Effect of tool pin length on interfacial behaviors and mechanical properties of friction stir-welded lap joints of 430/304 stainless steels

In this study, the effect of tool pin length on friction stir lap welding of 430 and 304 stainless steels has been investigated. The thermal history of interface was measured by thermocouple. The interfacial macro- and micro-structures under different tool pin lengths were analyzed via optical microscope and scanning electron microscope. Also, tensile shear tests were carried out to evaluate the strength of the lap joint. The results show that increasing the pin length from 0.5 to 1.3 mm increases the peak temperature of the interface, which has a significant effect on diffusion layer formation and hook geometry. The fracture mode of the joints shifts from shear fracture to tensile fracture with the increase of the pin insertion depth. When the pin length is 1.0 mm, i.e., a slight penetration into the 304 plate, the joint presents a maximum failure load of 2.86 kN. The findings reveal that the effective top plate thickness (Teff) used to describe the hook geometry is an important factor determining the joint strength and fracture mode.


Introduction
Ferritic stainless steel has low price and good stress corrosion resistance, and austenitic stainless steel has good plasticity, toughness, and high temperature stability. In practical application, austenitic stainless steel can be used in components that require high temperature stability, while ferritic stainless steel can be used in components that require low mechanical properties. The welding components of austenitic-ferritic stainless steels are mainly used in petrochemical, nuclear power plant, power plant, and automobile production, etc. [1,2].
There are many problems in welding dissimilar stainless steels by conventional fusion welding methods. For example, directional columnar crystals are easily formed in austenitic stainless steel, which will promote impurity segregation and produce solidification cracks. Ferritic grains in the heataffected zone tend to grow, which will lead to joint embrittlement [3]. In addition, it is difficult to control the heat input when welding thin plates by fusion welding, resulting in weld penetration and large weld deformation [4,5]. As a solid-phase joining technology, friction stir welding (FSW) has relatively low heat input and can effectively avoid the deformation in the thin plate during welding. In recent years, FSW has been employed to form a lap joint of dissimilar metals as a friction stir lap welding (FSLW) process. This method can avoid overheating of the lower-melting temperature alloy and wear on the tool in lap configurations, and obtain higher joint strength than the butt joints.
Chen et al. [6] implemented the welding of aluminum and magnesium sheets in lap configuration and obtained a sound joint. In lap joints of aluminum alloy and commercially pure copper [7,8], the tool pin was inserted into the copper plate which was placed under the aluminum plate, and the joint strength was close to that of aluminum alloy due to the combination of mechanical mixing and metallurgical bonding. When welding soft base metals by FSLW, the tool pin can be easily inserted into the lower plate, and it can obtain an integrated joint.
For lap joints made of soft and hard metals, the normal tool pin is usually not inserted into the bottom hard metal in order to reduce the wear of the pin. With this consideration, the FSLW of aluminum alloy and pure titanium is conducted by inserting the pin tip into the soft aluminum alloy, and the shear strength of the join could reach 62% of that of aluminum alloy [9]. Elrefaey et al. [10] investigated pure aluminum and low carbon steel lap welding and found that the depth of the pin tip to the bottom plate surface has the effects on the joint strength. When the pin depth did not reach the steel surface, the joint failed under low applied loads because of the interface cracks, and slight penetration of the pin tip to the steel surface increased the joint strength but only 89 MPa. It is difficult to obtain a dissimilar metals lap joint with good bonding quality using the smooth tool pin which could not be inserted into the bottom hard metal because of the wear of the pin. On the basis of this, Wei et al. [11] made lap joints of dissimilar metals of 1060 aluminum alloy and SUS321 stainless steel by using a stir tool consisting of a cutting pin and a concave shoulder. When the pin tip penetrated through the aluminum alloy and was inserted into the stainless steel to generate the plastic deformation, it was formed the defect-free joining at the interface with a visible mixed layer, and the strength of the joint is significantly improved. Patterson et al. [12] implemented FSLW of aluminum and steel by the scribe technology coupling the FSW tool with a carbide cutting tool which is responsible for machining the harder metal, and high quality joints were obtained by the thermo-mechanical effects at the interface. Therefore, the tool pin geometry also plays an important role in obtaining high strength joints.
