*3.1. Microstructure of NiTi/CuSn10*

The microstructure of these cuboids observed with the OM was shown in Figure 4. When *E* = 190.48 J/mm<sup>3</sup> , due to the fast-scanning speed, the cooling speed was also fast. So, a large temperature gradient was formed along the scanning route, thus a large thermal stress was formed. Cracks caused by thermal stress did not originate in CuSn10 alloy due to its good ductility. On the contrary, vertical cracks originated in the process of thermal stress release in NiTi alloy the ductility of which is relatively worse and extended to CuSn10 alloy, as shown in Figure 4a [25,26]. When *E* = 200.00 J/mm<sup>3</sup> , the scanning speed and energy density were moderate, so the interface quality was good and no obvious defects were observed, as shown in Figure 4b. When *E* = 222.22 J/mm<sup>3</sup> , vertical microcracks also occurred due to high scanning speed, as shown in Figure 4c. When *E* = 233.33 J/mm<sup>3</sup> and 250.00 J/mm<sup>3</sup> , it can be seen that with the increase of energy density, movement of the molten pool became more intense and Marangoni convection appeared. The two materials penetrated each other, and the generation of an island can be observed, as shown in Figure 4d,e [7]. When *E* = 266.67 J/mm<sup>3</sup> , movement of the molten pool was further intensified and the gas in the building chamber was involved in the part, forming large pores, as shown in Figure 4f. Two types of defects, pores, and vertical cracks appeared in the process of gradually increasing energy density. While the cracks in the direction perpendicular to the interface generally have little effect on the bonding strength of the joint, samples No.1–No.5 were selected for the follow-up tests [5].

**Figure 4.** Optical micrographs of interfaces: Sample (**a**) No.1; (**b**) No.2; (**c**) No.3; (**d**) No.4; (**e**) No.5; (**f**) No.6.

The microstructure of samples No.1–No.5 was observed with SEM after etching. Sample No. 1 was selected as a typical sample for microstructure analysis. In Figure 5a, the NiTi/CuSn10 interface was divided into four zones: CuSn10 region, upper area of interfacial region, bottom area of the interfacial region, and NiTi region, as shown in Figure 5b–e. CuSn10 region was composed of equiaxed grains. Upper area of the interfacial region was also composed of equiaxed grains, and both NiTi and CuSn10 could be seen in this zone. Bottom area of the interfacial region was composed of columnar grains, which formed along the molten pool boundary. Columnar grains formed due to the orientation of the thermal gradient. They were formed from CuSn10 to NiTi (along the direction of heat dissipation), playing the role of a stiffener. That is, the columnar grains were embedded into NiTi alloy, making the bonding firmer. So, it can be concluded that the interfacial bonding properties may be improved. NiTi region was also composed of columnar grains. The average grain size and standard deviation for each microstructure zone were listed

in Figure 5. The width of columnar grain was regarded as its grain size. It can be seen that the average grain sizes of upper area of the interfacial region and bottom area of the interfacial region were smaller than those of CuSn10 region and NiTi region. A possible explanation is that the mixing of multiple elements in the interfacial region may promote a large increase in the number of crystal nuclei and significantly refine the grains [5]. This may also improve the interfacial bonding properties.

**Figure 5.** The microstructure of (**a**) NiTi/CuSn10 interface (500×); (**b**) CuSn10 region (10,000×); (**c**) upper area of interfacial region (10,000×); (**d**) bottom area of interfacial region (10,000×); (**e**) NiTi region (10,000×).
