Solder Resist Crack Resistance Process Characterization in BGA Package for Automotive Grade Reliability

The numbers of electronic devices are increasing year on year basis in our cars. New additions such as Autonomous Self Driving, ADAS and e-mobility: electrification of cars will include charging and thereby extend further in future our expected requirements for IC’s. These factors have changed the overall requirement demanding more stringent conditions and higher density packages. The requirement for these semiconductor devices are also increasing. Automotive Industry Standards AEC- Q100 requires Grade 0, which indicates that these devices to be exposed to higher temperatures such as $175^{\circ}C$, higher than the regular $150^{\circ}C$ and for a longer duration. One of the packages that are used under the hood which is submitted to this severe environment, is the Ball Grid Array (BGA) package. The standard BGA package when submitted to these stringent reliability requirements due to the automotive Grade 0 starts to exhibit defects. The major concern seen are the cracks that are propagating through the solder resist level and in some cases propagating even further reaching to the copper traces and causing an open failure. These cracks in the solder resist are evident after the parts were stressed with extended duration at temperature cycling (TC) and also seen during power temperature cycle (PTC). In regards to PTC, more complex considerations are required to ensure proper stress is applied with respect to the device power activation. The cross section of the failed unit showed that the die attach material had resin rich area at the edges. Further failure analysis was carried out on the reject samples and it was found that the crack signature is matching to the peripheral area of the die edge location. Simulation was performed to identify the stress gradient within the region and the results showed that the die attach fillet edge has the highest stress point. Various designs of experiments were carried out to determine or rather establish a process window with the existing bill of material. The initial experiments were conducted by optimizing the fillet height, bond line thickness, epoxy coverage and also the optimization of the epoxy cure profile. Following that the experiments included different candidates of die attach epoxy. These candidates were selected based on their suitable material properties e.g. Tg and CTE. Phase 2 included the different candidates of Solder resist material. In the next step, new solder resist candidates were also evaluated to determine the overall bill of material robustness in regards to the stress and its performance on the cracks. The results are summarized in detail in this paper. In summary the combination of the given bill of materials are limited to existing boundary conditions and further stressing the units at extended stress or accelerated stress levels would certainly push the BGA package to its limits. Further measures must address the concept and design levels to eradicate these defects.