Under Bump Metallization Development for Eutectic Pb-Sn Solders
- PDF / 1,020,432 Bytes
- 5 Pages / 417.6 x 639 pts Page_size
- 49 Downloads / 204 Views
EXPERIMENT Several different UBM schemes were prepared by sputtering and evaporation. The UBMs consist of CuNi alloy or Cu/Ni double layer as a wettable layer and Cr or Ti based adherence layer next to silicon substrate (Fig. 3). The wafer curvature measurement method is used to determine the stresses of the UBM layers. The area array contact pads are prepared by basic lithography process and circular 25 mil UBM pads are patterned. Once patterned, small squares of contact pads were diced out of the wafer so that each chip contains only one UBM pad. The eutectic solder (37Pb/63Sn) balls with 16mil diameter are mounted on the pads. Small amount of flux is put on each pad to remove any oxide and any other compounds that may be present on the pads and to promote wetting at the soldering temperature. The solder balls go through two reflows. Each reflow was done at 220 oC for 1 minute. After the first reflow, the solder balls are melted on to the contact pads to form solder bumps. Two solder bumps are then aligned together and reflowed again to make final solder joint. The solder joints are tested on a micro-mechanical tester to simulate the fatigue failure due to CTE mismatch. The joint samples are first mounted to the tester and then cycled in simple shear. During the test, the solder joint is sheared between strains of 0 to 100% in cyclic mode until it breaks. The shear displacement of the joint is measured by a capacitance strain gauge and the load is followed by a load cell. Following the mechanical testing, the samples are then examined under SEM. By examining the fracture surface of the joint, the location at which the sample breaks is determined. This information is used to understand what caused the failure of the joint. RESULTS Four types of failure are observed. The first is a failure of the sample before mechanical testing. The second type is due to a failure of the interface between the solder and the substrate during mechanical testing. The third type is due to a failure at the silicon side. Lastly, a sample joint can fail due to a ductile fracture through the solder. Sample failure before mechanical testing arises generally because their respective UBM schemes exert such a high stress on the silicon substrate during fabrication that the UBM layer peels off immediately after deposition or during patterning process. Sample C shows this type of failure. UBM of sample C peeled off about 20 hours after the sputter deposition during lithography process. The second type failure involves a failure at the interface between the UBM and the solder. Weak adhesion between the metal alloys of the UBM and the sample substrate or a weak adhesion between the metal alloys within the UBM can cause this failure. Sample A exhibits this type of failure. A SEM picture of the fractured sample shows shinny surface of solder with intact UBM, which implies a weak bonding between the substrate and the UBM (Fig. 4). An example of the third type failure is given by sample B. Under mechanical testing, sample B failed into the silicon sid
Data Loading...
