Discontinuously reinforced aluminium (DRA) matrix composites exhibit improved stiffness, strength, and wear resistance and a reduced coefficient of thermal expansion over monolithic aluminium alloys and have potential structural applications in the aerospace and automotive industries ^[1]. Wide industrial applications of DRA depend on effective joining methods, which are still dependent on the use of a specific material and process. The weldability of these composites is significantly reduced by the addition of ceramic reinforcements. It is hard to achieve defect-free DRA welds ^{[2] and [3]}.

Friction stir welding (FSW) is a novel solid-state welding method applied particularly in the aerospace and automotive industries ^{[1], [2], [3] and [4]}. In the last few years, a number of investigations have been conducted to join DRA by FSW ^{[5], [6], [7] and [8]}. High joint efficiencies of 70%–90% could be obtained in the FSW DRA joints. Therefore, it is considered a promising welding technique for the DRA to avoid the drawbacks of fusion welding ^[3].

The microstructure of the FSW joints, especially the nugget zone (NZ), which suffers severe plastic deformation, changes significantly during FSW. Dynamic recrystallization (DRX) is a typical microstructure character in the NZ. For the DRA, recrystallized Al grains tend to nucleate firstly at the interface between the reinforcing particles and matrix. This would change the interfacial microstructure between the reinforcing particles and matrix. Furthermore, the size, shape, and distribution of the reinforcing particles in the NZ are also changed because the reinforcements are broken up and the sharp corners of the large particles are knocked off ^{[7] and [9]}.

The variation in the microstructure affects the strength of the DRA significantly. The interface between the reinforcing particles and the matrix is one of the important factors. The excellent bond integrity of the interface improves the load transfer at the interface, thereby increasing the strength of the DRA. The interfacial microstructure depends on the characterization of the particles and matrix as well as the fabrication process of the composites ^{[10], [11], [12], [13], [14], [15], [16] and [17]}. For powder metallurgy (PM) processed SiCp/Al composites, compounds such as Al₄C₃ and MgAl₂O₄ are formed on the interface depending on the chemistry of the Al matrix ^{[13] and [14]}. Furthermore, an amorphous layer was revealed on the interface between SiC and Al diffusion bonded at 586 °C ^[15] and in a PM processed SiCp/6061Al composite ^[16].

The size, shape, and distribution of the reinforcing particles are also important for the strength of the DRA. Flom and Arsenault ^[18] suggested that fine reinforcement increased the yield strength (YS) of DRA. Furthermore, Liu et al. ^[19] reported that the long axis of the SiC particles along the tensile direction could increase the strength of DRA. The theoretical models also prove that these factors play important roles in the strength. Wu and Lavernia ^[20] proposed a modified shear lag model that can predict the strength of the DRA well, with the size, shape, and distribution of the reinforcing particles as the important parameters.

However, for FSW DRA, the interface between the reinforcing particles and the matrix in the NZ was rarely concerned. Feng et al. ^[7] found that a number of Cu₂FeAl₇ particles were generated on the interface between SiC and Al matrix in the NZ of FSW SiCp/2009Al composite when using a steel tool. The strength of the T4-treated composite joint was considerably lower than that of the BM (base metal) with T4 tempering (solutionized at 516 °C for 1 h, water quenched, and naturally aged for seven days). Wearing of the steel tool and the formation of interfacial intermetallics made it difficult to determine the fine interfacial microstructure in the FSW composites. The interfacial microstructure and the factors that affect the strength of the FSW composites are not clear so far. Moreover, a detailed investigation is still lacking on the variation in the size, shape, and distribution of the reinforcing particles in the NZ and their effect on the strength of the NZ.

In this study, hot-rolled SiCp/2009Al composite plate was friction stir welded using an ultra-hard cermet tool to eliminate the effect of wearing of the steel tool. The microstructural evolution in the NZ and interfacial fine microstructures in both as-rolled and as-FSW composites were examined in detail. The aims were (a) to elucidate the evolution of the SiC particles and the interfacial microstructure between the SiC particles and Al matrix during FSW and (b) to investigate the factors that affect the strength of the NZ.

6 mm-thick 15 vol.% SiCp/2009Al composite plates were used in this study. The composite was produced by a PM technique, in which 50 μm 2009Al powder and 7 μm SiC powder were blended and hot consolidated at 580 °C. After hot rolling at 480 °C, the composite plates were subjected to T4 tempering (solutionized at 516 °C for 1 h, water quenched, and naturally aged for seven days). The plates were friction stir butt welded along the rolling direction at a welding speed of 100 mm/min and a tool rotation rate of 800 r/min. An ultra-hard cermet tool with a shoulder 20 mm in diameter and a conical pin 8 mm in diameter and 5.8 mm in length was used. After welding, the FSW sample was naturally aged for seven days. Furthermore, part of the FSW sample was re-treated to a T4 temper. The NZ of the as-FSW joint and T4-treated joint were named as as-FSW and FSW-T4, respectively.

