1. Semiconductor Equipment Fire Suppression Compliance
for
SESHA North Texas Chapter by
Matt Wyman
Semiconductor Division Manager
KOETTER FIRE PROTECTION
March 22, 2007

2. Koetter Fire Protection
Specialize in Semiconductor Equipment Fire Protection
Global Design & Install CO2 & FWS
Bulk Control Systems for CO2 & FWS
Corrosive Applications (Combustible Plastics)
SEMI Standards Fire Protection Task Force Leader
Specialize in Equipment Design & Integration
Code Review & Compliance Consulting
NFPA/FM Standards
SEMI Standards
Semiconductor Fire Protection Training Seminars
KFP Test & Training Bench
Industry Forms
Industry White Papers
Industry Presentations (SESHA)
7 Texas Locations & International

3. Overview
Are my new and/or existing fire suppression systems compliant?
“Quick” CO2 Design Review
Local Application Nozzles
Fire Detectors
NFPA Changes
Bulk Systems
Hatsuta Cabinex Equipment
Fine Water Spray
Fire Suppression Submittals

4. Fire Codes IFC2006 FM DS 7-7 “Semiconductor Fabrication Facilities”
Combustible Tool >4ft. Wide
Sprinkler Head Inside or Gaseous Fire Suppression
FM DS 7-7 “Semiconductor Fabrication Facilities”
CO2 or FWS (Fine Water Spray)
CO2 is most widely utilized
Extinguish 13kW Fire
NFPA 12 “Carbon Dioxide”
SEMI S14 “Safety Guidelines for Fire Risk Assessment & Mitigation for Semiconductor Manufacturing Equipment”
SEMI S2 “Environmental, Health, & Safety Guideline for Semiconductor Manufacturing Equipment”
Chapter 14 “Fire Protection”

5. CO2 Design Total Flood (TF) Enclosed Compartments
Mini-Environments
Plenums, Head Cases, Chemical Cabinets
50% CO2 Concentration
60 Second Discharge
Compensate CO2 Demand for:
Process Exhaust Ventilation
Damper Violates SEMI S2
Unclosable Openings

6. CO2 Design Local Application (LA) Open Compartments Rate by Volume
Working Surfaces
Rate by Volume
“Assumed” 3D Volume
30 Second Discharge
No Compensation for Exhaust Ventilation & Openings
Combination System (TF & LA)
Typically 30 Second Discharge

7. Existing CO2 Systems “Quick CO2 Review” Typical “Wet Bench”
No Calculations
Check that Supply = Demand
Typical “Wet Bench”
Exhaust Rates = ~125cfm/linear foot (typical)
“Quick” Demand Calcs
TF Demand = 11.5% of Volume (60sec)
Volume = Dimension + Exhaust (60sec)
LA Demand = Average
(8ft = 50lbs, 6ft = 35lbs, 4ft = 20lbs)

8. Quick CO2 Review
Example – 8’L x 4’D x 6’H Enclosed Plastic Bench-100lbs. CO2
Total Volume (TF) = 192cuft x 11.5% = 22.1 lbs.
Exhaust Volume (TF) = 125cfm/ft x 8ft x 11.5% = 115 lbs.
Total Demand (TF) = 137 lbs. CO2
CO2 Supply = NOT ADEQUATE
Example – 6’L x 4’D x 5’H Open Solvent Bench-100lbs CO2
Working Surface (LA) = 35 lbs. (Typical)
Plenum Volume (TF) = 6ft x 4ft x 3ft = 72cuft x 11.5% = 8.3 lbs.
Plenum Exhaust (TF) = 125cfm/ft x 6ft x 11.5%/2 = 43.1 lbs
Total Demand = 86.4 lbs.
CO2 System = ADEQUATE

9. Nozzle Comparisons LA Nozzles TF Nozzles
All have “cone” to absorb discharge pressure
Reflected discharge from orifice
Less exit velocity (same flow rate)
Nozzle absorbs most 600psi discharge pressure
TF Nozzles
“V” Nozzle is “Vent” nozzle (TF)
Straight discharge from orifice
Full effects of 600+psi discharge pressure on tool (especially baths)
Much higher velocity

10. Local Application Nozzles
Working Surface Areas (Wet)
Cone shaped for “Soft” discharge
Cone shape provides larger 2D coverage area
Nozzles pointed downward towards back wall
Provides “rolling” effect across Working Surface
Side View
Front View

11. LA Nozzle Listings/Approvals
All Local Application Nozzles Must Have Coverage Chart
2D Coverage of LA Nozzle
Ansul CR D Nozzle
Ansul D Nozzle
Identical Nozzles
FM Listing = FM testing

12. LA Nozzle Coverage Does this apply to other manufacturers and models?
Ansul
Has Other LA Nozzles
Not for Wet Bench (different coverage chart)
Hatsuta Cabinex-EWT
Has FM Approved LA Nozzle for Wet Bench
Has Own Coverage Chart
Chemetron
Has FM Approved LA Nozzle(s)
Kidde
Not for Wet Bench - (different coverage chart)
What about Kidde “V” Nozzle for LA?

