Michael D. Wedlake Principal Engineer, Process Integration Engineering Samsung Austin Semiconductor S.LSI Foundry Business

Transcription

1 Michael D. Wedlake Principal Engineer, Process Integration Engineering Samsung Austin Semiconductor S.LSI Foundry Business 1 0

2 Presentation Overview 1. Background a. CMP Market b. CMP Equipment Market ( ) c. Post CMP Clean Market d. End Customers 2. Significant Problems with Post CMP Cleans a. Published GLOBALFOUNDRIES research b. Examples from SAMSUNG 3. Post CMP Cleans Innovation Challenges 4. Post CMP Cleans Innovation Opportunities 1

3 Presentation Overview 1. Background a. CMP Market b. CMP Equipment Market ( ) c. Post CMP Clean Market d. End Customers 2. Significant Problems with Post CMP Cleans a. Published GLOBALFOUNDRIES research b. Examples from SAMSUNG 3. Post CMP Cleans Innovation Challenges 4. Post CMP Cleans Innovation Opportunities 2

4 1a. Background CMP Market Typical Total Cost of CMP Property, Plant, Equipment 28% Slurry 28% CMP Equipment 19% Pads 10% Variable Costs 10% Direct Labor 5% 100% Global CMP market: Worth $3.32B in 2014 Projected worth $4.94B in 2020 CAGR is 6.83% CMP Consumable market: Worth $2.25B in 2014 Projected worth $3.21B in 2020 CAGR is 6.10% CMP Equipment market: Worth $1.07B in 2014 Projected worth $1.73B in 2020 CAGR is 8.32% Consumable $ Millions % Slurry % Pads % Conditioners % PVA Brushes % Slurry Filters % Total % Takeaway: Post CMP Cleaning is a small portion of the cost of the entire process, but critical for success CMP Consumable Market Shares, Source: CMP Market Size Data, Source: 3

5 1b. Background CMP Equipment Market 4

6 1b. Background CMP Equipment Market Written by William O Mara in May 1998 in Solid State Technology, describing the situation in CMP 20 years ago. CMP is taking the IC manufacturing industry by storm. Market growth rates of up to 100%/year have been common As these new companies compete with early, more entrenched entrants, a shakeout is looming ahead. Top 4 firms control 86% of the market; 16 companies fight for 14%. Most wafer fab equipment categories are dominated by a single supplier, termed the "800-pound gorilla" of that market Who will be the 800-pound gorilla? 5

7 1b. Background - Inevitable Shakeout in CMP Merged in 1998, 47.5% market share, >1,000 polishers installed, $374.3 M Suppliers of CMP Polishing & Grinding Equipment Patterned Wafer 300 mm Patterned Wafer Si Wafer Fab Equipment Tool Available? Market Share Tool Available? Market Share Applied Materials* X 14.5% X X ~76% Aplex X Part of 1% Cybeq Nano Technology X 1.4% X Doosan Machinery X Part of 1% Ebara X 22.2% X X ~24% Fujikoshi X Part of 1% X ICMP X Part of 1% IPEC Planar X 26.2% Lam Research X Part of 1% X Lapmaster SFT X 1.3% X Obsidian X Part of 1% disk Okamoto X 1.4% X Peter Wolters X Part of 1% X Pressi X Part of 1% X Sony Precision Machinery X Part of 1% Speedfam X 21.5% X Strasbaugh X 7.7% X Sumitomoa Metal X 2.0% X Toshiba X 0.7% X Tokyo Seimitsu X Part of 1% X $517.7 million $1,356 million Sources: The Inevitable Shakeout in the CMP Equipment Market 05/01/1998 William O Mara, O Mara and Associates, Palo Alto, California Chemical Mechanical Polishing in Silicon Processing, S. Li, R. Miller, 2000, Chapter 2: Equipment, Thomas Bibby and Karey Holland, SpeedFam-IPEC Chandler, Arizona *Applied Materials entered the CMP market in 1995 with the Mirra-Mesa 200mm tool. $1,031 M $325 M 6

