2019 Volume 60 Issue 12 Pages 2506-2515
Friction stir welding (FSW) of high tensile strength steel plate formed by cold-rolling (SPFC980) was conducted by using SiAlON tools. Characteristics of SPFC980 work-pieces after FSW tests and durability of the SiAlON tools were evaluated. FSW tests were done under various tool rotation speed (VR) and traveling one (VT) to search appropriate FSW conditions. Sound joint conditions were found when temperature measured at the shoulder part of the tool ranged from 686 to 1080°C. The durability of SiAlON tools was strongly influenced by the measured shoulder temperatures. The wear of the shoulder progressed rapidly with increasing travel distance when the measured shoulder temperature reached more than 1080°C. On the other hand, the wear progress was hardly observed in the case when the average measured shoulder temperature was less than 700°C. The most likely mechanism of the tool wear was elimination of grain clusters due to softening of glass phases contained in the SiAlON tool at high temperatures. It was concluded that the measured shoulder temperatures ranging from 680 to 800°C were effective for SiAlON tools to be used for FSW of SPFC980.
Friction stir welding (FSW) is a welding technique developed in 1991.1) Frictional heat originating from tool rotation induces the plastic flow at the welding point, which enables solid-state welding of the work-piece. The FSW technique is an eco-friendly one because it reduces dusts and fumes compared to conventional fusion welding techniques such as arc welding. To take advantage of the above, the FSW technique has been mainly applied to the welding of light metals such as aluminum and its alloys.2–11)
In the last decade, concern about the welding of steel has been growing especially among automobile makers. They are trying to reduce cars’ weight to improve the fuel efficiency. High tensile strength steel plate formed by cold-rolling (SPFC) is one of the key materials to reduce the body weight because SPFC keeps high strength with a thin plate shape. In addition to the light weight, SPFC possesses high impact energy absorption capability for ensuring the cars’ crash safety. The welding of SPFC, therefore, is an important task for producing cars having good fuel efficiency and crash safety.
There are several studies on FSW of iron-based materials such as high carbon steel and stainless steel by using tools made of Ir alloys,12,13) Ni alloys,14) W-Re,15–25) cemented carbides,26–32) polycrystalline cubic boron nitride (PCBN),24,25,33–36) W-Mo,37) and Si3N4.38–41) Because iron-based materials have higher melting points than aluminum alloys, the tool material must have excellent wear resistance at high temperatures against iron-based work material for long distance stable welding. The tool materials described above have individual weak points such as they are very expensive or have low heat resistance.
SiAlON is one of the high-temperature refractory ceramics based on the elements of silicon (Si), aluminum (Al), oxygen (O), and nitrogen (N). Solid solution of Al and O elements in Si3N442,43) improves strength, thermal shock resistance, wear resistance, and oxidation resistance at high temperatures around 1000°C. Utilizing a SiAlON tool for FSW of SPFC may be a good way if considering the balance between production cost and heat resistance. In this study, we conduct FSW tests of SPFC using SiAlON tools under various conditions.
The SPFC products are divided into several types by tensile strength (namely, 440, 590, 780, 980 MPa). SPFC980 is desirable for light-weighting of automobiles because it especially has high tensile strength. On the other side, very strong force is applied to the tool during FSW of SPFC980. To weaken that force, the hardness of the work-piece during FSW should be reduced. This hardness strongly depends on welding temperatures, which can be controlled by changing several parameters of FSW conditions such as rotation speed and traveling speed of the tool. Sufficiently high (but not too high) welding temperatures facilitate the plastic flow of the work-piece, which leads to reducing the possibility of defects in the welded area. On the other hand, high-temperature welding may influence the mechanical properties of the work-piece especially in the welded area due to microstructural change accompanied with the formation of martensite phases.44) In addition to the influence on work-pieces, the welding temperature also influences the durability of SiAlON tools against tool oxidation and/or chemical reaction between the tool and work-piece.
Here, we investigate FSW of SPFC980 using SiAlON tools, which includes appropriate FSW conditions from the mechanical properties and microstructure of welded SPFC980 work-pieces, and evaluation for the durability of SiAlON tools in those FSW conditions.
