A numerical and experimental investigation of influential factors for solder joint reliability of power LEDs for automotive applications

A major issue that still concerns solid-state lighting, especially in power LEDs for automotive applications, is thermo-mechanical reliability. Thanks to advancements in the opto-semiconductor industry, LEDs have reached a degree of reliability in Level 0, LED die, and Level 1, packaged die [1], such that the LED device is able to survive standard cycling tests on the FR4 board. However, regarding Level 2 reliability on boards used in application which relies on the solder, there is still room for improvement. The problem affecting this Level is due to the difficulties involved in finding a balance between thermal management requirements and cost. For better thermal management (Level 0 and 1), ceramic is mostly preferred as the heatsink of the LED and, for more effective heat dissipation on Insulated Metal Substrates (IMS), namely Al-IMS, is used as a cost-effective option. However, this combination brings about solder reliability issues in 2
^nd
board reliability (Level 2) such as degradation and cracks induced by the mismatch in the Coefficient of Thermal Expansion (CTE) and the large operational temperature range T = (-40 to 125°C) that is required for automotive applications. Therefore, a reliable lifetime model for solder in such setups is required. In this study, by defining a comprehensive test matrix, which investigates influential factors such as substrate (Steel, Copper, Molybdenum), solder thickness (55 μm till 95 μm) and cycling temperature profile (different dwell time and ∆T), we were able to verify the simulation approach to the lifetime model. The test matrix was realized using dummy LEDs consisting of only ceramic which were subjected to temperature cycling. Samples were examined before and during temperature cycling (every 250 cycles) using Scanning Acoustic Microscopy (SAM) to track crack initiation and growth. Experimental results show good agreement with simulation results, indicating molybdenum with 55 μm solder thickness to be the best performing setup with no damage visible in any of the samples and steel with 35 μm solder thickness as the worst performing setup at the current 750 cycles where 95% of the devices have met our damage criterion in their electrical pad.