Issue link: https://iconnect007.uberflip.com/i/960594
50 SMT007 MAGAZINE I APRIL 2018 regressed against the DOE factors to assess impact on solder joint lifetimes. Results are summarized in Tables 6 and 7 for number of cycles and percentage, respectively. Factors not statistically significant at the alpha=0.05 level are designated with an "*". Each of the three metrics tells the same story. Examining Eta, the low CTE material added nearly 1000 cycles, or 22%, to the solder joint lifetime. The hybrids added about 1200-1300 cycles, or 30%. In combination, the lifetime was improved over 2200 cycles, ~50%. This statistical analysis confirms the impression from inspection of the Weibull plots. Cross-Section after Cycling After 3,000 cycles, one unmonitored unit per cell was removed from the chamber for crack growth characterization by cross- section. Each was sequentially cut, polished and imaged by SEM at BGA rows B, F and K. Row B was selected for cross-section analysis since it experienced more significant crack - ing than row A as discovered during previ- ous investigations [8] . Row K was along the die edge. From Figure 1, note rows B and F were all NSMD for both hybrid designs. Row K was all NSMD for Hybrid-B, but contained both types for Hybrid-A. The degree of cracking was calculated for each solder joint as described above. Figure 12 shows a color-coded mapping of the BGA arrays. Red represents joints with over 90% crack length, yellow for those with cracks between 50% and 90%, and green for those Figure 11: Weibull plot by experimental cell. Table 5: Summary of Weibull results. Table 6: Change in solder joint lifetime (# of cycles). Table 7: Change in solder joint lifetime (%). * Not statistically significant * Not statistically significant