1 Introduction

Lightweight alloys have been widely used in various industries for weight and cost reduction purposes. Titanium and aluminum alloys rank among the most important lightweight alloys because they provide excellent performance in many applications [1, 2]. Titanium alloys provide a high strength/weight ratio, superior corrosion resistance, and the ability to perform at high temperatures. Aluminum alloys are versatile in automobile industry applications where low cost, lightweight, and good plasticity are required. By appropriately joining the two alloys, a joint with complementary advantages of them will be obtained, which has a huge market in aerospace and other application fields [3, 4]. However, it is generally accepted that obtaining a good joining of Ti/Al is hindered by metallurgical incompatibility, the differences of their physical properties, and dominantly by the formation of brittle intermetallic compounds (IMCs) [5, 6].

Although the formation of the IMCs is necessary for the metallurgical joining of aluminum to titanium, their thickness must be controlled. Traditional fusion welding experiences a reaction between liquid aluminum and liquid titanium, which makes it impossible to avoid the massive formation of the brittle IMCs [7]. Solid-state joining processes such as explosive welding [8], diffusion welding [9], friction welding [10], and friction stir welding [11,12,13] rely on solid-state reactions and low-temperature thermal cycles, which can effectively suppress the IMCs. Welding-brazing processes skillfully avoid the difficulties caused by the different melting points of the two compounds, which has great advantages in joining dissimilar metals. Many joints were obtained by welding-brazing with arc [14,15,16,17,18], laser [19,20,21,22], and electron beam [23].

However, it is possible but more difficult to control the IMCs in the Al/Ti welding-brazing process based on a solid–liquid interface reaction in a few seconds. Generally, three dominant options could be employed to control the IMC growth in the above system. The first option is to employ appropriate alloying elements in base metals or fillers to adjust the interfacial reaction [24,25,26,27,28]. Another choice is to add an intermediate layer to hinder the mutual diffusion of atoms, such as Zn [29] and Nb [30]. Additionally, the growth of the IMCs can be controlled effectively by decreasing the heat input during Al/Ti welding. Yufeng Zhang et al. used a MIG/TIG double-sided arc welding-brazing to control the heat input, suppress the IMC layer to 2–6 µm, and increase the joint strength to 240.3 MPa [31]. Rui Cao et al. applied a low heat input welding technique—cold metal transfer (CMT) to Al/Ti joining for the first time [32]. Ming Gao et al. employed laser-CMT hybrid welding to reduce the heat input and control the IMC thickness, but the IMC thickness along the vertical direction was inhomogeneous [33].

Most previous studies on CMT joining of Ti/Al dissimilar alloys were based on the traditional single-wire CMT. Although these works have indicated a promising effect on controlling welding heat input, the performance of the Al/Ti joints still cannot fully meet the requirements of all the application fields. As an upgrade of the single-wire CMT, a CMT Twin technique based on the setup of two power sources, one torch, and two insulated contact tips (“Lead” and “Trail”) was developed. The CMT and pulsed processes can be combined, and better self-regulated than ever before can be accomplished. Especially, CMT Twin welding has more significant advantages than traditional single-wire CMT welding for joining aluminum to titanium, such as less heat input, independent adjustments of fillers, and welding parameters of the two incompatible alloys.

In this study, the CMT Twin process is applied to Ti/Al butt welding-brazing to study the interfacial IMC layer and joint strength, compare the mechanical properties of the joints obtained by the traditional single-wire CMT and CMT Twin processes, and discuss the relationship among the welding parameters, microstructure, and joint properties.

2 Materials and methods

The TC4 titanium alloy and 5083 aluminum alloy sheets used in the experiment were 3 mm × 100 mm × 60 mm in size, and a 45° single V-groove was cut on both sides of the two sheets. The filler metal was an ER4043 aluminum welding wire with a diameter of 1.2 mm. According to the data sheets provided by the producer, the chemical compositions of the materials are listed in Table 1. The shielding gas used was 99.999% ultra-high purity argon produced by the Air Products company. Before welding, oxide films on the surfaces of the alloys were cleaned by brush and acetone, and then, a 0.2–0.5-mm-thick non-corrosive KAlF4 flux was coated on the grooves as well as on the face and back surfaces of the titanium alloy.

