Interfacial fracture toughness of Pb-free solders

Abstract

Increasing environmental concerns and pending government regulations have pressured microelectronic manufacturers to find suitable alternatives to Pb-bearing solders traditionally used in electronics packaging. Over recent years, Sn-rich solders have received significant attention as suitable replacements for Pb-bearing solders. Understanding the behavior of intermetallics in Sn-rich solders is of particular concern as the microelectronics industry progresses towards Pb-free packaging. The formation of intermetallic compounds results from the reaction of the solder with the metallization on the substrate in the electronic package. While the presence of the intermetallic is an indication of good wetting, excessive growth of the intermetallic can have a dramatically adverse effect on the toughness and reliability of the solder joint. Understanding their fracture behavior will lend insight to their reliability under mechanical and thermomechanical strains.

We investigated the intermetallic compound growth associated with Sn–0.7Cu and Sn–4.0Ag–0.5Cu solders on Ni–Au, Ni–Pd, and Cu substrates. (Ni,Cu)₃Sn₄ was present at the Ni interface for both solders but was coarser in the case of Ni–Pd. Cu₆Sn₅ and Cu₃Sn were observed for both solder types. The Cu₃Sn layer was similar in thickness and appearance for both solders, but the Cu₆Sn₅ was smoother and rounder in the case of Sn–0.7Cu. Additional time above liquidus resulted in growth of the Cu₆Sn₅ layer and eventual spalling of the IMC grains. The effect of the intermetallic on the toughness (K_Q) of the solder joint was investigated using a modified compact tension specimen. Typical failure modes included bulk solder failure, intergranular separation, and intermetallic fracture, or cleavage. In some cases, additional time above solder liquidus was used to shift the dominant failure mode from that dominated by the bulk solder to interfacial delamination through the intermetallics. Solder joint fracture toughness was different between Ni–Sn and Cu–Sn interfacial intermetallics and was also affected by the relative intermetallic thickness. The relationship between solder and intermetallic microstructure and mechanical properties is discussed.

Introduction

The microstructure and mechanical behavior of traditional Sn–Pb solders has been well documented and is well understood [1], [2], [3], [4], [5], [6]. Recently, environmental concerns, legislation, and even customer preference are driving the microelectronics industry towards implementation of Pb-free solders in packaging [7], [8], [9], [10]. The electronics industry has settled on a few Sn-rich solders after careful consideration of mechanical behavior, cost, processibility, and integration. Sn–3.9Ag–0.5Cu, and Sn–0.7Cu are two common Pb-free solder candidates for replacing the traditional Sn–Pb solder systems.

Upon reflow, Sn-rich solders readily react with several metallizations to form interfacial intermetallics. Typical pad metallizations include Ni–Au, Cu, or Ni–Pd. When reflowing Sn-rich solders on such pad metallizations, several intermetallics may be formed, such as Cu₆Sn₅, Ni₃Sn₄, Cu₃Sn, and Ag₃Sn. While the presence of these intermetallic compounds is necessary to promote a proper bond and adhesion between the solder and metallization, at large thicknesses the intermetallic layer can significantly decrease the fracture toughness of the joint. Fig. 1 shows the relationship between apparent fracture toughness (measured using a compact tension geometry) and intermetallic thickness for several Pb-free and Pb-bearing solders reflowed on Cu undermetal (after Frear et al. [1]). Note that with increasing intermetallic layer thickness (particularly above 5–10 μm), the toughness of the joint decreases significantly. Thus, the brittle nature of most of these intermetallics, as well as the lower ductility of Pb-free solders compared to Pb–Sn, have caused some concern regarding the reliability of Pb-free solders. In particular, the growth rate of the intermetallics is accelerated for Sn-rich solders due to larger amounts of Sn available for reaction with the pad metallizations. Furthermore, the reliability of Sn-rich solders is degraded in cases of high strain rates (approximately >10⁻¹/s) that can be experienced primarily during pure mechanical shock rather than lower rates more common to thermal cycling. In the case of higher strain rate loading, failure analysis of the solder joint typically shows that fracture takes place at the intermetallics [1], [7].

