2023 Volume 64 Issue 8 Pages 1894-1901
Friction stir processing (FSP) has been performed on 2024 aluminum alloy sheets. The process was carried out using two different cooling rate i.e., Air Cooling (FSP-AC) and cooling using the liquid nitrogen (FSP-NC). In the current research, the effect of different cooling rates on microstructures and relating precipitates has been investigated. The results have been shown that the average grain size in the stirred zone (SZ) has been decreased significantly in the FSP-NC (1.75 µm) condition, with respect to the FSP-AC (3.3 µm). The misorientation angle of grain boundaries created the special structure. The grain boundaries were facet in FSP-AC and FSP-NC and the rough boundaries was observed in FSP-AC specimen.
TEM and EBSD analysis have been indicated the dissolution of Ω-Al2Cu precipitates in the SZ in the FSP-NC specimen, while FSP-AC caused formation of supersaturated copper solid solution with Ω-Al2Cu precipitates in the grain boundaries. Furthermore, in the both FS-processed specimens (FSP-AC and FSP-NC) fine and dispersed Al20Cu2Mn3 precipitates have been observed.
Advancement technology in the aerospace industry has been led to increased demand for lightweight aircraft structures. These achievements would result in reducing the fuel consumption and increasing the operational efficiency of the aircrafts. Aluminum alloys play a vital role in the construction of aircrafts and decreasing the whole weight of them.1,2) Increasing the strength of aluminum alloys using different mechanical methods specially by severe plastic deformation (SPD) would result in refinement of grains and precipitates and would led to desirable properties in these alloys.3,4)
Friction stir processing (FSP) as a subset of SPDs is a variant of friction stir welding (FSW), which has been introduced in 1991 by The Welding Institute (TWI).5) FSP is a novel and unique thermomechanical process that can alter the microstructure and mechanical properties of the processed material with maximum efficiency and minimum cost and time. Compared to other thermomechanical processes, this process has distinct advantages such as low cost, short operation time, environment friendly and capability of process automation.6,7) During the FSP, the temperature of the stir zone (SZ) has been increased8–10) and the material undergoes SPD at a high temperature.11–18) Consequently, this process would result dynamic recrystallization and so the finer grains have been formed in the SZ.19–22) The refined grains produced in the SZ are equiaxed and the dislocation density of the processed material has been decreased.6) By selection of FSP operation parameters, such as the tool (traverse & rotation) speed and using a suitable cooling rate, the microstructure of the processed zone and its metallurgical properties could be altered.23,24)
Aluminum alloys are still considered as primary material for airframe construction and contribute to 60% of structural weight in case of some passenger aircraft. Due to excellent combination of low density and high strength, 2XXX and 7XXX alloys are most widely used metallic systems in the aerospace industries.5) The grains of as-rolled AA2024 sheet are extended along the rolling direction such that their size in the rolling direction reaches 100 µm and the average grain size along the sheet thickness reduces to about 50 µm.25,26)
The main elements of the AA2024 are Cu and Mg. The precipitates formed in the AA2024 are complex and difficult to analyze.27,28) Various precipitates were reported in the microstructure of AA202429,30) so that these precipitates are divided to (a) heat-treatable and (b) non-heat-treatable precipitates. Heat-treatable precipitates, such as Al2Cu, θ, θ′, Ω precipitates, are generally fine and distributed randomly in the form of bar and irregular phases in the structure.31–33) Heat-treatable Al2CuMg precipitates have an orthorhombic structure.31,34) Al7Cu2Fe is a non-heat-treatable precipitate with a tetragonal crystal structure; that forms during solidification and precipitates in the form of needles.35) Non heat-treatable Al20Cu2Mn3 precipitates that are not coherent with Al matrix can pile up the dislocation movement.36)
The operating parameters of FSP have a critical effect on microstructure and grain size of AA2024 in addition of precipitates formation. The grain size can be varied from 2 µm to 10 µm.37,38) Increasing the cooling rate of FSP would result in sub-micrometer grain size in Al–Mg–Cu alloy.39,40)
The energy of microstructure could be controlled using misorientation angle between grains. At certain misorientations, $\theta_{\sum }$, which called coincidence misorientations, a part of the lattice sites of one lattice coincide with the lattice sites of another lattice forming a coincidence sites lattice (CSL). CSL is characterized by a parameter $\sum $.41) The $\sum $ would be increased by increasing the temperature.42) In the high angle boundaries, facets have low activation energy so that high facet mobility could be occurred. On the other hand more mobility would result in increasing the facet length and so it is clear that faceted boundaries move with low mobility.43) If two boundaries meet in the sharp edge (slope discontinuity) it would be called first order and in the smooth (no slope discontinuity) it would be named second order.44) Triple junction of intersection of boundaries can control the mobility of boundaries. So, the first order boundaries have an effective role on grain growth and mobility of grain boundary.41,43) In the grain growth phenomena, facet and rough boundaries can move by different speed and this could change the shape of grain boundary.45) Sursaeva et al.45) have been shown that the mobility of facets is greater than the mobility of rough boundary, but by increasing the temperature, this difference would be decreased. In another word the heat sensibility of rough boundary mobility is more than facets.45)
As the cooling condition has a key role on microstructure of FSP, so in this research the effect of chilling during the FS-process on size and orientation of grains, distribution of precipitates and type of intersection of grain/precipitation boundaries in AA2024 have been investigated.