The joint was cross-sectioned perpendicular to the welding direction for examination by optical microscopy (OM) and scanning electron microscopy (SEM, Quanta 600). The OM specimen was mechanically polished and etched by Keller's reagent. The interfaces between the SiC and Al matrix were analysed by transmission electron microscopy (TEM, Tecnai G2 20) and high resolution TEM (HREM, Tecnai F20). Thin foils for TEM were thinned by ion milling.

Tensile specimens with a gauge length of 5 mm and width of 1.4 mm were machined perpendicular to the welding direction, with the gauge length completely within the NZ. The tensile direction of the BM specimens was perpendicular to the rolling direction. Tensile tests were conducted using an Instron 5848 micro-tester at a strain rate of 1 × 10^-3 s^-1. The property data for each condition were obtained by averaging three test results. The fracture surfaces of the tensile specimens were observed using SEM.

Fig. 1 shows a cross-sectional macrograph of the FSW SiCp/2009Al joint. No defects were detected in the joint, indicating that a sound joint could be achieved by FSW using ultra-hard cermet tools. Similar to those in the other FSW DRA joints ^[7], the NZ in the present FSW SiCp/2009Al joint exhibited a basin shape. However, unlike in previous reports using steel tools ^[7], no onion structure was observed in the NZ in this study.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Optical macrograph showing cross-section of FSW SiCp/2009Al composite.

Fig. 2(a) shows the OM image of the SiCp/2009Al composite perpendicular to the rolling direction. The equiaxed grains of the Al matrix were ∼10 μm in size. For the as-FSW sample, the Al matrix in the NZ was characterized by fine equiaxed recrystallized grains of ∼6 μm in size, smaller than those in the BM sample. Furthermore, the edges and corners of some SiC particles became blunt in the NZ, due to the abrasion during welding ( Fig. 2(b)). The grain size of the FSW-T4 sample ( Fig. 2(c)) was similar to that of the as-FSW sample. This indicated that the growth of the Al grains would be inhibited by the SiC particles in the heat treatment process.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Optical micrographs of SiCp/2009Al composite: (a) BM, (b) as-FSW sample, (c) FSW-T4 sample.

Fig. 3 shows the results of quantitative measurements of the particle size distribution in the BM and the as-FSW sample. For the BM, the size diversity of the SiC particles was observed along the rolling and transverse directions, because the long axis of the SiC particles tended to be distributed along the rolling direction. The sizes of SiC particles in the as-FSW sample, the BM along the rolling direction, and the BM along the transverse direction were estimated to be 8.4, 9.0, and 7.5 μm, respectively. For the as-FSW sample, the size of the SiC particles was approximately the average value of the BM in the two directions. Furthermore, the fraction of SiC particles larger than 8.5 μm was higher than that of the BM along the transverse direction. Moreover, the average aspect ratios of the BM and the as-FSW samples were 1.7 and 1.5, respectively. These implied that part of the SiC particles flowed and rotated in the matrix with the stirring of the welding tool and some large SiC particles were broken up during FSW.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Quantitative measurements of the particle size distribution: (a) as-FSW sample, (b) BM perpendicular to the rolling direction, (c) BM along the rolling direction.

Fig. 4(a) illustrates the backscattered electron (BSE) image of the BM. Some circular white particles were observed in the matrix (marked as “A”). EDS (energy dispersive spectroscopy) analyses indicated that the particles contained 88.2 Al, 7.84 Cu, 1.57 Si, 1.68 Mg, and 0.73 Fe (at.%). The Fe element was introduced into the matrix in the fabrication process of the BM, forming the Al₇Cu₂Fe intermetallic compounds. In the hot rolling and subsequent cooling processes, the Al₂Cu phase nucleated on the Al₇Cu₂Fe particles and grew up ^{[21] and [22]}. When the sample was solutionized, most of the Al₂Cu phases were dissolved into the matrix. However, part of the Al₂Cu phases around the Al₇Cu₂Fe phases remained in the heat treatment process. Therefore, circular white particles were detected in the matrix.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. Backscattered electron image of SiCp/2009Al composite: (a) BM, (b) as-FSW, (c) FSW-T4.