Several studies have been performed to evaluate the fracture behavior of Pb–Sn solder joints. The majority of these studies have reported that Pb–Sn solder joints have a tendency to fail through the bulk solder, although some studies suggested that the intermetallics at the solder-undermetal interface acted as crack initiation sites [1]. In the case of the bulk solder failures, fracture was attributed to a coarsening of the Pb–Sn microstructure. Kang et al. [2] attempted to relate decreased fatigue life to the thickness of Ni₃Sn₄ intermetallics formed in a Pb–Sn solder joint on Ni metallization. They thermally cycled Si power transistors soldered to Ni using Pb–Sn solders. They found that failure sometimes took place near the solder/intermetallic to undermetal interface, but within the solder rather than the intermetallic. Keller [3] found that tensile testing of previously thermal cycled joints to failure resulted in fracture through the intermetallic region while tests of non-cycled joints showed joint failure through the bulk solder. Thermal cycling caused further growth and coarsening of the intermetallics making them more susceptible to fracture. Hall et al. [5] have shown similar results from aged joints using 60Sn–40Pb. It appears that thermal aging induced intermetallic growth in Sn–Pb on Cu solder joints increased interfacial failure and decreased failure through the solder itself.

Frear and Vianco [7] showed that we can expect a similar trend in Pb-free solders, but at an accelerated rate, due to faster intermetallic growth rate associated with Sn-rich solders. They also indicated that shear tests performed at a strain rate of round 6.6 × 10⁻⁴/s and with the intermetallic thickness less than 10% of the overall joint thickness, fracture took place primarily through the solder. In tests performed in tension, fracture in both the solder and within the interfacial intermetallic region was observed. More recently, Kurosaka et al. [11] related aging of Sn–Pb solder joints on Ni pads to lower tensile strength. They also reported on the aging effects related to Sn–3Ag–0.5Cu on Ni joints. In both cases, the failure mode shifted from bulk solder failure to brittle intermetallic failure along with a corresponding decrease in tensile strength. Again, intermetallic growth and coarsening brought on by thermal aging decreased the strength of the solder joint.

Understanding the kinetics of intermetallic formation in these solder and metallization systems is important to the prediction of solder joint reliability. For example, Kurosaka et al. [11] showed a significant change in the interfacial morphology of Sn–3.0Ag–0.5Cu on Ni when finishing the Ni with approximately 60 nm of Pd. They found the that the growth of (Cu,Ni)₆Sn₅ was minimized by the small layer of Pd. Microstructure characterization of subsequent pull tests revealed ductile failures occurring within the solder while samples made without the Pd finish showed evidence of less desirable brittle failures occurring within the intermetallics. It has also been reported that thick Cu₆Sn₅ tends to have deleterious effects on reliability while a thin, irregular layer of the same phase still indicates good solder wetting but without the decreased reliability [12], [13], [14]. Furthermore, intermetallic thickness has been shown to affect the fracture behavior of solder joints under shear stress. Deng et al. [15] found that total intermetallic (Cu₆Sn₅ and Cu₃Sn) thickness greater than 20 μm tended to cause fracture along the intermetallic interfaces under shear failure. Deng et al. [15] and Chawla and Sidhu [16] used Finite-element analysis (FEA) to predict shear stress–shear strain curves related to increased intermetallic thickness as well as the morphology. It was found that for both planar and nodular intermetallics, greater intermetallic thickness leads to higher shear stress at the same shear strain. They also showed that nodular morphology of the intermetallics resulted in higher shear strength than planar intermetallics of the same thickness. FEA modeling results of nodular intermetallics reported by Chawla and Sidhu [16] showed stress concentrations in the nodules suggesting that they serve as crack initiation sites.

The conditions for mechanical testing will also have a significant influence on the failure modes for Sn-bearing solders. While very low strain rates will tend to cause ductile solder failures, higher deformation rates resulting from mechanical shock or drop tests can cause the failure modes to take place in the IMC region. Frear [17] studied the affect of deformation rate on Sn–40Pb on Cu fractures in a shear orientation. Frear [17] discovered that fracture occurred through the intermetallics for shear strain rates greater than 6.6 × 10⁻⁴/s. Below that rate, failure tended to occur within the bulk solder. It should be noted that intermetallic thicknesses will also have a significant effect on failure location with failures generally shifting from bulk solder to intermetallic as the intermetallic grows.

While the intermetallics of Pb–Sn and Pb-free solders on different undermetals have been studied, as summarized above, comprehensive studies characterizing the deformation behavior of the intermetallics and the resulting fracture toughness seem to be missing. In this study we have characterized the fracture behavior of several Sn-rich, Pb-free solder joints under fracture toughness testing. Sn–4.0Ag–0.5Cu and Sn–0.7Cu and three metallizations, Cu, Ni–Au, and Ni–Pd, were investigated to create six combinations. The growth of the intermetallics over extended reflow times of 45, 90, and 180 s was also studied on selected combinations. The relative fracture toughness was measured using a modified compact tension geometry. The fracture behavior was studied and related to the intermetallic morphology and microstructure, as well as solder microstructure. This study presents a thorough evaluation of the relationship between the intermetallics formed during reflow (described in the companion paper, part I), their fracture behavior, and the resulting solder joint fracture toughness in Cu and Ni-containing metallization.