**3. Results and Discussion**

### *3.1. Analysis of Welds Macrostructure and Microstructure*

The macrostructure of the joints is shown in Figure 1 together with the scheme of the processing. The chemical etchant clearly reveals the different zones of the joints consisting of unaffected base material (BM), FZ and HAZ. In addition, highlighting the intermetallic layer (IMCL) at the interface. Some small round pores can be observed in the FZ, above all on the Al side.

The BM AA6061, supplied in the annealed condition, is a precipitation-hardenable aluminum alloy, containing magnesium and silicon as its major alloying elements. Figure 2a highlights the annealed grains of the alloy. The Vickers micro-hardness (HV0.3/15) is 52.5 ± 1.15. The Titanium base material is supplied in mill annealed condition and the microstructure is characterized BCC (body-centered cubic) beta phase (darker zone in Figure 2b) that outlines the HCP (hexagonal closest packed) alfa phase (lighter in Figure 2b). The Vickers micro-hardness (HV0.3/15) of the alloy is equal to 290.3 ± 10.2.

All the joints exhibit an aluminum fusion zone with columnar grains that grow from the HAZ/FZ interface towards the Ti side (Figure 2c). The growth of columnar grains, promoted by steep heat gradient, occurred in the opposite direction of the heat flow (perpendicular to the FZ/HAZ interface). In the HAZ of the aluminum alloy, coarsening or dissolution of soluble Mg2Si particles, developed due to annealing of BM, can happens [25–31]. The particles dissolution is more significant as the distance from fused zone is reduced since, being closer to the fusion zone, the HAZ thermal cycle is characterized by higher peak temperature [9,10]. Figure 3a shows the evolution of the 3B joint Al side microstructure in the different zones (FZ, HAZ) with respect to the BM.

In the titanium side, FZ, the microstructure is martensitic (α ), as shown in Figure 2d [5,27,30]. In fact, as reported in the literature [5,8–10,13,31], the microstructure of the laser fusion zone of Ti-6Al-4V alloy is due to the complete evolution of the beta phase microstructure to martensitic one due to the high cooling rate. The HAZ is a mixture of martensitic and primary α grain [5,8,13,31]. In the HAZ, during the welding heat thermal cycle, an increase of β phase occurs upon heating but, due to the lower peak of temperature, this transformation is never complete and some alpha phases remain unchanged, while the transformed beta grains develop the martensitic microstructure upon cooling. Figure 3a shows the evolution of the Ti side microstructure in the 3B joint in the different zones (FZ, HAZ) with respect to the BM.

**Figure 2.** Optical micrographs of the AA6061 BM (**a**), Ti-6Al-4V BM (**b**), AA6061 FZ (**c**) and Ti-6Al-4V FZ (**d**).

**Figure 3.** Optical micrographs showing microstructural evolution in the Al side (**a**) and Ti side (**b**).

### *3.2. Intermetallic Layer and Geometry of Welds Fusion Zones*

The IMCL generated during welding at the Al/Ti fused zone interface is due to the melting and mixing of the different alloys. The shape of the intermetallic layer change with process parameters. At fixed defocusing, the interlayer was more curved when the offset was small (Figure 1). At fixed offset, the shape of IMCL become flatter with decreasing defocusing (Figure 1). In fact, both in focus beam and low laser offset promote higher temperature close to the interface favoring convective mixing of the melted alloys, as well as atomic diffusion [8,13]. Consequently, the thickness of the interlayer was also lower with negative defocusing (samples B exhibit lower size of IMCL with respect to the samples T in Figure 4) and increased laser offset (samples 3 exhibit lower size of IMCL with respect to the samples 1 in Figure 4).

**Figure 4.** IMCL thickness size for joints 1T and 1B and joints 3T and 3B as function of the focus beam position.

For the same previously discussed role of energy density on the mixing of two alloys, acicular IMC particles could nucleate and grow mainly in the welds processed with focused beam (Figure 5a,b) thanks to the augmented titanium atoms concentration next to the interlayer (Figure 5).

