Effect of Corrosion on Mechanical Reliability of Sn-Ag Flip-Chip Solder Joint
2016 Volume 57 Issue 11 Pages 1966-1971
Details
2016 Volume 57 Issue 11 Pages 1966-1971
The shear strength and failure behaviors of the Sn-2.3 mass%Ag flip-chip solder joints before and after corrosion test were investigated. The relationships between the shear strength, corrosion amount, and fracture mode are elucidated in this study. The shear strength of the Sn-Ag solder bump joints decreased with increasing amount of corrosion, mainly due to the formation of brittle corrosion products. In addition, the shear strength was changed with corrosion site. After the shear test, the failure mode switched from a bulk-related ductile fracture to a corrosion-related brittle fracture, depending on the site and amount of corrosion. The top-side corroded bump did not affect the shear strength, whereas the shear strength was decreased for the partially corroded area at the side of the bump. After prolonged corrosion reactions, the joints had extremely low shear strength values and very brittle fracture surfaces. This result was discussed in terms of the relationship between the corrosion site, the shear height, and the resulting force-displacement (F-x) curves during the shear test.
In the electronic packaging technology, the reliability and strength of the solder joints generally depend on the interfacial structures between metallization and lead-free solder.
Since solder joints are often subjected to mechanical loadings during handling and system use, mechanical properties of solder joints, such as the fatigue, shear strength, and creep resistance, are the crucial issue for the reliability of the solder joint as well as the integrity of electronic packaging.1–7) The most popular method to evaluate the mechanical reliability of the solder joints is the bump (or ball) shear test due to its simple and convenient implementation.6,7)
Many studies have been performed on various properties of Pb-free solder joints such as wetting, reflow, interfacial reactions, intermetallic compound (IMC) growth, bump/ball shear test, bending test, drop test, electromigration test, and other reliability tests.1–13) However, corrosion reactions and their effect on the mechanical reliability of Pb-free solder joints are still uncertain. Although the corrosion of solder alloys is not a major concern for electronic packages used in a normal environment, it may become a serious problem when they are used in an excessive moisture environment. Especially, the corrosion behaviors of solder alloys are more important for the automotive, aerospace, and maritime industries because of more extreme use conditions.
Some researchers have studied the corrosion behaviors of conventional Sn-Pb solders and Pb-free solder alloys such as Sn-Ag, Sn-Ag-Cu, Sn-Cu and Sn-Zn.14–23) However, few studies have been conducted on the corrosion behavior of the Sn-Ag flip-chip solder bump and the corresponding bump mechanical reliability. The objective of this study was to evaluate the effect of corrosion on the bump shear strength and failure behavior (fracture mode). A representative Pb-free solder, Sn-Ag solder, was used to observe the effects in this study. The joint strength of the Sn-Ag flip-chip solder bump after corrosion test was evaluated from the bump shear tests and the test results are discussed in connection with the corrosion reactions.
Figure 1 shows the schematic of the cross-sectional bump structure used in this study. Sn-2.3Ag (in mass%) solder formed by electroplating are utilized as an interconnect material in flip chip assembly. A Ni(3 μm)-Cu(3000 Å)-Ti(1000 Å) under bump metallization (UBM) is used to connect the Sn-Ag solder bump and the Al bonding pad of silicon wafer. The reflow process was performed in a fluxless solder reflow machine (Geneva STP300, SEMIgear, Inc. USA) using formic acid. The formic acid is served as an oxide reduction agent. The gaseous formic acid is active at typical soldering temperature to react with metallic oxide in reflow process chamber. The properties of reflowed solder bumps depend mainly on the reflow temperature profile. Figure 2 shows the reflow temperature profile used in this study. Two key parameters of reflow profile are the peak temperature and dwell time. The dwell time means the duration time above the melting temperature of solder alloy. The peak reflow temperature was fixed at 240℃. The reflow time over 221℃, the eutectic temperature of the Sn-Ag solder, was approximately 60 sec. After reflow process, the bump height and diameter of the reflowed bump were approximately 75 and 90 μm, respectively.
Schematic of cross-sectioned Sn-Ag solder bump.
Reflow profile of Sn-Ag solder bump.
The distilled water (pH 5～7) is used as a corrosive solution in this study to evaluate the corrosion behavior of the reflowed bump. The reflowed wafers were dipped into the corrosive solution for up to 2 weeks (336 hours) at room temperature. The wafers were removed at 2, 5, 10, 24, 48, 168 and 336 hours. After immersion for the specific period, each specimen was taken out from the corrosion bath.
Bump shear test was performed before and after corrosion test to evaluate the effect of corrosion on the bump mechanical reliability. A bond tester (PTR-1100, Rhesca Co. Ltd., Japan) was employed for the bump shear test. The shear speed and shear height in the bump shear test are 150 μm/sec and 15 μm, respectively. The average shear strength (a. u.: arbitrary unit) of thirty solder bumps was taken for each condition. The fracture mode of each test site was examined after shear testing to evaluate the failure mode. The fracture surfaces were investigated thoroughly by scanning electron microscope (SEM, Inspect S50, FEI) and energy dispersive X-ray spectroscope (EDX).
