Issue link: https://iconnect007.uberflip.com/i/1307491
NOVEMBER 2020 I DESIGN007 MAGAZINE 41 in a consensus on a detailed layup and lay- out for the test article. We struggled with that because, as those of you that are familiar with the industry realize or have recognized in the past, if you put the exact same stackup lay- out design in front of two different fabricators, there are differences in their press operation, relief pattern, panel layout, and DFM projec- tion software. That will get you two different results. We finally had to give ground on one dielectric layer in each to get to an identical predicted structure. In February, the test labs that we are working with on this began con- sensus discussions will continue detailed sam- ple handling and testing. And in late April of this year, the project pro- posal was submitted to the U.S. Army's Cor- nerstone Group at Rock Island. I must stress that at this point, when this was submitted, many thousands of dollars and hours of time were invested because we were concerned about this problem. We have gone about as far as we can go off the cuff. We need to get some funding, and since this is aimed primarily at military applications and use, that's why we're going to Cornerstone with this. Objectives Project objectives were to provide publicly accessible information on some contributors to the weak interface failure mechanism. If you try to do an ideal design that covered every possible element on a fish-bone diagram, we would all be much older, grayer, and many millions of dollars poorer than we are today. We have taken the approach of getting this in bite-sized chunks. From a manageability stand- point, getting viable data was quickly deemed preferable to getting the ultimate answer to all problems and questions in 10 years. We have taken the mini-phase approach. We are going to help determine current limits on reliable design production methods and mate- rials for use in high-density interconnect microvia electronics and critical high-reliability applica- tions. We are not building cellphones or address- ing that. This is a narrow market segment. We are very eager to help enable improved correlation between predictive modeling, sim- ulation reliability, and actual results. This is one of the reasons that our friends from ANSYS are so willing to help with this; they very much want to improve their ability to handle all dif- ferent structural types of high-density stacked/ staggered microvia reviews and be much more confident that those are going to provide rea- sonably good predictability. Each phase of this project is designed to inform and direct subsequent phases in terms of specific structures, materials, manufactur- ing testing techniques, etc., and to succes- sively improve the predictive modeling. The first results will be compared with the predic- tive modeling, the predictive modeling will be readjusted, and then the second phase will make the same iterative loop. The test article is seen in Figure 1, includ- ing covers stacked and staggered from one to four. It is a little more aggressive than the iMac ESA Structure. It is probably less aggres- sive than Lockheed's structure. We purposely chose to limit it to a single material in this first phase. It will provide a readout, though, on a few commonly commonly employed structural choices during the design phase. It gives stag- gered microvias in varying offset staggered, and there are a couple of microvia sizes there. Methodology and Variables • "Shop A" vs. "Shop B" • Structure (stack height, stacked vs. staggered, stacked over buried, degree of offset, etc. ) • 17 different D structure coupons were required (when metallization included) We are very eager to help enable improved correlation between predictive modeling, simulation reliability, and actual results.