Introduction

In the automotive industry, there are economic and environmental motivations to reduce the weight of vehicles, thus increasing their energy efficiency and reducing energy costs [1]. Replacing steel parts with lighter metals, such as aluminum and its alloys, is one of the major weight-saving techniques employed [2]. Hence, the joining of dissimilar materials (specifically, aluminum to steel) has been extensively investigated [3,4,5,6,7,8,9]. However, it is difficult to produce durable welds between aluminum and steel using conventional fusion welding, as the high heat input during this process results in the formation of large amounts of intermetallic compounds (IMCs) that can reduce the strength and durability of the weld. Therefore, a lower heat input is essential for producing a sound joint between aluminum and steel.

Friction stir welding (FSW) [10, 11] is a relatively new joining technique that can achieve high-quality welded joints; hence, it has been applied to various materials, including aluminum alloy [12,13,14], steel [15, 16], magnesium alloy [17, 18], titanium alloy [19, 20], and nickel-based alloy [21, 22]. Additionally, FSW has recently attracted major interest as a promising technique for joining aluminum to steel because it has several advantages over more standard techniques, such as a low heat input, short welding time, and superior mechanical properties of the weld [23,24,25,26,27]. Fukumoto et al. [28] reported that a dissimilar joint between cast aluminum alloy and mild steel had approximately the same strength as that of the aluminum base material, which was achieved by optimizing the offset of the pin periphery from the butt interface. Morita et al. [29] and Tanaka et al. [30] also reported that dissimilar FSW butt joints between aluminum alloys and steels exhibited good bending properties and excellent deep-drawing formability. However, abundant IMCs were formed and degraded mechanical properties were observed when the FSW conditions were not optimized. For example, Kimapong et al. [31] reported that the strength of an FSW joint with an IMC thickness of ~ 7 μm was ~ 60% lower than that with an IMC thickness below 1 μm. Moreover, Tanaka et al. [32] indicated that strength increased exponentially with a reduction in the IMC thickness, where the strengths of FSW joints with IMC thicknesses of ~ 1 μm and ~ 0.1 μm were approximately 100 MPa and 330 MPa, respectively. Therefore, it is necessary to inhibit IMC formation and elucidate the IMC growth mechanism during FSW in order to develop reliable and strong FSW joints between aluminum and steel.

Tanaka et al. [33] analyzed the IMC growth mechanism in FSW butt joints during static welding (i.e., wherein the tool was inserted into the plate, but not moved), which is a process similar to friction stir spot welding. Use of this method was necessary in order to maintain a constant temperature for a long enough period to investigate the IMC growth rate. In their study, it was shown that IMC growth was dominated by diffusion that was similar to diffusion bonding and roll bonding, where rapid IMC growth occurred for a brief period during the operation of the welding tool. However, this period of rapid IMC growth was not investigated in their study. As it is very important to clarify the IMC growth mechanism under actual FSW conditions, in this study, the rapid growth mechanism of IMCs during the operation of a welding tool was investigated by combining a previously proposed theoretical formula with experimental results from microstructural observations and analyses of welding temperature. Additionally, the predominant factors controlling IMC growth during FSW were considered.

Materials and methods

Sheets of mild steel and A1050 aluminum alloy, both with a thickness of 3 mm, were used in this study. Figure 1 presents diagrams of the dissimilar friction stir welding process; as is shown, the aluminum and mild steel plates were located on the retreating side and advancing side, respectively. The advancing side of the weld was the side on which the rotation of the tool proceeded in the same direction as the motion of the tool itself. The opposite side is referred to as the retreating side. Moreover, the probe center was offset from the butt interface so that the tool probe contacted the steel over a width of ~ 0.2 mm. Friction stir welding was performed at a tool rotation speed of 1800 rpm and a weld travel speed of 100 mm min−1. An FSW tool with a screw thread probe made of SKD61 tool steel was used with a shoulder diameter of 12 mm, probe diameter of 4 mm, and probe length of 2.9 mm, as shown in Fig. 2.

