Browse Prior Art Database

Method for Electrical Determination of Interconnect Uniformity and Defect Distribution for Deep Sub-Micron Back End of Line (BEOL) Manufacturing

IP.com Disclosure Number: IPCOM000188017D
Original Publication Date: 2009-Sep-18
Included in the Prior Art Database: 2009-Sep-18
Document File: 5 page(s) / 104K

Publishing Venue

IBM

Abstract

A method for detection of non visual defects (NVDs) in a wafer is disclosed. The method includes electrical determination of interconnect uniformity and defect distribution induced by deep sub-micron Back end of line (BEOL) manufacturing process.

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 53% of the total text.

Page 1 of 5

Method for Electrical Determination of Interconnect Uniformity and Defect Distribution for Deep Sub-Micron Back End of Line (BEOL) Manufacturing

Disclosed is a method for electrically determining interconnect uniformity and defect distribution induced by Back end of line (BEOL) manufacturing process.

Due to the complexity of BEOL processes and integration schemes, a process induced uniformity and defect distribution issues could impose a great challenge for yield, reliability, and chip performance. Time Dependent Dielectric Breakdown (TDDB) reliability is by far the most challenging low-k BEOL reliability issue.

In order to determining the effect of area scaling on reliability and uniformity of low-k BEOL interconnects, typically Poisson area scaling is used. The Poisson area scaling property yields a vertical shift of the cumulative distribution in the Weibull Plot. However, as shown in Fig. 1, Poisson area scaling cannot be applied in various cases like via structures. The horizontal shift associated with the cumulative distribution in the Weibull plot can be derived from Poisson area scaling, but is can be used only when the vertical shift is valid, and it should generate the same Weibull shape factor as vertical shift does. Unfortunately, for many BEOL TDDB cases, those two methods couldn't generate the same Weibull shape factor due to the problem mentioned above. Further, implementation of Poisson area scaling based on as-designed area ratios underestimates reliability and makes it difficult to merge distributions based on as-designed area ratios. There is an inconsistency in the β value obtained from

vertical shift, log-log plot and individual fittings and the "β" value obtained from log-log

plot is not realistic and is higher than expected.

Figure 1

Fig. 2 illustrates a "shift and compare" method as proposed by the publication for determining interconnect uniformity and reliability. The "shift and compare" method

1

[This page contains 2 pictures or other non-text objects]

Page 2 of 5

computes a "fatal" via ratio that satisfies the vertical shift equation and the horizontal shift equation simultanesouly. To compute the "fatal" via ratio, the Weibull distributions are shifted vertically on the Weibull scale by an amount "n1". The Weibull distributions are shifted when the Poisson area scaling model fails for an as-designed area ratio. Subsequently, all data points are fit together to obtain an overall "β1a". Thereafter, a

log-log plot is constructed to obtain "β1b" using "n1".

Figure 2

If "β1a" is equal to "β1b", then "n1" is determined to be the actual "fatal" via ratio for via

TDDB area...