1. Introduction
Microstructural control is one of the key issues in friction-stir welding (FSW) [
1,
2]. This is primarily due to the extremely high sensitivity of the mechanical properties of welded joints to the underlying microstructure. Specifically, the microstructural changes induced by FSW (grain refinement, dissolution of second-phase precipitates, development of crystallographic texture, etc.) may exert a significant influence on weld strength and ductility [
1,
2].
Moreover, there is also an essential academic interest in FSW-induced microstructures. The unique characteristic of FSW is an exotic combination of very large plastic strains at high temperatures and comparatively high strain rates. As material behavior under such conditions is still not completely clear, investigation of FSWed joints may broaden our understanding of the fundamental microstructural mechanisms.
At present, grain-structure evolution during FSW is most studied in aluminum alloys. The extensive research over the last ~25 years has conclusively demonstrated the complexity of this process in such materials. Specifically, it typically consists of several stages and may involve several different microstructural mechanisms. Those include the geometric effect of strain [
3,
4,
5], continuous recrystallization [
6,
7,
8,
9,
10,
11], or geometric recrystallization [
3,
12,
13,
14,
15]. Remarkably, the stir zone material may experience secondary deformation associated with the tool shoulder [
16]. This effect is most pronounced at the upper weld surface and may produce a fine-grained microstructural layer in this area [
17]. After FSW, the welded material often experiences a static microstructural coarsening upon cooling from FSW temperature to ambient conditions [
3,
18,
19,
20].
It is well accepted that microstructural changes during FSW are significantly influenced by the thermal conditions of welding. However, the underlying mechanism is not completely clear. In fact, it is only well known that an increase in the FSW heat input leads to grain coarsening within the stir zone [
1].
In this context, of particular interest is an examination of FSW under high heat input conditions. Such FSW is usually realized using a combination of a high spindle rate and a low feed rate, thus being characterized by the highest welding temperature, the largest plastic strain [
16], and, perhaps, the highest strain rate. In other words, the high-heat-input FSW comprises the most extreme combination of deformation conditions.
To the best of the authors’ knowledge, this FSW range has not been studied systematically so far. Thus, to broaden our fundamental understanding of microstructural aspects of FSW, the present study was focused on the elucidation of a specific character of grain-structure evolution at high heat input conditions.
2. Materials and Methods
The commercial aluminum alloy 6013 was selected as a program material for the present study (
Table 1). This alloy belongs to the 6xxx series of aluminum alloys, whose FSW behavior is studied relatively well. Despite being a particle-containing aluminum alloy, this material is suitable for the purpose of the present study (i.e., examination of grain-structure evolution during FSW) because the second-phase particles are expected to completely dissolve at the high–heat input condition.
The program material was produced by semi-continuous casting using a KREMIA laboratory casting machine. The cast ingot was then homogenized at 550 °C for 4 h, with subsequent cooling in air. To enhance the effect of the high–heat input FSW on microstructure evolution, the program material was subjected to extensive cold rolling prior to the welding. To this end, the homogenized ingot was sliced along its longitudinal axis and then rolled at ambient temperature to a total thickness reduction of ≈80%. The produced material had a typical work-hardened microstructure, which consisted of pancake-shaped grains, a dense subboundary structure, and a high dislocation density (
Supplementary Figure S1a). Crystallographic texture was dominated by the typical rolling components, including S {123}<634>, Copper {112}<111>, Brass {110}<112>, and Cube {001}<100> orientations (
Supplementary Figure S1b). Throughout the manuscript, this material condition was referred to as
base material.The 2-mm-thick sheets of the base material were friction-stir welded in a bead-on-plate configuration using a commercial AccuStir FSW machine. To provide a high–heat input condition, a spindle rate of 1100 rpm and a feed rate of 0.5 inches per minute (≈13 mm/min) were employed. Given the relatively wide FSW processing window for aluminum alloys, such an extremely low feed rate is very rarely used in welding practice, to the best of the author’s knowledge. Moreover, to reduce heat loss during FSW, a 2-mm-thick Ti-6Al-4V interlayer was used between the aluminum plate and the steel backing plate.
Per analogy with the previous work [
21], the welding tool was manufactured from tool steel and consisted of a concave-shaped shoulder of 12.5 mm in diameter and an M5-threaded probe of 1.9 mm in length. FSW was conducted under the plunge-depth control mode, while the distance between the probe tip and the interlayer material was maintained to be as small as ≈50 μm. A typical convention of FSW geometry was adopted, which included welding direction (WD), normal direction (ND), and transverse direction (TD).
