Robustness Validation / Mission Profile Compared to AEC-Q100 Standard Qualification Flow Jürgen Gruber Zwolle Nördlingen Stuttgart Dresden Bath
Outline History of AEC-Q100 qualification procedure Why robustness validation Basis for your decision Limits RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 2
RoodMicrotec to be the Leading Independent European Company for Semiconductor Supply and Quality Services. RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 3
Services overview extended Supply Chain Management Test & Quality Engineering Production Test and Programming Qualification and Monitoring Burn-In Failure & Technology Analysis Optoelectronics Evaluation & Consultancy Automotive Competence Center RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 4
The begin of AEC-Q100 (end of 1970 s) Quality & reliability of ICs became a real concern IC-failure rate at incoming inspection was 0.5%, defined by AQL (Acceptance Quality Level) Burn-in needed at a failure rate of 2-5% Electronics in automotive became a tough challenge However these rates were an improvement compared to mechanical units RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 5
The begin of AEC-Q100 Lack of own experience in automotive industry MIL-standards became first basis of AEC-Q100 (SAE) MIL-STD 883 for test methods MIL-STD 38510 for test methods and procedures MIL-STD 105 for sampling plans MIL-STD S-19500 lot tolerance percent defective RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 6
AEC-Q100 flow RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 7
AEC-Q100 flow RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 8
AEC-Q100 flow Test conditions were in correlation to the 1970 s technology of integrated circuits The acceptance criteria of the qualification tests covered the accepted failure rates of (one) percent This allowed the use of a limited number of test samples (3 x 77) to get a statistical lot acceptance criteria No fails in 231 parts is equivalent to a maximum of 1% failures at a 90% confidence level far away from today s requirement of ppm failure rates RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 9
What happened meanwhile? The introduction of electronics, replacing mechanical parts continued with unexpected speed New functions became reality using new semiconductor technologies Safety functions (e.g. airbag, ABS, ESP) are standard ISO 26262 is now describing requirements for extremely safe relevant systems RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 10
ADAS trend RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 11
Transistors per die 3µm 2µm 1µm 0.8 µm 0.35 µm 0.25 µm 170 nm 110 nm 75 nm IC development 100G 10G 1G 100M 10M 1M 100k 10k 1k 1k 4k 16k 64k 8080 8085 4004 8008 8086 256k 286 1M 386 4M 486 Pentium Navigation Analog/Power 1970 1975 1980 1985 1990 1995 2000 2005 2010 16M 64M P2 256M P3 1G 2G Pentium 4 Itanium 4G Source: Bosch, IC Insights, Infineon RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 12
IC development Mid 1980s 1,0 µm, 10 MHz Early 2000s 0,1 µm, 3 GHz Today 22 nm FinFET Source: Sematech, Ted Dellin IRPS 2004 RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 13
What happened between 2004 and 2014 Increasing demand for low-cost but also high performance ICs for the consumer market pushed designs and technologies Consumer market is technology driver Automotive a fast follower Smartphone sales in 2014 was 1.5 billion Euro Automotive and other hi-rel applications market share is less than 10% RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 14
What are the consequences New technologies with dramatically reduced structures below 20 nm, together with the need for minimized packages are reducing the reliability margins from earlier "robust" designs The consumer market with short product life cycles accepts shorter lifetime with higher field failure rates of components than needed for automotive or other hi-rel applications RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 15
The bathtub curve RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 16
What are the consequences The historical AEC-Q100 qualification flow will not cover the tremendous changes in technologies, hi-rel applications and ppm requirements We need a method for a risk assessment to determine the safety margin between the remaining capability of a device to the cumulated / combined stress in the application (mission profile) RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 17
AEC Q-100 qualification conflict Appendix 7: Guidance on Relationship of Robustness Validation to AEC-Q100H / A7.1 Scope The present updated version of AEC Q100 (published in 2014) describes literally, also due to legal reasons, this conflict to the ppm-needs: "No fails in 231 parts (77 parts from 3 lots) are applied as pass criteria for the major environmental stress tests. This represents an LTPD (Lot Tolerance Percent Defective) = 1 meaning a maximum of 1% failures at 90% confidence level. This sample size is NOT sufficient or intended for process control or ppm evaluation. Much larger sample sizes would be needed but are not possible in terms of costs and time to market." 1% 231 (3 x 77) 100 ppm 23,027 1%o 2,304 10 ppm 230,260 RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 18
What are the consequences We need a new qualification strategy which demonstrates robustness of products against stress conditions during the application for a valid risk assessment and not only Passing a standard with unspecific tests and a 1%- acceptance criteria A standard has no correlation to the different applications with the different specific mission profiles RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 19
What are the consequences So AEC means Fit for the %-standard and not Fit for the application RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 20
Required information Risk assessments need more information than AEC can provide With no failure you don t know how good the part really is What is the real stress / level / type to failure What is the safety margin between the specification limit and the limit when failures will occur What is the failure physic RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 21
Required information Passing AEC-Q100 flow does not deliver needed information as Limits of design Limits of technology Limits of process capability Potential physics of failures relevant to the application in field and how far these limits are away from the application Interaction of different stresses RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 22
QFN Solder Joints QFN QFN QFP QFN in the centre of a pcb failed after 2 years in field with electrical short circuit. Similar component at the edge in same package showed no problems RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 23
QFN Solder Joints Metallographic cross sectioning QFN in the centre of the pcb, combination of vibration and temperature, destroyed solder joints, cold deformation of solder material shortened two connections RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 24
Robustness validation But relevant information and data regarding the against the later capability of a device stress conditions = mission profile are needed as basis for a risk assessment: robustness validation RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 25
Parameter B Failure mode a Robustness validation Robustness curve? Robustness margin? App I App II? Spec? Parameter A Source: Helmut Keller, ZVEI RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 26
Robustness validation Based on four key elements: Knowledge of the condition of use mission profile Knowledge of the device s capability and limits Knowledge of the relevant failure mechanisms and possible interactions between different failure mechanisms Knowledge of acceleration models for the failure mechanisms (to define accelerated tests) RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 27
Remote car key Intermittent open reported X-ray showed cracks in stitch bond area Source: Peter Jacob, ISTFA 2010 RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 28
Remote car key force PCB QFN Pushbutton QFN placed at narrowest point of the PCB Pushbutton adjacent to QFN Poor temperature profile during soldering process RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 29
Remote car key Delamination at interface leadframe / mold compound Shear off of stitch bond during bending of the PCB Small overlap mold / pin RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 30
Summary Statistical sampling plans with zero failure criteria will not cover ppm needs Testing against device specification limits is not sufficient Only the knowledge of the capability of a device compared to the mission profile allows a risk assessment and robustness validation New test strategies must be implemented (e.g. test till end of life) RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 31
Summary Bad news For today s ppm failure rate demands, AEC-Q100 test flow is no longer a sufficient qualification standard The easy time pass the standard and feel good is over Good news The global industry, under the roof of SAE and ZVEI published robustness validation guidelines and handbooks for your free download RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 32
Summary www.zvei.org/robustnessvalidation Systems 2010 Lifetime Measurement 2013 Implementation 2009 E/E-Modules 2008 revised 2013 Semiconductor 2007 revised 2012 MEMS 2008 RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 33
Thank you for your attention Jürgen Gruber Manager Failure and Technology Analysis Email: juergen.gruber@roodmicrotec.com RoodMicrotec - Jürgen Gruber - AESIN Conference - October 2015 www.roodmicrotec.com 34