Welding and Joining
Recrystallization Behavior of IF Steel at the Interface of Aluminum Junction
2022 Volume 62 Issue 7 Pages 1469-1477
Details
2022 Volume 62 Issue 7 Pages 1469-1477
Two cold-rolled IF steel sheets sandwiching an aluminum sheet were heat-treated at 650°C, and the recrystallization behavior in the interfacial region between the IF steel and pure aluminum was observed and analyzed. The tongue-like structure of the η-Fe2Al5 phase appeared in the reaction zone and the recrystallized structure of the IF steel was characterized by an equiaxed structure including subgrains in the region surrounded by the η phase. In contrast, a bcc-ferrite phase elongated with a pancake shape was mainly observed in the other regions. The η phase grew preferentially along the c-axis, and the growing η phase caused distortion of the surrounding iron due to the difference in the molar volume. This suggests that compressive strain is generated along both a-axis and b-axis, and tensile strain is generated along c-axis of the η phase. The growth of the η phase obviously changed the recrystallization behavior of the IF steel. The development of the γ-fiber crystalline texture (<111>/ND), a typical recrystallization texture of IF steel, was not observed in the region surrounded by the tongue-like η phase. The local misorientation in the vicinity of the two-phase interface could not explain the texture change. It is suggested that the growth of the tongue-like η phase caused constraint of the surrounding region and affected nucleation in the recrystallization of the ferrite.
In response to concern about global environmental problems, stricter regulations have been applied to CO2 emissions from automobiles, and the strength of automotive steel sheets has increased accordingly in order to achieve targets for weight reduction.1,2,3) However, the targets have become so strict that there is a trend toward multi-material structures to replace some parts with lightweight materials such as aluminum, magnesium, and carbon fiber-reinforced polymer (CFRP).4) In applying multi-material automotive parts, joining technologies have become essential because joining processes are needed in assembly lines. One representative joining technology is friction stir welding (FSW),5,6,7) in which materials are welded by pressing a rotating heat-resistant pin against the joint part. As the materials are subjected to extreme deformation at higher temperatures, recrystallization may occur together with deformation.
Examples of interfacial reactions among dissimilar metals include hot dip aluminizing. While hot-dip galvanizing is a major surface treatment process for automotive steel sheets, new plating techniques such as hot dip aluminum plating have also been studied recently8) due to concern about the depletion of zinc ore resources. Since aluminum has a higher melting point than zinc and brittle intermetallic compounds are formed in the reaction zone between steel and aluminum, alloying elements such as Si are added to the aluminum melt to improve the plating quality.9,10) Assuming the use of the same process as in galvanizing, dipping would be carried out after recrystallization of the steel. However, if the steels have a higher recrystallization temperature or its thermal history is changed, it is possible that recrystallization of the steel and the interfacial reaction may occur simultaneously. Thus, control of the interfacial reaction will become more important from the standpoint of multi-material joining technologies.
In the phase diagram of the Fe–Al binary system,11) for example, in the range of 600 to 700°C, there are several intermetallic compounds (IMCs): FeAl (α’) that is bcc ordered phases, FeAl2 (ζ), Fe2Al5 (η), and Fe4Al13 (θ). All of these IMCs have a high Vickers hardness of 1000 HV or more. In particular, the η phase is the hardest and most brittle, which causes interfacial defects such as weld cracking. Chang et al.10) studied the Fe/Al interfacial reaction of commercial AISI 1050 steel (Fe-0.05%C-0.24%Mn) by immersing a steel sheet in a pure aluminum melting bath with a temperature of 700°C. According to their research, the aluminide layer is composed of an outerly layered θ and an innerly thick η phases, in which the η phases grew in a tongue-like shape resulting in the complicated interface. The previous literature9) reported that the orthorhombic η phase grows preferentially along the c-axis, and suggested that residual stress also affects that growth. In the present study, a microstructural analysis was performed using IF steel with a higher recrystallization temperature than that of mild steels, and the effect of η phase growth on the recrystallization of the steel was studied.