Fig. 4(b) shows the BSE image of the as-FSW sample. There were also some white particles in the matrix. EDS analyses indicated that the irregular particles (marked as “B”) contained 87.7 Al, 7.6 Cu, 2.6 Si, 1.4 Mg, and 0.7 Fe (at.%), and the circular particles (marked “C”) contained 80.0 Al, 24.2 Cu, 3.1 Si, and 1.7 Mg (at.%).

Jariyaboon et al. ^[23] measured the peak temperature distribution during FSW of 2024Al-T351 alloy. They found that the temperatures adjacent to the NZ were 480 °C. Mahoney et al. ^[24] reported that the temperature adjacent to the NZ was measured as 420–470 °C during FSW of 7075Al-T651 alloy. For the SiCp/2009Al composite, the peak temperature of the NZ during FSW was possibly similar to the above temperature. After welding, when the sample was cooled to room temperature, the Al₂Cu phases would precipitate in the as-FSW sample and then form the circular particles (marked as “C”) and the fine white phases. Moreover, the Al₂Cu phases would also nucleate and grow on the Al₇Cu₂Fe phases. Finally, the irregular particles (marked as “B”) were formed.

For the FSW-T4 sample ( Fig. 4(c)), the number of white particles clearly decreased. The shape of the remaining particles was basically circular and the irregular particles almost disappeared. EDS results suggested that most of the particles contained Fe element. After the solution treatment, most of the Al₂Cu phases were dissolved. However, some Al₂Cu phases around the Al₇Cu₂Fe particles remained and formed the circular particles in the matrix, similar to those in the BM sample.

Fig. 5(a) shows a typical TEM image of large compounds on the interfaces between SiC particles and Al matrix in the BM. EDS analyses indicated that the phase contained 20.4 Al, 9.7 Cu, 33.2 Si, 4.2 Mg, 26.8 C, 5.8 O (at.%, marked as “D”). It was reported that impurities such as Si and oxide, adhering on the surface of original SiC particles, could react with Al, Mg, and Cu elements during the fabrication process ^{[25] and [26]}. Typical phases of Al₄C₃, MgAl₂O₄, and Mg₂Si formed on the interfaces. Furthermore, Al₂Cu phases could nucleate and grow on these compounds. Therefore, complex compounds would form on the interfaces. However, for some SiC particles, there were few impurities on the original surfaces. Therefore, a relatively clean interface with few small compounds was observed, as shown in Fig. 5(b).

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. TEM micrographs of T4-treated SiCp/2009Al composite showing: (a) large compounds and (b) small compounds on SiC/Al interfaces.

Fig. 6(a) shows a typical TEM image of the interfaces between SiC and Al matrix in the as-FSW sample. Some large compound particles were observed in the matrix around the SiC particles (marked as “E”). EDS analyses indicated that the particles contained 9.73 Al, 0.66 Cu, 27.5 Si, 25.6 Mg, 12.9 C, and 23.7 O (at.%). This indicates that the large compounds on the interfaces in the BM, as shown in Fig. 5(a), were broken off and dispersed into the matrix adjacent to the SiC particles during FSW. Therefore, most of the SiC/Al interfaces became clean after FSW.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. TEM micrographs of as-FSW SiCp/2009Al composite showing: (a) large compounds broken off from interface and (b) fractured small SiC particles with clean interface.

Fig. 6(b) shows the existence of a few fine SiC particles with relatively clean interfaces in the as-FSW sample. Such fine particles are believed to result from the knocking-off of sharp corners and edges of large SiC particles in the FSW process. Feng et al. ^[7] and Root et al. ^[8] reported similar phenomena in FSW SiCp/2009Al and Al₂O₃/7005Al composites. The new interfaces formed by the broken SiC were very clean without any compounds.

Fig. 7 shows the HREM images of the interfaces in the BM. There were three kinds of interfacial microstructures between the SiC particles and Al matrix. For the first kind, some small compounds formed on the surface of SiC particles and they were identified as Al₄C₃ phase with a crystallographic orientation relationship of(1¯016¯)SiC//(1¯1¯24)Al4C3,[6¯601]SiC//[5¯83¯3]Al4C3 ( Fig. 7(a)). However, the Al₄C₃ had no crystallographic orientation relationships with the Al matrix. For the second kind of interface, an amorphous layer was present on the interfaces between the SiC particles and Al matrix ( Fig. 7(b)). There were no crystallographic orientation relationships between the SiC particles and Al matrix on the two sides of the amorphous layer. A similar interfacial structure in the PM DRA was also reported by other researchers ^{[12] and [16]}. The third kind of interface is shown in Fig. 7(c). There were no reaction products or amorphous layer on the interface. However, no preferential crystallographic orientation relationship was observed between SiC particles and Al matrix.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. HREM micrographs of T4-treated SiCp/2009Al composite showing SiC/Al interfaces: (a) with compounds, (b) with amorphous layer, and (c) without crystalline relationship.

