Issue link: https://iconnect007.uberflip.com/i/1137649
110 SMT007 MAGAZINE I JULY 2019 Tables 6 and 7 show the pull test results of wire bonds made on scratches after reliabil- ity testing. The information about the width of the scratches is also included in the tables. The range of observed scratch widths in Tables 6 and 7 was from 17–33 mm. The widest scratch is slightly more than half of the diameter of the wire bond; however, wire pull tests after reli- ability testing did not show failures. In MIL- STD-883H METHOD 2010.12 [3] , wire bonds made outside of the bonding pads are consid- ered acceptable if >75% of the wire bond is within the bonding pad, which means that a maximum of 25% of the wire bond could be on a non-bondable surface. However, the area occupied by the scratch under the wire bond in this study is much more than 25%. By rough calculation, a scratch with 12.3-mm width would take about 25% of wire-bond area of the wire bond with 63-mm diameter if the scratch is aligned about in the middle of the circu- lar wire-bond area. For a scratch with 33-mm width, the scratched area occupied under the wire bond would be about 65% of the total wire-bond area. One of the reasons that wide scratches could be tolerable for the wire-bonding process in this study could be that scratches did not neg- atively affect the surface finish wire- bonding capability, as shown by the EDS results in Figures 14 and 15. Even though the gold layer was sig- nificantly damaged and removed, the pallidum layer—considered the wire- bondable surface—was not com- pletely removed. In other words, an entire scratched area may not nec- essarily be a non-bondable surface. This could be considered a benefit of using an ENEPIG finished surface over using an ENIG finished surface for wire bonding. Yield Analysis After Specification Update Based on the successfully performed exper- iments, changes were made to the internal specification. Magnification used for inspec- tion was reduced from 20X to 10X. Addition- ally, the specification was changed to allow for the presence of isolated nodules and scratches as per the criteria in the following paragraph and Table 8. When inspected under 10X magnification, individual nodules, such as the most commonly observed Type A and C nodules, and shallow scratches with no exposed nickel or copper in the wire-bond area are considered accept- able. In some cases, process variations during PCB manufacturing, or any nodules formed from random particles that form Type B nod- ules, may create unusual deviations resulting in exceptional nodules and scratches. In these cases, the following specification referred by Table 8 can be used. Since this new specification has been applied, the yield loss at the PCB manufacturer due to nodules and scratches has shown a signifi- cant decrease over 10 months. Fig- ure 18 shows the average yield loss due to nodules and scratches since Table 6: Pull Test results of wire bonds made on scratches after temperature/humidity test. Table 7: Pull test results of wire bonds made on scratches after thermal cycling test. Table 8: Specification for nodules and scratches.