The 2024-T3 aluminum alloy sheets with the dimensions of 300 × 100 × 3 mm used as the precursor in this study. The chemical composition of the specimens was analyzed using Belec spectrometer and has been shown in Table 1.
FSP has been carried out using A500-Gr.C C HSS steel as the FSP stirring tool. The shoulder diameter and cylindrical pin diameter were 25 mm and 2 mm respectively. According to the preliminary investigations, the tool rotation (clockwise) and transverse speed have been selected as 315 rpm and 2 mm/s respectively. Due to the high sensitivity of Al-2024 alloy to the temperature and different precipitations, two different FSP environments, natural cooling in air and forced cooling using liquid nitrogen have been used. During air cooling (named as FSP-AC) FSP was carried out in the usual laboratory atmosphere. But during the forced cooling (named as FSP-NC) the FSP has been done by pouring the liquid nitrogen on the processed specimens. To compare the temperature of SZ under two different cooling conditions, the temperature of the SZ has measured by type-K thermocouple that has been installed under the sheet.
The microstructural characterization has been performed using transmission electron microscopy (TEM) and electron backscatter diffraction (EBSD) methods. The 200 KV JEOL JEM-2100F, JEOL, Japan TEM has been used (equipped with TEM-EDS detector: X-MaxN100 TLE, Oxford Instruments, United Kingdom) for this reason. TEM specimens have been mechanically grinded to thickness of about 100 micrometers and have been punched to the 3 mm discs and then twin-jet polished using Tenupol-5, Struers, Germany via 30% vol. nitric acid in methanol at −30°C.
EBSD has been done on the FSP specimen using a 20KV FE-SEM (A Quanta 3D FEG instrument using working distance of 20 mm and the step size of 0.1 micrometer) investigating the grain boundary misorientations. EBSD specimens have been mechanically and electrochemically grinded.
The effect of the cooling condition during FSP-AC and FSP-NC has been shown in the Fig. 1. As could be seen in this figure, when the milling head or shoulder passed on the specimen during the FSP, the temperature of 370 ± 10°C has been recorded in the FSP-AC specimen. Cooling condition during the FSP-NC, has been caused to decreasing the maximum temperature more than 70°C compared to the FSP-AC specimen. Furthermore, the cooling rate after FSP-NC has been significantly higher than FSP-AC, i.e., temperature of SZ reduced to the 10°C immediately after FSP-NC, while the temperature of this region was more than 200°C in FSP-AC and after more than 35 s the temperature of FSP-AC was about 150°C. This time has an effective role on grain growth and resulting microstructure. The maximum temperature of the upper and bottom region of SZ could be more than 100°C depending on the sheet size, cooling type and different process conditions,40) so it is clear that peak temperature at the upper region of FSP-NC specimen would be less than the bottom region.
Temperature variation during of FSP specimens in FSP-NC and FSP-AC conditions.
The effect of cooling condition during FSP on the SZ grain morphology has been shown in the Fig. 2. Comparison between Figs. 2(a) (grain orientation of FSP-NC) and 2(c) (grain orientation of FSP-NC) reveals the significant effect of cooling rate on the grain morphology and size. EBSD analysis of the FS-processed specimens shows that about 84% of the grains formed during both FSP-AC and FSP-NC have grain boundaries with a misorientation angle larger than 15°. The recrystallized grains have been formed mainly with the high-angle grain boundaries (HAGB) and less than 16% with low-angle grain boundaries (LAGB). In Figs. 2(b) and 2(d), the blue lines and red lines represent LAGB and HAGB respectively.
EBSD image of the SZ; (a) grain orientation of FSP-NC.; (b) misorientation angle of the grain boundaries of FSP-NC and; (c) grain orientation of FSP-AC; (d) misorientation angle of grain boundaries of FSP-AC.
Figure 3 shows the grain size distribution in the FSP-AC and FSP-NC conditions. As is could be seen, decreasing the temperature, the small-sized grains gradually increases at the expense of large-sized grains. Furthermore, forced cooling during FSP reduces the average grain size down to 1.75 µm, so that the grain size of more than 5% of grains are less than 500 nm, while just 1.5% of FSP-AC specimen has grain sizes less than 500 nm. During the FSP, dynamic recrystallization and grain growth occur simultaneously.19) The high strain rates of the process, provide the necessary driving force for activating the dynamic recrystallization, while the heating generated during the process activates the grain growth.46) The dynamic recrystallization has been caused the formation of fully equiaxed grains in the structure.47) Similarly Fig. 2 shows the equiaxed grains in both of FSP-AC and FSP-NC specimens. Chilling conditions didn’t affect the dynamic recrystallization.
