Anomalous Mitigation in Phase Evolution Impacting Thermal Stability in a Rapidly Solidified AA5182 Al–Mg Alloy via Continuous Thin‐Strip Casting

This study, for the first time, reports an anomalous phase evolution and its impact on the thermal and oxidation behavior of rapidly solidified AA5182 Al–Mg alloy strip fabricated using a novel thin‐strip (TS) continuous casting technique. The microstructural analysis reveals a through‐thickness gradient microstructure of distinct types, morphology, and fractions of nonequilibrium Al–Mn–Fe intermetallic and Al–Mg eutectic phases. The rapid solidification experienced in TS casting effectively mitigates the formation of phases, particularly with the Al–Mg (β‐Al3Mg2) eutectic, being nearly absent from the near‐surface regions. The total fraction of formed phases is considerably lower than that in the slowly cooled direct‐chill counterpart. The solute macrosegregation also shows an inverse Mg segregation profile toward the strip surface primarily due to a higher degree of matrix supersaturation closer to the strip surface. Modeling of solidification successfully predicts the influence of cooling rate on the fractions of the nonequilibrium eutectic phase, agreeing well with the experimental data obtained from image analysis. Heat treating the samples over a broad temperature range unveiled unexpected improvements in oxidation resistance and excellent thermal stability, attributed to the absence of the eutectic β‐phase in subsurface regions. The research findings have practical implications for improving the properties of sheet Al–Mg alloys.


Introduction
[17] The most common method of producing ingots that are subsequently processed to sheets is direct chill (DC) casting.[26] Twin-roll casting and belt casting are developed continuous casting techniques that can produce aluminum sheets with improved mechanical properties. [27,28][31] In contrast, the cooling rate in DC casting is slower, ranging from 1 to 100 K s À1 near the mold surface and dropping to as low as 0.1 K s À1 in the center of the DC ingot. [18,31,32][35] A newly developed continuous casting method by Hazelett-CASTechnology, which we term thin-strip (TS) casting, represents a new variant of the belt casting technique which though simpler than DC casting still requires further thermomechanical processing via hot prior to cold rolling and heat treatments.As shown in Figure 1a, the TS method has a shorter production route compared to the conventional DC method, which typically involves intermediate thermomechanical processes such as scalping, homogenization, hot rolling, and cold rolling before the final sheet production. [36,37]In contrast, the TS method will be more cost-effective, efficient, and fast, requiring less energy and labor consumption than traditional methods.49] Figure 1.a) Schematic plots illustrating the TS and DC casting methods.b) The longitudinal cross section of a stack of 3 mm-thick strips, a panoramic OM image of the TS displaying distinct regions formed through the thickness in the cast direction, and micrographs depicting equiaxed grain structure of cast alloys in c) TS-R2 and d) midthickness (i.e., core) DC regions.
In the meantime, the interchangeable nature of Fe and Mn may lead to different designations for the same compound in the literature. [46,49,50]In Mg-bearing alloys, the stoichiometry of Al-Mn-Fe compounds is not entirely clear, and metastable phases such as Al x (Mn,Fe) with 4 < x < 4.4 are reported.In this regard, higher Mg content can promote the formation of Al 3 (Mn, Fe) at the expense of Al 6 (Mn,Fe). [47,49]Similarly, Al(Mn,Fe)Si compounds exhibit different stoichiometries and designations, such as β-Al 5 FeSi, α-Al 15 (Fe,Mn) 3 Si 2 , and α-AlFeSi. [45,51,52]The formation of eutectic phases in Al-Mg alloys during solidification is influenced by the partition coefficient, (k), which is the ratio between the composition of solute between the liquid and solid phases at thermal equilibrium and the cooling rate. [53,54][57] Cooling rate and solute supersaturation are closely related phenomena.[60] The cooling rate affects the supersaturation degree in Al-Mg alloys, leading to several benefits such as enhanced solid-solution strengthening, corrosion resistance, formability, weight reduction, and improved heat treatability and thermal stability. [61,62]his study reports a comprehensive analysis of the anomalies in microstructure evolution, eutectic phase formation, and thermal behavior of rapidly solidified AA5182 Al-Mg alloy strips fabricated using TS casting.We undertake a comparative analysis with a conventional DC cast counterpart, with the experimental procedure and solidification modeling approach being specifically utilized to evaluate the impact of the cooling rate difference between the TS and DC casting methods on the eutectic phase formation.The outcome of this exploration holds promise for paving the way for in-depth studies and developments in the future of sheet metal production.

Results and Discussion
Here, the evolution in microstructure, eutectic, and intermetallic phases in the TS AA5182 strip is investigated and compared to the DC counterparts.The segregation of solutes at both microand macrolevels in the TS cast is examined and the underlying mechanisms are discussed and highlighted with regard to the influence of the solidification cooling rate.Furthermore, a previously developed analytical model of rapid solidification is used to predict the fraction of the nonequilibrium Mg-rich eutectic phase and establish the effect of cooling rate, that is, between TS and DC casting.Furthermore, the oxidation resistance and thermal stability of the examined samples are evaluated, considering their respective microstructures and the extent of solute supersaturation/segregation.

Solidification Microstructure
Figure 1b depicts the longitudinal cross section of a stack of 3 mm-thick strips, that is, parallel to the casting direction.
As can be seen from the through-thickness optical image of one strip, the size and morphology of the second-phase particles (i.e., the nonequilibrium intercellular/intergranular eutectic and intermetallic phases formed during solidification) indicate a gradient structure consisting of three different regions.The first region, "R1", located right below the surface, contains ultrafine phases.The second region, "R2", extends from R1 to the proximity of the strip centerline and mostly comprises elongated and needle-shaped colonies of intercellular/intergranular phases.The third region, "R3", is situated around the strip centerline, containing widely spread and irregularly shaped colonies of eutectic and intermetallic phases.The approximate area fractions of R1, R2, and R3 regions are 0.30-0.33(%1 mm of the total thickness), 0.41-0.43(%1.25 mm of the total thickness), and 0.24-0.29 (%0.75 mm of the total thickness), respectively.R2 is the most extensive region throughout the thickness, while R3 with coarse particles has the lowest area fraction.The inset images in Figure 1b, marked by colored arrows, show the differences in size, volume fraction, and morphology of formed phases among various regions.The emergence of different characteristics of phases is indicative of a change in the cooling rate from the strip's surface toward the centerline.It appears that the throughthickness variation of the cooling rate results in distinct solute supersaturation and microsegregation levels which, in turn, give rise to the above differences in volume fraction, size, and morphology of the phases/colonies.According to the observations, it can be inferred that the regions R1 and R2, which constitute most of the strip volume, are likely to exhibit higher degrees of supersaturation and lower microsegregation compared to R3.
Figure 1c,d shows the etched grain structure found in the TS and DC samples in their as-cast state.The micrograph in Figure 1c was selected from the R2 region in the TS sample, and the DC micrograph (Figure 1d) was captured from a position between the surface and the center of the cast ingot.Microscopic images show the development of equiaxed grains in both samples, despite having different grain sizes.Mean grain diameter was determined using the line intersection method, where the results from several arbitrary lines were averaged.The mean grain diameter for regions R1, R2, and R3 was found to be 48.2AE 1.7, 82.8 AE 2.4, and 112.3 AE 3.1 μm, respectively.In contrast, the DC sample showed a relatively consistent grain structure from the ingot surface to the center with an average grain diameter of 135 AE 4.1 μm.The significant variation in the mean grain diameter could be linked to differences in the thermal history and solidification cooling rate across the three regions examined in the TS sample.In addition, it is important to consider the effect of grain refiners on the final grain size in cast alloys.As mentioned in the authors' earlier publication, [36] the TS method is efficient at utilizing grain refiners such as Ti particles added to the melt.Assuming that an equal amount of grain refiners is added to TS and DC melts, the TS melt would contain a higher density of active nuclei due to the less time available for the particles to agglomerate and settle.However, the potential partitioning of Ti during solidification could also be a factor affecting the grain refinement which will be addressed later in this context.
As stated earlier, the DC sample was cut from the ingot core (i.e., the central region of the cross section).However, being %95 mm thick, the ingot did not exhibit a significant difference in the overall microstructure between the core and next to the surfaces (i.e., the edge parts).This is due to an insignificant difference in the cooling rates experienced through the cross section/thickness (e.g., cooling rate at ingot surface vs. ingot core).On the other hand, larger ingots may show more significant differences and therefore different microstructures due to gradual changes in their through-thickness cooling rate.Figure 2 exhibits the microstructure of several DC samples which were prepared from regions close to the surface (Figure 2a), core part (Figure 2b), and areas between the two (i.e., quarter-thickness) (Figure 2c).As shown, the second-phase particles size and their volume fractions are consistent in all three samples denoting comparable intergranular/intercellular spacing formed during solidification, which subsequently controls the formation of intergranular/intercellular phases upon cooling in the solid state (i.e., Al-Mn-Fe intermetallic and Al-Mg β-phase eutectic phases).
The R3 region in the TS sample, which is %0.7-0.75 mm thick, was identified based on the presence of shrinkage porosity.In the TS casting process, the metal layers solidify in contact with the cooling belts pushing the molten metal toward the centerline, gradually closing in until the gap between the two belts reaches the final strip thickness.In the R3 inset of Figure 1b, solid black globular regions have formed as a result of the solidification shrinkage trapped in between the cells/dendrites from the two solidifying fronts closing in on the cast centerline.Figure 3 displays the optical microscopy (OM) and scanning electron microscopy (SEM) images of the centerline porosity at different magnifications.The images show a spread of irregularly shaped intercellular/interdendritic porosity with comparable dimensions to the eutectic/intermetallic phases.The isolated regions formed among solidification fronts are branched flower-shaped and are not connected to each other.This lack of connectivity likely, to some extent, mitigates stress concentration concerns from the reduced load-bearing section of the R3 region (compared to the other regions).Al-Mg alloys with considerable Mg content are more susceptible to the occurrence of defects and distortion due to increased solidification shrinkage.This means, in the casting processes, there is a tendency for porosity formation or other types of defects in the final product.However, this centerline porosity appears to be less pronounced in DC casting, that is, where slower cooling rates allow for a more effective compensation for shrinkage porosity with a continued flow of molten liquid.Nevertheless, it appears the centerline porosity can be still present even in DC cast ingots, as shown by Chaijaruwanich et al. [63] Therefore, tweaking the process parameters is likely to alleviate the extent of centerline porosity rather than its complete elimination.

Phase Evolution and Identification
The partition coefficients of alloying elements and the cooling rate can have a synergistic effect on the volume fraction of undesirable intermetallics in Al-Mg alloys during solidification.A higher partition coefficient means that the solute element has a greater affinity to segregate toward the solid phase, leading to the formation of more compounds.On the other hand, a higher cooling rate during solidification can suppress the formation of intermetallics (due to limited diffusion time which noticeably decreases the solute microsegregation/rearrangement), promoting the formation of a more refined and homogeneous microstructure.Therefore, the combination of a low partition coefficient and a high cooling rate can be favorable for producing Al-Mg alloys with a reduced amount of intermetallics, thus improving the alloy's mechanical properties.Mg has a relatively low partition coefficient in the aluminum melt (k < 1) and its atoms tend to segregate toward the liquid phase, resulting in a relatively high amount of solidification shrinkage (i.e., the degree of which depends on the Mg content). [64,65]That is why Mg-bearing aluminum alloys are difficult to cast due to their high susceptibility to porosity formation and hot tears during solidification.It is therefore feasible to more effectively maintain the Mg atoms in solid solution upon casting at higher cooling rates (i.e., suppressing the formation of Al-Mg eutectic phases).This opens opportunities for newly developed techniques such as the TS casting to promote Mg supersaturation, lessen the microsegregation, and lower the amount of undesirable intermetallics, thus resulting in enhanced mechanical properties of Al-Mg alloys.It, therefore, allows for a lower requirement for the Mg content in the alloy, hence reducing the production cost as well as the environmental impact.
In pursuit of understanding the impact of cooling rate on solute partitioning and macrosegregation, spark-optical emission spectroscopy (OES) analysis was conducted in several trials and the findings are presented in Figure 4.The bulk chemical composition of different regions (R1, R2, and R3) through the thickness of the TS was obtained by stepwise milling of coupons from both the top and bottom surfaces of the strip.Figure 4a shows the locations of the spark analysis at 0.5, 1, and 1.5 mm from the strip centerline on both sides, and Figure 4b-f presents the mean solute concentrations.Variation of Ti content along with the main alloying elements was explored to track the effect of grain refiner addition in the melt.It is worth noting that %0.008 wt% of Ti particles were added to the melt as a grain refiner for both TS and DC casting.However, the measured results were higher than the nominal addition due to the presence of the element in the feedstock (i.e., the commercial purity Al bars).The trend followed the general path based on the partition coefficient of Ti in Al-Mg alloys, which is to segregate toward the melt (k < 1).Meanwhile, the slightly higher amount of Ti in the R3 region together with coarser grain size suggests that the cooling rate is the dominating factor controlling the grain evolution.
Mg has a lower partitioning coefficient in the aluminum melt due to its limited solubility, whereas other solute elements, such as Fe, Mn, and Si, have relatively higher coefficients (k > 1), tending to partition toward the solid phase.[66][67][68] In DC casting, which has lower cooling rates, lower Mg content and higher Fe, Mn, and Si concentrations are typically present within the initial solidified layers compared to the terminal layers.In contrast, in TS (as shown in Figure 4b), the highest Mg content was obtained in the first solidified layer (i.e., for the R1 region), and the lowest for the R3 region, indicating an inverse segregation of Mg.As shown from the error bars, due to the presence of centerline porosity with nonuniform distribution, the R3 region exhibits the largest fluctuations in the measured composition.The concentration of Mn was somewhat uniform, and no significant macrosegregation was observed.A slight decrease in Si and Fe concentrations was observed toward the centerline, following the regular partitioning of these elements in aluminum.The inverse segregation of Mg is believed to be due to the higher degree of matrix supersaturation closer to the strip surface, together with an incomplete filling of the strip centerline with the Mg-rich liquid at the terminal stage of solidification (i.e., the formation of centerline porosity).The latter phenomenon is also potentially exacerbated by the two solidifying layers closing in and exerting pressure on the hot, remaining liquid in the centerline.The variation of Mg supersaturation in the matrix (i.e., within the α-Al phase) is further examined and confirmed below using point energy-dispersive X-ray spectroscopy (EDX) analysis together with analytical simulation of solidification.
Table 1 summarizes the results of point-EDX analyses conducted on various phases and regions of both TS and DC samples.The phases labeled in the inset SEM images are color coded correspondingly with their characteristic locations specified by the point number.As can be inferred from Table 1, the matrix composition as represented by the average values of points 5 and 6 shows a decreasing trend from the surface (%Al(Bal.)-Mg(4.30)-Fe(0.06)-Mn(0.29)-Si(0.05)for the R1 region) toward the centerline region (%Al(Bal.)-Mg(3.63)-Fe(0.04)-Mn(0.04)-Si(0.01)for the R3 region) where it exhibits a value closely matching that of the DC sample, that is, %Al(Bal.)-Mg(3.47)-Fe(0.03)-Mn(0.02)-Si(0.02).These values and their variations are in accordance with the evolution of solidus composition obtained from the analytical simulation of rapid solidification for a velocity-dependent Al-Mg phase diagram as presented below.
The supersaturation of the α-Al phase can be explained by the nonequilibrium binary Al-Mg phase diagram at the Al-rich side as a function of solidification velocity given in Figure 5.The binary diagram is reconstructed based on velocity-dependent parameters such as liquidus line slope (M V L ), solidus line slope (M V S ), solid composition at dendrite tip (C s ), liquid composition at dendrite tip (C t ), maximum solid solubility of Mg in Al (C SM ), and partition coefficient (k v ).The related equations for calculating required parameters are given later in this context (i.e., Equation ( 7)-( 13)).[71] The applied parameters and their values for the given equations are listed in Table 2: For simplicity, the equilibrium liquidus and solidus lines in the equilibrium binary Al-Mg diagram were estimated as linear.Moreover, the two solidification velocities (R) of %0.1 and %1 m s À1 , specified in Figure 5, represent the average values for the DC and TS casting, respectively.It is noteworthy that the casting speed in TS is 1 m s À1 , which more or less represents the average solidification velocity experienced across the strip thickness, that is, with the subsurface R1 region solidifying uuuuuuuuuuuuat a relatively higher velocity versus the centerline R3 region at a lower velocity.Given the typical cooling rates reported (%100 and %1000 K s À1 for DC and TS casting, respectively), a temperature gradient (G) of 10 3 K m À1 can be estimated for the phase diagram reconstruction under both TS and DC casting conditions (i.e., with cooling rate = G Â R).The nonequilibrium diagram anticipates a eutectic temperature drop from 723 to 703 and 672 K, respectively, for the DC and TS casting, that is, due to a deviation from the equilibrium condition.This causes an increase in the maximum Mg solubility from 17.4 to 20.6 and 24.1 wt% for the DC and TS casting, respectively, giving rise to the solidification of supersaturated α-Al cells/dendrites via retention of more solutes in the aluminum matrix.This is in accordance with the EDX results where a nearly complete supersaturation of Mg (i.e., a matrix Mg content of %4.30 wt% with respect to the average alloy composition of 4.5 wt%) has been measured in regions near the surface (i.e., the average of points 5 and 6 values in the R1 region) where the highest solidification velocity is experienced upon the melt's first contact with the cooling belts.Further down toward the center, the measured Mg content approaches that of the R3 region (%3.5 wt%), which also matches the value obtained for DC, both of which experience a velocity of %0.1 m s À1 .This also provides a higher driving force for the precipitation of Mg-rich particles (i.e., GP-zones and β 00 ) during postsolidification thermal cycles which can contribute to hardness and strength enhancement.EDX-SEM microchemical analyses were also used to investigate the composition of compounds.Based on the compositions obtained, it can be concluded that the Al-Mn-Fe intermetallic phases form at higher temperatures while the Al-Mg eutectic phases form at lower temperatures toward the end of the solidification process.The hierarchy of phase evolution during solidification has an impact on ultimate morphology and appearance.The results showed that the β-Al 3 Mg 2 stoichiometric phase was the main Mg-rich eutectic, being present in all regions identified in TS and DC samples.Elemental maps and point EDX analyses revealed that the β-phase could potentially contain a higher Si content than the Al-Mn-Fe compounds.It also showed a varying composition for Al-Mn-Fe particles throughout the thickness in the TS sample.The different particles in the TS-R1 and TS-R2 regions had a varied m ¼ C Mn þC Fe C Al ratio (C stands for composition in weight percent).As the concentration of Mn and Fe in Al-Mn-Fe intermetallics is considered interchangeable, their summation is taken into account.However, in the TS-R2 region, two different particle morphologies exist, including fine Chinese-script particles and elongated particles, with the fraction of elongated particles decreasing toward the TS-R3 region.The 2.5 < x < 5 range was observed for the elongated particles, indicating the formation of a metastable Al x (Mn,Fe) compound.On the other hand, for the Chinese-script particles, the 6 < x < 8.5 range was obtained representing the value for the formation of stable Al 6 (Mn,Fe) phase in Al-Mg alloys.From the literature, [1,9,49] a range of 4 < x < 4.4 is reported for the metastable Al x (Mn,Fe) phase, which is a slightly tighter range than the one obtained for the TS sample.This difference could be due to a higher cooling rate and less diffusion time available for the further transformation of these metastable particles.Al 6 (Mn,Fe) particles were not detectable in the TS-R1 region, but their fraction increased further away from the strip surfaces at the expense of Al x (Mn,Fe).In the TS-R3 region, the fraction of metastable Al x (Mn,Fe) particles is negligible.Therefore, it can be concluded that the local cooling rate affects both the stoichiometry and morphology of Al-Mn-Fe intermetallics, while it can influence only the morphology of the eutectic β-phase.Likewise, the EDX analysis showed that the metastable Al x (Mn,Fe) particles exhibit a higher Fe/Mn ratio than Al 6 (Mn,Fe) particles, which is consistent with the throughthickness segregation trend of Fe (shown in Figure 4c) dipping around the centerline.
To study the effect of cooling rate and solute partitioning on microstructure evolution, phase characterization and phase fraction analysis were carried out, that is, via obtaining semiquantitative phase fractions through image analysis to compare the degree of microsegregation in TS and DC samples.The results are presented in Figure 6 and Table 3, which include backscattered electrons (BSE)-SEM images captured from R1, R2, and R3   regions of the TS (Figure 6a-c), as well as from the mid-thickness region of the DC ingot (Figure 6d), with the corresponding elemental distribution maps obtained using EDX-SEM (Figure 6e-g).
The images of the same magnification and resolution (i.e., 1536 Â 1103 pixels) in Figure 6a-c show that the morphology and fraction of eutectic and intermetallic compounds vary substantially across the strip thickness in the TS sample, with the lowest fraction emerging in the R1 region (Figure 6a).Despite the higher macrosegregation and cooling rate, the microsegregation in the R1 region is evidently minimal among all regions, that is, as the compounds form in between the solidifying cells/dendrites, the local cell size impacts the evolution of phases.In a recent study by Yin et al., [36] the secondary dendrite arm spacing (SDAS) was measured in TS and DC castings of AA6005 alloy using a model developed by Kurz and Fisher (Equation ( 3)) [72] and correlated to the cooling rate.
where λ 2 is the SDAS (μm), T : [75] It was demonstrated that the size of secondary phases and the resulting microsegregation were significantly influenced by the SDAS.Specifically, the λ 2 value in the TS sample was found to be less than half of that observed in the DC sample, which led to the formation of finer  interdendritic phases and a higher level of solute supersaturation in the TS sample matrix. [36]BSE-SEM images in Figure 6 revealed the formation of two general groups of phase colonies in each of the identified regions, Al-Mn-Fe (white-colored chunks) and Al-Mg (black-colored phases) compounds.The same combination of phases was observed in the DC sample but with a coarser size and appearance.The morphology of the phases in the TS-R3 region was comparable to those in the DC sample, despite their size difference.Microscopy analysis showed that, in the TS-R1 region, fine, semispherical, and occasionally elongated Al-Mn-Fe intermetallics were formed in the intercellular/interdendritic regions.In the TS-R2, frequent needles and irregular chunks of the Al-Fe-Mn particles were formed.Moving away from the surfaces toward the strip thickness, the Al-Mn-Fe compounds became coarser and mostly exhibited a Chinese-script morphology, including fewer elongated phases (the TS-R3 region).For the Al-Mg eutectic phase, a similar trend was observed, commencing with the sparsely dispersed, fine, semispherical particles in the TS-R1 region.Further down, this gradually transforms into branched and elongated interdendritic/intergranular phases through the TS-R2/R3 regions, suggesting their formation at the terminal stages of solidification.
Scanning transmission electron microscopy (STEM) analysis was carried out to study the ultrafine phases in the microstructure of the TS sample, which could not be resolved using SEM.Thin discs were prepared from the TS-R2 region adjacent to the R1 region for capturing the images shown in Figure 7.The panoramic bright-field transmission electron microscopy (BF-TEM) image shows the presence of different types of ultrafine intergranular and transgranular particles.Submicrometer phases that were decorated with multiple particles were detected near the grain boundaries in the form of colonies in close proximity to the β-Al 3 Mg 2 phase, making them difficult to distinguish from the β-phase using SEM.Elemental maps obtained using high angle annular dark field (HAADF)-STEM confirmed the formation of Al-Si eutectic flakes with dendritic/branched morphology, which were surrounded by faceted Mg-bearing Al(Mn,Fe)Si (Figure 7b) and fine Al-Mn-Fe particles (Figure 7c).The higher Mg content in Al(Mn,Fe)Si particles than in Al-Mn-Fe possibly points to their relatively lower stability, that is, one that could transform into more stable Al-Mn-Fe particles by further rejecting their Si and Mg content.Colonies of coarser Al-Mn-Fe particles had been also previously detected via SEM as the primary intermetallic compounds (Figure 6).It can therefore be suggested that the formation of Mg,Si-free primary Al-Mn-Fe compounds at higher temperatures leaves behind a liquid that is enriched in Mg and Si, thus facilitating the formation of transient flakes and particles at lower temperatures, some of which can later transform into fine Al-Mg compounds with β-Al 3 Mg 2 stoichiometry.Accordingly, the likelihood of nonequilibrium Al-Si eutectics formation diminishes further away from strip surfaces due to the sufficient time available for solute diffusion and transformation.Moreover, ultrafine transgranular particles can be also observed in the BF-TEM images in Figure 7a and the STEM-HAADF image in Figure 7d.The sparsely distributed irregular particles, pointed to by yellow arrows in Figure 7a, are inclusions likely formed during solidification, and the ones by blue arrows are Al-Mn-Fe needles elongated along a preferred crystallographic orientation to the aluminum matrix.Some randomly oriented fine Al-Mg-Si needles were also present in the aluminum matrix.
The nanoscale area chemical composition obtained by STEM revealed that the aluminum matrix in the TS sample had Mg concentration ranging from 3.51 to 4.32 wt%, away from the fine particles.This indicates that the TS sample can achieve a retention rate of %78-96% of Mg in solid solution in comparison to its nominal composition.In contrast, the DC counterpart showed a lower Mg content of 2.91-3.60 wt% with an approximate retention rate of 64-80% in its matrix.This is in accordance with the higher volume fraction of eutectic phases in the DC sample.The lower volume fraction of eutectics in the TS sample increases the likelihood of forming metastable/transient Mg, Si-rich GP-zones, and β 00 -phase precipitates, as well as β 0 particles, during postsolidification heat treatments, that is, due to the higher solute levels in its matrix.This affects the grain boundary precipitation kinetics, leading to a higher fraction of ultrafine intergranular precipitates that can pin the grain boundaries, thereby enhancing the thermal stability and mechanical properties of the TS sample.Interestingly, although Al-Mg alloys are typically categorized as work-hardening alloys, the high solute supersaturation levels of Mg and Si in the Al matrix of the TS sample may also provide some precipitation hardening potential through artificial aging treatment.

Phase Fraction Analysis
The phase fractions of the TS sample were determined semiquantitatively using SEM images with the same magnification.The images of the same resolution (i.e., 1536 Â 1103 pixels) were processed to enhance their contrast and detect small details.The white-colored Al-Mn-Fe phases and black-colored β-phase were measured separately by masking the other phase to increase accuracy.The area fractions of different phases were calculated for different regions of the TS sample and compared to those of the DC sample, with the results listed in Table 3.The phase formation was found to occur more likely in regions with a lower cooling rate and a higher available diffusion time.Interestingly, for the TS-R3 region, the phase fractions were comparable to those of the DC sample.It is therefore important to explore the influence of cooling rate on the emergence of each phase type.The phase fraction ratios in the TS-R3 region over the TS-R1 region were measured as 16.44% and 7.11% for eutectic Al-Mg (β-phase) and intermetallic Al-Mn-Fe, respectively.This indicates that the β-phase is more sensitive to the cooling rate and that reducing the cooling rate leads to the formation of more Mgbearing compounds (i.e., at R3 vs R1 regions).Moreover, the Mg supersaturation level in the alloy is more sensitive to the cooling rate than the Mn and Fe levels due to the different temperature ranges at which these compounds form.The β-phase forms at a lower temperature and during the final stages of solidification, which allows for a longer period of time for Mg diffusion.Faster cooling rates not only reduce the time available for diffusion during solidification but also during the postsolidification cooling phase when the strip is still in contact with the cooling belts.In contrast, the diffusion of solutes for the Al-Mn-Fe phases, which have a higher melting point, is affected less by the postsolidification cooling rate.To compare the TS and DC samples, one can also consider the area fraction ratio of Al-Mn-Fe intermetallics to β-phase in each region.The data obtained indicate that the ratios in the R3 region of the TS and DC samples are similar, suggesting that the cooling rates experienced are comparable.Since the TS sample consists of regions with varying phase fractions, and to make a better comparison with the DC sample, it is necessary to differentiate between the regions and their respective thickness fractions.Using the rule of mixture, the total phase fraction can be calculated as follows.
where f T Ri denotes the thickness fraction of each identified region and A Ri β and A Ri AlMnFe refer to the area fraction of each compound in that region.When all regions are taken into account, the calculated phase fraction in the TS sample is less than half of that in the DC sample (2.50 AE 0.27% vs 5.61 AE 0.73%).In order to compare the phase fractions more accurately, two different methods were used to convert the area fractions into volume fractions, assuming the irregular shape of the phases.The lower boundary of the probable range was obtained by assuming that the whole phases fit into a sphere with the least surface-to-volume ratio, while the upper boundary was obtained by assuming they could fit into a cube.The conversion results are presented in Table 3.By converting the results, the volume fractions for the TS-R3 region and the DC sample were relatively similar.When all regions of the TS sample were taken into account with their thickness fractions, the average volume fraction of all phases in the TS sample was found to be 3.46 vol%, which is less than one-third of that for the DC sample (11.65 vol%).It is noteworthy to state that in the following section, the modeling of the eutectic phase fraction is discussed, and the results of the employed analytical modeling are presented in Figure 8.As shown, the volume fraction of the eutectic phase (i.e., β-phase) is plotted versus solidification velocity.Since the applied modeling method is based upon volume fraction, for dimensional consistency, the conversion of area to volume fraction was carried out for the experimentally obtained data.Nevertheless, despite the possible approximation errors, the assumptions applied in converting the experimentally obtained area fraction to volume fraction (referring to Table 3) resulted in values reasonably correlating with those predicted by the model.

β-Al 3 Mg 2 Eutectic Phase Fraction Modeling
To further explain the effect of the solidification cooling rate on the evolution of the terminal eutectic phase in the binary Al-Mg system, the analytical model developed by Sarreal and Abbaschian [69] is employed.As stated earlier, Al-Mn-Fe intermetallics form at higher temperatures as the initial phases during solidification.Considering the minor Fe and Mn solute contents in the nominal composition of AA5182 alloy and the absence of any eutectic phase in the (Fe,Mn)-Mg system, it is reasonable to assume that the formation of the Al-Mn-Fe compounds at higher temperatures depletes the melt from Fe and Mn such that the terminal melt solidifies in the binary Al-Mg system, that is, to form the β-Al 3 Mg 2 eutectic only.It is noteworthy that due to several limitations and drawbacks, the so-called Scheil model of solidification is unable to accurately predict the eutectic phase fractions and microstructure evolution under nonequilibrium, rapid solidification conditions. [69,71,73]First of all, from the experimental observations and analyses (Figure 6 and Table 3), the higher cooling rates in the TS casting compared to the DC casting lead to the formation of considerably lower eutectic phase fractions.Moreover, the TS has experienced a varying cooling rate through the thickness (regions of R1, R2, and R3).With regard to both cases, the Scheil equation that does not factor in the solidification velocity would be inapplicable.
The employed analytical model (Equation ( 5)-( 17)) is shown to return a reasonable accuracy under nonequilibrium, rapid solidification conditions. [69,71]The model considers back diffusion and dendrite tip undercooling from Burden and Hunt's model, [76] as well as the eutectic temperature decline formulated by Jackson and Hunt's model [77] to predict eutectic volume fraction as a function of solidification velocity (solid/liquid interface velocity).In addition, the model is capable of correlating some essential parameters such as solute redistribution and partition coefficient, as well as solidus and liquidus lines as a function of solidification velocity.In a study, Sarreal and Abbaschain demonstrated that the volume fraction of the eutectic phase and the aluminum solubility limit significantly deviate from the values predicted by equilibrium phase diagrams at a high cooling rate. [69]In this study, the model was employed to predict the volume fraction of the β-phase ðf AlÀMg E Þ in a binary Al-4.5Mg (wt%) alloy as follows.In the above equations, C t is the liquid composition at the dendrite tip, C s is the solid composition at the dendrite tip, M V s and M V L are the velocity-dependent solidus and liquidus line slopes of the aluminum side of the Al-Mg binary diagram, k v is the velocity-dependent partition coefficient, P c is the interface Peclet number for solute redistribution, and C SM is the velocity-dependent maximum solubility of Mg in Al.All other parameters along with their values are summarized in Table 2.It is noteworthy that for extracting the liquidus (T L ) and solidus (T s ) temperatures of binary Al-4.5Mg (wt%) alloy, as well as the equilibrium partition coefficient (k E ), solidus and liquidus lines were considered linear up to the eutectic point and their slopes (M L and M S ) were calculated from the equilibrium Al-Mg diagram (Table 2). [78]The factors considered and the assumptions used for the derivation of the above equations are also thoroughly explained in a recent study by Qin et al. [71] Here, in this study, the model differentiates between the TS and DC casting via the definition of solidification cooling rate ( Ṫ) estimated from the temperature gradient (G) along the solid-liquid interface and the solidification velocity (R).
Given the previously estimated values of R, %1 and %10 À2 -10 À1 m s À1 , and T : , %10 3 and %100 K s À1 , for TS and DC casting processes employed in this study, [30,31,36,79] respectively, both casting methods will be represented by a G of %10 3 -10 4 K m À1 for which the model's predicted results for f AlÀMg E are plotted as a function of R in Figure 8.The experimentally measured volume fractions of the β-phase in the DC sample, as well as in the R1, R2, and R3 regions of the TS sample, are also displayed in the form of ribbons with mean values to follow the ranges given in Table 3.As shown, at very low solidification velocities of <%10 À6 m s À1 , the model predicts no eutectic phase formation which is in accordance with the equilibrium phase diagram, given that the maximum solid solubility of Mg in Al exceeds the Mg content in the alloy (i.e., 17.4 vs 4.5 wt%).Upon increasing the solidification rate, with the emergence of a nonequilibrium solidification and thus a deviation from the equilibrium phase diagram, the model predicts the onset of the formation of the eutectic phase at solidification rates of >%10 À6 m s À1 (i.e., in the case of G = 10 3 K m À1 ), followed by a rapid rise in the eutectic volume fraction until a maximum, and then a gradual decline back to zero.As can be followed from the plot and the ribbons, the model predicts a eutectic volume fraction for the TS sample that is close to the one experimentally obtained for the R2 region, that is, at a solidification rate of %1 m s À1 which represents the overall casting speed.One would expect a higher-than-average solidification rate for areas solidifying next to the cooling belts, that is, within the R1 region, for which the model's prediction confirms the experimental results exhibiting a lower volume fraction of the eutectic phase.On the other hand, for the centerline region of R3, a lower-than-average solidification rate can be assumed, thus predicting a higher volume fraction.The experimentally measured eutectic volume fraction range for the DC sample partly overlaps with that of the R3 region of the TS sample, which is fairly close to the values predicted by the model, that is, at solidification rates of %10 À4 -10 À1 m s À1 , regardless of the temperature gradient.It is interesting to note the remarkable agreement between the model's predictions and the experimental results; first, what appears as a slight deviation from equilibrium via DC casting at slow cooling rates leads to the emergence of a relatively large volume fraction of the eutectic phase (which would not have existed in equilibrium) and, second, further deviation from equilibrium via TC casting (i.e., higher cooling rates) leads to a gradual decline in the volume fraction.The model results are therefore in good agreement with the experimental measurements, both quantitatively and qualitatively, demonstrating its effectiveness in predicting phase evolution in nonequilibrium solidification.

Oxidation Resistance and Thermal Stability
][82] Preserving the Mg atoms in a solid solution and reducing the grain boundary phases can hinder both the inward oxygen and outward Mg atoms diffusion, thus improving oxidation resistance.Grain boundary phases, especially those rich in Mg (i.e., β-Al 3 Mg 2 ), facilitate diffusion by providing easy oxygen pathways.[85] The oxidation behavior of both TS and DC samples was examined after exposure to atmospheric oxygen at different temperatures in the range of 150-600 °C for 72 h.Considering the DC ingot size, cubic coupons were all cut from the midthickness region to ensure their consistency in the microstructure.No specific surface preparation was performed prior to the tests.This was due to the primary objective of this study which was to compare the overall structural stability and surface alterations in both TS and DC specimens under potential high-temperature service conditions.Consequently, heat treatments were conducted on the surfaces in their as-manufactured state for practicality.
Figure 9a-i shows the surface images captured from the exposed samples at 50 °C temperature increments.The surface analysis of coupons revealed that the onset of visible discoloration (as an indication of surface oxidation products) was %400 °C (Figure 9e) and %450 °C (Figure 9f ) for DC and TS coupons, respectively, suggesting a delay in the phenomenon by %50 °C in the TS sample.The oxidation products in the form of fine surface nodules formed at 550 °C for the DC sample (Figure 9h).Moreover, exposure up to 600 °C resulted in severe blistering and slight dimensional change due to partial melting in the DC sample (Figure 9i).In contrast, no observable surface phenomena (including blistering, nodule formation, and melting) were detected for the TS coupons tested in the entire range of exposure temperatures demonstrating a remarkably higher resistance to oxidation compared to the DC coupons.
Figure 10a displays a surface shot of the DC sample's blisters and fine nodules, with the OM images of the polished cross section right below the blisters (pointed to by red dots) shown in Figure 10a 1 ,a 2 .As evident, the subsurface microstructure is associated with internal oxidation and massive porous regions of oxidation products, as well as deeply branched inward pathways.
According to their morphology, it seems that the pathways follow the grain boundaries containing coarse particles.Given the depth of the involved region, it can be inferred that the grain boundaries in the DC sample (originally containing a relatively large amount of the continuous β-phase, referring to Figure 6d) are capable of satisfying the oxidation mechanisms upon the high-temperature exposure.In contrast, given the finer size, lower amount, and the isolated nature of the intergranular compounds in the TS sample (specifically, β-phase in the R1 region), grain boundaries do not provide oxidation pathways.From Figure 10b, it is evident that the TS sample remains unchanged after being exposed to atmospheric oxygen at 600 °C for 72 h, with no signs of blisters, nodules, or any alteration in dimensions.Figure 10c shows the as-cast microstructure of the DC sample containing intergranular Al-Mn-Fe intermetallics (mainly Al 6 (Mn,Fe) in a brownish color) and β-Al 3 Mg 2 eutectic particles (in black), and Figure 10d depicts the microstructure of heat-treated DC sample at 600 °C far from the surface.As evident, the heat treatment leads to the coarsening of Al-Mn-Fe intermetallics, as well as the spheroidization and partial dissolution of the β-phase.The complete dissolution of the β-phase is therefore probable upon longer exposures to high temperatures, which may eventually halt the inward growth of oxidations pathways.However, one could argue that the oxidation kinetics, especially at higher temperatures, supersedes that of dissolution such that subsurface regions quickly form deeply rooted oxidation pathways.In contrast, with the overall lower microsegregation level and the much smaller fraction of the β-phase in the subsurface regions of the TS sample (i.e., the R1 region), the complete dissolution of the β-phase can occur faster, thus further contributing to a higher oxidation resistance in TS sample.After all, the β-phase has a nonequilibrium nature, making it thermodynamically unstable, given the fact that the AA5182 alloy contains only 4.5 wt% Mg, which is well below the maximum solubility of the α-Al phase for Mg (%17 wt%).The fraction evolution of the eutectic β-phase was well elucidated earlier in this context using an analytical modeling approach.Figure 10e shows the cross-section OM image captured from the subsurface region of the TS sample (i.e., R1 region) exposed to atmospheric oxygen at 600 °C for 72 h.As shown, neither oxidation products/porous regions nor the inward-growing oxidation pathways are formed, which are in good agreement with the surface appearance (Figure 9i).In the meantime, Figure 10f shows an OM image far from the strip surface (i.e., R2 region) at the same heat-treated TS sample as in Figure 10e.Both Figure 10e,f indicate coarsening of Al-Fe-Mn intermetallics (brownish-colored particles) and spheroidization of β-phase (appear as sparse black dots).It is evident that, even after long exposure, the TS sample exhibits a much lower coarsening rate and thus a significantly smaller particle size evolution.Also, partial dissolution along with spheroidization of the remaining β-phase can be observed in both the R1 and R2 regions of the TS sample.As a common observation in TS and DC samples, after exposure to 600 °C, a nonuniform coarsening of Al-Mn-Fe particles led to the formation of long needle-shaped particles denoting that despite their high-temperature nature, most of the particles are transformed.
The thermal stability of cast samples was evaluated by Vickers hardness measurement after heat treatment cycles applied in the  range of 150-600 °C.Several indentations were made on regions far from the surface to ensure the measurements were not affected by oxidation.The average hardness values are reported in Figure 11a,b plots the normalized hardness of the treated samples with respect to their initial as-cast state to more clearly highlight the hardness retention in the samples.As shown, a hardness increment with respect to the as-cast state was observed in the range of 200-250 °C for the TS sample, suggesting the possible formation of hardening phases of GP-zones and β 00 precipitates.A more stable and coarser β 0 /β-phase forms at higher temperatures (>250 °C), which results in a hardness reduction.In contrast, the DC sample shows a consistently decreasing trend with temperature, also exhibiting an increasingly steeper slope than that of the TS sample at temperatures above 250 °C, that is, indicating a faster kinetics and/or a lower volume fraction of such transformations.Given the relatively coarse-grained as-cast microstructure of both samples, grain coarsening during the heat treatment is most likely a minor contributor to the hardness decline.Therefore, it can be suggested that the coarsening of Al-Mn-Fe intermetallics along with the formation of transient hardening phases, as well as their subsequent transformation and coarsening into β 0 /β phases, are the main phenomena controlling the hardness evolution of the heat-treated samples.The final hardness retention in the treated samples reached 89.9% and 79.5% for the TS and DC samples, respectively, insinuating higher thermal stability of the as-cast TS sample compared to the DC counterpart, which is directly linked to their distinct microstructure and evolution of intergranular particles.It is indeed important to further understand the associated mechanisms of oxidation and possible products as a function of microstructural variations.The detailed discussion on the underlying oxidation mechanisms and products as well as phase evolution during the postsolidification heat treatments will be covered by the scope of another context which is the subject of an ongoing investigation.

Conclusions
This comprehensive study, for the first time, reports the anomalous mitigation in phase formation and microstructural evolution and their critical effects on the oxidation behavior, thermal stability, and properties of rapidly solidified AA5182 Al-Mg TS.The findings, summarized below, shed light on the unique characteristics and critical impacts of rapid solidification achieved through a novel TS continuous casting technique: 1) A distinct through-thickness gradient microstructure was revealed in the TS, with varying types, morphology, and fractions of nonequilibrium eutectic and intermetallic phases.The TS casting effectively mitigated the emergence of phases, particularly the β-phase, which were nearly absent from the strip's subsurface regions, with the highest cooling rate experienced.This was in contrast to the conventional DC casting, which exhibited a considerably higher total fraction of compounds; 2) In the TS sample, the chemistry and morphology of the Al-Mn-Fe intermetallic colonies formed within the microstructure were shown to vary based on the interchangeable amounts of Mn and Fe elements, that is, the sum of the two.In the DC sample, the morphology and the stoichiometry were found to be constant; 3) The solute macrosegregation study unveiled an anomalous Mg concentration profile inversely toward the strip surface, attributed to a higher degree of matrix supersaturation in the subsurface regions.This was further clarified by reconstructing a nonequilibrium binary Al-Mg diagram under rapid solidification conditions, where a higher maximum solubility for Mg was predicted for the subsurface regions of TS casting, experiencing the highest cooling rates; and 4) The thermal stability analysis demonstrated a remarkable oxidation resistance of the TS sample, that is, in significant contrast to the DC counterpart.Even when exposed to atmospheric oxygen up to 600 °C, no surface oxidation was observed in the TS sample, due to the unique isolated morphology of the eutectic phases and the absence of coarse intergranular β-phase.Additionally, the heat-treated TS coupons exhibited significant thermal stability throughout the temperature range of 150-600 °C.The TS sample, particularly, showed an anomalous hardness increase attributed to the formation of transient GP-zones and β 00 -phase in the range of 200-250 °C, followed by the transformation to β 0 /β-phase colonies.
These findings have significant implications for enhancing the properties, reducing production costs, and minimizing environmental impact in the manufacturing of Al-Mg sheets.Future research directions could focus on further optimizing the TS casting process to achieve even more controlled microstructures with enhanced mechanical performance.Additionally, exploring the potential of alloy modifications could provide opportunities for tailoring the properties of the TS to specific application requirements.Overall, this study contributes to the advancement of knowledge in the field of Al-Mg alloy development and paves the way for innovations in the production of high-performance cast alloys.

Experimental Section
The TS AA5182 Al-Mg alloy was cast at a strip width of 350 mm and a thickness of 3 mm, using a CASTechnology-Hazellett caster.The strip's chemical composition was Al(Bal.)-Mg(4.38 AE 0.01)-Mn(0.23AE 0.002)-Si(0.04AE 0.004)-Fe(0.11AE 0.001) (in weight percent).The newly developed caster enabled solidification cooling rates of %10 3 K s À1 to be achieved.To investigate microstructure evolution, an as-received DC AA5182 cast ingot with a 94 Â 205 mm 2 cross section, having a similar chemical composition of Al(Bal.)-Mg(4.51AE 0.03)-Mn(0.42AE 0.005)-Si(0.09AE 0.001)-Fe(0.17AE 0.003) (in weight percent), was compared with the TS cast strip.Both cast samples were analyzed in their as-cast state only.Bulk chemical compositions were determined using spark-OES (Spectro Analytical Instruments, SPECTROMAXx LMX09).Microstructure characterization was performed using an optical microscope (Olympus, BH2-UMA), a scanning electron microscope (FEI, NanoSEM 450) equipped with an energy-dispersive X-ray spectroscopy detector (Bruker, XFlash 6160), and a transmission electron microscope (FEI, Tecnai Osiris STEM) operated at 200 kV.To reveal the grain structure, a modified Barker's reagent was used under polarized light.Metallography steps were carried out using emery papers of different grits and finalized by colloidal silica suspension for polishing and reaching a mirror surface.For the TS specimen, a cross-section parallel to the strip casting direction was cut and ground, whereas, for the DC ingot, metallography samples were cut from the core (center), near-surface, and quarter-thickness regions within the ingot cross-section.For STEM sample preparation, manual mechanical grinding to <100 μm was performed, and then 3 mm diameter discs were extracted and smoothly polished by a twin-jet stream polisher using a mixture of methanol and nitric acid (with a 3:1 volume ratio) at 20 °C to create a uniform transparent sample.Constituent phase fractions were analyzed using ImageJ Java-based developed image processing software. [86]The processing and quantification were performed on SEM images with a resolution of 1536 Â 1103 pixels.In order to investigate the oxidation resistance, samples of TS and DC castings were subjected to exposure to atmospheric oxygen at the temperature range of 150-600 °C in 50 °C increments for 72 h.No specific treatments were applied to the sample surfaces, preserving them in their as-manufactured condition, and there was no regulation of the furnace atmosphere.Another set of samples underwent heat treatment cycles at the same temperature range and duration, and their thermal stability was assessed by measuring Vickers hardness with a 5 kgf indentation load and 10 s dwell time.Before indentations, the sample surfaces were ground, and multiple readings were taken from areas unaffected by any surface oxidation.

Figure 2 .
Figure 2. OM images showing the size, morphology, and volume fraction of second-phase particles obtained from different locations of cast DC ingot cross section: a) near-surface (edge) region, b) quarter-thickness region, and c) core (near center) region.(Note that chunks of the intermetallic Al-Mn-Fe phase appeared as gray and the β-phase having Chinese-script morphology appeared as black).

Figure 3 .
Figure 3. a,b) OM and c,d) SEM images obtained from the R3 region in the as-cast TS sample showing the emergence of intercellular/interdendritic centerline shrinkage porosity having irregular and isolated-branched appearance.

Figure 4 .
Figure 4. Spark-OES chemical analysis results exhibiting the effect of cooling rate on solute partitioning and macrosegregation throughout the thickness of the TS sample: a) schematic of stepwise coupon milling and test locations and b-f ) variations in solute concentrations from the centerline toward the strip surfaces.

Figure 5 .
Figure 5.The solidification velocity-dependent Al-Mg phase diagram in the Al-rich region.
in Al side of Al-Mg equilibrium phase diagram (M L ) À5.66 K wt% À1 8 Solidus line slope in Al side of Al-Mg equilibrium phase diagram (M s ) Mg in the solid/liquid interface (D i ) 2.5 Â 10 À8 m 2 s À1 12 Diffusivity of Mg in liquid Al (D L ) 5 Â 10 À8 m 2 s À1 13 Diffusivity of the solid at T E (D s ) 3.5 Â 10 À14 m 2 s À1 14 Gibbs-Thomson coefficient of Mg (Γ Mg ) 1 Â 10 À7 mK 15 Gibbs-Thomson coefficient of Al (Γ Al ) 1.96 Â 10 À7 mK 16 Constant for Al-Mg alloying system (

Figure 6 .
Figure 6.BSE-SEM images obtained from a) R1, b) R2, c) R3 regions of the TS sample, and d) the DC sample, displaying distinct morphologies and fractions of phases.EDX-SEM elemental maps acquired from the compounds formed in e) R2, f ) R3 regions of the TS sample, and g) the DC sample.

Figure 7 .
Figure 7. a) Collection of BF-TEM images captured from neighboring grains in the R1 region of the TS sample illustrating the formation of ultrafine intergranular and transgranular particles.HAADF-STEM images and elemental maps obtained from different regions showing the distribution of solutes: b) grain boundary Al-Si eutectic with flake morphology and faceted Al(Mn,Fe)Si particles, c) Al-Si eutectics decorated by Mn-Fe particles and d) short Al-Mg-Si needles and ultrafine parallel Al-Mn-Fe needles distributed with specific crystallographic orientation relationship (blue arrows) within the aluminum matrix.

Figure 8 .
Figure 8.The volume fraction change of the eutectic β-phase in AA5182 alloy with nominal Al-4.5 Mg wt% composition under the influence of altering thermal gradient and solidification velocity.The predictions are done utilizing a modified model developed by Sarreal and Abbaschian and the colored ribbons in the figure represent the range of measured eutectic β-phase fractions in the DC sample and different regions of the TS sample.

Figure 9 .
Figure 9. Captured surface images to compare the oxidation behavior of the TS and DC coupons after exposure to atmospheric oxygen for 72 h at different temperatures: a) 200 °C, b) 250 °C, c) 300 °C, d) 350 °C, e) 400 °C, f ) 450 °C, g) 500 °C, h) 550 °C, and i) 600 °C.

Figure 10 .
Figure10.a) Magnified surface appearance of the DC sample after exposure to atmospheric oxygen for 72 h at 600 °C exhibiting the formation of blisters and fine nodules as well as a dimensional change due to partial melting, (a 1 and a 2 ) polished cross section of the DC sample showing massive porous regions of oxidation products below blisters, and internal oxidation following pathways initiated from the surface and branched deeply inward implying the considerably large depth of material involved with oxidation (red dots show the surface blisters in (a) and the areas right below the blisters are shown in (a 1 ) and (a 2 )).b) Magnified surface appearance of the TS sample remains unchanged after being exposed to atmospheric oxygen at 600 °C for 72 h, with no signs of blisters, nodules, or any alterations in dimensions.c) OM image showing the as-cast microstructure of the DC sample prepared from midthickness of the cast ingot revealing the Al-Mn-Fe intermetallics (brownish-colored chunks) and β-phase eutectics (black-colored continuous phase).d) Microstructure evolution in the DC sample far from the surface after heat treatment for 72 h at 600 °C, revealing the coarsening of Al-Mn-Fe particles and spheroidization of β-phase.e) An OM image from the subsurface area in the TS sample heat treated for 72 h at 600 °C, showing no indication of oxidation products and inward oxidation pathways.f ) An OM image from the R2 region in the TS sample heat treated for 72 h at 600 °C, exhibiting coarsening of Al-Fe-Mn particles (brownish-colored particles), and due to the lower amount and finer size of the β-phase and its partial dissolution, spheroidization (black dots) is hardly discernible.

Figure 11 .
Figure 11.Thermal stability analysis of cast samples showing the a) average Vickers hardness values examined for TS and DC samples after different heat treatment cycles for 72 h, and b) normalized mean hardness values showing the hardness retention of alloys after treatment with respect to their as-cast nontreated state.

Table 1 .
Point EDX analysis results obtained from different phases/regions in TS and DC samples (the similar locations are color coded with the corresponding points).

Table 2 .
Input parameters applied for predicting the nonequilibrium β-Al 3 Mg 2 phase fraction in AA5182 alloy as a function of cooling rate.

Table 3 .
Measured values for area fractions by image analysis along with the conversion values obtained for the volume fraction of compounds in different regions/samples.