Microstructure evolution and mechanical properties of friction stir welded FeCrNiCoMn high-entropy alloy

Introduction

High-entropy alloy (HEA) is a new metal containing more than five elements in near-equiatomic concentration, and it has gained considerable attention in the recent years [1]. The HEAs usually have a simple solid solution structure and some excellent properties, and they are regarded as potential structural materials in the near future. For the manufacturing of HEA complex structural parts, welding technology is inevitable. Therefore, it is necessary to research the weldability of HEA and expand its industrial application. Friction stir welding (FSW) was invented by Thomas et al. as an initial attempt to join low melting point materials such as Al alloys and Mg alloys [2]. With the rapid growth of the welding tool in the past two decades, FSW has been demonstrated to be an effective method of joining high melting point materials as well, such as Cu, Ti and steel [3,4]. However, the use of FSW in HEA has been scarcely investigated [5,6]. It is well accepted that the most important character of FSW was grain refinement and mechanical properties’ modification, which was usually attributed to dynamic recrystallization caused by severe plastic deformation at elevated temperatures [7]. Previous studies found that the predominant deformation during the FSW process is simple shear deformation. According to the recent FSW studies about the HEA with face-centered cubic (FCC) structure, ultrafine grain structure with a significant B type {112}<110> shear texture was detected in the weld centre [6]. It is known that the microstructure and texture development are determined by processing history at the stirring stage [8]. Moreover, in our recent study, it was found that a post annealing effect occurred at the air cooling stage can further remarkably influence the grain structure which produced in the stirring stage [9], and thus it may affect final mechanical properties in the weld as well. In this study, therefore, the microstructural features and the mechanical properties of the weld produced at the stirring stage and the air cooling stage was separated for discussion. This work also provides a new insight for us to understand the mechanism of grain refinement and mechanical properties improvement of HEA during the FSW.

Experiments

HEA ingots with a nominal composition of Fe₂₀Co₂₀Ni₂₀Cr₂₀Mn₂₀ (at.-%, FCC structure [10]) were prepared by arc-melting a mixture of pure metals. Then, the HEA ingots were cold rolled to 3 mm thickness, and subsequently 900°C annealed for 90 min. The FSW was operated by using a location-control mode with rotation rate and welding speed of 600 rpm and 200 mm min^–1, respectively. The tool tilt angle was set as 2.5°. Figure 1 schematically shows the dimensions of the tool and the welding process. At the end of the welding process, the FSW machine's emergency stop button was pressed, stopping the tool from travelling and rotating suddenly. Immediately, liquid nitrogen was continuously sprayed out towards the keyhole to stop the texture transformation. For a more detailed understanding of this method, readers may refer to our previous article [11]. After that, electron backscattered diffraction (EBSD) measurements were operated on the specially selected regions surrounding the keyhole, which located at the liquid nitrogen-cooled area. EBSD measurement regions 1^#, 2^#, 3^# and 4^# indicating the microstructure evolution in the stirring stage. To investigate the microstructural transformation details in the air cooling stage, EBSD characterisation was further operated in region 5^#, which located 60 mm behind region 4^#. Moreover, to understand the substructure transformation in the air cooling region, region 4^# and region 5^# were further characterised by transmission electron microscope (TEM). Average Vickers hardness value was examined in the rapid cooing region and the air cooling region. The tensile specimen with gauge size of 15^L × 4^W × 2.5^T mm³ contains only the stir zone, cut according to the standard of GB/T 228-2002. The tensile tests were conducted at a crosshead speed of 1 mm min^–1 with the tensile axis parallel to the welding direction.

OPEN IN VIEWER

Figure 1

Schematic illustration of the FSW process and the tool dimensions. The normal direction, welding direction and transverse direction are indicated by ND, WD and TD, respectively. AS and RS mean advancing side and retreating side of the weld, respectively.

Results and discussion

Initially, the microstructure of the base material was investigated. The average grain size of the base material was about 25.4 µm (Figure 2(a)). Based on the histogram of misorientation angle distributions as shown in Figure 2(c), a remarkable distribution peak was identified. This distribution peak was located at 60° vicinity, indicating the formation of annealing twin boundaries [6]. The (111) pole figure as shown in Figure 2(d) indicated that {100}<001> recrystallization texture was produced in the base material due to the annealing treatment.

OPEN IN VIEWER

Figure 2

EBSD measurement results of the base material including (a) inverse pole figure, (b) grain boundaries map, (c) histogram of the misorientation angle distribution and (d) (111) pole figure. In (b), low angle boundaries (LAB, 2 ≤ θ°15°), high angle boundaries (HAB, θ ≥ 15°) and twin boundaries (TB, θ ° 60°) are depicted as blue, black and red lines, respectively.

