Wetting and Interfacial Chemistry of SnZnCu Alloys with Cu and Al Substrates

The results of wetting angle measurements on Cu substrates are shown in Fig. 1; one could see that for all the investigated solder/substrate pairs the effect of time on wetting angle is negligible. Wetting angles of Sn-Zn-Cu alloys on Cu are generally lower than the wetting angles of Sn-Zn on Cu. On the other hand, the results of different Sn-Zn-Cu alloys on Cu are within the experimental error, which is particularly well seen after 180 s of wetting. The present wetting angle data for Sn-Zn-xCu alloys are significantly lower than the data of (Ref 12) for alloys of similar composition. Similarly, different fluxes used in the present study are most likely the reason why wetting angles of Sn-8.8Zn alloy are lower than the angle of the same alloy on the same type of substrate reported earlier (Ref 13). Such a difference is an indication that deoxidation of metallic surfaces by the flux used earlier (Ref 13) is not sufficient.

Figure 2 illustrates the interfacial microstructure of Sn-Zn-1.0Cu/Cu couples after 15, 30, 60, and 180 s of wetting, respectively. Similar results were obtained for x = 0.5 wt.% Cu and x = 1.5 wt.% Cu. One could see that with increasing time of wetting the observed number and thickness of interlayers (formed of reaction products between liquid alloy and solid substrate) varies. More on reactive wetting can be found in review (Ref 14). Although there are a few intermetallic phases in the sub-binaries of Sn-Zn-Cu (namely: CuZn₄, CuZn, Cu₅Zn₈, CuSn₃, and Cu₆Sn₅), regarding the Sn-Zn/Cu interface Zn has stronger than Sn affinity for Cu, the Cu₅Zn₈ has the lowest Gibbs free energy of formation (Ref 15) and also low activation energy of growth (Ref 16). As shown in Fig. 2, there is a certain difference between the interfaces after 15, 30 s and the interfaces after 60, 180 s of wetting. In the first case (wetting time up to 30 s) a single interlayer can be observed. According to EDS analysis results collected in Table 1 its composition is close to ε-CuZn₄ phase assuming that Sn substitutes portion of Zn. In the second case (wetting time 60 s and more) there are two distinct interlayers. The upper layer is clearly scalloped from the side of the solder and according to EDS (Table 1) its composition is close to CuZn₄ phase. The thickness of the layer adjacent to the substrate is greater and increases with time of wetting. Its composition determined with EDS matches Cu₅Zn₈, although some content of Sn is observed. This phase is flat on both sides and is known to be a barrier for diffusion of Sn toward substrate thus preventing formation of Cu₆Sn₅. Since it is prone to fractures which enable diffusion of Sn, its growth should be controlled. According to thermodynamic calculations (Ref 17) Cu₅Zn₈ is the phase that should form first, i.e., before CuZn₄, at the Sn8.8Zn/Cu interface, however present results after 15 and 30 s of wetting do not confirm this. According to the present results, the ε-CuZn₄ phase is the first that is formed at the interface and the γ-Cu₅Zn₈ appears after 60 s of wetting and grows faster than the ε-CuZn₄. It can be speculated that γ-Cu₅Zn₈ is formed as a result of copper diffusion from solid substrate to ε-CuZn₄ and subsequent nucleation and crystallization within the ε-CuZn₄, as in the micrograph (Fig. 3) showing particles of ε-CuZn₄ in the matrix of Sn-Zn-1.5Cu solder after 180 s at 250 °C with grains of γ-Cu₅Zn₈ inside. Figure 4 shows the Sn-Zn-1.0Cu/Cu couple after 180 s at 250 °C with the line indicating the position of the original interface at the beginning of wetting. The fact that the thickness of intermetallic layer from the side of solder is approximately the same as thickness from the side of substrate suggests that diffusion of Cu from substrate into γ-Cu₅Zn₈ is similar to that of Zn from the solder.

Generally, if there is more than one intermetallic formed at the interface, the intermetallic in direct contact with liquid at the triple line determines wetting (Ref 18). The fact that wetting angles (Fig. 1) after 15 and 180 s are close means that intermetallic in contact with liquid is the same after 15 and 180 s, which is in agreement with EDS results (Table 1). Thickness of the intermetallic layers in Fig. 2 was measured at 10 spots at least at two separate micrographs per each solder/substrate couple with Axio Vision and later averaged. It has to be noted that for wetting times 15 and 30 s the intermediate layer is thin, as can be seen in Fig. 2(a), (b), with EDS analysis we found its composition to be close to CuZn₄. For this reason in Fig. 5 there is only one data bar at 15 and 30 s, while for 60 and 180 s the two data bars represent thickness of CuZn₄ and Cu₅Zn₈ layers. After 15, 30 s the thickness of intermetallic layer is similar. Starting from 60 s, with increasing time of wetting thickness of Cu₅Zn₈ layers significantly increases as shown in Fig. 5, whereas thickness of CuZn₄ remains small. Similar results were obtained for x = 0.5 wt.% Cu and x = 1.5 wt.% Cu.

