Cu Wire Bonding Survey Results inemi Cu Wire Bonding Reliability Project Team Jan 30, 2011
Outline About inemi Project Overview Survey Mechanism Survey Respondents Survey Results Technology Adoption Status by Device Type Implementation Plan by Package Major Concerns Reliability Test Challenges Common Failure Modes JEDEC Standard Failure Analysis Technique Preferred Wire Type Mold Compound Smallest Bond Pad Size & Wire Diameter Pad Structure Re-design Chip Pad Finish Interposer Surface Finish Surface Treatment inemi Project Expectation 1
About inemi Mission: Forecast and accelerate improvements in the Electronics Manufacturing Industry for a sustainable future. 5 Deliverables: Technology Roadmaps Deployment Projects Research Priorities Proactive Forums Position Papers 4 Major Focuses: Miniaturization Environment Alternative Energy Medical Electronics inemi is an industry-led consortium of more than 85 global manufacturers, suppliers, industry associations and consortia, government agencies and universities. Visit us at www.inemi.org. 2
Thrust Area: Miniaturization Goal: TIG: Packaging Understand key issues and concerns regarding reliability of fine-pitch Cu wire bonding for semiconductors. Strategy Issues Graphics 2-phased project. The first phase will focus on collecting information from the industry regarding the key processing and reliability issues pertaining to Cu wire bonding. The second phase of the project will perform necessary experimental work in the areas defined by Phase 1. Project Lead: Peng Su, Cisco Systems, inc. Project Co-Lead: TBD Copper bond wires are increasingly being used for a wide variety of components. Reliability needs to be collectively assessed by the industry in a quantitative manner. Standard reliability test methods and durations for Au wire device may not be sufficient for Cu. Mar-11 Tactics Milestones and/or Deliverables Plan Actual Phase 1 1. Industry Survey 2. Existing Data Review 3. DOE Plan for Phase 2 Phase 2 1. Finalize DOE 2. Procure Materials 3. Package Assembly 4. Reliability Tests 5. Failure Analysis 6. Summary Copper Wire Bonding Reliability Project Initiative Launch Date Project Inauguration - Two PS Signers Conduct survey, analyze inputs Plan DoE for phase 2 Mar-10 Sep-10 Nov-10 Dec-10 Mar-10 Sep-10 Jan-11 Mar-11 More information at: http://www.inemi.org/cms/projects/orgpkg/copper_wire.html 3
Cu Wire Bonding Reliability Project Members 4
Survey Mechanism The industry survey is Phase 1 of the inemi Cu Wire Bonding Reliability Project. Survey Objectives Understand the key processing and reliability issues pertaining to fine pitch Cu wire bonding Collect information to help outline an experimental study of reliability of components using copper wire bonding Survey Process May Aug 2010: Develop survey by project formation team Aug 23 Sept 30, 2010: Collect inputs from industry Oct 1 Nov 26, 2010: Review & analyze survey data Dec 3 Dec 24, 2010: Verify significant issues with follow-up questions Jan 2011 : Report survey results 5
Survey Respondents 58 individuals from over 40 companies completed the survey. Suppliers (32) Bonding wire manufacturers Interposer suppliers Equipment manufacturers Package assemblers IC IDM Total revenue ~40B USD 4 8 Respondent Statistics 6 15 OEM ODM/EMS Semiconductor IC assembly house Equipment supplier 5 Material supplier Other 9 11 OEM/EMS (20) System providers End product manufacturers Total revenue ~400B USD Others (6) came from consulting, academia, industry organizations Survey data is analyzed separately for suppliers and OEM/EMS. OEM/EMS Product Segment 3 4 3 2 2 3 4 4 Consumer/portable Computers and Peripherals High End Servers Industrial Netcom Automotive Medical Other * Military, Aerospace 6
Survey Results 7
Technology Adoption Status by Device Type Please select the current application status of Cu wire bonding of your company for different device types. Survey Results: Suppliers Only semiconductor suppliers and assembly houses are counted. Material or equipment suppliers are not included. Most suppliers are in the process of technical qualification or production. Cu wires are used for all device types but memory devices appear to have less usage. OEM/EMS Many OEM/EMS have no plan for Cu wire bonding product introduction at this time. Some OEM/EMS are conducting feasibility studies. Small percentage in production for memory, logic and power devices. Observation: Suppliers are already in the process of implementing Cu bond wires. OEM/EMS are just beginning to study feasibility and a few are in the early stage of production introduction. No plan Feasibility study Technical qualification Low volume production High volume production Suppliers (20 of 32*) 0 5 10 15 20 25 30 OEM/EMS (16 of 20*) No plan Feasibility study Technical qualification Low volume production High volume production 0 10 20 30 40 *1 st number = number of responses to question 2 nd number = total responses to the survey. Memory Logic Analog Low MPU/MCU Power Devices Memory Logic Analog Low MPU/MCU Power Devices 8