SPFC980 was prepared as a work-piece for FSW. The dimensions of the work-piece were 120 mm in length, 50 mm in width, and 1.2 mm in thickness. Concentrations of elements contained in a work-piece are shown in Table 1.
Figure 1 shows the microstructure of SPFC980 as the base material. The microstructure was observed by an optical microscope. SPFC980 has a dual phase: a main ferrite matrix and martensite as a secondary phase. However, a clear martensite phase which often could be observed due to the quenching effect45) was not observed.
Microstructure of SPFC980 as the base material.
Figure 2 shows a schematic illustration of FSW tools used in this study. The tool material was SiAlON (SX9, NGK SPARK PLUG Co., Ltd., Japan). The physical properties of the SiAlON used in this study are shown in Table 2. The diameter of the shoulder was 6 mm. The probe shape was a circular truncated cone. The diameters of the top surface and undersurface and probe height were 2 mm, 3 mm, and 1 mm, respectively. Figure 3 shows a schematic illustration of FSW tests. The entire tool assembly (tool and its holder) was inserted into a 1D FSW machine (1D-FSW, Hitachi Power Solutions Co., Ltd.), and a work-piece was clamped on the machine table.
Schematic illustration of FSW tools used in this study.
Schematic illustration of FSW tests.
The clockwise-rotating tool was plunged (1.0 mm/min) into the work-piece. After contacting the tool shoulder with the surface of the work-piece, the tool assembly traveled along the center line of the work-piece in the longitudinal direction. The tilting angle (the angle of the tool against the vertical line of the work-piece) and the tool travel distance per one welding were 3° and 85 mm, respectively. The tool rotation and traveling speeds were varied to explore appropriate FSW conditions. The temperature at the welding surface (tip of shoulder) of the tool was measured simultaneously by a radiation thermometer (FTK9Y-R220LA-50S11, Japan Sensor Corporation). The measuring spot of the thermometer was 1.5 mm in diameter and was focused on the tool shoulder part just above the surface of the work-piece throughout the tool traveling. The emissivity in the thermometer was set to be 0.9 as the maker’s recommendation value for Si3N4. In this study, the average value of measured temperatures obtained by the above method during the tool traveling was defined as the measured shoulder temperature. It is noted that the measured shoulder temperature is different from the peak welding temperature which can be estimated from the microstructure of stir zone in the welded work-piece. In some conditions, FSW tests were iterated up to 10 times by changing the welded work-piece to a new one.
The welded work-piece was cut into tensile specimens as shown in Fig. 4. Four specimens having a stirred part were obtained from a work-piece. One specimen was cut from the work-piece without the stirred part for evaluating the joint strength. When the tensile strength of the four specimens or the joint strength was not less than 90% of tensile strength for the base material (not-stirred part), the corresponding FSW condition was judged to be appropriate.
Information of specimens for tensile tests (a) Cut location of specimens from a work piece (b) Illustration of a specimen for tensile tests.
After FSW tests, microhardness, macrostructure, and microstructure of welded SPFC980 were evaluated. Microhardness profiles across the welded center were obtained from micro-indentations as shown in Fig. 5. The advancing side and retreating side are denoted as A.S. and R.S. in Fig. 5, respectively. Indentations were formed every 500 µm at depths of 0.3, 0.6, and 0.9 from the surface and at 200 gf load. The macrostructure of the SPFC980 was observed by a 3D measurement macroscope having a close-up function (VR-3000, KEYENCE Corporation, Japan). The microstructure of the stirred zone was observed by an optical microscope. The welded surface of the specimens was mechanically polished and then etched with a Nital solution. Mechanical polishing was performed by grinding with sandpaper of #1500 mesh and then polishing with a diamond paste of 3, 1, and 0.25 µm granulation.
Location of micro-indentations on the specimen for obtaining microhardness profile (A.S.: advancing side, R.S.: retreating side).
