3.1 weld appearance
Fig. 3 presents the weld surface aspect and cross-section profiles with different oscillation amplitudes. Smooth weld and full penetration was achieved for all the samples, with the absence of weld defect (Fig. 3a-c). It was clear that increasing the oscillation amplitude, the weld size was significantly expanded at the top and bottom surface, and the weld width increased from ~384 μm (Aosc=0) to 1248 μm (Aosc=1.3), and the weld shape was altered from hourglass to an inverted trapezoid shape. Compared to Fig. 3d and Fig. 3e, the weld width only had a slightly increase with the addition of oscillation for stirring (Aosc=0→Aosc=0.5), but it was evident with oscillation amplitude increased to 1.3 mm (Fig. 3f).
Previous studies showed that heat input had a significant effect on the weld width, depending on the weld speed, peak power, etc. In this study, the welding parameters were constant, but with various laser oscillation amplitudes. Compared to linear laser beam, the oscillating laser beam is in fact a spiral one since the beam is continuously moving in the weld direction. Fig. 4 presents the schema of the laser beam oscillation during a circular movement. For linear laser welding, the melting metal area was depended on the single laser beam, and its welding speed was equal to the moving speed of the steel plate. However, when applied oscillation, interaction area on the welded metal was increased, which was determined by the oscillation amplitudes. The greater oscillation amplitudes, the larger interaction area. As a result, a wider molten pool can be formed due to more melting welded metals. Thus, the top weld size had a corresponding results after welding cooling and solidification. While, the oscillation amplitudes was 0.5 mm, the weld width had a slightly increase compared to the case of without oscillating (Aosc=0). Laser energy distribution on weld appearance maybe the key, as shown as in Fig. 5, similar results were also presented in Ref. [20]. At Aosc=0.5, the peak energy in the center had a slightly reduce compared to Aosc=0, as well as no evident increase for energy interaction areas. Thus, the weld width was similar for Aosc=0 (0.38 mm) and Aosc=0.5 (0.4mm), but it was increased to 1.6 mm with oscillation amplitudes was 1.5 mm. Compared to Ref. [20], full penetration weld was achieved with different oscillation amplitudes, since the weld depth and width was also determined by other welding parameters, such as welding speed, oscillation frequency and so on.
3.2 Elements in the fusion zone
Ni foil was applied as interlayer during Al-Si coated PHS laser welding in this study. This present focused on the distribution of Ni and Al with different oscillation amplitudes, as shown as in Fig. 6. As mentioned in 3.1, the weld width was increasing with oscillation amplitudes increased. Actually, the weld width was also different from top surface to bottom surface, especially in a transition zone where the size had a sharp cut-down, located in the neck of fusion zone (Marked by red rectangle in Fig. 3). For linear laser (Aosc=0), the content of Ni and Al had a great variation in the fusion zone, and segregation was evident in the transition zone (Fig. 6a and d). With the oscillation amplitudes increased to 0.5 mm, although the weld shape was similar to linear laser, but the segregation was not visible in the transition zone (Fig. 6b and e). When the oscillation amplitudes was 1.3 mm, Al had a more uneven distribution than Aosc=0 and Aosc=0.5 (Fig. 6f). Furthermore, a homogeneous content profile for Ni was achieved in the fusion zone, as shown as in Fig. 6c.
Generally, the melt flow in the molten pool has a significant influence on the elemental distribution in the fusion zone. Based on the previous studies [23-25], this present established a module to understand the elemental flow in the molten pool during laser welding, as shown as in Fig. 7. Al-Si layer coated on the both surfaces of PHS was melted under the effect of laser radiation, and then diffused into the molten pool. Ni interlayer, was also melted with the base metals during laser welding. In this module, we regarded Ni and Al as spherical Ni-rich and Al-rich particles in the molten pool. Here, two melt flow loops were formed by the effect of Marangoni force and keyhole, respectively. It was upper flow loop in the upper half of the molten pool, and lower flow loop in the down half [23]. Ni-rich and Al-rich particles were flowing with melt metals. However, the presence of the transition zone was considered as the block for these particles flow due to the cup-shaped, resulted in an elemental stacking in this zone (Fig. 7a). Thus, it was clear that Ni and Al had a visible segregation in the transition zone with high content for the linear laser. With the oscillation applied, the laser beam has a rotary moving to provide a stirring force in the molten pool (Fig. 8). Hence, the stacking particles were driven to migrate to the whole molten pool (Fig. 7c), and the greater oscillation amplitudes, the more evident stirring effect. It provided an evidence for elemental distribution difference when the oscillation amplitudes increased from 0 to 1.3 mm. Although similar weld size and shape was for Asoc=0 and Asoc=0.5, but the presence of oscillation weakened the segregation in the transition zone (Fig. 7b), and a homogeneous element profile for Asoc=1.3 was achieved as a result.
