Wetting and interfacial reactions: Experimental study of the Sb-Sn-X (X = Cu, Ni) systems

Experimental studies of the Cu-Sb-Sn and Ni-Sb-Sn systems have been carried out by the wetting tests, followed by the analysis of the microstructural evolution occurring at the interface between the liquid alloy and solid substrate. The wetting experiments on the Sb30Sn70 / (Cu, Ni) and Sb38.4Sn61.6 / (Cu, Ni) systems have been performed by using a sessile drop apparatus. The wetting behaviour of the two alloys in contact with Cu-substrate differs from that observed in the case of Ni-substrate. The Sb-Sn alloy / substrate interface was characterised by SEM-EDS analyses. For each system, the solidliquid interactions and the phases formed at the interface were studied with the help of the corresponding phase diagrams.

The COST Action MP0602 "Advanced Solder Materials for High Temperature Application" (HISOLD) has been dedicated to the investigation and design of high temperature lead free solders and the results obtained are reported in [11][12][13]. Another important application of Sb-Sn alloys and/or Sb-Sn based composites is related to their use as anodic materials for Li-batteries. Indeed, high Li-storage capacity and good cycling stability of Sb-Sn alloy powders make them suitable for rechargeable Libatteries [14][15][16][17]. Among Sb-Sn alloy compositions, it seems that the use of SbSn intermediate phase at the equiatomic composition is beneficial during lithiation due to its separation into the more lithiated and/or not lithiated phases helping to relieve mechanical strain [15]. Although the investigation of the Sb-Sn system started more than one hundred years ago [18], until nowadays at least twenty assessments of its phase diagram are available [19][20][21][22][23][24][25][26][27][28][29][30][31][32], and even after recent experimental studies [31,32] some questions concerning the existence of a Sb 3 Sn 4 -SbSn two-phase field and of ordering phenomena in the SbSn phase are still open.
Thanks to long lasting experience related to the investigation of lead-free solders by wetting experiments [1,8,33,35], we can state that the microstructure evolution after wetting tests is qualitatively comparable to that of a real solder joint [36,37]. Therefore, the interfacial reactions between liquid alloys and solid substrates and the phases subsequently formed at the interface, identified by SEM-EDS analyses, can be analysed by means of the corresponding phase diagrams. On the other hand, lacking direct thermodynamic measurements as in the case of high melting temperature systems, the data describing the solid-liquid interface formed during the wetting tests can be used as the input to calculate the phase equilibria [38]. The wetting behaviour of any liquid in contact with a solid surface is expressed by the contact angle Ɵ that the liquid drop forms on the solid substrate. Another important parameter is the work of adhesion W A that depends on the bonding characteristics of both, the liquid and the solid phase as well as on the interactive forces at their interface formed during wetting tests [39]. As already mentioned above, there are lot of literature data describing the wetting of Sn-rich alloys in contact with Cu-substrate and only a few datasets for Nisubstrate [1]. On the contrary, concerning intermediate alloy compositions, the wetting behaviour of the Sb-Sn / (Cu, Ni) systems has not yet been investigated.
In the present work, the Sb 30 Sn 70 and Sb 38.4 Sn 61.6 (at %) alloys on Cu and Ni-substrates were used for wetting experiments that were carried out by the sessile drop method. Subsequently, SEM-EDS analyses were done and the results obtained were analysed in terms of the Sb-Sn-X (X = Cu, Ni) phase diagrams.

Materials
The Sb 30 Sn 70 and Sb 38.4 Sn 61.6 (in at % ) alloys were prepared from pure Sn bar (Newmet Koch, 99.9999 wt %) and Sb ingots (Newmet Koch, 99.999 wt %). After mechanical cleaning proper amounts of Sb and Sn were weighted with accuracy of  0.05 mg, encapsulated inside a quartz tube, sealed under Argon atmosphere, and induction melted for three times. The homogenization was achieved by shaking at different speeds. Subsequently the Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys were water quenched. All ingots were checked by SEM-EDS analyses to ensure proper composition and uniform structure and subsequently were cut into small samples of around 1 g.

