Interfacial Reactivity of the Filled Skutterudite Sm y (Fe x Ni 1 − x ) 4 Sb 12 in Contact with Liquid In-Based Alloys and Sn

: The study of the wettability of thermoelectric materials, as well as the search for the most proper brazing alloys, is of the maximum importance to get one step closer to the realization of a thermoelectric device. In this work, a wettability study of the ﬁlled skutterudite Sm y (Fe x Ni 1 − x ) 4 Sb 12 by Sn and In-based alloys is presented. Samples, having both p - and n - characters were prepared by the conventional melting-quenching-annealing technique and subsequently densiﬁed by spark plasma sintering (SPS). Afterward, wettability tests were performed by the sessile drop method at 773 K for 20 min. Scanning electron microscopy-energy dispersive spectroscopy (SEM-EDS) analyses performed on the cross-section of the solidiﬁed drops suggest quite a complicated scenario due to the coexistence and the interaction of a large number of di ﬀ erent elements in each analyzed system. Indeed, the indication of a strong reaction of In-based alloys with skutterudite, accompanied by the formation of the InSb intermetallic compound, is clear; on the contrary, Sn exhibits a milder reactivity, and thus, a more promising behavior, being its appreciable wettability, whilst coupled to a limited reactivity.


Introduction
Nowadays, the general lack of energy and its always increasing demand leads us to explore different fields and technologies with the aim of searching for new energy sources or energy-saving pathways. As a response, thermoelectricity is an attractive effect based on the ability of materials to directly convert thermal energy into electrical power. The thermoelectric approach can be employed in different fields dealing with fundamental problems of sustainability and eco-environmental compatibility. Thermoelectric generators (TEGs) can be used in every circumstance where it is essential to produce energy in small volumes and with neither moving parts nor working fluids [1]; alternatively, they can be coupled to traditional energy production technologies in order to recover waste heat where there is a lack of conversion efficiency [2,3]. Moreover, energy harvesting is of the primary importance for the development of the Internet of Things [4,5].

Preparation of the Brazing Alloys
Wettability of skutterudites was tested using two different alloys-an In-based alloy (80% In and 20% eutectic alloy Ag 0.62 Cu 0.38 , hereafter named AgCuIn), and pure Sn (melting point = 504.9 K, purity: 99.9999%, Goodfellow, Huntingdon, UK). Before the wetting tests, the AgCuIn alloy was prepared by combining appropriate amounts of In (purity: 99.9999%, Goodfellow, Huntingdon, UK) and eutectic AgCu (purity: 99.95%) and melting them in an arc melting furnace. The alloy was melted five times in order to ensure compositional homogeneity. Arc melting was carried out under an Ar atmosphere; before this procedure, a small drop of Zr was melted with the purpose of pick-up any residual oxygen. A weight loss of 0.001-0.002 g was quantified after the preparation, which indicates that evaporation of the molten alloy can be neglected. Both substrates and brazing alloys were carefully cleaned in ethanol using an ultrasonic machine before wetting tests.

Sessile Drop Experiments
Sessile drop wetting tests were carried out in a tubular alumina furnace (T max~1 800 K), fully described in [59], equipped with the ad hoc designed ASTRAView image analysis software (version 2006, running on NI-Labview environment, developed at CNR-ICMATE, Genoa, Italy), which allows obtaining contact angles and drop dimensions during each experimental run. Temperature is read by a type S thermocouple, located just underneath the test sample, which was previously calibrated using Metals 2020, 10, 364 4 of 13 an Sn piece located in the same position as the test specimen. The precision of the temperature data can be assessed in ±5 K. A magnetic manipulator allows samples to be introduced into the furnace once temperature and environmental conditions reached steady values. The whole experimental system is prudently aligned, and the contours of drop and substrate are recorded as back-lit images, using a high-resolution CCD camera (Foculus F0531B, Net-GMBH, Finning, Germany). Drops profiles are instantly acquired and then off-line elaborated. The intrinsic precision of contact angle data measured by the software is in the order of ±0.5 • .
For the tests described here, substrates were previously micrographically polished and weighed before and after the tests. Tests were performed under a vacuum of at least 5.0 × 10 −4 Pa at 773 K. The substrate/alloy couples were introduced into the hot zone of the furnace, and after 1200 s, samples were extracted from the hot region, moved to the cold sector, and cooled down to room temperature. The weight loss during the wetting tests resulted in being below 1%.
After the tests, solidified drops and micrographically polished surfaces of the cross-sectioned samples were observed and analyzed by SEM-EDS. Table 1 reports an overview of the experimental compositions of skutterudites as obtained from EDS analyses; all the samples present a composition which is very close to the nominal one [44]. Moreover, the small amounts of extra phases recognizable in annealed samples [Sb, (Fe,Ni)SmSb 3 , SmSb 2 , FeSb 2 ] cannot be detected in dense samples, meaning that the SPS treatment also contributes in improving phase homogeneity. SEM microphotographs taken on Fe100 skutterudites before and after SPS are reported in Figure 1, as representative examples. The densification treatment also induces a significant pore reduction, which reflects in a density value exceeding 92% for all the compositions [44]. experimental system is prudently aligned, and the contours of drop and substrate are recorded as back-lit images, using a high-resolution CCD camera (Foculus F0531B, Net-GMBH, Finning, Germany). Drops profiles are instantly acquired and then off-line elaborated. The intrinsic precision of contact angle data measured by the software is in the order of ±0.5°. For the tests described here, substrates were previously micrographically polished and weighed before and after the tests. Tests were performed under a vacuum of at least 5.0 × 10 −4 Pa at 773 K. The substrate/alloy couples were introduced into the hot zone of the furnace, and after 1200 s, samples were extracted from the hot region, moved to the cold sector, and cooled down to room temperature. The weight loss during the wetting tests resulted in being below 1%.

