Design for Reliability
advertisement
Design for Reliability © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 1 Reality of Design for Reliability (DfR) Ensuring reliability of electronic designs is becoming increasingly difficult Increasing complexity of electronic circuits Increasing power requirements Introduction of new component and material technologies Introduction of less robust components Results in multiple potential drivers for failure © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 2 Reality (cont.) Predicting reliability is becoming problematic © 2005 2004 - 2010 2007 Standard MTBF calculations can tend to be inaccurate A physics-of-failure (PoF) approach can be timeintensive and not always definitive (limited insight into performance during operating life) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 3 Why This DfR? The process of Design for Reliability (DfR) is achieving a high profile in the electronics industry DfR is even on Wikipedia! Part of an overall Design for Excellence (DfX) Numerous organizations now offer DfR training tools (courses, books, etc.) © 2005 2004 - 2010 2007 Response to market demand 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 4 Why This DfR? (cont.) However, many of these tools are Too broad in focus (not electronics focused) Too much emphasis on techniques (e.g., FMEA and FTA) and not answers Incorporation of HALT and failure analysis (HALT is test, not DfR; failure analysis is too late) © 2005 2004 - 2010 2007 FMEA/FTA rarely identify DfR issues because of limited focus on the failure mechanism Frustration with ‘test-in reliability’, even HALT, has been part of the recent focus on DfR 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 5 DfR Outline What this section will not be about Not supplier issues Not production issues Not testing issues Not MTBF / MTTF Strictly design and material selection © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 6 DfR Outline DfR at Concept / Block-Diagram Stage Part selection Specifications Derating and uprating Wearout mechanisms and physics of failure © 2005 2004 - 2010 2007 Predicting degradation in today’s electronics 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 7 Design for Reliability At Concept: Specifications © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 8 Concept / Block Diagram Can DfR mistakes occur at this stage? No………..and Yes Failure to capture and understand product specifications at this stage lays the groundwork for mistakes at schematic and layout © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 9 Specifications Important specifications to capture at concept stage © 2005 2004 - 2010 2007 Reliability expectations Use environment Dimensional constraints 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 10 Reliability Goals Reliability is the measure of a product’s ability to …perform the specified function …at the customer (with their use environment) …over the desired lifetime Typical reliability metrics: Desired Lifetime / Product Performance Desired lifetime Defined as when the customer will be satisfied Should be actively used in development of part and product qualification Product performance © 2005 2004 - 2010 2007 Returns during the warranty period Survivability over lifetime at a set confidence level Try to avoid MTBF or MTTF 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 11 Failure Rate Why is Desired Lifetime Important? Electronics: Today and the Future Electronics: 1960s, 1970s, 1980s Wearout! No wearout! Time © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 12 Desired Lifetime (IC Wearout) Process Variability confidence bounds 1000 100 Airplanes Mean Service 10 life, yrs. Computers laptop/palm cell phones 1.0 0.1 Technology 0.5 µm 0.25 µm 1995 130 nm 65 nm 2005 Year produced 35 nm 2015 Known trends for TDDB, EM and HCI degradation (ref: extrapolated from ITRS roadmap) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 13 Desired Lifetime (Capacitor Wearout) Ceramic chip capacitors with high capacitance / volume (C/V) ratios Can fail in less than one year when operated at rated voltage and temperature © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 14 Solder Wearout (cont.) Design change: More silicon, less plastic Increases mismatch in coefficient of thermal expansion (CTE) BOARD LEVEL ASSEMBLY AND RELIABILITY CONSIDERATIONS FOR QFN TYPE PACKAGES, Ahmer Syed and WonJoon Kang, Amkor Technology. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 15 Examples: Desired Lifetime Low-End Consumer Products (Toys, etc.) Do they ever work? Cell Phones: Laptop Computers: Desktop Computers: Medical (External): Medical (Internal): High-End Servers: Industrial Controls: Appliances: Automotive: Avionics (Civil): Avionics (Military): Telecommunications: © 2005 2004 - 2010 2007 18 to 36 months 24 to 36 months 24 to 60 months 5 to 10 years 7 years 7 to 10 years 7 to 15 years 7 to 15 years 10 to 15 years (warranty) 10 to 20 years 10 to 30 years 10 to 30 years 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 16 Product Performance: Warranty Returns Consumer Electronics Low Volume, Non Hi-Rel 1 to 2% Industrial Controls Table on right (1st 500 to 2000 ppm Year) Depends on complexity, production volumes, and risk sensitivity Automotive 1 to 5% (Electrical, Year) Can also be reported as problems per 100 vehicles Percent of revenue (1.2 to 4.5) © 2005 2004 - 2010 2007 1st Repair rate (%) [First 3 Yrs] Product Desktop PC 37 Laptop PC 33 Refrigerator: side-by-side (with icemaker and dispenser) 28 Washing machine 22 Refrigerator: top- and bottom-freezer (with icemaker) 17 Projection TV 16 Vacuum cleaner (excluding belt replacement) 13 Dishwasher 13 Clothes dryer 13 Microwave oven (over-the-range) 12 Electric range 11 Camcorder 8 Digital camera 8 Refrigerator: top- and bottom-freezer (without icemaker) 8 TV: 30- to 36-inch 7 TV: 25- to 27-inch 5 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Consumer Reports 2006 17 Product Performance: Survivability Some companies set reliability goals based on survivability Advantages Often bounded by confidence levels Example: 95% reliability with 90% confidence over 15 years Helps set bounds on test time and sample size Does not assume a failure rate behavior (decreasing, increasing, steady-state) Disadvantages © 2005 2004 - 2010 2007 Can be re-interpreted through mean time to failure (MTTF) or mean time between failures (MTBF) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 18 Limitations of MTTF/MTBF MTBF/MTTF calculations tend to assume that failures are random in nature Easy to manipulate numbers Tweaks are made to reach desired MTBF E.g., quality factors for each component are modified Often misinterpreted Provides no motivation for failure avoidance 50K hour MTBF does not mean no failures in 50K hours Better fit towards logistics and procurement, not failure avoidance © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 19 Identify Field Environment Approach 1: Use of industry/military specifications MIL-STD-810, MIL-HDBK-310, SAE J1211, IPC-SM-785, Telcordia GR3108, IEC 60721-3, etc. Advantages No additional cost! Sometimes very comprehensive Agreement throughout the industry Missing information? Consider standards from other industries Disadvantages Most more than 20 years old Always less or greater than actual (by how much, unknown) MIL HDBK310 IPC SM785 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 20 Field Environment (cont.) Approach 2: Based on actual measurements of similar products in similar environments Determine average and realistic worst-case Identify all failure-inducing loads Include all environments © 2005 2004 - 2010 2007 Manufacturing Transportation Storage Field 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 21 Failure Inducing Loads • Temperature Cycling – Tmax, Tmin, dwell, ramp times • Sustained Temperature – T and exposure time • Humidity – Controlled, condensation • Corrosion – Salt, corrosive gases (Cl2, etc.) • Power cycling – Duty cycles, power dissipation • Electrical Loads – Voltage, current, current density – Static and transient • Electrical Noise • Mechanical Bending (Static and Cyclic) – Board-level strain • Random Vibration – PSD, exposure time, kurtosis • Harmonic Vibration – G and frequency • Mechanical shock – G, wave form, # of events © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 22 Field Environment (Best Practice) Use standards when… Measure when… Certain aspects of your environment are common No access to use environment Certain aspects of your environment are unique Strong relationship with customer Do not mistake test specifications for the actual use environment © 2005 2004 - 2010 2007 Common mistake with vibration loads 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 23 Electrical Environments Often very well defined in developed countries Introduction into developing countries can sometimes cause surprises Rules of thumb © 2005 2004 - 2010 2007 China: Can have issues with grounding (connected to rebar?) India: Numerous brownouts (several a day) Mexico: Voltage surges 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 24 Temperature: Worst-Case (Ambient) Temperature Avg. U.S. Avg. U.S. Weighted by Registration CLIM Data (Source: Confidential) Phoenix (hrs/yr) U.S. Worst Case (hrs/yr) 95F (35C) 0.375% 0.650% 11% (948) 13% (1,140) 105F (40.46C) 0.087% 0.050% 2.3% (198) 3.8% (331) 115F (46.11C) 0.008% 0.001% 0.02% (1.4) 0.1% (9) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 25 Temperature: Closed Containers Container and Ambient Temperature 75.0 Temp. Variation 65.0 55.0 Temperature (°C) Trucking Container Container Temp (°C) Outdoor Temp (°C) 45.0 35.0 25.0 15.0 0 50 100 150 200 250 300 350 400 450 Hours © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 26 Temperature: Long-Term Exposure For electronics used outside with minimal power dissipation, the diurnal (daily) temperature cycle provides the primary degradation-inducing load Phoenix, AZ Month Cycles/Year Ramp Dwell Max. Temp (oC) Min. Temp. (oC) Jan.+Feb.+Dec. March+November April+October 90 60 60 6 hrs 6 hrs 6 hrs 6 hrs 6 hrs 6 hrs 20 25 30 5 10 15 May+September June+July+August 60 90 6 hrs 6 hrs 6 hrs 6 hrs 35 40 20 25 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 27 Humidity / Moisture (Rules of Thumb) Non-condensing Condensing Standard during operation, even in outdoor applications Due to power dissipation Can occur in sleep mode or non-powered Driven by mounting configuration (attached to something at lower temperature?) Driven by rapid change in environment Can lead to standing water if condensation on housing Standing water © 2005 2004 - 2010 2007 Indirect spray, dripping water, submersion, etc. Often driven by packaging 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 28 Dimensions Keep dimensions loose at this stage Case study: Use of 0201 chip components Large number of hardware mistakes driven by arbitrary size constraints Examples include poor interconnect strategies and poor choices in component selection Tight dimensional requirements push designer towards wholesale placement of 0201 components 0201 is not yet an appropriate technology for systems requiring reliability Result: Major issues at customers Use the Toyota approach © 2005 2004 - 2010 2007 Except with sudden acceleration 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 29 Toyota Approach Toyota's development engineers are 4X as productive as U.S. counterparts. Why? Focus on learning as much as possible Use of that knowledge to develop a stream of excellent products Western engineers © 2005 2004 - 2010 2007 Define several product concepts Select the one that has the most promise Draw up specifications and divide them into subsystems; Subsystems are designed, built and rolled up for system testing. Failures? Rework the specs and the designs accordingly (nonoptimized and confusing endeavor) Toyota engineers Efforts concentrated at lowest possible design level Thorough understanding of the technology of a subsystem so it can be used appropriately in future designs 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 30 Toyota Example: Radiators Traditional approach: Design radiator for a specific vehicle based on mechanical specifications written for that vehicle Toyota considers a range of