Cu and Ag additions affecting the solidification microstructure and tensile properties of Sn-Bi lead-free solder alloys

.


Introduction
During the last 10 years several alloys have been proposed as alternatives to eutectic or near-eutectic Sn-Pb electronic alloys. This has been done to avoid using lead (Pb) in the alloy due of its inherent toxicity [1,2]. Work into finding lead-free joints includes Sn-based alloys such as Sn-Bi [3], Sn-9Zn [4], Sn-10Sb [5], Sn-3.5Ag [6], Sn-0.7Cu [7] and Sn-3Ag-0.5Cu [8]. Some of these environmental-friendly solders are characterized by high melting temperatures (Sn-0.7Cu (227 o C), Sn-3.5Ag (221 o C) and Sn-3Ag-0.5Cu (217 o C)) which can damage the other electronic components attached to the printed circuit board (PCB) during reflow soldering operations.
Other drawbacks of the mentioned alloys involve void formation, large undercooling during solidification, rapid tendency for intermetallic compound (IMC) formation and spalling of interfacial IMCs during storage under moderate-to-high temperatures [9]. These problems are mainly caused by the high-Sn-content of the alloys. Therefore, it is crucial to find suitable low Sn content lead-free solder alloys and to also understand the inter-relations between the length-scale of the dendritic microstructures and the solidification thermal parameters (i.e. the growth rate (V L ) and the cooling rate (Ṫ L )).
Transient directional solidification (DS) techniques are very suitable for this kind of evaluation as comprehensive charts can be generated which establish the microstructure features-cooling rate type correlations. With the use of the DS technique, a single casting can be solidified enabling to access a large range of cooling rates, which can be further associated with a wide spectrum of microstructure sizes, morphologies and intermetallic particles either as part of the eutectic constituent or as the primary phase.
Sn-Bi lead-free solder alloys are considered alternatives that can provide relatively high tensile strength and good creep resistance [10]. The addition of small amounts of Ag is reported to decrease the brittleness of the Sn-58Bi solder as it leads to refinement of the microstructure [11]. One of the most promising low Sn-content 3 lead-free alloys is the Sn-35Bi-1Ag (SBA) alloy, as reported by Gain and Zhang [12]. It has been considered a suitable replacement for the Sn-37Pb alloy due to their comparable melting points [13]. According to Li and Hanna [13] the so-called SBA alloys are less prone to fail during electrochemical migration (ECM) than Sn-Ag-Cu solder alloys. As mentioned earlier this low melting temperature is highly desired as it can damage the electronic components during reflow soldering operations.
Although Sn-Bi solder alloys have high hardness due to the hard Bi phase, this also greatly reduces their ductility making them unsuitable for use [14,15]. The addition of third elements is a possible solution to improve this limited ductility [16,17].
Sakuyama et. al. [18] investigated the influence of 0.6wt% Cu/Ag additions on the microstructure and tensile properties of the eutectic Sn-57Bi alloys. It was found that the ductility of both the Sn-57Bi-Cu and Sn-57Bi-Ag alloys improved. Tensile tests were carried out at a strain rate of 2x10 -3 s -1 and after that the acquired tensile stress-strain diagrams were examined. These results revealed ductility values of 11%, 19% and 23% for the Sn-57Bi, Sn-57Bi-Ag and Sn-57Bi-Cu alloys, respectively. Improvement in ductility was attributed to the growth of ternary eutectic in combination with the fineness of the binary and the ternary eutectic constituents (either Sn+Bi+Cu 6 Sn 5 or Sn+Bi+Ag 3 Sn).
Despite the interest in lower Bi-content alloys, such as the mentioned Sn-35Bi-1Ag alloy, most of the investigations remain dedicated to examining the microstructure and the resultant mechanical properties for the near-eutectic or the eutectic compositions [19,20]. A number of studies examined microstructure, wettability and mechanical properties of Sn-Bi-based alloys, with and without small additions of alloys elements such as Ni, Cu, Zn and Ag [21][22][23][24].
According to Takao and co-authors [25] an optimized composition of Sn-Bi-Cu may improve the ductility since the Cu 6 Sn 5 fine particles are randomly distributed throughout the microstructure, even though their size was not mentioned in this research work. In this case, such particles have been recognized as part of a ternary 4 eutectic formed by Sn+Bi+Cu 6 Sn 5 . The ductility of the Sn-40wt%Bi-0.1wt%Cu alloy, for instance, is reported to be around 2.5 times higher than that characterizing the Sn-40wt%Bi-1.0wt%Cu alloy, when considering tensile tests that were performed at a strain rate of 1x10 -4 s -1 [25].