Compositional and Morphological Characterization of Skutterudites
After the tests, solidified drops and micrographically polished surfaces of the cross-sectioned samples were observed and analyzed by SEM-EDS. Table 1 reports an overview of the experimental compositions of skutterudites as obtained from EDS analyses; all the samples present a composition which is very close to the nominal one [44]. Moreover, the small amounts of extra phases recognizable in annealed samples [Sb, (Fe,Ni)SmSb3, SmSb2, FeSb2] cannot be detected in dense samples, meaning that the SPS treatment also contributes in improving phase homogeneity. SEM microphotographs taken on Fe100 skutterudites before and after SPS are reported in Figure 1, as representative examples. The densification treatment also induces a significant pore reduction, which reflects in a density value exceeding 92% for all the compositions [44].

AgCuIn Alloy
The In-based alloy in contact with skutterudites started upon melting to wet and spread over the surface (see Figure 2); a strong reactivity between the alloy and the substrate could be detected. The occurrence of remarkable amounts of solid phases during the test made it impossible to report a

AgCuIn Alloy
The In-based alloy in contact with skutterudites started upon melting to wet and spread over the surface (see Figure 2); a strong reactivity between the alloy and the substrate could be detected. The occurrence of remarkable amounts of solid phases during the test made it impossible to report a final contact angle. The strong interaction can be easily observed in Figure 2d, where the skutterudite substrate appears even deformed due to intense reactivity. This behavior is roughly the same whatever the Fe amount in the skutterudite.
Metals 2020, 10, x FOR PEER REVIEW 5 of 13 final contact angle. The strong interaction can be easily observed in Figure 2d, where the skutterudite substrate appears even deformed due to intense reactivity. This behavior is roughly the same whatever the Fe amount in the skutterudite.  Figure 3 shows the SEM pictures of the solidified drops deriving from wetting tests performed with AgCuIn. It can be easily recognized that reactivity was strong, because the classical shape of a solidified drop is not recognizable, and substrates appear heavily damaged.    Figure 3 shows the SEM pictures of the solidified drops deriving from wetting tests performed with AgCuIn. It can be easily recognized that reactivity was strong, because the classical shape of a solidified drop is not recognizable, and substrates appear heavily damaged. final contact angle. The strong interaction can be easily observed in Figure 2d, where the skutterudite substrate appears even deformed due to intense reactivity. This behavior is roughly the same whatever the Fe amount in the skutterudite.  Figure 3 shows the SEM pictures of the solidified drops deriving from wetting tests performed with AgCuIn. It can be easily recognized that reactivity was strong, because the classical shape of a solidified drop is not recognizable, and substrates appear heavily damaged.    Coming to microchemistry, as shown in Table 2, the scenario is quite complicated by the presence of a considerable number of elements; the most relevant feature is the formation of the InSb phase, which was found all over the drop interspersed in a solidified phase formed of Fe, Sb, Sm, Ag, Cu, In. At the skutterudite/solidified drop interface, an infiltrated zone was observed, which was formed of the same phases. Similar reactive microstructures were observed for all the skutterudites in contact with AgCuIn with an increasing reactivity taking place with decreasing the Fe amount ( Figure 5). Metals 2020, 10, x FOR PEER REVIEW 6 of 13 Coming to microchemistry, as shown in Table 2, the scenario is quite complicated by the presence of a considerable number of elements; the most relevant feature is the formation of the InSb phase, which was found all over the drop interspersed in a solidified phase formed of Fe, Sb, Sm, Ag, Cu, In. At the skutterudite/solidified drop interface, an infiltrated zone was observed, which was formed of the same phases. Similar reactive microstructures were observed for all the skutterudites in contact with AgCuIn with an increasing reactivity taking place with decreasing the Fe amount ( Figure 5).   Coming to microchemistry, as shown in Table 2, the scenario is quite complicated by the presence of a considerable number of elements; the most relevant feature is the formation of the InSb phase, which was found all over the drop interspersed in a solidified phase formed of Fe, Sb, Sm, Ag, Cu, In. At the skutterudite/solidified drop interface, an infiltrated zone was observed, which was formed of the same phases. Similar reactive microstructures were observed for all the skutterudites in contact with AgCuIn with an increasing reactivity taking place with decreasing the Fe amount ( Figure 5).