Adibeig et al. [13] carried out the FSW on poly methyl methacrylate (PMMA) by using a tool with a double-step shoulder. This new designed tool could compress the region near the weld line and prevent excess material from escaping. On the basis of this, they obtained welded joints with excellent mechanical properties by optimizing the process parameters.
In the recent studies on FSLW between soft-soft metals and soft-hard metals, the weld quality of the lap joints was improved by designing the stir tool geometry and the pin penetration depth. This is related with the interfacial microstructures which were dominated by the heat input and metal deformation. In this study, both ferritic and austenitic stainless steels have higher melting point and greater hardness. At present, there is no evidence that the above findings are applied to hard-hard metals FSLW joints. In order to control the heat input and interface behavior in the weld, the insertion depth of the pin can be varied in a wide range by designing tool pins of different lengths. In view of this, the present work focuses on the interfacial behaviors and mechanical properties of dissimilar lap joints of 430/304 stainless steels fabricated by different types of tool pins.

Experimental procedures
Three hundred four austenitic stainless steel and 430 ferritic stainless steel plates with 1-mm thickness were used in present study. The chemical compositions of base metals are listed in Table 1. Figure 1 illustrates the welding process. Two base metal plates were assembled into a lap joint, where the 430 plate was placed on the top because of its relatively low hardness. The two plates were welded by JIBIC-GWM850 FSW machine with configurations of the output power of 15 kW, the maximum rotational speed of 8000 rpm, and the maximum welding speed of 3000 mm/min. To carry out the welding process on stainless steels, a WC-based tool (90% WC + 10% Co) with 18-mm diameter shoulder was used. The tool pin was tapered from 6 mm at the pin root to 5 mm at the pin tip. The welding was carried out using 1200-rpm rotational speed, 50-mm/min welding speed, 0.1-mm plunge depth of the shoulder, and 0° angle of tilt.
When other welding parameters are constant, the depth of the pin inserted into the plates has a significant influence on the volume of the softened zone and the plastic flow of the material. For lap joints, whether the two base metals can form a reliable connection depends on the degree of plastic deformation and metallurgical reaction at the interface. In  this study, the depth of the tool pin inserted into the welded material was controlled by changing the length of the pin. According to the thickness of the welded plates, the pin length is set as 0.5, 1.0, and 1.3 mm. Accordingly, the pin end is inserted into the upper plate, lower plate surface, and lower plate respectively, as shown in Fig. 2. For the FSLW process, the thermal history has a significant effect on the welding results. In this study, K-type thermocouple close to the lower plate is used to monitor the real-time temperature variation along the tool traveling direction. Figure 3 gives the location of the thermocouple, 60 mm far away from the initial position of FSLW. After FSLW process, the joints were sampled by wireelectrode cutting for measuring the mechanical properties and preparing the metallographic specimens. After grinding and polishing, the metallographic specimens were etched in the mixture of 5 ml FeCl 3 + 5 ml HCl + 15 ml H 2 O. The microstructures of the specimens were observed under an Olympus GX51 optical microscope (OM). The elements of the interface were analyzed using a ZEISS field emission scanning electron microscope (SEM) and an energy-dispersion spectrum (EDS). The X-ray diffraction (XRD) was performed inside an X-Pert Pro MPD with an angle from 20 to 90° for the phases. The microhardness distribution of the joint interface was examined using a TMVP-1 at an interval of 0.2 mm, with 0.98 N load and 10-s hold time. The tensile shear test was performed using an UTM5305 electronic universal testing machine at a rate of 1.0 mm/min.