The durability of a SiAlON tool was evaluated for three FSW conditions at which joint strength indicated not less than 90% of tensile strength for the base material. Tool wear was measured by using the same macroscope as macrostructure measurements of the SPFC980. The change in tool appearance relating to the tool wear was observed and the distance between the tip of the probe and surface of the shoulder was measured. Microstructural observation and elemental analysis on the shoulder part after FSW tests were also conducted by using a scanning electron microscope (ERA-8900FE, ELIONIX INC., Japan) including energy disperse spectroscopy (Genesis EDAX Inc., USA). From the above analysis, the probability of oxidation and reaction of the tool material with the contents of the work-piece were evaluated.
Vickers hardness of the tool and the work-piece was measured in the temperature range from room temperature to 1200°C in the tool and to 800°C in the work-piece. These hardness data were useful to discuss the relationship between tool wear and the measured shoulder temperatures.
Figure 6 shows the welding parameter window for FSW of SPFC980. Tool rotation speed (VR) and traveling speed (VT) are used as welding parameters. The circles in Fig. 6 correspond to the appropriate conditions that the joint strength is not less than 90% of tensile strength for the base material, whereas the triangles correspond to below 90% of that. The threshold value of VT became larger with increasing VR, which positively correlates with frictional heat. Table 3 shows the welding parameters and measured shoulder temperatures in the conducted FSW tests.
The welding parameter window for FSW of SPFC980.
From Table 3, the measured shoulder temperature increased with decreasing VT and/or increasing VR. It was noted that appropriate FSW conditions were obtained from a wide temperature range between 686°C and 1080°C. Such a wide temperature range leading to sufficient welding strength is difficult with conventional cemented carbide tools26) especially in the case of high VR because cemented carbide tools would be fractured or deformed plastically due to high temperatures. At least it is suggested that a SiAlON tool persists under a wide range of welding temperatures from 686 to 1080°C without any tool fracture.
3.2 Evaluation of SPFC980 work-piecesFigure 7 shows the macrostructures in transverse section of welded SPFC980 specimens. The advancing side and retreating side are denoted as A.S. and R.S. in Fig. 7, respectively. The dark contrast in the center of the macrostructure corresponds to the stir zone and the lighter border lines correspond to the heat-affected zone. These border lines tended to move inside with decreasing VR and/or increasing VT, which is caused by a decrease in the frictional heat by unit time and volume.
The macrostructures in transverse section of welded SPFC980 specimen (a) (VR, VT) = (800, 100), (b) (VR, VT) = (1000, 100), (c) (VR, VT) = (1200, 100), (d) (VR, VT) = (1200, 200), (e) (VR, VT) = (1800, 200), (f) (VR, VT) = (1800, 300), (g) (VR, VT) = (2400, 300), (h) (VR, VT) = (3000, 300), (i) (VR, VT) = (3000, 500), and (j) (VR, VT) = (3000, 700). VR: rotation speed (rpm), VT: traveling speed (mm/min).
In some conditions, defects were observed (circles or ellipses in Fig. 7), indicating that the corresponding FSW condition was not optimum. Three categories of defects outside the optimum FSW conditions were reported in other articles.46) Among these categories, a cavity of groove-like defect caused by insufficient heat was most likely because a defect tended to appear in the case of relatively lower VR and/or higher VT, which suggests insufficient heat input.
Figure 8 shows microhardness (Hv) profiles in transverse sections of the welded area for three representative appropriate FSW conditions. Here, we denote the three representative FSW conditions as “High”, “Middle”, and “Low”. The corresponding VR and VT are shown in the figure caption. They are categorized by the measured shoulder temperatures. Concretely, the “High” condition is the one when the measured shoulder temperature exceeds 1050°C, the “Middle” condition is the one when the temperature ranges from 800 to 1050°C, and the “Low” condition is the one when the temperature is less than 800°C. Hereafter, various items are evaluated among the three conditions. It was clear that a harder area compared with the base material was formed around the stir zone for all three different FSW conditions. This area is a typical characteristic for FSW of steel alloys.47) The width of the harder area tended to decrease with decreasing measured shoulder temperature. The Hv of the harder area ranged from 450 to 550 kgf/mm2 for all three conditions. In the “High” condition, the width of the harder area hardly varied when the measured positions changed from upper to middle or lower positions (Fig. 8(a)). In the “Middle” and “Low” conditions, on the other hand, it became narrower with shifting the measured positions downward (Fig. 8(b) and (c)). The formation of the harder area suggested that martensite phases were transformed from austenite phases which had transformed from ferrite phases by frictional heating.