To understand the microstructural and textural transformation during the FSW, four EBSD measurements surrounding the keyhole were conducted. The obtained inverse pole figure and grain boundary maps are shown in Figure 3, and the corresponding histograms of misorientation angle distribution and (111) pole figures are shown in Figures 4 and 5, respectively. For region 1^#, the grains of base material were remarkably elongated towards the material flow direction caused by the highly rotating pin. The number fraction of LAB remarkably increased to 70% due to the generation of dislocations, and the grain boundary bulging also occurred in this region (black arrow in Figure 3(e)). Also, a new grain surrounded by massive LABs was detected in an initial grain interior (red arrow in Figure 3(e)). The nucleation of this new grain maybe attributed to the intense plastic deformation, and then grew up through HAB migration. These two features were usually observed during the discontinuous dynamic recrystallization (DDRX) [12]. Accordingly, the mean grain size decreased to 7.5 µm because of the partial recrystallization. The (111) pole figure as shown in Figure 5(a) exhibited a different orientation component compared to the base material as shown in Figure 2(d). Before the textural analysis, the deformation geometry during the FSW was initially introduced. The stir zone usually showed shear textures because of the simple shear deformation caused by the rotating pin [8]. The shear texture is usually defined by using a combination of crystallographic plane {hkl} and crystallographic direction <UVW> [13]. The <HKL> and the <UVW> are parallel to the shear plane normal (SPN) and shear direction (SD), respectively. The shear deformation changes as a function of the location surrounding the probe, as shown in Figure 6(a). For example, in the stir zone centre, the SPN and the SD are almost parallel to the TD and WD, respectively. Figure 6(b) summarises the typical shear textural components for FCC metals or alloys. To confirm the shear textural type, the original (111) pole figure was rotated to match the conventional reference frame of simple shear [14]. The rotated (111) pole figure indicates that a shear texture consisting of A* {111}<112> component produced in the region 1^#. At the initial stage of the stirring, {111}⟨112⟩ dislocation slip in the grains was activated by intense shear deformation, leading the elongated grains exhibiting a strong A* {111}<112> component dominant simple shear texture. A similar result was obtained in our previous research on the FSW of pure Cu [11]. In this region the initial {100}<001> recrystallization texture disappeared, and the simple shear texture begun to appear.

OPEN IN VIEWER

Figure 3

Inverse pole figures and grain boundary maps at different positions around the frozen keyhole, namely, region 1^# (a and e), region 2^# (b and f), region 3^# (c and g) and region 4^# (d and h). [To view this figure in colour, please see the online version of this journal.]

OPEN IN VIEWER

Figure 4

Histograms of misorientation angle distribution at different positions around the frozen keyhole, (a) region 1^#, (b) region 2^#, (c) region 3^# and (d) region 4^#.

OPEN IN VIEWER

Figure 5

Original and rotated (111) pole figures at different positions around the frozen keyhole, namely, region 1^# (a and e), region 2^# (b and f), region 3^# (c and g) and region 4^# (d and h).

OPEN IN VIEWER

Figure 6

(a) The conventional reference frame of simple shear during the FSW and (b) (111) pole figure showing important ideal orientations associated with simple shear deformation of face-centered cubic metals or alloys.

When the materials flow into region 2^#, a remarkable bimodal microstructure was detected. Many small grains surrounded by HABs or by LABs were observed in the large deformed grains (black-dotted circle in Figure 3(f)). These refined grains were generated through transformation from LAB to HAB by lattice rotation, which means the occurrence of the continuous dynamic recrystallization (CDRX) [12]. The mean grain size further decreased to 4.3 µm, and the number fraction of LAB decreased to 58% compared to those of region 1^#. The reason for these two changes was that the recrystallization process further promotes grain refinement, dislocation rearrangement or annihilation. As the materials entering region 3^#, much more fine equiaxed recrystallized grains can be observed at the original grain boundaries. The LAB number fraction and the average grain size deceased to 46% and 2.3 µm respectively. When entering region 4^#, uniform grain structure with mean grain size of 1.2 µm and LAB number fraction of 31% is obtained, which means that the recrystallization process has been completed. The LAB number fraction continuously reduced from region 1^# to region 4^# associated with materials flow from AS to RS due to the LABs transformed to the HABs by absorbing the dislocations under DDRX and CDRX. The (111) pole figures of region 2^# to 4^# indicated that all these positions exhibited a mixture texture including A* {112}<110> component and A {111}<110> component. This means that the crystal SP gradually align with the SP of the shear deformation [12,15–17]. Accordingly, the maximum texture intensity decreased from 4.521 in region 2^# to 3.324 in region 4^#. It suggested that although recrystallization process was completed in region 4^#, these fine recrystallized grains still retain strong shear texture due to severe shear deformation. Figure 7(a,b) shows the microstructural features of region 5^# located at 60 mm behind the frozen keyhole which experienced the whole air cooling stage. Clearly, the grain structure homogeneous and the average grain size was 3.4 µm, which slightly coarsened compared with that of the region 4^#. Figure 7(c) reveals a further reduced LAB number fraction (21%) and increased TB number fraction (23%). This structural morphology maybe caused by post annealing effect in the air cooling stage. This region experienced the whole air cooling stage and exhibited the final texture of the weld, and it showed a texture mainly consisting of B {112}<110> component, and the texture intensity increased to 4.676. A {111}<110> texture can also be detected, but its intensity was reduced (Figure 7(d)).

