Interface microstructure and formation mechanism of ultrasonic spot welding for Al–Ti dissimilar metals

1.   Introduction

A lightweight structure is proposed for performance enhancement and reduction of carbon dioxide emission in the transportation, aerospace, and automotive industries [1]. Titanium alloy (Ti) is a promising lightweight material with high strength ratio, and excellent thermal and corrosion resistance, but the high price limits its applications. Further studies have conducted the welding of aluminum (Al) and Ti dissimilar materials as a solution to balance property and cost.

Fusion welding is widely used in industries but not highly suitable for Al–Ti dissimilar metals. Intermetallic compounds (IMCs), which are brittle in most cases, tend to form at the interface of Al–Ti fusion welding joints due to a high degree of dissimilarity of metallurgical and physical properties [2–3]. In addition, problems of great deformation, internal stress concentration, and grain coarsening, which deteriorates the joint quality, generally occur because of high temperature.

With regard to brazing, Al–Si, Al-based, and Zn-based filler metals for Al–Ti joining have been investigated [4–6]. The thick IMC layer formation and residual thermal stress are the main factors that compromise the joint strength and restrict its application in Al–Ti connections. Welding–brazing generally selects laser or arc as a heat source and provides suitable welding energy and faster heating rate. This method has been regarded as effective in suppressing IMC thickness and controlling IMC morphology [7–8]. However, requirements for high-level operation and strict filler selection decrease the feasibility of this method to a certain extent.

Friction stir welding (FSW) exhibits advantages on Al–Ti joining because of low welding temperature and good ability of plastic material flow [9–10]. Zhao et al. [11] found that the thickness of IMC increased with the length of the probe, and cleavage fracture gradually dominated the fracture of the joints due to the IMC formation. Zhou et al. [12] studied the rotation speed influence and found that IMCs could form in the Al/Ti mixture and hook with the rotation speed of 1400 r/min. However, the probe wear problem during the process reduces joint stability and increases the cost [13]. Recently, Huang et al. [14] obtained excellent dissimilar Ti–Al joints with no probe worn by friction surfacing-assisted hybrid FSW. Another pressure welding method, explosive welding (EXW), is preferred for fabricating multi-layer composites [15–16]. The interface of Al–Ti joints welded by EXW is wavy or flat due to enormous pressure and high local temperature. A subsequent annealing process is needed to improve the properties for this method [17].

In recent years, many investigations have focused on using ultrasonic spot welding (USW) to join Al and Ti dissimilar alloys because of its easy operation and high efficiency compared with other welding methods [17]. Ultrasonic vibration and clamping force are combined to perform solid-state welding [18]. Parameter optimization studies considered welding energy, welding force, and displacement amplitude as crucial factors in joint quality [19]. The relationship between summit load and welding time was studied by Zhang et al. [20]. Wang et al. [21] found that adding a pure Al interlayer could increase the adhesion effect. However, no apparent IMC layer has been detected at the interface by USW even through transmission electron microscopy (TEM) [22–23]. Previous studies have shown the weldability of Al–Ti dissimilar alloys and the common phenomenon in which no IMC formation occurs at the interface by USW. However, most studies have focused on the influence and optimization of welding parameters, mechanical properties, and fracture mode of welded joints. Interface microstructure and formation mechanism of USW for Al–Ti dissimilar alloys are barely clarified [24]. The influence of welding time on Al–Ti ultrasonic spot welds were introduced in our previous study [24]. The current paper reveals the reasons for the absence of IMCs at the interface with kinetic analysis based on the microstructure studies and discusses the influence of welding heat input on joint formation.