Recently, we evaluated the corrosion behavior of the Sn-2.3 mass%Ag flip-chip solder bump in the corrosive solution.24) In the study, we obtained a fundamental understanding of corrosion behavior of the Sn-Ag solder alloy, and verified the morphological transition of the corrosion product. The morphology of corrosion product changed with corrosion time. At the early stage of corrosion, the dendrite-shape corrosion products are observed on the bump surface. The six dendrites-shape corrosion products converted into the six faceted-shape pyramids as the corrosion time increased. With the prolonged corrosion process, the cavities between adjacent pyramids are filled with corrosion products, resulting in the formation of the octahedron-shape corrosion products.24)
In this study, we performed the bump shear test to evaluate the effect of corrosion on the bump mechanical reliability. Figure 3 shows the shear strength variation of the Sn-Ag flip-chip solder bumps with corrosion time. In this case, we only tested and evaluated the un-corroded bumps (bumps without corrosion) on the wafers. The bump shear strength did not significantly change with corrosion time, and had constant values of approximately 30 gf (Fig. 3).
Bump shear strength as a function of corrosive solution exposure.
Figure 4 shows the top view SEM images of the corrosion behavior of Sn-Ag solder bump exposed to distilled water, the solder bump was dipped in various time ranges. The corrosion product was composed of polygon-shaped particle. As shown in Fig. 4(b), polygon-shaped particles were formed on the surface of the solder bump after dipping for 12 hr. Explosion of the solder bump also occurred due to severe corrosion reaction (see Fig. 4(c)). Finally, the solder bump was almost completely changed to polygon-shaped particles after dipping for 48 hour (see Fig. 4(d)). The mechanism of this severe corrosion damage was essentially a galvanic cell reaction.14)
Corrosion behavior of Sn-Ag solder bump exposed to distilled water; (a) before (b) 12 hr, (c) 24 hr, and (d) >48 hr.
Figure 5 shows the bump shear strength variation with site and amount of corrosion in the corroded bumps. It is interesting to note that the shear strength value was changed with corrosion site. In addition, the shear strength of the Sn-Ag solder bumps significantly decreased with increasing amount of corrosion. In other words, the site and amount of corrosion in the corroded bumps are critical factors for the bump mechanical reliability (bump shear strength). We performed the bump shear test on the five different parts of the tested wafers with amount of corrosion; normal bump without corrosion (before corrosion) (Fig. 5(I)), partially corroded bump at the top of the bump (Fig. 5(II)), partially corroded bump at the side of the bump (Fig. 5(III)), bump with medium portion of corrosion (Fig. 5(IV)), and completely corroded bump (Fig. 5(V)). The corresponding bump schematics are depicted in Fig. 5. In the case of the normal bumps without corrosion, the average shear strength was approximately 32 gf. On the other hand, the shear strength for the bumps with medium and large corroded portion were approximately 12 and 2 gf, respectively.
Bump shear strength variation with site and amount of corrosion product.
Another interesting feature to note from this study is that the corrosion location will impact the required shear strength. The top-side corroded bump did not significantly affect the shear strength of the Sn-Ag solder bump under shear loading. As shown in Fig. 5, the shear strength for the bumps with corrosion at the top of the bump had similar values to that for the bumps without corrosion. The corroded part of the bump did not affect the shear loading of the bump because the corrosion part is far from the shearing plane. The shearing planes are indicated with red line in the Fig. 5. On the other hand, in the case of the partially corroded bump at the side of the bump, the shear strength was approximately 20 gf. Compared to the bump without corrosion, the shear strength was decreased by approximately 25%. This correlates to the bump shearing height, which was 15 μm in this study. In the case of the partially corroded bump at the side of the bump, the corrosion parts are partially on the shearing plane, resulting in the deterioration of the bump shear strength during the bump shear test.
Generally, thick reaction layers at the interface between solder and substrate or brittle phases in the solder bumps (or balls) are very sensitive to external stress and provide sites of initiation and paths for crack propagation.1,25) In other words, the integrity and mechanical strength of the solder joint correlate to the solder microstructure and interfacial. In this study, the corrosion products acted as the brittle phases in the solder bump. From the results of bump shear test, we can conclude that the bump corrosion induced the deterioration of the bump mechanical reliability. The cracks easily initiated in the corroded bumps during the mechanical shear test, which will be mentioned in more detail later.
The variations in the shear strength could be explained by an in-depth study of their fracture surfaces and force-displacement (F-x) curves. Figure 6(a) and (b) shows the SEM micrographs of the Sn-Ag bumps with different corrosion site before and after shear test. The direction of the shear action is indicated by white arrows. In the case I of the normal bump without corrosion, the ductile solder fracture is observed on the fracture surface, as shown in Fig. 6(a) and (b). A similar result is observed in the bump with corrosion at the top of the bump. As mentioned before, the corroded part of the bump did not affect the shear loading of the bump, as shown in Fig. 6 case II. In this case, the corrosion part is far from the shearing plane.