Nb‐added ultra‐low carbon steel with the chemical composition of Fe-0.007%C-0.6%Mn-0.068%Nb (mass%) was used. In this steel, the Nb and C contents were higher than those of conventional IF steels in order to obtain a higher recrystallization temperature, and the atomic ratio of Nb/C was set to 1.05. The steel was prepared by vacuum melting and hot forging to produce a sheet bar with a thickness of 30 mm, which was solution-treated at 1250°C for 3.6 ks, hot-rolled at a finishing temperature of 900°C to a thickness of 3.5 mm, subjected to coiling treatment at 640°C for 3.6 ks and then cooled to room temperature. The hot-rolled steel sheet was ground on both sides to 2.8 mm, and cold rolled in one direction without intermediate annealing to obtain deformed steel sheets with a thickness of 0.28 mm. A cold-rolled aluminum sheet with a thickness of 0.54 mm was sandwiched between two cold-rolled IF steel sheets, the circumference was wound with wire, and the sample was sealed in a quartz tube under an Ar atmosphere. The encapsulated samples were heated to 650°C at a heating rate of 2°C/s in an infrared image furnace and held for 120 s or 600 s.
The heat-treated specimens were embedded in a conductive carbon resin. The surface of the cross-section was ground with sandpaper and then polished with a 0.25 μm diamond paste and oxide polishing suspension (OPU). The crystal orientation was measured by the electron backscatter diffraction method (EBSD). Orientation Imaging Microscopy (OIM) data correction by TSL Solution was used for data acquisition, and OIM data Analysis 6.2 by TSL Solution was used for the crystal orientation analysis.
After EBSD observation, a specific interface in the IF steel/η phase was obtained by a focused ion beam (FIB: FEI Helios Nano Labo) apparatus. Then, the substructure of the phase interface was observed with an image-aberration-corrected scanning transmission electron microscope (STEM; FEI Titan, 300 kV). The concentrations of the elements at the reaction interfaces were confirmed by elemental mapping and line analysis using an X-ray microanalyzer (FE-EPMA; JXA 8530 F-plus, JEOL) at an acceleration voltage of 15.0 kV and irradiation current of 1.0−7 Å.
Figure 1 shows the typical microstructures of the interfaces between the IF steel (upper side) and aluminum (lower side) in the specimens heat-treated at 650°C for 120 s and 600 s. The needle-like or rectangular crystals in the aluminum region at the center of the specimen in the vicinity of the interface adjacent to aluminum are considered to be the θ-Fe4Al13 phase. It assumed that the metastable θ phase is crystallized during solidification of the liquid phase supersaturated by iron. Two layers with different contrasts were observed at the interface, and they were identified as θ and η-Fe2Al5 phases respectively by FE-EPMA analysis. While the θ phase grew slowly over the holding time, the η phase grew preferentially to the IF steel side and kept a tongue-like shape. The average growth lengths of the η phase were approximately 40 μm at 120 s and 180 μm at 600 s, indicating that this phase grew by 4 to 5 times between 120 s and 600 s. Voids and cracks were observed in the η phase near the aluminum side. Since the η phase is very hard and residual stresses may remain after heat treatment as a result of the difference in the thermal expansion coefficient with that of the steel, cracks could easily be introduced during sample polishing.
Microstructures of the diffusion couple with heat treatment at 650°C. The holding times are 120 s (a,b) and 600 s (c,d). The η phase grows preferentially into the side of the IF-steel with a tongue shape.