Table 1 Chemical composition of the base and filler metals (wt%)
Full size table

A schematic of the CMT Twin welding-brazing process is illustrated in Fig. 1. Twin Fronius TPS CPT power sources and a KUKA arc welding robot were employed in this study. A groove gap of 1 mm was reserved between the aluminum and titanium alloys. The welding torch kept vertical to the plates. There was an angle of approximately 7° between the two wires in the CMT Twin process.

Fig.1
figure 1

Schematic of the CMT Twin welding-brazing process: a welding system, b welding-brazing process

Full size image

The CMT welding process was applied in the experiment, and a twin CMT mode was performed for welding. In one motion cycle, the front wire performs a jet transition in the first period of the cycle. In the next period, the droplet transfer is carried out by mechanical wire pullback, which is a typical CMT mode. The rear wire is in the CMT mode throughout the entire welding cycle. The welding robot controls the welding wire to start the arc at the beginning of the path and then moves to the end of the path along the preset track. During the welding, the two welding wires cooperate with each other under the regulation of the control system to complete the welding. The classical linear energy equation (single-wire CMT, Q = IU/v; CMT Twin, Q = 2*IU/v) was used to calculate the welding heat input.

After welding, one metallographic and three tensile test specimens were cut from each weld plate by wire electrical discharge machining. The macrostructure samples were polished to mirror-like surface aspects. Then the macrostructure of the joint was inspected using a Suzhou Huiguang HSH-200C plane test microscope. The metallographic structure was observed using a Zeiss Ario Imager M2m digital metallographic microscope after the samples were etched in Keller’s reagent for 5–10 s. The microstructure of the interfacial layer was observed using an ultra-high resolution field emission scanning electron microscope (FEI-Nova Nano SEM 430, The Netherlands). TEM samples of approximately 4 μm × 4 μm × 0.09 μm were prepared from the cross-section. The IMC at the interface was examined with a transmission electron microscope (FEI Talos F200X) equipped with an energy dispersive spectrometer (EDS). The reinforcement on the specimens was retained, and tensile tests were conducted using the SHIMADZU AG-IC universal testing machine.

3 Results

3.1 Welding parameters

The welding parameters of the CMT Twin and single-wire CMT were determined by pre-experiments, as shown in Table 2. For Ti/Al CMT welding-brazing, there is a reasonable welding heat input range that can obtain an acceptable weld formation, as illustrated in Fig. 2. Once beyond this range, excessive heat input will lead to the melting of titanium base metal and generate massive brittle IMCs, while inadequate heat input cannot guarantee the melting of the aluminum base and filler metals. Employing the CMT Twin extended the range of heat input by 15.4% and made the range shift to a negative direction. The expansion and shift of the welding heat input window help achieve a more stable welding process and a lower interface temperature.

Table 2 Welding parameters
Full size table
Fig. 2
figure 2

Ranges of welding heat input for different welding methods

Full size image

3.2 Macrostructure and microstructure

Visual inspection was performed on the samples. No obvious weld defects were observed in Fig. 3a. Figure 3b presents a typical cross-sectional view of the welding-brazing joint with the CMT Twin. Reinforcements at the face and root of the joint were observed. The aluminum alloy, which had a low melting point (570 ℃), was fused and mixed with the liquid filler metal to produce the welded seam with an obvious fusion line, while the titanium alloy, which had a high melting point (1650 ℃) and remained in the solid state during the welding-brazing process, was wetted by the liquid aluminum and reacted with it at the interface. Metallography of the fusion welding area of the aluminum alloy and the brazing area of the titanium alloy in this cross-sectional area are shown in Fig. 3c and d, respectively. A uniform IMC layer can be observed in the Ti/welded seam interface in Fig. 3d.

