II. EXPERIMENTAL PROCEDURE

AZ31-extruded bars provided by Applied Magnesium (Denver, CO) were cut into small pieces and melted using a high-frequency induction furnace (NORAX, Norax Canada, Inc., Saint-Romauld, Canada). The addition of Sr to AZ31 was carried out using a Sr–10wt% Al master alloy (Timminco, Haley, ON). The alloy and the additive were melted in a graphite crucible and held at 700 °C for 15 min for complete dissolution. SF₆–CO₂ gas mixture was used to prevent oxidation. The molten alloy was cast into cylindrical preheated steel dies at 720 °C. The chemical compositions of the cast samples were determined by inductively coupled plasma from the middle part of the cylinders, and the data are summarized in Table I. Alloy compositions are given in wt%, hereafter, unless otherwise indicated.

TABLE I.

Chemical compositions of the alloys.

* Composition for “AZ31 + 1.5%Sr” are taken from target and not from inductively coupled plasma-chemical analysis. Bal. = Balance.

For investigating the effect of temperature on precipitates, some of the samples were heated from 100 to 400 °C for 2 h in a Blue M Lindberg inert-gas mechanical convection-oven (Lindberg/MPH, Riverside, MI) and subsequently air-cooled. Samples for microstructural observations and analysis were sectioned from cast cylinders (Lindberg/MPH, Riverside, MI) and, after proper grinding, polishing was performed using 1 μm diamond paste. Picric-acetic solution was used to etch and reveal the microstructure of the specimens. Field-emission gun–scanning electron microscopy (FEG–SEM, Hitachi S-4700, Hitachi High Technologies America, Inc., Pleasanton, CA) was used for investigating the precipitates, and energy dispersive spectroscopy (EDS) was used for investigating the phase compositions. Nano-size precipitates in the microstructures were observed using transmission electron microscopy (TEM, Philips CM200, FEI, Hillsboro, OR) operating at 200 kV. The thin foil specimens were prepared from as-cast samples by grinding and were further thinned by jet polishing using a solution of 5% perchloric acid, 35% butanol, and 60% ethanol.

For more precise analysis of the Al–Sr and Al–Mn precipitate morphology and composition and to remove the effect of Mg matrix on the EDS results, intermetallics were extracted by dissolving the surrounding Mg matrix into a solution of 15-mL acetic acid and 150-mL water. The remaining precipitates were flushed with water and ethanol several times and finally collected using a vacuum filtering device. The extracted particles were analyzed using x-ray diffraction (XRD), SEM, and EDS.

III. THERMODYNAMIC CALCULATIONS

A pseudo-binary section for the AZ31 alloy with up to 3% Sr (Fig. 1) was calculated using the FactSage software and the FTLite database. The compositions of alloy used for this calculation are exactly the same as given in the scale of Fig. 1 (except the level of Mn, which changes in different calculations). The ratios of Mg/Al/Zn/Mn are kept constant and the Sr content is varied. According to the binary Al–Sr phase diagram, two different stable Al–Sr intermetallics (Al₂Sr and Al₄Sr) are likely to form during solidification. In a Mg–Al–Sr system, the affinity of Sr to Al is so strong that it can suppress the formation of β-(Mg₁₇Al₁₂) phase at a specific concentration of Sr.Reference Sadeghi and Pekguleryuz¹³ Both the Mg–Al and Al–Sr precipitates have large solidification ranges, which make it possible to dissolve other elements and form different complex stochiometries.Reference Janz, Gröbner, Mirković, Medraj, Jun Zhu, Chang and Schmid-Fetzer^16–Reference Janz, Gröbner and Schmid-Fetzer¹⁷ As can be seen in the phase diagram, thermodynamic modeling shows that the precipitation of Al–Mn particles is also correlated to the level of Sr. Calculations predict that the decrease in β-Mg₁₇Al₁₂ and the increase in Al₄Sr with increasing Sr is affected by the Mn content of the alloy [Fig. 2(a)]. Because Mn forms intermetallics with Al, the increase in Mn leads to the suppression of the β phase at lower concentrations of Sr (the cross shifts to the left). With increasing Sr, more Al is bound to Al–Sr precipitates and the formation of Al–Mn precipitates with high Al/Mn ratio (Al₉₉Mn₂₃, MnAl_4, and Mn₄Al₁₁) is suppressed in sequence. Figure 2(b) demonstrates the precipitation sequence of Al–Mn phases with increasing Sr, predicted by thermodynamic calculations. Nonetheless, the Al₈Mn₅ precipitate with the lowest Mn/Al ratio stays in equilibrium with the Al–Sr phases even at high levels of Sr.