Technology Adoption Status by Device Type Cont. Please select the current application status of Cu wire bonding of your company for different device types. Survey Results: OEM/EMS in different product segments Memory, Logic and Power devices are in volume production for some Consumer/Portable, Computer and High End Server products. Not in production for Netcom, Medical, Automotive products. No plan for Military and Aerospace. Consumer/Portable, Computer High End Server Netcom No plan Feasibility study Technical qualification Low volume production High volume production 0 1 2 3 4 5 No plan Feasibility study Technical qualification Low volume production High volume production 0 2 4 6 8 No plan Feasibility study Technical qualification Low volume production High volume production 0 1 2 3 4 5 6 No plan Feasibility study Technical qualification Low volume production High volume production Medical 0 2 4 6 8 10 12 No plan Feasibility study Technical qualification Low volume production High volume production Automotive 0 2 4 6 8 No plan Feasibility study Technical qualification Low volume production High volume production Military, Aerospace 0 1 2 9
Implementation Plan by Package What package types do you plan to adopt (or have already adopted) for Cu-wire technology? Survey Result: All package types are in planned or already adopted phase for Cu wires in the industry, but may vary for each company. xbga, QFN and xqfp packages have highest adoption rates. Observation Slow adoption by OEMs in contrast to suppliers, most likely because of reliability concerns (refer to results later in this report). Less implementation for memory devices, per suppliers input. 40 35 30 25 20 15 10 5 0 40 35 30 25 20 15 Power Devices Suppliers (24 of 32) Low MPU/MCU Analog Logic Memory OEM/EMS (8 of 20) xsop xqfp QFN xbga Power package (HSOP, etc.) xsop xqfp QFN 10 xbga 5 0 Power Devices Low MPU/MCU Analog Logic Memory Power package (HSOP, etc.) 10
What are the major concerns with Cu wire bonding? In-service reliability Process yield Throughput JEDEC reliability spec Unproven historical performance Manufacturing statistical process control Equipment and assembly process parameters Survey Results: Reliability and historical performance are major concerns for the OEM community. Suppliers have concerns in these areas as well but to a lesser degree. Suppliers have additional concerns about throughput and yield. Major Concerns 35 30 25 20 15 10 5 0 18 16 14 12 In-service reliability Suppliers (31 of 32) Process yield Throughput JEDEC reliability spec. Unproven historical performance OEM/EMS (18 of 20) N/A High Medium Low Manufacturing Statistical Process Control Equipment and assembly process parameters N/A High Medium Low When In-service Reliability is selected, the major reason is lack of use history (80%), followed by unknown or poor correlation of JEDEC testing with in-field performance (20%). 10 8 6 4 2 0 In-service reliability Process yield Throughput JEDEC reliability spec. Unproven historical performance Manufacturing Statistical Process Control Equipment and assembly process parameters 11
Reliability Test Challenges For your current production process, which reliability tests are the most difficult to pass? 60% 50% Suppliers (24 of 32) Survey Result: Suppliers see biased HAST as the most difficult reliability test to pass. OEM/EMS perceive biased HAST and thermal cycling as equally difficult. 40% 30% 20% 10% 0% Observation: HAST and thermal cycling are commonly used for package qualification. OEMs often have difficulty passing these items in qualification and would be aware of such failures. 80% HTSL Un-biased HAST Biased HAST PCT Thermal cycling OEM/EMS (10 of 20) THB MSL3 70% 60% 65% respondents made their selection based on actual test results, while the other 35% were based on what they expect would be difficult to pass. 50% 40% 30% 20% 10% 0% HTSL Un-biased HAST Biased HAST PCT Thermal cycling THB MSL3 12
