ABSTRACT
To quantify the design for manufacturability (DfM) success rate, one needs tangible metrics [1]. The cost and confidence level of parameters to choose from, including all aspects of reliability, yield, and cycle time, vary by orders of magnitude. The simplest ones are design rule checks (DRCs): A mandatory clean DRC report policy should ensure, at least theoretically, no fallouts until the end of the specified product lifetime. Manufacturing yield verification is hundreds of times more expensive and statistically less significant, as it is not possible to test every product for every aspect of its functionality. The situation is even more difficult at product qualification, for which small sample sizes and accelerated testing allow for very limited confidence in the validity of the data. For these reasons, physical DfM metrology is supported by conceptual parameters, such as criticality, occurrence, and detectability for failure mode and effect analysis (FMEA). FMEA enables problem anticipation and solving with only limited support of the actual physical data.