FIG. 1.

Mg–Al–Zn–Mn–Sr phase diagram.

FIG. 2.

Effect of increasing level of Sr on the phase amounts (in wt%) of (a) Al₄Sr and Mg₁₇Al₁₂ and (b) Al–Mn precipitates calculated at 25 °C and equilibrium conditions.

On the other hand, it is known that thermodynamic equilibrium conditions do not hold during castingReference Jing, Yangshan, Shan, Feng and Tianbai¹⁰; therefore, a kinetic model of fast solidification under Scheil cooling condition has been used to calculate the amount of precipitation in a nonequilibrium condition. Scheil cooling assumes a local equilibrium of the advancing solidification front at the solid–liquid interface. In this approach, there is limited solute back-diffusion into the solid and the solute is rejected into the liquid. Complete mixing of solute in the liquid as a result of convection and/or stirring is also assumed.

The relative amounts of phases calculated under fast solidification conditions in the Scheil mode are presented in Fig. 3. As shown by thermodynamic calculations, at all levels of Sr, out of the four equilibrium Al–Mn precipitates, only the Al₈Mn₅ phase is expected to precipitate. The addition of 0.05wt% Sr [Fig. 3(a)] leads to the formation of Al₄Sr, but the precipitation of the β (Mg₁₇Al₁₂) and φ (Mg₆(AlZn)₅) is not suppressed. At 0.4 wt% Sr [Fig. 3(b)], the β precipitate ceases to form and the MgZn phase precipitates instead of φ. Finally, at 0.8% Sr, no Mg-containing precipitates exist [Fig. 3(c)]. Although thermodynamic simulation predicts the precipitation of two different Al–Sr precipitates (Al₄Sr and Al₂Sr), the microstructures and cooling curve analysis of the alloys show only one precipitate during solidification.Reference Sadeghi and Pekguleryuz¹³ This is attributed to the capture of Sr atoms during solidification in the primary α-Mg dendrite interfaces and the possibility of its back diffusion in solid, which are not considered in equilibrium or even in Scheil conditions. Therefore, the concentration of Sr in the liquid phase during solidification is less in reality compared to that in calculations. As a result, the precipitate with the higher Sr/Al ratio (Al₂Sr) is not likely to form during solidification or, if it does, its amount and size are not sufficient for microstructural observation.

FIG. 3.

Phase amounts (atomic fractions) of each phase under Scheil solidification condition versus temperature: (a) AZ31 + 0.05% Sr (Mg−2.99Al−0.99Zn−0.39Mn−0.05Sr), (b) AZ31 + 0.4% Sr (Mg−2.98Al−0.99Zn−0.39Mn−0.40Sr), and (c) AZ31 + 0.8% Sr (Mg−2.97Al−0.99Zn−0.36Mn−0.80Sr).

IV. RESULTS AND DISCUSSION

The β-Mg₁₇Al₁₂ (determined via SEM/EDS) precipitate can be distinguished by its globular morphology in Figs. 4(a)–4(d) [A in Fig. 4(a)]. It is noted that small amounts of Sr (0.01–0.5%) seem to refine the β-Mg₁₇Al₁₂ particles and the grain size (from 174 μm for AZ31 to 115μm for AZ31 + 0.5% Sr). EDS mapping revealed the presence of dissolved Sr inside the β phase.Reference Sadeghi and Pekguleryuz¹³ The effect of Sr on the refinement of β precipitatesReference Koubichek¹ and the grain size of Mg–Al–Zn alloys has been previously reported by other researchersReference Zhao, Wang, Zhai and Zhu^12, Reference Guan, Zhu, Wang, Yang and Cao^14, Reference Jing, Yangshan, Feng, Shan, Jinga and Weijianb^15, Reference Gruzleski and Aliravci¹⁸ and also in our previous work.Reference Sadeghi and Pekguleryuz¹³

FIG. 4.