Common Failure Modes What are the most common failure modes for copper wire bonding? 45% All Respondents (38 of 58) Survey Result: Inputs from OEM/EMS and Suppliers are similar. Ball lift, IC damage and 2 nd bond defect are the major failure modes. OEM/EMS also cite wire break as a common failure mode. Observation: The responses for the ball lift failure mode most likely refers to ball lift defect on the bonding process, not interface cracking after acceleration tests. 40% 35% 30% 25% 20% 15% 10% 5% 0% Ball lift IC damage Short Wire break 2nd bond defect 13
JEDEC Standard Are current JEDEC standard acceleration tests sufficient for system application requirements? Survey Result: Both supplier and OEM/EMS responses indicate great uncertainty of the suitability of current tests. There are concerns on test duration, temperature range, and if current JEDEC standard can properly detect corrosion. A higher percentage of Suppliers believes different or new tests are needed for Cu wire device qualification. Observation: The word sufficient in the question may have been interpreted differently by different companies. Some may have interpreted it as sufficient to generate failures, while others may have interpreted it as sufficient to ensure product reliability. Suppliers (23 of 32) OEM/EMS (14 of 20) Yes, the requirements for copper wire bonding are the same as gold wire bonding. No. Please explain why and/or suggest what test should be added/modified. Not sure. Please explain why not sure. Yes, the requirements for copper wire bonding are the same as gold wire bonding. No. Please explain why and/or suggest what test should be added/modified. Not sure. Please explain why not sure. 14
Any Issue of Failure Analysis Technique? For Cu wire bonding, have you seen any issues for the following Failure Analysis (FA) techniques? 25 20 Suppliers (21 of 32) N/A No Yes Survey Result: Suppliers indicated that Decap, IMC thickness measurement and X-ray inspection of debonding / wire shifting are issues. X-ray inspection for debonding / wire shifting is an issue for OEM/EMS. C-SAM analysis for delamination is not an issue for either suppliers and OEM/EMS. 15 10 5 0 Decap IMC layer thickness measurement Delamination analysis (i.e. C- SAM) Pad cratering condition inspection X-ray inspecion for debonding or wire shifting Observation: This data indicates that suppliers and OEM/EMS have seen challenges for FA. 12 10 OEM/EMS (12 of 20) N/A No Yes 8 6 4 2 0 Decap IMC layer thickness measurement Delamination analysis (i.e. C- SAM) Pad cratering condition inspection X-ray inspecion for debonding or wire shifting 15
Preferred Wire Type What's the preferred type of wire material? Survey Result: More than 50% of Suppliers prefer pure Cu wire, followed by Pd coated Cu wire. Most OEM/EMS prefer Pd coated Cu wire Observation: Choice of Pd coated or pure copper is determined by reliability performance and cost considerations. Some commented that Pd-coated wires have better reliability test results than bare Cu wires, while others showed opposite results. Suppliers (24 of 32) OEM/EMS (8 of 20) Pure copper Pd coated copper Other wire composition Pure copper Pd coated copper Other wire composition 16
Mold Compound Have you adopted new specifications on mold compounds for Cu wire bonding? Suppliers (22 of 32) Survey Result: Some component suppliers are adopting materials (e.g. molding compound) with lower halogen content and more neutral ph. Others do not have special requirement on materials such as mold compound. Some (40%) stated no performance advantage was observed for materials with nominally better properties. Observation: Some OEMS do not have sufficient information on package BOMs and currently can not clearly specify material property requirements. Other Reduced specification on moisture absorption/penetration Changed specification on Tg Changed specification on ph Lower specification on halogen content We do not use special mold compounds for Cu 0 2 4 6 8 10 12 17
