By Deng Chen
The life estimations of solder interconnects are made using fatigue life models that are derived from standard temperature cycle tests. These temperature cycle tests usually have specific temperature cycle ranges, mean temperature, temperature transition rates, and hold times at the top and bottom temperatures of the temperature cycle. To create an effective model, results from tests that include various temperature ranges, mean temperatures, dwell times, and transition rates are needed. In general, larger temperature ranges, higher mean temperatures, and longer dwell times result in the shortest fatigue life times. While transition times play a role, they are usually less important with the exception of temperature rates that result in a change of failure mode from solder fatigue to intermetallic interface or copper trace failures. Once an effective model has been established, engineers can use the model to predict test outcomes. However, predicting field outcomes is more difficult.
Under field use conditions, temperature excursions are often variable with changing temperature ranges, mean temperatures, hold times, and transition rates. The variations found in the field applications requires an approach for life predictions based on temperature history. To make a field life prediction, one could review the anticipated field temperature history and use the most severe temperature cycles, such as the largest temperature range and longest dwell time. Alternatively, one could use a cycle extracting algorithm, such as a rainflow cycle count method, to extract individual cycles. Using the extracted temperature cycles, the time to failure for each extracted temperature cycle is predicted and the damage for each cycle based on the predicted time to failure is made. Further, current damage state can be predicted by summing the damage of each individual cycle. The expected future temperature cycling is typically assumed to follow the past temperature history, which can be used for remaining useful life prediction.
This presentation will review solder interconnect fatigue life model development, temperature cycle extraction from temperature time history, and damage accumulation methods for fatigue life prediction. In particular, non-linear damage accumulation methods will be explored.