Typical as-cast microstructures of (a) AZ31, (b) AZ31 + 0.01% Sr, (c) AZ31 + 0.1% Sr, (d) AZ31 + 0.5% Sr, (e) AZ31 + 1% Sr, and (f) AZ31 + 1.5% Sr, white arrows point to (A) Mg₁₇Al₁₂, (B) Al–Mn, and (C) Al–Mg–Sr precipitates.

In alloys with higher amounts of Sr [Figs. 4(e) and 4(f)], an interdendritic/grain-boundary phase of a different morphology is observed [C in Fig. 4(e)]. Al–Sr eutectic precipitates are seen in the grain boundaries but no globular β phase is seen inside the grains. The sample with the highest level of Sr (AZ31 + 1.5% Sr) reveals a higher amount of Al–Sr precipitates along the grain boundaries and inside the grains. In addition to the Al–Sr particles, Al–Mn phases are also present inside the microstructure. Al₈Mn₅ (with dissolved Si) and AlMn (a metastable precipitate with the lowest Al/Mn ratio) are among the particles that could be distinguished using optical microscopy (OM) and SEM inside the grains by their needle-shaped morphology [marked as B in Fig. 4(c)].

Figures 5(a) and 5(b) reveal the three-dimensional morphologies of extracted Al–Mn and Al–Sr precipitates after matrix-dissolution. EDS analysis on the extracted precipitates and TEM–EDS analysis on thin foils showed the composition of the precipitates to be more complex than simple Al₄Sr stochiometry, containing Zn and Mg. XRD analysis on the extracted precipitates reveals the characteristic peaks of Al₄Sr but with small deviation (Fig. 6), which suggests that the Al–Sr–Zn–Mg phase is actually Al₄Sr with small amounts of dissolved Mg and Zn, which has slightly distorted lattice parameters. However, it is also known that the Al₄Sr precipitate dissolves other elements in higher order systems (quaternary or higher) and assumes more complex stochiometry,Reference Baril, Labelle and Pekguleryuz^8–Reference Sadeghi and Pekguleryuz^13, Reference Janz, Gröbner, Mirković, Medraj, Jun Zhu, Chang and Schmid-Fetzer^16, Reference Janz, Gröbner and Schmid-Fetzer^17, Reference L’Esperance, Plamondon, Kunst and Fischersworring-Bunk¹⁹ potentially losing its thermal stability. A recent study reported that in an Mg–Al–Sr alloy the chemical composition of the complex phase (Mg₉Al₃Sr) changes by moving toward an equilibrium condition. During annealing, the Al/Sr ratio increases in the Mg–Al–Sr ternary phase, whereas its Mg content decreases. Consequently, the ternary intermetallic decomposes to α-Mg and Al₄Sr.Reference L’Esperance, Plamondon, Kunst and Fischersworring-Bunk¹⁹ It has also been observed by Liu et al.Reference Liu, Li, Wang, Su and Han²⁰ that the Al/Sr ratio increases during annealing and the complex stochiometry moves toward equilibrium Al₄Sr.

FIG. 5.

Scanning electron microscope image of extracted (a) Al–Sr and (b) Al–Mn precipitates from as-cast AZ31 + 0.75% Sr.

FIG. 6.

X-ray diffraction spectrum of extracted precipitates from as-cast AZ31 + 0.75% Sr.