Sn
When skutterudites were put in contact with pure Sn, the melting process proceeded with a significantly reduced reactivity in comparison to the previous case: drops appear considerably more rounded, and the formation of solid phases on the drops during tests was limited, as can be seen in Figure 6.
For this reason, it was possible to obtain the plots of contact angle vs. time (see Figure 7) and to measure contact angles (see Table 3). An increased wettability, corresponding to a lower value of the contact angle, was observed for the n-composition Fe50_SPS with respect to other samples. For all the Sn drops, spreading kinetics lasting several minutes from melting was observed, indicating that wetting is guided by interfacial phenomena. Asymmetry of the drop profile, resulting in different values of the right and left contact angles, was also observed, due to the pinning of the liquid at the triple line (i.e., the line of coexistence of the liquid, solid and vapor phases) as a consequence of porosities and asperities of the surface.

Sn
When skutterudites were put in contact with pure Sn, the melting process proceeded with a significantly reduced reactivity in comparison to the previous case: drops appear considerably more rounded, and the formation of solid phases on the drops during tests was limited, as can be seen in Figure 6. For this reason, it was possible to obtain the plots of contact angle vs. time (see Figure 7) and to measure contact angles (see Table 3). An increased wettability, corresponding to a lower value of the contact angle, was observed for the n-composition Fe50_SPS with respect to other samples. For all the Sn drops, spreading kinetics lasting several minutes from melting was observed, indicating that wetting is guided by interfacial phenomena. Asymmetry of the drop profile, resulting in different values of the right and left contact angles, was also observed, due to the pinning of the liquid at the triple line (i.e., the line of coexistence of the liquid, solid and vapor phases) as a consequence of porosities and asperities of the surface.    For this reason, it was possible to obtain the plots of contact angle vs. time (see Figure 7) and to measure contact angles (see Table 3). An increased wettability, corresponding to a lower value of the contact angle, was observed for the n-composition Fe50_SPS with respect to other samples. For all the Sn drops, spreading kinetics lasting several minutes from melting was observed, indicating that wetting is guided by interfacial phenomena. Asymmetry of the drop profile, resulting in different values of the right and left contact angles, was also observed, due to the pinning of the liquid at the triple line (i.e., the line of coexistence of the liquid, solid and vapor phases) as a consequence of porosities and asperities of the surface.   At variance with the In-based alloy, Sn exhibited a reduced reactivity towards skutterudites, and the shape of the solidified drops are better recognizable (see Figure 8).

Fe63_SPS
70° At variance with the In-based alloy, Sn exhibited a reduced reactivity towards skutterudites, and the shape of the solidified drops are better recognizable (see Figure 8). As shown in Figure 9, due to the reduced reactivity, the overall integrity of the substrate was preserved, and the skutterudite surface remained flat, while a large infiltration zone of the liquid into the solid was observed, as observable in Figure 10. The microstructural scenario was consequently quite simpler, as can be seen in Table 4: the EDS analysis revealed the formation of the Sb2Sn3 intermetallic both in the drop bulk and in the infiltration zone. The interfacial behavior and the final microchemistry were similar for all the skutterudite compositions; also, for these systems, a more intense reactivity was observed at lower Fe amounts, accounting for the increased wettability.  As shown in Figure 9, due to the reduced reactivity, the overall integrity of the substrate was preserved, and the skutterudite surface remained flat, while a large infiltration zone of the liquid into the solid was observed, as observable in Figure 10. The microstructural scenario was consequently quite simpler, as can be seen in Table 4: the EDS analysis revealed the formation of the Sb 2 Sn 3 intermetallic both in the drop bulk and in the infiltration zone. The interfacial behavior and the final microchemistry were similar for all the skutterudite compositions; also, for these systems, a more intense reactivity was observed at lower Fe amounts, accounting for the increased wettability.

