Transient Thermal Analysis as In-Situ Test Method for Reliability of High Power LEDs during Temperature Cycle Tests

Light emitting diodes (LEDs) are today a standard and mature method to produce light. Moreover, due to the advantages of LEDs compared to standard light bulbs, e.g. the high energy efficiency, small weight and size, they will become dominant and replace standard light bulbs in many applications. In addition, the LEDs can have a very long lifetime and therefore they have an advantage for all application where the exchange of the light source is very costly as for outdoor and building lighting. Also for automotive applications an important design advantage is achieved by LED light sources with “car lifetime” because they don’t need to be accessible for replacement any more. However, poor LED quality and wrong system design undermines the advantage and has put the reliability of the LED system on the agenda. Thermo mechanical stress due to temperature cycle causes failure of electronic systems. The electronic component itself or the interconnect device, e.g. printed circuit board (PCB) might fail. However, in many cases the interconnect between package and the board, e.g. solder interconnect, is the weakest link in the system. Cracking of the interconnect causes an open contact and the system fails. In the paper we compare the existing methods to investigate the failure of LED interconnects during temperature cycle tests, e.g. simple “light on test”, electric resistance measurement and shear test. We introduce the transient thermal analysis as measurement method for solder joint reliability and show that the resolution is significantly higher compared to the existing standard methods. Typically, the transient thermal analysis is used to measure the thermal resistance of a package. It is very often still seen as complicated and time intensive measurement method. However, we show that for package and interconnect integrity analysis during temperature cycle test the measurement effort is small, actually it’s almost as simple as an electric resistance measurement. The voltage to drive a defined current is measured time resolved for a short time interval. A current source with response time down to 1µs and a data acquisition system with sampling time of 1MHz is required. We explain the data evaluation, i.e. how to process the measurement data at the cycle number n and how to compare them with the zero cycle measurement data to obtain the information of the package integrity. We derive from theory, i.e. from the behavior of the transfer function of the thermal system, that the linear factors like power step and k-factor are not required and do not be measured to obtain accurate data. In the paper we present reliability data, i.e. thermal air to air thermal shock test 40°C/125°C, for ceramic high power LED packages on Al-IMS analyzed by electric resistance measurement, thermal analysis and cross-sections. We demonstrate the sensitivity of the thermal analysis to detect solder joint cracking and thermal degradation of the system.