Elsevier

Microelectronics Reliability

Volume 79, December 2017, Pages 119-123
Microelectronics Reliability

Nucleation and shrinkage of defects on the surface of tin induced by local strain distribution

https://doi.org/10.1016/j.microrel.2017.10.026Get rights and content

Highlights

  • Nucleation and shrinkage of defects on the tin-copper surface finish were observed by using atomic force microscopy.

  • Local strain distribution by digital image correlation is related to nucleation and growth of defects.

  • The surface energy based model can explain that the stress induced defect is hillock.

Abstract

This paper discusses nucleation of stress-induced defects on a tin surface finish. The spring back effect was used to generate compressive stresses in the tin layer. In-situ observation was carried out by means of atomic force microscopy (AFM). Nucleation of defects was found in the observed area. AFM revealed that not only nucleation but also shrinkage of defects took place. Cross-section profiles revealed that defects are classified into hillocks. Since the tin crystal structure has anisotropic properties, local stress distribution was developed by compressive stresses. The digital image correlation (DIC) approach revealed that defects were located at areas of severe strain gradient. The strain gradient corresponds with the nucleation and shrinkage of defects in in-situ observation. The surface energy based model is employed to explain the hillock formation.

Introduction

Small defects such as nodules, hillocks, and whiskers are often induced by stress and/or electric current on the surface of thin metal films. Although their size is less than several micrometers, they can be a threat to the reliability of microelectronics. Tin whiskers, which are needle-like crystals, cause electrical shorting since it can bridge to the adjacent connectors. The nodule is a small size defect involved in the nucleation process of the whisker. According to IEC60068–82, its aspect ratio of length to diameter is defined to be less than two. The hillock is a mountain-like crystal and its diameter increases with growth.

The growth mechanism of defects has been discussed in several past research studies [1]. The growth models have been proposed based on dislocation theory [2], recrystallization [3], compressive stress [4] and stress assisted diffusion [5], [6]. On the other hand, the nucleation process of the defect is still not clear. Since whisker formation is a long-term phenomenon and strongly depends on the microstructure of a film, it is not easy to capture the nucleation. Recently, in-situ observation of surface defects on thin solid films is one way to elucidate the mechanism of tin whisker growth [7]. Su et al. could control the position of spontaneous whisker growth by deliberately creating weak oxide locations, and, using consecutive observations, precisely determined the whisker growth rate [8]. These observations were performed in a vacuum using scanning electron microscopy (SEM). Onuki et al. carried out in-situ observations of nodule nucleation using atomic force microscopy (AFM) [9].

It is known that the growth of tin whiskers depends on the stress distribution in the film. Choi et al. used synchrotron scanning X-ray micro-diffraction to measure the stress distribution on a tin surface [10]. Sun et al. combined the digital image correlation (DIC) technique with SEM images to map the local strain field during whisker growth. By calculating the stress field chronologically, they revealed that the magnitude of the strain gradient plays an important role in whisker growth [11].

In this study, a possible mechanism for nucleation of defects induced by residual stresses is investigated. Defects were observed to be initiated on tin–copper surface finishes. The nucleation behavior was quantitatively evaluated based on AFM surface profiles. The DIC technique was also used to measure the local strain distribution.

Section snippets

Experimental procedure

The specimen consisted of an electroplated matte tin or tin–copper (Sn-2mass%Cu) surface finish, a nickel underlayer, and a phosphor bronze substrate as listed in Table 1. All specimens were tested without additional heat treatment. Fig. 1(a) and (b) shows inverse pole figures (IPFs) of tin and tin–copper surface finishes. The grain size of tin is larger than that of tin–copper. The texture of tin–copper is also different from that of tin. Since beta tin has anisotropic mechanical properties,

AFM observation after unloading

Nucleation of defects was observed on the surface of tin–copper surface finishes after unloading. Fig. 3 shows AFM images of specimen I, which is the tin–copper surface at 18, 117, and 189 min after unloading. The arrow in the figure indicates the bending direction. Five defects were found in the target area. Three defects (Defects C, D, E) were observed at 18 min, but two of them (C and D) disappeared, and another (E) became smaller at 117 min. Two defects (A and B) formed at 117 min and were

Discussions

The spring back effect was used to develop strains in the surface finishes. Since the melting point of tin is low, the surface finish creeps during the loading. Then, compressive strains were introduced after unloading. Defects were observed to form after unloading due to the spring back effect. Not only nucleation but also the disappearance of defects was observed.

Shibutani proposed the tin whisker growth model based on grain boundary diffusion creep [14] and nucleation theory [17]. The

Conclusion

This paper discussed the nucleation and shrinkage of defects induced by local stress distribution due to anisotropic tin texture. The spring back effect was used to generate compressive stresses in the tin surface finish and in-situ observation was carried out by using AFM. The conclusions of this paper are summarized as follows:

  • 1)

    In-situ observation of the surface showed that not only nucleation but also shrinkage of defects was found. The cross-section profile revealed that defects are hillocks

Acknowledgements

This work was supported by JSPS KAKENHI Grant Number 26630007. Authors would like to thank Takashi Maki for performing the image analysis of AFM.

References (20)

There are more references available in the full text version of this article.

Cited by (0)

View full text