Browse Prior Art Database

'Poisson Distribution Pattern' Recognition Test for Array Failure Analysis And Diagnosis

IP.com Disclosure Number: IPCOM000120016D
Original Publication Date: 1991-Mar-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 2 page(s) / 76K

Publishing Venue

IBM

Related People

Canniff, ML: AUTHOR [+2]

Abstract

Disclosed is an analysis technique based on the application of Poisson statistics to spatial regions of an array cell memory matrix. The technique analyzes occurrences of multiple failing cells which are not apparently related by functional organization or electrical distribution. It distinguishes their potential physical cause between a single underlying physical abnormality and multiple independent discrete defects. This delineation is invaluable to producing accurate defect counts and other more sophisticated diagnostic capabilities.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

'Poisson Distribution Pattern' Recognition Test for Array Failure
Analysis And Diagnosis

      Disclosed is an analysis technique based on the
application of Poisson statistics to spatial regions of an array cell
memory matrix.  The technique analyzes occurrences of multiple
failing cells which are not apparently related by functional
organization or electrical distribution.  It distinguishes their
potential physical cause between a single underlying physical
abnormality and multiple independent discrete defects.  This
delineation is invaluable to producing accurate defect counts and
other more sophisticated diagnostic capabilities.

      Methods now exist which enable specific electrical fail
signatures within an array memory matrix to be translated into their
physical defect causes.  Many of these electrical fail signatures,
such as bitline or wordline failing cells, are easy to recognize and
decipher into discrete defects. In other cases interpretation is more
ambiguous.  The same scattered failing cells signature may be caused
by a single physical abnormality extending over a large region of the
memory matrix, or it may consist of a large number of discrete
defects which are independent of one another. Distinguishing the
actual physical cause of such electrical signatures is key to an
accurate and effective diagnosis of array products.

      The method proposed is a key piece of a larger capability
called 'Cell Map Pattern Recognition for Arrays'. This Pattern
Recognition method decomposes the composite fail map (produced by a
number of physical defects) into a set of constituent shapes which
are relatable probabilistically to an underlying set of physical
defect causes.  After the composite cell map has been deciphered, and
all individual shapes have been extracted and analy...