Fig. 3
figure 3

Macroscopic and microscopic images of the Ti/Al butt joint: a top view of the joint, b transverse cross section, c metallography of position 1, d metallography of position 2

Full size image

The TEM microstructure of the IMC layer is shown in Fig. 4. The FIB selection area is shown in Fig. 4a and b. From Fig. 4c, the compound on the interface layer between the titanium alloy and the weld metal is a monolayer, growing from the titanium side into the weld bead in an irregular serrated shape. Based on the TEM diffraction results in Fig. 4c, the IMC is identified as a tetragonal TiAl3 phase with the lattice parameters of a = 0.3853 nm and c = 0.8584 nm [34].

Fig. 4
figure 4

IMC morphology and diffraction spots: a micrograph of the interface layer, b FIB sample preparation selection area, c TEM morphology, d TEM diffraction spot of TiAl3

Full size image

The composition of the intermetallic layers was measured using EDS of the TEM foils and is summarized in Table 3. The results show that a small quantity of Si is contained in the IMC. The IMC was thus determined to be Ti (Al, Si)3.

Table 3 Analysis of EDS (at%)
Full size table

3.3 Effect of heat input on IMC

The SEM images of the IMCs with different welding heat inputs are shown in Fig. 5. When the heat input is relatively low, the interface is a large smooth plane. As the heat input increases, the interface gradually evolves into a serrated shape composed of many small planes with different inclination angles.

Fig. 5
figure 5

SEM images of the IMCs under different welding heat inputs with CMT Twin (a–d) and single-wire CMT (e–h): a 180.0 J/mm, b 275.3 J/mm, c 379.6 J/mm, d 486.0 J/mm, e 275.3 J/mm, f 379.6 J/mm, g 486.0 J/mm, h 540.5 J/mm

Full size image

The thicknesses of the IMCs with varying heat inputs are illustrated in Fig. 6. The average thicknesses of the IMCs and standard deviation were calculated according to three different positions on the same specimen interface. With an increase of welding heat input from 180.0 to 486.0 J/mm, the IMC thickness at the CMT Twin welding interfaces increased constantly from 0.53 to 2.12 μm. Similarly, the thickness of the IMCs obtained by the single-wire CMT also increased from 1.83 to 3.22 μm with increasing heat input. However, under the same heat input, the IMCs obtained by the CMT Twin are thinner than those obtained by the single-wire CMT.

Fig. 6
figure 6

Relationship between welding heat input and thicknesses of IMCs

Full size image

3.4 Mechanical properties

The results of tensile tests with and without reinforcement are shown in Fig. 7. The error bars in the Fig. 7 represent the standard deviation calculated from the tensile strength of three parallel specimens. For the joints with reinforcement, the tensile strength by CMT Twin decreased from 297 to 259 MPa, while the strength by single-wire CMT reduced from 218 to 155 MPa with increasing heat input. On the other hand, for the joints without reinforcement, the strength by CMT Twin decreased from 258 to 189 MPa, while the strength by single-wire CMT reduced from 219 to 119 MPa with increasing heat input. The tensile strength of the joints obtained by CMT Twin welding is higher than that of single-wire CMT in both cases with and without weld reinforcement. The maximum strength of the CMT Twin joint is 32% higher than that of single-wire CMT.

Fig. 7
figure 7

Relationship between joint strength and heat input: a with reinforcement, b without reinforcement

Full size image

The microfracture profiles of the joints are shown in Fig. 8. First, for the joint with a relatively low tensile strength, the fracture originated at the root of the interface layer and extended to the weld. Furthermore, for the joints with higher strength, the crack evolved along the weld close to the fusion line. Finally, for the joint with the highest strength, fracture occurred at the base aluminum alloy.

Fig. 8
figure 8

Fracture positions of the joints by CMT Twin: a 259 MPa, b 295 MPa, c 297 MPa

Full size image

3.5 Relationship between IMC thickness and joint strength

Figure 9 presents the relationship between the IMC thickness and tensile strength. The tensile strength decreases significantly with increasing IMC thickness. Joints with or without reinforcement show the same trend. When the IMC thickness is 0.53 μm, the tensile strength of the joint reaches a maximum of 297 MPa. The fitting results of the strength variation are summarized in Table 4. The absolute slope values of the CMT Twin fitting lines are smaller than the values of the single-wire CMT, indicating that the decreasing speed of the joint strength with increasing IMC thickness obtained by the CMT Twin is slower than that of the single-wire CMT.

