Interfacial Phenomena in Al/Al, Al/Cu, and Cu/Cu Joints Soldered Using an Al-Zn Alloy with Ag or Cu Additions

The microstructures presented in Fig. 2 and 3 show the reaction occurring at the interface between the liquid solder and the Al pads, described in Ref 19, 20. That is a cyclic mechanism of dissolution and crystallization. The liquid solder dissolves the Al pads, and the resulting alloy crystallizes forming so called “scallops,” separated from each other by channels with liquid solder. The liquid solder in the channels causes further dissolution, and the alloy formed in this way yields crystallization. Despite the fact that the Zn-Al eutectic alloy with 12 at.% of aluminum and small additions of silver or copper was used as the solder, the joint contains mainly aluminum from the dissolved Al pads, especially in the case when the joints were prepared with a strong force (Fig. 3). Indirectly, this confirms the existence of the mechanism described in paper (Ref 20). The authors (Ref 20) studying the processes occurring at the interface between the liquid aluminum and the nickel pad noticed that there are three zones: d _x —dissolution zone, s _r—saturation zone, and s _s—supersaturation zone. In the first one, the liquid dissolves the pad, until the product of this reaction becomes a liquid having the N ₀ composition. Then the liquid diffuses and crystallizes. This zone is difficult to find as it is very small and does not exists for a long time. On the microstructures shown in Fig. 2b and 3a, one can extract the two other zones—s _r and s _s which differ in the content of dissolved zinc. Silver and copper which are added to the solder concentrate on the edge of “scallops,” and they are dissolved in the Al40Zn60 phase. Subsequently, it can be assumed that they react with the liquid zinc resulting from the previous reaction and form small grains of intermetallic compounds (IMCs): Cu₅Zn₈. The stronger pressure caused the decrease of the solder amount in the joint and hence a much larger part of the solder reacted with the pads forming three times larger structures. Moreover, in some cases (Fig. 3a), the formation of the delamination process was observed. The reaction products of the two Al pads were separated by the unreacted eutectic solder.

Figure 4 and 5 illustrate the microstructures of Al/solder/Cu joints prepared with the high-pressure (Fig. 4) and with the low-pressure (Fig. 5) springs. Similarly as before, the existence of s _r and s _s zones was noticed at the aluminum pad side. From the copper pad side, one can observe the creation of the intermediate layer consisting of the three intermetallic phases in the following sequence (starting from copper): β-Cu[Zn,Al] γ-Cu₅[Zn,Al]₈, and ε-Cu[Zn,Al]₄. It can be presumed that the above-described mechanism of dissolution and crystallization occurs (Ref 20). Cu and Ag additions cause the emergence of precipitates rich in Ag or Cu and the propagation of these elements to the eutectic. Moreover, the silver is dissolved in the Al₄₀Zn₆₀ phase. Higher pressure slows the rate of the phases’ growth. This is probably related to the increase of aluminum concentration which dissolved from the pads to the solder.

The Gibbs phase rule states that during the process of crystallization at the constant temperature of 500 °C in the ternary (Al, Zn, Cu or Ag) system, there are four phases in the equilibrium state. As previously mentioned, the intermediate layer composed of three intermetallic phases with different crystal lattices: β-Cu[Zn,Al], γ-Cu₅[Zn,Al]₈, and ε-Cu[Zn,Al]₄ forms at the solder-Cu pad interface. The fourth phase is the supercooled liquid (d _x ) with the N ₀ composition. The composition of N ₀ was calculated from the mass balance equation (1).where d ₁ , d ₂ , d ₃ are the thickness of the phases formed in one joint after a specific brazing time and N _i ^Cu is the average copper content in the “i” phase.

The calculations were carried out for all the solders after each brazing time and for different contact forces. The results showed that the copper content ranged from 28% for solder with a high content of alloying elements and the shortest time of brazing, to 35% for the pure eutectic after the longest period of time. It seems that the longer wetting time, the more important role that the phenomenon of diffusion played, and hence the difference of about 7% Cu occurred. The average value of N ₀ for all the measurements amounted to 31% Cu.