In addition, it was further found that the average content of Al and Ni was decreased with increasing oscillation amplitudes, as shown as in Fig. 9. In order to calculate the average elemental content in the fusion zone, this present regarded the weld as a simplified shape, as shown as in Fig. 10. Where, a is the width of top surface, mm; b is the width of bottom surface, mm; c is the width of half part, mm; t is the thickness of Al-Si coating, mm; h is the thickness of PHS plate,1.5 mm; d is the thickness of Ni foil, 0.06 mm; S is the area of weld cross-section. Based on this, the Ni content can be calculated as formula (1). It was clear that ωNi was determined by the area of cross-section of fusion zone, S, which depended on the oscillation amplitudes (d is a constant value, 0.06 mm). Thus, the average Ni content in the fusion zone decreased with oscillation amplitudes increased.
By this means, the Al content was calculated as formula (2), it equals the weight of Al divided by the weight of fusion zone. Combined with Fig. 10, the weld shape was altered from hourglass to an inverted trapezoid shape,was increasing with oscillation amplitudes increased, such that a decreasing Al content. When the weld altered to inverted trapezoid shape,
it means that the average Al content in the fusion zone will be a constant nerveless of any oscillation amplitudes. In contrast, the average Ni content will be decreased sharply due to the constant thickness of Ni foil. Thus, it was considered that the descent of average Al content had a slow trend compared to Ni content.
3.3 Microstructure and properties
Fig. 11 presents the microstructure in the fusion zone (FZ) center and fusion line (FL). Similar microstructure was observed in the FZ center and FL for the case of Aosc=0 and Aosc=0.5 (Fig. 11a, b, d and e). Martensite was confirmed by TEM results (Fig. 11g and h). With the oscillation amplitudes increased to 1.3 mm, a different microsturcture profile was achieved in the FZ center, maybe it was another phase. While, TEM image (Fig. 11i) cleared away this uncertainty, the visiable lath martensite was found in the weld center. In addition, δ-ferrite was persent in the fusion line[8,9,12,13].
In our previous studies[12-14], the formation of δ-ferrite was determined by the Al content in the molten pool since Al was an austenite stabilizers, while the presence of Ni can promot the formation of austenite. In this present, although the Ni interlayer has a constant thickness, but the oscillation amplitudes was altered from 0 to 1.3 mm, such that a decreased Ni content in the fusion zone, as mentioned as in 3.2. JMatPro was used to calculate the phase fraction of fusion zone, based on the average elemental contents, as shown as in Fig. 12. As can be seen in Fig. 12 a-c, the liquid phase completely transformed into austenite without any presence of high-temperature ferrite, and followed by transforming into martensite via shear phase transformation during solidification. Compared to Asoc=0 and Asoc=0.5, martensite phase had a different microstructure profile, which was occupied by different Ni content. However, δ-ferrite was persent along the FL with the oscillation amplitudes increased to 1.3 mm, the phase transformation was presented in Fig. 12d. It was showed that 8% of high temperature ferrite did not transform into austenite, and remained to the room temperature to form δ-ferrite. In this case, the Ni content was not sufficent to suppress the formation of δ-ferrite, since the Ni content in the fusion zone had a sharp descent compared to Al.
Hence, the microstructure of fusion zone was lath martensite with the oscillation amplitudes increased from 0 to 1.3 mm, but the increasing oscillation amplitudes also provided a potential for the formation of δ ferrite due to the sharp descent of Ni content in the fusion zone.
Actually, the laser tailored sheet of PHS is followed by hot-stamping. Thus, heat treatment was used in this present to get the welded joints similar to the actual manufacture. The fusion zone had similar average hardness (~498HV) for any oscillation amplitudes, due to the martensite microstructure. However, a very narrow area with low hardness (~299HV) was found near to the fusion line for Asoc=1.3, as shown as in Fig. 13. As mentioned, δ-ferrite formed in the fusion line in Fig. 11f, and it did not go through phase transformation during heat treatment, only with the grain boundary migration. Thus, the prsence of softer ferrtie had the reasonability for the hardness reduction due to lower hardness than martensite.
Fig. 14 presents the tensile properties for the welded joints. The results showed that the increasing oscillation amplitudes in this present did not have an influence on the tensile strength and elongation of welded joints, and fracture failed in the base metal (BM). Dimples were in the fracture surface, implying as ductile fracture. Previous studies [14, 26] had been confirmed that the presence of
δ-ferrite was the key for the strength decrease of welded joint. In this present, the fracture location of A
soc=1.3 was in base metal although little
δ-ferrite was in fusion line, maybe due to the strain constraint from surrounding hard phase. Anyway, the presence of
δ-ferrite in the fusion zone was not recommended in the actual manufacture.