Figure 1
figure 1

Diagrams of the dissimilar friction stir welding (FSW) process

Full size image
Figure 2
figure 2

Diagrams of the method of measuring the welding temperature a of the probe (Tp) and b of the shoulder (Ts) using K-type thermocouples

Full size image

The inclination angle during FSW was 3°, and temperatures were measured via K-type thermocouples at the tool probe (Tp) and tool shoulder (Ts) using a non-contact temperature measurement system (FSW-2000 Series telemetry system; Megastir, USA). Microstructural observations of the interface of the weld were performed on cross sections cut perpendicular to the welding direction (region A in Fig. 3) and perpendicular to the normal direction of the sheets (region B in Fig. 3) using field emission electron microprobe analysis (EMPA; JXA-8530F, JEOL Ltd., Japan). The thickness of the IMC layer at the joint interface was calculated as the mean of ten IMC thicknesses randomly measured from scanning electron microscopy (SEM) images at 10000× magnification. In order to characterize the properties of the developed IMCs, transmission electron microscopy (TEM), bright-field imaging, and selected area electron diffraction (SAED) analyses were performed using a JEOL JEM-2000FX transmission electron microscope operated at 150 kV (Jeol Ltd., Japan). The TEM sample of the interface region was prepared using a focused ion beam (FIB; FB-2000, Hitachi High Technologies Inc., Japan).

Figure 3
figure 3

Schematic showing the regions of the welds observed using scanning electron microscopy (SEM)

Full size image

Results and discussion

Microstructure and growth behavior of IMCs

A typical surface morphology of a weld is shown in Fig. 4a. The weld was successfully joined without obvious porosity or defects, though some splashes of material from the plate were observed. Figure 4b shows a micrograph of the cross section perpendicular to the welding direction (region A in Fig. 2), while fine fragments of the mild steel were distributed in the aluminum region and no voids or cracks were formed at the joint interface or in the weld zone.

Figure 4
figure 4

a Top surface view of the weld area and b typical SEM image showing the microstructure of the cross section perpendicular to the welding direction. The horizontal dashed line indicates the plane cut for observing the cross section perpendicular to the normal direction of the sheets

Full size image

Figure 5a shows a typical TEM image of the microstructure of the IMC. The IMC layer consisted of dual sub-layers denoted as Phase 1 (adjacent to the aluminum) and Phase 2 (adjacent to the iron). Phase 1 was composed of the equiaxed and fine grains with a grain size of less than 100 nm. Meanwhile, elongated grains with a layer thickness of ~ 300 nm were observed in Phase 2. Figure 5b and c shows a high-magnification TEM micrograph from the region near Phase 1 and the SAED pattern of grain A, respectively. Figure 5d and e shows a high-magnification TEM micrograph from the region near the Phase 2 and the SAED pattern of grain B, respectively.

Figure 5
figure 5

a Typical transmission electron microscopy (TEM) image at the joint interface, b high-magnification TEM micrograph near Phase 1, c the corresponding selected area electron diffraction (SAED) pattern for grain A, d high-magnification TEM micrograph near Phase 2, and e the corresponding SAED pattern for grain B

Full size image

The SAED analysis of the patterns shown in Fig. 6c and e confirmed FeAl3 (Al-rich part) and Fe2Al5 structures (Fe-rich part), respectively. Ogura et al. [24], Tanaka et al. [34], and Sun et al. [35] reported the formation of an amorphous structure at the joint interface. However, similar to our results, Van der Rest et al. [36] and Movahedi et al. [37] reported that the IMC was composed of two sub-layers of FeAl3 and Fe2Al5. The microstructures and composition of the IMC at the interface thus depend on the heat input and degree of plastic deformation.