Microhardness profiles in transverse sections of the welded area for three representative appropriate FSW conditions (a) “High” condition; (VR, VT) = (2400, 300) (b) “Middle” condition; (VR, VT) = (1800, 200) (c) “Low” condition; (VR, VT) = (1000, 100).
Figure 9 shows microstructures of the stirred zone in SPFC980 specimens. The observed position was the center of the stir zone in the cross-section of the welded work-piece as shown in Fig. 9(d). The microstructure of all the three conditions had lath martensite phases which were not observed in the base material. This indicated that the peak welding temperature exceeded the transformation temperature from ferrite to austenite phases which subsequently transformed into martensite phases in the cooling process for all the three FSW conditions. The transformation temperature from ferrite to austenite phases, AC1 (°C), is given from the mass present of various elements by the following empirical equation:48)
| \begin{align} A_{C1}& = 723 - \text{10.7Mn} - \text{16.9Ni} + \text{29.1Si} + \text{16.9Cr} \\ &\quad + 290A_{S} + \text{638W}. \end{align} | (1) |
Microstructures of the stirred zone in SPFC980 specimens (a) “High” condition (b) “Middle” condition (c) “Low” condition.
The formation of martensite phases not only depends on the temperature of AC1 but also the cooling rate. Minimum cooling time from 800 to 600°C for no martensite phase formation, tB (s), is given as below:49)
| \begin{align} t_{B} &= \exp [6.2 (C_{p} + \text{Mn/3.6} + \text{Cu/20} + \text{Ni/9} \\ &\quad + \text{Cr/5} + \text{Mo/4}) + 0.74] \end{align} | (2) |
The tool wear strongly depends on the shoulder temperature. Among the several tool parts, the shoulder part is subjected to the highest temperatures and worn out prior to other parts. Figure 10 shows configurations of the tool top as a function of the iteration number of the FSW tests. The maximum iteration number was ten for the three FSW conditions, which corresponds to 850 mm of travel distance (TD). The x-axis corresponds to the diametrical direction of the tool and the z-axis to the length direction. The probe length did not change with increasing iteration number, indicating that tool length was not changed for the three representative conditions, although probe diameter slightly decreased in the “High” condition. On the other hand, the wear of the shoulder part apparently progressed with increasing iteration number except for the case of the “Low” condition.
Configurations of the tool top as a function of the iteration number of the FSW tests (a) “High” condition (b) “Middle” condition (c) “Low” condition.
Figure 11 shows wear amount as a function of travel distance for the three representative FSW conditions. The wear amount (WA) of the tool was defined as the disappearance amount (depth) of the shoulder part. In the “High” and “Middle” conditions, the wear amount became larger with increasing travel distance (TD), whereas wear was hardly observed in the “Low” condition. Such difference was apparently based upon the difference of the shoulder temperatures. When the measured shoulder temperatures were below 700°C, the shoulder part was not worn for a short distance of 850 mm. In contrast, the wear of the shoulder part progressed when the measured welding temperatures reached more than 850°C.
Wear amount as a function of travel distance for the three representative FSW conditions.
On the other hand, mechanical force applied to the tool did not influence the tool wear so much in this study. Figure 12 shows the representative torque data in the direction of tool traveling for the three representative FSW conditions. The torque (Tr) data were acquired by converting the current passing through a motor monitored by a sequencer. The “High” and “Middle” conditions exhibited weak torque oscillation, whereas the oscillation of the torque was strong in the “Low” condition compared to the other conditions. Torque oscillation directly links to the oscillation of mechanical force applied to the tool. The largest difference between peak or valley mechanical force and the average one may promote tool wear. Among the three conditions, the “Low” condition had the highest difference (approximately 0.3 Nm), although tool wear was hardly observed in this condition. This result suggests that the difference in peak or valley torque (mechanical force) between the “High” and “Low” conditions (approximately 0.15 Nm) may be not large enough to influence the tool wear. Compared to the torque oscillation, the shoulder temperature was a more important factor influencing the tool wear for SiAlON tools. The dependence of the wear mechanism on the shoulder temperature is discussed in the next section.