Smallest Bond Pad Pitch & Wire Diameter Wire Diameter (mil) Wire Diameter (mil) In your CURRENT/FUTURE PRODUCTION, what is the smallest bond pad pitch and smallest wire diameter for copper wire bonding? Survey Result: General trend is moving to smaller wire diameter and pitch. 0.6 mil wire diameter is being considered for pitch ranging from 20 to 45 μm. Current (20 of 58) Pad pitch (um) 20 25 30 35 40 45 50 55 60 65 70 80 120 170 0.6 0.7 3 1 2 1 0.8 1 1 2 1 1 0.9 1 1 1 1 1 1.2 1.3 1.4 1 2 1 1 Observation: Fine Cu wires / fine bond pitch are already being used for current products and future trend is going toward even finer wires. Some data might have been mis-entered. E.g., Current: 65um pitch for 0.8mil wire, 40um pitch for 1.38mil wire 1.0mil wire for 30um pitch would be challenging. Data may be scattered due to die pad design differences or specific use for big pad size. Future (20 of 58) Pad pitch (um) 20 25 30 35 40 45 50 55 60 65 70 80 120 170 0.6 2 1 1 2 2 0.7 2 0.8 2 3 2 0.9 1 1 1 1.2 1 1.3 1.4 2 * Highlighted numbers in the table counts how many people input in this wire diameter & pad pitch size. 18
Pad Structure Re-design Needed? For the same device in your application, have the bond pad and under-pad structures been re-designed for copper wire bonding? Survey Result: While almost half of respondents replied no redesign was required for pad structure, some indicated the need for re-design. Some re-designed the structure to prevent bond pad damage. Suppliers (23) No Not sure Yes. Please explain the reason of re-design. Observation: Accuracy of response might have been limited by manufacturing experiences. Chip manufacturers might not disclose this information. OEM/EMS (8) No Not sure Yes. Please explain the reason of re-design. 19
Chip Pad Finish What is the preferred surface finish on the chip bond pad? NiPdAu Suppliers (20 of 32) Survey Result: No specific preference is indicated from suppliers input. Observation: OEM/EMS may not have the information to make comments. NiPd > 2um Al 1.2-2.0 um Al 0.8-1.2 um Al 0.0 1.0 2.0 3.0 4.0 5.0 OEM/EMS (4 of 20) NiPdAu NiPd > 2um Al 1.2-2.0 um Al 0.8-1.2 um Al 0.0 1.0 2.0 3.0 4.0 5.0 20
Interposer Surface Finish For different interposer types, what is the preferred surface finish? 25 Suppliers (20 of 32) Survey Result: Wide range of surface finishes are suggested. There is no clear consensus from suppliers or OEM/EMS. Flex / Rigid Substrate: Au finish is preferred Ceramic Substrate: No clear preferred finish Leadframe: Ag was preferred by suppliers, but OEMs have no clear preference. Observation: Variability of OEM/EMS inputs may have been limited by small sample sizes. Suppliers input shows similar surface finishes as presently used for Au wire bonding. 20 15 10 5 0 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Flex materials Rigid substrate Ceramic Leadframe OEM/EMS (5 of 20) Electrolytic NiPd Au Flash Au Ag ENEPIG ENEPEG Other Electrolytic NiPd Au Flash Au Ag ENEPIG ENEPEG Other Flex materials Rigid substrate Ceramic Leadframe 21
Surface Treatment Have you adopted new cleaning process for Cu wire bonding (e.g. plasma clean prior to bonding process)? Suppliers (20 of 32) Survey Result: Majority of respondents (75%) use the same cleaning process for Cu wire bonding as the current Au wire bonding process. The other 25% use slightly modified cleaning process. No Yes (please specify the process and why) OEM/EMS (5 of 20) No Yes (please specify the process and why) 22
Expectation for inemi Project Top 3 issues respondents expect inemi project to address: Reliability assessment of representative Cu wire bonding currently in production; Requirements for reliability testing (either new tests or modification of current standard JEDEC tests); Recommendations about wire materials, pad materials, and design rules for under-pad Si structures Several other issues were also identified as important. Suppliers (32 of 32) Other (please specify) Requirements on assembly quality tests; Requirements on reliability testing (either new tests or modification for current standard JEDEC tests); Standardized requirements on bonding process controls; Recommendations on wire materials, pad materials, and design rules for under-pad Si structures; Standardized requirements on packaging materials such as molding compounds; Reliability assessment of representative Cu wire bonding currently in production; Requirements on assembly quality tests; OEM/EMS (18 of 20) Other (please specify) 0% 10% 20% 30% 40% 50% 60% 70% Requirements on reliability testing (either new tests or modification for current standard JEDEC tests); Standardized requirements on bonding process controls; Recommendations on wire materials, pad materials, and design rules for under-pad Si structures; Standardized requirements on packaging materials such as molding compounds; Reliability assessment of representative Cu wire bonding currently in production; 0% 10% 20% 30% 40% 50% 60% 70% 80% 23
www.inemi.org inemi Contact: Haley Fu haley.fu@inemi.org 24