Evaluation of Additive Friction Stir Deposition for the Repair of Cast Al-1.4Si-1.1Cu-1.5Mg-2.1Zn

In many aluminum and magnesium alloys, traditional repair methods based on fusion (welding) techniques are prone to defects that can severely degrade the mechanical properties, including porosity, cracking, inclusions, and phase segregation [1–3]. In addition, in alloys like 6061 aluminum, selective vaporization can further diminish the strengthening response and corrosion resistance by the loss of zinc, magnesium, and silicon [4]. As a result, solid-state processes based on fiction stir technologies have been gaining increasing attention for repair applications [1,5]. Traditional friction stir welding (FSW) is a solid-state joining method that avoids the solidification problems that are common in fusion processes [6]. The high strains and comparatively low temperatures (typically 0.6–0.9T_melt) associated with FSW result in microstructural refinement that is associated with improved mechanical properties in the weld [7]. FSW is widely accepted to provide several benefits over traditional fusion welding, including favorable microstructures, and since the 1990s, several related technologies have been developed to extend these benefits to additive material applications [8–10]. One of these methods, additive friction stir deposition (AFSD), has the potential to enable high-deposition rates (>1000 cm³/h) during the manufacture of moderately complex geometries [11,12].

Known commercially as MELD [13,14], AFSD utilizes a rotating, nonconsumable, cylindrical tool through which a co-rotating feedstock material is fed, Fig. 1(a) [15]. During deposition, the rotating feedstock is forced against the substrate, where the dynamic contact at the interface results in frictional heating. This softens the feedstock material, which yields under the compression from the feeding apparatus and extrudes into the gap between the substrate and the tool head [9,16]. Robust metallurgical bonding occurs as a result of vertical mixing between the substrate and the feedstock, and this effect can be enhanced by the tool design [5,17–19]. Shear forces from the bottom surface of the tool head cause the extruded material to further spread between the tool head and the substrate, while translation of the tool head across the substrate leads to deposition of a continuous layer of material. Subsequent layers are created by incrementally increasing the tool height between traverses. The resultant material has been reported to exhibit fine, equiaxed microstructures that provide enhanced wear, corrosion, and mechanical properties [8–10,20].

While AFSD has obvious potential for repair applications, only two publications were identified that directly assess implementation strategies [1,21]. Griffiths et al. evaluated AFSD for hole and groove filling in 7075 aluminum plate and reported complete filling of the features combined with extensive feed/substrate mixing and good bonding in the upper portions of the repair. They also reported generally worse repair quality in the lower (deeper) portions of the features, with evidence of insufficient shear mixing at depths exceeding 1.3–1.7 mm below the original surface [21]. Finally, they observed up to 15% decrease in the hardness of the repaired volume relative to the original feed material and attributed this change to the effect of the thermomechanical history on dissolution of the hardening precipitates (e.g., loss of temper). Qi et al. evaluated a radial method to reduce the diameter of tolerance-exceeding holes in AZ31 magnesium alloy plate [3]. It should be noted that the repair material comprised plugs pre-inserted into the holes and not actively fed into the interface; thus, the method is not strictly relevant to the AFSD discussion. Nevertheless, they observed distinct microstructures associated with the stir zone (SZ), thermomechanical affected zone (TMAZ), heat-affected zone (HAZ), and base metal (BM), and documented an inverse correlation between hardness and the grain size associated with the different regions. Finally, they measured the tensile strength and compressive shear strength to be 94% and 75%, respectively, of the as-manufactured plate.

The present study represents an initial, exploratory effort to evaluate repair strategies using 6061-T6 feedstock to fill grooves in cast aluminum alloy Al-1.4Si-1.1Cu-1.5Mg-2.1Zn (Arconic/Alcoa Tradename Mic 6). The combination of cast substrate with AFSD alloy is novel within the context of additive repairs. Mic 6 is a cast alloy based on the 7XXX series of Al-Mg-Zn-Cu alloys and is primarily known as a dimensionally stable tooling plate. It is manufactured to maximize consistency between plates and is widely used in aircraft and automotive tooling, base plates, indexing tables, side plates, and various manufacturing applications. Mic 6 was selected to evaluate the performance of the AFSD process for repair of cast aluminum parts used in automotive, aircraft, and marine applications. In this context, the criteria for a successful repair include void- and crack-free deposition of well-bonded material with similar strength to the existing substrate.