Figure 2 shows the results of EBSD measurements of the specimen heat-treated at 650°C for 120 s. Figure 2(b) is an Inverse Pole Figure (IPF) map of the η phase in the Normal Direction (ND) vertical to the interface or sheet surface, where the growth direction shows a <001> texture. In the region where the η phase preferentially grows with a tongue-like shape, coarse grains had elongated to the interface normal. The η phase at the base of the tongue or on the aluminum side consisted of relatively fine grains and was equiaxed compared with that at the tip parts. Moreover, the grain orientations deviated slightly from the c-axis. Figures 2(c) and 2(d) are the IPF maps of the IF steel in the ND and Rolling Direction (RD; horizontal in the figure). In the region far from the interface with the η phase, a partial recrystallization microstructure was observed with elongated deformed grains and recrystallized grains. In the Nb-based IF steel used in this study, the nucleation for recrystallization was inhibited due to the added Nb,12) which appears to have led to elongation of the recrystallized grains along the RD. The crystal orientations mainly belonged to <111>// ND (blue) and <110>// RD (green), the typical recrystallized texture of IF steels. On the other hand, the steel parts bounded by the η tongues, i.e., the bay regions, consisted of equiaxed fine grains, and the crystal orientation was close to <110>//ND, which is different from those in the region far from the interface. This tendency was notable at the grains facing the side of the tongue-like η phase. Figure 2(e) is a Kernel Average Misorientation (KAM) map showing the local orientation difference between each measurement point and the surrounding measurement points. In the matrix IF steel, the KAM value was high in non-recrystallized grains, while being quite low in recrystallized grains. In contrast, in the steel sandwiched by the η phase, there were many grains which had high KAM values in spite of possessing an equiaxed structure, particularly in the vicinity of the interface with the η phase. Figure 2(f) shows the character of the grain boundary in the square area in Fig. 2(a). The red line indicates a grain boundary with a tolerance angle of 2 to 15°. Figure 2(g) maps the deviation angle from the average orientation of the grains in the same region. In the region bounded by the η phase, many subgrain boundaries were observed. The subgrain was characterized by the region with higher values of grain reference orientation deviation angle and also with higher KAM values.
EBSD analysis around the interface of the sample annealed at 650°C for 120 s; (a) image quality map, (b) inverse pole figure map (ND) of the η phase, (c,d) inverse pole figure map ((c); ND, (d); RD) of the steel, and (e) KAM map of the steel. Figs. (f,g) are the subgrain structure of the steel in the square region in fig. (a); (f) grain boundary character, (g) grain reference orientation deviation angle.
Figure 3 shows the results of the EBSD measurements of the heat-treated sample held at 650°C for a longer time of 600 s. The IF steel was more completely recrystallized than the steel held for 120 s, while a small number of non-recrystallized grains with orientations close to <100>//ND remained.13) The recrystallized grains had a pancake shape slightly extended in the rolling direction without orientation dispersion within each grain. Near the η tip and in the region surrounded by the η phase, however, a mixed microstructure composed of fine equiaxed grains and relatively large grains with orientation dispersion was observed, and the orientation dispersion was larger in the coarse grains.
EBSD analysis around the interface of sample annealed at 650°C for 600 s; (a) image quality map, (b) inverse pole figure map (ND) of the η phase, (c,d) inverse pole figure map ((c); ND, (d); RD) and (e) KAM map of the steel. Figs. (f,g) are the subgrain structure of the steel in the square region in fig. (a); (f) grain boundary character, (g) grain reference orientation deviation angle.
The texture properties of the IF steel heat-treated at 650°C for 120 s were analyzed by using the EBSD datum (Fig. 2) for two regions, the matrix far from the interface and the area surrounded by the η phase. The (200) pole figures and the three-dimensional crystal orientation density function (ODF) for each region are shown in Fig. 4. In general, the recrystallization texture in steel after rolling accumulates in <110>// RD (α-fiber: φ2 = 45°, φ1 = 0°) and <111>// ND (γ-fiber: φ2 = 45°, Φ = 54.7°), and the γ-fiber intensity becomes particularly strong in recrystallized IF steel sheets.14) The matrix in the present IF steel showed a texture with both the α- and γ-fiber, as expected. On the other hand, in Figs. 2(c) and 2(d), the texture was rather weak in the region surrounded by the η phase. Although a quantitative discussion would be difficult due to the small number of grains evaluated, some components near the rotated cube {100}<110>, which is well-known as a characteristic of slow recrystallization, were relatively strong, and ones that deviated from the γ-fiber were also observed.