Distribution of grain size of the SZ; (a) FSP-NC; (b) FSP-AC.
Yazdipour et al.48) proposed two mechanisms of (1) nucleation and (2) restoration for grain refinement in the SZ. In the first mechanism, the recrystallization is accompanied by the formation of new grain nuclei. The dynamic recovery provides the formation of nuclei which are essential for recrystallization to occur. During the mechanism of breaking and rotation of grains, low angle grain boundaries have been formed by high strain rate and would increase rapidly the density of structural defects, especially dislocations. Finally, the high strain and temperature in the SZ would result in coalescence and rotation of subgrains and the formation of new grains. Therefore, smaller grains are formed by splitting bigger ones and the formation of fine-grained structure is under an up-and-down procedure, i.e., an individual grain was formed with no need to form a nanometer-scale nucleus. Figures 3(a) and 3(b) show the nano grains (i.e., less than 500 nm) while all of them have HAGBs (Fig. 2). Subgrains are formed in larger grain size. So, it seems the recrystallization mechanism is controlled by nucleation and growth without subgrain coalescence.
The size of the recrystallization nuclei strongly depends on the temperature. Decreasing the temperature would reduce the size of stable recrystallized grain nuclei.49) The minimum grain size in FSP-NC specimens was less than 200 nm, while the minimum grain size in FSP-AC specimen was 400 nm. This variation of SZ grain size in the FSP-NC specimen could be related to the reduction of recrystallize nuclei.49) This is in good agreement with the reduction of peak temperature during the forced cooling condition (FSP-NC). The specimen of FSP-AC has been experienced more heat regime in compare of FSP-NC and more time duration in temperatures above 200°C, so it would be the major reason of its large grain size.
The microstructure prefers to reduce the grain boundaries energy by formation of coincidence site lattice (CSL). The misorientation angles of grains (Fig. 2(a), 2(c)) were in all direction. But most of them were 15°, 19°, 35° that could be known as critical misorientation angle41) in FCC. It is worth mentioning that based on CLS theory, special structure would be formed in critical misorientation angles. Grain boundaries of FSP-AC and FSP-NC were in faceting and roughing shape (Fig. 2) but smaller grain size of FSP-NC specimen creates more triple junction faceted boundaries in compare of FSP-AC. Grain growth has been caused to change boundaries from facet to rough. So, it is clear that rough boundaries in FSP-AC have been more than in FSP-NC. Most type of edge boundaries were first order type in both FSP-AC and FSP-NC specimen. Low angle misorientation grain boundaries (blue line in Fig. 2(b), 2(d)) were facet in FSP-AC and FSP-NC. Also, there have been some rough boundaries in FSP-AC specimen. This phenomenon is due to increasing the heat in FSP-AC process in compare to FSP-NC process.
From another point of view, the distribution of dislocations was non-uniform. As seen in Fig. 4, the density of dislocations was significantly high at dark-contrast grains, while in the light contrast grains the density of dislocation is low. Su et al.46) reported that the reduction in grain size provides dislocations with the opportunity of easier recovery processes. Although the generated heat facilitates grain growth in the FSP-AC specimen, the migration of dislocations from the grains interior to the grain boundaries will be less due to the larger grain size. Therefore, the migration of dislocations to the grain boundaries would be easier under FSP-NC conditions with smaller grain sizes. This was in good agreement with Humphreys49) that has been shown the grains with sizes smaller than 200 nm could be free from dislocations. As Yazdipour et al.48) proposed, the restoration mechanism has been activated because of grain boundary with dislocation aggregation. So, the dynamic recrystallization was controlled by both of nucleation and restoration mechanism.
TEM image of the SZ of: (a) FSP-AC and (b) FSP-NC specimen, D, L indicated to the dark contrast and light contrast grain respectively.
TEM studies indicated the significant effect of FSP on the distribution and shape of the precipitates. Sever plastic deformation (SPD) during FSP causes significant changes in the distribution and shape of the precipitates. Figure 5 displays the precipitates in the SZ. It is worth mentioning that 5 type of particles have been observed in both FSP-AC and FSP-NC. Figure 5(a) illustrates the Al2CuMg precipitate that has the orthorhombic S-phase crystal structure. This precipitate appears in near-spherical shapes in the structure. The average diameter of these precipitates was about 200 nm. The facet boundary in this precipitate was observed in [001]. Intersection edge of facet to rough was second order type.