The repair strategy, which was similar to that evaluated by Griffiths et al. [21], used a groove-filling repair geometry to simulate replacement of material after localized grinding was performed to eliminate cracked or corroded material, Fig. 1(b). Three crack geometries were investigated, “V” shaped, radiused (hemi-cylindrical), and square. The 6061-T6 filler was deposited by AFSD using process parameters that were informed by prior research, and process optimization was not pursued during this initial investigation. The resultant repairs were analyzed by metallography, microhardness, tensile and fatigue testing, and fractography. The goal was to explore the practical benefits and limitations of AFSD 6061 as an additive repair for cast aluminum alloys.

The substrate used in this study was a 12.7 mm thick plate of cast Al-1.4Si-1.1Cu-1.5Mg-2.1Zn, e.g., Arconic/Alcoa tradename Mic 6, and the filler metal was a 9.4 mm × 9.4 mm square bar of Alloy 6061-T6. Table 1 provides the chemical composition of the cast alloy, as determined by laser-induced breakdown spectroscopy (SciAps Z-200 Handheld LIBS Analyzer), along with the ASTM specification for the 6061 filler feedstock [22]. The specified Young's modulus, 0.2% offset yield strength, and tensile strength for the two alloys are tabulated in Table 2. As indicated in this table, these values were obtained from ASTM B209M and the relevant product datasheets [22].

The 450 mm × 305 mm substrate was prepared for AFSD by machining surface grooves with the different geometries shown in Fig. 2. The three geometries were designated as “V” for the truncated triangle (left), “R” for the hemispherical radius (middle), and “S” for the square shape (right). The V- and S-grooves were 6.4 mm wide, while the R-grooves were 12.7 mm wide, and all grooves were 6.35 mm in depth.

Process parameters for subsequent cover passes were adjusted by the operator in real time to optimize the process performance, in particular the formation of flashing caused by excessive extrusion of material from between the tool and the substrate. The relevant parameters are the tool rotation rate (RPM), feedstock feed rate (mm/s), the tool traverse rate (mm/s), and the interpass z-step (mm). These parameters determine the dependent parameters of heat input (via friction and material plasticity), force on the feedstock, and volumetric deposition rate. Some trial and error was employed during the second passes to determine the optimal parameters, which resulted in slight process variability. However, for the third and subsequent passes, Table 4, the process was stable, with little variation in the parameters applied to give a fill ratio, R_dep/R_layer, of 1.3×. Figure 3 shows the final result, where the V-, R-, and S-grooves are shown from left to right after filling followed by three (R-) or four (V- and S-) cover passes to homogenize the surface.

Metallographic samples were sectioned from several locations, mounted in a room temperature curing acrylic potting material, polished, and etched for inspection by light optical microscopy. Polishing was performed by sequentially grinding to 600 grit SiC paper, then polishing with 9, 3, and 1 µm diamond suspensions, and final polishing with 0.05 µm colloidal silica. Etching was performed by immersion for 30 s in a 10% solution of NaOH at 60 °C. Microstructures were imaged using an Olympus IX-50 inverted metallurgical microscope. Vickers microhardness depth profiles were measured on the metallographic cross sections at three locations: in the center of the AFSD feature and at approximately the quarter-width locations on either side. The testing used a 200 g load and a 13 s dwell time and was performed using a LECO LM-248 AT microhardness tester.