Wetting tests
The sessile drop method has been applied to evaluate the wetting behaviour of the Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys on Cu and Ni-substrates. A standardized experimental procedure has been described in detail for other alloys of the same system [1].
The contact angles were measured using Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloy samples, previously mechanically cleaned by scratching, chemically rinsed with pure ethanol in an ultrasonic bath, introduced at the centre of the chamber and placed on Cu or Ni-substrate (both with a purity of 99.9999 %, square plate, 13x13x1 mm). To reach a surface roughness of R a~ 0.05 µm, Cu and Ni-substrates were previously metallographically mirror polished. Before the test, the furnace was heated and degassed under vacuum (P = 10 -4 Pa), then pure Argon (ArN60) was introduced. The resulting oxygen pressure was around an average value of p o2 ≤ 10 -6 Pa. The isothermal measurements were carried out at different temperatures over the interval T liq + 50 K-900 K ( is the liquidus temperature). During experiments, the temperature was measured by a K-type thermocouple and kept constant within ± 2 K and the drop profile was acquired by a CCD camera with a precision of ± 1μm. The image was processed in LABview environment using home-made A.S.T.R.A.view® software [40] and the contact angle data were evaluated with an accuracy of ± 0.5°. The left and right contact angle values differ less than 3 % and the contact angle data given in Table 1 represent an average value of the two values.
After wetting experiments, the drop / substrate couples were cooled down to room temperature and the interfacial microstructure of each solidified sample was characterized by optical microscopy and SEM-EDS analyses on the cross-section mirror polished.

Results and discussion
Until now, only Sn-rich alloys containing up to 15 at % of Sb have been studied aiming to develop a series of new high temperature lead-free solders. Indeed, the wetting behaviour of a liquid Sb-Sn / solid Cu system has been studied in [1,2,4,[41][42], while in the case of Ni-substrate only two datasets were reported [1]. The aim of the present research is twofold: to determine the wetting behaviour of abovementioned systems and to substantiate the phase equilibria determination in the Cu-Sb-Sn and Ni-Sb-Sn ternary systems. Therefore, the wetting properties of the Sb 30 Sn 70 / (Cu, Ni) and Sb 38.4 Sn 61.6 / (Cu, Ni) systems were determined and the test parameters such as the initial (T 0 ) and the final (T f ) temperatures of the experiments together with the corresponding contact angles (Ɵ 0 ) and (Ɵ f ) respectively, and the duration of the experiments conducted under isothermal conditions (t f ) are given in Table 1. The liquidus temperatures (T liq ) of the alloys studied refer to the Sb-Sn phase diagram [31].
The microstructural characterisation of the Sb-Sn/(Cu, Ni) systems investigated has been done by SEM-EDS and the results obtained were analysed in view of the corresponding binary and ternary phase diagrams. The reaction layers formed at the interface between liquid Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys in contact with the two substrates have irregular morphology. In the case of Cu-substrate, a characteristic so-called "scallop" morphology is  particularly pronounced as it was already observed for various Sn-rich solders [1,8,43,44]. The estimation of layer thickness by means of SEM-EDS analyses is related to some errors even if the variations of thickness are very small. Higher uncertainties are introduced to estimate the thickness of interfacial layer with completely irregular morphology such as that of the scallop type. The thickness of each layer (Table 2) represents a mean value calculated from real thickness values measured in different points along a cross section, as it was reported in previous works [1,8].