TEM and SAED pattern of precipitates in the SZ structure: (a) Al2CuMg, orthorhombic S-phase; (b), (c) Al20Cu2Mn3, orthorhombic T-phase; (d) Al2Cu, orthorhombic Ω phase; (e) Al2Cu, tetragonal θ-phase; (f) Al7Cu2Fe, tetragonal ω-phase.
Figure 5(b) and 5(c) displays the Al20Cu2Mn3 precipitate that has an orthorhombic T-phase crystal structure. These precipitates are almost rod and spherical shape. There are researches26) that show the Al20Cu2Mn3 precipitates are observed with irregular shapes in the AA2024. Also, after FSP, these precipitates also appear with irregular shapes. The facet boundary in these precipitates were [200] and intersection of facet to rough was second order in Fig. 5(b). Figure 5(c) showed faceted boundary in [010], [011] and [011]. The intersection of facets was first order boundaries.
Figure 5(d) and 5(e) exhibits the Al2Cu precipitates that has an orthorhombic Ω and θ phase crystal structure. Ω-phase precipitates are polygonal or rod shape, whereas θ-phase precipitates are irregular shape.
The Al2Cu precipitates have a tetragonal and orthorhombic crystal structure,31) which in this study both of these structures have been observed. In addition, the heat-treatable Al2Cu precipitates are observed in irregular geometric shapes. The facet boundary was observed in [010] in Fig. 5(d). Intersection edge of facet to rough was second order type.
The SAED pattern of Fig. 5(e) indicates that the severe deformation could change the crystal structure. The ring pattern show the nano structure4) so, SAED pattern of Fig. 5(e) express the nanostructure of θ-phase.
Figure 5(f) shows the non-heat-treatable precipitate Al7Cu2Fe that has a ω-phase crystallographic structure. The average diameter of the precipitates was about 150 nm. There were three facet boundaries, [010] and [212] and [101]. All intersection of boundaries were second order type. TEM accompany with SAED analysis revealed that four different types of precipitate were formed in the stir zone, as shown in Table 2.
Figure 6 and Fig. 7 exhibit the elemental TEM-EDS map in the SZ of FSP-NC and FSP-AC specimens, respectively. According to these elemental distributions and chemical composition of different precipitates, the precipitants of each specimen have been summarized in Table 2. The S-phase (Al2CuMg) has been partially distributed at the SZ grain boundaries of FSP-AC specimen (Fig. 6), while in FSP-NC specimen, the S-phase mainly observed in the SZ (Fig. 7). In as-received Al-2024 sheet, S-phase has been precipitated on both grain boundaries and in dislocations.36,50) The mean size of S-phase is nearly in both FSP-AC and FSP-NC specimen (∼200 nm). Zamani et al.32) reported that the minimum dissolution temperature of S phase is 320°C. The maximum temperature measured during the process is about 370°C. Therefore, it is possible that S-phase artificially re-precipitated during the FSP. Also, the spherical like shape and uniform distribution of S-phase has been indicated to re-precipitation of this phase.
The distribution of various precipitates in the SZ of FSP-NC: (a) TEM image of SZ, (b) EDS map of Mn element, (c) EDS map of Fe element, (d) EDS map of Cu element, (e) EDS map of Mg element, (f) EDS map of Al element.
The distribution of various precipitates in the stir zone structure of FSP-AC: (a) TEM image of SZ, (b) EDS map of Mn element, (c) EDS map of Fe element, (d) EDS map of Cu element, (e) EDS map of Mg element, (f) EDS map of Al element.
Ω-phase Al2Cu are similar in shape and size at FSP-AC and FSP-NC specimens. Figure 7 clearly shows these precipitations have been precipitated at grain binderies. The θ-phase Al2Cu in FSP-NC are distribute at grains, while there are distribute at grain boundaries in FSP-AC.
ω-phase (Al7Cu2Fe) has been distributed in the SZ as fine spherical particles. These precipitates have similar distributions in the FSP-AC and FSP-NC specimens. The applied strain accompany with SPD causes refinement uniform distribution of second phase.51) So, it can be considered that the distribution of ω-phase because of high torsion force.
Figures 6 and 7 illustrate the uniform distribution of fine T-phase (Al20Cu2Mn3) precipitates in the SZ. Distribution and size of these particles were not affected by cooling conditions during FSP. The morphology of this particle is needle form. The T-phase precipitates in 2024 Al alloy have a pinning effect on the grain boundary and restrict the grain growth.36) Uniform and fine distribution of T-phase in both FSP-AC and FSP-NC specimen is important on final grain size and these particles don’t be solved during FSP. The microstructural changes in the stir zone FSP-AC and FSP-NC specimens have been summarized in Table 3.
The authors gratefully acknowledge the comments on this manuscript of Professor Pirouz at the Department of Metallurgy, Case Western Reserve University, Cleveland, Ohio, USA.
