Effects of zinc on the interfacial reactions of tin–indium solder joints with copper

Abstract

Sn-20.0 wt%In (Sn-20.0In) alloy is a promising base material in Pb-free solders for low-temperature applications. Zn is often used as an additive to Pb-free solders to reduce the extent of undercooling during reflow. Cu is the most commonly used substrate in electronics industry. Interfacial stability at Sn–In–Zn/Cu joints is crucial to reliability of electronic products. In this study, interfacial reactions between Sn-20.0 wt%In-x wt%Zn (Sn-20.0In-xZn) solders and Cu where x = 0.5, 0.7, 1.0, 2.0, 3.0, and 5.0 at 150, 230, and 260 °C were experimentally examined. It is found that the reaction phase formation and interfacial morphologies are strongly influenced by Zn concentrations. The reaction phases evolve from the Cu₆Sn₅ phase, CuZn and Cu₅Zn₈ phase, to Cu₅Zn₈ phase with higher Zn doping in the solders. The Cu₅Zn₈ phase acted as a diffusion barrier and suppressed the growth of the Cu₆Sn₅ phase. The results indicate that 2.0 wt%Zn addition resulted in the gentlest reactions during both soldering and solid-state ageing in Sn-20.0In-xZn/Cu couples.

Introduction

Soldering is the key step in various interconnection technologies including tape-automated bonding (TAB), flip chip-ball grid array (FC-BGA) packaging, and the more recent three-dimensional integrated circuit (3D IC) packaging in electronics industry [1, 2]. Reliable solder joints ensure the required lifetime of electronic products, while uncontrolled interfaces at solder joints can raise significant reliability concerns. Conventional eutectic Pb–Sn solder has been banned for consumer electronic products due to the prominent neural toxicity of Pb [3]. In the past decade, extensive efforts have been devoted to developing alternative Pb-free solders [1, 2, 4–6]. The Sn–In binary alloys, particularly the Sn-20.0 wt%In alloy, have been regarded as a base-alloy for low-temperature applications [7–14]. Zn is a frequently used additive to Pb-free solders. It can reduce the extent of undercooling of Pb-free solders [15, 16]. Cu is the most commonly used substrate materials in electronic packaging. The effects of Zn addition on interfacial reactions of Sn-20.0 wt%In-x wt%Zn (Sn-20In-xZn, x = 0.5, 0.7, 1.0, 2.0, 3.0, and 5.0) solders with Cu substrates at 150, 230, and 260 °C are examined in this study. Extremely sensitive responses of reaction phase formation, interfacial microstructure evolution, and reaction kinetics due to minor Zn additions are reported and discussed.

Experimental Procedures

Sn-20In-xZn (x = 0.5, 0.7, 1.0, 2.0, 3.0, and 5.0) alloys were prepared from pure Sn (99.98 %, Showa, Japan), pure In (99.9 %, Showa, Japan), and pure Zn (99.9 %, Showa, Japan). Proper amounts of constituent elements were cleaned, precisely weighed, and encapsulated in quartz tubes under a 10⁻² mbar vacuum. The capsules were annealed at 600 °C for a week to ensure homogeneity of alloys. After heat treatments, the capsules were quenched in icy water. The alloys were then sectioned into 2-g-weight ingots by a diamond saw. A 500-μm-thick pure Cu foil (99.98 %, Sigma-Aldrich, USA) was cut into 5 mm by 8 mm pieces and then metallographically ground and polished down to 0.3 μm Al₂O₃ powders. Each Cu substrate was coated with a thin layer of flux (water soluble, MEC-W-2326, MEC, Taiwan) and encapsulated together with the Sn-20In-xZn solder ingot in quartz tubes under a 10⁻² mbar vacuum. The capsules were annealed at 150, 230, and 260 °C for predetermined lengths of reaction time. At 230 and 260 °C, the Sn-20In-xZn/Cu reactions were liquid/solid (L/S) reactions. The L/S capsules were quenched in icy water at the end of reactions. However, the reactions at 150 °C were solid/solid (S/S) reactions. Prior to the heat treatments at 150 °C, the couples were prepared with a 30-s reflow process at 230 °C. After reactions, the couples were also quenched in icy water. The resultant couples were analyzed metallographically with a scanning electron microscope (SEM) (JEOL JSM5600, Tokyo, Japan). The compositions of the reaction phases were determined using electron probe microanalysis (EPMA) (JEOL JXA-8200SX, Tokyo, Japan). Pure elements were used as the standard specimens for calibration and the standard ZAF scheme was used for correcting the quantitative measurements in the EPMA. The thicknesses of the intermetallic compounds (IMCs) were measured with an image-processing-software (ImageJ, USA).

