No abstract

Aluminum films deposited at three different conditions, such that texture is the only microstructural variable, were tested for electromigration behavior. Texture analysis shows that random and (111) fiber texture components are present in the films deposited by both partially ionized beam (PIB), physical vapor deposition and sputtering. Two parameters are required to properly quantify the texture: (111) volume fraction and the distribution (half-width) of the (111) fiber component. As the (111) texture becomes stronger, the median time to failure increases, while the failure standard deviation decreases. Previous texture correlations are based on incomplete information, so they cannot predict electromigration behavior in all cases.

The effects of thin Ti, TiN, or Ti/TiN underlayers on the development of the crystallographic texture and the grain structure are explored. Metal layers ∼0.5 μm in thickness of Al-0.5Cu or of Cu are deposited on these underlayers and on amorphous SiO2 as a reference. A strongly textured underlayer such as Ti〈0002〉 or Ti〈0002〉/TiN〈111〉 induces a similarly strong 〈111〉 texture in the AlCu. In copper with 〈111〉, 〈200〉, and random texture components, an underlayer induces a stronger 〈111〉 component compared to an analogous film deposited on SiO2. A nearly random texture in TiN significantly weakens the texture in subsequent metal films. Grain size distributions in all AlCu films are monomodal reflecting a process of normal grain growth. The grain size distribution for Cu sometimes deviates from lognormal. The bimodal distribution implies that grain growth is abnormal even though the median grain size does not exceed a low multiple of the film thickness.

Tin-bismuth alloys may be an alternative to lead-based solders for low-temperature applications, but very little is known about their manufacturability and reliability. This article presents an overview of these issues. First, experiments to determine the wetting properties of the Sn-Bi solder are presented. The results show that Sn-Bi solders do not wet bare copper well, but that they do wet copper having a hot-dipped Sn-Bi coating. Next, the effects of aging on the microstructure of Sn-Bi solders are described. The results show that during aging, tin is de-pleted from the solder/base metal interface. The two-phase Sn-Bi microstructure coarsens during aging; the rate of coarsening can be slowed by adding 1.0 wt. % Cu to the solder. The aging also affects the shear strength of the solder jOints, where aged joints show an increase in maximum shear stress and ductility at failure.

No abstract