8 1b. Background CMP Equipment 76% market share Image Source: 7

9 1b. Background CMP Equipment 24% market share To enhance productivity and Cost-of-ownership the F*REX300SII has two more platens (4 in total), a layout which enables a variety of different parallel and serial polishing modes targeting Cu and other advanced applications. The layout enables a greater throughput especially for 2 step processes. The F*REX300SII system is equipped with 1 head per platen and 3 cleaning stations and with 4 FOUP load ports. Further options include the latest Profile head technology, Pad thickness and retainer ring thickness monitoring, Endpoint monitoring and Inline metrology. Source: Up to three stages cleaning module: The cleaning module consists of a sequence of brush an d pencil cleaning and rinse stations enabling a minimum defectivity after CMP. 8

10 1b. Background CMP Equipment Typical Attributes of Post CMP Cleaner 1. Megasonic Cleaning acoustic energy liberates defects. 2. Brush Box 1 (double-sided scrub brushes, typically cylindrical nodule covered brush made of hydrophilic porous polyvinyl alcohol (PVA), water is continuously pumped through the brush core and then through the porous brush. Chemical is supplied from spray bars. 3. Brush Box 2 (same configuration as BB1) 4. Spin Rinse Dry, more typically Marangoni or IPA Dryer. Reflexion LK Configuration Image SOURCE: Brush Images: SOURCE: 2013 Applied Materials Patent: 9

11 1c. Background Post CMP Clean Market PVA Brushes Market Share TAM ITW (Rippey, Texwipe) ~80% Aion (formerly Kanebo, manufacturer for Rippey) $30-50M Entegris ~20% BrushTek PCMP Cleaning Chemistry Suppliers Market Share TAM Air Products and Chemicals, Inc. 25% Anji Microelectronics, Inc. Entegris (ATMI Inc. > ESC) 27% BASF SE Cabot Microelectronics Corp. (Epoch Materials of Taiwan) DuPont EKC Technology JT Baker Kanto Chemical Company, Inc. Mitsubishi Chemical Corporation Shanghai Sinyang Semiconductor Materials Co., Ltd. Technic France Wako Pure Chemical Industries, Ltd. $90-140M Famous Clean Chems: Applied Materials s Electraclean (APCI) APCI CoppeReady CP72B (acid) & CP74 (base) ATMI Planarclean and ESC cleans DuPont s PCMP5510 J.T. Baker pcmp-800/850 Mitsubishi Chemical MCX-SDR4 Wako s CX-100 Source: 10

12 1c. Background Post CMP Clean Market Zeta Potential Key to Particle Removal Efficiency Source: Chemical Mechanical Polishing in Silicon Processing, S. Li, R. Miller, 2000 Chapter 7: Post-CMP Clean, Francois Tardif, LETI, Grenoble, France 11

13 1d. Background End Customers 12

14 1d. Background End Customers SOURCE: 13

15 1d. Background End Customers Company Type Location Market Share TSMC Pure-Play Taiwan % Globalfoundries Pure-Play U.S % UMC Pure-Play Taiwan % Samsung IDM South Korea / U.S % SMIC Pure-Play China % Powerchip IDM Taiwan % TowerJazz Pure-Play Israel % Fujitsu IDM Japan N/A Vanguard Pure-Play Taiwan N/A Hua Hong Semi Pure-Play China % Dongbu HiTek Pure-Play South Korea % SSMC Pure-Play Singapore N/A WIN Pure-Play Taiwan N/A Non-TSMC SOURCE: Revenue (US$ Million) $35,000 $30,000 $25,000 $20,000 $15,000 $10,000 $5,000 y = x - 6E+06 R² = Foundry Revenues y = 845.3x - 2E+06 R² = $ Year TSMC Globalfoundries UMC Samsung SMIC Powerchip TowerJazz Fujitsu Vanguard Hua Hong Semi Dongbu HiTek SSMC WIN Non-TSMC Linear (TSMC) Linear (Non-TSMC) 14

16 1d. Background End Customers 93% 54% 57% Source:

17 1d. Background End Customers Company Revenue ($M) Capacity (K w/m) 2015 Revenue ($M) 2015 Annual Capacity 2015 Avg $/wfr* Samsung ,140,000 $ 1, TSMC ,884,000 $ 1, Micron ,468,000 $ Toshiba/SanDisk ,312,000 $ SK Hynix ,040,000 $ 1, GlobalFoundries ,752,000 $ Intel ,628,000 $ 5, UMC ,312,000 $ Texas Instruments ,240,000 $ 1, STMicroelectronics ,844,000 $ 1, SOURCE: * Assumes 100% capacity is being used. 1 16