Wetting behaviour of Sn-Sb / Cu systems
The contact angle values of 43 and 29° were determined for liquid Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys respectively, in contact with solid copper (Table 1), indicating a good wettability of Cu-substrates. Neither there are available data on the wetting properties for the alloys investigated here nor for the alloys with similar compositions and therefore the results obtained cannot be compared, but the effect of Sb on their wetting behaviour can be analysed. For the experimental conditions applied, the contact angle values of the two alloy compositions indicate that higher Sb-content in the alloy results in an improvement in wetting. The same was deduced for Sb-Sn solders tested under similar conditions [1]. On the contrary, under different experimental conditions, some wetting tests performed on Sb-Sn solders showed that an increase in Sn promotes the wetting on Cu-substrate [2]. Sb-Sn melts wet and react with Cusubstrate arriving at low contact angles, but with significant modification of the initial metal-metal interface. The wetting was observed after alloy melting, and it was mainly driven by the high dissolutive reactivity of the Sb-Sn/Cu system and subsequent formation of a scallop type interface [1].
Interfacial reactions and the phases formed at the interface of the Sb 30 Sn 70 / Cu and Sb 38.4 Sn 61.6 / Cu systems are similar to those observed for Sn-rich alloys on Cu-substrates [1,8]. In both cases, Sn/Cu couples can simply describe the main reaction products [44] that are defined by the Cu-Sn phase diagram [26,45]. Indeed, the Cu-Sn system is characterised by the presence of the two energetically favoured phases, i.e. -Cu 3 Sn and -Cu 6 Sn 5 with narrow solubility ranges, i.e. 23.5 ≤ Sn (at %) ≤ 27.1 and 43.5 ≤ Sn (at %) ≤ 45.5, respectively. Usually the -Cu 3 Sn phase was experimentally found adjacent to Cu-substrate, while the -Cu 6 Sn 5 was adjacent to the solder [1,8,33,34,44]. Comparing the two phases -Cu 6 Sn 5 results more stable [46]. The experimental values of the growth-rate coefficients at T = 453 K are k ε = 4.40•10 -9 and k η = 7.76•10 -9 m•s -0.5 [47] and the growth of the two phases is diffusion controlled obeying the parabolic law [48]. Based on it, an increase in reaction time leads to a more pronounced growth of -Cu 6 Sn 5 with respect to that of -Cu 3 Sn [44,49]. The solid-liquid interaction phenomena observed during the wetting tests and the phases formed at the interface identified by SEM-EDS analyses can be interpreted in terms of the phase equilibria, taking into account the recent assessment of the Cu-Sb-Sn phase diagram [50][51][52]. The phase equilibria in the Cu-Sb-Sn system were studied by [50], using the samples equilibrated for long times, i.e. Sn-rich alloys were annealed for about 2 months at T = 523 K. Differential scanning calorimetry (DSC) and Electron Probe Micro Analyses (EPMA) were employed to study the Cu-Sn-Sb system along the Sn:Sb=1:1 isopleth and to identify five invariant ternary reactions [51,52]. Among the primary solidification phases in the Cu-Sb-Sn liquidus projections, the -Cu 3 Sn, -Cu 6 Sn 5 , (Sn), Sb 2 Sn 3 , SbSn, Sb 4 Sn 3 and (-Cu 3 Sb, Cu 4 Sn) were also detected at the interfaces of the Sb-Sn/Cu systems investigated by the wetting tests. The present results

R.M. Novaković et al. / JMM 54 (2) B (2018) 251 -260
253 suggest that an increase in Sb-content leads to a decrease in the thickness of the interface layer formed between the Sb-Sn alloy and Cu-substrate ( Table 2), indicating that Sb is not actively involved in the interfacial reactions, as it was already observed [1,53]. The solubility of Sb in Cu 6 Sn 5 phase up to 6.5 reported in [54] differs significantly from that of 11 (both in at %) obtained from the equilibria studies of the Cu-Sb-Sn system at T=533 K [50]. After the last isothermal steps of wetting tests ( Until now, no available experimental data on the Cu-Sb-Sn phases formed at these temperatures.