radiator solutions based on cooling capacities and the cooling demands of various engines that might be used. How the radiator actually fits into a vehicle would be kept loose so that Toyota's knowledge of radiator technology could be used to create the optimum design Toyota's system is "test & design" rather than the traditional "design & test." © 2005 2004 - 2010 2007 Toyota engineers test at the fundamental knowledge level so they don't have to test at the later, more expensive stages of design and prototyping 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 31 Design for Reliability Part Selection © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 32 Part Selection The process of creating the bill of materials (BOM) during the ‘virtual’ design process Before physical layout For some companies, this is during the creation of the approved vendor list (AVL) © 2005 2004 - 2010 2007 Design-independent 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 33 Part Selection (cont.) KIS: Keep it Simple New component technology can be very attractive Not always appropriate for high reliability embedded systems Be conservative Reality: Marketing hype FAR exceeds actual implementation © 2005 2004 - 2010 2007 Component manufacturers typically use portable sales to boost numbers Claim: We have built 100’s of millions of these components without a single return! Actuality: All sales were to two cell phone customers with lifetimes of 18 months 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 34 Part Selection (cont.) Even when used by hi-rel companies, some modifications may have been made Example: State-of-the-art crystal oscillator required specialized assembly to avoid failures one to three years later in the field Prior examples of where care should have been taken © 2005 2004 - 2010 2007 New technologies: X5R dielectric, SiC diodes, etc. New packaging: Quad flat pack no lead (QFN), 0201, etc. 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 35 Part Selection (cont.) As technology progresses, functional performance has become a limited aspect of the part selection process Other concerns are increasingly taking center stage Moisture sensitivity level (MSL) Temperature sensitivity level Electrostatic discharge (ESD) classification Manufacturability (Design for Assembly) Plating material Lifetime / Long-term reliability © 2005 2004 - 2010 2007 Sometimes Physics of Failure is required 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 36 Critical Components Most small to mid-size organizations do not have the resources to perform a thorough part selection assessment on every part Does not excuse performing this activity Requires focusing on components critical to the design Critical Components: A narrowed list of components of most concern to the OEM Sensitivity of the circuit to component performance Number of components within the circuit Output from FMEA / FTA Past experiences Complexity of the component Industry-wide experiences © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 37 Critical Components (Industry Experience) Optoelectronics Low volume or custom parts Non-volatile memory has limited data retention time and write cycles Parts with mechanical movements (switches, relays, potentiostats, fans) Part is no longer a commodity item Memory devices High volume controls not always in place Wearout can initiate far before 20 years Depending on environment, wear out can initiate far before 20 years Surface mount ceramic capacitors © 2005 2004 - 2010 2007 Assembly issues 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 38 Critical Components (cont.) New technologies or state-of-the-art Electronic modules Part is a miniature assembly (no longer a commodity item) Power components Fuses At the limit of the manufacturer’s capabilities MEMS, 45-nm technology, green materials, etc. Susceptible to quality issues Electrolytic capacitors © 2005 2004 - 2010 2007 Depending on environment, wear out can initiate far before 20 years 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 39 Part Selection (ESD) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 40 Part Selection for ESD Industry movement to decreasing feature sizes and high frequency technology 90nm → 65nm → 45nm GaAs / SiGe desirable at high GHz Increasing ESD risks © 2005 2004 - 2010 2007 More parts are ESD susceptible ESD sensitivity is increasing (is Class 0 still sufficient?) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 41 Parts Selection for ESD (cont.) Know the ESD rating for each part and select parts with the best ESD rating Identify all ESD Sensitive Parts on drawings Mark Locations of ESD Sensitive parts on the Board with the ESD symbol Be aware that the appropriate ESD rating is driven by part location © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 42 ESD Sensitivity and Location Use ESD Protection on all susceptible parts Box or System I/O ESD Rating < Class 2 HBM IEC (4000V, 150pf, 330 Ohm) MANDATORY Internal Components (not exposed to outside connectors) ESD Rating <= Class 1 ANSI (0-999V) MANDATORY ESD Rating < Class 2 ANSI (2000V) WHEREVER POSSIBLE High Speed, RF and GaAs parts will be sensitive to ESD [Class 0 (<250V) or Class 1A (<500V)] © 2005 2004 - 2010 2007 Place ESD sensitive components and traces to avoid locations where the board may be handled Avoid Coupled ESD events – Do not route traces to ESD sensitive parts near lines connected to the outside world Install protective devices before ESD sensitive parts (Class 1 or lower) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 43 Evaluate Potential ESD If ESD sensitive parts are used in design, the circuitry connected to device pins should be evaluated Insure that it provides “attenuation” to prevent voltage in excess of the parts ESD rating from developing in case the pin or connected traces are contacted during board handling or system assembly. Often the recommended circuit components for operation of the part will provide adequate ESD protection. This should be verified by analysis or simulation and extra protection added as required to limit the voltage seen at the part. Assumptions for analysis/simulation © 2005 2004 - 2010 2007 2000V,1.5K, 100pf for Internal circuits 4000V, 330 Ohms, 150pf for I/Os 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 44 Part Selection (Plating Material) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 45 Tin Whiskers – Failure Modes Courtesy of P. Bush, SUNY Buffalo Direct Contact Requires growth of sufficient length and in the correct orientation Electromagnetic (EM) Radiation Causes an electrical short Emits or receives EM signal and noise at higher frequencies Deterioration of signal for frequencies above 6 GHz independent of whisker length DfR Solutions Debris Whisker breaks off and shorts two leads (primarily during handling) © 2005 2004 - 2010 2007 Observation of tin whisker debris as reported to NASA from Sanmina-SC 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 46 Tin Whiskers – Root Cause Whiskering behavior is speculated to be driven by stress gradients Results in preferential diffusion paths Stresses can be extrinsic or intrinsic Extrinsic: pressure, bending, corrosion Intrinsic: plating, intermetallics The grains that whisker are supposedly based on orientation and oxide morphology © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 47 What Do We Know about Tin Whiskers? It can take days to years before tin whiskers start to grow Growth rates can vary by orders of magnitude Maximum lengths can vary by orders of magnitude Usually 1 to 5 µm Current carrying capability Highly dependent upon plating type and substrate material Bright (?) tin on steel - 25 mm; Matte tin on copper - 0.5 mm Diameters of 6 nm to 7 µm recorded 0.1 to 200 microns per day Typically 10 - 35 mA, up to 75 mA observed Plasma arcing, up to 200 A possible Plating type plays a very important role Source GEB-0002 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 48 Bright vs. Matte: Plating Differences Bright tin can be “bad” plating and matte tin can be “better” plating, not necessarily “good” plating Problem: quantitative definitions of bright tin and matte tin can vary What are the differences? Appearance Sulfate, stannate, methane sulfonate (MSA) or fluoro-borate baths Carbon content in plating (organic brighteners and other agents) Grain size and grain orientation (lattice structure) Grain Size Carbon Content © 2005 2004 - 2010 2007 Bright Tin Matte Tin Tyco < 1 µm ~ 3 µm iNEMI 0.5 – 0.8 µm 1 – 5 µm Tyco < 0.3% < 0.05% iNEMI 0.2 – 1.0% 0.005 – 0.05% 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Current terminology considered ineffective in capturing whisker risks 49 What Else Do We Know? Temperature cycling primarily accelerates whiskering when there is a large mismatch in thermal expansion Example: Tin-plated Alloy 42 (Ni-Fe alloy) Tends to be less effective for tin-plated copper Temperature/Humidity Tends to be more effective for tin-plated copper ‘Corrosion’ sites, areas of condensation, tend to accelerate whisker growth Electric Field No effect Vibration / Mechanical Shock Whiskers do not break © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 50 Tin Whiskering and Alloy 42 Industry seems to be unable to prevent whiskering of tin plating over discrete (‘loose’) Alloy 42 components Standard mitigation techniques are not available (NiPdAu) or have limited influence (nickel underplate, anneal) Whisker lengths > 67 microns (Class 1 as per JESD201) when exposed to JESD 22A121 temperature cycle Exposure to Pb-free reflow (peak temp > 232C) may provide mitigation (Micron) Maximum whisker lengths may saturate at values lower than tin over copper © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 51 What Don’t We Know? Tin whiskering is a complex problem Drivers are a combination of Materials and Environment Materials: metals, intermetallics, organic inclusions, affect grain size, grain structure, grain boundary behavior Depends on: substrate materials, substrate thickness, substrate geometry, plating bath composition, plating thickness, plating material composition, plating microstructure Environment: temperature, humidity, corrosion, stress gradients Compressive stress gradients (global vs. local) play a very important role © 2005 2004 - 2010 2007 Intermetallic formation, residual stresses in tin plating, external stress, mechanical loading, coefficient of thermal expansion (CTE) mismatches, corrosion or oxidation, damage to surface 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 52 Don’t Know (cont.) No organization has identified a technique to predict maximum whisker length for a particular component No organization has identified a plating parameter or additive (other than Pb) that will consistently (under all conditions) prevent plated tin from whiskering What’s left? © 2005 2004 - 2010 2007 Mitigation 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 53 Tin Whiskering (Mitigation) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 54 How to Mitigate Tin Whiskers? The first step is to focus mitigation on critical components Critical components are based upon three pieces of knowledge © 2005 2004 - 2010 2007 The overwhelming majority of tin plated electronic parts are matte tin over copper Matte tin over copper produces whiskers of a finite length Whiskers tend to only break off during handling 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 55 Maximum lengths – Matte Sn over Cu Reference Environment Time Period Maximum Length JEITA[10] 25C/50%RH 4000 hours 150 microns Peng (Freescale)[1] 60C/93%RH 3000 hours 70 microns Hashemzadeh (Linköping)[2] -60C/60C 500 cycles 130 microns Okada (Murata)[3] -40C/85C 2200 cycles 85 microns Gedney (iNEMI)[4] 60C/95%RH 3000 hours 60 microns -55C/85C 3000 cycles 35 microns iNEMI – DOE3)[5] 60C/93%RH 8000 hours 270 microns Brusse (NASA)[6] -40C/90C 500 cycles 250 microns 60C/93%RH 5000 hours 450 microns Room Conditions 8500 hours 100 microns -55/85C 3000 cycles 40 microns Romm (TI)[8] 51C/85%RH 3634 hours 34 microns Dittes (E4)[9] 30C/60%RH 450 days 275 microns -55C/85C 3000 cycles 35 microns Hilty (Tyco)[7] [1] Peng Su, Min Ding, and Sheila Chopin, "Effects of Reflow on the Microstructure and Whisker Growth Propensity of Sn Finish", Proceedings of the 55th ECTC Study of Tin Whisker Growth and their Mechanical and Electrical Properties, Moheb Nayeri Hashemzadeh, Undergraduate Thesis, Linköping University, 2005 [3] S. Okada, et al, "Field Reliability Estimation of Tin Whiskers Generated by Thermal Cycling Stress", Capacitor and Resistor Technology Symposium (CARTS) Europe, October 2003 [4] R. Gedney, J. Smetana, N. Vo, G. Galyon, "NEMI Tin Whisker Projects", Second International Conference on Lead Free Electronics, June 21-23, 2004 (Amsterdam) [5] “Tin Whisker Test Committee Results (Phase 3)", Presentation at ECTC NEMI workshop, June 1, 2005 [6] J. Brusse, "Tin Whisker Observations on Pure Tin-Plated Ceramic Chip Capacitors", Proceedings of the American Electroplaters and Surface Finishers (AESF) SUR/FIN Conference, June 24-28, 2002, pp. 45-61 [7] Tin Whiskers in Electronic Components, Bob Hilty, www.tycoelectronics.com/environment/ leadfree [8] D. Romm, D. Abbott, S. Grenney, M. Khan, "Whisker Evaluation of Tin-Plated Logic Component Leads", Texas Instruments Application Report SZZA037A, February 2003 [9] Standard for Whisker Acceptance Level in Pb-free Lead Finish. A Step towards Green Packaging for Consumer Products to Drive Future European Strength, Final Report. Contract No: IST-2001-37826. 