(200) pole figures, (a,c), and ODF φ2=45° sections, (b,d), of IF steels with heat-treated at 650°C for 120 s. Figs. (a,b) are analyzed with the redion in the matrix region. Figs. (c,d) are in the region surrounded by η.
This study examined the microstructure evolution in conditions under which recrystallization of Nb-added IF steel competes with the growth of the η phase near the steel-aluminum interface. The steel regions surrounded by the tongue-like η phase consisted of an equiaxed grain structure deformed with some orientation dispersion, and possessed a texture apart from <110>//ND, which is not observed in conventional recrystallized IF steel sheets.
There are three potential factors providing this unique orientation, (1) the diffusion of aluminum penetrating from the η phase, (2) the difference in the recrystallization nucleation sites in the steel between the interfacial area and the matrix region, and (3) the constraint of the surrounding steel by the tongue-like η phase. These are discussed in detail below.
It will be discussed whether the diffusion of aluminum from the η phase affects recrystallization of steels. Figure 5 shows the FE-EPMA concentration map for each element in the interfacial region acquired from the sample heat-treated for 120 s. Figures 5(a), 5(b), 5(c), and 5(d) correspond to the maps of Fe, Al, Mn, and Nb, respectively. In the concentration maps of Al and Fe, θ is distinguished from the η phase, the thickness of the θ phase is not uniform, and its growth rate is much slower than that of the η phase. In Fig. 5(c), some small amount of Mn is detected in the η phase, and the concentration tends to gradually increase toward the direction of the IF steel matrix. This is thought to be caused by a small amount of residual Mn in the η phase as a result of the partitioning of Mn at the interface when the η phase grows into the IF steel. A similar trend was observed for Nb, but was not clear because of the low concentration of Nb. Figure 6 shows the concentration line profiles of iron and aluminum near the interface. Line 1 is from the valley region of the tongue where the growth of the η phase was delayed, and line 2 is from the tip region where the η phase grew preferentially. Because the aluminum concentration in the IF steel near the interface was close to zero in both the tip and the valley, it is unlikely that the diffusion of aluminum into ferrite is the main factor that changes the recrystallization behavior and texture. The α′ (B2) ordered bcc phase between the η phase and the IF steel could not be confirmed, and if present, it would be very thin, being below the resolution of the EPMA. In the line 2, the slight fluctuation of the aluminum concentration in the η single-phase region was due to the signals of the electron beam, which penetrated the interior below the measured surface. After this kind of noise was removed, the aluminum concentration in the η phase decreased slightly approaching the tip. This concentration gradient is considered to yield the driving force for the growth of the η phase.
Concentration maps of the elements by Fe-EPMA of the specimen heat-treated at 650°C for 2 min. Figs. (a), (b), (c), (d) correspond to the maps of Fe, Al, Mn, Nb, respectively.
Line profiles of the concentration of both Mn and Al near the steel/Al interface on the specimen heat-treated at 650°C for 120 s. Fig. (b) is the profile of the line 1 in fig. (a): fig. (c) is the profile of the line 2 in fig. (a).
It will be secondly discussed whether recrystallization nuclei at the η/α interface could lead to the unique textures. It is possible that the η/α interface itself becomes the recrystallization nucleation site. The nucleated orientation could be different from the one caused by recrystallization of mild steel, in which nuclei are generated on deformation bands of the cold-rolled material. However, since the entire region surrounded by the η phase consisted of equiaxed structures and the unique orientation was observed even at a distance from the η phase, this factor alone cannot explain the texture change observed in this experiment.
