Kirkendall voids

Sometimes interdiffusion between two metals is uneven and may lead to the creation of vacancies or voids. This type of imbalance is the result of possible unequal mobilities between a metal couple. These voids occur individually near the common interface. The voids, like bubbles, coalesce, resulting in porosity and loss of strength. Many thin-fihn couples exhibit this phenomenon, which is referred to as Kirkendall void creation. Al-Au, Cu-Pt, and Cu-Au are just a few examples. To be specific, it has been found (7), for instance, that in the case of Au-Ni, about five times more Ni atoms diffuse into Au than Au atoms diffuse into Ni. 

These are depicted schematically in Figure 18.4 in the case of metal A deposited on metal B. Bulk diffusion, as noted above, is the transfer of B into A or A into B through the crystal lattice. This is characterized by the coefficient D in the figure. Defect path diffusion is the migration along lattice defects such as grain boundaries, characterized by the coefficient D in the figure. Ordered A B, possible phases are indicated between the metals. Finally, Kirkendall void porosity is indicated and will be expected to be present if the interdiffusion rates from one metal to the other are not equal in both directions. 

The Kirkendall effect (8) is time and temperature dependent, and with some metal couples, it takes place even at room temperature. For instance, adhesion of solder to gold is damaged by heating to about 150°C for about 5 minutes, due to the formation of Kirkendall voids. Naturally, the formation of Kirkendall voids is accelerated by increased temperature and dwelling time. 

Kirkendall void formation can, however, be prevented from occurring by choosing the right metal species. For example, whereas platinum coating on copper is subject to the Kirkendall void creation process, the same coating on electrodeposited nickel is free of it even if heated to as high as 600°C for many hours (more than 10 hours ). 

In electronic applications, where it is common to deposit copper and/or copper alloy and tin in sequence, with a nickel diffusion barrier layer, 0.5 fim thick, between the layers present, no failure occurs. Without the nickel layers between bronze/-copper/tin layers themselves, for instance, intermetaUic brittle layer(s) and Kirkendall voids are formed, leading eventually to separation of the coated system and substrate. 

In practice, clean (very clean ) cleaved or otherwise smoothed metal surfaces should be made to effect a firm mechanical contact using a strong force but one that is still insufficient to cause macroscopic deformation even at an elevated temperature. This will have to be been done, usually, in vacuum or at least in an inert atmosphere. The problems of hard to get to (inaccessible) joints and possibly objectionable thermal conditions and the resultant undesired microstructures such as Kirkendall voids, for instance, are minimized, if not eliminated all together. Thus, good-quality, distortion-free joints requiring no additional machining or other posttreatment can be achieved. 

The rapid diffusion causes impoverishment of A1 in the TiAl3 layer leading to breakaway oxidation, because the A1203 scale formed is not maintained anymore. It also causes Kirkendall voids near the coating/substrate interface during the treatment and/or oxidation. This will reduce the adherence of coating to the substrate. 

There are, however, limitations on the degree to which the matrix/NbaSn volume ratio can be reduced in these high-tin bronze composites. For example, the niobium cores need to be spaced sufficiently for uniform reaction throughout the filament groups and niobium filaments must not be engulfed by Kirkendall voids or non-a-bronze phases [ ] during the homogenization heat treatment, which distributes the tin prior to reaction. A purpose of this study was to determine the quantity of matrix material necessary to fulfill these requirements. 

The Kirkendall effect is commonly accompanied by the Frenkel effect, the void formation in the diffusion zone. In foreign literature, the Frenkel effect is often referred to as Kirkendall voiding, which is rather confusing, as Kirkendall and Frenkel effects are competitive vacancies annihilating at the dislocation kinks and causing the Kirkendall shift, cannot be used for Kirkendall voiding, and vice versa. 

Nanoshells, produced in the diffusive reactions of nanoparticles within the ambient phase with a simultaneous formation of Kirkendall voids inside, are unstable in principle, but the shrinkage time can be very large due to their cubic dependence on the radius of the nanoparticle. The mechanism of shrinkage is the out-diffusion of vacancies from the void due to the curvature effect. 