Results

Interfacial reactions in Sn-20In-xZn/Cu couples reacted at 230 °C

Figure 1a–e shows the backscattered electron image (BEI) micrographs of the Sn-20In-0.5Zn/Cu and Sn-20In-1.0Zn/Cu couples reacted at 230 °C for 30 min, and of Sn-20In-1.0Zn/Cu, Sn-20In-2.0Zn/Cu, and Sn-20In-3.0Zn/Cu couples reacted at 230 °C for 2 h, respectively. As shown in Fig. 1a, a scallop-type intermetallic layer was observed at the interface. The composition of the interfacial layer was determined to be Sn-55.6 at.%Cu-7.0 at.%In-6.0 at.%Zn. Since the major constituents are Cu and Sn, the phase equilibria of the Cu–In–Sn and Cu–Sn–Zn ternary systems were considered. According to the 250 °C isothermal sections of both Cu–In–Sn [17] and Cu–Sn–Zn [18] ternary systems, Cu–Sn and Cu–In binary phases have large mutual solubility between In and Sn, while there are about 6 and 10 at.% Zn solubility in Cu₆Sn₅ and Cu₃Sn phases, respectively. Moreover, as the Cu–Sn phase regions in both Cu–In–Sn and Cu–Sn–Zn systems are approximately in parallel with the In–Sn and Sn–Zn binary boundaries, i.e., the Cu compositions remain unchanged in the homogeneity ranges of these ternary Cu–Sn compounds [17, 18], In and Zn atoms are likely to substitute the Sn atoms in the ternary Cu–Sn phases. As a result, the quaternary phase with 55.6 at.%Cu is presumed to be the Cu₆Sn₅ phase with 7.0 at.%In and 6.0 at.%Zn. Similar results were found in 0.7 wt%Zn-doped couples. Only a scallop-type Cu₆Sn₅ phase with 9 at.%In and 6.4 at.%Zn was observed at the interface. The Cu₆Sn₅ layers in both couples grew thicker and ripened with increasing reaction time. The reaction paths across the interfaces in 0.5 and 0.7 wt%Zn-doped couples are liquid/Cu₆Sn₅/Cu at 230 °C.

Fig. 1

BEI micrographs of the Sn-20In-xZn/Cu couples reacted at 230 °C: a x = 0.5 for 30 min, b x = 1.0 for 30 min, c x = 1.0 for 2 h, d x = 2.0 for 2 h, and e x = 3.0 for 2 h. At 230 °C, the Cu₆Sn₅ and Cu₅Zn₈ phases formed in Sn-20In-xZn/Cu couples when x ≤ 0.7 and x ≥ 1.0, respectively

As shown in Fig. 1b, the interfacial morphology changed when the Zn doping level was increased to 1.0 wt%. Instead of the scallop-type layer, a loose reaction layer with fine grains was observed at the interface. Its composition was determined to be Cu-64.8 at.% Zn without detectable In or Sn content. According to the Cu–Zn binary phase diagram [19], the phase is presumed to be the Cu₅Zn₈ phase. As shown in Fig. 1c, the loose Cu₅Zn₈ layer became denser after a 2-h reaction. Similar results were found in Sn-20In-2.0Zn/Cu and Sn-20In-3.0Zn/Cu couples. As shown in Fig. 1d and e, planar reaction layers were formed at the interfaces. The compositions of the layers were determined to be Cu-66.0 at.%Zn and Cu-66.5 at.%Zn, respectively. Both of them are presumed to be the Cu₅Zn₈ phase, same as those formed in Sn-20In-1.0Zn/Cu couples. In addition, since Zn is the major constituent in the solder that participated in the reactions in these couples, faster reaction-progressions were expected for the couples with higher Zn doping levels, which resulted in formation of thicker and flatter reaction layers, as shown in Fig. 1c–e.

In summary, when the doping level of Zn ≤ 0.7 wt% in the Sn-20In-xZn/Cu couples reacted at 230 °C, the reaction product is the Cu₆Sn₅ phase. On the other hand, when the Zn content ≥1.0 wt%, the Cu₅Zn₈ phase would form at the interface. The reaction products in the Sn-20In-xZn/Cu couples at 230 °C are summarized in Table 1.

Table 1 Summary of reaction products between the Sn-20.0 wt% In-x wt% Zn solders with Cu, Ni, or Ag substrates reacted at 230, 260, or 150 °C