2.   Experimental

AA6061 Al alloy (1.5 mm thick) and commercial pure Ti sheets (1 mm thick) were used as experimental materials. The nominal composition (wt%) of AA6061 Al was 0.85 Si, 0.75 Mg, 0.7 Cu, 0.3 Mn, 0.25 Mn, and Al balance. Output power of ultrasonic metal spot welder can reach 3.6 kW, and the vibration frequency is 20 kHz. The knurl pattern of sonotrode tip and bottom anvil with size of 10 mm × 10 mm are shown in Fig. 1(a). The specimens were cut as rectangles with dimensions of 65 mm × 20 mm, and Al sheet was placed on top of the Ti sheet with a 20 mm overlapped area, as described in Fig. 1(b). Specimen surfaces were ground with abrasive papers, cleaned with acetone, and then dried in ambient air before welding. The sonotrode tip was set at the center of welds for the duration of the USW process. The welding energy was selected, ranging from 1100 to 3200 J, which is proportional to welding time. The vibration amplitude and clamping pressure were performed at 32 μm and 15 MPa, respectively. The K-type thermocouples of 0.5 mm diameter were inserted into welds centered between the Al and Ti sheets to estimate the thermal cycle, as shown by the schematic in Fig. 1(b).

Fig. 1.  Illustration of detailed pattern and dimension of sonotrode tip (a), and configuration of welding sheet setup (b).

DownLoad: Full-Size Img PowerPoint

Metallographic specimens were cut perpendicular to the welding direction using an electric discharge machine. The specimens were ground and polished with different grades of abrasive paper and 1 μm diamond paste, respectively, and then etched with a 10wt% aqueous NaOH solution at 60°C for 5 min. Then, the specimens were washed with a 5wt% nitric-acid aqueous solution. OM, field emission scanning electron microscope (SEM) equipped with an energy-dispersive X-ray spectroscopy (EDS) detector and FEI Tecnai-G2 F20 TEM were used to observed the joint appearance and interface microstructure. Thermodynamic and kinetic analyses were conducted to reveal the mechanism of the absence of IMC formation.

3.   Results and discussion

3.1   Joint appearance and microstructure

The joint morphology of the Al–Ti dissimilar alloys produced by USW with welding energy of 2150 J is shown in Fig. 2. The surface of Al and Ti base metal both had a serrated indentation area with a size of 10 mm × 10 mm, as shown in Fig. 2(a). During the process, the sonotrode tip was pressed into the Al sheet to conduct the ultrasonic energy. Friction occurred between the welding tip and Al, Al and Ti, and Ti and bottom anvil through the high-frequency vibration of the welding tip. No obvious welding defects were found at the joint appearance. Fig. 2(b) shows the cross-section of the joint marked with a yellow line as shown in Fig. 2(a). Compared with fusion welding, which has high thermal input, USW did not result in an apparent fusion zone and heat-affected zone at the joints. According to the cross-section, the indention of Al was deeper than that of Ti due to Al’s inferior strength and hardness.

The cross-sections of joints were observed by optical microscope (OM) as shown in Figs. 3(a)–3(c) corresponding to the white rectangles marked in Fig. 2(b), respectively, from top to bottom. The microstructure change of the Al side is the main focus of the study because the Ti side experienced no apparent deformation due to high strength. Plastic flows occurred on the upper surface of Al under the friction and compression of sonotrode as shown in Fig. 3(a). The microstructure of the middle region of Al transformed from fibrous to equiaxed, as shown in Fig. 3(b). The grains contiguous to the interface were refined by the intense friction between base materials, as shown in Fig. 3(c). The aforementioned microstructure changes could be explained by the dynamic recovery and dynamic recrystallization of Al, which easily occurred under the welding temperature and high stress at the USW duration [25–26]. The refined effect may also be associated with the pinning effect by second-phase particles (Mg–Si) that are inhomogeneously distributed along the boundaries and within the grains.

Fig. 2.  Joint morphology at welding energy of 2150 J: (a) weld appearance of a typical Al–Ti USWed joint; (b) macrostructure of joint cross-section marked by the dotted line in (a).

DownLoad: Full-Size Img PowerPoint

Fig. 3.  Typical joint microstructure from upper part of Al to interface: (a) microstructure at position I in Fig. 2(b); (b) microstructure at position II in Fig. 2(b); (c) microstructure at position III in Fig. 2(b).