18 1d. Background End Customers Source: 17

19 1d. Background End Customers 18

20 1d. Background End Customers Source: mm-wafers-less-than-half-using-200mm-wafers/ 1 19

21 1d. Background End Customers 20

22 1d. Background Ultimate End Customers 21

23 Presentation Overview 1. Background a. CMP Market b. CMP Equipment Market ( ) c. Post CMP Clean Market d. End Customers 2. Significant Problems with Post CMP Cleans a. Published GLOBALFOUNDRIES research b. Examples from SAMSUNG 3. Post CMP Cleans Innovation Challenges 4. Post CMP Cleans Innovation Opportunities 22

24 2a. Significant Problems with PCMP Cleans Last year at SPCC 2017 Source: 23

25 2a. Significant Problems with PCMP Cleans For several years GLOBALFOUNDRIES has presented many interesting papers on PCMP Cleaning. The Origin of Defects in Replacement Metal Gate (RMG) CMP Hong Jin Kim, Bohra Girish, Huey-Ming Wang, Venugopal Govindarajulu, Dinesh Koli CAMP Conference 2016, Lake Placid, NY 24

26 2a. Significant Problems with PCMP Cleans Brush breakin procedure is very important for removing initial contamination from the brush. Brush rotational speed strongly influences defectivity Slowly touching to the wafer can reduce transfer of particles from the brush to the wafer. 25

27 2a. Significant Problems with PCMP Cleans Study on Cu CMP Defects Failure Mechanism from the Viewpoint of Cleaner Module Design -- Ji Chul Yang (GLOBALFOUNDRIES), CAMP Conference in Lake Placid, NY Fundamental study on reducing Cu metal flakes. Fundamental Study of Cleaner Module Design. Megasonic, Jet Spray, Pencil Brush, Horizontal Type Brush, Vertical Brush+Dryer. Vertical Brush. Brush only, Pencil + Brush, Meg + Brush, Brush + Brush. Particle removal efficiency: Jet Spray, Pencil Brush, One Brush, Two Brush. A Frictional Analysis of Various Types of PVA Brushes and an Attempt to the Direct Observation of its Contact Condition -- Toshiyuki Sanada (Shizuoka University), CAMP Conference in Lake Placid, NY, 2016 Pen type brush is manufactured differently from the conventional PVA roller brush. A dense skin layer exists on regular PVA brushes while no skin layer is present on the pen brush. Air and water are incorporated into the brush during cleaning. ICPT 2015 My question: If the pen brushes performs so much better, how do we remove the skin from the regular PVA brushes? 26

28 2b. pcmp Clean Examples from Samsung We have had our shares of struggles and triumphs: 1. Organic residue issues (source: pad, clean chemistries) 2. Alumina slurry contamination 3. Visible metallic byproduct contamination linked to 3D Nanotopography (2016 CAMP) 4. Brush loading / cross-contamination 5. Invisible organic/metallic byproduct contamination 6. IPA Spray bar contamination 7. Clean chemistry attack 8. Brush torque correlation to yield 9. Studies on light induced corrosion I will review the examples in red. 27

29 2b. pcmp Clean Problem #1 Alumina particle contamination example Particle Defect Count Comparison of Defects / Density Time Density Map Composite Wafer Map Composite Die 28

30 2b. pcmp Clean Problem #2 Brush Loading Contamination 1 29

31 2b. pcmp Clean Problem #2 IPA-Dryer Only 5-sec DI BB1 5-sec DI BB2 5-sec DI BB2 5-sec DI BB1 5-sec DI BB1 10-sec DI BB1 20-sec DI BB1 40-sec DI BB1 5-sec DI BB2 10-sec DI BB2 20-sec DI BB2 40-sec DI BB2 5-sec HF BB1 10-sec HF BB1 20-sec HF BB1 40-sec HF BB1 5-sec NH4OH BB2 10-sec NH4OH BB2 20-sec NH4OH BB2 40-sec NH4OH BB2 POR, 5-sec HF +5-sec NH4OH Reverse POR, 5-sec NH4OH + 5-sec HF POR, 10-sec HF +10-sec NH4OH Reverse POR, 10-sec NH4OH + 10-sec HF Simulated Double BB2 (5+5 NH4OH) Simulated Double BB1 (5+5 HF) Experimentation with Brand New Brushes: Heavy organic residue IPA Dryer has some contamination Water only scrubbing is much worse than chemical cleaning NH4OH brush (BB2) is much cleaner than HF brush (BB1) HF brush has random straight line defects Cleanest results achieved with HF, then NH4OH SEM and EDX Reviewed 30