The third factor is mechanical constraint of the IF steel by the growing η phase, as will be discussed next. As vacancies occupy the aluminum sites in the c-axis of orthorhombic structures, the preferential growth of the η phase along the [001] direction has been explained by the rapid diffusion of the η phase in that direction.15) It has also been reported that the growing η phase confines the surrounding region and the priority growth to the c-axis is lost with an extended holding time. In the present study, the recrystallization texture also changed in the region surrounded by the η phase, suggesting that the stress condition around the growing η phase might affect the recrystallization behavior.
Since bcc α-iron contains two iron atoms in a unit cell with a lattice constant of 2.866 Å16) and the volume of the unit cell is 23.54 Å3, the volume occupied per iron atom is 11.7 Å3. On the other hand, the η phase has an orthorhombic oC24 structure with a lattice constant of a = 7.66, b = 6.42, c = 4.22 Å,17) a slightly short c-axis, and a unit cell containing 15.2 atoms. The volume of the unit cell is 207.2 Å3, and the occupied volume per metal atom is 13.65 Å3. Since the occupied volume per atom greatly changes in the transformation from α to η, the growth of the η phase distorts the surrounding iron. In particular, compressive stress may be applied in the a- and b-axis directions. Thus, it is possible that constraint by the η phase changes the recrystallized texture of the surrounding steel.
Figures 2 and 3 showed that the IF steel surrounded by the η phase has an equiaxed grain structure, while the orientation is slightly dispersed in each grain and the KAM is also locally high near the interface with the η phase. To clarify these phenomena, the substructure was investigated by TEM observation. For the specimen at 120 s, a specific region surrounded by the η phase was cut out by FIB (Fig. 7). Figures 7(a), 7(b) shows the results of the EBSD measurement and the FIB cutting position. Figure 7(c) shows a secondary electron micrograph of the appearance of the TEM specimen cut out by FIB. The TEM bright-field image near the interface is shown in Fig. 8, where the incident electron beam was aligned with the growth direction of the η phase. The stripe contrast in the η phase indicated by the arrow in Fig. 8(b) is thought to be derived from the order-disorder transformation during the cooling process after heat treatment.18,19) Although it was not clear in the EPMA analysis, a thin phase structure with a width of about 100 nm was observed on the interface between the η and bcc phases. The composition determined by WDS (wavelength dispersive X-ray spectroscopy) analysis was Al: 29.2%, Mn: 0.34%, Fe: 70.5% in atomic fraction, and the region was identified to be α′, B2 phase. Although a heavily deformed structure with a high density of dislocations was not observed on the IF steel side as a whole, a high density of dislocations had accumulated locally near the interface (B), which is clearly different from the recrystallized structure. This implies that recovering or recrystallizing steel grains near the interface could be distorted by the constraint accompanying the growth of the η phase.
Preparatory EBSD analysis for TEM observation of the sample which was cut out from the joint part, annealed at 650°C for 120 s: (a) inverse pole figure map (ND) of the steel, in which cut-out part indicated by a red rectangular, (b) inverse pole figure map (ND) of η phase, (c) cut-out specimen by FIB.
TEM micrographs of joint part of the specimen annealed at 650°C for 120 s. Fig. (a) is a bright field image. Fig. (b) is the magnified BF image of fig. (a). Fig. (c) is the diffraction pattern of the point A in fig. (a). Fig. (d) is the diffraction pattern of the point B in fig. (a).
Figure 9 shows the plot of the misorientation profiles near the interface for parts “B” and “C” in Fig. 7. The misorientation between the neighboring measurement points is indicated by solid lines, and that between the interface (original) orientation and each measurement point is shown by broken lines. In the IF steel, the orientation changes in the vicinity of the interface, rotated by 3 to 4°, compared with the interior region, whereas the orientation change in the η phase is much smaller than that in the IF steel. This difference may be related to differences in the hardness and elastic modulus between the η phase and the IF steel.