Thermal histories during FSLW
During the FSLW process, when the length of the tool pin is different, the plastic deformation degree, element diffusion, and metallurgical reaction at the interface are different. These interfacial behaviors have great influence on the bonding strength of the joint. Because the interfacial temperature variation is a key factor in the interfacial reaction, it is worth to analyze the heat generation mechanism during the FSLW process. The heating sources are mainly derived from three parts, namely the friction heat generated between the shoulder and the upper surface of the 430 plate, the friction heat generated between the pin and the inner metal, and the plastic deformation heat of metal in the stirred region. The real-time temperature variations curves under different pin lengths but the invariable welding parameters are given in Fig. 4.
It can be seen from Fig. 4 that the pin length has obviously effect on the temperature variation close to the interface. Based on the heat generation model of conical tool pin [14], the heat generation power of the whole welding process is deduced, as shown in Eq. (1).
where μ is the friction coefficient between the stir tool and base materials, n and P are the rotational speed and welding pressure during FSLW. R 1 , R 2 , and R 3 are the shoulder radius, the root radius, and top end radius of the pin respectively, and L is the pin length. Also, W shoulder and W pin are the heat generation powers of the shoulder and the pin, and W  1), W has the positive correlations with L, suggesting that the heat input increase with increasing the depth of pin inserted into base metals. Therefore, the relationship between the pin length and the peak temperature during FSLP is verified as well. In addition, the holding time at high temperature is also different, which affects the formation of the interfacial reaction layer.

Macroscopic structures of joints
A characteristic feature of the friction stir lap-welded joint is the formation of a geometrical defect originating at the interface of the two welded sheets, called as "hook," which has a significant effect on joint strength [15]. Schematic diagram of FSLW joint is shown in Fig. 5. Song et al. [16] believed that this structure was mainly generated on both sides of the pin tip plane. During welding, the softening degree of the lower layer material was low, and it squeezed upward the upper layer material with a higher softening degree, which resulted in bending to form a hook structure. From this, they proposed the suck-squeeze theory. Liu et al. [17] studied FSLW of 7B04-T74 aluminum alloy and found that the formation of hook defect was caused by the combined effect of the shoulder and the pin. No matter which formation theory is related to the different softening degree of materials in each region during welding, and softened materials are extruded to form a curved hook structure. The curved lap interface morphology generated on advancing side and/or retreating side plays a decisive role in the effective top plate thickness (T eff ) and effective bonding width (W eff ), and the T eff usually used to describe the hook geometry (H h ) [16]. Figure 6a shows the cross-sectional macromorphology of FSLWed 430/304 stainless steels joint obtained by 0.5mm long pin. Since the 430 plate is 1 mm thick, the pin tip was inserted into the upper plate without reaching the surface of the lower plate. Thus, the contact area between the pin surface and 430 steel is small, and the limited friction heat results in the peak temperature of the interface below 884 ℃, shown in Fig. 4. Because the plastic deformation only occurred in 430 steel, the joint still retained a straight interface.
When the pin length is 1.0 mm, having a slight penetration into the 304 lower plate, the peak temperature of the interface increases to about 1100 ℃. Due to the proper heat input, the two plates were tightly bonded without holes at the interface, as shown in Fig. 6b. Squeezed and stirred by the tool pin, a low step was formed at the interface of advancing side. The 304 steel on retreating side was squeezed upward into the 430 steel because the upper plate had a higher degree of softening, and it then bent towards the stir zone to form a hooked structure with a height (H h ) of 0.76 mm. Thus, T eff (0.34 mm) and W eff (2.7 mm) were formed at the interface of the two welded plates. Suitable hook size can produce a mechanical interlocking effect, which plays a key role in the tensile shear performance of the joint.
