May_EDFA_Digital

edfas.org 1 1 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 22 NO. 2 images of the C4 bumps closest to the die edge and the die-C4 interface, respectively. As in the case of the SEM images in Fig. 6b, c, d, and e, no obvious charging or image distortion is observed, even in the right die corner that is further away from the conductive epoxy section. Rather, obvious locations with delamination are observed in the BEOL that may have been difficult to discern if the sample surface had been charging. The delamination is localized and does not propagate through the length of the cross- section, suggesting that the defect was already present in the sample and not induced due tomechanical polishing. The polishing finish of the samples and the image quality appear to be comparable to Fig. 6b, c, d, and e, suggesting that embedding the sample in mounting wax does not have any adverse effects on the polishing process or the SEM imaging. COMPARISON ACROSS ALL MODIFIED PUCK DESIGNS A final check was conducted to verify changes in polishing time, if any, due to the design iterations of adding the conduc- tive epoxy in Config2 or the slotted puck with mounting wax and conductive epoxy in Config3. For this, the pucks for Config0, Config1, Config2, and Config3 were indi- vidually polished from the center of one row of C4 bumps to the center of the next row. Because the samples were identical, the spacing between consecutive rows of C4 bumps was also the same. Figure 8 shows a comparison of the time taken to reach from the center of one row of C4 bumps to the center of the next row for all four pucks. As expected, the polishing time for Config1 was much lower than that of Config0, with a reduction of 85%. The polishing times for Config2 and Config3 are comparable to Config1 and showapolishing time reduction of 89% and 85% as compared to Config0. Figure 8 proves that the polishing time was consistent across the different iterations of the modified puck and that the incorpora- tion of additional materials like conductive epoxy sections and mounting wax did not adversely affect the polishing rate. CONCLUSIONS AND FUTURE WORK These experiments have successfully demonstrated that, in the caseof epoxypuck based mechanical cross-section analysis of packaged samples, by modifying the epoxy puck and reducing the cross-sectional area to be polished, the material removal rate Fig. 7 (a) Optical image of the sample in the slotted modified puck with conductive epoxy sections, Config3. Low magnification SEM images of the (b) leftand (c) right corners of the die in the cross-section. (d) Higher magnification SEM image of three C4 bumps closest to the die corner in (b). (e) High magnification image of the die-C4 interface. Fig. 8 Comparison of the time taken to reach from the center of one row of C4 bumps to the center of the next consecutive row for all four epoxy pucks.