The pursuit of lightweight structural materials is a central theme in modern manufacturing, particularly for aerospace and automotive applications where weight reduction directly translates to improved fuel efficiency and performance. Among metallic materials, magnesium alloys stand out as the lightest structural option. Within this family, the Mg-Al-Zn system, encompassing widely used alloys like AZ91 and AM60, has been extensively researched and applied. While alloys such as AM60 offer good ductility and AZ91 provides high strength, there remains a significant demand for materials that can offer an optimal balance of both strength and toughness. Research has indicated that medium-aluminum, medium-zinc compositions, like Mg-6Al-4Zn (AZ64), can exhibit promising high-strength, high-ductility properties in permanent mold casting. This work, therefore, shifts focus to the sand castings process, a versatile and economically vital method particularly suited for producing complex, low-volume components such as those found in aerospace prototypes and specialized machinery. The study investigates a series of Mg-6Al-xZn (AZ6x) alloys to systematically understand the influence of zinc content on their solidification pathways, second-phase formation, and resultant grain structure under the relatively slow cooling conditions characteristic of sand castings.
The solidification microstructure of a casting is the primary determinant of its final mechanical properties. In sand castings, where heat extraction is slower compared to metal or die casting, the solidification sequence and the formation of intermetallic phases occur over longer timescales, potentially leading to coarser microstructures. A critical aspect of microstructural control is grain refinement. The concept of the Growth Restriction Factor (Q), or GRF, is pivotal here. It quantitatively describes the ability of a solute element to restrict the growth of dendrites ahead of the solidification front, thereby promoting grain refinement. For a binary alloy, Q is defined as:
$$ Q = m_L C_0 (k – 1) $$
where \( m_L \) is the liquidus slope, \( C_0 \) is the solute concentration, and \( k \) is the partition coefficient. A higher Q value signifies a greater constitutional undercooling zone, which can limit grain growth and promote nucleation events. However, for ternary or multicomponent systems like Mg-Al-Zn, the interactions between solutes make the direct application of this simple formula inaccurate. The true Q value must account for the coupled effects of all solutes and can be accurately derived from thermodynamic data using the relation related to the evolution of constitutional undercooling, \( \Delta T_{cs} \), with solid fraction, \( f_s \):
$$ Q_{true} = \left( \frac{\partial (\Delta T_{cs})}{\partial f_s} \right)_{f_s \rightarrow 0} $$
Furthermore, the point during solidification when a coherent dendritic network forms—the Dendrite Coherency Point (DCP)—is a critical transition. Before DCP, the semi-solid material behaves like a slurry; after DCP, it begins to develop mechanical strength as the solid dendrites interconnect. The solid fraction at DCP (\( f_s^{DCP} \)) is a key parameter influencing the development of casting defects such as hot tearing, macrosegregation, and shrinkage porosity. Understanding the relationship between alloy composition, Q, grain size (d), and \( f_s^{DCP} \) is therefore essential for optimizing the quality and properties of sand castings.
This investigation employs a combination of experimental characterization and thermodynamic modeling to unravel the solidification behavior of sand-cast AZ6x alloys. Experimental techniques include two-thermocouple thermal analysis for capturing real-time solidification events and determining \( f_s^{DCP} \), scanning electron microscopy (SEM) for microstructural observation, and electron backscatter diffraction (EBSD) for quantitative grain size measurement. Complementary thermodynamic calculations using the Pandat software and the PanMg database are performed to construct phase diagrams, simulate non-equilibrium (Scheil) solidification, and compute accurate Q values for these multicomponent alloys. The overarching goal is to establish clear linkages between composition, solidification parameters, and the final as-cast microstructure in sand castings.
Experimental Procedures and Thermodynamic Framework
The alloys for this study, with nominal compositions of Mg-6Al-xZn (x=0, 2, 4, 6 wt.%), were prepared from high-purity Mg (99.9%), Al (99.9%), Zn (99.9%), and an Al-10Mn master alloy. Melting was conducted in a mild steel crucible under a protective sulfur dioxide atmosphere to prevent excessive oxidation of the molten magnesium. No grain refiners or modifiers were added to isolate the effect of Zn content. Upon reaching a temperature of 735°C, the melt was poured into cylindrical resin-bonded sand molds. The mold cavity had an internal diameter of 100 mm and a height of 120 mm, with a mold wall thickness of 30 mm, ensuring a cooling regime representative of typical sand castings. The actual chemical compositions of the castings were verified using inductively coupled plasma atomic emission spectroscopy (ICP-AES).
To monitor the solidification process, a two-thermocouple thermal analysis setup was employed. Two K-type sheathed thermocouples (3 mm diameter) were positioned within the mold cavity: one at the geometric center and one near the mold wall, both at a height of 50 mm from the bottom. Temperature data was recorded at a high frequency (20 Hz). The Dendrite Coherency Point (DCP) was identified as the moment when the temperature difference between the wall and center thermocouples reached its first maximum. Cooling curves from the center thermocouple were analyzed to determine characteristic reaction temperatures (onset of nucleation for different phases). The solid fraction evolution during solidification, \( f_s(T) \), was derived from the cooling curve using the Newtonian baseline method, where the baseline represents the hypothetical cooling path in the absence of latent heat release.
Microstructural characterization was performed on samples extracted from the region adjacent to the center thermocouple tip. SEM was used for observing the morphology and distribution of second phases. For accurate grain size measurement, EBSD analysis was conducted. Samples were prepared by standard grinding and electropolishing. The average grain size was determined using the linear intercept method on EBSD orientation maps, with second phases at grain boundaries accounted for in the measurement.
Thermodynamic calculations served as a powerful complement to the experiments. The Pandat software with the PanMg database was utilized for three primary purposes:
- Equilibrium Phase Diagram Calculation: Vertical sections of the Mg-Al-Zn-Mn quaternary system were calculated to understand stable phase fields and equilibrium solidification paths.
- Non-Equilibrium Solidification Simulation: The Scheil-Gulliver model was applied to simulate solidification under negligible diffusion in the solid and complete mixing in the liquid. This model effectively predicts the formation of non-equilibrium eutectics, which are common in sand castings despite their slower cooling.
- Growth Restriction Factor (Q) Calculation: The accurate \( Q_{true} \) values for the multicomponent alloys were computed directly from the thermodynamic database using the methodology that evaluates the derivative of constitutional undercooling at the very beginning of solidification.
Solidification Sequence and Second-Phase Formation in Sand Castings
The analysis of cooling curves and their first derivatives, combined with SEM microscopy, revealed a distinct evolution of the solidification sequence with increasing Zn content in these sand castings. The key thermal events and corresponding microstructural constituents are summarized below.
| Alloy | Thermal Analysis Peaks | Identified Phase Reactions (Non-Equilibrium) | Major Phases in As-Cast Microstructure |
|---|---|---|---|
| AZ60 | Peak A (α-Mg), Peak B (Eutectic) | L → α-Mg; L → α-Mg + γ-Mg17Al12 | α-Mg, γ-Mg17Al12 (eutectic), Al-Mn phases |
| AZ62 | Peak A (α-Mg), Peak B (Eutectic) | L → α-Mg; L → α-Mg + γ-Mg17Al12; (Minor Φ-phase) | α-Mg, blocky γ-Mg17(Al,Zn)12, minor Φ-Mg21(Al,Zn)17 (eutectic), Al-Mn phases |
| AZ64 | Peak A (α-Mg), Peak B, Peak C | L → α-Mg; L → α-Mg + γ-Mg17(Al,Zn)12; L + γ → α-Mg + Φ-Mg21(Al,Zn)17 | α-Mg, blocky γ-phase, substantial Φ-phase (eutectic), Al-Mn phases |
| AZ66 | Peak A (α-Mg), Peak B, Peak C | L → α-Mg; L → α-Mg + γ-Mg17(Al,Zn)12; L + γ → α-Mg + Φ-Mg21(Al,Zn)17 | α-Mg, blocky γ-phase surrounded by continuous Φ-phase network, Al-Mn phases |
For the AZ60 alloy, solidification proceeded with the primary nucleation and growth of α-Mg dendrites, followed by a non-equilibrium eutectic reaction forming α-Mg and the γ-Mg17Al12 intermetallic phase. The γ-phase exhibited a classical lamellar eutectic morphology. Al-Mn intermetallics (predominantly Al8Mn5) formed at temperatures above the α-Mg liquidus but their thermal signature was not detectable due to their low volume fraction.
With the addition of 2 wt.% Zn (AZ62), the microstructure changed significantly. While the thermal analysis still showed only two major peaks, SEM revealed the presence of a small amount of a second intermetallic, the Φ-Mg21(Al,Zn)17 phase, alongside the dominant blocky γ-phase. The γ-phase morphology changed from lamellar to more continuous and blocky. At 4 wt.% Zn (AZ64), a third thermal peak (C) became clearly visible, corresponding to the formation of the Φ-phase via a peritectic-type reaction: L + γ → α-Mg + Φ. Microstructurally, the amount of Φ-phase increased, appearing with a fine eutectic-like structure, often surrounding the blocky γ-phase. This trend continued in the AZ66 alloy, where the Φ-phase became the dominant secondary phase, forming an extensive interconnected network that encapsulated the remaining γ-phase particles. This microstructural progression underscores how Zn content dramatically alters the non-equilibrium solidification path and phase selection in sand castings.
Thermodynamic equilibrium calculations provide crucial context for heat treatment potential. The calculated vertical section revealed that for AZ60 to AZ64 alloys, both the γ and Φ phases reside in a three-phase field (α-Mg + Al11Mn4 + Liquid) at temperatures just below the solidus. This implies that with appropriate solution heat treatment (typically between 350-420°C, depending on composition), these non-equilibrium second phases can be completely dissolved into the α-Mg matrix, enabling subsequent age hardening. For AZ64, this requires precise temperature control. In stark contrast, the AZ66 alloy’s equilibrium solidus is around 357°C, and the Φ-phase is stable at all temperatures below it. Therefore, the extensive Φ network in as-cast AZ66 sand castings cannot be dissolved without causing partial melting, severely limiting its strengthening potential through conventional T6 heat treatment.
Quantitative Analysis of Grain Size and Growth Restriction
The effect of Zn addition on the macroscopic grain structure was quantified using EBSD. The average grain sizes, measured via the linear intercept method, are presented below alongside the calculated true Growth Restriction Factor (Q) values for each alloy composition.
| Alloy Designation | Zn Content (wt.%) | Average Grain Size, d (μm) | Calculated Q Value | 1/Q |
|---|---|---|---|---|
| AZ60 | 0 | 557 ± 45 | 21 | 0.0476 |
| AZ62 | 2 | 275 ± 32 | 28 | 0.0357 |
| AZ64 | 4 | 271 ± 28 | 34 | 0.0294 |
| AZ66 | 6 | 235 ± 25 | 43 | 0.0233 |
The data reveals a clear trend: increasing Zn content leads to a significant increase in the Q value and a concomitant decrease in the average grain size. The initial addition of 2% Zn causes a dramatic grain refinement, halving the grain size from 557 μm to 275 μm. Further Zn additions continue to refine the grain structure, but the effect diminishes, with grain size decreasing only slightly to 271 μm and 235 μm for AZ64 and AZ66, respectively. This relationship can be visualized by plotting grain size (d) against the reciprocal of the Q value (1/Q). The classic grain refinement model often suggests a linear relationship of the form \( d = a + b / Q \), where ‘a’ and ‘b’ are constants related to the potency and density of nucleant particles. The data from these sand castings shows a deviation from a simple linear fit, indicating that while Q is a dominant factor, other mechanisms may become influential at higher solute levels. These could include changes in the dendrite tip growth kinetics, the potential for growth restriction saturation, or subtle variations in the effective nucleation undercooling as the solidification path and solute fields become more complex. The high Q value in AZ66 (43) confirms Zn’s potent role as a growth restrictor in the Mg-Al system during solidification of sand castings.
The Dendrite Coherency Point and Its Correlation with Microstructure
The solidification evolution was further analyzed to determine the Dendrite Coherency Point (DCP). The DCP temperature (\(T_{DCP}\)) was identified from the two-thermocouple thermal analysis data. Using the solid fraction curve \( f_s(T) \) derived from the cooling curve analysis (and validated against Scheil model predictions for the early stages of solidification), the solid fraction at coherency (\( f_s^{DCP} \)) was determined.
| Alloy | DCP Temperature, \(T_{DCP}\) (°C) | Solid Fraction at DCP from Thermal Analysis, \( f_s^{DCP-TA} \) (%) | Solid Fraction at DCP from Scheil Model, \( f_s^{DCP-Scheil} \) (%) |
|---|---|---|---|
| AZ60 | 610 | 35 | 36 |
| AZ62 | 604 | 27 | 27 |
| AZ64 | 593 | 26 | 31 |
| AZ66 | 587 | 25 | 23 |
The results show a consistent decrease in both \(T_{DCP}\) and \( f_s^{DCP} \) with increasing Zn content. The AZ60 alloy, with the lowest Q value, achieves coherency at a relatively high solid fraction of ~35%. This suggests that in the absence of strong growth restriction, dendrites grow rapidly not only in length but also in thickness, requiring a higher volume of solid to form a continuous network. The good agreement between the experimental and Scheil-predicted \( f_s^{DCP} \) for AZ60 and AZ62 validates the methodology for these compositions.
As Zn is added, the Q value increases. The constitutional undercooling zone ahead of the dendrite tips widens, which slows down their lateral growth (thickening) more effectively than their longitudinal growth. Consequently, the dendrites remain thinner and more branched. With a larger number of nucleation events (finer grain size) and thinner dendrite arms, the point at which these arms impinge and form a coherent network occurs at a lower overall solid fraction. This is clearly observed as \( f_s^{DCP} \) drops to around 25-27% for the AZ62, AZ64, and AZ66 sand castings. The Scheil model prediction for AZ64 shows a higher value, which may be attributed to the model’s assumptions about diffusion and the specific complexities of the ternary peritectic reaction becoming prominent around that solid fraction range. Nevertheless, the overall experimental trend is unambiguous.
This interplay can be conceptually summarized. Let \( G \) be the thermal gradient and \( v \) the growth velocity. The dendrite tip radius \( R \) and primary spacing \( \lambda_1 \) are governed by models that include the solute diffusion coefficient \( D_L \) and the strength of the solute field, which is related to Q. A higher Q generally leads to a smaller tip radius and potentially finer spacing. The coherency condition is met when the extended lengths of the primary arms, which are a function of the number of grains (N) and their growth geometry, first contact each other. If grains are more numerous (higher N, smaller d) and arms are thinner (affected by Q), impingement happens earlier in the solidification process, i.e., at a lower \( f_s \). Thus, we can posit a functional relationship:
$$ f_s^{DCP} \propto \frac{d \cdot \lambda_1}{F(Q, G, v)} $$
where \( F \) is a function describing the dendrite arm morphology. For these sand castings under similar slow cooling conditions (similar G and v), the decreasing trend of both d and \( f_s^{DCP} \) with increasing Q is consistent with this conceptual framework.
Implications for the Design and Processing of Sand Castings
The findings of this study have direct implications for the design and optimization of Mg-Al-Zn alloy components produced via sand castings. The following points summarize the key takeaways:
- Microstructure Control: Zn is a powerful microstructural modifier in Mg-6Al based sand castings. It not only refines the α-Mg grain size but also fundamentally alters the type, amount, and morphology of the strengthening intermetallic phases, shifting from a γ-dominated to a Φ-dominated microstructure as Zn increases from 0 to 6 wt.%.
- Heat Treatment Potential: Designers must carefully consider the Zn content if post-casting solution heat treatment is planned. Alloys like AZ60-AZ64 offer a window for dissolution of second phases and subsequent precipitation hardening. In contrast, the AZ66 composition, while offering the finest grain size and highest Q value, is largely non-heat-treatable due to the stable, insoluble Φ-phase network, limiting its peak strength potential but possibly offering good stability at elevated temperatures.
- Solidification and Defect Formation: The lower \( f_s^{DCP} \) in higher-Zn alloys means the coherent solid network forms earlier during solidification. This can have contrasting effects. On one hand, an earlier forming skeleton can help resist hot tearing during the vulnerable pasty zone. On the other hand, it may influence feeding characteristics and the formation of microporosity. Simulation of feeding and stress development in sand castings of these alloys should account for this composition-dependent shift in mechanical coherence.
- Performance Prediction: The quantitative relationships established—between Zn content, Q value, grain size (d), and \( f_s^{DCP} \)—provide a dataset for calibrating multi-scale solidification models. This enables better prediction of the as-cast microstructure and properties for new Mg-Al-Zn-X compositions intended for sand castings.
Conclusion and Future Perspectives
This comprehensive investigation elucidated the solidification behavior and microstructural evolution of Mg-6Al-xZn (x=0, 2, 4, 6 wt.%) alloys under the slow cooling conditions of sand castings. Through integrated experimental thermal analysis, advanced microscopy, and thermodynamic modeling, several key conclusions were drawn:
- The solidification path follows a non-equilibrium sequence, with Zn addition promoting the formation of the Φ-Mg21(Al,Zn)17 phase at the expense of the γ-Mg17Al12 phase. The morphology evolves from a lamellar γ-eutectic in AZ60 to a microstructure where blocky γ-phase is increasingly surrounded by a continuous network of Φ-phase eutectic in higher-Zn alloys.
- Thermodynamic equilibrium calculations delineate the heat-treatability of these alloys. AZ60-AZ64 alloys can be solution treated to dissolve secondary phases, whereas the AZ66 alloy contains a stable Φ-phase that cannot be dissolved, defining its processing limitations.
- Zinc acts as a potent grain refiner in this system. The calculated true Growth Restriction Factor (Q) increases monotonically with Zn content (from 21 to 43), leading to a significant reduction in average grain size from 557 μm to 235 μm. The relationship between grain size d and 1/Q is not perfectly linear, suggesting the involvement of additional growth kinetic factors at higher solute levels.
- The Dendrite Coherency Point (DCP) is strongly influenced by composition. Higher Zn content (higher Q) results in a lower solid fraction at coherency (\( f_s^{DCP} \)), decreasing from ~35% in AZ60 to ~25% in AZ66. This is consistent with a microstructure comprising a larger number of finer grains with thinner dendrite arms, which impinge and form a continuous network earlier in the solidification process.
Future work stemming from this research could explore several promising avenues. Firstly, the mechanical properties (tensile, fatigue, creep) of these as-cast and heat-treated sand castings need to be correlated with the quantified microstructural parameters (grain size, phase volume fraction, connectivity of Φ-network). Secondly, investigating the effect of cooling rate variations within the sand castings process (e.g., using chills or different sand binders) on the Q-\( f_s^{DCP} \)-d relationships would be valuable. Thirdly, extending the thermodynamic-kinetic approach to model the explicit development of the Φ-phase network and its impact on feeding and hot tearing susceptibility would significantly enhance casting simulation accuracy. Finally, exploring the role of micro-alloying additions (e.g., Ca, Sr, RE) on modifying the Φ-phase morphology and distribution in these Zn-containing sand castings could open new pathways for microstructure and property optimization.