Fe80_SPS
80° 80° Fe100_SPS 80° 105° At variance with the In-based alloy, Sn exhibited a reduced reactivity towards skutterudites, and the shape of the solidified drops are better recognizable (see Figure 8). As shown in Figure 9, due to the reduced reactivity, the overall integrity of the substrate was preserved, and the skutterudite surface remained flat, while a large infiltration zone of the liquid into the solid was observed, as observable in Figure 10. The microstructural scenario was consequently quite simpler, as can be seen in Table 4: the EDS analysis revealed the formation of the Sb2Sn3 intermetallic both in the drop bulk and in the infiltration zone. The interfacial behavior and the final microchemistry were similar for all the skutterudite compositions; also, for these systems, a more intense reactivity was observed at lower Fe amounts, accounting for the increased wettability.      Figure 9 (regions A and B) and 10 (regions C and D); at least five spots were analyzed for each phase.

Discussion
As observed from microstructures, the main feature of the interfacial phenomena during wetting tests was the skutterudite dissolution with the formation of new compounds, namely InSb and Sb 2 Sn 3 , formed of Sb from the skutterudites, and from the base metal in the alloy.
Regarding the tests with AgCuIn, the InSb formation was accompanied by the solidification of the drop at the testing temperature of 773 K. Looking at the binary In-Sb phase diagram [60], one can see that InSb exhibits congruent melting at 800 K; this means that the liquid In alloy dissolves the skutterudite, and the solid InSb compound is rapidly formed-it was observed as the solid phase in our experiments. The presence of other elements, namely Ag and Cu in the alloy, and Fe and Ni in the skutterudite, does not change this reasoning to a great extent; these elements were dissolved in the In-based melt without forming any other interfacial compound, and they were found as solidification phase in the microstructure (phases B and D in Figures 4 and 5).
When moving to liquid Sn in contact with skutterudites, the situation is quite different. In fact, while a slight dissolution and interfacial reactivity were still present, no solidification phenomena were observed, and liquid drops maintained their shape during the high-temperature experiments. The process of spreading lasted several minutes ( Figure 7) with kinetics, which are typical of dissolutive spreading processes [61], thus demonstrating that wetting and infiltration of the liquid into the solid substrate were guided by the dissolution of the substrate. Looking at the binary Sb-Sn phase diagram [62], it results that no solid intermetallic compounds can be formed at the testing temperature; for this reason, the interfacial compound Sb 2 Sn3, that undergoes peritectic reaction at 596 K, was formed upon cooling. The fact that the β-SbSn phase was not found suggests that the liquid composition remains rich in Sn (X Sn > 0.82), indicating that the dissolution of the skutterudite happens just to a slight extent. According to microstructures, Fe released from skutterudite and introduced into the liquid reacted with Sn to form FeSn 2 , while, for the skutterudites containing Fe and Ni, the phase (Fe, Ni)Sb 2 with Sn partially substituting Sb was detected.
To summarize, from the wetting tests, it has been shown that In is unfeasible for any joining process due to the strong reactivity with skutterudites, which led to the destruction of the starting substrates. On the other hand, the weaker reactivity of Sn with skutterudites, accompanied with good wettability, was more promising, provided that appropriate solutes were added to Sn in order to reduce the interfacial reactivity while maintaining the overall adhesion properties.
Therefore, further studies aimed at obtaining reliable joints through the transient liquid phase diffusion bonding (TLPB) technique should focus on the selection of the most proper interlayer compositions and thicknesses, as well as of process parameters (temperature, time, thermal cycles) in order to preserve the integrity of the skutterudite, and to assure good adhesion between the adjoining materials.

Conclusions
In this work, a wettability study of the filled skutterudite Sm y (Fe x Ni 1−x ) 4 Sb 12 system by Sn and In-based alloys was performed by the sessile drop method in order to find a proper brazing alloy able to connect the thermoelectric material to the device. The temperature range of a possible joining method is limited by the operating temperature of the thermoelectric device (~773 K) and by the temperature at which Sm y (Fe x Ni 1−x ) 4 Sb 12 undergoes the peritectic decomposition (873 K).
Due to the lack of reliable soldering/brazing alloys in the cited temperature range, the transient liquid phase bonding approach was attempted by using pure Sn and an In-based alloy as brazes. SEM-EDS analyses carried out on the cross-section of the solidified drops suggest a quite complicated scenario in both cases, due to the coexistence and the interaction of a large number of different elements in each analyzed system. Indeed, the clear indication of strong reactivity of the In-based alloys with skutterudite, accompanied by the formation of the InSb intermetallic compound, leads to consider In as inappropriate, and to exclude In-based alloys from further studies; on the contrary, Sn exhibits a more promising behavior, being its reactivity limited while coupled to good wettability and to an appreciable adhesion to the skutterudite substrate.