DownLoad: Full-Size Img PowerPoint

The microstructure of the interface varied with different rates of welding energy. The observation results are exhibited in Fig. 4. Continuous gaps between Al and Ti could be observed from Fig. 4(a). The welding energy in this condition was 1100 J, which is not enough for effective joint completion. As the energy increased to 1625 J, the interface was partially joined, as shown in Fig. 4(b). With the further increase in the energy of welding, the Al softened and the connection between Al and Ti stabilized. The Al–Ti dissimilar alloys exhibited reliable joining under sufficient welding energy of 2150 J, as shown in Fig. 4(c), whereas the excessive welding energy led to crack formation at the interface so that the strength of joint decreased, as shown in Figs. 4(d)–4(e). Welding cracks were affected by the stress concentration resulting from the friction and fatigue effect by ultrasonic.

Fig. 4.  Microstructures of joint interface at different welding energy: (a) 1100 J; (b) 1625 J; (c) 2150 J; (d) 2675 J; (e) 3200 J.

DownLoad: Full-Size Img PowerPoint

3.2   Interface structure

To verify the interface structure more precisely, SEM–EDS and TEM–EDS observations under a welding energy, which is 3200 J, are shown in Figs. 5 and 6, respectively. IMC is most likely observed under the Maxim welding energy. The interface condition and element distributions are studied.

Fig. 5.  SEM–EDS results of interface at welding energy of 3200 J: (a) interface morphology; (b) element distribution at interface.

DownLoad: Full-Size Img PowerPoint

Fig. 6.  TEM–EDS analysis of elements at interface at welding energy of 3200 J: (a) interface morphology; (b) aluminum; (c) titanium; (d) magnesium; (e) oxygen; (f) silicon; (g) iron.

DownLoad: Full-Size Img PowerPoint

The interface of Al/Ti dissimilar metals was obvious and no apparent reaction layer was found according to the SEM images. A wave-shaped deformation at the interface was observed, as presented in Fig. 5(a). Sufficient welding energy is considered as the important factor in the interface deformation and effective joining between dissimilar metals. The EDS line scan analysis approximately demonstrated atomic diffusion. Diffusion distance between Al and Ti was approximately 4 μm, as shown in Fig. 5(b). No IMC was detected even with the largest welding energy in this study. The reasons are associated with kinetic analysis in the following.

The good bonding conditions of the Al–Ti joint are shown in Figs. 6(a)–6(c). Element distributions of Mg, O, Si, and Fe are presented in Figs. 6(d)–6(g). The distributions of Mg were consistent with oxygen according to Figs. 6(d) and 6(e). The segregation of Mg and O occurred at the interface. The concentrated Mg with high activity led to the growth of the oxygen absorption rate of Al. On the other hand, the friction between the faying surfaces of Al–Ti at the process duration caused the temperature to become high enough to form the oxide film of Mg, as confirmed by Field et al. [27]. Silicon (Si) originally in the Al base metal diffused toward the interface, as shown in Fig. 6(f). The Fe, which is the element with the highest impurity content in Ti, was few at the interface, as shown in Fig. 6(g).

The greater the negative chemical enthalpy of elements X and Y, the greater is the chemical attraction between them. The chemical enthalpy between Si, Mg, Fe, and base materials Al and Ti are reported in Table 1. According to the table, the enthalpy change $\Delta {H}_{\mathrm{T}\mathrm{i}-\mathrm{M}\mathrm{g}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$(16 kJ∙mol⁻¹) is higher than $\Delta {H}_{\mathrm{A}\mathrm{l}-\mathrm{M}\mathrm{g}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$ (–2 kJ∙mol⁻¹), which reveals that the segregation of Mg was not attracted by Ti but was spontaneous to the Al surface. The value of $\Delta {H}_{\mathrm{T}\mathrm{i}-\mathrm{S}\mathrm{i}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$ is –66 kJ∙mol⁻¹, which is much lower than that of $\Delta {H}_{\mathrm{A}\mathrm{l}-\mathrm{S}\mathrm{i}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$(–19 kJ∙mol⁻¹). The migration of Si to the Al–Ti interface from Al alloy could be attributed to the greater affinity between Si and Ti. The difference of $\Delta {H}_{\mathrm{T}\mathrm{i}-\mathrm{F}\mathrm{e}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$ and $\Delta {H}_{\mathrm{A}\mathrm{l}-\mathrm{F}\mathrm{e}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$ is not significant, so the driving force is insufficient for Fe diffusion to the Al–Ti interface, which explains the small amount of Fe at the interface. The analysis results are consistent with Fig. 6. In general, element segregation has a negative effect on the mutual diffusion between Al and Ti.