Tensile and fatigue testing were performed using a 100 kN load cell on an Instron 8801 servo-hydraulic fatigue testing system. Strain measurements less than 2% were made with an Instron 2620-series dynamic axial clip-on strain gauge extensometer using a 12.5 mm gauge length. Tensile tests were paused at 2% strain to remove the extensometer, and strains >2% were estimated from the crosshead displacement. The geometry of the tensile specimens was derived from the guidelines provided by ASTM B557 for rectangular sub-sized specimens [23], except that the grip ends were shortened to decrease the overall sample length due to the spacing of the AFSD repairs on the substrates. The geometry of the fatigue specimens was similarly based on the continuous radius geometry found in ASTM E466 [24], and the surfaces were finished by lightly sanding to 400 grit (44 µm). Tensile testing used a constant crosshead speed of 0.085 mm/s (5.08 mm/min), and fatigue testing used load-controlled, completely reversed cycling (R = −1) at 15 Hz. Figure 4 shows the geometry of the tensile and fatigue samples, which were 89.4 mm long, 19.05 mm wide, and 6.35 mm thick. The samples were oriented perpendicular to the grooves such that the gauge length was centered on the AFSD repairs. The thickness of the samples, 6.35 mm, corresponds to the depth of the repaired grooves as illustrated in the figure. For reporting purposes, the samples were designated by the groove geometry (“V,” “R,” or “S”), the test type (“T” or “F”), and a sample number. Thus, sample designation VT1 corresponds to the first tensile sample from the V-groove.

The initial, as-received substrate microstructure was documented as a baseline for comparison to the AFSD-repaired material. Figure 5 shows the microstructure of the cast Mic 6 plate to contain large, equiaxed grains, often exceeding 125 µm. Colonies of intermetallic inclusions can be seen forming at the grain boundaries, and to a lesser extent the grain interiors, while a fine dispersion of plate- or needle-like particles is present in the grain interiors. Limited shrinkage porosity was observed throughout the microstructure.

Cross sections of the three AFSD-filled grooves in the Mic 6 substrate are shown in Fig. 6. In these images, the deposited material has been machined down so that the top surface is coplanar with the original surface of the substrate. In all three images, the SZ extends to a depth of 2.3–2.6 mm below the surface. The SZ shows a slight asymmetry related to the relative directions of rotation and translation of the tool, which interact such that one side of the tool is rotating toward the translation direction (called the “advancing” side) and the other is rotating away from the translation direction (called the “retreating” side). In Fig. 6, the advancing side in the original fill pass was on the left side of the images. Consideration of the relative size of the filler metal and the tool head, 13.3 mm along the diagonal and 38 mm in diameter, respectively, reveals that the width of the SZ is associated with the diameter of the tool rather than that of the filler material.

The SZ is one of four distinct microstructures observed in the samples: SZ, TMAZ, HAZ, and BM [25,26]. The SZ comprises heavily deformed filler alloy that was deposited during the AFSD process. It often has a striated morphology associated with material flow, and refined, equiaxed grains resulting from dynamic recrystallization [21,26]. The TMAZ refers to regions where the substrate material has been plastically deformed at elevated temperature. It is characterized by evidence of material flow, breakup of inclusions, and distortion of interfaces or boundaries. The HAZ is similar to that of a welding process, but due to the relatively low temperature of the AFSD process, the resultant HAZ is significantly smaller and less pronounced. The four microstructures (SZ, TMAZ, HAZ, and BM) are identified at locations 1–4 in Fig. 6(b) and are shown in more detail in Fig. 7.

Figure 7(a) shows the microstructure of the filler alloy to be typical of aged 6061 [27,28], containing finely distributed particles of excess Mg₂Si [4,25,27], and coarser Fe-containing intermetallics (typically > 1µm) that range in composition (e.g., Al₇Cu₂Fe, Al₅FeSi, Al₁₂(Fe,Cr,Mn)₃Si) [17,27,29]. The TMAZ microstructure in Fig. 7(b) is pronounced, consisting of severely deformed substrate with highly aligned inclusions and clear evidence of breakup of the larger intermetallic colonies described with respect to Fig. 5. Due to the relatively low process temperature, the HAZ in Fig. 7(c) is subtle, while the underlying substrate microstructure (Fig. 7(d)) is consistent with the base material shown in Fig. 5. Due to the coarseness of the cast microstructure, Figs. 6(b)–6(d) are presented at lower magnification than Fig. 7(a).