01.06.2002 - 30.11.2004, Marc Dittes, Infineon Technologies [10] Tin Whisker Workshop, ECTC2005, Tin Whisker Test Committee Results (Phase 3) [2] © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 56 When Do Really Long Whiskers Occur? Usually bright tin and/or tin plating over a substrate material other than copper (brass, bronze, steel, etc.) NASA (Leidecker): 25 mm over +10 years Bright tin over BeCu(?) or steel(?) © 2005 2004 - 2010 2007 DfR (Fischer): +2 mm over 6 months Tin (matte?) over brass 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 57 Tin Whiskers: Effect of Substrates Sakuyama (FUJITSU Sci. Tech. J., 2005) Sn over Phosphor-Bronze 10X longer than Sn-Cu over Copper © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 58 Tin Whiskers: Critical Components JEDEC JESD201 Spacings of less than 500 microns Contact points (connector flex circuitry) Stress gradients could change maximum length Welds (electrolytic capacitors) Parts with 0.8 mm lead pitch or less 0201 chip components Metal can housing Stress gradients could change maximum length Note: some organizations specify the critical spacing as 350 microns (e.g., EU) © 2005 2004 - 2010 2007 0.65 mm pitch or less 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 59 How to Mitigate? There are three basic approaches to mitigation Data Gathering and Monitoring Part Manufacturer Mitigation Steps offered by your suppliers Equipment Manufacturer (OEM) Mitigation © 2005 2004 - 2010 2007 Steps you have to perform yourself 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 60 Data Gathering and Monitoring Driven by iNEMI and JEDEC (JESD22A121.01, JESD201, JP002) Industry recommended qualification tests Ambient (25ºC/50%RH, 4000 hrs) Elevated (55ºC/85%RH, 4000 hrs) Cyclic (-40 to 85ºC, 1500 cycles) Shorter test times for consumer products Use manufacturer’s data, require thirdparty testing, or perform your own © 2005 2004 - 2010 2007 Visual inspection should be performed properly http://nepp.nasa.gov/whisker/video/ inspection/index.html 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 61 JESD201: Potential Issues Document allows for test coupons versus actual components (JESD 22A121-01, sec. 5.1) Document does not specify if components are discrete (‘loose’) or board-mounted Temperature cycling and elevated humidity exposure of non-preconditioned discrete parts may not be representative of storage conditions Board-mounted parts may not be representative of hand assembly or repair conditions Test times are not based on a failure model © 2005 2004 - 2010 2007 Unable to correlate to time in the field for various use environments 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 62 Part Manufacturer Mitigation Nickel underplate between the tin plating and the copper leadframe Some question about effectiveness (IBM vs. TI) Some question about minimum thickness iNEMI article recommends minimum of 2 microns (Smetana, Global SMT, 6/07) ATIS requires minimum of 2 microns PC manufacturer requires 1.2 microns JP-002 requires 0.5 microns What if the nickel underlayer cracks? Might not help for ceramic substrate due to CTE mismatch Anneal for 1 hour at 150ºC within 24 hours of plating This is the approach for Freescale © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 63 Part Manufacturer Mitigation (cont.) Fusing (melting of tin through dipping in a hot oil bath) Excellent field history; must be performed soon after plating Minimum plating thickness © 2005 2004 - 2010 2007 Some question about minimum thickness Telecom manufacturer requires 10 microns JP-002 recommends 7 microns ATIS requires 4 microns (with nickel underplate) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 64 Mitigation (Sn-Bi and Sn-Ag) Some component suppliers promoting bismuth and silver additives (1 – 3 wt%) Not as accepted as other mitigation techniques However, increasing acceptance by some OEMs (e.g., Rockwell Collins) Concerns Low melting point SnPbBi alloys Some claim test results demonstrating this behavior Others state thermodynamically impossible below 6 wt% Gordon, Northrop Grumman © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 65 Part Manufacturer Mitigation (cont.) Some component manufacturers claiming proprietary whisker-free plating formulation Be skeptical; require Statistical Process Control Request palladium (Pd) plating – NiPdAu © 2005 2004 - 2010 2007 Increasingly offered as an option, even to low volume customers (medical, industrial controls, etc.) Most manufacturers have moved to Pd as a standard plating for fine-pitch components 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 66 Mitigation: Lead Frame Platings Company Amkor Package Plating Intel QFP / TSOP Sn[1] Samsung QFP / TSOP NiPdAu Texas Instruments QFP / TSOP NiPdAu TSOP (Discretes) NiPdAu TSOP (Memory) SnAg or SnCu TSOP (LSI) NiPdAu or SnAg or SnBi QFP / TSOP NiPdAu QFP Sn or SnPb TSOP NiPdAu QFP Mostly Sn-Cu, Sn-Bi; some NiPdAu TSOP Mostly NiPdAu, with some Sn-Cu, Sn-Bi QFP / TSOP Pd or SnPb QFP Sn TSOP NiPdAu Hynix TSOP SnBi Freescale QFP / TSOP Sn NEC QFP / TSOP Sn, SnBi, or NiPdAu Micron TSOP Sn, SnPb QFP Pd Toshiba STMicroelectronics Infineon Renesas Technology Sony Philips/NXP Palladium and SnBi are seeing an increasing market share due to concerns with tin whiskering Matsushita/Panasonic TSOP SnBi AMD QFP Sn, SnCu, or SnPb IBM QFP N/A Qualcomm N/A N/A QFP SnBi TSOP SnBi QFP SnBi, NiPdAu TSOP SnBi Fujitsu Sharp © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 67 OEM Mitigation Four Options Procurement / Design Procurement, Re-packaging, Post-plate or dip, Conformal coat Select only components with SnPb or Pd plating May require complete change in circuit design if alternative component required Rarely performed (functionality trumps reliability) Subcontract packaging or Re-packaging © 2005 2004 - 2010 2007 SnPb or Pd plated leadframes Rarely performed (cost, risk of damage) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 68 OEM Mitigation: Post-Plate or Dip Options Reflowing solder Replating (passive components / AEM) Needs bake for enhancing interdiffusion Hot-tin dipped coating Stripping and refinishing with whisker-free plating material $0.10 - $0.40 per part (4000 part minimum) Avnet considering stocking this option (Rollins / Minter) Lead (Pb) overplating Application of solder over the entire exposed lead Whiskers do not occur where solder is present Found not be to effective (iNEMI) Hot solder dip with SnPb © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 69 OEM Mitigation: SnPb Solder Dipping Initially relevant only to products not covered by RoHS GEIA-STD-0006 (Draft) Gap at component body could still result in whiskers Risk of solder bridging and popcorning on fine pitch; thin components with large die areas will be problematic Some programs have erroneously assumed ‘qualification by similarity’, both process and part, and ended up with damaged parts (Rollins / Minter) Costing (Rollins / Minter) Requirements for Using Solder Dip to Replace the Finish on Electronic Components Problems may outweigh solution Fine-pitch components now exempt from RoHS $2.00-3.50 per part for 100, plus $450-$900 set-up fee Avnet considering stocking this option (Rollins / Minter) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 70 OEM Mitigation: Conformal Coating Mitigation #1: Prevention or delay of whiskering Some indication of 2 - 6 year delay in whiskering (Rollins / Minter) NASA reported ~50um polyurethane coating prevented whisker penetration after 9 years of office storage Short tin whiskers will eventually penetrate all current conformal coatings © 2005 2004 - 2010 2007 No definitive trend in regards to coating properties 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 71 Conformal Coating (cont.) 2nd International Symposium on Tin Whiskers Mitigation #2: Buckling Based on calculations (Leidecker, NASA) Experimentally proven? (Brusse, NASA) Limitations Insufficient coverage at leads (gravity) Problems with conformal coat may outweigh possible risk avoidance Fine pitch components may have no air gap! Current status Development of whisker-resistant coating © 2005 2004 - 2010 2007 Patents filed Assessment of single vs. double coating 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Leadframe Conformal Coating Tin Whisker Conformal Coating Leadframe 72 Conformal Coating (NEW) Woodrow, 2008 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 73 Conformal Coating (NEW) Recent GEIA-Industry Survey (Touw, 2010) Strong mitigators Parylene C Epoxy > 3 mil Reasonable mitigators Uralane >1mil Urethane 1-2mil Epoxy 1-3mil Arcylic >2mil Silicone >3mil ALD Weak mitigators Silicone 2-3mil Acrylic 1-2mil © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 74 Strong Mitigators (>=4) (NEW) Physical Barrier (1resp <=1) Hard potting (2 resp <=1) SnPb dip with Chemical dip (4 resp <=1) Soldered with SnPb Solder with qualification / verification of coverage (with or without sampling) (4 resp <=1) Some Conformal Coats Parylene C (1 resp <=1) Epoxy >3mil (2 resp <=1) Spacing >500mils (2 resp <=1) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 75 Reasonable Mitigators (>3) (NEW) Many Conformal Coats Uralane >1mil (1 resp <=1) Urethane 1-2mil (4 resp <=1) Epoxy 1-3mil (4 resp <=1) Arcylic >2mil (5 resp <=1) Silicone >3mil (5 resp <=1) ALD (2 resp <=1, low response) Soft potting or encapsulation (6 resp <=1) Soldered with SnPb without verification on bottom termination or small leadless chip parts (6 resp <=1) Spacings 100-500mils (2 resp <=1) SnPb with 1% to 3% Pb (errors in 3% process) (7 resp <=1) Component Level Redundancy (2 resp <=1) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 76 Weak Mitigators (>2) (NEW) Spacings 50-100mils (4 resp <=1) Nickel Underplate (6 resp <=1) SnPb dip with standoff (7 resp <=1) Soldering with SnPb on large leadless chip and thru-hole (12 resp <=1) SAC finish or dip (with or without standoff) (13 resp<=1) SnAg finish (11 resp <=1) Heat treatment above fusing temperature (10 resp <=1) Some Conformal Coats Silicone 2-3mil (7 resp <=1) Acrylic 1-2mil (8 resp <=1) Styrofoam fillers (9 resp <=1; low response rate) Avoiding compressive stress (8 resp <=1) Medium Current (11 resp <=1) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 77 Very Weak or Non-Mitigators (<2) (NEW) Soldering with SnPb without verification on surface-mount (4 resp >=4) Spacing <50 mils (10 resp >=4) Hot dipped tin (7 resp >=4) Silver Underplate (1 resp >=4) SnBi finish (4 resp >=4) Heat Treatment below fusing (1 resp >=4) Large grain size or claimed matte tin (1 resp >=4) Thick Tin (2 resp >=4) Immersion Tin (1 resp >=4) <1% Pb (1 resp >=4) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 78 Tin Whisker Field Failures (post RoHS) Crystal Oscillator Competitors used nickel instead Tin over steel Time to failure: N/A Length: ~1.5 mm Product: Industrial server Bright tin over copper Time to failure: N/A Length: ~1.5 mm Product: Electrical distribution Connector contact crimped onto a flex circuit Tin over nickel over copper Time to failure: Less than 2 years Length: N/A Product: Medical What can we learn from these failures? Time to failure: ~6 months Length: ~2-3 mm Product: Appliance controls D-style connector RJ45 Connector Tin over brass (internal surface) Concept of critical components works! All failures are wrong plating (bright), wrong substrate (steel, brass) or involve contact points Keep a close eye on connectors © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 79 FFC-Touch pad-Interposer board Connector on Touch pad TP to interposer FFC Interposer to MB FFC FFC connector on interposer (Elco) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 80 Sn Whiskers on SnCu plated Flex (FPC) Evidence shows this is where most of the risk lies. Connector contact points © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 81 Sn Whisker on Flex Cables • JAE evaluated many Sn based finishes but all showed unacceptable whisker growth. • Whisker growth was acceptable only if Sn plating thickness was reduced to below 0.5 microns but then contact resistance degraded too much with high temp aging. • They concluded the following: Recommendation for high value electronics is to only use gold plating on flex cables and mated connectors. Note: a tin plated connector to gold flex cable could result in fretting issues or galvanic corrosion. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 82 Part Selection (Misc) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 83 Ceramic Capacitors (Voltage Effect) High C/V caps can display a significant voltage effect X5R/X7R (1 µF) Operating at 80% of rated voltage can result in a 75% drop in capacitance 0603 case size Similar performance for 6.3, 10, and 16 VDC Same capacitor © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 84 Ceramic Capacitors (Cyclic Voltage) Reports of field failures of MLCC in AC or pulsed DC voltage Piezoelectric effect Variant voltage will vary internal stresses, potentially inducing fatigue behavior With high frequency ripple current, capacitor can vibrate (resonate). Fatigued specimens can contain scattered microcracks Decrease in capacitance; increase in leakage current Concern at hundreds of kHz http://www.avxcorp.com/docs/techinfo/parasitc.pdf Decreases with increasing capacitance, X7R -> Y5V Avoid or use AC-rated capacitors Case Size English Metric 3025 7563 2220 5750 1812 4532 1210 3225 1206 3216 0805 2012 0603 1608 0402 1005 Resonance Frequency 250 - 750 kHz 300 - 900 kHz 400 - 1200 kHz 600 - 1200 kHz 600 - 1600 kHz 900 - 1800 kHz N/A N/A Nippon Chemi-con, CAT.No.E1002l © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Sang-Joo Kim and Qing Jiangy, Microcracking and electric fatigue of ferroelectric ceramics, Smart Mater. Struct. 5 (1996) Phone (301) 474-0607 polycrystalline Fax (240) 757-0053 www.DfRSolutions.com 85 Resistors (High Resistance) Board surfaces can be susceptible to periodic SIR drops Can interfere with high resistance resistors Especially with no-clean Duration as short as 1 min Down to 1 megaohm Fine pitch, high voltage especially susceptible Especially chip resistors Intermittent in nature Avoid values > 500 kohms if used for sensing or calibration © 2005 2004 - 2010 2007 Consider lower values in series Use guard banding or cutouts 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 86 Sulfide Corrosion of Thick Film Resistors Sulfur dioxide (SO2) and hydrogen sulfide (H2SO4) in environment Sources: Black rubber, industrial pollution Attacks silver material under passivation/ termination Creates nonconductive silver sulfide Drivers Cracking/separation of coating/termination Potting or conformal coating Poor manufacturing Thermal shock Seems to act as a ‘sponge’ Holds SO2 molecules in place Electrical opens within 1-4 years Avoidance Orient parallel to solder wave © 2005 2004 - 2010 2007 Entrance side can experience thermal shock Avoid hand soldering/rework Sulfur-resistant PdAg material (KOA) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 87 Tantalum and Polymeric Capacitors Tantalum capacitors are selected for volumetric efficiency Older technology can be susceptible to ignition Requires aggressive derating (50% or greater) Sensitive to higher temperatures (>85C) and certain circuits Newer, polymeric capacitors are available © 2005 2004 - 2010 2007 Significant reduction in ESR Less derating No risk of ignition 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 88 Electrolytic Capacitors Voltage Maintain a minimum of 25-33% of rated voltage (maintains the dielectric) Temperature Maintain adequate distance from ‘hot’ components 105C rated capacitors can be an issue at lower temperatures (below -40C) ESR increases 500X; capacitance decreases 80-90% Ripple Current Power resistors, IGBTs, etc. Seems to accelerate time to failure and can induce explosive rupturing Up to 100% or greater of rated ripple current Need to calculate/measure case temperature rise Equivalency on bill of materials is often not maintained Equivalent Series Resistance (ESR) © 2005 2004 - 2010 2007 Often not specified on the component data sheet 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 89 Separable Connectors Separable Connectors One of the most common failure sites First thing maintenance checks (plug / unplug) Hardware Design Rules Blind insertion increases risk of damage or mismating (consider flex or rigid-flex) All connectors should be keyed Prevents reversal of I/O pins Use positively retained connectors Avoid use of sockets Specify material and thickness © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 90 Separable Connectors [Gold (Au)] Connections and environments Hi-speed digital or critical connections Low voltage (< 5V), low current (< 10mA) Corrosive environment (gases such as H2S, SO2, Cl2) Risk of micromotion (< 2.5 µm) Material specifications © 2005 2004 - 2010 2007 Be-Cu or P-bronze base pins Nickel underplate (250 µin) Soft gold (Au) plating 10 µin (single insertion); 30 µin (50 insertions); 70 µin (hi-rel) Porosity spec No gold flash Contact specifications 50-100 grams contact force Minimum of 2 contacts; 4 preferred Adequate contact wipe – 0.010” min. No mating with tin plating 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 91 Separable Connectors [Tin (Sn)] Connections and environment Power connections Benign Tin plating design specification © 2005 2004 - 2010 2007 100 grams-force, 100 microinches (Tin Commandments) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 92 DfR (Misc.) Components taller then 1 inch © 2005 2004 - 2010 2007 Use of staking compound to adhere to board 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 93 Derating and Uprating © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 94 Component Ratings Definition A specification provided by component manufacturers that guides the user as to the appropriate range of stresses over which the component is guaranteed to function Typical parameters © 2005 2004 - 2010 2007 Voltage Current Power Temperature 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 95 Derating Derating is the practice of limiting stress on electronic parts to levels below the manufacturer’s specified ratings Guidelines can vary based upon environment (“severe, protected, normal” or “space, aircraft, ground”) One of the most common design for reliability (DfR) methods Goals of derating Maintain critical parameters during operation (i.e., functionality) Provide a margin of safety from deviant lots Achieve desired operating life (i.e., reliability) Sources of derating guidelines rd parties Governmental organizations and 3 OEM’s Component manufacturers Derating is assessed through component stress analysis © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 96 Derating Guidelines Government Agencies Air Force AFSC Pamphlet 800-27 (1983) Component Manufacturers Heavily legacy based EEE / Mechanical Parts Management & Implementation Plan for Space Station NASA Goddard MSFC-3012 (1999) NASA Johnson Space Center SSP 30312 (1999) Parts Requirement and Application Guide Heavily based on MIL standards 3rd Parties RIAC (Quanterion) Electronic Parts, Materials, and Processes for Space and Launch Vehicles NAVSEA SD-18 (2000) Canceled in 1998; Replaced with NASA Parts Selection List (NPSL) MIL-HDBK-1547 (1997) MIL-STD-975 (1994) Acquisition Management Part Derating Guidelines OEMs Boeing (MF0004-400) Multiple other OEMs; especially telecom EEE Parts Management and Control for Space Flight Hardware NASA Jet Propulsion Lab JPL-D-8545 (2003) © 2005 2004 - 2010 2007 Derating Guidelines 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 97 Derating Guidelines (Examples) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 98 Criticality of Component Stress Analysis Failure to perform component stress analysis can result in higher warranty costs, potential recalls Eventual costs can be in the millions of dollars Perspective from Chief Technologist at major Original Design Manufacturer (ODM) “…based on our experience, we believe a significant number of field returns, and the majority of no-trouble-founds (NTFs), are related to overstressed components.” © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 99 Derating Failures Where are the derating mistakes? Problem #1: Designers do not derate Failure to perform component stress analysis Problem #2: Derating does not have a practical or scientific foundation © 2005 2004 - 2010 2007 Extraordinary measures are taken when inappropriate Derating is excessive: ‘The more, the better’ rule 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 100 Failure to Derate Analog / Power Designs Derating is typically overlooked during transient events Especially turn-on, turn-off Digital © 2005 2004 - 2010 2007 Excessive number of components and connections tends to limit attempts to perform component stress analysis 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 101 Failure to Derate (Case Study) Customers of embedded system industrial controls experiencing intermittent failures Intensive failure analysis effort ($) and loss of customers Parametric drift measured on output control lines Many months to identify the problem Failure site identified as thick film resistor © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 102 Resistors (Pulse Conditions) Minimal variation when operated within specification Thin film and thick film resistors use laser trimming Power on, brown-outs, etc. Up to 350C Failure mode Resistance adjustment Can induce hot spots during transient conditions 0.5% at 70C for 30 years Gradual increase in resistance Two approaches © 2005 2004 - 2010 2007 Obtain peak pulse guidelines Design larger copper bond pads 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Courtesy of S. Silvus 103 The Foundation of Derating To be effective, derating must have a practical and scientific foundation Problem: Manufacturer’s ratings are not always based on a practical and scientific foundation Manufacturers’ viewpoint Ratings are based on specific design rules based on materials, process, and reliability testing The reality Ratings can be driven by tradition and market forces as much as science Best practice Based on data from field returns Based on test to failure qualification (especially for new suppliers) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 104 Manufacturer’s Derating (example) Tantalum capacitor Derating based on desired failure rate MnO2 cathode 10 ppm at startup Why not 10 ppm failure rate at rated voltage? Was 0.3% failure rate acceptable? 