers only
• Users are deciding what level of turn key
  solution desired
• Mixture of central and local applications are
  expected for both factory level and ad hoc data
  collection
• Equipment implementations can cause client issues
  during data collection
• When collecting at high data rates
• When collecting large amounts of data
• When multiple clients are collecting data from
  the equipment
• When other applications run in the same computer

9
EDA Client Lessons Learned

• Metadata content
• Missing elements in the metadata definition in
  the equipment caused the host application to
  reject messages
• Alarms without severity or parameters missing
  elements
• Connectivity and client server installation
• Physical location, computer size, noise, and
  accessibility of server
• Learning curve to operate new client
• GUI, commands, configurations, screens,
• Time synchronization between client and equipment
  critical to analyze results
• Re-synchronize the equipment with the client
  every hour or the results could not be analyzed
• Due to equipment computer clock drift

10
IC Maker Data Quality Issues

• Issues Experienced
• Equipment interface has software defects
• Documentation does not match software installed
• Reports send wrong format of data
• Equipment processing affected by data collection
• Short data collection period with a high rate can
  take tool down
• Missing data, including context data
• Reporting inaccurate data
• Data report and event latency
• Unclear level of data collection limits
• Not clear what the equipment is capable of
• Context data missing, e.g., linking data with a
  wafer
• Non-existent time synchronization
• Data reports cannot be used to correlate events,
  alarms, or process issues
• Inconsistent point of time-stamping

Time Synchronization Time-Stamping -gt Data
Quality
11
Data Quality Scope

• Goal
• Define Data Quality Guidelines and terminology to
  guide equipment suppliers, 3rd party software
  suppliers, and IC makers to measure, produce, and
  verify data quality
• Scope
• Data used for purposes of collection, control,
  configuration, and diagnostics, that is, control
  variables/signals, sensor values, events, and
  equipment health monitoring data available via
  SECS/GEM, EDA interface, or other standard
  interfaces
• Data Quality Guidelines specified in this
  document are associated with the following

• Data Attributes and Characteristics
• Accuracy
• Latency
• Repeatability
• Context
• Content
• Resolution
• Drift
• Sampling Rate
• Jitter

• Commands/Responses
• There is a reply for each a request
• Event reports are send in the right order
• Data reported is correct (Traceability)
• Events are accurate
• Message is well formed
• Alarm includes data

12
Equipment Data Quality Principles

• Data must be provided with sufficient accuracy,
  resolution, and sampling frequency to allow high
  fidelity extraction of relevant data features for
  process/equipment characterization, fault
  detection, failure diagnosis, and process control
• Event data and context information must be
  complete, consistent, and correct as well as
  reflect the actual time and conditions pertaining
  to the occurrence of the indicated event
• Timely transfer of data is necessary to achieve
  fault interdiction
• A time synchronization solution is the initial
  focus

13
Time Synchronization

• Objective
• Require accurate equipment and factory clock
  synchronization for future implementations, while
  maintaining compatibility with legacy systems
• Goals
• Factory time source
• Factory clock synchronization precision
• Embedded clocks within equipment need to be
  synchronized with factory clock
• Factory applications timing requirements
• Common time-stamp format/basis
• Consistent point of time-stamping
• Synchronization traceability/quality for
  verification
• Facilitate data merging and analysis

14
Time Synchronization Guidelines
To complement the proposed new time
synchronization standard and provide guidance in
establishing a factory time synchronization
architecture for realizing effective data
collection and time-stamping

• Goals
• Facilitate the establishment of effective factory
  time synchronization architecture
• Ensure synchronization quality
• Provide recommendations and methods for meeting
  the upcoming SEMI Time Synchronization standard
  requirements
• Guidance on data time-stamping for ensuring data
  quality
• Guidance on current and upcoming time
  synchronization accuracy requirements