Interfacial reactions in Sn-20 In-xZn/Cu couples reacted at 260 °C

Figure 2a–e shows the BEI micrographs of Sn-20In-0.7Zn/Cu couple reacted at 260 °C for 2 h, and of Sn-20In-1.0Zn/Cu couples reacted at 260 °C for 30 min, 2, and 6 h, and Sn-20In-2.0Zn/Cu couple reacted at 260 °C for 2 h, respectively. As shown in Fig. 2a, scallop-type IMC layers were found in the 0.7 wt%Zn-doped couple. Its composition was determined to be Sn-53.7 at.%Cu-6.9 at.%Zn-8.0 at.%In and the IMC layer was presumed to be the Cu₆Sn₅ phase with some In and Zn contents. Similar results were found in Sn-20In-0.5Zn/Cu couples reacted at 260 °C. For the 2.0 wt%Zn-doped couples, as shown in Fig. 2e, the IMC layer composed of smaller scallop-type grains was observed. According to its composition of Cu-66.0 at.%Zn, this IMC layer was presumed to be the Cu₅Zn₈ phase without detectable In and Sn contents. Similar results were found in Sn-20In-3.0Zn/Cu couples reacted at 260 °C. These results of interfacial reactions in 0.5, 0.7, 2.0, and 3.0 wt%Zn-doped couples reacted at 260 °C are all similar to those in couples reacted at 230 °C. Both Cu₆Sn₅ and Cu₅Zn₈ phases were formed when Zn doping level ≤0.7 and ≥2.0 wt%, respectively, as summarized in Table 1.

Fig. 2

BEI micrographs of the Sn-20In-xZn/Cu couples reacted at 260 °C: a x = 0.7 for 2 h, b x = 1.0 for 30 min, c x = 1.0 for 2 h, d x = 1.0 for 6 h, and e x = 2.0 for 2 h. At 260 °C, the Cu₆Sn₅ and Cu₅Zn₈ phases formed in Sn-20In-xZn/Cu couples when x ≤ 0.7 and x ≥ 2.0, respectively, while the transient CuZn phase formed when x = 1.0

The reactions in 0.5, 0.7, 2.0, and 3.0 wt%Zn-doped couples reacted at 260 °C were similar to those in couples reacted at 230 °C; however, the reactions changed drastically in 1.0 wt%Zn-doped couples, i.e., Sn-20In-1.0Zn/Cu, reacted at 260 °C. As shown in Fig. 2b, after a 30-min reaction at 260 °C, a loose IMC layer was formed at the interface. Its composition was determined to be Cu-40.8 at.%Zn-5.8 at.%Sn-2.4 at.%In. Since In and Sn have similar atomic sizes and crystallographic properties, great mutual solubility between Cu–In and Cu–Sn compounds has been found in the Cu–In–Sn ternary system [7, 17]. Therefore, the contents of In and Sn were considered together as the Sn content and the IMC layer was presumed to be the CuZn phase with 2.4 wt%In and 5.8 wt%Sn according to the Cu–Sn–Zn 250 °C isothermal section [18]. When the reaction time was increased to 2 h, the CuZn layer grew thicker and became denser, as shown in Fig. 2c. In addition, a bright thin layer could be found between the previously formed CuZn layer and the Cu substrate. The composition of the newly formed bright IMC layer was determined to be Cu-21.5 at.%Zn-17.8 at.%Sn-5.6 at.%In. For similar reasoning mentioned above, the Cu–Sn–Zn phase equilibria were employed with the Sn content taken from the combined contents of In and Sn; however, it falls within the CuZn + Cu₆Sn₅ two-phase region. In order to identify the phase, the reaction was prolonged to 6 h for a larger reaction zone. As shown in Fig. 2d, the thickness of the CuZn layer did not increase much, but the bright layer grew thicker. As shown in the magnified image in Fig. 2d, the bright phase is composed of fine dark grains embedded in a bright matrix. Therefore, the dark grains and bright matrix are presumed to be the CuZn and Cu₆Sn₅ phases, respectively. As a result, the diffusion path across the interface is liquid/CuZn/Cu in the initial stage of reaction and then transforms to liquid/CuZn/CuZn + Cu₆Sn₅/Cu after a longer reaction time in Sn-20In-1.0Zn/Cu couples reacted at 260 °C. The Zn-doping effect upon Sn-20In-xZn/Cu interfacial reactions at 260 °C is summarized in Table 1.

Interfacial reactions in Sn-20In-xZn/Cu couples reacted at 150 °C

Figure 3 shows the BEI micrograph of the Sn-20In-0.5Zn/Cu couple reacted at 150 °C for 480 h. Prior to the solid-state ageing, a thin IMC layer has been found at the interface due to the fast reaction between Sn–In–Zn solders and Cu substrates. The IMC formed at the early stage of reactions is presumed to be the Cu₆Sn₅ phase with 14 at.%In and 4 at.%Zn according to its composition of Cu-26 at.%Sn-14 at.%In-4 at.%Zn. The thickness of the Cu₆Sn₅ layer increased with longer ageing time, such as the 480 h-reacted couple shown in Fig. 3. In addition, the Zn content in the Cu₆Sn₅ layer gradually changed from 4 wt% near the solder to 12 wt% adjacent to the Cu substrate, indicating that Zn is likely the dominant diffusion species towards the Cu substrate in the Cu₆Sn₅ layer.

