A customer conducted accelerated fatigue tests of a circuit board assemble and wanted to analyze the failed components and decide whether the damage created was of the same mode (cyclic thermal fatigue) as field failures that they had already seen.

The EMPF noted some difference between the lead geometry and bend radius of the field components was apparent (Figure 1). An appreciable amount of conformal coating, however, was noted fewer than one of the components in Group “T”.

Wetting of the solder up the leads was present for components from Group “T” and indicated solder profile issues. A majority of the solder had been pulled up the lead from the base of the solder joint. The lack of solder under the foot has contributed to the lack of solder joint strength and hence the failure mechanism. IPC-A-610D 8.2.5.5 only states that the target condition for maximum heel fillet height is that solder extends above the lead thickness but does not fill the upper lead bend, and also that the solder does not contact the component body. In this case, solder was present beyond the upper lead bend and had also contacted the component. This condition is not an IPC defect by definition, but can reduce reliability.

Cross sections of the Group “T” solder joint showed that a solder fracture was present. It was deformed and ran through the Pb-rich matrix Figure 2. These are characteristics of a cyclic fracture. Figure 3 shows backscatter (left) and secondary electron (right) image show that the fracture was through the bulk solder and near the intermetallic of the component. The fracture appeared to have initiated at the toe of the solder joint and propagated under the lead to the heel of the solder joint. At the component interface, intermetallic measurements indicated some growth (from typical initial solder thickness of 1-1.2um) to 3.5 um. This indicated a high temperature application.

Observations of the microstructure show that there was severe microvoiding present in the Pb-rich regions of the solder. There was little, if any grain growth and no grain elongation. The severe micro voiding had enabled fracture initiation locations. The response of solder to cyclic loading under thermo mechanically driven environments is complex. However, above 100°C, solder decreases in strength rapidly. The combination of application temperature of 130°C and the inherent low tensile strengths has had a severe influence on the solder microstructure and joint reliability. ACI offered the following materials engineering improvements ranking in order of ease and importance.

Modification of PCB board material to higher Tg. Thermount, because of its low CTE, is expected to minimize the effects of CTE mismatch for components that are silicon (bare die) or have a large fraction of their volume (die/packaging volume ratio) as silicon. Polyimide is expected to minimize temperature effects in general because of its high Tg (glass transition temperature). Also below are relative cost increases from a standard grade FR-4.

Substitution of PbSn eutectic. Replacement of solder that considers the application high temperature environment such as 95Sn 5Pb. Board components will need to be examined to determine if they can withstand the higher soldering temperatures.

From the Help Line customers fatigue testing, Figure 2 is an SEM showing that one of them is cracked throughout the entire lead. The fatigue test is proceeded rapidly, and at 1.5 mil peak-to-peak displacements, a sample failed at 680 cycles. The crack appears to have initiated at the toe and shows some distortion, Figure 3. There is also some compression action within the crack along the intermetallic. The compression action and distortion of the crack appear that it is beginning to conform and model the expected behavior. The bulk solder of the broken (left) and opposite (right) joint at high magnification, Figure 4., show micro porosity and a soft distorted solder. Even though there has been stress on the solder joint, there is no alignment of the grain structure. Even though there has been temperature exposure, there is no grain enlargement.

Comparing the microstructures of the original failed samples of Group “T” with the fatigue samples, it was observed that the latter was beginning to approximate the damage that had occurred with the failed samples. In this way, test fixture setup and displacement profiles will be determined that allow the customer to test assembly designs and increase lifetime and reliability.