Abstract
IC bond pads having thin pad aluminum (Al), less than 0.7μ thick, were probed multiple times by a variety of cantilever probe tips. Probe marks were created by an experimental high force probe card having various tip diameters. A finite element model of the probe tip was matched to the probe mark scrub length to gain a little more academic understanding of what to expect as probe parameters vary. Simulations using this model will be helpful in the future for situations where physical experiments are difficult or costly to run. Bond pads in the experiments include various ON Semiconductor circuit under pad (CUP) structures having Al metallization and silicon dioxide (SiO2) interconnect, previously demonstrated to be robust against cracking when compared against traditional IC bond pads. Smaller ball size and bond contact area are desirable as pads shrink on future products, but this will heighten any adverse effects of probe marks due to the relative % area increase beneath the bond. Experiment evaluations include bond pull strength (BPS), and bond shear (BS) for gold (Au) ball bonds of varying ball diameters over the wide range of probe marks, to begin checking the validity of a customary “probe mark area” maximum limit in wirebonding. Data indicates that large, deep probe marks can indeed bond ball lift fails, especially for non-optimized bonding recipes. It appears that probe mark depth, not area, is the most adverse factor in bond reliability. No issues relating to bonding over probe marks are expected in the more controlled and “gentle” manufacturing situations.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have