The microstructure of the FZ and HAZ of all the joints does not change with process parameters. Nevertheless, the size and geometry of the different zones exhibit some differences, as shown in Figures 6 and 7. Regarding the FZ and HAZ sizes, the joints 1T and 1B are characterized by a lower FZ and HAZ with respect to the joints 3T and 3B (Figure 6a). Moreover, for the Ti side of joints 1T and 1B, those sizes slightly decreased with defocusing. In the joints 3T and 3B, due to the higher distance of laser beam from the Al alloy, the heat input is better confined in the Ti side. The lower heat conductivity of this alloy [5,9,10,27,29] leads to higher time in temperature and promotes higher size for the FZ and HAZ (Figure 6b). This effect is strengthened by the lowering of energy density with negative defocusing, as better shown below in Figure 7, that leads to lower temperature gradient [9,10].

For the joints 1T and 1B, the heat input better involves the aluminum side, which easily conducts away the heat due to its higher heat conductivity [5,9,10,27,29]. Therefore, the size of the FZ are reduced with respect to the joints 3T and 3B (Figure 6). The high thermal conductivity of aluminum alloy also compensates the effect of the negative defocusing, so the Al side fused zones do not change significantly with defocusing, while Ti FZ and HAZ are decreased.

**Figure 5.** Intermetallic layer at a half of thickness in the welds cross sections of joints 1T (**a**), 1B (**c**), 3T (**b**) and 3B (**d**).

**Figure 6.** FZ and HAZ size for joints 1T and 1B (**a**) and joints 3T and 3B (**b**). Considering both focused (T) and defocused beam (B), the joints 1T and 1B (**a**) are characterized by a lower size of the FZ with respect to joints 3T and 3B (**b**).

**Figure 7.** Width of the FZ side along the thickness for joints 1T and 1B (**a**) and joints 3T and 3B (**b**): the size of joint 3B along the thickness is always higher than the size of 3T joint (**b**), while the opposite is true for joints 1T and 1B (**a**).

The geometry of the joints FZ at different process parameters was evaluated by measuring the width of FZ along the joints thickness. Three measurements were made for each welds, respectively, at 200 μm, 1000 μm and 1800 μm from the top surface, as shown in Figure 7. The defocused beam strengthens the effect of Ti and Al heat conductivities on the size of FZ; therefore, the size of joint 3B along the thickness is always higher than the size of 3T joint (Figure 7b), while the opposite happens for joints 1T and 1B (Figure 7a).

For the joints processed with the laser offset equal to 0.3 mm, the highest difference in size with defocusing width was found to be 200 μm from the top (in the Ti side, it was equal to 700 micron for 3T vs. 960 micron for 3B). Even for the Al side, the highest increasing of width is registered at 200 μm from the top surface, but the width of Al FZ is more uniform along the joint thickness.

To summarize the results in Figures 6 and 7, the FZ size is decreased (or increased) with decreasing (or increasing) laser offset. The negative defocusing strengthens the effect of laser offset by increasing (decreasing) the size of FZ at the high (low) laser offset. Therefore, the highest sizes of FZ and HAZ were observed for the 3B joint (Figures 6 and 7). The most significant change in the geometry of the FZ size was observed for the Ti side of joint 3, where the width of FZ decreases moving from the top toward the bottom of the joints. The opposite happens for 1T/1B joints, but less significantly.

### *3.3. Intermetallic Particles: Count and Distribution*

The effect of the process parameter on the occurrence and distribution of intermetallic particles in aluminum matrix was also investigated. The number and the size of particles at the Al–Ti interface was evaluated and showed as number of particles for classes of area (Figure 8). For all the welds, the highest number of particles had an area of lower than or equal to 10 μm. The number of particles is reduced by defocusing and, at fixed defocusing, it increases for the welds processed with laser offset equal to 0.1 mm, confirming the role of the peak temperature on the amount of IMC particles. Moreover, it is interesting to note that the number of particles decreases by 64%, comparing the 1T to the 3T joint (as well as from 1T to 1B) and decreases by 78% from the 1B to 3B joint (as well as from 3T to 3B). Therefore, the shift of 0.2 mm in the laser offset position leads to the same effect of a 2 mm shift in defocus distance.

