Browse Prior Art Database

Improving the Resistance to Electromigration of Al Cu Stripes

IP.com Disclosure Number: IPCOM000073790D
Original Publication Date: 1971-Feb-01
Included in the Prior Art Database: 2005-Feb-23
Document File: 2 page(s) / 13K

Publishing Venue

IBM

Related People

d'Heurle, F: AUTHOR [+3]

Abstract

The scatter in data of lifetimes for a group of commonly prepared stripes (thin film conductors) is measured by a parameter 0, which heretofore has been found to be relatively high for stripes annealed at a relatively high temperature, e.g., 530 degrees C for 20 minutes. The lifetimes are measures of how long stripes at a given current and temperature will continue to carry this current before there is a catastrophic failure due to an open circuit caused by electromigration. The medium lifetime for a group of stripes is essentially an average value and individual members of the group will have a distribution of lifetimes, i.e., one member will have the shortest lifetime and one member will have the longest lifetime.

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

Page 1 of 2

Improving the Resistance to Electromigration of Al Cu Stripes

The scatter in data of lifetimes for a group of commonly prepared stripes (thin film conductors) is measured by a parameter 0, which heretofore has been found to be relatively high for stripes annealed at a relatively high temperature, e.g., 530 degrees C for 20 minutes. The lifetimes are measures of how long stripes at a given current and temperature will continue to carry this current before there is a catastrophic failure due to an open circuit caused by electromigration. The medium lifetime for a group of stripes is essentially an average value and individual members of the group will have a distribution of lifetimes, i.e., one member will have the shortest lifetime and one member will have the longest lifetime. Due to practical limitations, accelerated reliability tests are usually conducted with small numbers of samples, e.g., 9, so that the first stripe to fail corresponds to 1 over 9+1, i.e. 10% failure level. However, for application purposes, it is necessary to extrapolate data obtained in such a way to much smaller failure levels <<1%. Thus, lifetimes under machine operating conditions from two sets of stripes with the same median lifetimes, as obtained from accelerated reliability tests, will be greater for that set which exhibits the less spread in lifetime results; namely, that set with the smaller Sigma.

In determining the usefulness under current stress of thin film conductors for microcircuit application, it is important to consider both the median time (t(50) to failure of a group of samples, e.g., 12, and the distribution (a) of individual failure time about this mean. Groups of n stripes are tested together, and the logarithm of the individual failure time is plotted against the cumulative percentage failure number on probability scale (if n = 9 the first failure would be at 10%, the next one at 20%, etc.). Sigma is a measure of the slope of a line drawn through these points and is given by Sigma = 2.3 log /t/50% over /t/15.9%.

In Al-Cu (5% Cu) stripes, whic...