The limits of automatic systems for defect location using e-beam testing

Abstract A lot of development and research has been done in order to make the e-beam analysis automatic. However, in the case of the defect location on ICs whose internal structure is unknown, automatic methods are generally not efficient. In this paper, based on a real example of defect location through e-beam testing, we show that in cases of poor information, an efficient (i.e. quick and cheap) method cannot be fully automated: is has to use the experience of the analyst and must include some advanced external tests so as to reduce the number of e-beam observations.