**Figure 8.** Distribution of IMC particles as function of particles area in 1T, 1B, 3T and 3B joints.

A more detailed analysis on IMC particles distribution was developed, considering the number and distribution of the IMC particles along the thickness. The FZ of the Al side was divided in three rectangular window 1000 × 600 μm. The middle window was centered at a half of the thickness in the aluminum FZ. The other two windows are, respectively, on the top side (top window) and the bottom side (bottom window) of the central window, as in the schemes of Figures 9 and 10. Inside each window, the number and size of the particles were counted. The analysis of the results shows that the amount of particles was always highest at the top of all the joints, even if, as showed previously, the maximum number decreases with negative defocusing and increased laser offset. The negative defocusing lead to energy loss and so lower average temperature at the interface; the same occurs when increasing the laser offset. In the T samples, the highest number of particles at the top of the joints is justified by the highest energy density and temperature, which favors a better mixing of the two alloys. For the negative defocusing, the increasing number of IMC particles at the top of the joints could be due to the divergent beam at the top, which imposes a longer time for cooling, favoring atomic diffusion [9,10,29].

The role of the energy density on hardness evolution was investigated by Vickers indentation (0.3/15 s) in the cross section of the joints along a line perpendicular to joint interface. For joints T and B, the hardness measurements were conducted, respectively, at 200 and 1800 μm away from the upper surface (Figure 11). These distances were chosen to investigate the role of energy density in the focus point (joints T) or defocused positions (joints B). By processing the welds using negative defocusing, the hardness values close to the AA6061/Ti interface were reduced. The beam defocusing reduces the hardness imposing lower cooling rate due to lower energy density and the consequent microstructure coarsening. Moreover, in 3T and 3B welds, the hardness values were closer to each other for the two different focus conditions (Figure 11b) in both the Al and Ti sides of the welds.

**Figure 9.** IMC particles distribution along the thickness of joints 1T and 1B.

The effect of the defocused beam on the hardness values is mitigated as laser offset increases, probably due to the prominence of the Ti heat conductivity effect on the cooling rate. With increasing laser offset, Ti heat conductivity dominates the cooling rate and the microstructure evolution and, therefore, the gap in hardness due to beam defocusing is less significant with respect to the joints processed with laser offset equal to 0.1 mm.

### *3.4. Mechanical Properties: Microhardness and Tensile Test of the as Welded Samples*

Table 3 shows the average results after performing the tensile test on two specimens of each type. 3B and 3T welds exhibited the best tensile performances. As shown by the previous analysis, this behavior can be justified by the lowest thickness of IMCL, the lowest number of IMC particles and the lowest gap of hardness values at the interface (Figures 4, 8 and 11). Table 3 highlights the main limit of the joints that is the low value of ductility. Ductility values depend on microstructure [9,10] and intermetallic compounds developed in dissimilar Al–Ti welds that are very brittle [5,27,29,30], but also affected by residual stresses. Residual stresses are significant in the high energy density process (such as laser or electron beam welding) due to the severe temperature gradients and the high heating/cooling rate [9,10]. Residual stress leads to brittle fracture and reduced fatigue life and can be reduced both by heat treatments [9,10,12–16] and mechanical treatment [17–25].

**Figure 10.** IMC particles distribution along the thickness of joints 3T and 3B.

### *3.5. Microstructural Evolution and Mechanical Properties due to SRHT*

For the analyzed joints, a feasible PWHT should reduce the residual stress without hardening the Ti side or softening the Al side in order to not further limit the ductility and toughness of the welds. A significative control of the Ti side microstructure requires heat treatments at temperatures higher than the Beta Transus Temperature of Ti-6al-4V (900 ◦C). Those values of temperature should lead to a completely BCC beta microstructure that allows one to obtain the appropriate microstructure and hardness values upon cooling. Such a temperature is higher than the melting point of AA6061 Al Alloy [27,28,30]. Moreover, the highest heat treatment temperature and holding time suitable for Al alloy must also avoid grain size coarsening. Therefore, the different properties of the alloys limit the effectiveness of the stress relief heat treatment. Considering that the liquidus T of the AA6061 alloy is equal to 582 ◦C [32], the temperature of the SRHT was chosen to be 530 ◦C and the holding time equal to 2 h, followed by air cooling. The treatment was applied to joint 3, characterized by the best mechanical performance (both strength and ductility). Hardness measurements were conducted on the cross sections of the heat-treated joints, at a half of thickness. The results indicate that the HAZ and FZ of Ti6Al4V harden significantly after SRHT, while in the AA6061 side, the increasing of hardness is not significant (five

point more or less). In the Ti side, the highest increases of hardness, obtained close to the weld interface, were equal to 60 HV in 3B joint and 110 HV in 3T Joint (Figure 12).