Representative torque data in the direction of tool traveling for the three representative FSW conditions.
According to the above results, it was found that the wear of a SiAlON tool during FSW of SPFC980 was strongly influenced by the measured shoulder temperature. According to the previous reports and conventional consideration, the possible tool wear mechanisms are listed as follows:
Heat cracks generated during the FSW may promote the elimination of SiAlON grain clusters, which can be regarded as tool wear. Figure 13 shows the appearance of the tool top and its elemental maps in the “Middle” condition. Although iron was adhered on the part of the shoulder, no micro-chipping seemed to be observed on the tool top. From those results, heat microcrack did not seem to be the main mechanism of tool wear. This interpretation should be reasonable because SiAlON is a well-known material having high heat shock resistance due to low thermal expansion, high toughness, and so on.53) Furthermore, in this study, the cooling rate of the tool was moderate because no coolant was used during the FSW tests.
Appearance of the tool top and its elemental maps after FSW tests in the “Middle” condition. (travel distance 850 mm).
A SiAlON tool may react with iron contained in the work-piece. If so, the adhered iron on the tool can diffuse into the internal material of the tool and react with silicon (Si) to generate brittle ferrosilicon such as Fe3Si. The generation of ferrosilicon weakens the bonding strength of SiAlON grains and promotes tool wear. Figure 14 shows elemental maps of the shoulder part for three representative FSW conditions. The measuring area was on the top face of the shoulder.
Elemental maps of the shoulder part (top face) for three representative FSW conditions (a) “High” condition (b) “Middle” condition (c) “Low” condition (SEI: Secondary electron image).
In the “High” condition as shown in Fig. 14(a), Fe element about 10 µm in thickness was detected on the top surface of the shoulder, but the location of Fe did not match that of Si, indicating that iron did not diffuse into the SiAlON tool material. Although the measured shoulder temperature was approximately 1080°C, the real shoulder temperature could reach more than 1100°C. So, we were not sure whether the chemical reaction of iron and SiAlON occurred or not.
From the viewpoint of thermodynamics, the temperature of Fe3Si formation can be estimated. Fe3Si formation Gibbs energy ΔG (J/mol) from the reaction of Fe and Si3N4 is given by the following equation:54)
| \begin{align} &\text{9Fe(s)} + \text{Si$_{3}$N$_{4}$(s)} = \text{3Fe$_{3}$Si(s)} + \text{2N$_{2}$(g)} + \Delta \text{G},\\ &\Delta\text{G} = 299880 - 215\,\text{T}. \end{align} | (3) |
Furthermore, Simoo et al. reported the reaction of iron and Si3N4, which is also useful to consider the reaction between iron and SiAlON.55) The starting reaction temperature between iron and Si3N4 was more than 1077°C and Fe–Si solid solution and Fe3Si were formed at more than 1200°C and 1300°C, respectively.
Considering the traveling speed, however, the time approaching the peak welding temperature is too short for iron to diffuse into SiAlON even if the temperature is high enough for the reaction. The exposure time at high temperature equal to the traveling times is 160 s in the “High” condition. Such short exposure time at high temperatures could not influence the tool wear. Furthermore, O and Al elements in SiAlON improve reaction resistance with iron compared to pure Si3N4.53) Therefore, it was reasonable that diffusion of iron into the SiAlON hardly occurred even in the “High” condition.
In the “Middle” and “Low” conditions as shown in Fig. 14(b) and (c), diffusion of iron into the SiAlON was not observed either, because the measured shoulder temperatures were much less than the reaction temperature of iron and Si3N4. It was noted that a crack was observed in Fig. 14(c). This crack was not a heat crack, because it tended to propagate in the direction of the tool expansion which is a typical characteristic of a heat microcrack.