50% derating is a legacy Courtesy of Kemet © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 105 Scientific Approach to Derating (Ta Caps) High Impedance Circuits Self healing in Ta capacitors involves leakage paths in the MnO2 being healed by the transformation to the higher resistance compound Mn2O3 Process requires enough current to allow internal temperatures to reach 500°C Small amounts of current (< 50 uA) will prevent self healing Leads to degradation and potential component failure Avoid use in circuits with impedances greater than 100 kΩ © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 106 Derating Decision Tree Step 1: Derating guidelines should be based on component performance, not ratings Test to failure approach (i.e., HALT of components) Quantifies life cycle cost tradeoffs For smaller OEMs, limit this practice to critical components © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 107 Derating based on Test to Failure Derived Derating Limit Manuf. Rating Test Results Stress OEM was concerned with voltage rating of tantalum capacitors after 2 reflows and use on low resistance line Performed step stress surge test (SSST) Derived voltage derating based on a sub-ppm failure rate © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 108 Derating Decision Tree (cont.) Step 2: Derating guidelines should be based on recommendations from the component manufacturer They built it; they should know it Don’t trust the manufacturer? Use someone else Step 3: Derating guidelines should be based on customer requirements Step 4: Derating guidelines should be based industry-accepted specification/standard Be flexible, not absolute © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 109 Wearout Mechanisms and Physics of Failure (PoF) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 110 Physics of Failure: Reliability Assurance Reliability is the measure of a product’s ability to …perform the specified function …at the customer (with their use environment) …over the desired lifetime Physics of failure therefore requires an understanding of the desired lifetime and the use environment © 2005 2004 - 2010 2007 Desired lifetime: When the customer will be satisfied Use environment: Must include assembly, transportation, storage, operation 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 111 PoF-Based Reliability Prediction All physics-of-failure (PoF) based models are semi-empirical Calibration testing should be performed over several orders of magnitudes The basic concept is still valid Requires calibration Allows for the derivation of constants The purpose of PoF is to limit, but not eliminate, the influence of material and geometric parameters © 2005 2004 - 2010 2007 E.g., Solder: Testing must be re-performed for each package family (ball array devices, gullwing, leadless, etc.) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 112 Failure Rate Why is PoF Important? Electronics: Today and the Future Electronics: 1960s, 1970s, 1980s Wearout! No wearout! Time © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 113 Physics of Failure (PoF) Algorithms ⎛ ~ 0.51eV ⎞ T f ∝ exp⎜ ⎟ × exp(~ −0.063% RH ) kT ⎠ ⎝ n t1 ⎛ V2 ⎞ E ⎛1 1⎞ = ⎜⎜ ⎟⎟ exp a ⎜⎜ − ⎟⎟ t 2 ⎝ V1 ⎠ K B ⎝ T1 T2 ⎠ ⎛ L ⎛ 2 −ν ⎞ ⎞ h h L ⎟⎟ ⎟ + + s + c + ⎜⎜ ⎟ ⎝ E1 A1 E2 A2 As Gs Ac Gc ⎝ 9 ⋅ Gb a ⎠ ⎠ (α 2 − α1 ) ⋅ ∆T ⋅ L = F ⋅ ⎜⎜ Can be mind-numbing! What to do? © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 114 PoF and Wearout What is susceptible to long-term degradation in electronic designs? © 2005 2004 - 2010 2007 Ceramic Capacitors (oxygen vacancy migration) Memory Devices (limited write cycles, read times) Electrolytic Capacitors (electrolyte evaporation, dielectric dissolution) Resistors (if improperly derated) Silver-Based Platings (if exposed to corrosive environments) Relays and other Electromechanical Components Light Emitting Diodes (LEDs) and Laser Diodes Connectors (if improperly specified and designed) Tin Whiskers* Integrated Circuits (EM, TDDB, HCI, NBTI) Interconnects (Creep, Fatigue) Industry-accepted models exist Plated through holes Solder joints 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 115 Part Selection (Lifetime) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 116 Lifetime Example: Memory Devices Limited lifetime based on read-write cycles and retention time Parameter Minimum Guarantee Units Endurance 100,000 Data changes per bit Store cycles 1,000,000 Store cycles Data retention 100 Years Some memory devices provide data retention time for different operating temperatures (20 years at 125°C and 10 years at 150°C). © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 117 Light and Laser Diode Wearout Increasing importance with adoption of optical communications and LCD backlight Standard model for on-die wearout is: ⎛ Ea ⎞ t f = A( J ) exp⎜ ⎟ ⎝ kT ⎠ n where A is a constant, J is the current density, n is an exponent Expression applies for units run in automatic current control (ACC), or constant current. n =1.5 - 2 for a large number of different LED structures n = 6 – 7 for laser diodes with facet passivation Units run at constant output power (APC), power substitutes current density (n may be higher) Some models will combine power and current density Note: Model does not apply to die attach fatigue © 2005 2004 - 2010 2007 A risk in high power, cyclic applications 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 118 Light and Laser Diode Wearout (cont.) Estimated lifetime is not always provided When lifetime is provided, it is MTBF at room temp. Time to 5% failure can be half the time Time to failure at 40C can be half the time 50K hrs can turn into 12.5K hrs Lifetime is not always be equivalent to failure © 2005 2004 - 2010 2007 50% reduction in intensity 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 119 Lifetime Example: Relays/Switches Relays are an electromechanical switch Minimum of four I/Os Control voltage Signal voltage What are the major concerns in regards to relay reliability? © 2005 2004 - 2010 2007 Number of cycles to failure Long-term non-use Power dissipation and contact resistance (heating and voltage drops) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 120 Relays/Switches (cont.) Selection of appropriate plating Idle for long periods of time: Gold contacts Numerous cycles: AgCd contacts Sealed packages if cleaning operations Use of protective devices Diode, resistor, capacitor, varistor, etc. Prevents arcing during switching (accelerates degradation) Must be nearby Temperature rise Wide range of contact resistance in specifications Ensure margin between design life requirements and manufacturer’s specifications © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 121 Wearout of Relays/Switches 100% Cumulative Failures % 90% Catalog Life = 100,000 Cycles Mean Time Between Failures = 102,900 80% 70% 60% 50% Characteristic Life = 113,545 40% 30% 20% 10% 0% 60,000 70,000 80,000 90,000 100,000 110,000 120,000 130,000 Number of Cycles © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 122 Lifetime Example: Ceramic Capacitors Ceramic caps are typically not expected to experience ‘wearout’ during normal operation n Ea ⎛ 1 1 ⎞ t1 ⎛ V2 ⎞ ⎜⎜ − ⎟⎟ = ⎜⎜ ⎟⎟ exp t 2 ⎝ V1 ⎠ K B ⎝ T1 T2 ⎠ where t is time, V is voltage, T is temperature (K), n is a constant (1.5 to 7; nominally 4 to 5), Ea is an activation energy (1.3 to 1.5) and KB is Boltzman's constant (8.62 x 10-5 eV/K) Lifetime may be limited for extended value capacitors © 2005 2004 - 2010 2007 Sub-2 micron dielectric thickness Greater than 350 layers (increased failure opportunity) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 123 Voltage Derating Temperature Mfg. A Mfg. B Mfg. C Mfg. D 0.8 60oC 145 340 655 285 Difference between MTTF and t5% is a factor of 40 For a design with 20 ceramic capacitors, all product could fail within 3 - 15 yrs Randall, et. al., CARTS 2003 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 124 Voltage Derating Temperature Mfg. A Mfg. B Mfg. C Mfg. D 0.8 60oC 145 340 655 285 Extrapolation can result in a factor of 250 difference between MTTF and t1% 1% failure rate in less than 10 years in benign conditions (0.5 derating, 45oC) Randall, et. al., CARTS 2003 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 125 Ceramic Capacitor Wearout (cont.) When to be concerned? Dielectric thickness less than 2 - 3 microns Approx. 2 - 10 µF/mm3 http://www.murata.com/articles/ta0463.pdf Case size/capacitance guide X7R X5R/Y5V © 2005 2004 - 2010 2007 0402 > 0.05 µF; 0603 > 0.2 µF; 0805 > 1 µF; 1206 > 5.6 µF 0402 > 0.5 µF; 0603 > 2 µF; 0805 > 5.6 µF; 1206 > 16 µF 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 126 Desired Lifetime (IC Wearout) Process Variability confidence bounds 1000 100 Airplanes Mean Service 10 life, yrs. Computers laptop/palm cell phones 1.0 0.1 Technology 0.5 µm 0.25 µm 1995 130 nm 65 nm 2005 Year produced 35 nm 2015 Known trends for TDDB, EM and HCI degradation (ref: extrapolated from ITRS roadmap) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 127 IC Wearout: Background Guaranteeing the conventional 10-year life of ICs is going to become increasingly difficult….the progressive degradation of the electrical characteristics of transistors and wires will start dominating over abrupt functional failures. Furthermore, the mean-time-to-first-softbreak will significantly diminish." Antonis Papanikolaou, IMEC, Leuven, Belgium, 2007 …the progress of Moore's Law means that transistor wear-out and statistical performance issues are beginning to cross over from the realm of academic and hypothetical discussion to real-world R&D engineering. EE Times Europe, 2007 ”The notion that a transistor ages is a new concept for circuit designers,” says Chris Kim (U of Minnesota). Transistor aging has traditionally been the bailiwick of engineers who design the processes that make transistors; they also formulate recipes that guarantee the transistors will operate within a certain frequency and other parameters typically for 10 years or so…But as transistors are scaled down further and operated with thinner voltage margins, it’s becoming harder to make those guarantees… transistor aging is emerging as a circuit designer’s problem. IEEE Spectrum, June 2009 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 128 IC Wearout (cont.) 10 FIT (000) 8 Dielectric breakdown failure rate vs. feature size Future… In design 6 4 It is becoming more challenging to achieve very high reliability for products made with advanced technologies (90nm and smaller) Phil Nigh, IBM Microelectronics Estimated In production 2 0 Measured 180 130 90 65 45 Feature size, nm “failure rate increases as we scale to smaller technologies…hard failures will present a significant and increasing challenge in future technology generations.” Pradip Bose, Jude A. Rivers, et al., IBM T.J. Watson Research Center Increasing need for Repairable electronics Disposable electronics Fault-tolerant electronics Figure adapted from industry published data, 2008 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 129 Industry Testing Falls Short Limited degree of mechanism-appropriate testing Only at transition to new technology nodes Mechanism-specific coupons (not real devices) Test data is hidden from end-users Questionable JEDEC tests are promoted to OEMs © 2005 2004 - 2010 2007 Limited duration (1000 hrs) hides wearout behavior Use of simple activation energy, with incorrect assumption that all mechanisms are thermally activated, can result in overestimation of FIT by 100X or more 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 130 Generic CMOS Wearout Models EM: HCI: TDDB: NBTI: © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 131 Electromigration Occurs when collisions of electrons cause metal atoms in a conductor to dislodge and move downstream of current flow (proportional to current density) Electromigration primarily controlled through design rule checks Certain architectures are immune © 2005 2004 - 2010 2007 Polygons around an angle versus sharp angles Increasing concerns with via connections 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 132 Time Dependent Dielectric Breakdown Degradation occurs when electric fields from charges trapped in various parts of the oxide exceed dielectric breakdown Feature scaling Gate dielectric (oxide) thickness is scaled down Power supply voltage is approximately the same as previous node Increase in electric field on gate dielectric Breakdown is becoming a constant failure rate (independent of time) © 2005 2004 - 2010 2007 Weibull Beta values as observed by Wu et al (IBM) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 133 Electron Tunneling Similarities of HCI and NBTI Dependent on the electric field from the gate Phenomenon of electrons and holes (carriers) gaining sufficient energy to overcome the direction of induced current and become injected in the gate oxide Degradation exponentially increases after the first injection occurrence Depiction of hot carrier effects on MOSFET © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 134 Hot Carrier Effects Hot Carrier Effects or Injection (HCI) Characteristics HCI has inverse