Higher magnification images of the three groove geometries from Fig. 6 are shown in Fig. 8. The images, which were stitched together from multiple overlapping photographs, reveal all three grooves to have been completely filled, but with incomplete bonding in the lower portion of the grooves. This is consistent with the results reported by Griffiths et al., where they reported poor repair quality at depths exceeding ∼1.7 mm during groove- and hole-filling in 7075 aluminum [21]. The images also show a mixture of substrate and filler material in the grooves, with obvious microstructural evidence of severe plastic deformation in the entrained substrate material. In the bottom 2.5 mm of the V-groove shown in Fig. 8(a), the interface between the substrate and the filler metal is nearly undistorted, and exhibits open gaps on both sides (indicated by the arrows). This indicates a lack of deformation of the substrate at that depth, and a resultant lack of mixing and/or bonding at the interface. In between the top of the unbonded interface and the top of the image, the microstructure exhibits increasingly severe deformation, with concurrent distortion of the original V-groove geometry. At the top of the image, the microstructure is fully stirred, as evidenced by the absence of any residual evidence of the original groove location or geometry.

Figure 8(b) shows the interface at the bottom of the R-groove to be in intimate contact and free of voids, but without any distortion, that would indicate shear mixing. As with Fig. 8(a), there is a section of highly deformed substrate material that has been entrained into the groove, with some evidence of mixing provided by the jagged and diffuse appearance of the interface with the filler metal. The features indicated by the arrows at the top of the image are gouges introduced during sample preparation and should not be interpreted as part of the structure. In combination, the void-free interface and the extent of intermixing suggest promise for this repair geometry. In contrast, AFSD filling of the S-groove shown in Fig. 8(c) resulted in a substantial crack in the substrate. This crack, partly shown at the left side of the image, extends well below the base of the groove as indicated by the arrows shown in Fig. 6(c). In addition to the crack in the substrate, the image shows lack of bonding along part of the substrate–filer interface (arrows).

Hardness depth profiles performed at the locations indicated in Fig. 6 are presented in Fig. 9. For reference, the microhardness was measured for the as-received filler and the substrate plate, and those results are provided in Table 5. The baseline data indicate the hardness of the as-received 6061 filler and the Mic 6 plate to be 105 HV0.2 and 70 HV0.2, respectively. Figure 9 shows the measured hardness at the surface of the AFSD filler metal to be approximately 70 HV0.2 for the V- and R-groove geometries and slightly lower for the S-groove. This measured hardness in the deposited alloy is consistent with values reported in a prior investigation of AFSD of 6061 processed under similar conditions, 55–73 HV [4,30]. The ∼30% decrease in hardness relative to the as-received alloy is attributed to dissolution of the β” hardening phase that occurs above approximately 500 °C and has been documented in the stir zone during both FSW and AFSD [4,31]. It should be noted that the decrease in hardness is expected to correlate with a concurrent loss of strength. In fact, from Table 2, the measured hardness is similar to a -T4 temper, which has a σ_YS of 110 MPa. This suggests that, under appropriate processing conditions, the AFSD 6061 could approach the nominal σ_YS of 105 MPa of the Mic 6 substrate.

Figure 9 shows that the hardness of the filler tends to decrease with the increasing depth inside the groove, which is an artifact of a gradient in the cooling rate. Beneath the SZ, the hardness of the substrate increases significantly in the TMAZ, where it has experienced refinement of the microstructure and breakup of the inclusions, reaching 100–110 HV0.2 adjacent to the filler metal. At increasing depths, the hardness of the substrate decreases and approaches that of the as-received substrate, 70 HV0.2, which is also indicated in the plots. Careful inspection of the placement of the indents reveals that the excursions from this behavior that occur at 5.2 and 5.6 mm in Fig. 9(a), at 2.8 mm in Fig. 9(b), and at 5.6 and 6.0 mm in Fig. 9(c) correspond to individual tests where the profile crossed a filler–substrate interface for only one or two indents.