Table  1.  Chemical combination of alloying elements (Si, Mg, Fe) with matrix (Al, Ti)

Diffusion couple	$\Delta {H}_{\mathrm{X}-\mathrm{Y}}^{\mathrm{m}\mathrm{i}\mathrm{x}}$ / (kJ·mol–1)
Al–Ti	–30
Al–Mg	–2
Al–Si	–19
Al–Fe	–11
Ti–Al	–30
Ti–Mg	16
Ti–Si	–66

 | Show Table

DownLoad: CSV

The observation results indicate the absence of IMC at the interface. Thermodynamics and kinetics are discussed in the analysis of this phenomenon. According to the Al–Ti binary phase diagram, many intermetallic compounds have the possibility to form between Al and Ti, including TiAl, TiAl₃, and Ti₃Al. The formation of reaction products is associated with free energy difference. The relationship between free energy difference and temperature, based on the Gibbs–Helmholtz equation, is as follows:

In the formula, ${\Delta _{\mathrm{r}}H}_{\mathrm{m}}^{\ominus}$ is the standard molar reaction enthalpy at temperature T (kJ·mol⁻¹), and ${\Delta _{\mathrm{r}}S}_{\mathrm{m}}^{\ominus}$ is the standard molar reaction entropy change at temperature T (kJ·K⁻¹·mol⁻¹).

where $\Delta _{\mathrm{r}}{C}_{\mathrm{p},\mathrm{m}}$ is the constant pressure molar heat tolerance of reactants and products (kJ·K⁻¹·mol⁻¹), and p and q are the number of atoms in the reaction.

The functions of Gibbs free energy of different IMCs varying with the temperatures were provided by Kattner et al. [28], as shown in Table 2. Based on data from the literature, the Gibbs free energy of Ti–Al IMCs in the range of 273 to 1300 K is shown in Fig. 7. The TiAl₂ has the largest free energy change, as shown in Fig. 7, but the TiAl₂ needs to be formed on the basis of Al–Ti. Thus, TiAl₃ has the largest possibility to form during the welding process. As TiAl₃ was not observed in all conditions, the following kinetic analysis is necessary.

Table  2.  Functions of Gibbs free energy of IMC in series of Ti–Al varies with temperature

Compounds	Free energy of formation / (kJ·mol–1)
TiAl	–37445.1+16.79376T
Ti3Al	–29633.6+6.70801T
TiAl2	–43858.4+11.02077T
TiAl3	–40349.6+10.36525T
Ti2Al5	–40495.4+9.52964T

 | Show Table

DownLoad: CSV

Fig. 7.  Curves of Gibbs free energy of Ti–Al intermetallic compounds varying with temperature.

DownLoad: Full-Size Img PowerPoint

At the range of USW temperature, the growth of the TiAl₃ layer is determined by diffusion. The thickness of the reaction layer and time satisfied the parabolic law at a certain temperature, which is usually simplified as follows:

where K is the rate constant, t is the reaction time, and n is the kinetic index.

The transformation rate depends on the frequency of the atom reaching the active state, as determined by the following Arrhenius equation:

where A is the pre-factor, ΔG^a is the growth energy of the reaction layer, and T is the absolute temperature.

A series of metastable phases may appear during the transformation process as long as the Gibbs free energy of the phase transitions decrease. Fig. 8 shows a free energy diagram of a single atom transition from a metastable state to a lower energy steady state during phase transition, where G₁ and G₂ are the metastable and steady state Gibbs free energy, respectively. The atom must receive additional free energy to reach the activation state and complete the transition. Therefore, activation energy determines the difficulty of the entire diffusion system. The USW of Al–Mg, Al–Fe, and Al–Ti all belong to solid–solid reactions, which encounter difficulty in forming IMC compared with liquid–liquid or liquid–solid status, but IMCs are not observed only in the Al–Ti interface [29]. The probable explanation is that Al–Ti has a diffusion activation energy with 250–300 kJ·mol⁻¹, which is much higher than that of Al–Mg and Al–Fe with diffusion activation energy of 60–70 and 190 kJ·mol⁻¹, respectively [30–31].