Figure 6
figure 6

a Typical SEM image showing the microstructure of the cross section perpendicular to the normal direction of the sheets. b Intermetallic compound (IMC) thickness profile as a function of distance from the center of the probe hole

Full size image

In this study, the tool rotation speed was higher than in previous studies in which the formation of an amorphous structure at the joint interface was reported [24, 34, 35]. Due to the higher heat input, no amorphous structure formed, but instead two IMCs formed at the joint interface. The existence of excessively thick IMC layers deteriorates joint properties; however, the IMC layer drives the metallurgical bonding of aluminum alloys and steels, resulting in a better weld quality [24, 38]. Although it is important to prevent IMC formation to improve joint properties, controlling IMC formation is difficult. Therefore, a method for controlling IMC growth during FSW was considered with respect to the IMC growth mechanism during FSW.

A micrograph of the cross section perpendicular to the normal direction of the sheets is shown in Fig. 6 and corresponds to region B in Fig. 2. The observed plane is shown by the dashed line in Fig. 4b. It was found that the welding line shifted toward the aluminum side after the tool probe contacted the steel (Fig. 6a). As shown in Fig. 3, the tool shoulder contacted and applied pressure to the steel plate during FSW due to the inclination angle of 3°. Therefore, the shift in the welding line indicates that the steel was plastically deformed by the tool shoulder pressure. Additionally, the effect of this pressure was found to be significant from a distance of ~ 2 mm from the center of the tool probe hole, as shown in Fig. 6a. Figure 7a and b shows high-magnification SEM images of the joint interface at distances of 1.7 and 5.8 mm from the center of the tool probe hole, respectively; these correspond to the observation positions X and Y shown in Fig. 6a. The mean IMC thicknesses shown in Fig. 7a and b are ~ 0.7 μm and 1.7 μm, respectively.

Figure 7
figure 7

Scanning electron microscopy images (backscattered electron imaging mode) showing the microstructure of positions aX and bY, as indicated in Fig. 5

Full size image

The relationship between the mean IMC thickness and standard deviation to the distance from the center of the tool probe is shown in Fig. 6b. Point 0 in Fig. 6b corresponds to the probe-hole center in Fig. 6a, where their distance scales are the same. It can be seen from Fig. 6b that IMC growth during FSW was not constant, but occurred in two rapid growth steps. First, IMC growth was observed immediately after the tool probe contacted the steel, and then growth was inhibited. Subsequently, a second rapid growth stage began 2.2 mm from the center of the probe hole, where plastic deformation of the steel under the pressure of the tool shoulder became significant (Fig. 6a). However, IMC growth was not observed over a distance of approximately 4.5 mm. As shown in Fig. 2, the tool used in this study had a curved 1 mm radius to the corner of the tool shoulder. Therefore, the lack of IMC growth is because the peripheral part of the tool was not subjected to the downward pressure of the tool shoulder on the steel plate due to the curvature relative to the tool corner. Additionally, the tool was extracted from the plates quickly after the tool arrived at the final location. The reduction in heat input was also expected to result in the reduction in the IMC growth rate. Consequently, the growth rate trend was consistent with the microstructural changes observed due to contact between the tool and steel plate.

Temperatures during FSW

It is well known that IMC thicknesses increase with increasing welding temperature. Thus, temperatures near the probe and underneath the tool shoulder (Ts and Tp, respectively) were measured in order to clarify the difference between the two rapid IMC growth processes. Figure 8 shows the two temperature profiles measured by K-type thermocouples. Both temperatures increased immediately after the welding tool came into contact with the plates, and remained nearly constant for ~ 6 s. Additionally, the temperatures were nearly unchanged while the tool was traveling along the weld. The Ts and Tp were approximately 780 K and 730 K, respectively. We interpret these results to mean that the temperature during the FSW of aluminum and steel is strongly influenced by contact between the tool and steel plate. As shown in Fig. 1, the tool probe contacted the side of the steel plate due to the tool offset, while the tool shoulder applied pressure to the surface of the steel plate predominantly due to the inclination angle (3°). Therefore, the different temperatures produced by these tool contacts appear to affect the IMC growth mechanism during FSW.