Fig. 3

BEI micrograph of the Sn-20In-0.5Zn/Cu couple reacted at 150 °C for 480 h. The Cu₆Sn₅ phase formed in Sn-20In-0.5Zn/Cu couples reacted at 150 °C

Figure 4a–e shows the BEI micrographs of the as-jointed Sn-20In-0.7Zn/Cu couple, and the couples reacted at 150 °C for 5, 18, 50, and 600 h, respectively. As shown in Fig. 4a, two thin IMC layers had been formed prior to solid-state reactions for the 0.7 wt%Zn-doped couple. According to compositional analyses, the dark and bright IMC layers are presumed to be the CuZn phase with 9.6 at.%Sn and 3.0 at.%In and the Cu₆Sn₅ phase with 14.0 at.%In and 6.0 at.%Zn, respectively. The reaction phases across the interface were initially solder/CuZn/Cu₆Sn₅/Cu in Sn-20In-0.7Zn/Cu couples. After a 5-h reaction at 150 °C, as shown in Fig. 4b, the Cu₆Sn₅ layer grew thicker and part of the CuZn layer grew thicker as well, while the other part remained a thin layer with new Cu₆Sn₅ layer formed at the opposite side of the CuZn layer. As shown in Fig. 4c, when the reaction time was increased to 18 h, the CuZn layer became very thin and the interface became a Cu₆Sn₅/CuZn/Cu₆Sn₅ sandwich structure. With even longer reaction time of 50 h, the entire IMC became a single Cu₆Sn₅ layer, as shown in Fig. 4d. With further prolonged reaction up to 600 h, the Cu₆Sn₅ layer was still the only IMC at the interface, as shown in Fig. 4e.

Fig. 4

BEI micrographs of the Sn-20In-0.7Zn/Cu couples reacted at 150 °C: a as-jointed (30 s at 230 °C), and for b 5 h, c 18 h, d 50 h, and e 600 h. The CuZn and Cu₆Sn₅ phases formed in Sn-20In-0.7Zn/Cu couples reacted at 150 °C, while the CuZn phase transformed into the Cu₆Sn₅ phase for longer reaction time

Figure 5a–d shows the BEI micrographs of the Sn-20In-1.0Zn/Cu couples reacted at 150 °C for 120, 240, 480, and 600 h, respectively. No noticeable IMC could be found in the as-jointed Sn-20In-1.0Zn/Cu couple. After a 120-h reaction at 150 °C, several special features can be found at the interface. According to compositional analyses; they are (i) solder/Cu₆Sn₅/Cu₅Zn₈/Cu, (ii) solder/Cu₅Zn₈/Cu₆Sn₅/Cu, and (iii) solder/Cu₆Sn₅/Cu₅Zn₈/Cu₆Sn₅/Cu, as indicated by arrows in Fig. 5a. Similar solder/Cu₆Sn₅/Cu₅Zn₈/Cu₆Sn₅/Cu interfacial phase sequence has also been reported in Sn-9Zn-3Bi/Cu interfaces reacted at 170 °C [20]. When reaction time was increased to 240 h, in addition to the three types of interfacial phase sequence in the 120-h-reacted couple, two other feathers were found. They are (iv) solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu and (v) solder/Cu₆Sn₅/Cu, as indicated by arrows in Fig. 5b. When reaction time was further increased to 480 and 600 h, the entire interfaces became the combination of types (iii) and (v), as shown in Fig. 5c and d, respectively; that is, solder/Cu₆Sn₅/Cu₅Zn₈/Cu₆Sn₅/Cu where the Cu₅Zn₈ layer was discontinuous at some regions. Figure 6a–c shows the BEI micrographs of the Sn-20In-2.0Zn/Cu couples reacted at 150 °C for 50, 120, 240, and 600 h, respectively. Similar to the result in the as-jointed Sn-20In-1.0Zn/Cu couple, no noticeable IMC could be found in the as-jointed Sn-20In-2.0Zn/Cu couple. After a 50-h reaction at 150 °C, two types of interfacial morphology can be found. They are (i) solder/Cu₅Zn₈/Cu₆Sn₅/Cu and (ii) solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu, as shown in Fig. 6a. With longer reaction time up to 600 h, both type (i) and (ii) interfaces remained and the Cu₆Sn₅ layer grew thicker, as shown in Fig. 6b–c.