As explained previously, annealing has an effect on phase composition: the Al–Sr precipitate increases its Al content as it moves toward the equilibrium Al₄Sr. The AlMn precipitate is also a nonequilibrium precipitate and is not stable when it is subjected to heat. Experiments show that as the annealing temperature increases, the Mn/Al ratio of the AlMn precipitate drops, whereas this ratio is almost constant for the equilibrium Al₈Mn₅ precipitate (Fig. 7). The excess Mn dissolves into the primary Mg grains. If the concentration of Mn exceeds the solid solubility, it is likely to bind with the dissolved Al in the primary α-Mg grains and form Al–Mn phases (probably Al₈Mn₅) nano-sized precipitates [Fig. 8(a)]. The precipitation of Al–Mn phase in Mg alloys at elevated temperatures has been reported by other researchers.Reference Kunst, Fischersworring-Bunka, L’Esperanceb, Plamondonb and Glatzelc⁷ Moreover, as a result of exposure to heat, the large Al₈Mn₅ precipitates start to reject their dissolved Si and their Si/Mn ratio drops (Si/Mn ratio in Fig. 7). TEM shows the presence of Al–Mn precipitates inside the grains of the as-cast and annealed samples [Fig. 8(b)].

FIG. 7.

Change of Al/Mn and Si/Mn ratio of Al₈Mn₅ and AlMn precipitates of AZ31 + 0.35% Sr by annealing at different temperatures.

FIG. 8.

Nano-sized (a) Al–Mn precipitate with Si (most likely Al₈Mn₅) and (b) Al–Mn precipitate without Si (most likely AlMn) of AZ31 + 0.75% Sr at 350 °C.

The addition of Sr also affects the solid solubility of other elements in α-Mg (Fig. 9). The values have been computed at 500 and 350 °C by the thermodynamic model considering the Scheil cooling conditions: 1 wt% increase in Sr results in 36% reduction in Al solid solubility and 27% in Zn solubility. The trend in Al has been confirmed by TEM–EDS analysis of the matrix between the precipitates, which shows ∼10% decrease in aluminum content from 2.8 at% in AZ31. However, a change in Zn concentration was not detected. It has been previously observed that Zn is surface active in Mg and segregates to grain boundaries,Reference Becerra and Pekguleryuz²¹ which could explain the EDS results. Evidence of formation of pure Zn dendrites has also been found in this work by precipitate extraction (Fig. 10). Depletion of Al inside the grains as a result of Sr in AJ series has been previously reported.Reference Baril, Labelle and Pekguleryuz⁸ In contrast, it is known that Al reduces the solid solubility of MnReference Avedesian and Baker²²; hence, with the increase in the level of Sr and the accompanying decrease in Al concentration, the solid solubility of Mn should increase, although the Mn content in solid solution is very low. The calculation predicts ∼23% increase in mol fraction (from 0.025 to 0.032 at 350 °C) when 1 wt% Sr is added to AZ31 (TEM verifies the increase from 0.01 to 0.03 at%). The level of dissolved Sr in Mg is increased as well, but the concentration is very low even at its maximum level.

FIG. 9.

Calculated change in interstitial concentration by increase in Sr level using thermodynamic computation.

FIG. 10.

(a) Pure Zn dendrites after precipitate extraction and (b) energy dispersive spectroscopic analysis of point A.

AZ31 alloys containing 0.01–1.5 wt% Sr can be divided into three groups on the basis of their microstructure. The first group contains trace levels of Sr (<0.1 wt%) and the main precipitate is β-Mg₁₇Al₁₂ similar to AZ31, but the grains and precipitates are refined. In these alloys, a large portion of the precipitates would be dissolved at high temperatures and provide good workability at elevated temperatures, while the precipitates would provide strength at lower temperatures. The other set of alloys have high Sr levels (>1 wt%) and contain a web of thermally stable precipitates at the grain boundaries and interdendritic regions. Although higher creep performance is expected (same as AJ series), severe deformation at high temperatures might result in brittleness and surface cracking. The last group of alloys fit in between the two groups and possesses a combination of their characteristics. The amount of thermally stable precipitates is not high enough to cause significant brittleness but they would enhance mechanical properties both at high and low temperatures. Moreover, as the precipitates do not dissolve at elevated temperatures, they could potentially act as nucleation sites for dynamic recrystallization and cause particle-stimulated nucleation during hot deformation, contributing to texture weakening.