The effect of the solidification cooling rate (Ṫ L ) on the formation and evolution of the microstructure of ternary Sn-Bi-X alloys is still an open topic, especially if one aims to characterize the representative length scales of the formed microstructures. In the case of Sn-Bi alloys some of the main aspects are related to the length-scale of the microstructure observed in different magnified views, which include the secondary dendritic spacing ( 2 ) and the eutectic spacing E ). In the case of Sn-based solder alloys it is common to note the growth of dendritic arrays within the solder fillets. As such it is essential to characterize the scale of the dendritic spacing. Here a careful examination and quantification of the microstructures is required by using complementary microscopy techniques, such as scanning electron microscopy (SEM) and light microscopy (LM).
It is a common knowledge that the solidification thermal parameters (V L and Ṫ L ) control the final dendritic arrangement of a certain alloy [26][27][28][29][30]. Therefore it is essential to expand such knowledge for low-Bi Sn-Bi alloys to see how the final microstructure is influenced by the solidification cooling rate. This knowledge is important, as the formed microstructure will govern the mechanical properties of the solder. It is worth noting that there has been little research into understanding the mechanical behavior of these alloys with and without added elements.
The present study is focused on the systematic measurement of the main microstructural parameters characterizing directionally solidified Sn-Bi-X alloys. These parameters will be thoroughly correlated with the solidification cooling rate, with emphasis on three distinct cooling rate levels. The microstructures of the alloys will be discussed by focusing on the size, distribution and nature of the formed phases.
Ultimately the influence of alloy additions (Cu and Ag) and microstructure length-scale 5 parameters on the tensile strength and tensile ductility will be examined for Sn-Bi-X alloys.

Experimental procedure
Directionally solidified (DS) specimens of the Sn-34wt%Bi, Sn-34wt%Bi-0.1wt%Cu, Sn-34wt%Bi-0.7wt%Cu and Sn-33wt%Bi-2wt%Ag alloys were made using a water-cooled solidification setup to allow for the transient heat flow regime to be reached during solidification [31]. All Cu-containing, Ag-containing and non-modified Sn-Bi alloys were melted in-situ by radial electrical wiring heating the cylindrical stainless steel split mold. When melt temperatures reached ~20% above the liquidus temperature simultaneously the furnace coilings were disconnected and the external water flow at the bottom of the container was to begin the cooling down procedure. A number of J-type thermocouples were laterally placed along the length of the castings, with their tips placed in the center of the container so that the thermal profiles could be acquired. The alloys of interest were directionally solidified into a low-C (SAE 1020) steel bottom mold. The surface of the bottom-part molds has been finished with a 1200 grit SiC abrasive paper.
The recorded temperature-time data was used as the basis to experimentally determine the values of growth rate (V L ) and cooling rate (Ṫ L ) along the length of the DS castings. Experimental plots of position (P), from the water-cooled surface of the casting, and the corresponding time (t L ) of the liquidus isotherm passing by each thermocouple, i.e., P=f(t L ), gave rise to experimental fitting functions for each alloy examined. A time-derivative of these fitting functions was carried out so that the growth rate (V L ) could be calculated in the form of V L =f(t L ). By replacing t L =f(P) with t L inside the equation for V L , resultant equations of the form V L =f(P) have been obtained. The thermal data recorded immediately after the passage of the liquidus isotherm by each thermocouple was considered to compute the cooling rates (Ṫ L ) along the length of the castings. For each performed experiment at least 6 thermocouples were inserted along 6 the length of the casting, being strategically spaced between each other until they were 80mm from the cooled bottom of the casting. Post-mortem examination with regard to the exact positions of the thermocouple tips was carried out.
Longitudinal and transverse samples were extracted, for each DS Sn-Bi-X alloy casting, from the positions 5mm, 10mm, 15mm, 20mm, 30mm, 40mm, 50mm, 60mm, 70mm & 90mm from the water-cooled surface of the casting. These samples were then prepared using conventional metallography techniques. A light etching procedure with the solution of 92% (vol.) CH 3 OH, 5% (vol.) HNO 3 and 3% (vol.) HCl was applied for 10-30s, so that microstructural characteristics regarding the morphology and size of the -Sn matrix could be revealed. Also, such procedure was used to provide contrast between Bi-rich and Sn-rich phases when using scanning electron microscopy (SEM) and light microscopy (LM).
The linear intercept method [32,33] was applied to measure the secondary dendritic arm spacing (λ 2 ) on the longitudinal samples, the eutectic spacing on the transverse samples ( E ) and the spacing between Bi precipitates (λ p ) on the transverse samples. At least 40 measurements were performed for each selected position along the length of the DS castings.