For the typical interfacial microstructure between the unbroken SiC particles and Al matrix in the as-FSW sample, a small Al₄C₃ particle was detected on the surface of SiC particles ( Fig. 8(a)). It had a crystallographic orientation relationship with the SiC particle,(11¯02)Al4C3//(1¯100)SiCand[246¯1]//[112¯0]. No other large compounds were observed on the interfaces. In other places on the interfaces, the SiC particle did not have an orientation relationship with the Al matrix. Furthermore, the amorphous layers in the BM, as shown in Fig. 7(b), were also observed on some interfaces in the NZ (not shown). Fig. 8(b) shows an HREM image of a typical interfacial microstructure between the broken SiC particles and Al matrix. An Al grain several nanometres in size was observed near a SiC particle and had a crystallographic orientation relationship with the SiC particle, i.e.(1¯12)SiC//(022¯)Aland[6101]SiC//[111]Al (marked by the white curve). This means that the Al matrix nucleated on the surface of fractured SiC particles in the DRX process.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. HREM micrographs of as-FSW SiCp/2009Al composite showing: (a) small Al4C3 particle on SiC/Al interface and (b) crystallographic relationship between SiC and Al.

shows the strength of the as-FSW sample and FSW-T4 sample. For comparison, the strength of the BM is also listed. The yield strength (YS) and ultimate tensile strength (UTS) of the as-FSW sample were lower than those of the BM, since the precipitates coarsened during the thermo-mechanical cycle of the FSW process (shown in Fig. 4(b)). Meanwhile, the precipitation of the coarse Al₂Cu phases in the Al matrix resulted in some solute-free regions near the interface between the SiC and matrix, where the voids were prone to nucleate and grow near the interface during tensile process ^[27]. Therefore, the elongation of the as-FSW sample decreased compared to that of the BM. In the FSW-T4 sample, the YS decreased slightly compared to that of the BM sample; however, the UTS of the FSW-T4 sample increased by 22 MPa. The elongation of the FSW-T4 sample also increased slightly compared to that of the BM. This will be discussed below in detail.

Table 1.

Table 1.

Table 1. Tensile property of SiCp/2009Al composites Sample YS (MPa) UTS (MPa) Elongation (%) BM 343 537 10.1 as-FSW 306 441 5.4 FSW-T4 331 559 11.4

Table 1. Tensile property of SiCp/2009Al composites

Fig. 9(a) presents a typical tensile fracture surface of the BM sample. The fracture surface was characterized by dimples, showing the characteristic of ductile fracture, and fractured SiC particles. Similar to the previous result ^[7], the surfaces of the SiC particles were smooth, indicating that the SiC particles were broken up in the tensile process. In the as-FSW sample ( Fig. 9(b)), the fracture surface was similar to that of the BM sample. Although the strength of the as-FSW sample was lower than that of the BM sample, particle fracture was also the dominant damage mode. A similar result was reported by Pandey et al. ^[27]. In the FSW-T4 sample ( Fig. 9(c)), the fracture surface was similar to that of the BM sample.

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. SEM images of fracture surface of SiCp/2009Al composite: (a) BM, (b) as-FSW, (c) FSW-T4.

(1) Defect-free FSW joints of 6 mm hot-rolled SiCp/2009Al composite plate were successfully achieved using an ultra-hard material tool.

(2) The size and shape of SiC particles in the NZ changed because the particles were broken up and rotated during the severe plastic deformation process of FSW.

(3) During FSW, the large compound particles on the interfaces between SiC and Al matrix fell off, whereas the small compound particles and amorphous layers on the interfaces were unchanged.

(4) The dynamically recrystallized grains nucleated on the surface of broken SiC particles to reduce the free energy of the interfaces and formed nano-sized grains. A crystallographic orientation relationship between broken SiC particles and Al matrix,(2¯12)SiC//(022¯)Al,[6101]SiC//[111]Al, was observed.

(5) The YS of the FSW-T4 sample decreased slightly due to the variation in the size, shape, and distribution of the SiC particles. The clean interface between SiC particles and Al matrix improved the load transfer and then raised the UTS of the FSW-T4 sample.