Fig. 8.  Free energy change process from initial state to final state transition.

DownLoad: Full-Size Img PowerPoint

Solid solubility is a prerequisite for atom diffusion. The calculations of dissimilar metal solubility are presented in Fig. 9. The solid solubilities of Al–Mg and Al–Fe are high although the solubility of Fe in Al is low which is similar to that of Ti in Al, but the solubility of Al in Fe is the highest among them, which could also explain that IMCs exist in Al–Mg and Al–Fe but not in the Al–Ti interface. A common way to diffuse in the multi-crystalline solids is grain boundary diffusion [32]. Relationships between the effective diffusion coefficient of polycrystalline solids and bulk diffusion coefficient and grain boundary diffusion coefficient are as follows [32]:

Fig. 9.  Mutual solid solubility between Al–Ti, Al–Mg, and Al–Fe

DownLoad: Full-Size Img PowerPoint

where ${D}_{\mathrm{g}\mathrm{b}}$ is the grain boundary diffusion coefficient, ${D}_{\mathrm{L}}$is the bulk diffusion coefficient, $g$ is the grain boundary volume fraction, δ is the grain boundary width, d is the grain size, and q is the numerical factor of the grain shape.

The size and configuration of grains affect the effective diffusion coefficient. Diffusion rate increases with the grain refinements. According to the analysis of microstructure at the joint interface, the grain sizes of Al became coarse with the increasing heat input due to the decreasing effective diffusion coefficient. Thus, the IMC growth driving force is insufficient because the diffusion amounts from Al to Ti encounter difficulty in achieving the solid solubility limit during the short welding time.

Similar to element segregation, which would deteriorate the joint mechanical properties and prevent the growth of the interaction layer to a certain extent, the residual oxide film formed at the Ti surface as a diffusion barrier hindered the movements of the elements at the initial stage of diffusion. As the vibration amplitude of USW was small (<37 μm) and the welding tip width was 10 mm, the broken oxide film at the center of the joint could not be moved outside the interface.

3.3   Thermal cycle

At the USW process duration, the thermal cycle varying with the welding parameters has a close relationship with the microstructure and mechanical properties of joints. The thermal cycles of the weld center corresponding to welding energy of 1100–3200 J are shown in Fig. 10(a). All the temperatures of the center welds rose rapidly to the maximum value, and then gently decreased to room temperature. Magnification of the thermal cycle curve at the welding energy of 3200 J is shown in Fig. 10(b), which presents the details clearly. Three parts are divided to describe the thermal cycle, which are temperature rise period, high temperature holding period, and cooling period. The temperature of the workpiece first rose rapidly under a heating rate of 648 K/s due to friction and clamping pressure. It cost approximately 0.8–1 s to achieve the maximum temperature. When the time exceeded 1 s, the ability to generate and dissipate heat reached equilibrium, so the thermal cycle achieved the predetermined temperature holding period. Once the set time is achieved, the heat input stopped and the temperature of the workpiece started to decrease. The high temperature holding period plays a critical role in the plastic deformation and element diffusion. The longer the high temperature holds, the more sufficient elements diffuse in theory.

Fig. 10.  Thermal cycling curves at different welding energies: (a) trend of thermal cycling during welding process; (b) a partial enlarged view of thermal cycle variation curve at 3200 J welding energy

DownLoad: Full-Size Img PowerPoint

3.4   Welding joint formation mechanism

Welding heat input is considered as an important factor in joint formation. Therefore, the mechanism of the joint formation by USW is explained by analyzing the joint formation with different heat inputs as follows.

(1) When the heat input is low.