The transverse specimens were prepared according to specifications of the ASTM Standard E 8M/04 and tested in an Instron 5500R machine at a strain rate of 1 × 10 -3 s -1 . In order to ensure reproducibility of the tensile results, three specimens were tested for each selected position. The ultimate tensile strength, the yield tensile strength and the elongation-to-fracture were determined for specimens extracted from different positions along the length of the castings, as can be seen in the sketch of

Results and discussion
The experimental evolutions of the solidification thermal parameters V L and Ṫ L can be seen in Fig. 2. This data for the four Sn-Bi alloys was plotted in a single set of graphics to help compare the V L and Ṫ L . While a single experimental trend was found to encompass all experimental cooling rate scatters of the four Sn-Bi alloys (see Fig.   8 2b), two tendencies seem to be necessary to represent the growth rate values along the length of the castings. In this case, a single experimental adjustment was adopted for the Cu-containing alloys.
Based on the distribution of the cooling rates, three distinct regimes were adopted as indicated in Fig. 2b. For the DS Sn-Bi-X alloys, a fast cooling down regime has been associated with 6.0<Ṫ L <12.0K/s, an intermediate regime with 0.6<Ṫ L <6.0K/s and slow cooling conditions with Ṫ L <0.6K/s. This parameterization of cooling rates will be suitably mentioned from here on through this research work. Takao et al. [25] observed, similarly, a very fine distribution of Cu 6 Sn 5 particles in the Sn-Bi eutectics of the Sn-40Bi-0.1Cu alloy. As can be seen in Fig. 5, the analysis of the Ag 3 Sn eutectic phase in the Sn-33wt%Bi-2wt%Ag alloy reveals the growth of coarser and continuous layers of such phase in the microstructure. Fig. 3, Fig. 4 and Fig. 5 show the presence of Bi precipitates which occurred due to the supersaturation of Bi atoms in the Sn phase, and thus enabled precipitation of this Bi phase during the cooling stage that occurs after solidification. In Fig. 7a Fig. 9a and Fig. 10a), yield tensile strength ( y : Fig. 9b and Fig. 10b) and elongation to fracture (Fig. 9c and Fig. 10c).
Considering the binary Sn-34wt.%Bi and the ternary Sn-34wt.%Bi-0.1wt.%Cu alloys, it can be seen in Fig. 9  As established in Fig. 7, the alloy containing 0.1wt.%Cu has been characterized by slightly higher eutectic and Bi precipitates spacings when compared with the binary Sn-34wt%Bi alloy. In contrast, lower λ 2 values were found for the Sn-34wt.%Bi-0.1wt.%Cu alloy which can be seen in Fig 6. It seems that the balance of these microstructural parameters observed at different magnifications, may be conducive to similar tensile properties as can be seen in Fig. 9. Overall the best balance of strength 17 and ductility is that obtained with 1/ 2 This difference is possible if it is considered that the addition of trace amounts of Cu may result in a very fine distribution of phases within the Sn-Bi-Cu eutectic phase.
Despite higher alloy Bi content, a fine eutectic structure is formed by alternate Bi-rich, Sn-rich and Cu 6 Sn 5 phases. Correspondingly an enhancement effect on the sliding of the interfaces between adjacent phases is attained due to the increase in contact area between the phases [34]. Such complex interaction may occur especially during necking, with provides a direct beneficial effect to the ductility.
As can be seen in Fig. 10 tensile properties were those of the Sn-34wt.%Bi-0.7wt.%Cu alloy, which is due to the relatively high  p and  E . Other important aspect that seems to reduce the tensile properties is the development of coarse and faceted Cu 6 Sn 5 IMCs in the microstructure (see Fig. 5/left images). According to a previous investigation [18], the addition of Cu in 18 Sn-Bi alloys can expand the region of solid-liquid coexistence, which may induce coarsening of the primary Cu 6 Sn 5 IMC crystals.
The Ag-modified alloy shows higher strength values inside the fast cooling zones of the graphics in Fig. 10, i.e., for 1/ 2 1/2 >0.27. This is due to the similar sizes of the eutectic and for the Bi precipitates if compared with those associated with higher cooling rate-samples of the binary Sn-34wt.%Bi solder alloy (gray lines in Fig. 10).  p values were 0.9m and 1.3m whereas  E were 1.8m and 1.6m for the Sn-34wt%Bi and Sn-33wt%Bi-2wt%Ag solder alloys, respectively. A more homogeneous distribution As a matter of fact, the Sn-34wt%Bi and Sn-34wt%Bi-0.1wt%Cu alloys surfaces show larger areas of ductile bulges, shown in Fig. 11a and Fig. 11b. This behavior can be associated with the lengthier stress-strain curves of these alloys, as shown in Fig.   8a. The prolonged shape of these curves indicates not only a higher fracture toughness but also higher ductility.
In contrast, the fracture surface of the Ag-containing Sn-Bi alloy is entirely brittle in nature. Here the coarse Ag 3 Sn IMCs seem to operate as favored points inducing the occurrence of premature fissures during tensile loading of the specimen. It is worth noting in Fig. 12   Sn-33wt%Bi-2wt%Ag solder alloys emphasizing the fractures that occurred from the primary intermetallic particles.