Analyzing event data: Where Does All the Time Go?

www.cimetrix.com
14th European Advanced Process Control
And Manufacturing (APC/M) Conference
Rome, Italy
Alan Weber
Cimetrix, Incorporated
Analyzing Event Data:
Where Does All the Time Go?
1

 Background and motivation
 Why events?
 Product Time Measurement (PTM)
 Methodology extensions
 Future vision
Outline
2

 APC applications have traditionally focused on analog
data
 Metrology results
 Recipe settings
 Equipment trace data
 Device performance parameters
 Yield statistics
 Event data analysis limited to
 Trace data “framing” for FDC
 Feature extraction for EHM/PHM (health monitoring and
predictive maintenance)
 Product movement for WTW (Wait Time Waste)
Background
3

 Cycle time is one of the most important metrics for
semiconductor factory performance
 Most factories cannot identify and quantify all the
components of cycle time
 We believe this is a problem worth solving, and have
worked closely with SEMATECH to do this
Motivation
4
Delivery
Performance
Revenue
Cycle Time
OEE
Capacity

 When something happened
 How long something took
 How often something happened
 How long something waited for something else
 And with a little more work…
 Historical statistics on all the above
 When something similar might happen again
 How different events seem to be related
Why events?
What they can the tell us…
5
“The only reason for time is so that everything
doesn’t happen at once.” - Albert Einstein

 Timestamp – universal time, high resolution
 Event name – unique text string
 Description – unambiguous meaning
 Also…
 Alarms – special cases (bi-state: Set/Clear)
Event data characteristics
SEMI Standards – the basics
6

 Source (who/what generated it)
 Related material, if any (substrate, lot)
 Location (where it happened)
 Process/recipe/step (what else was going on)
 Trigger conditions (why then)
 Destination, if any (who wants to know)
 Other information – purpose-specific
Context data
Making events truly useful
7

 Equipment
 Direct, indirect (related to a tool)
 Material
 Product, consumables, fixtures
 People
 All manufacturing roles
 Factory applications
 Wide variety… and growing
Broad range of event data sources
Interesting things are happening all the time!
8

 Deduce them from other sources!
 Messages (“Gentlemen, start your engines”)
 State variable changes (Pump Off / On)
 Condition testing (When V >= V0)
 Boolean logic (If A and B happen, then C)
 Timers (1.45 seconds after D, trigger E)
 Operator commands (“Do it now!”)
 …
But what if…
…the events I need are not available?
9

Equipment capability monitoring
Using counter value changes as events
Gate valve open-to-
close time (precedes
change in WaferID)
WaferID of
interest
GateValveCloseCount
GateValveOpenCount

11
Product Time Measurement (PTM)
Methods and Metrics

SEMATECH WTW project objective
12
• Objective: Develop methods and metrics to measure the
time a product spends active and waiting to identify time
waste
• Cover the whole manufacturing life cycle of the product
‒ Contiguous lot-level and substrate-level time elements
• Use event-based time elements
‒ Each time element begins and ends with a time-stamped event
• Extract data from current SECS-II communication logs
‒ Non-disruptive to production, no new messages required
• Demonstrate data extraction, transformation, and
visualization
• Develop into industry standard
Breadth
Focus

How Much Time Is Wasted?
13
Time required for all operations – “Total Cycle Time”
How much of your Total Cycle Time is spent waiting?
July Aug Sep
500+ Operations

Product Time Measurement
Generating TimeElements
14
Etch
Event
Processor
$avings
SECS
Messages
Material Movement
Events

WTWRI Process Flow
How it works…
15
Step 2
Step 1
Step 3
Factory
Message
Logs
IF1
IF2
WTW
Data
Message
Filters
Time Element
Definitions
Factory
Applications
WTW
Analysis
WTWRI
Visualization Charts, Graphs,
Reports
Step 0
SF File
SML Dialect
Definition
Event Report
Definitions
Supplier Data
Dictionaries
Standard Event
Mappings
WTWRI
Configuration.xml
(a single file)
15

Time Element definition template
Shows standard event list for each element
16

Configuration Process
TimeElement definition (Start and Stop events)
17
 Select active version
of Time Element set
among available
alternatives
 Associate start and
end event(s) with
each Time Element
 Set other attributes
for analysis (index,
level, wait/active,
etc.)

Results Files
WTWData (TimeElements)
18
• This file/tab has one row per TimeElement, sorted by Start Event Time
• Fields include context data useful for sorting/filtering (lot, substrate, loc)
• Calculations include TimeElement Duration, and Wait/Active designation

TimeElement summary statistics
For a given substrate, lot, or timespan
19
• Generated by Excel macros from SubstrateSummary file

TimeElement summary statistics
For given substrates, lots, or timespans
20
• Generated with Excel charts from SubstrateSummary file

More TimeElement Visualization
State, Location, and TimeElement (per wafer)
21
Note cursor positions….

New Visualization Possibilities
Material Location statistics (between cursors)
22

Raw Event Visualization
Event sequence for each Substrate
23
StandardEventName
(sequence)SubstrateID
StandardEventName
at red cursor location

Etch
Full production capability
24
Event
Processor
Etch
Etch
A
M
H
S
Track
Scanner
Event
DB
Event
Generator
Factory
DBs
Result
DB
Job
Manager
Equipment
Modeler

25
Extensions of
Methodology

 Fleet-based variability
 Comparing results across a group of identical (or similar) equipment
 Timespan-based variability
 Comparing results for a single equipment over some period of time
 Context
 Set of parameters used to group results in a way that they can be
meaningfully compared
Variability analysis
Key concepts
Etch
Etch
Etch
Scanner
Context Vector
 Process ID
 Recipe ID
 Exp type

 Recipe step
 A contiguous span of time with a process recipe in which equipment settings
are held constant
 May or may not correspond to changes in the value of a recipe step counter
variable
 Feature
 Parameter (direct or derived) of interest during a recipe step
 Context
 Set of parameters used to group results in a way that they can be
meaningfully compared
Recipe step-level timing analysis
Key concepts

 Equipment component
 Element of a tool that provides some useful function (often a field-
replaceable unit)
 Component cycle
 Sequence of individual operations that the component executes to achieve
its function
 Command/response signals
 Parameters that indicate when a component cycle is supposed to begin and
when it is complete, respectively
Time-based fingerprinting
Key concepts

Application integration support
Key concept – state machine alignment
29
SUBSTRATES
SUBSTRATE LOCATIONS
LOAD PORTS

Identify where all the time goes
in your factories
Quickly and automatically
Using equipment and factory data
from any available source
Future Vision
30

Analyzing event data: Where Does All the Time Go?

More Related Content

Similar to Analyzing event data: Where Does All the Time Go? (20)

More from Kimberly Daich (12)

Recently uploaded (20)

Analyzing event data: Where Does All the Time Go?