Performance Evaluation of Serial Photolithography Clusters: - - PowerPoint PPT Presentation

▶
performance evaluation of serial photolithography clusters
SMART_READER_LITE
LIVE PREVIEW

Performance Evaluation of Serial Photolithography Clusters: - - PowerPoint PPT Presentation

Aug 13, 2023 353 likes •929 views

IEEE/SEMI Advanced Semiconductor Manufacturing Conference Performance Evaluation of Serial Photolithography Clusters: Queueing Models, Throughput and Workload Sequencing James R. Morrison Central Michigan University Department of Engineering

slide-1
SLIDE 1

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

1

Performance Evaluation of Serial Photolithography Clusters:

James R. Morrison Central Michigan University Department of Engineering and Technology Beverly S. Bortnick Donald P. Martin IBM, Systems and Technology Group

Queueing Models, Throughput and Workload Sequencing

slide-2
SLIDE 2

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

2

Presentation Overview

  • Basic serial processing cluster tool model
  • Queueing models and the mean cycle time
  • Throughput and common practical events

– Empty modules – Completely idle cluster – Lot specific processing times

  • Concluding remarks
slide-3
SLIDE 3

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

3

Serial Processing Cluster Tools

Wafers Enter

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9 m16

Wafers Exit Wafer Handling Robot

  • Wafers arrive in lots containing W wafers each
  • M processing modules each with process time = D
  • Wafers proceed sequentially from one module to the next
slide-4
SLIDE 4

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

4

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D
slide-5
SLIDE 5

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

5

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

1

Time:

slide-6
SLIDE 6

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

6

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1

Time: D

slide-7
SLIDE 7

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

7

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3

Time: 2D

slide-8
SLIDE 8

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

8

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4

Time: 3D

slide-9
SLIDE 9

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

9

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5

Time: 4D

slide-10
SLIDE 10

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

10

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6

Time: 5D

slide-11
SLIDE 11

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

11

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7

Time: 6D

slide-12
SLIDE 12

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

12

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8

Time: 7D

slide-13
SLIDE 13

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

13

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1

Time: 8D

slide-14
SLIDE 14

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

14

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2

Time: 9D

slide-15
SLIDE 15

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

15

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2 3

Time: 10D

slide-16
SLIDE 16

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

16

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2 3 4

Time: 11D

slide-17
SLIDE 17

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

17

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2 3 4 5

Time: 12D

slide-18
SLIDE 18

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

18

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2 3 4 5 6

Time: 13D

slide-19
SLIDE 19

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

19

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 1 3 4 5 6 7 8 1 2 3 4 5 6 7

Time: 14D

slide-20
SLIDE 20

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

20

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

MD Time: 15D

First wafer exits the tool

slide-21
SLIDE 21

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

21

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

3 4 5 6 7 8 1 2 3 4 5 6 7 8 1

Time: 16D

slide-22
SLIDE 22

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

22

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

4 5 6 7 8 1 2 3 4 5 6 7 8 1 2

Time: 17D

slide-23
SLIDE 23

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

23

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

5 6 7 8 1 2 3 4 5 6 7 8 1 2 3

Time: 18D

slide-24
SLIDE 24

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

24

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

6 7 8 1 2 3 4 5 6 7 8 1 2 3 4

Time: 19D

slide-25
SLIDE 25

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

25

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

7 8 1 2 3 4 5 6 7 8 1 2 3 4 5

Time: 20D

slide-26
SLIDE 26

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

26

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

8 1 2 3 4 5 6 7 8 1 2 3 4 5 6

Time: 21D

slide-27
SLIDE 27

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

27

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7

(M+W-1)D Time: 22D

First lot is complete

slide-28
SLIDE 28

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

28

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

Time: 23D

slide-29
SLIDE 29

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

29

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

3 4 5 6 7 8 1 2 3 4 5 6 7 8 1

Time: 24D

slide-30
SLIDE 30

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

30

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

4 5 6 7 8 1 2 3 4 5 6 7 8 1 2

Time: 25D

slide-31
SLIDE 31

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

31

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

5 6 7 8 1 2 3 4 5 6 7 8 1 2 3

Time: 26D

slide-32
SLIDE 32

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

32

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

6 7 8 1 2 3 4 5 6 7 8 1 2 3 4

Time: 27D

slide-33
SLIDE 33

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

33

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

7 8 1 2 3 4 5 6 7 8 1 2 3 4 5

Time: 28D

slide-34
SLIDE 34

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

34

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

8 1 2 3 4 5 6 7 8 1 2 3 4 5 6

Time: 29D

slide-35
SLIDE 35

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

35

Basic Serial Cluster Tool Behavior

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9

Serial cluster tool with:

  • M = 15
  • W = 8
  • Process time per module = D

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7

Time: 30D WD time

units after last lot end

Next lot is complete

slide-36
SLIDE 36

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

36

Basic Serial Cluster Tool Behavior

For a serial cluster tool with:

  • M modules
  • W wafers per lot
  • Process time per module = D

75 . 2 1 1 ||*           W M

Example: M = 15, W = 8

D  D    22 ) 1 ( W M P

D  D  8 W T

P is 2.75 times as large as the time between lot completions

Total process time for each lot: Time Between Lot Completions: Maximum parallelism:

 D

   1 W M P D W T

             W M W W M T P 1 1 1 : ||*

slide-37
SLIDE 37

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

37

Queueing Models

  • Consider that lots of W wafers arrive as a Poisson process of rate a
  • When the first module is empty an available lot may begin service
  • A variant of the Pollaczek-Khinchin formulae for batch arrivals

reveals the mean waiting time until a batch (lot) of wafers enters service

  • Where r  a D is the loading on the system

m1 m2 m3 m4 m5 m6 m7

Lots arrive as a Poisson process

  • f rate a

 

r r  D  1 2 W Time Waiting

slide-38
SLIDE 38

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

38

Mean Cycle Time

  • Mean cycle time for a serial

processing cluster tool

 

P          r r 1 1 1

*

|| 2 1

  P

        r r 1 1

2 1

Mean Cycle Time Comparison: M/D/1 and Cluster Tool

10 20 30 40 50 60 70 80 0.2 0.4 0.6 0.8 1

System Loading (Utilization of Capacity) Mean Cycle Time M/D/1 Mean Cycle Time Cluster Mean Cycle Time (M = 15, W = 8, D = 1, ||* = 2.75)

  • P = WD and r  a D
  • Mean cycle time for an M/D/1

queue (a cluster with one module)

  • P = (M+W-1)D and r  a D
slide-39
SLIDE 39

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

39

Normalized Mean Cycle Time

  • It is common to normalize by

the process time P, let r  a D

  • Cluster tool normalized mean

cycle time

  • M/D/1 tool normalized mean

cycle time

 

         r r 1 1 1

*

|| 2 1

 

        r r 1 1

2 1

Normalized Mean Cycle Time Comparison: M/D/1 and Cluster Tool

1 2 3 4 5 6 7 8 9 10 0.2 0.4 0.6 0.8 1 System Loading (Utilization of Capacity) Normalized Mean Cycle Time

M/D/1 Normalized Mean Cycle Time Cluster Normalized Mean Cycle Time

(M = 15, W = 8, D = 1, ||* = 2.75)

slide-40
SLIDE 40

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

40

Maximum Throughput

  • Once all of the processing modules are full one lot exits every

WD time units

  • Maximum throughput:
  • Events disrupting the steady exit of wafers may reduce

throughput

D  D   1 W W unit time wafers

*

m1 m2 m3 m4 m5 m6 m7 m8 m15 m14 m13 m12 m11 m10 m9 2 1 3 4 5 6 7 8 1 2 3 4 5 6 7

slide-41
SLIDE 41

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

41

Throughput and Empty Modules

  • Empty modules between lots

– Absence of lots to run – Failure to load a lot promptly – Delays internal to the tool

  • Delay between first lot completion and second lot

m1 m2 m3 m4 m5 m6 m7 m15 m14 m13 m12 m11 4 5 6 7 8 1 2 3 4 5 6 7 Empty modules

Throughput is reduced

slide-42
SLIDE 42

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

42

Throughput and Empty Modules

  • Let tem denote the average

time lost between each lot

  • Throughput with empty modules
  • Example:

– W = 8, D = 1 minute, – Mean lost time tem = 3 minutes

         

DW

em

t

  1 1

*

73 . 1 1

8 1 3 *

          

 

Throughput in the Presence

  • f Empty Modules

50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% 1 2 3 4 5

Mean Time Lost to Empty Modules Per Lot (Minutes) Percent of Maximum Throughput Achieved

slide-43
SLIDE 43

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

43

Throughput and Idle Tool Events

  • A completely idle tool (“flush and fill event”) occurs

– Following maintenance/repair activities – Lack of product

  • Wafers are not completed until the tool is once more filled

Throughput is reduced

time Production

  • ut

Wafers  

slide-44
SLIDE 44

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

44

Throughput and Idle Tool Events

  • Let fit denote the fraction of lots

facing a completely idle tool

  • Throughput with idle tool events
  • Example:

– ||* = 2.75 (W = 8, M = 15) – Lots facing idle tool fit = 0.2

 

1 || 1

* *

  

it

f  

 

74 . 1 75 . 2 2 . 1 1

*

     

Throughout with Idle Tool Events

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 20% 40% 60% 80% 100%

Fraction of Lots Facing an Empty Cluster Percent of Maximum Throughput Achieved

slide-45
SLIDE 45

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

45

Lot Specific Module Speeds

  • Different lots may have different processing speeds
  • Suppose the lot in module B dictates the rate of wafer

movement

  • A fast lot (green) following a slow lot (red) cannot proceed as

quickly as is normal

Module B

slide-46
SLIDE 46

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

46

Lot Specific Module Speeds

  • After some time, the slow lot (red) relinquishes control of

module B

  • Even though the slow lot (red) is still in the tool, the pace of

production is dictated by the fast lot (green)

