Robotics for Electronics Manufacturing Presented to the IEEE Consultants Network of Silicon Valley (IEEE-CNSV) Tuesday, 8 June 2010 by Dr. Karl Mathia 1
Abstract Dr. Mathia will present topics from his new book 'Robotics for Electronics Manufacturing' (Cambridge University Press, 2010) and will focus on semiconductor manufacturing, the most challenging cleanroom manufacturing application for robotics as a benchmark example. The specific topics discussed are: industrial robotics, cleanroom robotics, the design of atmospheric and vacuum robots, test and characterization. The presentation will begin with an overview of early substratehandling robots and how the current cleanroom automation technology evolved over time. Real-world examples illustrate the design and use of cleanroom robots for both atmospheric and vacuum environments. The compliance of robotic systems with industry standards and de-facto standards is emphasized. Finally, future trends are outlined. 2
Speaker Biography Dr. Karl Mathia studied in Germany and the U.S. and holds advanced degrees in Electrical and Computer Engineering. He has over 20 years of experience in research & development in robotics, control systems, and automation. He held management positions at leading robotic firms, including Brooks Automation, PRI Automation, and Newport Corporation. Dr. Mathia recently published the book 'Robotics for Electronics Manufacturing - Principles and Applications in Cleanroom Automation' (Cambridge University Press, 2010). He has also published numerous articles in the areas of robotics, controls, and intelligent systems, and taught short courses in industry. Dr. Mathia is a senior member of the IEEE, the chair of the IEEE Control Systems Society in Santa Clara Valley, and a member of the Consultants' Network of Silicon Valley (CNSV). 3
Questions to the Audience 1. Robotics experience? 2. Cleanroom experience? 3. Semiconductor experience? 4. Other industries? 4
Robotics in Silicon Valley 5
Robotics in Silicon Valley Silicon Valley has a rich history of research in robotics Two famous examples: Shakey Stanford Arm 6
Robotics in Silicon Valley Shakey (1967-1972): AI research at SRI International (then the Stanford Research Institute) Mobile robot, perceived and modeled its environment First electronic person (Life Magazine, 1970) Now at the Computer History Museum in Mountain View, CA. 7
Robotics in Silicon Valley Stanford Arm (1969): Designed by Victor Scheinman at Stanford He started Vicarm Inc. in 1973 and sold it to Unimation. The arm became the PUMA robot. 8
Robotics in Silicon Valley Unimation was sold to Westinghouse (1980) and Stäubli (1988). The original PUMA prototype is now at the Smithsonian. 9
Robotics in Silicon Valley Cleanroom Robotics: innovation that enabled the semiconductor boom: Adept Technologies Applied Materials Asyst Technology (bankruptcy) Cybeq Systems (acquired by IDE, now defunct) Equipe Technologies (acquired by PRI/Brooks Automation) Genmark Automation Hine Design (acquired by Asyst Technology) Kensington Laboratories 10
Industrial Robotics 11
Industrial Robotics Operational stock of industrial robots (*estimate). Source: World Robotics 2008, IFR, 2008 1,200,000 1,000,000 Financial crisis Industrial robots 800,000 600,000 400,000 200,000 0 1973 1983 1990 1995 2000 2005 2007 2011* Year 12
Industrial Robotics Reasons for success: three characteristics Programmable: flexible and versatile (not special-purpose) Superior physical abilities over humans: long time periods, uncomfortable or hazardous environments Accurate and consistent performance 13
Industrial Robotics Price index for industrial robots 1990-2005 Source: World Robotics 2005 (IFR, 2006) 100 90 80 Price index [%] 70 60 without quality adjustment 50 40 30 with quality adjustment 20 10 with labor cost increases 0 1990 1995 2000 2005 Year 46% 78% 10% 14
Industrial Robotics Largest robot companies 2005: Fanuc (Japan) Motoman (Japan, US based) ABB (Sweden) Kuka (Germany) 15
Cleanroom Robotics 16
Cleanroom Robotics Industrial cleanroom robotics: study, design, and use of robot systems in industrial cleanroom environments 17
Cleanroom Robotics Shipments of industrial robots by application Source: World Robotics 2008 (IFR, 2008) 25,000 20,000 15,000 10,000 5,000 0 Welding Handling Cleanroom Assembly Dispensing Processing Industrial robots Americas Asia Primarily electronics industry (IFR) Application 18
Cleanroom Robotics How clean are we? Semiconductor Automation Cleanroom Robot Applications Summary 19
How clean are we? Not clean enough for modern semiconductor manufacturing Human contamination at different levels of motion. *Particle size: 0.3 µm and larger Human Motion Heat emission (kw) Moisture emission (gram/hour) Particle emission* (particles/min) At Rest 0.12 90 100,000 0.50 Light Work 0.18 180 1,000,000 1.00 4.8 km/h 0.3 320 5,000,000 2.15 6.4 km/h 0.4 430 10,000,000 2.55 Breathing requirements (m 3 /hour) 20
