3.2. Al Solute Element Distribution
With the help of point spectrum scan, ten points were selected from the top, middle, and bottom parts of the cross-sectional areas perpendicular to weld, respectively, and the same analysis was conducted along the weld. The distributions of Al element in the welded joints of three samples are shown in Figure 2
From Figure 2
a, it can be seen that the concentration trends of Al element in different regions were basically consistent in sample I. The concentration of the Al element tended to increase from the weld center to the fusion line, but tended to decrease crossing the fusion line to HAZ, especially in the top of the welded joint. The highest concentration peak appeared in the vicinity of the fusion line. It is known from Figure 2
d that along the center line, the lowest concentration of Al element was in the middle of the welded joints.
As illustrated in Figure 2
b, the highest concentration of Al element in sample II appeared in the center of the weld, which was gradually reduced along the direction of the base metal. Near the fusion line, the concentration of Al tended to ascend slightly, and then continued to decrease along the direction of the heat-affected zone. However, the concentrations of Al element changed mildly in the top and bottom parts of weld joints with respect to the middle part. Thus, the effect of the Al element transfer was greater than that of the diffusion in the top and bottom of the weld. As the laser welding process is a rapid melting and solidification process, a large number of aluminum powder were preset on the Ti–6Al–4V titanium specimen in sample II before the welding test. Hence, aluminum element content in the weld upper was higher, and the bottom part was also higher than that of sample I and sample III, which is shown in Figure 2
The basic trend of Al element concentration (Figure 2
c) in the regions of the weld in sample III was similar to that in sample II. Due to the addition of Si element, which increased the fluidity of the molten pool, the movement of liquid metal in the molten pool was intense, which led to the distribution of Al element being more uniform along weld-width direction, and the range of Al element in the weld changed less than that in the sample II weld (see Table 4
The change of Al solute element concentration in the welded joints indicated that burning loss of alloy elements is not the main reason for the uneven distribution of solute elements. Instead, the self-diffusion of solute elements and the flow of molten metal have a great influence on alloy elements distribution in laser welding. The convective motion of molten metal in the top and bottom is far more violent than that of the middle of welded joints. Due to the higher temperature, the solidification time in the top of the weld is longer. The violent movement of molten metal and the longer diffusion time make the distribution of solute elements more uniform in the top of welded joints. As a result, the motion of molten pool in laser deep-penetration welding is studied in this paper to explain the distribution rule of solute elements.
In laser welding of titanium alloy plate, the high-energy laser beam was applied on the surface-coating layer of titanium plate. Due to the large emissivity of Al, only part of the laser energy is absorbed by material. When the laser energy is sufficiently large, the temperature rises to the melting point of alloy elements. Then the surface material starts to melt, leading to intense evaporation and molten pool formation. Solute elements transfer mainly depends on violent convection in molten pool at the formation stage of keyhole in laser deep-penetration welding. As shown in Figure 3
a, the surface with fluctuation of molten pool exhibited two convection rings [17
], namely Marangoni flow [19
], which was caused by the surface tension gradient. The velocity of high-energy laser beam welding is so fast, and then produces very high temperature gradient in the molten pool. The nonuniform distribution of the temperature field in the molten pool triggers a higher surface tension gradient of liquid metal. The surface tension gradient forces solute elements to be transported from the zone with lower surface-tension gradient to the higher one, resulting in the height difference of the liquid metal. In the vertical welding seam, the liquid metal in the middle of the molten pool flows directly to the free surface. The closer to the free surface, the higher flow rate the liquid metal has. The molten pool temperature around the keyhole is higher, and the surface-tension gradient of the liquid metal gradually decreases with increasing temperature, which means the surface tension gradient is lower in the center of the molten pool and relatively higher on the edge of the molten pool. Therefore, the liquid metal flows from center to the edge of the molten pool. Correspondingly, it is equally applicable to solute elements. Finally, the solute elements are formed in the high-concentration zone near the fusion line.
However, in the closure stage of keyhole, only one convection ring was observed in the top of the molten pool, as shown in Figure 3
b. The main reason is that without laser heat source, residual heat still remained. Under the influence of the tension gradient caused by the temperature gradient, convection of the molten pool was maintained in the upside of the weld. However, the liquid metal flow velocity in the middle and bottom of the molten pool was very low, especially in the middle. There was not enough driving force to maintain the movement of liquid metal in the middle of the molten pool. Therefore, stationary state was observed. The intense convective motion of molten metal in the top and bottom is more violent than that in the middle. The influence of molten pool convection on solute elements transfer is greater than that of the solute diffusion, leading to the solute elements transfer from the weld center to the fusion line. However, the liquid metal in the middle of the welded molten pool is almost kept stationary. The solute elements are only transferred by diffusion, and the molten pool often starts solidifying from edge to the center. So solute elements of the middle often are enriched in the weld center. The flow of the molten pool can be significantly improved by adding Si, which makes the movement of the molten pool more violent, so that the distribution of the solute elements in the weld is more uniform.
The microstructure of the welded joint is shown in Figure 4
. The Ti–6Al–4V -welded joints consisted of coarse β columnar grains and small acicular α martensite. The reason for the coarse β columnar phase formation is that the columnar grains are formed in the opposite direction of the heat flux, due to the low thermal conductivity and excessive thermal sensitivity of titanium alloy [20
]. In welding, the β columnar grains in the middle of the weld are grown from the fusion line to the weld center, parallel to the direction of weld width. However, the top and bottom of the molten pool are in a relatively complex temperature and flow field, which leads to the growth of the β columnar grains different from the growth in the middle of the weld. As the top of the molten pool has large contact area with air, the heat exchange is relatively stronger, resulting in the β columnar grains in the top near the surface growing from the fusion line to the weld surface. At the bottom of the weld, the size of columnar structure is relatively fine due to the fast heat loss and short residence on high-temperature time during welding.