Fig. 5

BEI micrographs of the Sn-20In-1.0Zn/Cu couples reacted at 150 °C for a 120 h, b 240 h, c 480 h, and d 600 h. Five types of interfacial phase formed at interface of Sn-20In-1.0Zn/Cu couples reacted at 150 °C. They are i solder/Cu₆Sn₅/Cu₅Zn₈/Cu, ii solder/Cu₅Zn₈/Cu₆Sn₅/Cu, iii solder/Cu₆Sn₅/Cu₅Zn₈/Cu₆Sn₅/Cu, iv solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu, and v solder/Cu₆Sn₅/Cu. The Cu₅Zn₈ phase transformed into the Cu₆Sn₅ phase for longer reaction time

Fig. 6

BEI micrographs of the Sn-20In-2.0Zn/Cu couples reacted at 150 °C for a 50 h, b 240 h, and c 600 h. Two types of interfacial phase formed at interface of Sn-20In-2.0Zn/Cu couples reacted at 150 °C. They are i solder/Cu₅Zn₈/Cu₆Sn₅/Cu and ii solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu

Figure 7 shows the BEI micrograph of the Sn-20In-5.0Zn/Cu couple reacted at 150 °C for 600 h, respectively. No noticeable IMC could be found in the as-jointed Sn-20In-5.0Zn/Cu couple. After reactions at 150 °C, only one Cu₅Zn₈ layer formed at the interfaces and the Cu₅Zn₈ layer grew thicker with longer reaction time, as shown in Fig. 7. The reaction phase formation in Sn-20In-xZn/Cu couples with Zn doping levels of 0.5, 0.7, 1.0, 2.0, 5.0 wt% are summarized in Table 1. The interfacial reactions and microstructural evolution in the solid-state reactions are much more complicated than those in L/S couples. The reaction mechanisms, diffusion paths, phase transformation, and microstructural evolution of both the L/S and S/S Sn-20In-xZn/Cu reactions will be discussed in the following sections.

Fig. 7

BEI micrograph of the Sn-20In-5.0Zn/Cu couple reacted at 150 °C for 600 h. The Cu₅Zn₈ phase formed in Sn-20In-5.0Zn/Cu couples reacted at 150 °C

Discussion

Reaction mechanisms of the L/S Sn-20In-xZn/Cu reactions

Figure 8a shows the schematic Cu–In–Sn–Zn isothermal tetrahedron at 250 °C [17, 18]. As summarized in Table 1, the IMCs formed at Sn-20In-xZn/Cu interfaces at 150, 230, and 260 °C are either quaternary Cu–Sn compounds with some In and Zn contents or binary Cu–Zn compounds. As shown in Fig. 8b, the homogeneity ranges of Cu–Sn compounds in the Cu–In–Sn ternary system are generally in parallel with the Sn–In binary boundary, which indicates that In substitutes Sn in the ternary Cu–Sn compounds. Hence, when Cu–Sn compounds were the predominated reaction products, the combined Sn and In content in these quaternary Cu–Sn compounds can be considered together as the Sn content in Cu–Sn–Zn phase equilibria. On the other hand, when binary Cu–Zn compounds were the predominated reaction products, Sn and In do not participate in the reactions and thus the Cu–Sn–Zn phase equilibria can naturally be employed. Therefore, the Cu–Sn–Zn isothermal sections can be used for evaluating the Sn-20In-xZn/Cu reactions.

Fig. 8

a The schematic 250 °C isothermal tetrahedron of the Cu–In–Sn–Zn quaternary system; b the 250 °C isothermal section of the Cu–In–Sn ternary system [17]; and the isothermal section of the Cu–Sn–Zn ternary system at c 250 °C and d 180 °C [18] with diffusion paths superimposed

The diffusion paths of the L/S Sn-20In-xZn/Cu reactions are superimposed on the 250 °C isothermal section of the Cu–Sn–Zn ternary system as shown in Fig. 8c [18]. When the Zn doping level is 0.5 ≤ x ≤ 0.7, the Cu₆Sn₅ phase is the predominated reaction phase for reactions at both 230 and 260 °C and the diffusion path is liquid/Cu₆Sn₅/Cu, which is similar to the results of Sn-20In/Cu reactions [8, 9]. When the Zn doping level is 1.0 ≤ x ≤ 3.0 for reactions at 230 °C and 2.0 ≤ x ≤ 3.0 for reactions at 260 °C, the Cu₅Zn₈ phase is the predominated reaction phase and the diffusion path is liquid/Cu₅Zn₈/Cu. It is clear that the IMC formation was the results of the competition between Cu–Sn and Cu–Zn reactions; i.e., when Zn doping level is high, Cu–Zn reactions dominate and when Zn doping level is low, Cu–Sn reactions dominate. Between the two extremes, e.g., the Sn-20In-1.0Zn/Cu reactions at 260 °C shown in Fig. 2b, the supplement of Zn was not sufficient for the formation of the Cu₅Zn₈ phase and the local equilibrium at liquid/Cu interface was shifted to the CuZn phase, which has less Zn content, and the diffusion path became liquid/CuZn/Cu, as shown in Fig. 8c. For longer reaction times, upon the depletion of Zn in the molten solder and the formation of the CuZn phase at the interface, the local equilibrium at CuZn/Cu interface was broken. The Cu₆Sn₅ phase was formed via the reaction CuZn + Sn + In + Cu = Cu₆Sn₅ and resulted in the CuZn + Cu₆Sn₅ two-phase layer between the CuZn layer and Cu substrate, as shown in Fig. 2c–d.