Tensile testing was performed in accordance with ASTM B557-15 [23] on selected locations from each AFSD-filled groove. The material suitable for testing was limited by the prevalence of unbonded interface noted above, especially for the V- and S-grooves, and by the length of the filled grooves (∼200 mm). As a result, the R-groove geometry was predominately used for fatigue testing, and the V- and S-groove samples for the tensile testing. As described with respect to Fig. 4, the tensile samples were sectioned from the top 6.35 mm of the substrate so that the central portion of the gauge length was composed of AFSD material and excluded the underlying substrate alloy. The values for Young's modulus were calculated from the initial slopes of the stress versus strain curves, the yield strengths were calculated at a 0.2% offset using the same slopes, and the tensile strengths were determined from the maximum load attained during each test. Because most samples exhibited ductile behavior without a clear fracture point, the elongation at failure was determined from the value of the strain when the load fell below 10% of the maximum [23]. Table 6 summarizes the results and shows Young's moduli ranging from 39.6 to 71.5 GPa, yield strengths from 56.8 to 115.1 MPa, and tensile strengths 93.2 to 149.3 MPa. For comparison, the strength values are also given as a percentage of the baseline values for Mic 6 provided in Table 2. The single R-groove sample yielded the highest strengths, exceeding the maximum V-groove and S-groove values for the yield and tensile strengths by >36% and >23%, respectively. This is consistent with the observations made with respect to Fig. 7(b), where the R-groove/substrate interface was shown to be void/gap free, and lacking any obviously unbonded sections. It is likely that the decreased measured strength in the V- and S-groove samples resulted from incomplete bonding between the filler and substrate, Figs. 7(a) and 7(c), so that the actual cross-sectional area was less than the sample cross section, thus increasing the actual applied stress in the material. It is interesting to note that the yield strength of 115 MPa measured in RT1 is within 5% of the 6061-T4 value of 110 MPa, corresponding to the measured hardness in the AFSD filler alloy. For the tensile strength, the difference between the measured value and the -T4 specification is larger, close to 27%. This larger discrepancy results from the increasing contribution of poorly bonded regions of the filler/substrate interface at the higher applied stresses.

Figure 10 shows stress versus strain curves for the tensile samples with the highest and lowest ultimate strength. All the tensile samples exhibited some variation of the discontinuous behavior shown by the curves in Fig. 10. The discontinuity manifested by the transient decrease in the stress is attributable to partial failures that occurred during loading, and fractography suggests that the observed behavior is associated with kissing bonds. Kissing bonds are a specific type of solid-state bonding defect where the two surfaces are in contact with only weak bonding [27]. In FSW of aluminum alloys, the phenomenon has been associated with failure to break through the native oxide layer [28]. Since the oxide film is fragmented by shear stresses generated by the tool motion, kissing bonds are associated with insufficient stirring as evidenced by a lack of deformation in the groove geometry. This occurs in the present samples between the well-stirred region near the surface and the unbonded region near the base of the grooves.

The fracture surface from the highest tensile strength sample, RT1, is shown in Fig. 11(a). The image is oriented such that the top edge of the sample corresponds to the original surface of the plate, and the bottom edge of the sample corresponds to the base of the groove (recall Fig. 4). Figure 11(a) provides an overview of the fracture by optical microscopy, showing two regions of fine and coarse ductile failure, designated by (b) and (c) in the Figure, and a smooth region near the base of the groove, designated by (d). Figures 11(b) and 11(c) show SEM micrographs taken from the locations indicated in Fig. 11(a), revealing dimpled structures that confirm ductile behavior. In contrast, the surface from near the base of the groove, shown in Fig. 11(d), shows evidence of quasi-cleavage fracture typical of failure at oxide or intermetallic layers. Combined with the previous observation of poor interfacial mixing at the base of the grooves, illustrated by the sharp interface in Fig. 8(b), the absence of ductile failure suggests that the portion of the fracture surface represented by the microstructure shown in Fig. 11(d) was not well bonded during the AFSD process. The shape of the stress strain curve for the sample, Fig. 10, suggests that the bond strength exceeded the yield strength of the nearby alloy, but limited the overall tensile strength of the sample. In contrast, the curve for sample ST3 suggests that unbonded sections of the interface, and possibly additional cracking in the substrate (Fig. 8), decreased the effective cross-sectional area, thus increasing the actual stress on the intact material. This caused the sample to fail at an apparent (engineering) stress that was lower than the actual stress in the material.

The R-groove geometry was selected for fatigue testing based on the metallographic observations and the tensile result. The number of specimens precluded construction of a full S-N curve, so the tests were performed at a fixed stress amplitude, σ_a, of 80.6 MPa, which is equal to 70% of the yield strength, σ_YS, of RT1. The testing utilized a fully reversed (R = −1), load-controlled stress cycle, and the individual tests were run to failure or to one million (N = 10⁶) cycles, whichever occurred first. The results in Table 7 show a bimodal behavior, with the cycles to failure, N_f, either less than 10⁴, or greater than 10⁵ cycles. Fractography was performed to determine the source of this bimodality.

