In-situ Determination of Specimen Temperature during Electromigration Testing of Solder Joint

摘要

Accelerated reliability testing under high current density, such as in electromigration (EM) tests of solder joints, gives rise to Joule heating, which can reduce the reliability performance in microelectronics. The determination of the temperature of the solder joint under Joule heating is reported for an SAC305 solder joint in a flipped PCB configuration during an EM test. Knowledge gains can be useful for appropriate temperature selection for EM tests to understand the contribution of failures from Joule heating. We developed a custom PCB circuit allowing for multi-wire access to two specimens which were SAC305 solder joints. Also developed was a method to determine the solder temperature based on the temperature coefficient of resistance (TCR) and thermal resistance ($R_{ext{th}}$) of the solder joints. The average TCR at 20 °C and $R_{ext{th}}$ for solder joints were 0.00441 1/K and 736 K/W, respectively. Subsequently, one of the solder joints was cross-sectioned and its $R_{ext{th}}$ was reduced to 559.1 K/W. During an example EM test in an ambient temperature of ≈60 °C, the local temperature of this solder joint rose from 32 K above ambient to 45 K above ambient. After 135 h the solder resistance had increased by over 40% to 0.81 milliohm, but did not show any signs of complete failure yet. However, the joint was weakened which only became evident when it completely broke due to thermomechanical stresses invoked by test end and cooling down. During the test the current was 10 A, and the current density estimated to be 22.9 kA/cm^{2. Our approach to estimate the solder joint temperature was dependent on the joint resistance at any given time. This allowed for capturing temperature changes during the EM test, unlike other techniques which are based on initial solder joint resistance that only provide a constant estimated temperature value and fail to capture temperature changes during the test. Our results demonstrated a steady increase in the solder joint temperature as the resistance of the joint increased during current stressing as a result of void nucleation and growth. Our approach can be extended to solder joints of varying geometries and materials, since the approach accounts for sample-to-sample variations using the thermal resistance term, which is unique for each joint.}