**Figure 11.** Micro hardness profile as a function of distance from the weld interface at 200 and 1800 μm from upper surface for 1T and 1B weld (**a**) and 3T and 3B welds (**b**).

**Table 3.** Average results coming from tensile test of two specimens for joints 1T, 1B, 3T, 3B.


The increase in the hardness of the Ti side is due to the aging of the martensite, which leads to the precipitation of hardening α + β particles [5,27,30,33]. In addition, some amount of the retained β phase could possibly exist in the FZ and HAZ welds due to rapid cooling. In this case, the effect of the heat treatment should lead to the microprecipitation of the α-phase platelet, it being the α phase most stable at low temperature. Therefore, from the transformation of the retained β, a further hardening effect is obtained, with the β-phase being comparatively soft and ductile relative to the α phase [5,27,30,33].

**Figure 12.** Micro hardness profile as a function of distance from the weld interface at a half of thickness for joints before and after SRHT.

Moreover, it was also observed that the IMC thickness changes with the SRHT and both IMCL and particles in the Al matrix close to the interface grow up upon the SRHT. The average growth of IMCL was equal to 2.60 ± 0.6 μm and 1.30 ± 0.3 μm, respectively, for the 3T and 3B joints. The growth was evaluated along the thickness and it is also easily observable due to its lighter color in the optical micrographs, as shown in Figure 13. SEM micrographs at high magnification (Figure 14) allow one to better observe the growth of the IMC particles due to SRHT. In the SEM micrographs, the new compounds appear darker. According to the EDS analysis on the new nucleated particles (particle 1 in Figure 14f, Table 4) and the previous study of P. Leo et al. [13], the composition of the new nucleated and grown compounds is Al3Ti, being a thermodynamically favored compound at the highest percentage of aluminum [5,13,27,29].

**Figure 13.** Optical micrographs before (**<sup>a</sup>**,**b**) and after (**<sup>c</sup>**,**d**) SRHT for welds3T(**<sup>a</sup>**,**<sup>c</sup>**) and 3B (**b**,**d**).

Due to the microstructure evolution induced by SRHT, both the ductility and strength of the welds were reduced, as shown in Figure 15.

It is noteworthy that a previous study [33] showed that aging heat treatments of Al alloy in dissimilar Al–Ti welds could increase the tensile strength, but only with a strongly reduction of the ductility. In that work, the aging treatment at 180 ◦C required, as a first step, a solution treatment at a temperature higher than 500 ◦C. The low temperature heat treatment increased the strength of the Al side by the precipitation of a small and uniformed dispersion needle shaped Mg5Si6 particles named β" [14,27,29,32], but the strong decrease in weld ductility could be mainly due to the first high temperature step. That first step is responsible for the increases in IMCL thickness/IMC particle size (Figures 13 and 14) and for the strong increase in the hardness of the Ti side, as shown in Figure 12; as a consequence, the gap in the mechanical property at the weld interfaces results increased leading to reduced both fracture strength and ductility [33].

**Figure 14.** Scanning electron micrographs before (**<sup>a</sup>**,**b**) and after (**<sup>c</sup>**,**d**) SRHT for 3T weld at 7500× (**<sup>a</sup>**,**<sup>c</sup>**) and 10,000× (**b**,**d**). 3T sample microstructure after SRHT at 2000× and 7500× with circled the zone of EDS investigation (**<sup>e</sup>**,**f**), together with the points of analysis (**f**).

**Table 4.** EDS result on particle (point 1) and aluminum matrix (point 2) shown in Figure 14f.

**Figure 15.** Tensile curves for joints 3B (**a**) and 3T (**b**) as welded, after SRHT and after UPT. Fractured cross section of B sample after tensile test in as welded state (**c**), after UPT (**d**) and after SRHT (**e**).

The main difficultly in providing good service performance of the dissimilar welds involves the low toughness and ductility. Therefore, the PWHT of the Al–Ti welds at temperatures higher than 500 ◦C is not the correct way to adjust tensile properties. Therefore, a possible application of UTP treatments is proposed in this study.