To record the weld thermal cycle, two thermocouples were placed beneath the aluminum plane in close proximity to the probe tip. A schematic of the thermocouple layout is given in
supplementary Figure S2. The measured peak temperature was close to ~500 °C or ≈0.9 T
m (where T
m is the melting point) (
Figure 1); hence, the actual temperature within the stir zone was likely even higher. Moreover, the cooling rate was also found to be low; specifically, it took about 3.5 min to cool the weld from ≈500 to ≈100 °C (
Figure 1). Thus, despite the fact that the term “high heat input” is not strictly defined in FSW, it could be concluded that this welding condition was provided in the present study.
On the other hand, the tracking of the Z-force evolution showed that material flow during FSW was relatively unstable. In addition to the expected fluctuations near the thermocouple positions, the loading force was found to significantly vary along the weld path (
Figure 1).
For microstructural examinations, the produced weld was sectioned in the ND × TD plane. Attempting to provide a link with temperature measurements and avoid the influence of the unstable material flow, the microstructural sample was machined between the thermocouple positions (
Figure 1). It was then mechanically polished in a conventional fashion, with the final step comprising 24-h vibratory polishing using the commercial OP-S suspension.
Microstructural observations were performed using the electron backscatter diffraction (EBSD) technique. To this end, an FEI Quanta field-emission-gun scanning electron microscope (FEG-SEM) (FEI Company, OR, US) was employed. The FEG-SEM was equipped with TSL OIM
TM software (Version 8.0) and operated at an accelerating voltage of 20 kV. To provide a broad view of the FSW-induced microstructure, a series of sample-scale maps with a comparatively coarse scan step size of 2 μm were acquired from the entire stir zone. On the other hand, to get a close inspection of microstructures in particular areas of interest, a number of higher-resolution maps with scan step sizes of 0.5 and 0.2 μm were also obtained. To improve the fidelity of EBSD data, the small grains comprising either one or two pixels were “cleaned” from the maps using the standard grain-dilation option of EBSD software (Version 8.0). Considering the limited angular accuracy of EBSD, grain boundaries with a misorientation angle below 2° were excluded from consideration. To differentiate the low- and high-angle boundaries (LABs and HABs, respectively), a 15-degree threshold was applied. Grain size was measured using the equivalent circle-diameter approach [
22].
To evaluate the effect of the post-weld aging treatment, some of the welded joints were artificially aged at 190 °C for 4 h [
23].
In order to get a broad insight into the microstructure distribution within the welded material, a series of microhardness profiles were measured across the weld’s mid-thickness. The Vickers microhardness data were collected employing a Wolpert 402MVD microhardness tester by applying a load of 200 g, a dwell time of 10 s, and a step size of 0.5 mm.
4. Discussion
4.1. Microstructural Evolution during FSW
The purpose of this work was the examination of the effect of the high–heat input FSW on microstructural evolution and material flow. From the experimental results summarized in
Section 3, the following four important findings were derived:
- (i)
The base material underwent almost complete static recrystallization in the heat-affected zone. Thus, the initial microstructure was fundamentally altered before its direct contact with the FSW tool.
- (ii)
Material flow during FSW was primarily governed by the tool shoulder.
- (iii)
FSW resulted in the formation of a pronounced layer of fine-grained microstructure at the upper surface of the stir zone.
- (iv)
The microstructure within the stir zone contained a significant fraction of the survived remnants of coarse grains, i.e., the recrystallization process was far from being completed.
Hereafter, each of these effects will be considered in detail.
The static recrystallization in the heat-affected zone was obviously related to the high-temperature condition of the applied FSW process. On the other hand, this effect was also significantly contributed to by the heavily deformed state of the base material. It is expected that a material in a well-annealed initial temper would not experience recrystallization (or even notable grain growth) during FSW in the same thermal environment. Therefore, the static recrystallization in the heat-affected zone was unlikely an intrinsic characteristic of the high–heat input FSW.
The dominant role of the tool shoulder in material flow during FSW was presumably associated with a comparatively large plunging depth of the shoulder into the welded material. This increased the friction effect between the shoulder and the material and thus promoted both an increase in the welding temperature and the contribution of the shoulder to material flow. However, the scale of the latter phenomenon was also substantially influenced by the relatively small thickness of the welded workpieces employed in the present study (i.e., 2 mm). It is fairly likely that an increase in material thickness (and the concomitant increment in the length of the tool probe) will reduce the contribution of the tool shoulder to global material flow. Hence, the shoulder-induced material flow was probably also not inherent to the high–heat input FSW.