13. Kidde “V” Nozzle Kidde “V” Nozzle from Compliance review
1st - Listed & Approved by UL/FM as TF Nozzle ONLY
No LA Coverage Chart in Design Manual
NOT LISTED/APPROVED FOR USE AS LA NOZZLE
“V”=“Vent” type is designed for “high velocity” discharge to overcome exhaust ventilation.
2nd - CAN CREATE A SEVERE SPLASH HAZARD
Many documented cases where discharge splashed acid baths
Dangerous Exposure to Personnel
3rd - Kidde began manufacturing and listing of Stainless Steel “V” nozzle in June 2005.
Previously only available in brass.
Previous SS installations were not listed/approved components
4th - Not supported by Kidde for LA Coverage

14. Kidde “V” Nozzle
Copy of Letter from Kidde regarding use of “V” Nozzle for LA Coverage
Dated May 26, 2006

15. SS Corrosion Protection
Halar Coating for Stainless Steel Nozzles, Fittings, & Tubing in Corrosive Areas
CO2 Tubing
CO2 Fittings
CO2 Nozzles

16. Detection Optical Flame Detection (13kW Fire) Heat Detection
Solvent Areas
Corrosive Areas
High Exhaust Rate Areas
Heat Detection
Low Exhaust Rates – 400cfm limit
Slower Response = Tolerate Larger Fire
Linear Heat Detection (Corrosive Areas)
Cross-Zoning Flame & Heat Detectors (Asia)
Prevents “False Discharges”
But Negative Affect to Performance

17. Flame Detectors Infrared (IR) & Ultraviolet (UV) Sensors UV IR
Quartz Window
Can be “Masked” by
Gases, Smoke, Fumes IR
Near Band (4.3µ)
Hydrocarbon Fire
Wide Band ( µ)
Visible ( µ)
Sierra DCR1(S)
UV/IR
Cheapest
Leakage Problems HF

18. Flame Detectors Cabinex SX-7000 Det-Tronics PM-5MP Fire Sentry FS7
Triple Band IR (Near)
Reliability?
Chemical Compatibility?
Det-Tronics PM-5MP
Dual Band IR (Near)
Really Fast
False Alarm Issues
Fire Sentry FS7
Near Band, Wide Band™, & Visible IR
Hydrocarbon & Non-hydrocarbon
Alarm & Pre-Alarm Outputs
5-sec Alarm Verification (Flash Fires)
FirePic History
Most Reliable Flame Detector!

19. Flame Detector Reliability
False Alarm Immunity
KFP TEST
Compare Reliability of FS7 vs. PM-5MP to “Steam”
Blackbody IR Radiation
Alarm Signal (False Alarm)
300ºF - 600ºF
6” – 36” Distance
Det-Tronics Technical Note
Alarm from Mini 2” w/o window
Alarm from 1-3” with & w/o window
Alarm from Quartz Halogen 3ft.

20. Spot Smoke Detectors 2 Types
120VAC “First Alert” Spot Smoke Detector
UL Listed as RESIDENTIAL detector
24VDC Fire Alarm System Spot Smoke Detector
UL Listed for use with compatible “cross-listed” FACP
SEMI S2 Section Fire Detection, Alarm, & Controls
&2 All detectors, alarms, and controls be accepted and listed by an accredited laboratory (example UL or FM) and installed in accordance with the terms of that acceptance and appropriate national or international standards (ex. NFPA 72)
Spot Detectors are NOT Compliant (NFPA/SEMI/UL)
Without connection to Fire Control Panel
Spot Detectors require ~2.5% obs./ft. concentration
Not effective in Exhausted or “Open” Compartments

21. NFPA 12 2005 NEW Changes to NFPA 12 2005 Edition
Retroactive upon AHJ
Labels must meet ANSI Z535
Orange “Warning” Color
“Warning” Triangle Symbol
Aspiration Symbol
Outside Protected Space
At Manual Activation
SEMI S1 Compliant