It can be concluded that in the L/S Sn-20In-xZn/Cu reactions, the Cu₅Zn₈ phase is the reaction product when Zn doping level is high, the Cu₆Sn₅ phase is the reaction product when Zn doping level is low, and the transient CuZn phase forms when the Zn doping level is between the two critical values. These critical Zn doping levels depend on reaction temperatures as well as the chemical compositions of major contents of solders. The critical window of Zn doping levels that causes formation of the transient CuZn are around 1.0 < x < 2.0 for Sn-20In-xZn/Cu reactions at 230 °C, around 0.7 < x < 2.0 for Sn-20In-xZn/Cu reactions at 260 °C, around 0.5 < x < 2.0 for Sn-xZn/Cu reactions at 250 °C [21], and around 1.0 < x < 1.5 for Sn-3.5Ag-0.7Cu -xZn/Cu reactions at 260 °C [22]. Note that these critical windows are summarized based on a limited number of Zn doping levels being conducted in the literatures. The actual ranges of the critical windows may be smaller for individual cases.

Reaction mechanisms of the S/S Sn-20In-xZn/Cu reactions

The diffusion paths of the S/S Sn-20In-xZn/Cu reactions are superimposed on the 180 °C isothermal section of the Cu–Sn–Zn ternary system as shown in Fig. 8d [18]. In the Sn-20In-0.5Zn/Cu couples, the Cu₆Sn₅ phase is the only reaction product and the diffusion path is solder/Cu₆Sn₅/Cu, as shown in Fig. 8d. The Zn dopants all incorporated with the Cu–Sn reaction forming the Cu₆Sn₅ phase. When the Zn doping level is increased to 0.7 wt%, the initial diffusion path is solder/CuZn/Cu₆Sn₅/Cu, but it shifts to solder/Cu₆Sn₅/CuZn/Cu₆Sn₅/Cu and ends up becoming solder/Cu₆Sn₅/Cu as that in Sn-20In-0.5Zn/Cu couples, as shown in Fig. 8d. These results indicate that (1) the growth front of the Cu₆Sn₅ layer was at the solder side and Cu was the dominant diffusion species across the CuZn layer; and (2) the CuZn phase was not a stable interfacial compound and it transformed to the Cu₆Sn₅ phase due to the depletion of Zn in the solder matrix in Sn-20In-0.7Zn/Cu couples.

As presented in the “Results” section, there are five types of diffusion paths in the Sn-20In-1.0Zn/Cu couples as shown in Fig. 5, i.e., (i) solder/Cu₆Sn₅/Cu₅Zn₈/Cu, (ii) solder/Cu₅Zn₈/Cu₆Sn₅/Cu, (iii) solder/Cu₆Sn₅/Cu₅Zn₈/Cu₆Sn₅/Cu, (iv) solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu, and (v) solder/Cu₆Sn₅/Cu. According to morphological evolution at interfaces of the joints, the thin Cu₅Zn₈ layer is presumed to form first, so the Cu₅Zn₈/Cu contact could be found in type (i) regions. Owing to the formation of the Cu₅Zn₈ phase, the Zn content near the interface was depleted and the Cu₆Sn₅ phase formed subsequently. The growth of the Cu₆Sn₅ phase relied on the inter-diffusion of Cu from the substrate and Sn from the solder. Since the Cu₆Sn₅ phase was found in the solder side (type i), in the substrate side (type ii), and at both sides simultaneously (type iii) of the Cu₅Zn₈ layer, it was likely that the diffusion rates of Cu and Sn atoms were generally comparable and the growth fronts of the Cu₆Sn₅ layer was governed by the local diffusion pathways. The two Cu₆Sn₅ layers both grew thicker with longer reaction time; however, the Zn content in the Cu₆Sn₅ layers adjacent to the Cu substrate was higher (7 at.%), while that at the solder side was lower (3 at.%), indicating that Zn outflow from the Cu₅Zn₈ layer towards the Cu substrate participated in the growth of the Cu₆Sn₅ layers adjacent to the Cu substrate. Hence, the Cu₅Zn₈ phase gradually dissociated after longer reaction time. The joint would eventually become solder/Cu₆Sn₅/Cu at the interface of Sn-20In-1.0Zn/Cu couples reacted at 150 °C. Regardless the kinetics reasoning, the microstructural evolution can be comprehended with the diffusion path shifting from solder/Cu₆Sn₅/Cu₅Zn₈/Cu to solder/Cu₆Sn₅/Cu, as shown in Fig. 8d.