32 2b. pcmp Clean Problem #2 IPA-Dryer Only 5-sec DI BB1 5-sec DI BB2 5-sec DI BB2 5-sec DI BB1 5-sec DI BB1 10-sec DI BB1 20-sec DI BB1 40-sec DI BB1 5-sec DI BB2 10-sec DI BB2 20-sec DI BB2 40-sec DI BB2 5-sec HF BB1 10-sec HF BB1 20-sec HF BB1 40-sec HF BB1 5-sec NH4OH BB2 10-sec NH4OH BB2 20-sec NH4OH BB2 40-sec NH4OH BB2 POR, 5-sec HF +5-sec NH4OH Reverse POR, 5-sec NH4OH + 5-sec HF POR, 10-sec HF +10-sec NH4OH Reverse POR, 10-sec NH4OH + 10-sec HF Simulated Double BB2 (5+5 NH4OH) Simulated Double BB1 (5+5 HF) Experimentation with Brand End-of-Life Brushes: Heavy metallic residue IPA Dryer has some edge contamination (9:00 in particular) Water only scrubbing is much worse than chemical cleaning NH4OH brush (BB2) is much cleaner than HF brush (BB1) HF brush has random straight line defects Cleanest results achieved with HF, then NH4OH SEM and EDX Reviewed 31

33 2b. pcmp Clean Problems #2 IPA-Dryer Only 10-sec DI BB2 10-sec NH4OH BB2 5-sec DI BB1 5-sec DI BB2 5-sec DI BB2 5-sec DI BB1 5-sec DI BB1 10-sec DI BB1 20-sec DI BB1 40-sec DI BB1 5-sec DI BB2 20-sec DI BB2 40-sec DI BB2 5-sec HF BB1 10-sec HF BB1 20-sec HF BB1 40-sec HF BB1 5-sec NH4OH BB2 POR, 5-sec HF +5-sec NH4OH Reverse POR, 5-sec NH4OH + 5-sec HF Reverse POR, 10-sec NH4OH + 10-sec HF Simulated Double BB2 (5+5 NH4OH) 20-sec NH4OH BB2 40-sec NH4OH BB2 POR, 10-sec HF +10-sec NH4OH Simulated Double BB1 (5+5 HF) Experimentation with Brand End-of-Life Brushes: Heavy metallic residue IPA Dryer has some edge contamination (9:00 in particular) Water only scrubbing is much worse than chemical cleaning NH4OH brush (BB2) is much cleaner than HF brush (BB1) HF brush has random straight line defects SEM Cleanest results achieved with HF, then NH4OH and EDX Reviewed 32

34 2b. pcmp Clean Problem #2 80 Tool Audit: Cleaner Partition 70 Oxide Removal (A) Slot15-40-sec (HF) Slot14-20-sec (HF) Slot13-10-sec (HF) Slot22-10-sec (HF) Slot23-10-sec (HF) Slot25-5-sec (HF) x2 Slot12-5-sec (HF) Slot20-5-sec (HF) Slot21-5-sec (HF) Radius (mm) 33

35 2b. pcmp Clean Problem #2 1.6 Tool Audit: Cleaner Partition 1.4 Normalized Oxide Removal (A) Slot12-5-sec (HF) Slot13-10-sec (HF) Slot14-20-sec (HF) Slot15-40-sec (HF) Slot20-5-sec (HF) Slot21-5-sec (HF) Slot22-10-sec (HF) Slot23-10-sec (HF) Slot25-5-sec (HF) x Radius (mm) 34

36 2b. pcmp Clean Problem #2 HF Etch Rate, 9 data points y = x R² = Oxide Etch Amount (A) AVG Linear (AVG) sec + 5-sec HF Exposure Time (sec) 35

37 2b. pcmp Clean Problem #2 80 Tool1 vs. Tool2 vs. HF Exposure Time Oxide Removal (A) Tool1-40-sec HF Tool2-40-sec HF Tool1-20-sec HF Tool2-20-sec HF Tool1-10-sec HF Tool2-10-sec HF Tool1-5-sec HF Tool2-5-sec HF Radius (mm) 36

38 2b. pcmp Clean Problems #2 Tool1 vs. Tool2 HF Etch Rate y = x R² = Oxide Etch Amount (A) y = x R² = HF Exposure Time (sec) 37

