An Evaluation of Gold and Copper Wire Bonds on Shear and Pull Testing

Abstract

In microelectronic packaging technology wire bonding is a common interconnect technique. The quality and reliability of wire bonds are generally evaluated by ball shear and stitch pull testing. From the load versus time and load versus tool tip displacement plots of the shear test, three regions can be observed. Region I primarily exhibits elastic-plastic deformation occur, while crack nucleate in region II which propagates in region III which finally ends in a catastrophic failure. Fractographs reveal in the case of gold ball bonds shows fracture occurs in Al bond pad metallization close to Au-Al intermetallics. In Cu ball bonds of 1, 2, and 4ml wire sizes also Al bond pad metallization cracks but penetrate deeper into the pad which indirectly shows that the bonding layer is stronger than that of gold ball bonds. Optical microscopic observation of the sheared copper bond surfaces reveal sticking of Al which provides qualitative information of the area of the bond between the ball bond and the bond pad. In thermally aged gold ball bonds, the gold above the intermetallic layer fractures. The energy required to fracture a gold or copper ball bond of 1ml wire size is around 370J∕m2, while an aged gold ball bond consumes about 520J∕m2. Void nucleation and coalescence mechanism of ductile fracture takes place in the ball and stitch bonds, however, silicon particles may be the preferential void nucleation sites in bond pad aluminum metallization failures. To understand the second bond strength, a stitch pull test was conducted and the results showed the neck of the stitched wire cracks thus leaving behind a tail bond on the lead finger.