SEM micrographs of the Sn-Ag bumps; (a) before and (b) after shear test, (c) corresponding F-x curves and (d) shear strength.
A different fracture surface was observed in the bump corroded at the side of the bump. Shown in Fig. 6 case III are SEM images of the partially corroded bump at the side of the bump before and after shear test. As shown in Fig. 6 case III, the shape of the fracture surface was an incomplete circle-shape. The damage of solder bump due to corrosion was observed near the pad edge as indicated by a blue arrow in the Fig. 6 case III. This area at the end of the bump in the shear direction is well match with the corrosion site of the bump before shear test. This damage weakens the solder bump and induced the shear strength drop during bump shear test. Such differences in the fracture modes and fracture surfaces shown in Figs. 6(a) and (b) corresponded to the different shapes of the F-x curves, as shown in Fig. 6(c). Thus, we investigated the F-x curves during bump shear test. We confirmed similar F-x curves for case I and case II, as shown in Fig. 6(c). On the other hand, in the case III, a low peak shear strength value and early drop are observed in the F-x curve. The low shear strength value and early shear strength drop are related to the corroded bump part at the end of the bump in the shear direction.
During the shear test, the solder joint fractures along the weak points, which indicate the failure mode of the solder joint. In this study, the failure mode of the Sn-Ag flip-chip bump joints switched from a bulk-related ductile failure before corrosion test to a corrosion-related brittle failure after corrosion test. Namely, fractures occurred inside the bulk solder for samples before corrosion test while the joints failed at the corrosion site within the bump after corrosion test. This transition can be directly linked to the formation of brittle corrosion product. Figures 7(a) and (b) show the SEM micrographs of the bump with medium portion of corrosion before and after shear test and the corresponding F-x curve, respectively. The top view micrographs before and after shear test in Fig. 7(a) showed that much of the corrosion products occupied the fractured surface. Conversely, we also found the remaining small portion of the solder on the fractured surface. The original bump position before shear test is indicated with a white dotted circle in the fracture surface micrograph of Fig. 7(b). We also clearly confirmed two distinctive parts in the F-x curve of Fig. 7(b). These two parts correspond to the only corrosion product part without solder and solder-containing fractured part, respectively. For ease of understanding, we presented the relation between fracture surface and F-x curve by using black lines, as shown in Fig. 7(b). These fracture surface and F-x curve are well matched to each other. Compared with the relatively smooth fracture surface of un-corroded solder bump, the fracture surface of the corroded solder bump appeared much coarser in the shear test (see Fig. 7).
SEM micrographs of the bump before and after shear test and corresponding F-x curve.
Figure 8 show the SEM micrographs of the bumps with large portions of corrosion before and after the shear test and the corresponding F-x curves. As shown in Fig. 8, the corroded bumps had extremely low shear strength values of approximately 2～4 gf. Close examination of the fracture surfaces disclosed that the Sn-Ag bumps completely converted into the corrosion products, as shown in Fig. 8 (a). Compared to the neighbor bumps, the corroded bump joints had very brittle fracture surfaces. Consequently, the brittle property of the corroded bumps increased the sensitivity of the bump to stress during the shear test.
SEM micrographs of the fracture surfaces after shear test and corresponding F-x curve.
Figure 9 shows the summary of the bump shear strength variation and fracture modes transition with site and amount of corrosion. The shear strength of the solder bumps decreased with increasing amount of corrosion. In addition, the corrosion site in the corroded bumps is a critical factor for the bump mechanical reliability. In this study, the correlation between mechanical shear strength, microstructures and fracture surfaces clearly shows that the formation of the corrosion product causes brittle solder bumps.
Effects of surface morphology of Sn-Ag solder bump on shear strength.
In this paper, we evaluated the effect of corrosion on the shear strength and failure behaviors of the Sn-Ag flip-chip solder joints. We performed the bump shear test of the Sn-Ag flip-chip solder joints before and after corrosion test, and successfully discussed the test results in connection with the corrosion reactions. The failure mode switched from a bulk-related failure (ductile fracture) to a corrosion product-related failure (brittle fracture), depending on the site and amount of corrosion. The shear strength of the Sn-Ag solder bumps changed with corrosion site. In addition, the shear strength significantly decreased with increasing amount of corrosion. The top-side corroded bump did not affect the shear strength of the Sn-Ag solder bump. Alternatively, the shear strength decreased for the partially corroded bump at the side of the bump, because the corrosion parts were partially located on the shearing plane. After prolonged corrosion reactions, weak corroded part/un-corroded part interfaces led to a failure mode of brittle fracture for all corroded solder joints, due to the formation of brittle corrosion products. In cases of the completely corroded bumps, the joints had extremely low shear strength values and very brittle fracture surfaces. It could be concluded that the corrosion time had little effect on the bump shear strength for the bumps without corrosion, whereas the site and amount of corrosion in the corroded bump played an important role in the bump mechanical reliability.