39 2b. pcmp Clean Problem #2 2 Tool1 vs. Tool2 Normalized Etch Profiles HF Induced Losses (A) Tool1-5-sec HF Tool1-10-sec HF Tool1-20-sec HF Tool1-40-sec HF Tool2-5-sec HF Tool2-10-sec HF Tool2-20-sec HF Tool2-40-sec HF Radius (mm) 38

40 2b. pcmp Clean Problem #2 1.5 Tool1 Normalized Etch Profiles HF Induced Losses (A) Radius (mm)

41 2b. pcmp Clean Problem #2 BAD GOOD GOOD BAD

42 2b. pcmp Clean Problem #2 BAD GOOD GOOD BAD

43 2b. pcmp Clean Problem #2 Image Source: 42

44 2b. pcmp Clean Problem #3 Pre Post Adders IPA spray bar was replaced to stop the presence of these silicon dioxide spheres. 43

45 2b. pcmp Clean Problem #4 Bin Loss (Lower is Better) Torque Key: Tool1 Tool2 Tool3 Tool4 Time Low-Torque Group Has High Yield Across Whole Wafer Yield Bin Yield Delta Torque Torque Radius Spatial 1 44

46 Presentation Overview 1. Background a. CMP Market b. CMP Equipment Market ( ) c. Post CMP Clean Market d. End Customers 2. Significant Problems with Post CMP Cleans a. Published GLOBALFOUNDRIES research b. Examples from SAMSUNG 3. Post CMP Cleans Innovation Challenges 4. Post CMP Cleans Innovation Opportunities 45

47 3. Post CMP Cleans Innovation Challenges 1. Equipment: Applied Materials and Ebara a. For best results, we must do CMP+Cleans b. Proprietary hardware c. Proprietary software d. How do we introduce new hardware for cleans? 2. More direct engagement on technical challenges a. Fab engineering challenges b. Equipment manufacturer c. Replacement parts d. Brush suppliers e. Clean chemical suppliers f. Third parties with new cleans technology g. Metrology suppliers encountering more Non Visual Detectable Defects 3. Logistical challenges a. Safety approvals b. Tool time c. Chemical delivery system d. Metrology (defectivity scans, elemental analysis, etc.) 4. More Fab process engineering attention / specialization a. Post CMP Cleaning is often afterthought b. New equipment to study and calibrate cleans c. More research into how to maximize performance d. More research into hardware e. More research into consumables (chems / brushes) 5. Funding for R&D a. PVA and chemicals are a small part of CMP budget b. Funding would likely need to come from Fabs or Equipment suppliers 46

48 Presentation Overview 1. Background a. CMP Market b. CMP Equipment Market ( ) c. Post CMP Clean Market d. End Customers 2. Significant Problems with Post CMP Cleans a. Published GLOBALFOUNDRIES research b. Examples from SAMSUNG 3. Post CMP Cleans Innovation Challenges 4. Post CMP Cleans Innovation Opportunities 47

49 4. Post CMP Cleans Innovation Opportunities Factory 1 is R&D sponsor for larger company Receives funding from all other Factories and Corporate R&D Allocated space for R&D tools and facilitation Conducts research geared toward needs of entire company Large factory economy of scale reduces costs CMP hardware supplier Factory 5 Facilities Supplier (In/Out) CMP + Clean Cleaner hardware Factory 4 Company Factory 1 R&D Center R&D Chemical Supplier Brush supplier Factory 3 Factory 2 CMP + Clean R&D Center Dedicated tool(s), Facilitization Access to best metrology technology in Fab Access to world class Failure/Surface Analysis Labs Sensor and insitu analytics overkill Safe Space for Creative Personalities Run new CMP process to generate Yield, pfa 48

