February_EDFA_Digital
edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 20 NO. 1 26 Shear strengthmeasurements were conducted for the copper pillars themselves (Fig. 4b), in an effort to examine the relative strengthof the copper pillar and the aluminum pad/pillar bond. The results revealed the strength of the copper pillar/aluminum pad bond, because all failures were observed to be in the copper pillars (Fig. 10b). Values for the shear strength of copper pillars themselves were all similar in the as-fabricated state; shear strengths of 130 MPa were observed (Fig. 10a). Accelerated thermal cycling between − 40 and 125 °C was conducted to characterize the interconnect reliability of the copper pillar non-underfilled assemblies. Comparisons were made between the performance of silicon/high-CTE glass interposer assemblies and silicon/ silicon interposer assemblies. These copper pillar assem- blies had 50-µm-diameter copper pillars and Sn1.8Ag solder. Failure of the copper pillar solder joints was observed to occur via crack propagation through a relatively thin solder layer remaining between intermetallic compounds after assembly and ATC testing (Fig. 11). Such a failurewas similar to those observed in more conventional surface- mount geometries, such as chip-scale packages. [24] Similar results were found for both corner and chained end joints (Fig. 12). These ATC results are presented in Fig. 12 in the form of two-parameter Weibull plots of cumulative percentage of failures (percent) versus number of cycles before failure. Fits to these datasets provided estimates of the shape parameter (slope), β and characteristic lifetime (scale), and η for each dataset. Lifetimes weremore than an order ofmagnitude larger for the silicon/silicon assemblies than for the silicon/high-CTE interposer assemblies (Fig. 12). CONCLUSIONS In the current study, relations between processing, as-received samples was generally significantly higher than that for aged samples, with Sn1.8Ag samples exhib- iting values above 50 MPa compared to approximately 38 MPa for Sn0.9Ag samples. Lower values of shear strength for Sn2.4Ag samplesmay reflect the possibility of formation of primary Ag 3 Sn precipitates in these samples, thus decreasing the silver available for the fine secondary Ag 3 Sn precipitates (Fig. 5, 6). However, counts of numbers of secondary Ag 3 Sn precipitates in these samples were similar to those for Sn1.8Ag solder joints (Fig. 7). Values of the shear strength of as-reflowed, 100-µm-diameter copper pillar solder joints were distinctly lower than for the 30-µm-diameter copper pillar solder joints (Fig. 9); values were close to those of the aged solder joints. This was despite a similar spacing between Ag 3 Sn precipitates in both as-received 100 µm and 30 µm copper pillar SnAg solder joints. Fig. 10 (a) Shear strength of pillars after reflow. (b) Optical micrographs showing failure of the pillar after shearing Fig. 11 Backscattered SEM images of a failed sample after ATC testingwith cycles between − 40 and 125 °C. Failurewas observed in the remaining solder. The solder compositionwas Sn1.8Ag, and the pad diameter was 50 µmwith a silicon interposer. (b) (a)
Made with FlippingBook
RkJQdWJsaXNoZXIy MjA4MTAy