When the pin length was increased to 1.3 mm, the pin tip penetrated through the 430 plate and was inserted into the 304 plate. As a result, the metals on advancing side present obvious streamline structure, and ferrite grains are smaller, as shown in Fig. 6c. But holes were formed on advancing side; this is because the soften 430 stainless steel was subjected to greater resistance when it flowed. When the pin was inserted deeper into the lower plate, the peak temperature at the interface reached 1176 ℃, and the metal was excessively softened. The large 304 steel particles extruded by the pin were further embedded into the softer 430 base metal. As a result, lager step and hook were formed on the advancing side and retreating side respectively, and the H h is 1.17 mm. Meanwhile, the T eff is decreased to 0.18 mm and, the W eff is almost unchanged, which is 2.75 mm. Obviously, the deeper insertion of the tool pin led to an increase in the hook geometry and a decrease in effective top plate thickness. Also, the 304 steel has a greater linear expansion coefficient. These led to serious deformation, which deteriorated of the joint performances.   Figure 7 is the enlarged view of interface region marked with white rectangular boxes in Fig. 6a. The metal of the interface did not undergo plastic deformation and dynamic recrystallization, so ferrite grains were as coarse as the parent metal. The EDS results of elements at the interface are listed in Table 2. From the content of Ni at positions A-C, the Ni content in 304 steel near the interface is 7.25 wt.%, slightly less than that in 304 base metal (8.08 wt.%), and the Ni content in the interface layer and near 430 steel are 1.13 and 0.63 wt.%, respectively. Due to the small insertion depth of the pin, the heat generated at the interface is limited, and the thermal activation energy is small, so a slight diffusion of Ni atom occurred in the interface region. As a result, a diffusion layer of 0.8 μm was formed at the interface. In addition, small pressure and low temperature at the interface caused micropores, which could damage the joint strength. Figure 8 shows the interface structure produced by 1.0mm long pin and its location is marked in Fig. 6b. Under the pressure applied by the pin end, the 304 steel at the interface of the advancing side generated plastic deformation and formed a 50-μm high step (shown in Fig. 8a). As shown in Fig. 8b, the banded structures on retreating side were bent towards the weld center, which is due to the higher softening 430 steel being squeezed by large volume of 304 steel. During welding, 430 steel deformed continuously with the rotation of the tool pin to form a strip structure. The streamline structures indicated that 430 steel reached a thermoplastic state and was well mixed with 304 steel. Figure 8c gives the interface characteristics of the stir zone. It is observed that two base metals are tightly bonded, and the grains at the interface are refined due to dynamic recrystallization. EDS analysis results of elements at the interface are listed in Table 2. Compared with the interface of 0.5-mm long pin, the contents of elements in the interface region are higher. White precipitates discontinuously distributed at ferrite grain boundaries (shown by red arrows). The XRD test results shown in Fig. 9 confirm that the second phase particles are Cr 23 C 6 and Cr 2 N. During the holding time of high temperature, C, N, and Cr atoms rapidly moved towards ferrite grain boundaries and formed second phases. According to the EDS and XRD results, W element was detected, indicating that severe friction occurred between the pin and the base metals. Meanwhile, a small amount of martensite was formed in the interface region. Thus, a 1.6μm thick mixed layer composed of α-Fe, γ-Fe solid solution, and precipitated phases was formed at the interface.

Microstructure morphology and bonding mechanism of interfacial region
When the pin length is 1.3 mm, the metal near the interface underwent severe plastic deformation, resulting in a higher step on advancing side and a larger hook on retreating side, shown in Fig. 10a and b. Besides, holes are found on both sides of the interface and reduce the effective bearing area of lap joint. The interface of stir zone has the characteristic of tight bonding, as shown in Fig. 10c. It can be seen from Table 2 that the diffusion of elements, especially the change of Ni content (0.63-6.31 wt.%), is obvious from 304 to 430 steel near the interface. It is confirmed that the diffusion coefficient increases with high strain rate and high temperature, and a diffusion layer of 1.7-μm thickness was formed. During welding, the top surface of the tool pin was obviously worn, as shown in Fig. 11.