How clean are we? Humans need cleanroom suits But in ultra-clean environments cleanroom robots must replace humans 21
Cleanroom Robotics How clean are we? Semiconductor Automation Cleanroom Robot Applications Summary 22
Semiconductor Automation Semiconductor manufacturing process per conductive layer: - Many process steps - Many metrology steps (not shown) - Multiple loops per conductive layer - Robots are used at every step 23
Semiconductor Automation Semiconductor manufacturing is characterized with the wafer size used. 450 400 350 Wafer Size [mm] Area Increase [%] 250 200 Wafer Size [mm] 300 250 200 150 100 50 150 100 50 Area Increase [%] 0 0 1970 1975 1980 1984 1988 1995 1998 2012 Year 24
Semiconductor Automation Early Factories ( Fabs ), 1950s and 1960s Laboratory-type operations Manual wafer and substrate handling, e.g. vacuum wands and unsealed cassettes Cleanliness not a big issue (large geometries) Foto: wafer cart Foto: Fairchild Semiconductor 1962 (transistors production) Foto: vacuum wand 25
Semiconductor Automation Early Automation: 100 150mm (late 1970s) Lack of technology limited automation Lack of standards limited technology integration Early tools used customized automation 26
Semiconductor Automation Customized Automation: Example Autoetch 490 (200mm), used with permission of Lam Research 27
Semiconductor Automation 50 years: From innovation to mass production in moderns fabs 1962 2010... 28
Semicondutor Automation Semiconductor automation became mainstream in 200mm fabs, and is critical for 300mm fabs. Three levels of automation: 1) Interbay automation 2) Intrabay automation 3) Tool automation (cleanroom robotics) 29
Semiconductor Automation Cleanrooms are isolated environments with a controlled humidity, temperature, and particulate contamination. 30
Semiconductor Automation Fabrication facility ( fab ): Tool Automation (Robotics) Intrabay (Reticle Handling) Interbay (AMHS) Fab Mgmt (Software) Services (Operators) 31
Semiconductor Automation Video: Intel s 300mm Fab 32 in Chandler, AZ 32
Semiconductor Automation International Technology Roadmap for Semiconductors 2010: CD=45 nm, critical particle size=23 nm Class 1: contamination limit=100/m 3 CD, Critical Particle Size [nm] 100 90 80 70 60 50 40 30 20 10 0 Technology Node (CD) Critical Particle Size 2009 Detection Limit 2009 detection MIT research 06: Transistor with CD = 60nm 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year 33
Semiconductor Automation International Organization for Standardization: ISO Standard Ultra-high vacuum mfg. process Manufacturing area with operators Examples: -Thickness of human hair: 50-120 µm, average is 120 µm -Normal room (office, residential) has Class 9 cleanliness -Ideal gas: 2.5 10 19 molecules/cm 2, Class 1: ~10 6 molecules/cm 2 (Earth s moon) 34
Cleanroom Robotics How clean are we? Semiconductor Automation Cleanroom Robots Cleanroom Robot Applications Summary 35
Cleanroom Robots Atmospheric cleanroom robots operate in cleanroom environments at the ambient atmospheric pressure. Substrate handlers 36
Cleanroom Robots Vacuum robots operate under vacuum pressures below the ambient atmospheric pressure. Either the entire robot or a portion of the robot resides in vacuum. 37
Cleanroom Robots Handling in atmosphere or vacuum 3 to 5 axes of motion SCARA-type arm is most common Substrates are 150 300 mm wafers, reticles, 38
Cleanroom Robots Radial motion only: polar coordinates, 3 axes of motion 39
Cleanroom Robots Articulated robot: 4 axes of motion 40
Cleanroom Robotics How clean are we? Semiconductor Automation Cleanroom Robots Cleanroom Robot Applications Summary 41
Cleanroom Robot Applications Semiconductor manufacturing is the most demanding cleanroom application Cleanroom robots are critical for automating these process tools Standardization is essential for the global semiconductor industry There are standards for the devices, manufacturing processes, tools, and automation systems 42
Cleanroom Robot Applications Atmospheric robots for automation in ambient atmosphere: Equipment Front End Module (EFEM): interface between the factory and tools Chemical mechanical polishing (CMP) tool: removes material from a substrate Inspection and metrology tools: quality control between process steps Rapid thermal processing (RTP) tool: modifies deposited films 43
Cleanroom Robot Applications Atmospheric Robots EFEM (2 doors) Vacuum Cluster Tool Inspection Tool EFEM (4 doors) 44
Cleanroom Robots Video: wafer-handling robots 45
Cleanroom Robot Applications Vacuum robots for automation in vacuum: Deposition tools: deposit insulating or conductive materials on the substrate (ALD, CVD, PVD, epitaxial) Etch tool: removes material from a substrate Ion implementation tools: implement dopants on the substrate 46
Cleanroom Robot Applications Vacuum Robots Blueshift PVD tool (Anvil Automation) CVD tool (INDEOtec) 47
Design Guidelines Overview Video: vacuum robot 48