Sb 30 Sn 70 / Cu
The contact angle variations have been determined under isothermal conditions at four temperatures (640, 690, 745, 795 K) and for times of about 10  2 min at each temperature (Fig. 1). During heating up to T = 640  1 K, at the beginning of the first isothermal cycle, the contact angle value drop down from Ɵ = 130° to Ɵ = 48°, observed, after 12 min. A further decrease in the contact angle values, i.e. from Ɵ = 48°t o Ɵ f = 43° was observed at T = 795 K after 60 min, when the last isothermal cycle was completed ( Fig. 1; Table 1).
The solidified Sb 30 Sn 70 / Cu sample was metallographically prepared and the cross section was characterised by SEM-EDS analyses. As in the case of Sb-Sn solders [1], Cu-solubility in the liquid alloy and subsequent dissolution of Cu-substrate determine the interfacial microstructure of the Sb 30 Sn 70 / Cu system. Significant dissolution occurs over the whole interface area and the formation of the crater with a maximum depth of about 40 m under the drop was observed (Fig. 2a). Because of this, the final value of the contact angle, Ɵ f = 43° can be considered as an "apparent" contact angle [55].
As it is shown in Fig. 2a and Fig. 2b (an enlarged interface area), the formation of two reaction layers was observed. Close to Cu-substrate an interface layer of the Cu-2.4Sb-24.4Sn phase (30 m) having the composition close to -Cu 3 Sn was detected and adjacent to this layer, the Cu-14.4Sb-47.5Sn phase (18 m), with composition close to -Cu 6 Sn 5 , was identified ( Fig. 2b; Table 2). Taking into account that the solubility of Sb in -Cu 6 Sn 5 increases with temperature, the value of 14.4 at % found at T= 795 K ( Table 2) might be considered comparable to that of 11 at % determined at T= 533 K [50]. As it was already observed for Sn-rich alloys in contact with Cu-substrates [1,8], the thickness of both scallop type interfacial layers is not uniform. The brightest regions in Fig. 2a and Fig. 2b correspond to the Sb-(56.1 ± 0.4) Sn phases indicating the presence of SbSn solid solutions [31,32].
The analysis of the solidified drop border composition near Cu-substrate (Fig. 2c) indicates the presence of phases based on  -Cu 3 Sn with an average composition of Cu-6.5Sb-19Sn. Due to a strong exothermic reaction that occurs during wetting experiments on Cu-substrate, small areas and spots having the compositions of Cu-(3.5 ± 0.5)Sb-(13.8 ± 2) Sn were found. On the top drop Sb 3 Sn 4 based phases having the composition of Cu-(40.5 ± 1.0)Sb-(57.6 ± 1.0) Sn and containing up to 2.2 at % Cu were identified (Fig. 2c). The alloys based on  -Cu 3 Sn and -Cu 6 Sn 5 phases together with the 25.7Sb-74.3Sn are slightly different from the nominal alloy composition, i.e. the Sb-70Sn as well as Sn-rich alloy compositions containing up to 6 at % of Cu were also detected on the top drop (Fig. 2d).

Sb 38.4 Sn 61.6 / Cu
The contact angle variations have been determined under isothermal conditions at two temperatures (695, 785 K) and for times of about 10  2 min at each temperature (Fig. 3). During heating up to T = 695  1 K, at the beginning of the first isothermal cycle, the contact angle values drop down from 126 to Ɵ = 44°, observed after 11 min. The contact angle further decreased from Ɵ = 44° to Ɵ = 29°, observed at T = 785 K after 38 min, when the last isothermal cycle was completed ( Fig. 3; Table 1).
As in the case of the Sb 30 Sn 70 / Cu dissolutive wetting was observed and the formation of a groove with a maximum depth of about 20 m under the drop (Fig. 4a) indicates that the contact angle of Ɵ f = 29°i s the "apparent" one ( Table 2). The two reaction layers formed at the Sb 38.4 Sn 61.6 / Cu interface were detected by SEM-EDS analyses ( Fig. 4a and Fig. 4b).
Close to Cu-substrate, an interface layer based on  -Cu 3 Sn having the composition of Cu-5.5Sb-21.2Sn (25 m), and the second layer with the composition of  Table 2). The last one is based on -Cu 6 Sn 5 , and at T=785 K it has significant Sb-solubility of 19.4 at %, that might be comparable to that of 11 at % observed at T= 533 K by Chen et al. [50]. Both interfacial layers have a non-uniform thickness related to characteristic scallop type morphology (Fig. 4b) described everywhere [1,8,44]. The brightest regions in Fig. 4a and Fig. 4b correspond to the Sb-(50.4 ± 0.6)Sn phases, identified on the Sb-Sn phase diagram as SbSn solid solutions [31,32]. The analysis of the solidified drop border composition close to Cu-substrate (Fig. 4c)     solid solutions) containing up to 1.9 at % Cu (Fig. 4c), and Sb 3 Sn 4 based phase with the composition of 42.1Sb-57.9Sn (Fig. 4d) were identified.