OPEN IN VIEWER

Figure 7

EBSD measurement results of the region 5^# including (a) inverse pole figure, (b) grain boundaries map, (c) histogram of the misorientation angle distribution and (d) (111) pole figure.

To claim the microstructural development in the air cooling stage, the stir zone microstructure of the air cooling region and the rapid cooling region was comparatively investigated by TEM examination and careful EBSD characterisation, and the obtained results are shown in Figures 8 and 9, respectively. As shown in Figure 8, for the region 4^# dislocation tangles which formed in the stirring stage can be detected. For the region 5^#, in contrast, the dislocation density reduced due to the static recovery, and twinning structure generated associated with grain growth. This kind of twin with fewer dislocations inside was usually identified as an annealing twin [6,18]. This TEM examination suggests that the post-annealing effect resulted in dislocations annihilation and annealing twin formation. In addition, the selected grain growth in air cooling stage made the texture transform from the A {111}<110> to B {112}<110>. The grain growth was mainly related to the movement of grain boundaries which produced in the stirring stage. The driving force of grain growth was the difference dislocation density near the grain boundary. The dislocation density strongly depends on the grain orientation in the stirring stage. It is difficult for the grains with B {112}<110> component to receive a high dislocation density in the grain interior due to the crystallographic symmetry [18]. As shown in Figure 9, the region of low dislocation density (Grain B) accelerates the grain growth, and thus B {112}<110> component becomes dominant in the air cooling stage.

OPEN IN VIEWER

Figure 8

Bright field TEM image of (a) region 4^# and (b) region 5^#.

OPEN IN VIEWER

Figure 9

Details of selected grain growth in the air cooling stage. LAB and HAB are indicated by gray and black lines, respectively.

Figure 10(a) shows the average Vickers hardness values of the weld zone as well as base material. The average hardness of rapid cooling region was higher than that of the air cooling region due to the high dislocation density and refined grain structure. It is known that for polycrystal metals or alloys, the relationship between grain size and microhardness can be descried by the Hall–Petch formula, HV = HV₀ + kd^−0.5, where HV is the microhardness of a polycrystalline metal or alloy, HV₀ and k are constants, and d is the grain size. Hall–Perch relationship predicts that the microhardness increases with the decrease of grain size. In Figure 10(a), the Hall–Petch line for HEA (FeCrNiCoMn) was extrapolated according to Liu et al.'s result [10]. For base material and air cooling region, the microhardness matches this relationship. However, the measured microhardness data of rapid cooling region was obviously higher than that predicted by the Hall–Petch relation due to the relatively high dislocation density. In our recent study, similarly, the microhardness and grain size in the weld of pure Ti do not match the Hall–Petch line of annealed pure Ti due to the high dislocation density caused by incomplete recovery under rapid cooling conditions [19]. Figure 10(b) shows the tensile test results of the base material and the stir zones which experienced air cooling and rapid cooling. All the stir zones showed increased strength and decreased ductility compared to the base material. Especially, for the stir zone of rapid cooling region exhibited significantly enhanced strength compared to the stir zone of air cooling. More importantly, as the strength of the stir zone increased, the ductility slightly decreased for about 1%. A relatively good strength and ductility combination was achieved. The reasons can be attributed to the appearance of twin structure in region 4^#. Twin boundary can provide adequate barriers to dislocation motion for strengthening and creates more local sites for nucleating and accommodating dislocations, thereby elevating ductility and strain hardening during the tensile test [20].

OPEN IN VIEWER

Figure 10

The hardness and grain size relationship of the FSW HEA and (b) tensile test results of the SZ of the FSW HEA joints with different cooling conditions.

Conclusion

FeCrNiCoMn high-entropy alloy (3-mm-thick) was successfully joined by FSW. To investigate themicrostructure development during the FSW, the stirring stage and the air cooling stage were separated for discussion. The grain refinement mechanism in the weld can be regarded as the combination of CDRX and DDRX in the stirring stage, and static recovery and selected grain growth in the air cooling stage. Accordingly, the texture component changed from A* {112}<111> via A {111}<110> to B {112}<110> component, as the grains transformed from elongated shape to equiaxial shape. This work demonstrated that the selected grain growth in the air cooling stage remarkably influenced the microstructure and mechanical properties that formed in the stirring stage. For microstructure development investigation, the post annealing effect should not be neglected. This study also provides us a good strategy to further enhance the FSW weld strength without ductility loss by using rapid cooling technology during the FSW.