$FFXUDWHO\'HWHUPLQLQJ%ULGJLQJ'HIHFWV IURP/D\RXW

As yield improvement becomes more important the differences between the modeled faults and actual defects in the diagnostic simulation process gets in the way. The need to diagnose the failure mechanisms closer to reality is driving the test industry towards more accurately modeling of defects as much as possible. Most failures occur as shorts and opens. Bridging faults represent shorts much more accurately however the sheer number of them overwhelms the diagnostic simulations. Methods defined in prior research have reduced the number of bridges modeled in the simulations by extracting the most probable bridges based upon proximity of lines. As a result diagnostic accuracy depends upon the right set of bridges being extracted from layout information. Traditional inductive fault analysis methods that rely upon proximity and length of lines take into account the defect size but do not indicate an efficient way to rank their contribution to bridges induced by random defects. In this paper we show how coupling capacitance can take defect size into account, even when nets change metal levels, along with the proximity of the lines.