Arrhenius relationship Caused by a single carrier gains sufficient kinetic energy, or Multiple carriers undergoing collisions that force them out of the directional path of electric field Activation energy -0.2 to -0.1 eV Lower temperatures ( ~35 to ~55°C) increase vulnerability Probability of HCI increases with increases in bias voltage and cyclic changes in bias voltage (switching on/off) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 135 Bias Temperature Instability Combined bias-temperature stress is required for activation Affects transistors that are turned-on Fluctuations in temperature Noticeable influence on failure rate in submicron nodes Trap formation from electric fields are worse under negative bias High temperatures cause molecular instability Requires lower electric fields than HCI overall device + self heating property PMOS transistors with negative gate voltage The molecular instability strongly increases with temperature and higher negative gate bias © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 136 Multiple Mechanism Theory Simultaneous influence of multiple failure mechanisms Electromigration (EM) Time Dependent Dielectric Breakdown (TDDB) Negative Bias Temperature Instability (NBTI) Hot Carrier Injection (HCI) Each mechanism is driven by different conditions Activation energy Electrical and thermal conditions Transistor use conditions © 2005 2004 - 2010 2007 switching vs. constant bias 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 137 Validation Study Field return data was gathered from a family of telecommunication products 56 different ICs comprised 41.5% of the failed part population The validation activity was utilized failure data from 5 integrated circuits CAP 1% TRANSISTOR 19% RESONATOR 6% CARD 44% RESISTOR 8% INDUCTOR 0.07% IC 21% FUSE 0.54% © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com DIODE 0.13% FILTER 0.02% 138 Failure Rate Calculation Failure rate was calculated from raw data Environmental conditions to determine in-field operating temperature Thermal measurements to determine power dissipation Cumulative failure distributions Weibull Exponential Probability Plot for Days Probability Plot for Days Exponential Censoring Column in Censor - LSXY Estimates Weibull - 95% CI Censoring Column in Censor - LSXY Estimates Table of S tatistics S hape 1.02661 S cale 118401 M ean 117136 S tD ev 114111 M edian 82853.1 IQ R 127575 F ailure 96 C ensor 9099 A D* 2501.362 C orrelation 0.989 Probability 0.2 0.05 0.02 0.01 0.95 0.8 Table of S tatistics M ean 141824 StD ev 141824 M edian 98305.0 IQ R 155810 F ailure 96 C ensor 9099 A D* 2501.362 0.5 0.2 Probability 0.95 0.8 0.5 0.05 0.02 0.01 0.0001 0.0001 0. 1 0 1. .0 10 0. 10 0 00 10 .0 Days 0 0. 00 10 0 10 0. 00 0 0 10 00 00 .0 1 10 100 1000 Days 10000 100000 1000000 Failure distribution graphs for part: Freescale Microcontroller © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 139 Lifetime Prediction Determine functional group breakdown (component complexity) Calculate HTOL test ambient operating temperature Thermal resistance (junction to ambient) Power dissipation (PD, or Vdd(max) * Idd) Uses operating conditions and HTOL test conditions Part Number MT16LSDF3264HG-10EE4 M470L6524DU0-CB3 HYMD512M646BF8-J MC68HC908SR12CFA RH80536GC0332MSL7EN © 2005 2004 - 2010 2007 Field Temperature (°C) 42 42 42 40 58 Calculated Test Temperature (°C) 62.28 70.00 66.68 77.40 99.70 Vdd Field (V) Vdd Test (V) 3 2.5 2.6 5 1.276 3.3 2.7 2.7 5.5 1.34 Functional Groups DRAM: 268435456 bits DRAM: 536870912 bits DRAM: 1073741824 bits microcontroller: 512B RAM, 10-bit ADC microprocessor: 2080KB RAM 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 140 Results of Validation Study Results demonstrate the accuracy and repeatability of the multi-mechanism model to predict the field performance of complex integrated circuits © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 141 Solder Wearout (1st and 2nd Level) Component Packaging Requirements Higher frequencies and data transfer rates Tradeoff between ‘hard’ and ‘soft’ underfills More inside less Lower voltage, but higher current Solder wearout vs. cracking of high-K dielectrics Higher densities Lower resistance-capacitance (RC) constants Joule heating is I2R Has resulted in less robust package designs © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 142 Solder Wearout (cont.) Elimination of leaded devices Provides lower RC and higher package densities Reduces compliance Cycles to failure -40 to 125C © 2005 2004 - 2010 2007 QFP: >10,000 BGA: 3,000 to 8,000 CSP / Flip Chip: <1,000 QFN: 1,000 to 3,000 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 143 Solder Wearout (cont.) Design change: More silicon, less plastic Increases mismatch in coefficient of thermal expansion (CTE) BOARD LEVEL ASSEMBLY AND RELIABILITY CONSIDERATIONS FOR QFN TYPE PACKAGES, Ahmer Syed and WonJoon Kang, Amkor Technology. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 144 Solder Wearout (cont.) Hotter devices Increases change in temperature (∆T) tf = ∆Tn n = 2 (SnPb) n = 2.3 (SnNiCu) n = 2.7 (SnAgCu) Characteristic Life (Cycles to Failure) 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0 50 100 150 200 o Change in Temperature ( C) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 145 Alternatives to PoF Step 1: Rules of Thumb Step 2: Best Practice Follow part selection guidelines Step 3: Norris-Landsberg © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 146 Rules of Thumb (Constant Temperature) Electrolytic Capacitor lifetime becomes an issue when ambient temperatures begin to exceed 40C on a constant basis 85C/2000 hour ratings tend to be insufficient for more than 5 year life Many companies limit solder joint temperature to a maximum of 75ºC – 85ºC © 2005 2004 - 2010 2007 Some limit IC junction temperature to a similar range 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 147 Rules of Thumb (Temperature Cycling) In nominal environments, solder joint wearout is unlikely Low power, diurnal cycling ∆25C, 1 cycle per day Lifetime of less than 10 years Greater concerns in more severe environments © 2005 2004 - 2010 2007 Diurnal heat sources with sufficient fluctuation (∆40C) Diurnal power dissipation of ∆40C and greater Power cycling greater than 4 cycles/day (mini-cycling) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 148 Rules of Thumb (Temp Cycling)(cont.) If a solder joint fatigue is a concern, manage package styles © 2005 2004 - 2010 2007 MELF parts (SMA and SMB available) Crystals on ceramic substrates (especially large ones) Chip resistors greater than 1812 or capacitors greater than 2225 Large memory devices (44, 56, 66 I/O) with Alloy 42 leadframes Large I/O (≥ 44) quad flat pack no-lead (QFN) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 149 Rules of Thumb (Vibration) Maintain high board natural frequency Two to three times greater than low frequency peaks (>250-300 Hz) Use of attachments, stiffer rail guides When peaks in the power spectral density (PSD) curve exceeds 0.01 G2/Hz © 2005 2004 - 2010 2007 Lower threshold for higher frequency peaks 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 150 Vibration (cont.) Failures primarily occur when peak loads occur at similar frequencies as the natural frequency of the product / design Natural frequencies © 2005 2004 - 2010 2007 Larger boards, simply supported: 60 – 150 Hz Smaller boards, wedge locked: 200 – 500 Hz Gold wire bonds: 2k – 4kHz Aluminum wire bonds: >10kHz 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 151 Norris-Lanzberg (SnPb) 1/ 3 No ⎛ fo ⎞ = ⎜⎜ ⎟⎟ AF = Nt ⎝ ft ⎠ “o” refers to operating environment and “t” refers to test environment Provides comparison of test results to field reliability ⎞⎤ ⎟⎥ ⎟ ⎠⎥⎦ f is cycling frequency, ∆T is change in temperature, and Tmax is the maximum temperature ⎡ ⎛ 1 ⎛ ∆Tt ⎞ 1 ⎜⎜ ⎟⎟ exp ⎢1414⎜ − ⎜ ⎢⎣ ⎝ ∆To ⎠ ⎝ Tmax,o Tmax,t 2 Usable if the component manufacturer provides accelerated life testing (ALT) results for 2nd level interconnects Warning: Component manufacturers can cheat (use very thin boards) Can not provide an absolute prediction © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 152 Norris-Lanzberg (SAC) N o ⎛ ∆Tt ⎞ ⎟⎟ AF = = ⎜⎜ Nt ⎝ ∆To ⎠ ⎛ tt ⎞ ⎜⎜ ⎟⎟ ⎝ to ⎠ 0.136 ⎧⎪ ⎞⎫⎪ ⎛ 1 1 ⎟⎬ exp⎨2185⎜⎜ − ⎟⎪ T T ⎪⎩ max, o max, t ⎠⎭ ⎝ t is the hot-side dwell time, ∆T is change in temperature, and Tmax is the maximum temperature 2.65 “o” refers to operating environment and “t” refers to test environment Not yet widely accepted Found to be inaccurate within some datasets 1 N. Pan et al, “An Acceleration Model For Sn-Ag-Cu Solder Joint Reliability Under Various Thermal Cycle Conditions”. pp. 876-883, SMTAI, September 2005, Chicago, IL © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 153 Long-Term Reliability Rules of Thumb, Best Practices, and NorrisLandzberg are not always sufficient Good first pass When the risk is too high, physics of failure (PoF) calculations are irreplaceable © 2005 2004 - 2010 2007 Need to understand the basis for PoF models (broad range of test results) Trends are more important than single set of test data PoF Example: Pb-Free Solder 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 154 Physics of Failure (Pb-Free Solder – Thermal Cycling) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 155 Long-Term Reliability of Pb-free How to develop PoF for Pb-free Solder? As specified in GEIA-STD-0005-1: Test and analysis Knowledge of environmental and operating conditions Reliability data (test or in-service) Extrapolation of reliability data to applicable conditions (acceleration factor) Translation: Understand how Pb-free material fails and determine if / when these mechanisms will initiate in your design in the customer’s use environment over the lifetime of the product © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 156 Thermal Cycling Thermo-mechanical fatigue of solder joints is one of the primary wear-out mechanisms in electronic products May be changing Especially true in products used outside of commercial/ consumer environments © 2005 2004 - 2010 2007 Longer required lifetime More severe operating conditions 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 157 Thermal Cycling: SnPb vs. SAC Where does SnPb outperform Pb-free? Leadless, ceramic components Leadless ceramic chip carriers (crystals, oscillators, resistor networks, etc.) SMT resistors Ceramic BGAs Severe temperature cycles -40 to 125ºC -55 to 125ºC Syed, Amkor © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 158 Thermal Cycling: SnPb vs. SAC (Cont.) Pb-free alloy exhibits a higher acceleration factor © 2005 2004 - 2010 2007 Failures under accelerated life testing (ALT) equivalent to longer field life 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 159 Thermal Cycling: SnPb-SAC Transition I. Kim, ECTC 2007 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 160 Thermal Cycling: Stress Relaxation Pb-free alloys demonstrate higher creep resistance When will Pb-free be less reliable in the field? Stress Results in greater durability under accelerated testing (fast ramps, short dwells) Exception: Very high temperatures (>125oC), high stress loadings (leadless, ceramic) SAC SnPb Temperature Time © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 161 Thermal Cycling: Effect of Dwell Time 10 Based on creep laws developed by Schubert and damage model developed by Syed SAC life / SnPb life 8 6 2512 Resistor on FR4 25 to 80C (modeling, Blattau, 2005) 4 Ceramic BGA on FR4 0 to 100C (experimental data, Bartello, 2001) 2 0 0 100 200 300 400 500 Dwell Time (min) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 162 Test Life Requirement Test Spec ∆ Temperature Thermal Cycling: When is a Failure Not a Failure? Field Condition SnPb Pb-Free Time to Failure © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 163 Thermal Cycling: Study on SnAgCu Solder Data gathering Time to failure data for Pb-free solder under a variety of standard test conditions Two parameter Weibull model (characteristic life, shape parameter) Focused on: © 2005 2004 - 2010 2007 BGA (area array devices – CSP, PBGA, CBGA) Thin-Shrink Small Outline Package (TSSOP) Leadless chip resistors, 2512 and 1206 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 164 Resistor – Cycles to Failure 1206 Case Size 2512 Case Size Characteristic Life (Cycles to Failure) 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0 50 100 150 200 Change in Temperature (oC) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 165 TSSOP – Cycles to Failure Characteristic Life (Cycles to Failure) 10000 Type 1 Type 2 Type N/A 9000 8000 7000 Type 1 6000 5000 4000 3000 2000 1000 Type 2 0 0 50 100 150 200 Change in Temperature (oC) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 166 BGA Results Normalized Characteristic Life 10 Life as a function of delta T from tests in which two or more thermal profiles were used, includes CSP, PBGA and CBGA data y = 301550x-2.7427 1 0.1 10 100 1000 Change in Temperature (oC) AN ACCELERATION MODEL FOR Sn-Ag-Cu SOLDER JOINT RELIABILITY UNDER VARIOUS THERMAL CYCLE CONDITIONS N. Pan, G. A. Henshall, F. Billaut, S. Dai, M. J. Strum, R. Lewis, E. Benedetto, J. Rayner © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 167 Thermal Cycling: Effect of Dwell Time Normalized time to failure as a function of dwell time at maximum temperature for SAC solder • 40% to 60% drop in the number of cycles to failure as dwell is increased past 8 hours • As the CTE mismatch decreases and the part becomes more compliant, the effect of dwell decreases © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 168 Long-Term Dwell Flip-chip BGA (-25C) to 100C Both SAC and SnPb experienced shorter time to failure when exposed to long-term dwells Rate of degradation was faster for SAC SAC still far superior to SnPb Intel Technology Journal, Materials Technology for Environmentally Green Micro-electronic Packaging, Volume 12, Issue 01, 2-21-2008 Vasu Vasudevan et al., "Slow Cycle Fatigue Creep Performance of Pb-free (LF) Solders," ECTC, 2008. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 169 SnPb vs. SAC: Resistors 10000 SAC (75%) SnPb SAC (50%) Time to 1% Failure (Cycles) 9000 8000 7000 Time to 1% failure for 2512 resistors attached with SAC or SnPb solder and subjected to long dwells (~8 hours) 6000 5000 4000 3000 2000 1000 0 0 20 40 60 80 100 o Change in Temperature ( C) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 170 SnPb vs. SAC: TSOP 10000 SAC (75%) SnPb SAC (50%) Time to 1% Failure (Cycles) 9000 8000 7000 Time to 1% failure for TSOPs attached with SAC or SnPb solder and subjected to long dwells (~8 hours) 6000 5000 4000 3000 2000 1000 0 0 20 40 60 80 100 o Change in Temperature ( C) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 171 SnPb vs. SAC: Area Array Devices 2.5 SAC Life / SnPb Life 2.0 1.5 1.0 0.5 0.0 0 2000 4000 6000 8000 10000 12000 14000 16000 Characteristic Life (Cycles) Ratio of SAC/SnPb reliability for area array devices as a function of characteristic lifetime of the SAC component © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 172 SnPb vs. SAC: Findings 2512 Resistors and TSOPs with Alloy42 leadframes have limited lifetimes 500 cycles of -40 to 125ºC 1500 cycles of 0 to 100ºC Long dwells up to 8 hours would be expected to reduce lifetime between 40 and 60% Once long dwell and differences in shape parameters are taken into account, there is likely to be minimal statistical difference between SAC and SnPb in most operating environments © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 173 2nd Generation Solders Reliability data on 2nd generation Pb-free alloys (non-SAC) are relatively scarce DfR recently completed accelerated life testing on Ni-modified SnCu (SN100C) Three package styles Three test environments © 2005 2004 - 2010 2007 Leadless: 2512 Resistor Leaded: 44 I/O TSOP Ball: 98 I/O CSP Temperature cycling Vibration Mechanical shock 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 174 Results: Thermal Cycling - Resistor Resistor samples exposed to ∆T of 75°C (blue), 100°C (red and green) and 165°C (violet). Dwell time has a minimal effect on time to failure. Time to 1% failure of 2000 cycles matches expectation of SnPb and SAC305 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 175 Results: Thermal Cycling – TSOP, all solders TSOP samples exposed to ∆T of 165°C soldered with SAC (blue), SN100C (green) and SnPb (red). Characteristic life (eta, η) and shape factor (beta, β) are nearly identical. Tails may indicate infant mortalities (defect driven). © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 176 Results: Thermal Cycling – CSP, all solders CSP samples exposed to ∆T of 165°C soldered with SAC (blue), SN100C (green) and SnPb (red). Characteristic life (eta, η) is nearly identical for SAC and SnPb, with SN100C marginally higher. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 177 Expected Field Failure Time (1%)(NEW) n = 2.18 n = 1.75 n = 1.55 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 178 Estimation for 1% Field Failure Rate © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 179 2nd Gen Pb-Free Solders (cont.) Study of certain Pb-free solders (SAC and SNC) strongly suggests they should provide similar or better reliability than SnPb Requires no manufacturing defects Does not include environments with drop/mechanical shock SNC may be an appropriate middle ground between SnPb and SAC © 2005 2004 - 2010 2007 Not as brittle as SAC More creep resistant than SnPb Widely accepted in wave solder and HASL processes 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 180 Flip Chip (1st Level Interconnects) There is limited test data on thermal cycling reliability of Pb-free flip chip solder bumps Scenario 1 Partially driven by existing exemption Underfill properties dominate solder bump performance Changes from high-Pb to Pb-free would be a second-order effect Scenario 2 © 2005 2004 - 2010 2007 Potential for 1 – 3 addition reflows (SMT, rework) Could have detrimental effect through dissolution of plating, thick intermetallics, etc. 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 181 Physics of Failure (Pb-Free Solder – Mechanical Cycling) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 182 Fatigue: Overview Mechanical fatigue failures occur due to low cycle/high amplitude events (shock) and high cycle/low amplitude events (vibration) Limited number of studies completed to assess the performance of Pb-free under high-cycle fatigue Initial understandings of failure behavior based on limited number of publications and results of internal testing Two approaches © 2005 2004 - 2010 2007 Bend cycling Vibration testing 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 183 Mechanical Cycling (Background) Lifetime under mechanical cycling is divided into two parts Low cycle fatigue (LCF) High cycle fatigue (HCF) Fatigue of Structures and Materials, J. Schijve, Springer, 2001 LCF is driven by plastic strain (Coffin-Manson) ε p = ε f (2 N f ) c -0.5 < c < -0.7; 1.4 < -1/c < 2 HCF is driven by elastic strain (Basquin) εe = σf E (2 N ) b f -0.05 < b < -0.12; 8 < -1/b < 20 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 184 Fatigue: Background Most test results plot life as a function of boardlevel strain Equivalent to solder strain only in the elastic regime Bend cycling is ½ to ¼ less severe than vibration © 2005 2004 - 2010 2007 Not fully reversed 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 185 Fatigue: Bend Cycling (KAIST) Test sample 1.27mm pitch 256 I/O BGA (27 x 27 mm) 90-mil (2 mm) printed circuit board Test setup © 2005 2004 - 2010 2007 Span of 100 mm Frequency N/A Loads 25 to 60 N 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 186 Fatigue: Bend Cycling (KAIST) 100 SnPb nSnPb = 2.9 nSAC = 4.2 Applied Load (N) SAC405 Seems to display transition or high-cycle behavior (good fit to MIL-STD-810) 10 1.E+03 1.E+04 1.E+05 1.E+06 Time to Failure (Cycles) I. Kim, ECTC 2007 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 187 Fatigue: Bend Cycling (Motorola) Test sample Test setup 0.8mm pitch 179 I/O BGA (14 x 14 mm) 22-mil (0.56 mm) printed circuit board Span of 44 mm Frequency of 1Hz Displacements 0.2 to 0.5 mm Results SnPb ≈ SAC Board Level Strain (microstrain) SnPb SAC Motorola 1400 1200 1000 800 600 400 200 0 1,000 10,000 100,000 1,000,000 Fatigue Life (cycles) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 188 Fatigue: Bend Cycling (DfR Solutions) Test sample 2512 chip resistors (6.25mm x 3mm) 62-mil (1.5 mm) PCB Strain (µε) SnAgCu SnPb 800 2,929,200 1,870,00 1200 69,399 26,395 2400 10,508 7,732 19,485 Test setup © 2005 2004 - 2010 2007 SnPb Span of 100 mm Frequency 4 Hz Board level strain 800 to 2400 µε SAC305 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 189 Fatigue Predictions Using Coffin-Manson SnPb SAC305 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 190 Vibration (JGPP) Random Vibration 9.8 to 28 Grms (0.07 to 0.5 G2/Hz) Natural Frequency (72 Hz) Results With BGA’s, SnPb solder always outperformed lead-free Less conclusive for leadless/ leaded parts Some MAS companies are using these results to incorrectly conclude SnPb is better than SAC in vibration All BGA-225 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Woodrow, IPC/APEX 2006 191 Fatigue Exponent – Coffin Manson Assume the solder strain is directly proportional to the board level strain Coffin – Manson N field N test Fit to data provides c = 1.5 ⎡ ε test ⎤ =⎢ ⎥ ⎢⎣ ε field ⎥⎦ c Exponent value is too low; representative of low-cycle fatigue High-cycle fatigue exponent is typically 4 to 6 or higher MIL-STD-810, Steinberg © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 192 Fatigue: Intel Vibration Study Sinusoidal Vibration 0.75g to 4.0g input Note change in Pb-free behavior at high loads, compared to SnPb © 2005 2004 - 2010 2007 S.F. Wong, ECTC 2007 Similar behavior observed at DfR 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 193 Fatigue: Intel Vibration Study (Cont.) Normalized Board Strain 1 SAC405 SnPb 0.1 0.01 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07 1.E+08 Normalized Time to Failure (Cycles) Results do not display power-law behavior © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 194 Experimental Procedures – Test Matrix (NEW) One component type : 15 mm x 15 mm CBGA Three preconditioning states No preconditioning Thermal Cycling from -20C to 85C 120 cycles Isothermal 150C for 100 hours Seven loading conditions 208 I/O perimeter array 0.8 mm pitch 12.7 x 12.7 mm die, CSP Four harmonic vibration levels Three shock levels Two alloys © 2005 2004 - 2010 2007 Sn3.0Ag0.5Cu 63Sn37Pb 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 195 Vibration Results – Analysis (NEW) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Large variation in time to failure due to multiple stress states present at each load level 196 Vibration Results – Analysis (NEW) FEA to determine the stress levels 6 Stress levels Identified for each vibe level U1 U20 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 197 Vibration Results – Analysis (NEW) SnPb Vibe Results: Stress/Location Corrected 1.00E+08 Cycles to Failure 1.00E+07 SnPb Fatigue Relationship No Precondition σ = 116 .8 ⋅ N −.116 Thermal Cycling σ = 68.2 ⋅ N −.081 Isothermal σ = 79.0 ⋅ N −.097 1.00E+06 y = 2E+14x -5.9271 y = 5E+14x -6.3281 y = 3E+14x -6.3956 SnPb NPC SnPb TC SnPb ISO 1.00E+05 1.00E+04 10 15 20 25 30 35 Stress (MPa) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 198 Vibration Results – Analysis (NEW) SAC305 Vibe Results: Stress/Location Corrected Fatigue Relationship No −.058 Precondition σ = 52.3 ⋅ N 1.00E+08 y = 6E+11x-4.2163 Thermal Cycling 1.00E+07 Isothermal -3.692 Cycles to Failure SAC305 y = 1E+11x 1.00E+06 y = 4E+11x-4.0694 σ = 78.1 ⋅ N −.09 σ = 71.0 ⋅ N −.082 SAC305 NPC SAC305 TC SAC305 ISO 1.00E+05 1.00E+04 10 15 20 25 30 35 Stress (MPa) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 199 Vibe Results – 2X Safety Factor SnPb (NEW) High Cycle Fatigue S-N Curve for SnPb, 3 Preconditions 35 SnPb NPC SnPb TC SnPb ISO 2X Safety Factor 30 Solder Alloy Stress (MPa) 25 No Safety Factor 15 2X Safety Factor Determined at 108 cycles Fatigue Exponents Precondition None 8.628 Temperature Cycling 12.34 Isothermal Aging 10.31 SnPb fatigue exponent 8.333 [1] © 2005 2004 - 2010 2007 20 6.51 10 Steinberg uses 6.4 8.427 5 1.00E+04 7.42 1.00E+05 1.00E+06 1.00E+07 1.00E+08 1.00E+09 Cycles to Failure 6.34 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 200 Vibe Results – 2X Safety Factor SAC (NEW) High Cycle Fatigue S-N Curve for SAC305, 3 Preconditions 