22. NFPA 12 2005 Lockout Valves (Not Required)
Required except where personnel can’t enter space ( )
Where CO2 can migrate to expose personnel ( )
System should “Lockout” during maintenance or testing ( )
“Lockout” Valve should be supervised
Pressure Switch installed between Supply and Lockout Valves ( )
Disable/By-Pass Switch (Required)
Electrically operated systems should have service disconnect switch ( )
Switch should be supervised (keyed)

23. Bulk Systems High Pressure or Low Pressure CO2 or FWS Benefits
Large Supply of Agent to Protect Multiple Systems
Main & Reserve
Controlled Release of CO2
Benefits
Floor Space
Controls
Tool Uptime
Life Cycle Cost
Negative
Higher Install Cost
Major Retrofit or New Fab Project

24. Bulk CO2 Compliance What is wrong with this design?
40 Tools on 1 HP Manifold System
1st - Largest FACP has maximum approval of 10 independent releasing hazards
Solution – Minimum of 4 releasing panels

25. Bulk CO2 Design FACP Approval/Listings
Limited to maximum of 10 independent releasing hazards
10 Tools

26. Bulk CO2 Compliance Anything else? 2nd - Location of Selector Valves
AT EACH TOOL?
Where does CO2 go?
What about phase change in pipe?
How do you Calc this?
CAN’T!!

27. Bulk CO2 Design Selector Valves Must Be @ CO2 Supply Location
Minimize CO2 Loss
Per Listed Hydraulic Calculation Programs

28. HP Manifold Configuration
LCD80 Wet Deck Annunciator
Horn / Strobe Light
Deck
Manual Pull
Keyed By-Pass Switch
STI
CO2 Line
Nozzle(s)
Sensor(s)
Owner: Tech Serv
Fab Level
Selector Valve(s) & Flow Switch(s)
Slab
Owner: Fac Ops
SLC Loop (Device Signals)
Other Decks (10 max)
CO2
Cylinder
Bank
RS485 Comms (Signal Data)
Copper (Comms)
Fiber Network (Comms)
Valve Control Signal
Manifold
Control Panel
Manifold Network Panel(s)
Building Panel Interface
Sub Fab Level

29. HATSUTA CABINEX-EWT FM Approved for Semiconductor Tools
September 2004
Manufactured in Osaka, Japan
Specifically designed for Semiconductor
Total Flood & Local Application
Excellent OEM product!
Packaged cabinet for cylinder(s) & panel
Worldwide Compliance!
CABINEX-EN = NOT COMPLIANT IN USA

30. Cabinex-EWT Control Cabinet Up to 2, 100lb. Cylinders Integrated Panel
50 & 100-lbs.
1.8’ x 1.7’ x 7.0’ Cabinet
Integrated Panel
Up to 32 “Addressable” Inputs
Cross-Zoning Capable
Single Release Zone
Solenoid & Gas Cartridge Actuators
Popular with OEMs
TEL, DNS, AMAT

31. Fine Water Spray (FWS) Securiplex Extinguishing Agent Low Pressure
Wet Bench Application
FM Approved
Extinguishing Agent
Fab DI Water (DIW)
Fab Nitrogen (N2)
Mixed at nozzle
Dual piping
Unlimited Supply
Low Pressure
PVDF or PFA Nozzles & Tubing

32. Submittal Documentation
Stamped Drawings of CO2 Mechanical & Electrical Systems including Isometric
FPE or NICET Level 3
CO2 Hazard Design Demand Calculations
CO2 Hydraulic Flow Calculations
Power Supply & Battery Backup Calculations
Bill of Material List
Equipment Data Sheet Submittals
Sequence of Events
Acceptance Test & Certification Form

33. CO2 Solid Modeling

34. Questions Any Questions? Matt Wyman Semiconductor Division Manager
KOETTER FIRE PROTECTION
Dallas, Texas
– cell

35. Discharge Tests Demonstration Various Discharges VIDEO
Mock of Open Wet Bench
8’W x 1.6’W x 2.5’H
Recessed baths
Exhaust Ventilation
Various Discharges
Various Configurations
VIDEO

36. Test Results CO2 Concentrations Analysis See Table
All Extinguish Fire Source
IPA/PP Pan Fire
Measure CO2 concentrations at various points
Vent type nozzles had significant splashing results
Not recommended over open baths
Front Left
Front Middle
Rear Lower Middle
Test #1
36%
48%
51%
Test #2
28%
52%
39%
Test #3
29%
44%
24%