Cleanroom Robot Applications Vacuum and Atmospheric Robots Stocker AMHS Process/metrology tool Mono rail EFEM VCT 49
Cleanroom Robot Applications Flat Panel Displays (FPDs): FPD manufacturing is similar to semi FPD substrates are significantly larger Robots are scaled up accordingly 6 5 Surface area of FPD substrates by generation Area [m 2 ] 4 3 2 1 0 1 2 3 4 5 6 7 8 Generation 50
Cleanroom Robot Applications Solar Cells (S.C.): S.C. manufacturing is similar to semi S.C. substrates are smaller and thinner Robots use special end-effectors 51
Cleanroom Robots ###Video### 52
Cleanroom Robotics How clean are we? Semiconductor Automation Cleanroom Robot Applications Summary 53
Design of Atmospheric Robots 54
Design of Atmospheric Robots Design Guidelines Overview Clean Materials Preventing Electrostatic Discharges (ESD) Surface Finishes Clean Drive Trains Arm Compliance Design Example 55
Design Guidelines Overview Two types of contamination must be addressed during the design process: Airborne molecular contamination (AMC) from particles and outgassing generated by the robot Surface contamination of the substrate from contact with the robot 56
Design Guidelines Overview Minimize particle generation: best practices Suitable, clean materials Minimal number of moving parts All moving parts below the substrate Use internal robot cabling only Coated or treated external surfaces Cleanroom approved lubricants Enclose bearings and the robot interior Evacuate generated particles 57
Clean Materials Clean materials: Minimize particle generation from contact, friction, out-gassing Stainless steel: excellent, expensive Aluminum: cheap, easy to machine Plastics: small parts, harsh environm. Ceramics: excellent, very expensive Composites: instead of metal, ceramics 58
Preventing ESD Electrostatic discharge (ESD): Rapid charge transfer between objects of different electrical potentials Two types of ESD: From an ESD-sensitive device To and ESD-sensitive device Of interest here (up to 30 kv!): From substrate to robot end-effector, or From robot end-effector to substrate 59
Preventing ESD ESD Test: ESD guns simulate events using the models. Most relevant: HBM and CDM Normalized current. Voltage 30 kv. ESD gun 60
Preventing ESD Prevention: Grounding Ground path from end-effector to ground (avoid electronics!) Conductive/dissipative parts within 0.3m of ESD-sensitive device Conductive and dissipative surfaces 61
Clean Drive Trains Typical design: SCARA-type robot Drive train: motor, belts, pulleys, shaft, lead screw (Z), bearings 62
Clean Drive Trains Design guidelines: Maintainability: easy access for repair to minimize tool downtime Minimize number of moving parts: minimize particle generation Moving parts are below substrate: generated particles don t fall on the substrate Motor selection: brushless motors generate fewer particles, no maintenance need 63
Atmospheric Robots Evacuating airborne particles Z-motion Z-motion Belts Motors Air flow Fan Evacuated particles 64
Clean Drive Trains Design guidelines: Contact seals for rotary bearings: sealing lip against rotating shaft Non-contact seals: flinger, labyrinth Contact type Non-contact: flinger labyrinth 65
Design Guidelines Overview Vacuum robot inside vacuum cluster tool Vacuum chamber Load lock Vacuum Ambient atmosphere Chuck Static vacuum barrier (gasket, O-ring, etc.) Area where dynamic vacuum barrier is located 66
Design Guidelines Overview Vacuum cluster tool: vacuum robot inside 67
Design Guidelines Overview Robot design: engineering challenges Static vacuum barrier, separating vacuum and atmosphere Dynamic vacuum barrier that transfers motion from atmosphere to vacuum Clean drive train Prevention of external and virtual leaks Surface finishes that don t contaminate the vacuum chamber 68
Summary Long-term result of cleanliness strategy: sources of wafer contamination (1997) Cleanroom Process Equipment Humans Excellent 100% 90% 80% Relative Contamination 70% 60% 50% 40% 30% 20% 10% 0% 1985 1990 1995 2000 Year 69
Possible Trends in Cleanroom Robotics 70
Possible Trends Future possibilities: 450 mm wafers Will 450 mm happen? (So far only Intel, TSMC, and Samsung support it.) Who will pay for the wafer size transition? (300 mm is not paid for yet [SEMI]) Consequence for robotics: technical challenge would be to scale up the robots and increase their reliability. (Not a significant challenge) 71
Possible Trends Future possibilities: 3-dimensional devices Are 3D devices a possibility, with an increased number of metal layers and denser 300 mm circuitry? Risk: more process steps, increased processing time, therefore a higher risk of reduced yield. Cost per wafer would increase (cost per process step is constant) Consequence for robotics: same form factors, but higher speed, tighter cleanliness requirements 72
Book Karl Mathia (2010). Robotics for Electronics Manufacturing Principles and Applications in Cleanroom Automation. Cambridge University Press. 73
Q & A 74