The development of the pronounced fine-grained layer at the upper surface of the stir zone was obviously attributable to the stirring action of the tool shoulder. As shown in some experimental works [
17] and numerical simulations [
16], the stir zone material should experience a secondary strain induced by the tool shoulder; this effect was due to the backward tilting of the FSW tool. This event is characterized by an extreme combination of large plastic strain, the highest temperature, and (presumably) the highest strain rate, and thus should lead to a drastic microstructural modification at the upper weld surface. Under a high–heat input condition, the stir zone material should be comparatively soft. Hence, the secondary deformation should result in a pronounced surface layer, as has indeed been found in the present study (
Figure 5). This surface layer may provide abnormal grain growth during post-weld annealing treatment [
17], thus playing an important role in the service performance of welded joints.
The continuous recrystallization within the stir zone was in good accordance with the scientific literature [
6,
7,
8,
9,
10,
11]. On the other hand, the revealed incompleteness of this process was perhaps the most interesting result. Given the high–heat input condition of FSW, which implies the largest strain at the highest temperature, this effect appears to be surprising. One of its possible explanations may be an enhancement of recovery at high temperatures. This should reduce dislocation density and thus inhibit the development of deformation-induced boundaries. In other words, the observed retardation of recrystallization within the stir zone was likely a result of competition between the recrystallization and the recovery. As this phenomenon is most pronounced at the highest temperatures, it should be intrinsic to the high–heat input FSW.
4.2. The Post-Weld Aging Behavior
As the 6013 aluminum alloy belongs to heat-treatable materials, its service properties rely heavily on aging treatment. Hence, to estimate the industrial performance of the high-heat-input FSW weld, its aging behavior was evaluated. The obtained results are outlined in the present section.
The microhardness profile measured across the midthickness of the aged weld is shown in
Figure 11a. From the comparison with the profile measured in the as-FSW’ed condition, it was found that the aging treatment exerted no significant influence on the material
outside the diameter of the tool shoulder. On the other hand, significant material strengthening was revealed
within the shoulder diameter, with the effect being most pronounced within the stir zone.
It is expected from the above observations that the tensile behavior of the aged FSW joint should be essentially degraded by material softening within the heat-affected zone (
Figure 11a). Thus, an elucidation of the negligible aging effect in this microstructural area is of interest. To assist in the interpretation of the underlying microstructural changes, the evolution of the constituent secondary particles was predicted for 6013 aluminum alloy using ThermoCalc calculations with the TCAl8:Al-Alloys v.8.1 database. The simulation results are summarized in
Figure 11b.
Given the long-time and high-temperature character of the weld thermal cycle (
Figure 1), it is possible that the particles could
precipitate in the heat-affected zone during FSW. As discussed in
Section 2, the peak FSW temperature within the stir zone presumably exceeded 500 °C. Assuming that the temperature in the heat-affected zone ranged between 200 and 450 °C, it could be suggested that the secondary particles in this region were represented by a mixture of Al
2Cu, Mg
2Si, and Q phases (
Figure 11b). Accordingly, the post-weld aging at 190 °C provided only a small precipitation effect, and therefore material strength remained nearly unchanged (
Figure 11a).
On the other hand, considering the highest FSW temperature within the stir zone, the precipitated particles within this region may then
dissolve, leaving only a minor fraction of the Mg
2Si phase (
Figure 11b). Hence, the post-weld aging at 190 °C should promote extensive precipitation of Al
2Cu and Q phases (
Figure 11b), thus giving rise to the revealed hardening effect (
Figure 11a).
Despite the fact that the above hypotheses seem reasonable, it is important to emphasize that they are based only on ThermoCalc predictions and thus lack experimental verification. Hence, appropriate microstructural observations of the particle behavior are necessary. However, this issue is outside the scope of the present work.
5. Summary
This work was undertaken to evaluate the influence of the high–heat input condition of FSW on microstructural evolution and material flow. Considering the extreme combination of deformation conditions inherent to such a welding regime (including an extremely large plastic strain, the highest temperature, and a high strain rate), it was expected to reveal an unusual structural response. The most important findings included: (i) the formation of a pronounced fine-grained layer at the upper weld surface; and (ii) a retardation of the recrystallization process within the stir zone.
The development of the surface layer was attributed to the material softening within the stir zone at elevated temperatures. As a result, the secondary strain induced by the tool shoulder has to result in essential microstructural modifications at the upper weld surface.
The retardation of continuous recrystallization within the stir zone was explained in terms of the concurrent influence of recovery. The activation of the latter mechanism at elevated temperatures should reduce dislocation density and thus slow down the development of deformation-induced boundaries.