Fig. 9
figure 9

Relationship between IMC thickness and joint strength

Full size image
Table 4 Fitting results of the strength variation
Full size table

4 Discussion

4.1 Microstructure and phase identification

In this work, a dominant layer (TiAl3) between aluminum and titanium was observed. There are other IMCs, such as Ti3Al, TiAl, TiAl2, and Ti2Al5, that may be present during the process but are not found in the final microstructures. Referring to the calculation results, the Gibbs energy of TiAl3 is lower than that of TiAl and Ti3Al and higher than that of TiAl2 and Ti2Al5 [35, 36]. Although the compounds TiAl2 and Ti2Al5 have lower free energies of formation than TiAl3, TiAl3 is still the dominant IMC layer at the joining interface. Raman and Schubert may explain this phenomenon that is the formation of TiAl2 and Ti2Al5 depends on the formation of TiAl as one of the starting phases [37]. Moreover, when the welding heat input was high enough, the IMC at the interface shifted from a single TiAl3 phase to TiAl + TiAl3 [31]. In this study, the formation of the TiAl was suppressed, possibly due to the effective control of the welding heat input by the CMT Twin welding technique, presenting a single thin layer of TiAl3.

On the other hand, based on the composition of the intermetallic layer from the EDS, the IMC can also be expressed as Ti (Al, Si)3. For the Ti–Al-Si system, since the silicon and aluminum atoms are very similar in atomic radius (111 pm and 118 pm, respectively [38]) and can easily undergo substitution reactions, [Si] + TiAl3 → Ti (SixAl1-x)3, where silicon atoms replace some of the Al atoms in TiAl3 to generate Ti (Al, Si)3 [39].

4.2 Relationship among heat input, IMCs, and mechanical properties

It is generally recognized that welding heat input plays an important role in the growth of IMCs. During the CMT Twin welding-brazing of Ti/Al dissimilar metals, the aluminum alloy melts and the titanium alloy remains in solid, forming a solid titanium-liquid aluminum reaction system. In theory, there are two important views on the formation of IMCs in solid alloy-liquid aluminum systems. One is that IMCs began to grow during the dissolution of a solid metal, and heat input affects the thickness of IMCs by affecting the diffusion rate and interfacial reaction rate [40]. Another view is that the IMCs formed during cooling and heat input influences IMCs by affecting atomic diffusion [41]. Both theories admit that diffusion is the most important factor controlling the growth of IMCs, and the diffusive mechanism of particles during the solid–liquid phase interfacial reaction is a non-stationary diffusion process, which can be briefly described by referring to Fick’s second law for the diffusion of titanium atoms into the weld [9]:

$$\frac{\partial c}{\partial t}=D\frac{{\partial }^{2}c}{{\partial x}^{2}}$$

where c is the concentration of Ti atoms, t is the diffusion time, x is the distance of Ti atoms relative to the interfacial layer, and D is the diffusion coefficient. Assuming that the Ti atom content in the weld metal at the beginning is “c0,” the following derivation can be made:

$$c={c}_{0}-{c}_{0}*\mathrm{erf}\left(\frac{x}{\sqrt{2Dt}}\right)$$

where erf (β) represents the value of the error function corresponding to different β values, due to the diffusion coefficient D:

$$D={D}_{0}exp\left(-\frac{Q}{RT}\right)$$

D0 is the diffusion constant determined by the diffusion mechanism, and R is the gas constant and is a constant value. T is the absolute temperature, and Q is the activation energy per mole of atomic diffusion. The higher the reaction temperature, the higher the welding heat input, and the more intense the diffusion movement of Ti atoms. The welding speed, wire feeding rate, and other process parameters determine the growth of IMCs by affecting the welding heat input. Some works have reported that the IMC thickness at the Al/Ti interface increases with increasing heat input in arc welding-brazing [14] and laser welding-brazing [42] processes. The results presented in Fig. 6 of this study are consistent with these results. A higher heat input indicates a higher interface temperature and a longer reaction time, which leads to a larger IMC thickness.