During the welding process, the sonotrode and bottom anvil caused jagged indentations on the surface of the upper and bottom workpiece, respectively, to conduct the ultrasonic vibration to the interface of two sheets. The ultrasonic energy conducted by the sonotrode induced the intense mutual friction and pressure concentration on some projections of the base material. When the heat input is low, the micro-bondings are few and the strength of biting facets is low. The micro-bonding is rapidly destroyed by the shear stress of ultrasonic vibration. Meanwhile, the heat input is insufficient to break the oxide film; thus, the dissimilar metals cannot contact directly. Consequently, the metallurgical connections are difficult to form, as shown in Fig. 11(a).

Fig. 11.  Illustration of joint formation under different heat input conditions: (a) insufficient welding heat input; (b) material plastic flow at interface; (c) appropriate welding heat input; (d) optimal welding heat; (e) excessive welding heat input.

DownLoad: Full-Size Img PowerPoint

(2) When the heat input is suitable.

When the welding process improves, the micro-bonding regions plastically deform. The wave-like deformation at the interface, under the pressure and shear force generated by relative sliding and the vortex plastic flows at the micro-connections, improve the mechanical connections as shown in Fig. 11(b). During the friction process by ultrasonic vibration, the temperature rapidly increases while the deformation resistance of materials decreases with the area of friction expansion. At the same time, the oxide film is broken, displaced, and dispersed. The atoms on the surface of Al–Ti dissimilar alloys are close enough to attract each other, the elements such as Al and Si diffuse quickly through these passages, and the joints with effective bonding areas are formed, as shown in Fig. 11(c).

Under the stress and shear force generated by ultrasonic vibration, the plastic flows continue and the broken oxide films disperse into the inner position of the base material. As the friction process goes on, the number of biting points increases and the effective connection area expands continuously. When the bonding force at the weld is higher than the shear stress caused by the ultrasonic mechanical vibration, the workpiece is no longer cut by the shear stress and a strong joint is formed. When the heat input reaches the optimum point, the joint strength is the highest, as shown in Fig. 11(d).

(3) When the heat input is extremely high.

Since the defection of sonotrode netting, the difference in the thickness of plates and weld deformation, the forces on the weld spot are not uniform during the welding, which leads to incomplete physical contact, thereby resulting in an interspace between the plates. As the heat input continuously increases, the diffusion distance extends. Furthermore, the stress concentration intensifies due to the shear stress and fatigue effect with high-frequency ultrasonic vibration and accumulated deformation resistance internal stress, which causes the appearance of cracks at the joint, as shown in Fig. 11(e).

4.   Conclusions

The microstructure of the Al–Ti joints by USW was investigated. The reasons for the absence of visible IMC were analyzed, and the mechanism of joint formation was discussed. The main conclusions are the following:

(1) AA6061 Al (1.5 mm thick) and commercial pure Ti sheets (1 mm thick) were used to obtain a strong joint welded by USW. As the welding energy increased, the base material plastically deformed with the grain, resulting in gradual coarsening and recrystallization. Under high stress and friction, the upper part of the Al and the area near the interface part had a plastic flow with grain refinement. The reaction layer of the intermetallic compound was not observed under all welding parameters.

(2) The IMC of Al and Ti was difficult to form by USW mainly due to the high energy barrier of Al–Ti. Another important reason is that the solid solubility limitation of Al–Ti is inaccessible during the short welding process because of the element segregation, residual oxide, and decreasing effective diffusion coefficient with coarsening grains on the Al side.

(3) During the USW process, the weld center temperature with different welding parameters rapidly rose to the maximum of approximately 0.8–1 s, and then decreased to room temperature slowly. The heating rate was 648 K/s. Temperature rise, high temperature holding, and cooling periods can be observed from the thermal cycle.

(4) Under the combined action of static pressure and ultrasonic mechanical vibration, mechanical occlusion occurred between the surfaces of Al–Ti dissimilar alloys, and the interface was gradually expanded through discontinuous micro-connections until an effective joint formed. When the welding heat input was appropriate, the mechanical fitting and metallurgical bonding between the materials were excellent.

Acknowledgements

We are grateful for the financial support provided by the National Natural Science Foundation of China (Nos. 51974100 and 51605117).