Figure 8
figure 8

Thermal history of the tool probe and shoulder

Full size image

IMC growth process during FSW

There have been several studies in which the IMC layer growth kinetics of aluminum and steel have been analyzed [39,40,41,42]. However, the growth mechanisms of IMC layers are diffusion bonding and isothermal heating, which are also favorable for temperature measurements; yet, very little information is available on IMC growth kinetics during FSW due to the difficulty of measuring temperatures synchronously. Tanaka et al. [33] reported on the growth rate of an IMC during FSW. The two rapid IMC growth steps revealed by microstructural observations in this study were considered from a growth kinetics perspective, using the temperatures measured experimentally and the diffusion rate equation. Tanaka et al. [33] also reported that the growth rate of an IMC during FSW could be calculated using the following equation, based on metallic diffusion:

$$ K = \frac{{d^{2} }}{t} = 9.6 \times 10^{ - 3} \exp \left( { - \frac{144000}{RT}} \right), $$
(1)

where K is the growth rate constant, d is the thickness of the IMC, t is the heating time, R is the universal gas constant [8.314 J/(mol. × K)], and T is the absolute temperature. The relationship between the mean IMC thickness and standard deviation to the welding time at temperatures of 730 K and 780 K was calculated using Eq. (1), and the results are shown in Fig. 9. The welding time was calculated based on the welding speed and distance from the center of the probe hole, while the IMC thicknesses measured from the SEM images are also plotted in this figure. As shown in Fig. 9, the experimental data for the first stage of rapid IMC growth agreed well with the calculated values for 730 K, which is the Tp in the region of 2 mm from the center of the probe hole. This indicates that the first rapid stage of growth was strongly influenced by contact with the tool probe.

Figure 9
figure 9

Square of IMC thickness as a function of the heating time and distance from the center of the probe hole

Full size image

In the region from 1.3 s (location of measured Tp) to 2.8 s (location of measured Ts), the temperature was expected to gradually increase from 730 to 780 K; however, in this study, the calculated IMC growth line is shown in Fig. 9, assuming that the temperature was 780 K. After ~ 1.3 s, the slope of the growth rate was similar to that calculated for a temperature of 780 K (Ts), thus demonstrating that the second growth stage was strongly influenced by the tool shoulder. These findings, which are based on the phenomenon of metallic diffusion, demonstrate that the two rapid growth processes occurred during FSW and were dominated by the contact between the tool probe and shoulder with the metal plates.

In this study, an experimentally obtained IMC growth rate equation during FSW was used in order to verify the existence of two rapid IMC growth processes, but was unable to clarify the IMC growth mechanism or to quantify the growth kinetics during the FSW of aluminum and steel because the temperatures measured experimentally may be inaccurate. It remains very difficult to measure accurate temperatures during FSW because the processed zone is not exposed to the surface and is under isothermal conditions. However, Wang et al. [39] reported that an Fe2Al5 layer grew faster than an FeAl3 layer because the diffusion of Fe atoms into Fe2Al5 is higher than that of Fe atoms into FeAl3. Additionally, they claimed that the local strain or stress introduced by the welding process contributed to either chemical reactions or interdiffusion because high deformation can provide higher diffusion vacancies and shorter diffusion distances. Therefore, further investigation is needed to clarify the IMC growth mechanism, as well as to quantify the growth kinetics during the FSW of aluminum and steel.

Conclusions

The rapid growth of the IMC during the dissimilar FSW of aluminum alloy and steel was divided into two steps according to the tool contact. Contact between the side of the steel plate and the tool probe caused the temperature to increase, resulting in the first stage of rapid growth. Friction between the tool shoulder and surface of the steel plate also caused a temperature increase that resulted in the second stage of rapid growth. Since the temperature during the second stage was higher than in first, it appears most important to control contact between the tool shoulder and steel plate, along with the plunging pressure, in order to inhibit the growth of IMCs during FSW. Such a strategy could help improve the strength of welds during the fabrication of lightweight materials in the automotive industry. Furthermore, these results could be applied to other dissimilar FSW welds that form IMCs at joint interfaces and thus could be valuable for various industrial applications.