Issue link: https://iconnect007.uberflip.com/i/271756
March 2014 • The PCB Magazine 35 SPHERICAL BEND TESTING continues the substantial additional thermal excursions related to hot gas rework of BGA devices do af- fect the resistance of the material to damage. This might be attributed to higher levels of cross-linking in the polymer, which reduce the compliance of the material. The additional thermal exposure of lami- nates and solder pads caused by forced rework of large BGA devices also produces changes in the board side solder interfaces and where ENIG pad plating is present. A second failure mode, Mode 8— Fracture at the NiP/NiSn IMC layer at the board side was identified as a second failure mechanism but Mode 10 remains domi- nant. The occurrence of two failure modes did not significantly affect the distribution of the data or the calculated limits. Results: DOE3 The six lots processed for this experiment were selected from a larger group of tested lami- nates with the intention of covering the range of responses that have been encountered. We have endeavored to select a laminate from the worst and best performers as well as one from the middle of the pack. All lots meet the ex- perimental expectations for Beta values based on testing from that significant number of laminate lots. The selected lots exhibit the well- formed distributions that are expected for single failure modes and meet our condition for a well defined mixed response zone. The Weibull distributions are plotted in Fig- ure 11. As expected there is a benefit to char- acteristic life by moving to a more compliant package construction. It is also notable that the rank order of laminates in terms of sur- vivable strain has not changed under the more compliant system. The plot of 95% CC for the Weibull distri- bution parameters displayed in Figure 12 clear- ly demonstrates the increase in characteristic life for each individual laminate which are de- fined by color. There is some variation in Beta value driven by the proximity of the boundary condition for laminate B. The top ends of both of these distributions are very close to or at the strain level where mixed failure modes would be expected to occur. Summary The spherical bend test methodology has been shown to provide informative compari- sons between sample lots created by varying Figure 11: weibull distributions—doe3.