SEMICON West 2006 STEP Methods to Measure/Improve Equipment Productivity

1
SEMICON West 2006 STEPMethods to Measure/Improve
Equipment Productivity

• SEMI Equipment Performance Standards Integration

Sal DiIorioSemi-Tech Group sdiiorio_at_semitechgroup
.com
2
Equipment Performance Metrics Process

• Factory sources (automated and/or human) provide
  the inputs
• Standards define the process and equations
• Metrics are output consistently
• Same meaning for all metrics regardless of
  location

Metrics
Factory Data
SEMI Standard
3
Factory Data Sources

• Automated Systems
• Production data, state history
• Equipment Maintenance and state history
• Equipment Cell Controllers
• Equipment E116 or E58 state information
• Non-automated Systems
• Manual state change histories
• Manual input to support E58
• Other state and history logs

4
Eq. Performance Standards Relationship
5
E10 RAM

• Central to all Equipment Metrics Standards
• First developed in 1986 and continuously improved
  to meet new requirements of the semiconductor
  industry
• Defines the basic equipment states and metrics
  which act as input to other performance metrics
• E79 Overall Equipment Efficiency
• E35 Cost of Ownership
• E124 Factory Level Productivity

6
E10 Breakdown of Time
7
E10 Process
8
E116 Process
9
Mapping the States
NON-SCHEDULED (MES) AND ANY STATE
(EPT) MANUFACTURING (MES) AND BLOCKED
(EPT) ENGINEERING (MES) AND BLOCKED
(EPT) UNSCHEDULED DOWN (MES) AND ANY STATE
(EPT) SCHEDULED DOWN (MES) AND ANY STATE
(EPT) ENGINEERING (MES) AND BUSY
(EPT) ENGINEERING (MES) AND IDLE
(EPT) MANUFACTURING (MES) AND IDLE
(EPT) MANUFACTURING (MES) AND BUSY (EPT)
E10 States
NON-SCHEDULED TIME
UNSCHEDULED DOWNTIME
SCHEDULED DOWNTIME
ENGINEERING TIME
STANDBY TIME
PRODUCTIVE TIME
ANY STATE (EPT) means that the EPT state is not
a factor in determining the E10 state.
10
E58 Process
11
E58 vs. E10 State Examples

• Equipment tracks its own E58 state...
• But it doesnt know its own E10 state details
• User input is required, in order to provide
  accurate data

12
E58 How it works
Information exchange regarding equipment downtime
state and material type.
MES
Operator input to trigger E58 state changes.
E58 state transition Event reports.
Equipment
Host Computer
Host input to trigger E58 state changes.
Metrics
13
Now that we have E10 Metrics, what do we do with
them?

• Continuous
• Improvement
• Activities
• Lean Mfg
• TPM
• RCM
• 6 Sigma
• Rel Eng
• Etc.

14
Using E10 Metrics

• Use them in purchase acceptance specifications
• MTBFp, Uptime, MTTR, etc.
• Monitor our own performance and drive continuous
  improvement activities
• Look for improvements in metrics when new
  procedures or processes are implemented.
• Benchmarking functions with other companies or
  between factories
• Use them as inputs to other SEMI Standards for a
  better understanding of equipment COO or
  productivity

15
SEMI E35 COO

• COO The full cost of embedding, operating, and
  decommissioning in a factory environment
  equipment needed to accommodate the required
  volume of production units.
• Among the many inputs used by E35, two come from
  E10
• Operational Uptime
• E10 metrics (MTBF, MTTR) to calculate maintenance
  labor hours required for scheduled and
  unscheduled downtime

16
E79 OEE

• OEE calculations are stated in terms that are
  consistent with SEMI E10.
• OEE Availability Efficiency x Operational
  Efficiency
• x Rate Efficiency x Quality Efficiency
• Where E10 directly provides the values for
• Availability Efficiency Equipment Uptime /
  Total Time
• Operational Efficiency Production Time /
  Equipment Uptime
• Rate Efficiency Theoretical Production Time for
  Actual Units
• 
  / Production Time
• Quality Efficiency Theoretical Production Time
  for Effective Units / Theoretical Production Time
  for Actual Units
• Notes
• Production Time is not defined in E10, but
  happens in Productive Time.
• Rate Efficiency x Operational Efficiency
  Performance Efficiency

17
The Relationship Between OEE and E10
18
E79 Process
In addition to OEE several other efficiencies are
also defined to enable users to assess more
specific aspects of equipment productivity.
Other data can come from MES or other Factory
sources.
19
E10 States vs. E79 Productivity Losses
20
What is needed to determine E10 State?
21
What is needed to calculate E10 Metrics?
22
Automation vs. Manual Data Collection

• What data do people capture worst?
• Transition from STANDBY to PRODUCTIVE
• Transition from PRODUCTIVE to STANDBY
• Transition to UNSCHEDULED DOWNTIME
• Wafer/Cycle Counts
• Failures
• Cluster Tool Module level states

• What data do people capture best?
• Transition to and from SCHEDULED DOWNTIME
• NON-SCHEDULED time
• Reason for Failures
• Equipment dependent vs. non-equipment dependent
• Transition and reasons for Maintenance Delay

23
Automation vs. Manual Data Collection

• What data does automation capture worst?
• Transition to and from SCHEDULED DOWNTIME
• NON-SCHEDULED time
• Reason for Failures
• Equipment dependent vs. non-equipment dependent
• Transition and reasons for Maintenance Delay

• What data does automation capture best?
• Transition from STANDBY to PRODUCTIVE
• Transition from PRODUCTIVE to STANDBY
• Transition to UNSCHEDULED DOWNTIME
• Wafer/Cycle Counts
• Failures
• Cluster Tool Module level states

24
Is automation absolutely necessary?

• Automation is superior for the following
• Accurate data regarding Productive / Standby time
• More accurate OEE Operational Efficiency values
• Accurate data about cluster tool module level
  state
• Essential for CT RAM and CT OEE metrics
• Accurate quantitative data about units processed
  and other parametric data for scheduling
  maintenance.
• Some level of automation is absolutely
  preferable, even if limited.

25
Do we really need automation?

• Yes, primarily if
• Detailed information is absolutely required.
• CT RAM and OEE metrics
• The cost of implementing and supporting
  automation can be justified.
• Full automation requires large investments in HW,
  SW and IT and is never done it continually
  needs updates and ongoing commitment of resources
  (people and ).
• Lesser degrees of automation can provide
  significant improvements in data accuracy.
• MES and CMMS (CIM) systems provide excellent
  sources for E10 data.

26
What if full automation is not possible?

• Equipment level E10 metrics can be maintained.
• Daily and Weekly summaries can be analyzed.
• Look for changes and trends
• Equipment to equipment
• Time period to time period
• Simple OEE metric may be possible.
• Even if detailed rate and quality data are not
  available
• COO calculations are still meaningful.

27
Conclusions

• SEMI equipment performance metrics standards work
  together to provide meaningful information about
  the RAM, utilization, productivity, and COO of
  semiconductor equipment.
• While high levels of automation are required for
  the most accurate data and many detailed metrics,
  manual data can still provide useful information.
• For newer, multi-billion dollar factories the
  return provided by small increases in
  productivity may quickly exceed the cost of
  automation.
• Older, smaller factories can achieve meaningful
  results with less automation, as long as absolute
  accuracy is not needed.