35 SAC305 NPC SAC305 TC SAC305 ISO 30 2X Safety Factor Stress (MPa) 25 Solder Alloy No Safety Factor 15 2X Safety Factor Determined at 108 cycles Fatigue Exponents 10 Precondition None 17.24 Temperature Cycling 11.11 Isothermal Aging 12.195 © 2005 2004 - 2010 2007 20 10.5 5 1.00E+04 7.83 1.00E+05 8.36 1.00E+06 1.00E+07 1.00E+08 1.00E+09 Cycles to Failure 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 201 Vibe Results (NEW) Solder Alloy SnPb SAC305 Precondition Fatigue Relationship −6.4 Cycles to Failure © 2005 2004 - 2010 2007 ⎛σ⎞ Nf =⎜ ⎟ ⎝116⎠ −7.8 ⎛σ ⎞ Nf =⎜ ⎟ ⎝71⎠ 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 202 Fatigue: Interpretation of Pb-Free Data SAC is ‘stiffer’ than SnPb Low-cycle fatigue (plasticity driven) For a given force / load, it will respond with a lower displacement / strain (elastic and plastic) Under displacement-driven mechanical cycling, SnPb will tend to out-perform SAC (e.g., chip scale packages [CSP]) Under load-driven mechanical cycling, SAC will tend to outperform SnPb (e.g., leads of thin scale outline packages [TSOP]) High-cycle fatigue (elasticity driven) © 2005 2004 - 2010 2007 SAC will tend to out-perform SnPb for all package types 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 203 Fatigue: Conclusions (NEW) If you have a problem with SnPb solder and vibration, you will have a problem with SAC solder and vibration If you do not have a problem with SnPb solder and vibration, you will not have a problem with SAC solder and vibration © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 204 Mechanical Shock SAC can be significantly less robust than SnPb Crossover into board failure Not always consistent Very strain-rate dependent Plating is an important driver SnNi vs. SnCu intermetallics Chai, ECTC 2005 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 205 Results: Mechanical Shock Testing Unaged CSPs (blue) versus CSPs exposed to 125°C (green) and 150°C (red) for 10 hours demonstrated that aging reduces drops to failure. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 206 Results: Mechanical Shock Testing Unaged CSPs (blue) versus CSPs exposed to 125°C (green) and 150°C (red) for 100 hours demonstrated that aging reduces drops to failure. Formation of “tails” indicates that infant mortality (due to randomized induced defects) is present. © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 207 Results: Mechanical Shock Testing Unaged CSPs (blue) versus CSPs exposed to 125°C (green) and 150°C (red) for 1000 hours is evidence for a transition from ductile failures (wearout) to brittle (random). © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 208 High Temp Storage: Embrittlement Mechanical shock results after extended aging (1000 hrs/150ºC) of CSPs: SN100C and SAC305 exhibit similar performance after exposure SnPb slightly superior Failures occur predominantly at or near CSP interface Mixed failures typical, with brittle fraction increasing with exposure time © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 209 Experimental Procedures – Shock (NEW) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 210 Shock Results (NEW) Table 8: Summary of Weibull Variables for SAC305 Shock Testing. Solder Precondition Effective Shock Level Characteristic Life (η) (tower level) Shape Factor (β) SAC305 None 694 (1000) 56 1.48 SAC305 None 570 (1500) 59 1.59 SAC305 None 520 (750) 160 1.86 SAC305 150°C Isothermal 694 (1000) 49 1.23 SAC305 150°C Isothermal 570 (1500) 79 2.29 SAC305 150°C Isothermal 520 (750) 111 2.14 SAC305 -40°C - 85°C Cycled 694 (1000) 55 2.33 SAC305 -40°C - 85°C Cycled 570 (1500) 83 3.22 SAC305 -40°C - 85°C Cycled 520 (750) 136 1.83 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 211 Shock Results (NEW) Table 9: Summary of Weibull Variables for SnPb Shock Testing Solder Precondition Effective Shock Level (tower level) Characteristic Life (η) SnPb None 694 (1000) 41 1.60 SnPb None 570 (1500) 78 1.47 SnPb None 520 (750) 154 2.17 SnPb 150°C Isothermal 694 (1000) 25 1.08 SnPb 150°C Isothermal 570 (1500) 44 1.90 SnPb 150°C Isothermal 520 (750) 61 1.48 SnPb -40°C - 85°C Cycled 694 (1000) 56 2.26 SnPb -40°C - 85°C Cycled 570 (1500) 76 2.65 SnPb -40°C - 85°C Cycled 520 (750) 184 1.81 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com Shape Factor (β) 212 Shock Results – Analysis (NEW) Pulse Time Frequency Frequency Ratio Amplification Factor Adjusted G Load 1500 G 0.5 mS 1000 Hz 0.17 0.38 570 1000 G 1.00 mS 500 Hz 0.34 0.694 694 750 G 1.00 mS 500 Hz 0.34 0.694 520 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 213 Shock Results – Analysis (NEW) 750 700 Shock Level (G) SAC305 NPC 650 SnPb NPC SAC305 150C SnPb 150C 600 SAC305 TC SnPb TC 550 500 0 50 100 150 200 Num ber of Drops © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 214 Long-Term Aging Intermetallic Growth Aging before thermal cycling Can influence drop / shock performance Rates in SAC seem faster; SnPb / SN100C may be equivalent Work by HDPUG and UMD seem to suggest minimal difference (-10%) Dwell © 2005 2004 - 2010 2007 Work by Intel demonstrates degradation of both SAC and SnPb 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 215 PoF Example: SnAgCu Life Model Modified Engelmaier Semi-empirical analytical approach Energy based fatigue Determine the strain range (∆γ) LD ∆γ = C ∆α∆T hs C is a correction factor that is a function of dwell time and temperature, LD is diagonal distance, α is CTE, ∆T is temperature cycle, h is solder joint height © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 216 PoF Example – SAC Model (cont.) Determine the shear force applied to the solder joint ⎛ L ⎛ 2 −ν ⎞ ⎞ hs hc L ⎟⎟ ⎟ (α 2 − α1 ) ⋅ ∆T ⋅ L = F ⋅ ⎜⎜ + + + + ⎜⎜ ⎟ ⎝ E1 A1 E 2 A2 As Gs Ac Gc ⎝ 9 ⋅ Gb a ⎠ ⎠ F is shear force, L is length, E is elastic modulus, A is the area, h is thickness, G is shear modulus, and a is edge length of bond pad Subscripts: 1 is component, 2 is board, s is solder joint, c is bond pad, and b is board Takes into consideration foundation stiffness and both shear and axial loads © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 217 PoF Example – SAC Model (cont.) Determine the strain energy dissipated by the solder joint F ∆ W = 0 .5 ⋅ ∆ γ ⋅ As Calculate cycles-to-failure (N50), using energy based fatigue models for SAC developed by Syed – Amkor N f = (0.0019⋅ ∆W ) −1 © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 218 Validation – Chip Resistors Cycles to Failure (Predicted) 10000 1000 100 100 1000 10000 Cycles to Failure (Experimental) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 219 PoF Example – SAC Reliability (cont.) How to ensure 10 year life in a realistic worst-case field environment for industrial controls? American Southwest (Phoenix) Dominated by diurnal cycling o o Month Cycles/Year Ramp Dwell Max. Temp ( C) Min. Temp. ( C) Jan.+Feb.+Dec. March+November April+October 90 60 60 6 hrs 6 hrs 6 hrs 6 hrs 6 hrs 6 hrs 20 25 30 5 10 15 May+September June+July+August 60 90 6 hrs 6 hrs 6 hrs 6 hrs 35 40 20 25 +10C at max temperature due to solar loading © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 220 PoF Example – SAC Reliability (cont.) Total damage in desert environment over 10 years 0.02604 Total damage in one cycle of -40C to 85C test environment 0.00012 Total cycles at -40C to 85C to replicate 10 yrs in desert 222 cycles At 1 cycle/hour, approximately 1 day of test equals 1 year in the field © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 221 Pb-Free DfR (Microstructure / HALT) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 222 Microstructural Evolution: SnPb It is well known that SnPb coarsens when subjected to thermal cycling Morphology provides valuable data on failure behavior Drivers for grain growth Larger grains -> fewer grain boundaries -> lower free energy Thomson-Freundlich solubility relationship The solubility in the matrix in contact with a particle increases as the radius of the particle decreases Solute diffuses in the direction of largest particles and away from smallest particles Small particles dissolve to compensate Thermo-mechanical cycling (TMC) increases grain growth rate (Ostwald ripening) Plastic deformation during TMC results in an excess concentration of vacancies, which increases rate of diffusion Relatively high solid solubilities of Pb in Sn and vice versa lead to microstructural instability (changes in the miscibility of the two phases over the temperature range) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 223 Microstructural Evolution: SnPb 200 hours © 2005 2004 - 2010 2007 1000 hours 1800 hours 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 224 Microstructural Evolution: SAC Reflowed SAC solder joints tend to contain relatively few grains Plastic deformation during TMC and high-angle grain boundaries creates initiation sites for recrystallization Starts at highest stress regions Grains get smaller! Behavior similar to brass After reflow After 1000 cycles Occurs faster and more uniformly in solder attached to copper-based platings Leads to inter-granular fracture © 2005 2004 - 2010 2007 SnPb tends to be intra-granular Mattila, J. Mat. Res. 2004 -40 to 125C 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com After failure (~1500 cycles) 225 HALT/HASS and Pb-Free As discussed in previous sections, SnPb will outperform Pb-free solders in high temperature/high vibration amplitude environments Thus, Pb-free assemblies may perform worse under HALT HASS conditions must be re-evaluated However, Pb-free assemblies will likely have longer lifetimes in actual field environments HALT/HASS continue to offer value in highlighting weak links of a design and/or process HALT/HASS results should not be used for lifetime prediction without substantial studies in less accelerated conditions © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 226 Design for Reliability (Conclusions) © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 227 Transition to Pb-Free: Other Issues Be aware of the ‘greening’ of packaging materials Recent examples of failing to capture new failure behavior when new materials were introduced More than just Pb-free Do not accept ‘form-fit-function’ Red phosphorus flame retardants Low ESR electrolyte Why? © 2005 2004 - 2010 2007 Strong drivers in industry to limit the value of testing and product qualification Hear no evil, see no evil 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 228 New Technology: Supply Chain Failures Failure in experimental design Failure to properly pre-condition No attempt to understand how the new technology could fail and adjust qualification procedures accordingly Testing not performed after reflow Most component testing not performed after exposure to all realistic manufacturing stresses Failure to use appropriate samples © 2005 2004 - 2010 2007 Must be representative of realistic worst-case (materials, manufacturing, use) 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 229 Outliers and Technology Qualification Range in field failure rates (approximately 1.5%) Environmental Stress Failures © 2005 2004 - 2010 2007 Material Strength No Failures! 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 230 Failures of Inadequate Testing Failure to monitor relevant behavior E.g., leakage currents not monitored during temperature / humidity / bias (THB) testing Assumed a particular failure mechanism (Kirkendall voiding, which induces a voltage drop) Functional tests not performed in-situ Failure of testing THB testing not performed Failures during integrity testing not always communicated to customers Interfered with launch date Not aware of new compound Strong impetus to report zero failures Changes in materials not always communicated to the customer © 2005 2004 - 2010 2007 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 231 Summary To avoid design mistakes, be aware that functionality is only the beginning Be aware of industry best practices When to use heuristic rules; when to use physics of failure Maximize knowledge of your design as early in the product development process as possible Do not overly rely on supplier statements © 2005 2004 - 2010 2007 Their view: Reliability is application dependent 5110 Roanoke Place, Suite 101, College Park, Maryland 20740 Phone (301) 474-0607 Fax (240) 757-0053 www.DfRSolutions.com 232