15
Application Timing RequirementsOn-tool Examples
16
Time Synchronization and Time-Stamping Guidelines
Overview

• Time Synchronization Architecture
• UTC time scale
• Reference time source and traceability
• NTP strata and modes
• Fault tolerance
• Time Synchronization Quality
• Clock quality
• Synchronization frequency
• Security
• Monitoring
• Accuracy and Precision Requirements
• Current and upcoming requirements based on
  working group discussions
• General agreement is about 1 ms accuracy for
  e-Manufacturing applications
• Effective Time-Stamping
• Timestamp format and time base
• Point of time-stamping

17
Implementer Roles Responsibility
18
Time Synchronization Traceability
The reference UTC source should be traceable to
either an international or national time
reference standard

• A measurement is traceable if it is made based on
  a global or national standard reference and the
  uncertainty of the measurement is known and
  documented
• International
• Bureau International des Poids de Mesures (BIPM)
• National
• NIST maintains an ensemble of cesium oscillators
  to provide a UTC source, which can serve as both
  a time and frequency reference
• NICT (National Institute of Information and
  Communications Technology) is Japan's official
  institution responsible for the national
  frequency and time standards

http//www2.nict.go.jp/cgi-bin/JST_E.pl
19
Time-Stamping Data
Time-stamps coupled with a piece of data shall
reflect the time the data was measured, generated
as closely as possible to provide a sufficient
level of data quality.

• Include accuracy information on the time-stamp,
  including estimated clock deviation from UTC and
  estimated time-stamping latency from when the
  event occurred to when it was actually
  time-stamped

Data Generation
Subsystem
Sensor
Data Collection System
Tool
Data Packaging
Data Queue
20
SEMI E148 Time Synchronization Standard

• What is specified in the new standard?
• Equipment and factory applications shall support
  synchronization using NTP
• Time-stamp reporting format (ISO 8601)
• Format to be used by other standards
• Defines a clock object (TS-Clock)
• To readily obtain time information from equipment
• Method to query date/time
• Accuracy/precision of internal clock
• Time synchronization status
• Uses internal clock
• Accessed via SECS-II or EDA

21
SEMI Standards Planned Updates
SEMI E148 timestamp format among all standards
Reference time synchronization standard
References new timestamp format
Equipment
FICS
Control Client
E5 SECSII
E30 GEM
E40 PM
Data Client
E134 EDA
Sensor clock synchronization and time-stamping
First standard to use the format
E54 SAN
Specify time accuracy and reference E148 clock
object
E116 EPT
Sensors
E127 Integrated Metrology Module
References new timestamp format
22
Data Quality Guidelines

• Data reported by the equipment including event,
  alarms, and data collection shall satisfy the
  following requirements
• Equipment Data Availability Documentation and
  Report
• Equipment Data Collection Report Reliability
• Data Interpretability Transfer from Equipment
• Reported Data Accuracy and Resolution
• Reported Data Sample Period and Throughput
• Data Report Latency and Buffering
• Data Sampling Period Uniformity and Skew
• Equipment Reported Data Management
• Data Time Stamping
• Data Reporting Units and Magnitude

23
Equipment Data Availability Documentation and
Report

• The equipment supplier shall document and report
  all equipment data and its associated parameters
  for the purpose of data collection, control, and
  configuration
• Information shall include data attributes,
  limitations, and how to access that information
• Who Implements Equipment suppliers
• Who Uses Equipment suppliers and device makers
• Standards TBD
• Remarks
• For accessibility purposes, the messages to be
  sent and received from the equipment depend on
  the interface used to communicate with the
  equipment
• IC makers will evaluate potential data
  availability with respect to this documentation.
  IC makers and suppliers will evaluate actual data
  availability by testing against this
  documentation
• In some cases the supplier should document what
  data is not available and why
• If data is used to control the equipment then it
  should be available external to the equipment
• Data Collection and analysis enables
  understanding data and its process impact
• Documentation of all data elements available from
  the equipment

