Browse Prior Art Database

Interactive Fault Density Display, a First Step Towards Designing for Yield

IP.com Disclosure Number: IPCOM000037311D
Original Publication Date: 1989-Dec-01
Included in the Prior Art Database: 2005-Jan-29
Document File: 2 page(s) / 51K

Publishing Venue

IBM

Related People

Gandemer, S: AUTHOR [+3]

Abstract

Known programs for computing the random chip yield are based on an estimation of the average number of faults - on a chip. In these programs, an important piece of information is missing, namely, the geographical distribution of faults on the chip. This is the object of our invention to provide an interactive display of the fault density on a given design.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 64% of the total text.

Page 1 of 2

Interactive Fault Density Display, a First Step Towards Designing for Yield

Known programs for computing the random chip yield are based on an estimation of the average number of faults - on a chip. In these programs, an important piece of information is missing, namely, the geographical distribution of faults on the chip. This is the object of our invention to provide an interactive display of the fault density on a given design.

A solution has already been proposed [1] to throw random defects on a design, screen the one causing faults with the appropriate shape checking code, then display the local density of faults using a grid array. We have shown that a large population of defects (usually more than 100 defects/mm2) must be thrown to obtain the correct statistical data on the fault density. Accuracy has to be traded off for CPU time, which limits the use of that solution to relatively small designs or highly repetitive ones such as memory cells. The display part is represented on Fig. 1.

In Fig. 2, we have represented the critical areas Ac1 and Ac2 corresponding to radiuses R1 and R2 of the defect distribution. Let us recall [2] that if the center of a defect of radius R falls within the associated critical area (corresponding, of course, to the same radius), it will create a fault.

One other way to rewrite the same definition is to say that every defect of the distribution, such as R /R1, has a probability of failure equal to unity everywhere within the cri...