Void Nucleation in Passivated Interconnect Lines: Effects of Site Geometries, Interfaces, and Interface Flaws
- PDF / 502,159 Bytes
- 10 Pages / 612 x 792 pts (letter) Page_size
- 78 Downloads / 178 Views
MATERIALS RESEARCH
Welcome
Comments
Help
Void nucleation in passivated interconnect lines: Effects of site geometries, interfaces, and interface flaws R. J. Gleixner, B. M. Clemens, and W. D. Nix Department of Materials Science and Engineering, Stanford University, Stanford, California 94305-2205 (Received 18 October 1996; accepted 27 January 1997)
Stress driven nucleation of voids in passivated aluminum interconnect lines is analyzed within the context of classical nucleation theory. A discussion of sources of tensile stress in such lines leads to an upper limit of 2 GPa. Calculations suggest that even at this high stress, nucleation rates are far too low to account for observed rates of voiding. Void formation at a circular defect at the line/passivation interface is then considered. In this case, a flaw on the order of nanometers in size may develop into a void under the imposed stress. These results strongly suggest that void nucleation in aluminum interconnect lines can be controlled by eliminating defects in the line/passivation interface.
I. INTRODUCTION
The failure of interconnect lines due to thermal and electromigration stress voiding has become a major issue in the design of reliable integrated circuits.1–3 Voids are observed to nucleate within a line, then grow until the line is severed and the interconnect fails. Although the driving force for void growth is well understood, the phenomenon of void nucleation has received very little attention. There are several reasons for this apparent lack of study. First, a large number of potential nucleation sites exist in interconnect lines, making it difficult to model the nucleation process. Next, nucleation is essentially an atomic scale process, where the use of bulk properties is questionable. Additionally, due to its inherently stochastic nature, nucleation theory has been an area of materials science in which experimental observation and verification have been quantitatively difficult. The lack of well-established models for void nucleation remains a limitation in modeling the failure of interconnect lines. In the case of thermal stress voiding, the line is subjected to large hydrostatic tensile stresses upon cooling from high passivation deposition temperatures. Here voids are often assumed to have nucleated during this cooling process, and the growth process is of concern.4–6 Under electromigration conditions, the stress in a line increases slowly with time. In this case, voids are not observed for a substantial fraction of the time to failure. This is frequently modeled as an incubation period for void nucleation, ending when the maximum stress in the line reaches a critical value.7,8 When this stress is reached, voiding is assumed to occur, and the subsequent growth is analyzed. In both of these cases it is assumed that the critical stress required for void nucleation as well as the location of the void are known. J. Mater. Res., Vol. 12, No. 8, Aug 1997
http://journals.cambridge.org
Downloaded: 18 Mar 2015
A better understanding of the nucle
Data Loading...
