Microvoid formation and Kirkendall effect in lead-free solder joints

The reliability of solder joints is dependent on the stability of intermetallic layers formed between copper and bulk solder. Microvoids observed to sometimes form at the copper-solder interface or in the intermetallic phase during annealing are also found to affect the reliability of these solder joints. In order to understand the reasons and mechanisms for the formation of these microvoids, two different studies were conducted. Diffusion studies were performed to determine the intrinsic diffusivities of Cu and Sn in the Cu3Sn and Cu6Sn5 phases at 200°C to examine the contribution of the Kirkendall effect to microvoid formation. In addition, the role of electrodeposition conditions and impurity incorporation on microvoid formation was examined using a wide range of processing conditions. Solid-solid diffusion couples assembled with disks of copper, tin and intermetallics (Cu3Sn and Cu6Sn5) were employed to investigate the Kirkendall effect in the copper-tin system at the temperature of 200°C. In the Cu(99.9%)/Sn diffusion couple, inert alumina particles used as markers were identified in the Cu6Sn5 phase, while microvoids were observed at the Cu/Cu3Sn interface. The Cu(99.9%)/Sn and Cu(99.9%)/Cu6Sn5 diffusion couples annealed at 200°C for 10 days were analyzed for intrinsic coefficients of Cu and Sn in the Cu 6Sn5 and Cu3Sn phases, respectively, with due consideration of changes in molar volume. Interdiffusion, integrated and effective interdiffusion coefficients were also calculated for the intermetallic phases. Diffusion couples annealed at 125°C–400°C for various times were analyzed for the kinetic parameters such as growth rate constants and activation energies for the formation of Cu3Sn and Cu6Sn 5 phases. Uncertainties in the calculated intrinsic diffusivities of Cu and Sn arise mainly from the non-planar morphologies of the interfaces and the non-planar distribution of the markers. Intrinsic diffusion coefficients based on average locations of the marker plane indicate that Cu is a faster diffusing component than Sn in both the Cu3Sn and Cu6Sn 5 phases. The electrodeposition conditions that may contribute to the microvoid formation during isothermal aging of solder-electrodeposited copper were evaluated in a systematic series of experiments. A screening design of experiment was used to identify the important plating parameters affecting the microvoid formation. The Hull cell was used to plate a large number of samples in short time. Brighteners, suppressor and bath age were identified as significant plating variables affecting the formation of microvoids. Then, 2K full-factorial design of experiment was used to study the effect of these pre-screened electroplating parameters on the propensity of microvoiding at solder-copper interface after annealing at 125°C for 40 days. All the pre-screened parameters and their two-way interactions were found to have significant effect on the microvoiding. Time of Flight-Secondary Ion Mass Spectroscopy and Glow Discharge Spectroscopy allowed us to establish a correlation between impurities incorporated in copper during electroplating and microvoiding at the solder-copper interface. This study showed that the microvoids observed in the solder joints at copper-Cu3Sn interface or in the Cu3Sn phase are not the Kirkendall voids and the main reason of formation of these microvoids is the presence of impurities in the electroplated copper substrate