Line profiles of misorientations along the lines near the phase boundaries. Figs. (a) and (b) correspond to the regions, B and C in fig. 7, respectively. (Online version in color.)
The recrystallization behaviors with and without erosion by the η phase are schematically illustrated in Figs. 10(a) and 10(b), respectively, where the crystal orientations are presented in grayscale, and the high and low angle boundaries are indicated by thick and thin lines, respectively. This indicates the strong influence of the constraint by the η phase, which was suggested by the high KAM values of the IF steel surrounded by the η phase in Figs. 2 and 3, the substructures in Fig. 8 and the orientation changes in Fig. 9. However, because the orientation change of 3 to 4° in Fig. 9 is smaller than would be expected from the texture analysis in Fig. 4, it is possible that the η phase affects the entire steel region surrounded by the η phase as well as the region near the interface. That is, the entire region would be deformed and rotated, changing the recrystallization texture. In addition, the recrystallized grains might be further deformed by the growth of η, and subgrain structures or local orientation changes may develop after heat treatment. Here, we consider the mixed microstructures of the IF steel surrounded by the η phase at 600 s. In Fig. 2, the thickness of the η+α two-phase region was about 40 μm at 120 s, and that of the η single-phase region facing the valley was 50 to 60 μm at 600 s. This means that the η+α two-phase region at 600 s has not yet been affected by the η phase at 120 s, when the recrystallization of the IF steel started. The η phase eroded directly into the pancake-like recrystallized grains, causing a mixed grain structure with coarse grains.
Schematic image of recrystallization behaviors of the IF steel with and without growing η phase.
Regarding the formation of the tongue-like η phase, Takata et al.9) proposed a mechanism in which the stress field associated with the α→η transformation yields the tongue shape (in their paper, called a “saw-tooth” shape). Due to the difference in the molar volumes of the η phase and iron, a tensile stress field is generated in the growth direction, and a compressive stress field is generated in the perpendicular direction. Vacancies move to the compressive stress region, and aluminum atoms migrate toward the tip region, as in diffusion creep. As a result, η crystal grains elongating to the steel direction are formed. In the EBSD analysis in the present study, fine equiaxed grains that deviated from the c-axis were observed in the valley part, suggesting the effect of the crystal orientation in addition to the stress field. The η nuclei are thought to have relatively random orientations in the early stage of the diffusion bonding reaction. Since they preferentially grew toward their c-axis, grains with other directions collided with each other, and the crystal grains oriented with the c-axis to the interface normal become dominant, resulting in the formation of the tongue-shaped structure. Because the η phase actually grows in a protruding shape from the joint interface in three dimensions,10) a three-dimensional analysis of the internal stress and strain is necessary to clarify the mechanism of texture development during the synchronous growth of the η phase and the recrystallization of iron.
In this study, the interfacial reaction at the bonding junction between cold-rolled Nb-added IF steel sheets and an Al sheet was studied under a condition in which the interfacial reaction and the recrystallization of the IF steel progressed simultaneously. The following knowledge was obtained by analyzing the microstructure changes in this process.
(1) A tongue-shaped η phase grew preferentially with the c-axis to the reaction interface normal. The steel bcc phase surrounded by the η phase changed from a deformed structure to an equiaxed structure during heat treatment. With holding time, recrystallization in the steel region progressed and a grain structure mixed with coarse grains was obtained.
(2) The crystal orientation in the steel region surrounded by the η phase deviated from the conventional recrystallization texture of IF steels. In addition to dispersion of the orientation inside the grains, the KAM values were locally high, especially in the grains facing the η phase.
(3) This texture change in the region surrounded by η was not considered to be the effect of the Al content, because the Al concentration of the IF steel in the vicinity of the phase interface was almost the same as that of the matrix.
(4) The texture of the entire region surrounded by the η phase cannot be explained by the local orientation change in the vicinity of the phase interface alone. It is suggested that the growth of the tongue-like η phase caused constraint of the surrounding region, and this also affected the recrystallization orientation.