50 4. Post CMP Cleans Innovation Opportunities CMP + Clean R&D Center Objectives: Demonstrate ROI Develop efficiency metrics Lower defectivity Higher Yield Longer consumable lifetimes Longer tool uptime Turn waste into re-sellable product Air Products and Chemicals, Inc. Anji Microelectronics, Inc. Entegris (ATMI Inc. > ESC) BASF SE Cabot Microelectronics Corp. (Epoch Materials of Taiwan) DuPont EKC Technology JT Baker Kanto Chemical Company, Inc. Mitsubishi Chemical Corporation Shanghai Sinyang Semiconductor Materials Co., Ltd. Technic France Wako Pure Chemical Industries, Ltd. Facilities Supplier (In/Out) Applied Materials and Ebara cooperation is critical to this project s success. With 450 mm postponed indefinitely, major tool upgrades to the large installed 300mm base could be a lucrative revenue stream. Learning from 300mm improvements could pass to 200mm. Critical first question: Open or Closed Cleaner System. Fixed or Modular. The CMP + Clean R&D Center might have an one or more R&D tools with many modules evaluation modules and transportation schemes. The cleaner component of the tool might even be located in a separate tool with pass through. The objective would be to have a versatile tool for screening components and hardware that produce the lowest defectivity. Chemical Supplier CMP hardware supplier CMP + Clean R&D Center Brush supplier Cleaner hardware Applied Materials and Ebara, SCREEN Holdings Co., Ltd. (ex DNS), Modutek Corporation, Cleaning Technologies Group, Akrion Systems LLC, Speedline Technologies, Inc., ONBoard Solution Pty Ltd., Falcon Process Systems, AP&S International GmbH, MEI Wet Processing Systems and Services LLC, Tokyo Electron Limited (FSI International), SEMES and Lam Research Corporation ITW (Rippey, Texwipe), Aion (formerly Kanebo), Entegris, BrushTek 49

51 4. Post CMP Cleans Innovation Opportunities Relative Advantages in HVM Factory Supplier R&D Problem Identification X Pathfinding for Solutions / X Continuous Improvement / \ Cutting edge defect inspection X Cutting edge metrology X Economies of scale X -- UPW X -- Cleanroom X -- Maintenance X -- Silicon (NPWs, Short loops, Patterns) X Electrical Test X Analytical Labs X Wet Labs X Accumulation Problems (Marathon) X Hardware design expertise / X Control over existing platform X New cleans technology X Innovative new consumables X Domain knowledge, expertise, and insights / X Outside the Box Thinking X Permitted time to deeply study issues X Development of Analytical Test to Prove Theory / X 50

52 4. Post CMP Cleans Innovation Opportunities 1. Sematech ADK 14nm Test Chip a. Test chip for modern era b. Open design enables open communication 2. CMP Center in Albany, NY a. Joint project of SUNY Polytechnic Institute and Sematech b. Air Products and Mitsubishi Chemical 3. IMEC 4. Engagement with Equipment Suppliers a. Possibility of modifying the CMP Cleaner a. Incremental improvements b. Complete replacement c. Modular replacement b. Modularity would lead to easier modification of hardware c. Potential for supplier to purchase their own hardware d. Modification leads to improved cleaning performance 5. Meetings Like This a. Surface Preparation and Cleaning Conference (SPCC) b. CMP User s Group c. CAMP Conference d. International Conference on Planarization Technology (ICPT) 51

53 4. Post CMP Cleans Innovation Opportunities EXAMPLES OF HARDWARE IMPROVEMENTS: 1. Horizontal oriented brush cleaning 1. More even etching / cleaning 2. Better cleaning efficiency 3. More efficient transportation and use of time 2. Standalone equipment for brush break-in and brush regeneration 3. Brush cycling (front side, back side, cleaning/regeneration) 4. Additional chemical lines / choices within Brush Box 5. Pad Cleaning 52

54 4. Post CMP Cleans Innovation Opportunities Revisit horizontal wafer processing. Benefits: More even distribution of chemical. More compact layout. Potential for adding multiple cleaning stations, evaluation of new consumables. Source: 53

55 4. Post CMP Cleans Innovation Opportunities Add multiple brushes to improve cleaning regardless of die level orientation and redundancy. Add automatic torque adjustment for brushes (brush gap adjustment APC) Adding quick swap-out of brushes so brushes can be de-loaded of contaminants. Source: US B1 54

56 CONCLUSIONS 1. Effective Post CMP Cleans are critical for Yield 2. Post CMP Cleans can t be treated like an afterthought; we can t continue to coast 3. CMP s equipment shakeout has occurred; how do we innovate inside these tools? 4. Non-Visible Defects (NVD) are a major concern now and into the future 5. Time to develop solutions is shortening 6. Focused R&D Centers within HVM with tight coordination among all suppliers may be the fastest, cheapest solution if you are interested I d like to speak with you. 55

57 THANK YOU FOR YOUR ATTENION 56