### *3.6. Role of UPT on Tensile Properties and Corrosion Resistance*

The tensile curve of the 3B and 3T joints after UPT (Figure 15) are characterized by higher elastic modulus with respect the as welded ones. These results could be due to the reduction of porosity induced by UPT, as shown by Chawla et al. [34] and also verified by Tian et al. [18] for a AA6061 Al Alloy. The increases in fracture strain observed after UPT could be justified by the reduction of porosity together with the reduced strain hardening of the UPT samples (compare the tensile curves in Figure 15) [18,34]. In fact, even if the ultimate tensile strength of UPT is treated and welded samples do not differ significantly, the work hardening behavior is different (before and after UPT) because it is affected by the presence and the amount of residual stress [34,35]. To describe the mechanical behavior of the joints after UPT, residual stress analysis and work hardening characterization should be developed. As shown by Zhoua et al. [36], the reconstruction of the residual stress field associated with the work hardening is of grea<sup>t</sup> importance for an accurate assessment of the mechanical performances. Therefore, further detailed studies are necessary to define the relationship between mechanical properties and UPT treatment for Al/Ti dissimilar joints and possibly introduce further improvements. The data presented in this study are only intended to provide a hitherto unexplored chance to overpass the main limit of the dissimilar joints that is the low ductility.

In all the tensile samples, fracture cracks initiate in the Al FZ, at the interface with IMCL or IMC particles, and propagate mainly along the IMCL, as shown in Figure 15c–e by the micrographs of the tensile fracture cross sections of the 3B sample in a welded state, after UPT and after SRHT.

Figure 16a shows the variation of the OCP aluminum alloy before and after UPT, joint before and after UPT and Ti-6Al-4V specimens, in the near-neutral aqueous 3.5% NaCl solution, monitored for 60 min. As expected, the OCP value for the investigated AA6061 alloy is about −0.75 V vs. Ag/AgCl [37]. It is also evident that the OCP value of the treated sample becomes less negative during immersion until reaching a constant potential, probably due to the coarsening of the oxide film. A clear difference can be observed in the behavior of the two samples with an ennobling of the material obtained after the treatment. The significantly higher OCP values of titanium indicate its more noble behavior. The OCP values of the untreated and treated joints are similar to those of the untreated and treated Al6061 samples, respectively, which represent the least noble element of the joints. The potentiodynamic polarization behavior of the investigated samples is reported in Figure 16b. The cathodic branch is mainly controlled by oxygen diffusion, indicating that the corrosion mechanism of the samples is analogous. The anodic branch shows a continuous increase in current density, indicating susceptibility to pitting corrosion until pseudo-passivation region is reached, where the current density increases much more slowly as the potential increases. From the polarization curves, the values of corrosion potential (Ecorr) and corrosion current density (icorr) can be calculated using the Tafel extrapolation method and are summarized in Table 5. Polarization shows much lower Ecorr and higher icorr for the aluminum specimen, both alone and in the joint, indicating an enhanced corrosion resistance in the aggressive solution containing chloride ions after UPT treatment. The slightly more negative values of Ecorr and slightly higher of the icorr, evaluated in the joints compared to the aluminum specimens, may be due to a galvanic coupling effect with Ti-6Al-4V.

In the pictures of Figure 17, the corroded windows for the Al samples and 3B welds before and after UPT are shown. In the optical micrographs, the edge between the corrosion window and the Al alloy (shown in the circle of each picture) is highlighted. The dark zone in the micrographs is the corroded zone, while the lighter one is the Al alloy. The effect of the UPT is evident on the treated Al surfaces (Figure 17c,d) compared to the untreated ones (Figure 17a,b).

**Figure 16.** (**a**) Open circuit potential and (**b**) potentiodynamic polarization curves of AA6061 before and after UPT, 3B joint before and after UPT, Ti-6Al-4V samples tested in 3.5% NaCl solution.

**Table 5.** Corrosion potential and corrosion current density extracted from potentiodynamic polarization curves reported in Figure 16b.


**Figure 17.** Pictures of corroded windows of the samples and optical micrographs of one edge of the corrosion window (circled in the pictures) for Al sample before UPT (**a**), Al sample after UPT (**c**), 3B weld before UPT (**b**) and 3B weld after UPT (**d**).
