| The long term life of a high reliability microelectronics product is commonly limited by fatigue of one of the interconnects. For the quantitative assessment of this the industry commonly relies on accelerated tests in which assemblies are cycled with a fixed amplitude until failure. Most realistic applications, on the other hand, obviously involve varying amplitudes and often more than one type of loading. Vibration testing is special in that frequencies are usually varied, leading also to variations in amplitude, but these variations are not relevant to most service conditions either.;Predictions of life under realistic service conditions usually rely on the implicit assumption of the Palmgren-Miner rule of linear damage accumulation, but we show this to break down in a systematic fashion when it comes to lead free solder joint fatigue. Extrapolations of the observed trends suggest that life in long term service may be much shorter than predicted based on Miner's rule. Cycling of individual BGA joints in a micromechanical tester allows us to monitor the solder stiffness and the inelastic energy deposition, offering an explanation for this and suggestions as to what would be a `worst case' scenario.;At high strain rates a few cycles with a higher amplitude were indeed seen to reduce the effective joint stiffness. Part of this effect was permanent, leading to faster inelastic energy deposition in subsequent mild cycles as well. However, the effect was strongly reduced or absent for lower strain rates, suggesting that the break-down of Miner's rule may no longer be a concern there. |
