Effect of Mg 17 Al 12 Fraction on Mechanical Properties of Mg-9 % Al-1 % Zn Cast Alloy

In the current study it was observed that the offset yield point of Mg-9%Al-1%Zn alloy was strongly influenced by the connectivity of Mg17Al12. It was suggested that an increase in the fraction of Mg17Al12 from 8% to 11% could lead to the formation of a Mg17Al12 network which resulted in a higher offset yield point. In addition, it was observed that elongation to failure of the Mg-9%Al-1%Zn alloy strongly depended on the fraction of Mg17Al12. Moreover, the apparent toughness showed a strong inverse relation to the secondary dendrite arm spacing. This approach might be extended to forecast the behavior in other magnesium alloys forming a network of the Mg-Al phase.


Introduction
Cast Mg-9%Al-1%Zn (AZ91D) has a good combination of castability and mechanical strength with a fair corrosion resistance, which makes it attractive for use in automotive components [1].Significant progress has been seen in casting capability and understanding of the relation between the microstructure and mechanical properties of AZ91.These relations were established experimentally [2][3][4] and modeled for quantitative prediction [5][6][7].The microstructure of AZ91D consists of the primary α-Mg and a highly divorced eutectic consisting of α-Mg and Mg 17 Al 12 .The fraction and morphology of the intermetallic phase, Mg 17 Al 12 , is heavily influenced by the non-equilibrium nature of the solidification and thus also by the choice of casting method and its associated process conditions [8].It has been demonstrated that the morphology, distribution and volume fraction of Mg 17 Al 12 affect the mechanical strength of AZ91D [9].Moreover, it was suggested that a highly interconnected network of Mg 17 Al 12 may account for the increased creep strength of other Mg alloys [10].Nagasekhar et al. [11] studied the three-dimensional characteristics of Mg 17 Al 12 in High Pressure Die Cast (HPDC) AZ91 and showed that the connectivity of Mg 17 Al 12 intermetallic could be seen in the as-cast material.
It has also been implied that increasing the area fraction of Mg 17 Al 12 may lead to an increased microhardness and macrohardness of HPDC AZ91D alloy [3,12].Cao et al. [13], on the other hand, reported that there is no evidence that Mg 17 Al 12 affects the hardness of AZ91.In addition, there is no clear influence of Mg 17 Al 12 content on the yield point of AZ91D alloy.Cáceres et al. [3] implied that there is no meaningful Hall-Petch grain size effect on the yield stress of HPDC AZ91.Dini et al. found that the yield point displayed a strain rate dependence but this depended on the fraction of Mg 17 Al 12 [14].
The current knowledge of the mechanical behavior of AZ91D clearly indicates a gap in our understanding of (i) the effect of Mg 17 Al 12 connectivity and (ii) the effect of grain size as well as (iii) secondary dendrite arm spacing (SDAS) on mechanical properties.Hence, the main objective of the current study is to improve our understanding of the effect of the microstructure on the tensile behavior of AZ91D by casting a range of microstructures corresponding to the coarse microstructure found in slowly cooled, sand-cast components to the fine microstructures typical of the higher cooling rates found in HPDC-processed AZ91D.The focus of this study is to understand the relationship between, in particular, SDAS, grain size, Mg 17 Al 12 and yield stress.In addition, the relation of the elongation to fracture and the fraction of Mg 17 Al 12 was determined for the temperature range of 25 • C up to 190 • C.