Wetting behaviour of Sn-Sb / Ni systems
The contact angles of 37 and 46° were determined for liquid Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys respectively, in contact with solid nickel (Table 1), indicating a good wettability of Ni-substrates. The spreading behaviour of the two Sb-Sn alloys investigated is similar to that observed for Sn-rich solder alloys in contact with Ni-substrate [1,8,35] indicating that the contact angle decreases in a stepwise manner with increasing temperature, as observed in [1]. For the alloy compositions investigated, no available literature data for a comparison. Due to some experimental problems, the wetting test on the Sb 38.4 Sn 61.6 / Ni was carried out under isothermal conditions at only one temperature, which was about 100 K lower in comparison to other operating temperatures, and for very short time with respect of the other wetting tests given in Table 1. Therefore, the effect of Sb-content on the wetting properties of the systems investigated cannot be deduced.
Significant difference in wetting behaviour of liquid Sb-Sn alloys on Ni-and Cu-substrates is related to a lower dissolution rate of solid Ni [56] controlled by diffusion in the liquid phase in comparison to that of solid Cu [57]. On the other side, the diffusivity data [56,57] indicates that Ni atoms diffuse slower into Snrich melts with respect to those of Cu and, in the case of the Sb-Sn/Ni interfaces, the intermetallic compounds with lower thickness were formed   alloys in contact with Ni-substrates the macroscopically planar interfaces consisting of thin reaction layers were observed ( Fig. 6a and Fig. 6b). Sn/Ni couples describe the main reaction products [44,58] and the phases detected by SEM-EDS (Table  2) were analysed in terms of the Ni-Sn [59], Sb-Sn [31,32] and Ni-Sb-Sn phase diagrams [60,61]. The information on the Ni-Sb-Sn system were summarised in [61] and until now they are not yet completed. The present results obtained at temperatures of 790 and 672 K can be helpful to determine the equilibrium phases at the corresponding isothermal sections.

Sb 30 Sn 70 / Ni
The min at each temperature ( Fig. 5; Table 1). During heating up to T = 650  1 K, the initial contact angle was about 127° and it decreased gradually to Ɵ = 70°m easured at the beginning of the first isothermal cycle at 650 K. A further decrease to Ɵ = 57° was observed after 10 min, followed by a stepwise decrease to Ɵ = 37° at T = 790 K after 56 min, when the last isothermal cycle was completed ( Fig. 5; Table 1).
The solidified Sb 30 Sn 70 / Ni sample was metallographically prepared and its cross section was characterised by SEM-EDS analyses. At thr interface one reaction layer with the composition of Ni-17.0Sb-10.7Sn having irregular morphology was observed. The Ni-17.0Sb-10.7Sn is based on the Ni 3 Sn phase, which has been found at 773 K in the samples subjected to isothermal solidified without quenching [62]. Near the interface the Ni-34.7Sb-18.5Sn phase was detected (Fig. 6a).
Similar composition described by Ni-35.2Sb-19.3Sn was also found. These phases belong to continuous solid solutions that exist between the two stoichiometric NiSb and Ni 3 Sn 2 compositions, and are denominated as ζ -phase [1,61]. Increasing the distance from the interface, the SbSn intermediate compositions of (42.1 ± 1.0)Sb-(57.9 ± 1.0)Sn, 9.1Sb-90.9Sn and 40.6Sb-59.3Sn (Sb 2 Sn 3 ) respectively, were identified (Fig. 6a). An enlarged interface area is shown in Fig. 6b. A very thin irregular interfacial layer (< 6 m) of Ni-18.2Sb-9.7Sn phase, which is based on Ni 3 Sn, and close to the interface, Ni-34.1Sb-19.6Sn solid solution known as ζ -phase (10 m) were found ( Fig. 6b; Table 2). The final operating temperature of 790 K (Table 1) and the formation of ζ -phase are substantiated by the studies reported in [12,60,61]. Near the interface, Sb-56.0Sn (SbSnphase) was detected, while the two Sn-rich compositions, i.e. 15.6Sb-84.4Sn and 9.0Sb-91.0Sn, were identified as bulk phases (Fig. 6b). The presence of Ni-28.9Sb-23.2Sn and Ni-33.4Sb-18.5Sn alloy compositions (ζ -phases) on the solidified drop border with Ni-substrate and in the vicinity of the interface, respectively, was detected. On the top drop Ni-43.2Sb-52.4Sn (SbSn based phase) and Ni-57.1Sn-39.0Sb (Sb 2 Sn 3 based phase) were identified (Fig. 6c), in agreement with [12,61]. Grytsiv et al. [60] reported negligible solubility of Ni in SbSn phase, while significant Ni-solubility was taken into account to reproduce invariant reactions in the temperature range between 523 and 673 K [61]. SEM-EDS data obtained after the wetting tests at 790 K indicated that about 4.4 at % Ni was dissolved in the SbSn phase (Fig. 6c).

