Issue link: https://iconnect007.uberflip.com/i/869081
58 SMT Magazine • September 2017 Package Reball Figure 8 shows a schematic of the preform process used to reball the WLPs. After desolder- ing the package-side, residual debris and flux were cleaned using a flux-off spray and a coarse- bristled brush. A custom solder preform was placed ball-side up on a fiberglass-resin fixture. A thin coat of flux was applied to the package- side, and the WLP was placed on top of the pre- form. To improve ball attach and prevent the part from blowing away, a metal weight was placed on top of the WLP and preform. The weight, WLP, and preform stack were reflowed using the rework system, then the weight and preform were carefully removed after the re- balled package had cooled. The entire stack-up was profiled by attach- ing a thermocouple between the preform and the fixture. The top and bottom heater temper- atures and times were adjusted to meet the fol- lowing critical process parameters: 60-90 s soak between 150°C and 217°C, and 60-90 s reflow between 217°C and 250°C. Results and Discussion Several board-assembled WLP units were used for rework and reball development activ- ities. A demount recipe was created by profil- ing a PCB-mounted WLP and adjusting the top heater and bottom heater settings. Figure 9 shows the temperature profile obtained during demount. WLPs were successfully demounted using a peak temperature of ~235°C and a 70-s time above liquidus. After demount, a recipe was developed to vacuum scavenge the solder from the package. A WLP was placed front-side up in a spring-loaded snugger inside a fiberglass-resin composite fix- ture. A thermocouple was attached to the sili- con backside and used to measure the package temperature. Top and bottom heater set points were adjusted until the package temperature re- mained between liquidus and 240°C. Figure 10 REWORK AND REBALL CHALLENGES FOR WAFER-LEVEL PACKAGES Figure 8: Schematic showing the reball stack-up. Figure 9: Graph showing the package temperature as a function of time during demount. Figure 10: Representative microscope image showing WLP after successful vacuum scavenging. Only a thin layer of solder remains on the under- bump metallization.