24
Reported Data Resolution and Accuracy

• Data shall be provided with an absolute accuracy
  of no less than .05 of full scale or .05 of
  normal operating range, whichever is finer
  resolution
• Who Implements Equipment Supplier
• Who Uses Equipment Supplier, 3rd Party
  Applications, and ICMs
• Remarks
• Parametric data must have sufficient resolution
  and accuracy in order to distinguish signal noise
  from potentially subtle trends in the data.
  Typical order of magnitude of process tolerances
  are in the single digits less than 10. Assuming
  1 process sensitivity, a resolution one order of
  magnitude lower is required, or 0.1
• Data must be provided with a resolution of no
  less that 0.1 of full scale or 0.1 of normal
  operating range, whichever is finer resolution
• For example, for a sensor that has a full scale
  operating range between 0 and 12 volts, a
  resolution of at least 0.012 volts is needed. If
  the normal operating range is between 9 and 11
  volts (a 2 volt range), a resolution of 0.002
  volts is needed.
• A resolution of 0.1 requires an absolute
  accuracy of 0.05 is required for the least
  significant bit of the data to have meaning

25
Data Quality Standard Approach

• Accessibility
• First step for assessing data quality, only
  pursued if data is relevant and available to the
  user
• Interpretability
• Drives the depth of data quality assessment
• Will determine method and depth of data
  assessment
• Usability
• Once the data is usable it can be assess for
  additional data attributes such as latency,
  sampling rate, resolution, accuracy, etc.

26
Usability Data Quality Attributes

• Formal definitions and common language seek to
  allow open communication between supplier and
  consumer of data when addressing data quality
  issues

27
Equipment Data Quality Issues
28
Data Management Key Messages

• Data producers are the equipment suppliers they
  are the only ones that can improve data quality
• Data consumers are IC makers, who are at the
  mercy of the producers the quality of the
  decisions made is wholly dependent on the quality
  of the data
• Semiconductor equipment generates data critical
  to improving equipment and factory productivity
• Data must be provided with sufficient accuracy,
  resolution, and sampling frequency for process/
  equipment characterization, fault detection,
  failure diagnosis, and process development/control
  
• SEMI Task Force defining data quality standard
• Contact harvey.wohlwend_at_ismi.sematech.org

Garbage in, garbage out
29
Summary
EDA will bring a new generation of factory
applications and capabilities

• Implementation of EDA requires cooperation
• Equipment suppliers, third-party software
  suppliers, factory integrators, and IC makers
• OEM/IC maker cooperation will lead to success and
  profitability
• ISMI members froze standards deployment at 1105
  level
• e-Manufacturing needs higher volume of quality
  data
• Process learning, tracking, and control requires
  fine grained data
• Time Synchronization standard defined
• Data Quality standard on-going effort
• Guidance and tools provided to the industry by
  ISMI
• Guidance Documents, EDA Evaluation Method, Data
  Quality Evaluation Method

30
EDA References

• SEMI standards E120, E125, E128, E132, E134,
  E138
• www.semi.org/standards
• e-Diagnostics Guidebook, v2.1 ismi.sematech.org/do
  cubase/document/4153deng.pdf
• EEC Guidelines, v2.5 ismi.sematech.org/emanufactur
  ing/docs/eecguidebook.pdf
• EEC High-level Requirements for APC
  ismi.sematech.org/emanufacturing/docs/EECReqs.pdf
• EDA Usage Scenarios, 04104579B-TR
• ismi.sematech.org/docubase/abstracts/4579btr.htm
• ISMI EDA Evaluation Method V3.0
• ismi.sematech.org/docubase/abstracts/4664atr.htm

31
Data Quality/Time Synch References

• SEMI Standard
• E148 Time Synchronization and Definition of the
  Clock Object
• www.semi.org/standards
• ISMI Reports
• Data Quality Guidelines
• ismi.sematech.org/docubase/abstracts/4843beng.htm
  
• Factory and Equipment Clock Synchronization and
  Time Stamping Guidelinesismi.sematech.org/docubas
  e/abstracts/4781aeng.htm
• Using Network Time Protocol (NTP) Introduction
  and Recommended Practicesismi.sematech.org/docuba
  se/abstracts/4736aeng.htm
• Semiconductor Factory and Equipment Clock
  Synchronization for e-Manufacturingismi.sematech.
  org/docubase/abstracts/4557aeng.htm
• ISO 86012004
• Internet standard for date/time format
• www.w3.org/TR/NOTE-datetime.html
• NTP
• www.ntp.org