Module B

slide-47
SLIDE 47

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

47

Throughput and Workload Sequence

  • For simplicity, restrict attention to two classes of lots

– Fast have module process time = DF – Slow have module process time = DS

  • Without interaction, the mean module process time would be

– where fS and fF are the fraction of lots that are slow and fast, respectively

  • Assuming no other losses, the throughput without interaction

would be

F F S S

f f D  D  D

D  1 

slide-48
SLIDE 48

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

48

Throughput and Workload Sequence

  • With two classes of lots

– Fast have module process time = DF – Slow have module process time = DS – Fraction of slow lots (out of all lots) that precede a fast lot = fSF – Fraction of fast lots (out of all lots) that precede a slow lot = fFS

  • The throughput with sequencing interaction is given as
  • This expression generalizes for more than two classes of lots

  

SF FS F S

f f W B M  D  D       D    1  

slide-49
SLIDE 49

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

49

Invariance to Workload Sequence

  • If the tool never idles:
  • Each fast-to-slow has one slow-to-fast following it (fFS = fSF)
  • For continuous production with two classes of lots
  • If the tool does idle, throughput does depend on the sequence

  

Time Wafers exiting the tool fast to slow fast to slow slow to fast

slide-50
SLIDE 50

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

50

Example of Photolithography Cluster Analysis

  • A photolithography cluster tool has the following parameters

– Processing is received from 21 modules (M = 21) – some multipath – Typical lots have 25 wafers per lot (W = 25) – The bottleneck module process time is 45 seconds (D = 45 s)

  • Fundamental parameters for the cluster include

– Process time P = (M+W-1)D = 33.75 minutes – Time between lot exits T = WD = 18.75 minutes – Parallelism ||* = P/T = (M+W-1)/W = 1.8

  • Maximum throughput for the cluster

hour wafers 80 0125 . 1 45 1 1

*

   D  s 

slide-51
SLIDE 51

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

51

Example of Photolithography Cluster Analysis Continued

  • Assuming a random (Poisson)

lot arrival process

  • The parallelism is ||* = 1.8
  • Normalized mean cycle time

behavior is given as

        r r 1 722 . 1

Normalized Mean Cycle Time Comparison: M/D/1 and Cluster Tool

1 2 3 4 5 6 7 8 9 10 0.2 0.4 0.6 0.8 1 System Loading (Utilization of Capacity) Normalized Mean Cycle Time

M/D/1 Normalized Mean Cycle Time Cluster Normalized Mean Cycle Time

slide-52
SLIDE 52

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

52

Example of Photolithography Cluster Analysis Continued

  • If the cluster experiences idle tool and empty module events
  • One in five lots are started on a idle tool (maintenance and idle)
  • Remaining lots average four empty modules between them

 

862 . 16 . 1 1 1 || 1 1 Throughput Achieved

* *

    

it

f 

886 . 128 . 1 1 1 1 Throughput Achieved

*

           

DW

em

t

slide-53
SLIDE 53

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

53

Example of Photolithography Cluster Analysis Continued

  • Two classes of lots

– 50% are fast lots with DF = 30 seconds – 50% are slow lots with DS = 60 seconds – Module 5 dictates the pace of the wafer flow (B = 5) – Fast preceding slow lots fFS = 0.3 – Slow preceding fast lots fSF = 0.1 – The other 50% either follow a like lot or precede an idle tool

s 45  D

  

904 . 1 1 Throughput Achieved

*

  D  D       D   

SF FS F S

f f W B M 

slide-54
SLIDE 54

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

54

Example of Photolithography Cluster Analysis Continued

  • A photolithography cluster tool has the following parameters

– Processing is received from 21 modules (M = 21) – some multipath – Typical lots have 25 wafers per lot (W = 25) – The bottleneck module process time is 45 seconds (D = 45 s)

  • Approximately (ignoring interactions between the events), the

throughput is reduced

– 1/5 lots face an idle tool  PPH (0.862) * – Four empty modules  PPH (0.886) (0.862) * – Different D’s (B = 5)  PPH (0.904)(0.886) (0.862) *

  

pph 55 (0.690) Throughput Achieved e Approximat

* 

  pph 80 Throughput Ideal

* 

 

Over 30% reduction

slide-55
SLIDE 55

IEEE/SEMI Advanced Semiconductor Manufacturing Conference

May 22-24, 2006 ASMC 2006 – Boston, Massachusetts

55

Concluding Remarks

  • Serial processing cluster tools can serve to model clustered

photolithography tools

  • Queueing models reveal mean cycle time performance
  • Throughput may be dramatically reduced by common events

– Empty modules – An idle tool – Diversity of processing times between lots

  • Examples of the methods
  • Future directions: Combination of the events and optimization