Here is the abstract you requested from the Wirebond_2010 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.
|Automatically Better Pull- and Shear Tests|
|Keywords: Wire bonding, Quality Control, Automatic Testing|
|By far the most important test method for wire bonds is destructive pull- and shear testing. Surprisingly enough, though, tests are almost exclusively performed manually. The two main drawbacks are high cost and low quality. Low quality: there is a high risk of operator influence on the test result, both inadvertent and purposely. Even more importantly, the standard pull tests normally give entirely misleading results. Pull tests are supposed to be standardized to a 30° angle of the bond wire to the horizontal plane at both bond pads. It is well known that most loops in a given component will have a height such that the bond angle is different from 30°, resulting in a lower reading for lower loops and too high a reading for higher loops. While this geometrical effect is, in principle, easily corrected for, this is hardly ever done in practice because the angle is not known, or would require knowing the distance from source to destination pad and the loop height. The effect is well known in practice but usually underestimated. The second, and even larger, problem is that the bond wires in a given component usually have different loop heights, leading to different pull test readings even though the bond strengths may in fact be identical. Therefore the result spread is much larger than it needs to be, and the bond quality is underrated. Because the central quality criterium for many customers is the cpk value, the increased standard deviation leads to an apparently lower cpk than the process actually has. Furthermore, a skilled operator can influence the measurement by placing the test hook closer to one bond which will lower the pull force felt by the other bond. This makes it possible to pass a weaker bond as OK even though it may be far below the threshold value. An automatic pull tester which is programmed much like an automatic bonder can not only prevent such manipulations but can also automatically correct pull test values. This makes the distribution more uniform and very often increases cpk values from around 2 to better than 3.5 – without any changes done to the bond quality itself. Thus, the bond quality is seriously underestimated by today's standard procedure. We present this general problem and our solution to it by means of a fully automatic testing system.|
|Dr. Josef Sedlmair,
F&K Delvotec Semiconductor GmbH