From the microstructures shown in Fig. 4, 5, 6, 7, and 8, one can conclude that from the liquid N ₀ composition, the intermetallic phases crystallize with the following sequence: β-Cu[Zn,Al], γ-Cu₅[Zn,Al]₈, and ε-Cu[Zn,Al]₄. The Al-Zn-Cu phase diagram (Ref 21) shows that the excessive amount of zinc (in relation to the calculated from the mass balance) is needed to keep the dissolution zone (d _x ) liquid. During the crystallization process, the intermetallic layer of the total N ₀ composition is formed, and the excess of zinc returns to the solder. The areas with high concentration of zinc can be seen on the microstructures shown in Fig. 8. According to (Ref 21) aluminum replaces zinc in the intermetallic phases, and the process is the easiest in the γ phase. Silver is most easily dissolved in the ε phase (Fig. 6).

Occurrence of IMPs layers was improved by XRDs measurements presented in Fig. 9(a) ZnAl + 1.5Ag and (b) ZnAl + 1.5Cu, respectively, in both case for connections Cu/solder/Cu. Measurements were conducted for three area of interactions pads with solder, at the edges and in the middle of connections. In both cases, the XRDs results show occurrence of IMPs, γ-Cu₅[Zn,Al]₈, ε-Cu[Zn,Al]₄, and for alloys ZnAl with Ag the AgZn₃. As is apparent from phase diagram (Ref 21), and XRDs measurements in the γ phase dissolution Al were observed.

The dependence of the change of thickness of the phase over time can be described by the following relationship:where δ is the average thickness of the reaction layer at time t, k is the growth rate constant, and n is the time exponent.

Generally, the obtained value of n determines the types of layer growth at the interface and the possible reaction mechanisms. Takaku et al. (Ref 12) suggested that for reaction of Al-Zn and Al-Zn-1%Cu solders with Cu pads, the exponent n is 0.5, which means that growth of phase is controlled by the volume diffusion. Table 2 shows the values of the coefficient k calculated under the assumption that n = 0.5. These data relate to solders with different contents of Cu and Ag, and connections obtained by different degrees of pressure. These data indicate that the greatest rate of increase has phase γ and the lowest β. Additives Cu to eutectic Al-Zn cause an increase in the rate of formation phase γ and ε, the greater share of Cu the faster growth of phases, while the Ag additives generally do not change the k value. The rate of formation of the phases is affected by different degrees of pressure; for the higher pressure, the growth of γ phase will be faster.

The kinetics of intermetallic phases growth at the liquid solder eutectic ZnAl and ZnAl with 1% Cu—copper pad on Fig. 10, was presented. The obtained results were compared with literature data (Ref 11) interface at low pressure. The kinetics of intermetallic phases’ growth at the liquid solder eutectic ZnAl and ZnAl with 1% Cu-copper pad interface at 500 °C is presented in Fig. 10. The obtained results were compared with literature data (Ref 12). Authors (Ref 12) obtained higher values of the coefficients k for phases, γ and β, and the lower for ε phase.

Analyzing Fig. 10, we can see that the value of the exponent of time n for the phases, ε and β, oscillates around the value of 0.5, which suggests that the growth of these phases is controlled by the volume diffusion mechanism. However, the γ phase values of n differ considerably from 0.5.

Considering the above, in Fig. 11(a-c) and 12(a-c), dependence of thickness of the phases is shown as a function of time, with the calculated coefficients n and k. These coefficients were calculated using the program GRAPHER_6, using Eq 2. The results of calculations shown in Fig. 11 and 12 show that the exponent of time n for phase γ in the solders of different contents of Ag and Cu ranges from 0.86 to 0.9. This suggests that the growth of γ phase is controlled not only on the volume diffusion but also on diffusion of the grain boundaries. This may result from the fact that this phase has the lowest activation energy and the lowest Gibbs’ free energy (Ref 13, 17). Figure 13 shows a comparison of the kinetics of the phases’ growth with high and low pressure applied. In the case of low pressure, there was a large amount of liquid solder to react. This resulted in a higher rate of ε phase growth (Fig. 13a) and did not cause the inhomogeneity of Zn concentration (Fig. 7). However, in the case of high-pressure application, the regions rich in zinc appeared, and the growth of the γ phase was faster. This can be probably related to the fact that in this case, the diffusion processes were responsible for the growth of the phase.