In the Sn-20In-2.0Zn/Cu couples, two types of diffusion paths in the Sn-20In-1.0Zn/Cu couples were found as shown in Fig. 6, i.e., (i) solder/Cu₅Zn₈/Cu₆Sn₅/Cu and (ii) solder/Cu₅Zn₈/solder/Cu₆Sn₅/Cu. The type (i) phase sequence is similar to one of the resultant morphologies in Sn-20In-1.0Zn/Cu couples reacted at 150 °C. Based on the discussion on Sn-20In-1.0Zn/Cu couples, it is likely the Cu₅Zn₈ phase also formed first between solder and Cu substrate. However, the Cu₆Sn₅ phase formed only at the substrate side of the Cu₅Zn₈ layer, indicating that Sn was the dominant diffusion species across the Cu₅Zn₈ layer, which acted as an effective diffusion barrier for Cu. However, the type (ii) phase sequence was rather peculiar, with the unreacted solder in the reaction diffusion zone between the two IMCs. The inflow of Sn and growth of the Cu₆Sn₅ phase in contact with Cu₅Zn₈ might have created stresses on the rigid Cu₅Zn₈ layer, or in the three-dimensional point of view, on the rigid Cu₅Zn₈ plan. The stresses upon the solder near the Cu₅Zn₈ plan would promote the inflow of mass towards the substrate. If the outflow of Cu was insufficient for Cu₆Sn₅ formation, Sn would then accumulate between the Cu₅Zn₈ and Cu₆Sn₅ phases. Therefore, the non-uniform interface with mixed type (i) and (ii) phase sequences would form. Although the diffusion path shifting in Sn-20In-1.0Zn/Cu and Sn-20In-2.0Zn/Cu couples is similar, as shown in Fig. 8d, comparing the results in these couples shows that thicker Cu₅Zn₈ layer formed in the early stage of reactions could stop the outward diffusion of Cu, as shown in Fig. 6, while Cu atoms could penetrate the thinner Cu₅Zn₈ layer, which would eventually dissociate within the Cu₆Sn₅ layer, as shown in Fig. 5.

In the Sn-20In-5.0Zn/Cu couples, the Cu₆Sn₅ phase no longer forms at the interface. The diffusion path is simply solder/Cu₅Zn₈/Cu, as shown in Fig. 8d. Not only Cu but also Sn was not able to penetrate the thick Cu₅Zn₈ layer. It is clear that the thickness of the Cu₅Zn₈ layer formed in the early stage of reactions is critical to the subsequent reactions and morphological evolution.

The interesting and sensitive interfacial morphology evolutions upon Zn doping levels were due to the two facts: (1) the high activity of Zn that Cu–Zn reactions dominate when Zn doping level is high; and (2) the limited supplements of Zn that depletion of Zn in solder breaks the local equilibrium between Cu–Zn compounds and Cu substrate and shifts diffusion paths. As shown in Fig. 8d, the diffusion path of Sn-20In-xZn/Cu reaction evolves with Zn doping levels. It is solder/Cu₆Sn₅/Cu or solder/Cu₅Zn₈/Cu without shifting when Zn doping level is low enough (x ≤ 0.5) or high enough (x ≥ 5.0), respectively. For medium Zn doping levels, transient Cu–Zn compounds form, such as CuZn (x = 1.0) and Cu₅Zn₈ (1.0 ≤ x ≤ 2.0), and both of them will end up transforming into the Cu₆Sn₅ phase due to the depletion of Zn in solders.

Kinetics of interfacial reactions in Sn-20In-xZn/Cu couples

Figure 9a shows the plot of IMC thickness as a function of Zn content in Sn-20In-xZn/Cu couples reacted at 230 and 260 °C for 2 h. Although the reactions at 230 and 260 °C are both L/S reactions, the IMC thicknesses are not always thicker in the couples reacted at higher temperature. It is because the type of IMC has changed due to the slight difference in temperature (30 °C). For the reaction conditions that the Cu₆Sn₅ phase is the predominated reaction phase, i.e., Sn-20In-0.5Zn/Cu and Sn-20In-0.7Zn/Cu couples reacted at 230 and 260 °C, the Cu₆Sn₅ layers were all thicker in the couples with higher Zn contents. However, the IMCs were thinner for even higher Zn doping levels when the predominated reaction phase was changed to the Cu₅Zn₈ phase; i.e., when the Zn doping level is higher than 1.0 wt% and 2.0 wt% for reactions at 230 and 260 °C, respectively, as summarized in Table 1. This is likely because the Cu₅Zn₈ phase was with approximately 65 at.%Zn and without detectable contents of Sn and In. Hence, it not only accommodates the segregated active Zn atoms without excess IMC growth, but also acts as a diffusion barrier between Cu and Sn or In. Nevertheless, the effects of Zn content upon the thicknesses of both Cu₆Sn₅ and Cu₅Zn₈ phases show the same trend; i.e., the higher the Zn contents were doped in the solders, the thicker the IMCs would be formed. The singular peak in Fig. 9a is the result of Sn-20In-1.0Zn/Cu reaction at 260 °C. Under such reaction condition, the predominated reaction phase is the CuZn phase. Although the CuZn phase contains about 40 at.%Zn, Sn and In also participate in the formation of the CuZn phase, leading to a faster IMC growth rate. During the Sn-20In-xZn/Cu soldering process, proper amount of Zn doping that causes formation of a minimum amount of the Cu₅Zn₈ phase is desired. In summary, as shown in Fig. 9a, Sn-20In-1.0Zn/Cu and Sn-20In-2.0Zn/Cu showed superior IMC growth rate at 230 and 260 °C, respectively.