We use the term Kirkendall effect to include both Kirkendall shift and Kirkendall voiding (or Frenkel voiding). We present a detailed analysis of the interaction of the Kirkendall effect and the inverse Kirkendall effect in nanoscale... 

Package and PWB surface finish Black Pad, Brittle Failure, and Kirkendall voiding... 

The presence of pervasive Kirkendall voids could reduce the strength of the solder joint. A mechanically applied strain could result in the sort of brittle solder joint failure shown in Figs. 58.19 and 58.20. 

FIGURE 58.18 Solder joint intermetallic formation, (a) The constituent material stack up in a BGA substrate land pad. (b) After the ball is attached, a number of intermetallics are formed between the copper pad and the SnPb solder joint. Kirkendall voids are known to form at the phosphorus-rich Ni layer and Sn-Ni intermetallic interface, (c) After board level reflow, the interface region thickens and several more intermetallics are formed. (Reprinted with permission from Renesas Technology Corp./ [2003] lEEEECTC)... 

With this approximate curve fit, extrapolating to end-use conditions can give an estimate of the time it would take for the voiding to reach 50 percent of the interface area. For instance, at 50°C nominal operating temperature, it would take 6000 days (16.5 years) to reach 50 percent of the interface area. It is important to note that the mechanisms that control the propensity and pervasiveness of Kirkendall voiding are not clearly known at this time. There isn t enough experimental data in the industry at this time to conclusively validate the aforementioned relations. 

The occurrence of Kirkendall voids is not limited to the package/solder joint interface. If OSP is used as the PWB surface finish, the same phenomenon is observed to occur on the PWB side. ... 

Kejun Zeng, Roger Stierman,Tz-Cheng Chiu, Darvin Edwards, Kirkendall void formation in eutectic SnPb solder joints on bare Cu and its effect on joint reliability, /ourwa/ of Applied Physics,2005, 97,024508. 

Zequn Mei, Mudasir Ahmad, Mason Hu,Gnyaneshwar Ramakrishna, Kirkendall Voids at Cu/Solder Interface and Their Effects on Solder Joint Reliability, Electronic Components and Technology Conference, 2005, pp. 415-420. 

This chapter reviews Uteratme on the microstructure of solder joints and the interactions of solders with substrates and related solder joint reliability issues. The substrates here are limited to Cu, Ni-coated Cu, electroless nickel/ immersion gold (ENIG), and hot air solder leveled (HASL) Sn-Pb. The solders are mainly Sn, eutectic or near eutectic Sn-Ag, and Sn-Ag-Cu alloys. Specific reliability issues discussed here include black pads of ENIG, gold embrittlement, compatibility of Pb-free solders with Pb-containing surface finish, and Kirkendall voids. 

Drop and shear tests were conducted of EGAs with near eutectic Sn-Ag-Cu solder balls on Cu pads after thermal aging at 100, 125, 150, and 175 °C (212, 257, 302, and 347 T) for 3, 10, 20, and 40 days (Ref 80). Kirkendall voids were observed at the Cu/Cu3Sn interface. Voids occupied 25% of the pad/solder interface after only 3 days of 125 °C (257 °F) aging. The void density increased with the aging time and temperature. The drop performance degraded 80% from time 0 to 10 days of 125 °C (257 T) aging. 

Failure mechanisms include phosphorus segregation, corrosion in immersion gold bath, brittle Ni-Sn intermetallics, and Kirkendall voids. 

Kirkendall Voids, Micro-voids at the interface between Ni(P)" and the Ni-Sn intermetallic were observed (Fig. 13) (Ref 93). These voids were thought to have been introduced by Kirkendall effects. It was proposed that the presence of high tensile stresses in the Ni-P" layer due to the volume reduction as a result of the phase change from the amorphous Ni-P to crystalline M3P in Ni-P, creates mud cracks in the Ni-P" layer. Propagation of these mud cracks along the Kirkendall voids results in the brittle fracture... 

---