Sb 38.4 Sn 61.6 / Ni
Due to some problems related to the melting, the alloy sample has been heated under non-isothermal   (Fig. 7). After the melting of sample, the initial contact angle was about 143°, it decreased monotonically to Ɵ = 80° at the beginning of the isothermal cycle at T = 672 K and after 8 min, the contact angle Ɵ = 46° was measured ( Fig. 7; Table 1). The Sb 38.4 Sn 61.6 / Ni interface was characterised using SEM and EDS analyses. The interface between the Sb 38.4 Sn 61.6 alloy and Ni-substrate remains nearly macroscopically planar. Thin interfacial layer of Ni-46.0Sb-31.7Sn (< 4 m) and close to it, the phase of similar composition, i.e. Ni-48.0Sb-29.5Sn were detected (Fig. 8a).
The presence of a novel ternary phase, denominated as t -phase, was found in a nonequilibrium four-phase mixture with binary phases [60]. Therefore, the two aforementioned phases shown in Fig. 8a belong to the Sb-Sn-NiSb-Ni 3 Sn 2 system, and they can also be termed as t-phase. The 44.2Sb-55.8Sn (SbSn phase) was identified in the bulk drop (Fig. 8a). t-phase, with the composition of Ni-30.7Sb-29.6Sn was also detected on the solidified drop border close to Ni-substrate (Fig. 8b).

Conclusions
The sessile drop method has been used to determine the wetting properties of liquid Sb 30 Sn 70 and Sb 38.4 Sn 61.6 alloys on Cu and Ni-substrates. The two Sb-Sn alloys exhibit a good wetting behaviour characterised by the contact angle values varying between 30 and 45°, respectively, similarly to those measured for Sn-rich solders. After solidification, in the case of Cu-substrates, the microstructural characterisation performed by SEM-EDS analyses indicates the formation of two reaction layers based on the -Cu 3 Sn and -Cu 6 Sn 5 intermetallic phases, while in the Sb-Sn / Ni systems, Ni 3 Sn based interfacial layer was found. For the last mentioned systems, after short time wetting experiments, the presence of t-phase at the interface was detected. The interfacial layers grown on Cu substrates are thicker with respect to the layers of Ni 3 Sn 4 or t-phase formed on Ni substrates due to faster diffusion rate of copper in liquid Sn. For the systems investigated, an increase in temperature and/or reaction time increases the thickness of interfacial layer and the interfacial reaction kinetics controls the growth of intermetallic compounds. The experimental data on the solid / liquid interactions occurring at the wetting temperature together with the data on subsequent phases formed at the interface can be used to calculate the phase equilibria in the systems tested.