Fig. 9

Plots of IMC thickness in the Sn-20In-xZn/Cu couples reacted at a 230 and 260 °C for 2 h and b 150 °C for 120, 240, and 600 h, respectively, as function of Zn content in wt%. 2 wt% was found to be the optimal Zn doping level for Sn-20In solders to bond with Cu substrate

Figure 9b shows the plot of IMC thickness as a function of Zn content in Sn-20In-xZn/Cu couples reacted at 150 °C for 120, 240, and 600 h. The thickness of the reaction layer formed depends strongly on the Zn content in the couples during solid-state thermal ageing. It is again ascribed to the type of IMC formation. As discussed in previous sections and summarized in Table 1, Cu₆Sn₅ and Cu₅Zn₈ phases were the only reaction phase for Sn-20In-0.5Zn/Cu and Sn-20In-5.0Zn/Cu couples, respectively. However, except for these two couples, the interfacial reactions in Sn-20In-0.7Zn/Cu, Sn-20In-1.0Zn/Cu, and Sn-20In-2.0Zn/Cu couples all involved transient stages with either the CuZn or Cu₅Zn₈ phase, which transformed to the Cu₆Sn₅ phase after prolonged reactions. As shown in Fig. 9b, with shorter reaction times of 120 and 240 h, the couple with the Cu₆Sn₅ phase as the only reaction phase, namely the Sn-20In-0.5Zn/Cu couple, showed the thickest IMC, while those involving Cu–Zn compounds had relatively thinner IMCs. The formation of Cu–Zn compounds suppressed the excess growth of the Cu₆Sn₅ phase. However, with prolonged reactions up to 600 h, the couple with the greatest Zn doping, namely the Sn-20In-5.0Zn/Cu couple, had the Cu₅Zn₈ phase as the only reaction product and showed drastically thicker IMC than those in the other couples with transient Cu–Zn IMC formation. It is likely that phase transition played an important role in mitigating IMC growth. In Sn-20In-0.7Zn/Cu couples, the CuZn phase formed first and transformed to the Cu₆Sn₅ phase with longer reaction times. However, it did not mitigate the interfacial reactions. As shown in Fig. 9b, among these couples with Cu–Zn IMCs, the 1.0 and 2.0 wt% Zn-doped solders showed the gentlest reactions with Cu. In these couples, the Cu₅Zn₈ phase formed first and detached from the Cu substrate due to the newly formed Cu₆Sn₅ phase and/or the inflow of Sn and In. Therefore, it can be concluded that in the Sn–In–Zn/Cu system, the Cu₅Zn₈ phase can act as the diffusion barrier and suppress the growth of the predominated Cu₆Sn₅ phase.

Conclusions

Zn content has a significant effect on the reaction phase formation and interfacial microstructural evolution in Sn-20In-xZn/Cu couples reacted at 150, 230, and 260 °C. The Cu₆Sn₅ phase forms in Sn-20In-xZn/Cu couples with the following Zn doping levels: x ≤ 0.7 for reactions at 230 and 260 °C and x = 0.5 for reactions at 150 °C, while the Cu₅Zn₈ phase forms in Sn-20In-xZn/Cu couples with the following Zn doping levels: 1.0 ≤ x ≤ 3.0 for reactions at 230 °C, 2.0 ≤ x ≤ 3.0 for reactions at 260 °C, and x = 5.0 for reactions at 150 °C. Between the two extremes, the transient CuZn phase forms in Sn-20In-xZn/Cu couples with x = 1.0 for reactions at 260 °C and x = 0.7 for reactions at 150 °C, while the transient Cu₅Zn₈ phase forms in Sn-20In-xZn/Cu couples with x = 1.0 and 2.0 for reactions at 150 °C.

In summary, 2.0 wt%Zn is found to be an optimal amount for addition to the Sn-20In solder that can result in transient formation of the Cu₅Zn₈ phase in solid-state ageing without forming the undesired CuZn phase during soldering, and the Cu₅Zn₈ phase can suppress the growth of the predominated Cu₆Sn₅ phase.