Issue link: https://iconnect007.uberflip.com/i/960594
44 SMT007 MAGAZINE I APRIL 2018 number of cycles for a 1% failure rate, and (3) first failure. Each of these metrics were linearly regressed versus the DOE factors. Package Characterization Solder ball height, diameter and coplanarity of the unmounted packages were measured at room temperature using an RVSI LS8000 scan- ner. Ball height was measured relative to the soldermask surface. The widest portion of the ball was taken as the ball diameter. Coplanarity was calculated according to the seating plane method described in JESD22-B108A [5] . Package warpage was measured using an Akrometrix TherMoire PS400 according to JEDEC standard JESD22-B112A [6] , and reported here for the temperature range of interest: -40°C to 125°C. Measurements were made on the bottom (substrate) side of the package after removing the solder spheres. Crack Propagation Analysis Two methods were used to examine crack propagation in the solder joints during cycling: dye-and-pry and cross-section. Dye-and-pry was a quick and simple method to obtain an overall view of cracking quantity, degree and distribution. A dye was applied to the solder joint array to mark crack loca- tions, followed by a forced separation of pack- age from board. Cracks formed during cycling were stained with ink, and were distinguish- able from fracture surfaces created merely because of the forced pry [7] . Dye-and-pry had some limitations. First, it only revealed one crack interface in each solder joint, whichever cleaved first during pry. Some- times a solder joint cracked along both pack- age and PCB sides simultaneously. Addition- ally, PCB pads often ripped out during peel, even when solder joint cracking had occurred. In these incidences it was assumed the degree of cracking was low (<50%) since the solder joint strength was greater than PCB pad adhe- sion. For cross-section, standard potting, section- ing, grinding and polishing techniques were used to prepare and study solder joint crack growth. SEM images were captured for the measurements. In all cases, sections were made through the joint center line. Cross- sections provided a more definitive picture of crack propagation location in the solder joint than dye-and-pry. Often cross-section identi- fied multiple cracks within a joint. One caution: crack front propagation may be at any arbi- trary angle to the cross-section plane, thereby distorting crack length measurements. Samples were studied for degree of crack growth, location within the solder joint, and distribution across the array. For solder joints examined by cross-section, degree of crack- ing was calculated as the percentage of visible crack length divided by apparent pad diameter (all linear). For dye-and-pry, it was the crack area (red die visible) divided by pad area for each pad where a crack surface was revealed. See Figure 4 as an example. RESULTS: Ball Size and Coplanarity Characterization Solder ball diameter and height of the unmounted packages are summarized in Figures 5 and 6, respectively. Distributions for the SMD and NSMD pads are plotted sepa- rately. Twenty packages were measured for Figure 4: Example dye-and-pry image and degree of cracking calculation.