New modeling technique for combined hygroscopic and thermal stress analysis

1. New modeling technique for combined hygroscopic and thermal stress analysis Delft Institute of Microsystems and Nanoelectronics Sau Koh1, Willem VD Driel2, C.I.M Beenakker1, C,A. Yuan3 G.Q. Zhang2 1Beijing Research Center,, Beijing, 2Philips Lighting, Eindhoven, 3Institute of Semiconductor,Beijing Abstract In this research, numerical analysis capable of modeling nonlinear thermal and hygroscopic effect has been developed. Using this methods, this paper will compare the failure of light emitting diode due to thermal mechanical with or without the additional hygroscopic stresses This is also important since accelerated testing using increase temperature may not be sufficient due to the very narrow temperature range that it can be used to perform a valid test. Hence, by understand hygroscopic stresses induced failure in solder layer; it may enable another parameter to increases the accelerated factors during these tests. • State of Art • Although the failure of SSL system is due to both thermal and moisture failure, • there is only limited studies that investigates the effect of the degradation of SSL’s life due to the hygroscopic stress degradation Validation Constant ΔT and no moisture absorption Uniform temperature and a moisture distribution w(x,y) Uniform temperature cycle with saturated moisture which could be reduces to • Since most of the commercial FE software does not have the inherent moisture diffusion modeling capability, thermal-moisture analogy technique developed by Wong et al is normally used • this method required that either thermal and moisture stress to be model individually. • limited to linear elastic analysis where linear superposition of hygroscopic and thermal stress is possible Objective • Developed an numerical analysis capable of modeling nonlinear thermal and hygroscopic effect Theory When the packages is subjected to both temperature change and moisture absorption, the total strain can be represented Model used Result Computed 1st principle strain distribution with only thermal loading to 85C • εmechis the mechanical strain stress • εthermal is the thermal strain which is a product of the coefficient of thermal expansion, α, and the change in temperature ΔT • εhygro is the hygroscopic strain which is a product of coefficient of moisture expansion β and the change in concentration ΔC Computed 1st principle strain distribution with only thermal loading to 85C and with 168 hrs 85C/85%RH Using transformation of variables Since the thermal and hygroscopic strain are independent on each other, equivalent strain εeqv can be introduced as • Max 1st principle strain of the dome increases from 40MPa to 62MPa • It explianed why cracking of silicone is observed when humidity test is performed before thermal cycling test and not the other way around Methods Approximated the ramping up of the temperature during each time step as a step function Divided the loading into smaller time step Updated αeqv and performed combine thermo-hygroscopic analysis into FEA Perform structure analysis to obtain the induced stress for this time step Repeat step 3 to 4 until end of time step Conclusion A new numerical analysis capable of modeling nonlinear thermal and hygroscopic effect has been developed The order at which humidity test and thermal cycling test is important is explained