Thermal fatigue life prediction: consequences of cycle reduction and material property variation

RF power amplifier demonstrators containing each one GaN-on-SiC, HEMT, CHZ015A-QEG, from UMS in SMD quad-flat no-leads package (QFN) were subjected to thermal cycles (TC) and power cycles (PC) followed by electrical, thermal and structural evaluation. Two types of solders i.e. Sn63Pb36Ag2 and lead-free SnAgCu (SAC305) and two types of TIM materials (NanoTIM and TgonTM 805) for PCB attachment to liquid cold plate were tested for thermo-mechanical reliability. Changes in electrical performance of the devices namely reduction of the current saturation value, threshold voltage shift, increase of the leakage current and degradation of the HF performance were observed as a result of an accumulated current stress during PC. No significant changes in the investigated solder or TIM materials were observed.

The paper investigates the effect of variations both in temperature cycling profile and in SAC305 solder Young's modulus in the PBGA256 package on the thermo-mechanical reliability. FE simulations quantify the effect of cycle reduction and counting techniques by introducing different temperature profiles having identical dwell- and period time characteristics. A difference of 30% in predicted accumulated creep strain energy density per cycle has been determined for the studied profiles. Under the provided modelling assumptions and simplifications, the maximum variation of the thermal fatigue life of SAC305 solder joints is within 30% as the result of experimentally determined Young's modulus variation in as-delivered packages.

Surface insulation resistance (SIR) measurements have become necessary to perform, especially due to the introduction of lead free solders. It has been reported that SIR and other test methods have to be used for evaluating the reliability of printed wiring boards. This paper presents an investigation on how the SIR test environment influences the test results. The factors varied were temperature and humidity. The temperature was varied between 40 and 85 degrees C, and the relative humidity was varied between 60 and 85%. Furthermore, the influence of different types of process chemicals on SIR was evaluated. Seven lead free solder pastes, of which five were no-clean and two were water soluble, were compared. The influence of using conformal coating was also studied. These solder pastes were used to verify the developed method. A proper test method for SIR measurements is suggested in the paper, as well as a recommendation for use of conformal coating.