Based on the present experimental results the following order of events at the interface of liquid Sn-Zn-xCu (x up to 1.5 wt.%) with Cu substrate is proposed. Initially, liquid Zn from the solder reacts with Cu dissolved from solid substrate and thin CuZn₄ layer is formed (Fig. 6a). The most likely reason that it is CuZn₄ rather than Cu₅Zn₈ is that the influx of Zn toward substrate exceeds the dissolution rate of Cu. Therefore, each Cu atom is instantly consumed as the fresh Zn atoms are coming from the bulk of the solder. Next (Fig. 6b), Zn diffuses through the CuZn₄ toward substrate (at the same time Cu diffuses the other way) and the Cu₅Zn₈ is formed adjacent to the substrate. Because of diffusion of Zn from the solder the Cu₅Zn₈ keeps growing and its growth is faster than the CuZn₄ as the system tends to reach the minimum energy (Fig. 6c).

In the case of alloys containing 1.0 and 1.5 wt.% Cu after 180 s of wetting, isolated precipitates of CuZn₄ phase are present in the solder particularly numerous in the vicinity of the interface. According to Huang et al. (Ref 19) for Sn-9Zn-xCu alloys (x = 0.5-3.0 Cu) Cu₅Zn₈ is the primary crystallizing phase. Although the CuZn₄ precipitates seem to be concentrated in the vicinity of interface they are randomly distributed in the solder. If the samples were kept in liquid for a long time sedimentation would be observed due to a density difference between precipitates (for instance, Cu₅Zn₈ is ~8.0 g/cm³) and the solder matrix (~7.0 g/cm³) as explained by Song (Ref 15). Song (Ref 15) studied sedimentation behavior of Cu-Zn precipitates in the above-mentioned solders held at 250 °C for 20 min. Since the temperatures at which the intermetallics from Cu-Zn crystallize are much higher than 300 °C, it was assumed (Ref 15) that at 250 °C the originally precipitated “as-cast” IMPs would sediment. They observed both CuZn₄ and Cu₅Zn₈ in the SnZn-0.5Cu solder but only Cu₅Zn₈ in solders containing ≥1 wt.% Cu. In their experiment they found that the “as-cast” IMPs dissolve into Sn-Zn matrix at 250 °C and precipitate again. After 20 min of holding at 250 °C they found rounded precipitates of ε-CuZn₄ which sediment in 0.5 and 1.0 wt.% Cu samples, and dendritic precipitates of Cu₅Zn₈ randomly distributed in Sn-9Zn matrix. They explained this dissolution-precipitation behavior by high solubility of Zn in liquid Sn.

The wetting angles of Sn-Zn-1.0Cu on Al, shown in Fig. 7, are much lower than the wetting angle of the respective alloy on Cu. Considering wetting of Sn-Zn-1.0Cu on Al substrate, the present data are close to those of Sn-8.8Zn and Sn-Zn-In reported earlier on the same type of substrate and in the presence of the same flux, but longer time of wetting (Ref 20). Figure 8(a) and (b) illustrates interfacial microstructure of Sn-Zn-1.0Cu/Al couples after 15 and 180 s of wetting, respectively. One could see that the interface is much different than in the case of Sn-Zn-Cu/Cu couples, i.e., no interlayer is observed at solder/substrate interface in agreement with Al-Sn and Al-Zn phase diagrams.

The interface between Sn-Zn-1.0Cu solder alloy and Al substrate is rather rough and the substrate roughness increases with increasing time of wetting. This system is an example of dissolutive wetting, theory and examples of which can be found in review (Ref 21). It is observed that at 180 s, solder grooves and penetrates the substrate along grain boundaries leading to some grains being separated, at least partially, from the surrounding substrate. This kind of dissolution of substrate may lead to Al grains detached and floating near the substrate as shown in Ref 11, 22. The Sn-Zn-1.0Cu/Al interface after 180 s of wetting resemble the Sn-Zn/Al and Sn-Zn-In/Al interfaces after 300 s of wetting and more, discussed in our earlier work (Ref 20, 22). The EDS analysis performed near the interface confirmed that Al dissolved in Sn-Zn-1.0Cu alloy. Nevertheless, as pointed before, one has to be cautious regarding the EDS results because of small size of microstructure features and the associated errors of measurement. Figure 9 shows the Sn-Zn-1.0Cu/Al couple after 180 s at 250 °C. No significant dissolution of the substrate, except for grooves, can be observed. This and the fact that the wetting angle after 180 s is close to that after 15 s is an indication that dissolution kinetics is much slower than the kinetics of wetting.