4.3 OxidationOxidation also promotes the elimination of SiAlON grain clusters. As shown in Fig. 14(a), the location of O element matched a part of Fe element attached to the resin, indicating that adhered iron on the tool was partly oxidized at high shoulder temperatures (at least 1100°C), but the surface of the shoulder part was not oxidized. The SiAlON tools were not oxidized for the other conditions, either. Therefore, oxidation can be excluded from the main mechanism of tool wear in this study.
4.4 Elimination of grain clusters due to softening of glass phasesSiAlON contains glass phases as a sintering aid among grain boundaries for promoting densification.50) The hardness of the glass phases is very important because it influences the holding force of SiAlON grains. Figure 15 shows hardness of the SiAlON tool as a function of temperature. The hardness Hv decreased gradually with increasing temperature T, which may be strongly attributed to the softening of the glass phases. This tendency of the tool hardness is understandable from the fact that the tool wear strongly depended on the measured shoulder temperature. Tool wear should progress when the glass phases holding SiAlON grains are softened leading to grain elimination. According to the secondary electron image (SEI) of Fig. 14, the surface of the shoulder part did not wear uniformly, which suggested that the elimination unit from the shoulder surface was cluster level rather than a single grain.
Hardness of the SiAlON tool as a function of temperature.
In the “High” condition, the measured shoulder temperature was nearly 1100°C and real shoulder temperature should be higher than the measured shoulder temperature and such high temperature promotes tool wear due to the reduction of grain holding force. In the “Low” condition, the holding force did not decrease so much, which may suppress tool wear. It was reasonable that elimination of grain clusters due to softening of glass phases may be the main mechanism influencing the tool wear in this study.
4.5 Appropriate FSW conditionAccording to the above discussions, the main mechanism for wear of the SiAlON tool at high temperatures must be elimination of grain clusters due to softening of glass phases contained in the SiAlON tool. Especially, in the “High” and “Middle” conditions in which the measured shoulder temperature was more than 1000°C, the force of holding SiAlON grains became so weak that the elimination of grain clusters could occur, and the wear of the shoulder part progressed.
In addition, tool fracture is important for considering tool durability. The risk of tool fracture is increased due to stress concentration at the worn shoulder part. During the FSW, the tool is subjected to strong force resisting the tool traveling, which supplies the risk of tool fracture. Especially, the strong oscillation of mechanical force may promote tool fracture as well as tool wear. The strong oscillation of mechanical force can be demonstrated from the difference in peak or valley torque measured during the FSW. According to Fig. 12, the “Low” condition exhibited larger difference in peak or valley torque than the “High” and “Middle” conditions. Therefore, the risk of tool fracture in the “Low” condition may become higher compared to the other two conditions, but such low-temperature welding provides a powerful merit for suppressing tool wear.
The difference of torque oscillation among three FSW conditions as shown in Fig. 12 should have originated from the difference in the hardness ratio of work-piece to tool. Figure 16 shows hardness of a SPFC980 work-piece as a function of temperature. The hardness of work-piece Hv drastically decreased when the temperature T exceeded 500°C. Around the temperature of the “Low” condition (775°C), hardness of the work-piece became about 1 GPa, which was 1/15 as hard as that of the SiAlON tool at the same temperature. When the temperature reached more than 900°C, hardness of the work-piece became less than 0.5 GPa, which was 1/28 less than that of the tool. This difference in the hardness ratio of work-piece to tool should influence torque oscillation during the FSW. Actually, torque oscillation was relatively high in the “Low” condition because the hardness ratio in the “Low” condition was largest of the three conditions. However, tool wear was hardly observed over at least 850 mm of traveling distance in the “Low” condition due to the low shoulder temperatures (around 700°C).
Hardness of a SPFC980 work-piece as a function of temperature.
From the viewpoint of tool wear, the “Low” condition is the best condition, although we must take into account the probability of tool fracture which depends on both the degree of torque oscillation and the shoulder temperature. It is good for SiAlON tools to keep the measured shoulder temperature ranging from 680 to 800°C from the viewpoint of tool wear, although the real shoulder temperature may be a little higher than the measured one.
In this study, friction stir welding (FSW) of high tensile strength steel plate formed by cold-rolling (SPFC980) was conducted by using SiAlON tools. The following conclusions can be obtained:
