Effect of Post-Deposition Heat Treatment on the Mechanical Behavior and Deformation Mechanisms of a Solid-State Additively Manufactured Al–Mg–Si Alloy

Additive manufacturing (AM) has generated interest in industrial sectors including aerospace and automotive due to the potential savings the process provides. Because AM technologies can produce near-net shape components, AM reduces the amount of post-processing steps needed in manufacturing and the amount of waste material generated. Additionally, transportation industries are interested in lightweight materials for the purpose of reducing fuel costs and improving fuel efficiency. Aluminum alloys have been studied for applications of lightweight material due to their machinability and excellent strength-to-weight ratios [1].

The majority of AM is currently facilitated using fusion-based techniques, such as selective laser melting or laser-engineered net shaping. However, fusion-based techniques have been documented to have difficulty manufacturing aluminum components [2–4]. Aluminum alloys experience a phenomenon known as solidification cracking due to the intense thermal gradients present in fusion-based manufacturing including AM and welding [3,5]. Recent advances in solid-state additive technologies, such as additive friction stir deposition (AFSD) [6–13], have demonstrated the capacity to produce fully dense depositions while avoiding solidification cracking or other volumetric defects.

The AFSD process is a friction-based AM process [6] that provides the ability to produce fully dense, lightweight alloys such as WE43 [11,14], AA6061 [7,15,16], AA7075 [17–19], aluminum metal matrix composites [12], Ti-6A-4V [20], and copper [21]. In the AFSD manufacturing process, a feed rod is pushed through a hollow rotating tool against a substrate. Concurrently, the tool may traverse along the substrate to deposit a layer of material via severe plastic deformation. The deformation behavior of the process has recently been studied [22] and found to experience a strain rate on the order of 10¹. Subsequent layers may be built upon the first to build a deposition into the net shape of a component. The layer-to-layer interface has also recently been investigated [23] and revealed that the nature of effects on the microstructure of materials printed via AFSD is material-dependent. Many of the aspects of this friction-based AM are still being investigated [13]. The AFSD process relies on severe plastic shear deformation and frictional heating to achieve dense metallurgical bonding without liquefying the deposition. Therefore, issues hampering fusion-based additive manufacturing that result from the intense thermal gradients involved during solidification, e.g., hot cracking and columnar grain growth, do not affect AFSD [24]. A recent study by Stubblefield et al. [25] was conducted to understand how best to model the AFSD process. By using experimentally-backed simulations, Stubblefield et al. were able to determine that the Fraser–Kiss–St-George constitutive model was the most effective for AFSD AA6061.

Since the AFSD technique is relatively new and not fully mature, related solid-state processes like friction stir welding (FSW) can be used to inform some understanding of the process. Friction stir welding of AA6061 produces several distinct regions of grain refinement in the general proximity of the pin tool's path [26–28]. The regions are generally referred to as the nugget zone (NZ)—the region where the most refined grains exist, directly in the tool path; the thermo-mechanically affected zone (TMAZ)—the region immediately adjacent to the tool path that has less refined grains than the NZ; and the heat affected zone—the least refined region of grains that transitions between the TMAZ and the base material. Dynamic recrystallization occurs in the NZ in FSW as evidenced by the observed high-angle grain boundaries (grain misorientation ≥ 15 deg) coupled with a low dislocation density [27,29–31]. Additionally, Feng et al. observed the dissolution of the strengthening phase β″ when examining FSW AA6061 via transmission electron microscopy (TEM) [27]. The implications of this observation is that the temperatures involved in FSW, which are greater than 250 °C, will have a significant effect on AA6061 since the β″ precipitates act as the chief strengthening phase in the T6 temper [32].

Several studies have been conducted regarding AFSD on AA6061 [7]. Phillips et al. performed a thorough investigation of process-structure relationships in as-deposited AA6061 by varying three process parameters: the rotating speed, the traversing speed, and the actuator feedrate. One key finding of the Phillips et al. study was a ratio between the traversing speed and the feedrate for acceptable builds. If the ratio became too traversing speed dominant (200 mm/min versus 85 mm/min feedrate), the deposition would be starved of material resulting in galling and voids. If, on the other hand, the ratio became too feedrate dominant (110 mm/min versus 140 mm/min traversing velocity), excess flash in the deposition would result, as can be seen in Fig. 2(b) of Phillips et al. Other key findings from the work of Phillips et al. were the grain size refinement of 93% from 200 µm to 15 µm as well as the dissolution of β″ in as-deposited AA6061 was observed for the following process parameters: rotational speed of 300 rpm, traversing speed of 127 mm/min, an actuator feedrate of 66 mm/min, a layer thickness of 1 mm, and a Z velocity of 5 mm/min.

The study performed by Rutherford et al. [15] was focused on the structure-performance relationship of as-deposited AA6061. The results of the monotonic testing of as-deposited AA6061 revealed similar strength observed in the direction parallel to the build layers (longitudinal direction) and the direction perpendicular to the build layers (build direction (BD)). The mechanical response of both directions of the deposition was similar in strength and elongation to failure to AA6061-O, which is not surprising due to the observed dissolution of the β″ precipitates reported in previous studies [7]. Furthermore, the study by Rutherford et al. looked at the fatigue behavior of the longitudinal and build directions, and found similar behavior to wrought material (AA6061-T651) as well as a similar deformation mechanism to the wrought, namely stress concentrations from the presence of constituent particles. Another key finding of Rutherford et al. is that AA6061 constituent particles are refined by the AFSD process similar to the grains, which is in agreement with what was observed in AFSD Inconel 625 [10].

In a recent study, Beck et al. [33] examined the effects of solutionization and solutionization followed by artificial aging on the tensile properties of AFSD AA6061. They reported improved strength and ductility for the solutionized and aged specimens and observed abnormal grain growth (AGG) in the material after solutionization. Furthermore, they found evidence of the precipitation of the strengthening phase, β″, via TEM investigation.

The focus of this work was to investigate and quantify the fatigue behavior of a standard ASM heat treatment [34] applied to AA6061-T6 AM build processed with AFSD [34]. In this work, the correlation between microstructural features and mechanical properties was examined. Additionally, post-mortem analysis of the fatigue samples was carried out to investigate the deformation mechanisms present in the post-deposition heat-treated samples. Finally, a microstructure-sensitive fatigue model was presented to quantify and elucidate the analysis of deformation mechanisms of cyclic deformation of post-deposition heat treatment of AFSD AA6061.

The material for this study was rolled AA6061-T651 that was cut into parallelepipeds with dimensions of 304.8 mm × 9.5 mm × 9.5 mm and processed into a deposition via AFSD. The parameters used in the AFSD process were a rotational speed of 300 rpm, traversing speed of 127 mm/min, an actuator feedrate of 66 mm/min, a layer thickness of 1 mm, and a Z velocity of 5 mm/min. These parameters were chosen to be consistent with the parameters used in previous studies [7,15]. The deposition was 419.1 mm long (from the center of the tool at the start of a pass to the center of the tool at the end of a pass) and was 63 layers tall. The samples for this study were cut from the deposition via wire electro-discharge machining in the orientation shown in Fig. 1. Also shown in Fig. 1 is the geometry of the samples tested, which has been modified from ASTM-E606.

The heat treatment procedure used in this study is based on the ASM Handbook Volume 4 for AA6061 [34] and the results of another study on AFSD AA6061 [33]. The as-deposited AFSD AA6061 samples were solutionized at 560 °C for 50 min followed by a water quench. After the quench, the samples were stored at 0 °C to prevent natural aging until the heat treatment process could be performed. The artificial aging was carried out at 160 °C for 15 h [34]. Evidence from the literature shows that as-deposited AFSD and FSW AA6061-T6 experience the dissolution of strengthening precipitates [7,27]. Although, this suggests the possibility of heat treating without solutionization, the repeated thermal cycles of the AFSD process and uncontrolled cooling of the build could result in irregular or non-homogeneous Guinier-Preston (GP) zone formation in the as-deposited state. Therefore, solutionization was implemented along with quenching to safeguard against unwanted natural aging and inconsistent heat treatment.

Samples for microstructural analysis were ground and polished incrementally to a 1 µm diamond suspension polish that was followed by a vibratory polish using 0.5 µm colloidal silica. The grain size analysis was performed in compliance with the guidelines outlined in ASTM 2627-13 [35]. The electron backscatter diffraction (EBSD) analysis was performed on a TESCAN LYRA3 field emission scanning electron microscopy with an EDAX Hikari Super EBSD camera.

Mechanical testing was performed on a servo-hydraulic load frame. The extensometer used to measure axial strain had a nominal gauge length of 5 mm. Monotonic tensile testing was run under quasi-static loading conditions in displacement control at a rate of 0.005 mm/s at room temperature. The yield strength of each monotonic tensile test sample was measured using the 0.2% offset method. Fatigue testing was conducted under strain-controlled conditions. The fatigue tests were tested under uniaxially fully-reversed conditions (R = − 1) and run at five different strain amplitudes: 0.2%, 0.3%, 0.4%, 0.5%, and 0.6%, all with a frequency of 1 Hz. After the cyclic hysteresis response stabilized, the test was continued under load-controlled conditions based on the stable hysteresis response of that sample at a frequency of 5 Hz. All mechanical testing was conducted with three specimens tested for each parameter to ensure repeatability within statistical bounds.

The multi-stage fatigue (MSF) model implemented in this study has been demonstrated to capture the fatigue behavior and help elucidate mechanisms in AM and welded materials [10,17,36,37]. The model was originally proposed by McDowell et al. [38], and has been expanded upon for other materials, processing methods including fusion-based AM and AFSD [37,39–44]. To aid the understanding of the reader, the basis of the model is briefly described. The overall fatigue life is modeled as three discrete parts. The first is the incubation phase, which is the portion of the fatigue life that is characterized by the buildup of damage prior to the nucleation of a crack. This portion of the model is informed by micromechanics on the local plasticity of microstructural features. The second portion of the model is the microstructurally small crack (MSC/PSC) growth phase, which is characterized as the portion of the fatigue life in which a crack has been nucleated but is on a length scale similar to that of an individual grain in the material. The MSC/PSC phase is informed by empirically observed crack growth near crack initiation sites on fracture surfaces. The final portion of the model is the long crack growth phase. This phase is characterized by empirical Paris Law data from linear elastic fracture mechanics.

The results of the EBSD scan are shown in Fig. 2. The grain size in the heat-treated AFSD AA6061 samples was not calculated via the EBSD software, as many grains were significantly larger than the scan area itself as shown in Fig. 2(c). Figure 2(c) also demonstrates a bimodal grain size distribution, as some areas retained the refined grains of the AFSD process, while others were consumed into very large grains during the heat treatment process. This indicates abnormal grain growth from the 15 µm observed in the as-deposited grain size [7], which is shown in Fig. 2(b) for reference and occurred because of the heat treatment process. Figure 2(a) also shows the reference feedstock material made from a rolled AA6061-T651 plate.

Grain growth due to heat treatment was expected to occur and demonstrates a clear case of highly abnormal grain growth. While some regions of grains were still on the order of 15–25 µm, other grains were larger than the scan window, which is almost 400 µm wide. Grain size calculations could not be made with statistical confidence for the heat-treated AFSD AA6061 due to the large disparity between the size of small grains in certain regions and the grains larger than the scan window. One clear takeaway is that the large grains that resulted from the abnormal grain growth are much larger than those of the wrought material negating any benefit of grain refinement on the mechanical behavior of the heat-treated AFSD AA6061. The EBSD analysis of the grain boundary characteristics revealed most grain boundaries for both large and small grains in the heat-treated AFSD AA6061 material to be high-angle grain boundaries, defined here as a grain boundary with a misorientation angle of 10 deg or more.

Due to the microstructural changes observed in heat-treated AFSD AA6061, examination of the intermetallic particles was performed to determine if there are any effects of heat treatment on the size and distribution of the intermetallic particles with respect to the wrought AA6061. Figure 3 shows a representative micrograph of (a) the wrought AA6061 and (b) the heat-treated AFSD AA6061. The particle distribution and the distribution of the aspect ratios of the particles in the wrought AA6061 are shown in Figs. 3(c) and 3(d), respectively. Likewise, the particle distribution and aspect ratio distribution of the particles in the heat-treated AFSD AA6061 are shown in Figs. 3(e) and 3(f), respectively. When comparing the particle size distributions in Figs. 3(c)–3(e) it is observed that the heat-treated AFSD AA6061 possesses more particles in the 0–5 µm² bin, and fewer particles in the 5–10 µm² and 10–15 µm² bins than the wrought AA6061. When comparing the distribution of the aspect ratios of the particles for both wrought AA6061 and heat-treated AA6061, we observe that the distributions of the aspect ratios are very similar with no major differences in the nature of the distribution. The statistical results from the analysis have been compiled in Table 1.

The monotonic tensile test results are shown in Fig. 4. The wrought AA6061-T651 used as the feedstock of the deposition is shown as a comparison to the heat-treated AFSD AA6061. The yield strength of the wrought material shown here was measured as 295.8 ± 1.8 MPa, while the ultimate tensile strength of the wrought was measured as 316.5 ± 2.2 MPa. The elongation to failure of the wrought sample was observed to be approximately 35%. The yield strength of the heat-treated AFSD AA6061 was recorded to be 308.5 ± 12.0 MPa, and the ultimate strength was observed to be 332.8 ± 8.1 MPa. The elongation to failure of the heat-treated AFSD samples was measured to approximately 30%.

From the monotonic results, we demonstrated that the effectiveness of the post-deposition heat treatment on the AFSD AA6061 successfully recovered the strength of the T6 temper possessed by the wrought AA6061-T651 material. The heat-treated AFSD samples exhibited a 4.3% increase in yield strength and a 5.2% increase in ultimate strength compared to the wrought material. It is worth noting the heat-treated AFSD AA6061 experienced approximately 5% less elongation to failure compared to the as-deposited in the longitudinal direction.

The results of the monotonic tension tests in conjunction with the results from the microstructural investigation strongly suggest that the primary strengthening phase, β″, was precipitated by the heat treatment. The monotonic results clearly show a recovery in strength in the longitudinal orientation for the heat-treated specimens, even more so than the wrought AA6061-T651 material. Furthermore, the bimodal nature of the microstructure resulting from abnormal grain growth with very large grains no longer benefits from the grain refinement strengthening mechanism. No additional alloying elements were added to the alloy during the heat treatment process, thereby ruling out a solid-solution strengthening mechanism. Likewise, the specimens were not work hardened prior to monotonic testing, which rules out work hardening as a strengthening mechanism. Therefore, the most reasonable source of the strengthening observed in the monotonic strength of the heat-treated AFSD AA6061 is the precipitation strengthening from the heat treatment. In a recent study, Beck et al. [33] used a similar heat treatment, but with a slightly different artificial aging schedule, on AFSD AA6061 and observed similar effects on monotonic strength. In that same study, TEM observations revealed the presence of β″ precipitates.

Figure 5 shows the results of the fatigue testing of the heat-treated AFSD AA6061 specimens. The heat-treated specimens tested at 0.6% experienced a fatigue life on the order of 10³ cycles to failure. At 0.5% and 0.4%, the heat-treated specimens had fatigue lives on the order of 10³–10⁴. The strain amplitude of 0.3% revealed that heat-treated specimens failed generally around 50,000 cycles, and the specimens tested a 0.2% failed on the order of 10⁶ or more cycles. An arrow indicates the specimen did not fail before reaching the runout threshold of 5 million cycles.

Overall, the heat-treated AFSD specimens exhibited similar fatigue behavior to the as-deposited AFSD across the entire range of strain amplitudes investigated. At the strain amplitude of 0.2%, we note that the heat-treated AFSD specimens did not reach a runout of 5 million cycles, while the as-deposited did for every specimen tested. This is likely due to the smaller grain size in the as-deposited specimens that act to prevent crack nucleation. Without the mechanism of smaller grains, it would be reasonable to expect both heat-treated AFSD and wrought AA6061-T651 to perform better than the as-deposited, as both possess higher load-bearing capacity.

Figure 6 exhibits the stress amplitude response to the applied strain amplitude for the various specimens tested in this work. The data shown here demonstrated relatively quick stabilization of the stress amplitude for the specimens tested at 0.2% and 0.3% strain amplitude. For the specimens tested at 0.6%, all those tested failed with the extensometer still attached. The percentage of the fatigue life spent in strain control for 0.5% is approximately 70%, and the percentage of the fatigue life spent in strain control for 0.4% is approximately 31%. The general trend observed for the post-deposition heat-treated AFSD AA6061 was mild cyclic hardening except for the 0.2% specimens, which exhibited very mild softening before quickly stabilizing under 100 cycles.

The Coffin–Manson strain-life curve generated by the post-deposition heat-treated AFSD AA6061 data is shown in Fig. 7. The Coffin–Manson curve is generated by decomposing the total strain amplitude ɛ_t into the elastic strain amplitude, Δɛ_e/2, and the plastic strain amplitude, Δɛ_p/2. The power law fit of the elastic strain data, which makes the dotted red line shown in Fig. 7, is given by the equation: $Δεe2=(σf′E)(2Nf)b$⁠, where $σf′$ is the fatigue strength coefficient, E is the modulus of elasticity, and b is the fatigue strength exponent. Similarly, the power law fit of plastic strain amplitude, shown in Fig. 7 as the gray dashed line, follows the equation: $Δεp2=εf′(2Nf)c$⁠, where $εf′$ is the fatigue ductility coefficient and c is the fatigue ductility exponent. The Coffin–Manson equation is, therefore, a combination of the elastic and plastic power law fits of the applied strain amplitude as given by: $εa=(σf′E)(2Nf)b+εf′(2Nf)c$⁠. The material constants and measured properties from the Coffin–Manson equation, i.e., $σf′$⁠, b, $εf′$⁠, and c, as well as the modulus of elasticity for post-deposition heat-treated AFSD AA6061 are provided in Table 2.

Figures 8(a) and 8(b) show the fracture surface for samples tested at ɛ_a = 0.3% (N_f = 47,470 cycles) and ɛ_a = 0.5% (N_f = 1432 cycles), respectively. Both fracture surfaces exhibit a large area fraction with severe shear dimpling indicative of final fracture. Additionally, both surfaces have regions with vertically aligned void coalescence. From previous studies [7,15], the depositions made with these parameters have been shown to be fully dense, so these voids have developed as a result of cyclic deformation and rupture of the specimen. Interestingly, these different regions are typically spaced out along 1 mm horizontal increments, which correlate to the deposition layer height of 1 mm. Thus, it is likely that the observed void coalescence during cyclic deformation occurred along the layer boundaries. Despite this deformation, it appears that the cracks that precipitated failure began near the free surface and closer to the corner on both specimens. It is important to note that the interlayer void coalescence was not observed in the as-deposited AFSD AA6061 [15], so it is likely a result of heat treatment process on the treated AFSD AA6061.

Some of the fatigue fracture surfaces investigated showed signs of mixed ductile and brittle failure modes, as shown in Fig. 9. Figure 9 shows the representative fracture surface of a specimen tested at ɛ_a = 0.2% (N_f = 1,535,618 cycles). While most of the fatigue fracture surface shown exhibits shear dimpling characteristic of ductile fracture, a region toward the left side of Fig. 9 exhibits cleavage planes with smooth faceted features indicative of brittle failure. Brittle fracture characteristics were not observed in the as-deposited AFSD AA6061 [15], therefore, it is likely that post-deposition heat treatment resulted in the introduction of new, competing deformation mechanisms. While the fracture observed in Fig. 9 did not initiate from the brittle region, the mechanism driving the mixed failure mode may still have influenced crack growth and final fracture of the sample. Further investigation into the micro- and nano-scale mechanisms will be needed to fully understand this phenomenon.

For the fatigue samples observed post-mortem, the fast fracture region of the fracture surface dominated the area fraction of the failure surface. This failure mode is common in components tested with higher nominal cyclical loads [45]. Shear dimpling can be observed in proximity to the top middle surface, but not at the free surface itself or within ∼50 µm in Fig. 10. This suggests the crack growth portion of the fatigue life was either relatively short in terms of the number of cycles or had a relatively slow crack growth rate. Once the crack reached the critical length, final rupture occurred, and the critical length of the crack was relatively short. The fact that this behavior is normal for high nominal stresses agrees with the monotonic results that showed a recovery of the strength of the AA6061-T6 temper.

One possible justification for the brittle failure regions as well as void coalescence in the fatigue samples is the abnormal grain structure that occurred as a result of the heat treatment. For instance, a crack propagating along the interface of two large grains or one large grain and many small grains would appear intergranular, i.e., brittle in nature, but other all areas of the fracture surface may experience void coalescence characteristic of ductile fracture. The void coalescence may be caused by a crack propagating into the boundary of a large grain and growing around the grain, since the grain boundary provides sufficient resistance to crack propagation. This crack would then coalesce with a dominant crack that leads to failure.

The occurrence of AGG has been observed in friction stir processed AA6061 when subjected to heat treatment in previous studies [46–52]. Due to the anisotropic nature of the microstructure after experiencing AGG, as well as the potential changes to the dominant damage mechanism, eliminating the occurrence of AGG is desirable. The three causes of AGG, as reported by Humphreys [53] are (i) anisotropic grain boundary energy and mobility, (ii) reduction in pinning forces due to coarsening or dissolution of particles, and (iii) grain size distribution leading to thermodynamic driving forces that result in AGG. Studies have shown that higher rpm rotational speeds in friction stir processed (FSP) AA6061 can reduce AGG [46,47], while different heat treatments [49], percentage of low-angle grain boundaries [51], and rolling pre-straining [52] were all shown to be effective at reducing AGG as well. Unfortunately, the current literature does not have a clear causal link to completely prevent AGG from occurring in solid-state processed materials [47].