The strength of Ti/Al dissimilar joints mainly depends on the morphology and thickness of the interfacial IMCs [19, 43, 44]. The difference in the thermal expansion coefficient between 5083Al (2.6 × 10–5 ℃−1) and TC4 (7.9 × 10–6 ℃−1) will lead to the generation of residual stress, and the presence of brittle inclusions is prone to stress concentration under the influence of welding thermal stresses and other factors, which becomes the origin of fracture cracks. Under the influence of the inherent brittleness, the excessively thick IMC may not only spontaneously initiate fracture cracks under the action of residual stress but also provide a channel for crack propagation and aggravate the fracture trend of the joint. Therefore, to obtain a sound Al/Ti joint, the IMC thickness has to be kept to the lowest possible value [45].

4.3 Comparison of the single-wire CMT and CMT Twin processes

The CMT Twin can be used for high-speed welding under a relatively low heat input and avoid common defects such as undercuts and humps. According to the classical double ellipsoidal heat source model and a developed model for double-wire welding, the temperature field of the double-wire heat source is elongated along the welding direction, the temperature gradient during the welding process is small, the pool volume is increased, and the liquid metal can fully flow and wet the titanium alloy [46,47,48]. On the other hand, under the action of surface tension, the liquid metal flows from the center of the two heat sources to the edge of the molten pool at the same time, and the liquid convection is more stable, avoiding the appearance of the defects [49]. Thus, the CMT Twin welding speed can reach 2–3 times that of the conventional single-wire CMT, which is of great significance to realize low heat input, limit the growth of IMCs, and improve the joint strength.

Under the same heat input, the IMC of the CMT Twin joint is significantly thinner than that of the single-wire CMT, possibly attributed to the calculation equation of the heat input, the difference of interfacial thermal cycles, and the status of the welding pool. First, the heat input is calculated according to the linear energy formula Q = UI/V without considering the thermal efficiency η, which has a different value between 0.5 and 0.8 depending on actual welding methods. In addition, for the two welding sources (single-wire CMT and CMT Twin), the temperature field shapes, temperature gradients during cooling, and welding pool agitation are different, which will affect the formation and growth of the IMCs, resulting in different IMC thicknesses.

For previous studies of the traditional single-wire CMT on Al/Ti joining, the thickness of IMC obtained by R. Cao was 5 μm, and the tensile strength of the joint was 199 MPa [32]. In this research, the strength of the single-wire CMT joint is 225 MPa with an IMC thickness of 1.83 μm, while the highest strength of the CMT Twin joint can be up to 297 MPa with a minimum IMC thickness of only 0.53 μm. The CMT Twin shows advantages over the single-wire CMT in terms of a higher welding speed, lower heat input, thinner IMC thickness, and higher joint strength.

5 Conclusions

A CMT Twin technique was applied to Al/Ti joining and obtained a sound joint. With the CMT Twin, the welding heat input and IMC thickness were successfully controlled, and the joint strength increased to a new level that equals the strength of the base aluminum. The following conclusions can be drawn:

  1. (1)

    By employing the CMT Twin on Al/Ti joining, the reasonable range of the welding heat input was extended and shifted to a negative direction compared with the single-wire CMT.

  2. (2)

    TEM analysis of the Al/Ti welding-brazing interface showed a dominant layer of Ti (Al, Si)3, which grew in a serrated pattern from the titanium alloy to the weld seam.

  3. (3)

    By using the CMT Twin on Al/Ti welding, the thickness of the IMC was controlled to a minimum value of 0.53 μm, and the corresponding maximum tensile strength of the joint was up to 297 MPa.

  4. (4)

    With decreasing heat input, the IMC thickness decreased, while the joint strength increased, and the fracture position transferred from the IMC layer to the base aluminum.

  5. (5)

    The CMT Twin shows advantages over single-wire CMT in terms of higher welding speed, lower heat input, thinner IMC thickness, and higher joint strength.