Skip to Main Content
Gold ball-bonding is still used for manufacturing over 90% of the worlds of electronics packages and has moved towards finer pitch in response to increasing demands for more I/O's in smaller spaces. However, as commonly observed when reducing the size of components or features in semiconductor assemblies, reliability decreases. Wirebonding behaves similarly in terms of isothermal bake testing and it is observed that the maximum time ballbonds can be baked without observing high strength ball lifts and low strength ball lifts (HSBL and LSBL respectively) decreases. One of the reasons for this is the need to accommodate smaller bonded balls (by a change of capillary geometry) with diameters not much larger than the wire diameter at pitches of 40 and 35μm and even a small loss of bonded ball area by intermetallic degradation can lead to competition between failure at the pad or neck. Previously at pitches >50pm, the larger ball size/wire diameter ratio favoured more robust ballbonding. Understanding the process parameters that affect the occurrence of HSBL and LSBL failures is useful, even if the science behind the root causes of the ball lift phenomenon in ultra-fine pitch applications is currently lacking or at best, developing. An easily controlled and measured parameter is ballbond squash height, which depends on the plastic deformation of the free air ball (FAB) and which might therefore result in ballbond residual mechanical stresses that may couple with geometrical, epitaxial and non-stoichiometry stresses during intermetallic compound growth during ageing. Another factor of interest is the bake temperature, temperatures of interest usually being 150°C, 175°C and 200°C. This paper briefly reports the effects of squash height and bake temperature on the occurrence of ball lifts in a 4N gold wire at 40μm bond-pad pitch.