Here is the abstract you requested from the IMAPS_2007 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|The Impact of Planarization on Bump Reliability in Flip Chip Technology|
|Keywords: Bump reliability, Flip chip technology, Failure modes|
|Bumping is one of the key processes in flip chip technology. Understanding bump failure modes will provide better criteria for reliability tests and for the improvement of bumping processes. This study primarily focuses on the failure analysis of failed bumps to explore the major bump failure modes and to discuss the impact of planarization on failed bumps. The failure analysis of failed bumps shows the breakage occurs between dies and bumps, mainly at the interfaces between redistributed Al traces and UBM, moreover, most of UBM was remained on the failed bumps. The EDX analysis indicates the major elements are Al and Ti on the surfaces of failed dies and Cu, Ni with small amount of Ti on the surfaces of failed bumps. From the results of bump shear test, it was found that the bump shear forces of the mechanical dies and real dies undergoing ultrasonic cleaning drop about 6% and 40% respectively, compared to those without ultrasonic cleaning. Without ultrasonic cleaning, the bump shear forces of the real dies are slightly lower than that of the mechanical dies, but after ultrasonic cleaning, the former one dramatically drops to two-third of the latter one. The failed surfaces of the real dies with ultrasonic cleaning show topographic morphology after bump shear tests, which is quite different from the other three cases such as mechanical dies with and without ultrasonic cleaning and real dies without ultrasonic cleaning. The sub-layers of UBM between redistributed Al traces and solder bumps provide the functions of barrier (Ni, Pt, Pd), wetting (Cu, Au) and adhesion (Ti, Cr, TiW). The real dies exhibit slightly lower bump shear forces than the ones of the mechanical dies without ultrasonic cleaning, which can probably contribute to the rougher surfaces of real dies causing thinner and more fragmented sputtered layers of UBM. From the results of failure analysis and bump shear tests, it was observed that during ultrasonic cleaning Ti tends to diffuse and react to form inter-metallic layers of TixAly where the consumption of the adhesion layer of Ti causes bump shear forces to drop and eventually leading to bump failure. However, for the case of mechanical dies, thicker and more uniform sputtered UBM still has enough Ti to form an adhesion layer after ultrasonic cleaning. Therefore, the bump shear forces of mechanical dies only decrease about 6% after ultrasonic cleaning and the breakage occurs at one-third of sheared bumps. On the other hand, the thinner and more fragmented sputtered UBM of real dies only has a small amount of Ti remained after ultrasonic cleaning which reduced the adhesion between redistributed Al traces and bumps leading to dramatic drops in bump shear forces and showing topographic morphology on the surfaces of failed dies. The results indicate that implementation of ultrasonic cleaning in bumped dies will have great negative impact on bump reliability due to the formation of Al-Ti inter-metallic layers between redistributed Al traces and UBM. Therefore, the cleaning processes should introduce washer/ dryer or non-clean flux instead of ultrasonic cleaning during flip chip assembly. Planarization of wafer surfaces before fabricating redistributed traces is strongly suggested.|
|Shu-Ching Ho, Engineer
ChipMOS Technologies, Inc.