As mentioned above, the bonding strength of lap joint mainly depends on the mechanical mixing degree of base metals and metallurgical reaction of the interface. The formation process of interface structure of 430/304 stainless steels lap joint is shown in Fig. 12. On the one hand, when increasing the insertion depth of the pin, the deformation degree of the metal is enhanced, resulting in the increase of the H h . However, too large hooks  where D 0 is diffusion constant, Q is diffusion activation energy, R is gas constant, T is thermodynamic temperature, and D is diffusion coefficient.
Under the combined action of "stirring-extrusion" of the tool pin, defects such as vacancy and dislocation are generated in the interfacial microstructure, and lattice distortion occurs in the metal. High energy and small diffusion activation energy are beneficial to the diffusion of elements, and friction heat and deformation heat are generated at the same time. According to formula (2), the higher the temperature T is, the greater the diffusion coefficient D is. These conditions are conducive to the diffusion of elements at the interface. Therefore, the shorter the pin is, the lower the interface temperature is, and the elements diffusion at the interface is not sufficient. The longer pins increase the heat input and the element diffusion. Combined with the microstructures of the joints, the thickness of diffusion layer at the interface is increased with the pin length.

Mechanical properties of the joints
The microhardness test was carried out on the cross section of the joint along the thickness direction. The hardness distribution of weld under the pin of different lengths is shown in Fig. 13. Hardness of the joint bottom is the same as that of the base metal because the welding process has little  First, the metal in the interface area was extruded by the pin to produce work hardening. Second, dynamic recrystallization resulted in microstructure refinement. Also, precipitates such as carbides, nitrides, and martensite were formed in the interface area. The hardness of 430 steel side decreased slightly, and the hardness value changed greatly due to the different deformation degree in different areas of 430 steel. Figure 14 shows the fracture morphology and corresponding failure loads of lap joints produced by pins of different lengths. The fractured sample of 0.5-mm long pin failed at the weld interface displayed in Fig. 14a, and its shear failure load was only 1.72kN. A noticeable growth in the tool pin length had a significant influence on the strength of the lap joints. The sample of 1.0-mm long pin fractured in 430 base metal and the maximum failure load of 2.86 kN was obtained. When slightly increasing the pin length to 1.3 mm, the sample fractured in 430 stainless steel of the weld zone, where the crack propagated along the minimum plate thickness (T eff ) and the failure load decreased to 2.62 kN. It is noted that the difference in failure loads is significant with increasing the pin length, which is related to the H h , T eff , and W eff . Badarinarayan et al. [15] and Bozzi et al. [18] also reported that the hook geometry had significant effect on the failure load of the joint.
In this work, for the joint with 0.5-mm long pin as shown in Fig. 6a, no hook was found at weld interface, and only a very thin diffusion layer was formed at the interface of the lap joint. Thus, the weak bonding strength led to the shear fracture. With increasing the pin length from 0.5 to 1.3 mm, the interface behavior during welding changes from simple metallurgical reaction to combined metallurgical reaction and macro-interlocking. The fracture mode shifts from shear fracture to tensile fracture. In the joints of 1.0-and 1.3-mm long pin, the bonded zone width (W eff ) is approximately equal, while the joint of 1.0-mm pin has a small H h but a large T eff , indicating it has greater resistance to tensile loading. Badarinarayan et al. [19] also found that the T eff offered the resistance to tensile loading. Song et al. [16] obtained similar result that the T eff , rather than the W eff , is the major effect on the failure load. Thus, the T eff is the predominant factor that determines the strength of lap joint.
In this study, two different fracture modes are observed during tensile shear loading: shear fracture under 0.5-mm long pin, and tensile fracture under 1.0-and 1.3-mm long pins. This can be directly attributed to the weld geometry, and the fracture mode was governed mainly by the hook height (H h ) or effective top plate thickness (T eff ). As the tool