I'm in the high reliability (exempt) sector and have thermal cycled hundreds of boards in the past few years. As far as which is more reliable, SnPb or SAC305, my data is similar to what the technical literature is saying, "It Depends". In low strain situations such as small components, QFPs, SOICs, large pitch BGAs, etc, SAC305 has similar performance to SnPb. In high strain situations such as large ceramic discretes, certain QFNs, etc, SAC305 is dramatically less reliable than SnPb. Something I've noticed is that even in parts that have equivalent fatigue life between SnPb and SAC305, the SAC305 joints have earlier crack initiation. Regardless of when the joint fails electrically, our customers don't like seeing cracks in cross-sections taken from units pulled from validation at 1/3 the expected life. The big question that hasn't been adequately addressed is acceleration factor. Some data is showing that SAC305 has a higher acceleration factor so, in theory, it can fail a thermal cycle test earlier than SnPb but still have a longer field life.
We can design around a lot of the fatigue issues. The thing that worries me more is Sn whiskers. We (the high reliability sector) don't have enough pull to force component makers keep supplying non-Sn finishes or use expensive mitigation such as nickel barriers.
reply »