In the laser welding process, the welded metal was heated to above the phase transition temperature. During the rapid cooling process in the fusion zone, the alloy elements in β phase do not have enough time to transform into the α equilibrium phase by diffusion. The remote migration of β-phase atom brings about the shear transformation and generates the acicular martensite α’ (supersaturated solid solution of the alloy elements in the equilibrium phase). The α’ phase nucleates and grows both on the boundaries and on the inside of β phase at the same time. The single or parallel α’ phases are firstly formed, which spread throughout the entire grain until the grain boundary [21
]. A series of smaller second-order α’ phases keep forming, before encountering the grain boundaries or the first-order α’ phases. Therefore, the weld shows a typical basketweave microstructure [23
] (Figure 5
). In conclusion, the length of acicular martensite α’ reflects the width of the β columnar grain to some extent, so different lengths of acicular martensite α’ in different regions of weld joints could be chosen as a way to measure the inhomogeneity of the welded joints.
Comparisons of the distribution of acicular martensite α’ length in the weld of the three samples are shown in Figure 6
. The length of the acicular martensite decreased gradually from the top to bottom of the weld in sample I. The length of martensite in the middle of the weld was longer than that in other regions in sample II. In sample III, the longest α’ martensite was found in the bottom of the weld. The martensite length in the top was slightly longer than those in the middle. The result showed that with the addition of alloy powder, the length of acicular martensite in the top of the weld was obviously changed. However, the change in other regions was much smaller, and the minimum acicular martensite was found in sample III. As thermal conductivity of Al is higher than Ti, Al could speed up the cooling rate of the fusion zone, and when the β phase crystal transformed into martensite, nucleation increased. In addition, nucleus do not have enough time to grow up with high cooling speed until the completion of organization transformation, and finer acicular martensite structure can finally be formed [24
]. All of those illustrate that the distribution of solute elements has an effect on the microstructure of the weld.
3.4. Mechanical Properties
Microhardness distribution profiles of the top, middle, and bottom of the weld along weld-width direction are shown in Figure 7
. The addition of alloy powder changed the distribution of microhardness in the weld-width direction. The microhardness value exhibited a smooth transition, and the microhardness distribution along the weld width direction exhibited a “saddle shape” in each layer without any addition of alloy powder. The highest microhardness value was in the region near the fusion line. However, the change of microhardness of Ti–6Al–4V weld with 100% Al added showed a “convex shape” in each layer. The microhardness in the middle of weld first reached the peak value, and then decreased along the fusion line to the base metal. After adding Al and Si powder, different trends in different layers were found as follows: distribution of the hardness in the top exhibited “saddle shape”, and in the middle and bottom exhibited “convex shape”. The distribution of hardness of the weld is related to the distribution of the Al solute element. Thus, the distribution of the solute elements has a great influence on the hardness distribution of the weld.
On the other hand, the hardness value of the weld was significantly improved with the addition of alloy powder, especially in the top. The higher the content of Al, the greater the microhardness. The maximum hardness value of sample I was lower than that of sample II and III, and the differences were about 189 and 130 HV, respectively. The main reason is that during laser deep-penetration welding, the addition of Al element makes the grain finer. Moreover, large amounts of Al were dissolved in α phase, resulting in solid solution strengthening. Thus, the microhardness of the welded joints was improved.
shows the average tensile test results of each sample. It can be seen that the tensile strength was significantly decreased after adding different alloy powders. Adding 100% Al powder made the tensile strength decrease to only 180.95 MPa. The tensile strength of the welded joint with 90% Al and 10% Si added was about 410.55 MPa, but the tensile strength of the welded joint without any powder was 589.85 MPa. The reason for this phenomenon may be attributed to that the aluminum and titanium form hard-brittle Ti–Al phase, which will lead to the reduction of weld strength due to the uneven distribution. The addition of Si makes Al distribution and the Ti–Al-phase distribution uniform, which will strengthen the weld and improve the tensile strength.
All samples were fractured in the welded joints, and tensile fracture morphology is shown in Figure 9
. It can be seen that the fracture surfaces were relatively homogeneous and perpendicular to the loading direction, without necking phenomenon. The macromorphology of the fracture surface in sample I exhibited an obvious “herringbone”, indicating the brittle fracture. The micromorphology showed a clear fan-shaped river pattern, which is the typical cleavage fracture. The macromorphology of the fracture surface in sample II was granular with few rough radiation shapes, indicating it was brittle fracture. And the fracture microstructure with discontinuous river pattern and the “tear ridge” crack expansion connection surface exhibited the characteristics of quasi-cleavage fracture. A row of round porosity was found in the middle of the weld, the walls of these porosities were in a wavy pattern, which indicates that those were the hydrogen holes. The porosities without a wavy pattern may have been the vapor pores. The porosity occupied a certain volume of the weld metal, which reduced the effective working area of the welding seam. These porosities can be stress concentration areas and even form cracks, leading to low tensile strength. In sample III, the fracture surface morphology was similar to that of sample II. The fracture was relatively flat and radial, and there were particles, showing a brittle fracture. Observing the microscopic morphology (Figure 9
f), it was found that the fracture showed a rock-like pattern, indicating the characteristics along the fracture of the crystal.