Figure 12 presents the fracture surfaces from the fatigue samples exhibiting the highest (excluding RF1) and lowest cycles to failure, RF3 and RF5, respectively. In the images, the top edge of the sample corresponds to the top of the filler material, while the bottom edge is at the approximate bottom of the groove. Figure 12(a) shows the fracture surface from RF3, which failed at 6.32 × 10⁵ cycles. The surface shows a thin ligament of fractured substrate alloy near the bottom of the image, multiple fatigue initiation sites indicated by the arrows, and typical flat fatigue fractures extending to a pronounced shear lip in approximately the top third of the sample. The initiation sites are at the bottom of the filled groove and at the approximate axial location indicated by the dashed line in Fig. 8(b). The observed morphology indicates that the fracture initiated near the bottom of the groove and propagated though the filler alloy rather than along the filler/substrate interface. In contrast, the fracture surface from sample RF5, which failed after only 3.7 × 10³ cycles, shows initiation sites near the top of an unbonded section of filler–substrate interface, as indicated by the arrows in Fig. 12(b). Below this level in the groove, the fracture surface shows the previously described characteristics of poor bonding, follows the groove geometry, and deviates substantially from the flat morphology demonstrated in RF3. This illustrates that, under these test conditions, the performance of the samples with the shortest fatigue life was dominated by incomplete interfacial bonding. As a result, the actual stress amplitude was higher than expected due to the decreased effective cross section, and the fatigue fractures initiated in the middle of the samples where the bond strength improved. This mode of failure occurred in two of the samples, RF5 and RF6.

Additive friction stir deposition of aluminum alloy 6061 was investigated as a means to fill machined grooves in a substrate plate of cast aluminum alloy Mic 6 (Al-1.4Si-1.1Cu-1.5Mg-2.1Zn). The test geometries were intended to simulate a repair in which damaged material was mechanically removed and replaced using an additive process. Three groove geometries were evaluated by metallographic inspection, and tensile and fatigue testing. The results indicated that, for the process conditions and tool/groove geometry used in this study, the effective repair depth was limited to 2.3–2.6 mm. Below that depth, the interface between the filler and substrate materials exhibited poor bonding associated with insufficient shear deformation. Within the effective depth limit, the groove shape does not have a significant impact; however, the geometry of the R-groove appears to have some advantages if the process is to be optimized for deeper repairs. For example, the R-groove geometry avoids stress concentrating corners that led to cracking in some of the S-groove samples.

The thermomechanical path experienced by the 6061-T6 filler alloy during the AFSD effectively erased the artificial age-hardening associated with the -T6 specification and left the filler alloy with a hardness equivalent to the naturally aged -T4 condition. Combined with the results of tensile testing, this suggests that under appropriate processing conditions, the deposited 6061 can approach the strength of the Mic 6 substrate. Furthermore, the results of fully reversed axial fatigue testing revealed performance that was generally consistent with FSW 6061. This suggests that there is potential to utilize AFSD 6061 as a viable repair process for cast Mic 6 alloy.

In closing, the results of this investigation support further development of repair methods based on AFSD as a means to replace damaged material. The limitations of the method that were identified here, the depth of effective repair and the loss of temper in precipitation hardened filler alloy, may be mitigated with further optimization of tool and groove geometry and the selection of filler material to match substrate properties. In addition, process parameters that control the heat input and rate of shearing at the interface, including tool rotation rate, feedstock feed rate, tool translation speed, and tool–substrate gap (z_step), should be further evaluated for their impact on the properties of the deposited layers. Finally, the loss of feedstock temper during deposition suggests that there is an opportunity to evaluate active (in process) cooling and/or postdeposition heat treatment.

The authors acknowledge MELD Manufacturing Corporation (Christiansburg, VA) for sample fabrication.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request. The authors attest that all data for this study are included in the paper.

This material is based upon work supported by the Office of Naval Research under Award No. N00014-18-1-2339. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Office of Naval Research.