The results of the MSF model are presented in Fig. 11. The model shows good correlation with the data throughout the various strain amplitudes tested, demonstrating reasonable accuracy for fatigue behavior of post-deposition heat-treated AFSD AA6061. We note, similar to a previously published work [15], the MSF model parameters were informed by updating those of a previously published study on AA6061 [44]. By adjusting the experimentally measured values, such as grain size, particle size, modulus of elasticity, ultimate tensile strength, and yield strength, as well as incorporating fatigue material properties like crack tip opening displacement, the MSF model was calibrated for post-deposition heat-treated AFSD AA6061.

Figure 12 exhibits the MSF model fit for the heat-treated AFSD AA6061, as well as the contributions of the incubation portion of the fatigue life and the MSC/PSC growth portion of the fatigue life. At the larger strain amplitudes tested in this present work, namely 0.006 and 0.005 mm/mm, the contribution of the incubation phase is relatively small compared to that of the MSC/PSC portion of the fatigue life. This is reasonable, since, at larger deformation, more energy is imparted into the material, which can allow a crack to quickly incubate, at which point the remainder of the life is dominated by MSC/PSC growth. If the specimens tested had been larger, it is possible the long crack growth portion of the MSF model would have had a noticeable effect. However, due to the relatively small length and width of the specimen geometry, any contribution of the long crack growth portion of the model was limited to a handful of cycles at most. Furthermore, at the lower strain amplitudes tested, primarily 0.002 mm/mm and 0.003 mm/mm, the contribution of the incubation phase was observed to dominate the small crack growth phase. Again, this finding is reasonable, as the smaller applied deformation applies much less energy to the specimen. Therefore, the number of cycles to incubate a crack to nucleation is increased. Once a crack is nucleated, it grows in the MSC/PSC phase, and while this phase is longer in terms of cycles than it was for the 0.006 mm/mm strain amplitude, it is still significantly smaller in terms of cycles than the incubation phase.

Figure 13 shows the bounded plot of the MSF model. To generate the upper and lower bounds of the plots shown in Fig. 13, the particle size distribution was considered. For this analysis, it is assumed that the particle distribution in the post-deposition heat treatment is unchanged from the as-deposited builds [15]. The upper bound is generated by holding all other constants equal and reducing the average particle size by one-half the standard deviation of the particle distribution. Likewise, the lower bound is created by holding all other constants equal and increasing the average particle size by one-half the standard deviation of the particle distribution. As can be seen in Fig. 11, the bounding fits for the MSF prove reliably capable of capturing most of the fatigue data from the post-deposition heat-treated AFSD AA6061.

This work sought to evaluate the fatigue behavior of a post-deposition heat treatment of AA6061 processed via AFSD. Based on the observations, experimental results, and the analysis performed the following conclusions can be drawn:

1. The post-deposition heat treatment process proved viable to recover the strength of the T6 temper in AFSD AA6061.

2. The strain-controlled fatigue performance of the post-deposition heat-treated AFSD AA6061 remains comparable to the as-deposited AFSD AA6061, and the load-bearing capacity of the post-deposition heat-treated AFSD AA6061 improved compared to the as-deposited AFSD AA6061.

3. The post-deposition heat treatment of the AFSD AA6061 resulted in abnormal grain growth, which is significantly larger compared to the grain size of the wrought AA6061-T651 feedstock. Additionally, the abnormal grain growth resulted in a bimodal grain size distribution throughout the material.

4. An investigation of the post-deposition heat-treated AFSD samples exhibited deformation mechanisms not observed in the as-deposited samples, however, these deformation mechanisms did not have a significant impact on the fatigue lives.

5. A microstructure-sensitive fatigue model was employed and found viable to model the performance of the post-deposition heat-treated AFSD AA6061.

This research was funded by the US Department of Defense Strategic Environmental Research and Development Program WP18-1323. The authors would like to thank the Alabama Analytical Research Center (AARC) and the Alabama Transportation Institute (ATI) for their support of this project. Permission to publish this information was granted by the Director, the Geotechnical and Structures Laboratory.

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.