Material and Methods
In this study, the design of experiments (DOE) and D-optimal method was used as a scientific approach for exploring the effect of variables.The DOE approach allows the study of the influence of individual factors and the mutual interactions between factors on desired properties within the chosen concentration working range.DOE provides a route to obtain the optimal formulations, and to get the needed information with the minimum number of experiments.The DesignExpert™ software (Stat-Ease, Minneapolis, MN, USA) was used to assist in the DOE and the statistical evaluation of collected data.The variable factors were Bridgeman furnace pull rate ranging from 0.3 mm/s to 6 mm/s (identical to the solidification speed), tensile properties from room temperature (RT) up to 190 • C and strain rates form 10 −4 to 10 −1 1/s.The established experimental plan obtained by DOE is shown in Table 1.Each particular test condition is identified by a run number (R (run number)) and the number of replicates is shown in brackets.For example, R1:(4) is identified as the run 1 and the replications of this run is 4 times.
Commercial grade AZ91D ingots were melted in a mild steel crucible, under dry air atmosphere with 0.5% SF 6 gas.Rods for remelting in a Bridgeman furnace, were fabricated by sucking the melt into 1000 mm long, 10 mm diameter mild steel tube, preheated to 250 • C.This method, ensured the fabrication of clean AZ91D rods for further processing.The fabricated rods were cut into lengths of 170 mm and re-melted in a Bridgman furnace under the protective atmosphere (0.5% SF 6 in dry air).These rods were held at 650 • C sufficiently long to ensure complete melting and subsequently resolidified unidirectionally in the Bridgmen furnace under constant solidification speed.In this study, the solidification rate was controlled by adjusting the pull rate of the furnace from 0.3 mm/s to 6 mm/s.The tensile properties were measured according to ASTM B557 [15] on cylindrical sub-sized tensile bar with a reduced section of 6 mm and a gauge length of 36 mm using a ZWICK-ROLL™ Z100 tensile testing machine (ZwickGmbH & Co. KG, Ulm, Germany) equipped with a Laser extensometer.Tensile tests were performed at room temperature (RT) up to 190 • C and at strain rates ranging from 10 −4 up to 10 −1 1/s.
Samples for microstructural characterization were prepared following standard procedures [16].Specimen cross-sections were etched at room temperature using 10 mL HF (48%) for 1-2 s to etch and color the Mg 17 Al 12 dark [17].According to ASTM E112 [18] when determining the particle size (diameter), two transversal lengths having perpendicular directions to each other were drawn on each particle in the micrograph, and the average of at least six lengths was reported as the diameter of the particle [19,20].Normally, at least 100 diameter measurements were made in each sample.The area fraction of phases was measured by using an Olympus Stream™ motion 1.8 image analysis system (Olympus, Tokyo, Japan), based on image contrast in five images per sample.The shape factor of particles was obtained from the 4πA/p 2 relation where A and p are the area and perimeter of particles respectively following ISO, 9276-6: 2008 (E) [21].The SDAS was determined using light-optical microscope (Olympus, Tokyo, Japan) by measuring more than 50 dendrite arms for each specimen (for sufficient accuracy).
To measure the actual grain size electron back scatter diffraction (EBSD) was employed by using the JEOL JSM-7001F field emission scanning electron microscope (JEOL, Tokyo, Japan).The scanning step size of 0.8 µm was employed to distinguish between secondary arms and primary grains.
Table 1.Experimental conditions, microstructural feature size and mechanical properties obtained from a variety of pulling rates between 0.3 mm/s and 6 mm/s.Each particular test condition is identified by a run number (R (run number)) and the number of replicates is shown in brackets.For example, R1:(4) is identified as run 1 and the replications of this run is four.

Results and Discussion
All quantitative data are collated in Table 1, together with the experimental conditions.Figure 1a,b shows typical microstructures obtained for the solidification rate of 6 mm/s and 0.3 mm/s, respectively.The observed equiaxed microstructures contained α-Mg (white phase) surrounded by Mg 17 Al 12 (dark gray phase).Depending on the subsequent cooling rate, the amount of back diffusion in the solid phase may vary, giving rise to different amounts of remaining liquid for the eutectic reaction.The expected outcome is that a low solidification rate gives a result closer to equilibrium.The cooling rate has an effect on the secondary dendrite arm spacing (SDAS), the grain size, and the fraction of Mg 17 Al 12 [22].The morphology of the divorced eutectic structure is dependent on the cooling conditions.It has been reported that an increased drawing rate in Mg-Al alloys promotes the formation of a more divorced eutectic [13,23,24].It was observed that increasing the solidification rate led to a reduction in SDAS (see Table 1).

Results and Discussion
All quantitative data are collated in Table 1, together with the experimental conditions.Figure 1a,b shows typical microstructures obtained for the solidification rate of 6 mm/s and 0.3 mm/s, respectively.The observed equiaxed microstructures contained α-Mg (white phase) surrounded by Mg17Al12 (dark gray phase).Depending on the subsequent cooling rate, the amount of back diffusion in the solid phase may vary, giving rise to different amounts of remaining liquid for the eutectic reaction.The expected outcome is that a low solidification rate gives a result closer to equilibrium.The cooling rate has an effect on the secondary dendrite arm spacing (SDAS), the grain size, and the fraction of Mg17Al12 [22].The morphology of the divorced eutectic structure is dependent on the cooling conditions.It has been reported that an increased drawing rate in Mg-Al alloys promotes the formation of a more divorced eutectic [13,23,24].It was observed that increasing the solidification rate led to a reduction in SDAS (see Table 1).1c,d.The maps illustrate a lack of texture, and the assumption of isotropic behavior in tensile testing could thus be made.Similar to SDAS, the higher solidification rate leads to finer grain sizes (see Table 1).
Mg17Al12 precipitated in a different morphology in the inter-dendritic regions.The size of the Mg17Al12 particles was measured.For the high solidification rate (6 mm/s), the microstructural scale for Mg17Al12 ranged from 0.8 µ m to 16.5 µ m.For the low solidification rate (0.3 mm/s), this range was from 1.3 µ m to 24.2 µ m.The shape factor (a value of one corresponds to a perfect sphere) of Mg17Al12 for different solidification speeds appeared to be independent of the solidification rate (0.60 ± 0.2), but the area fraction showed a significant variation.The wide range of the shape factor suggests that Mg17Al12 has a strongly elongated shape and is present at the intergranular regions.It was interesting to note that an increased solidification rate resulted in a decreased fraction of Mg17Al12 (see Table 1).
Figure 2a,b shows typical true stress-strain curves for different SDASs and fractions of Mg17Al12 at room temperature and 190 °C, respectively.The chosen curves illustrate the most critical phenomena observed.The offset yield strength (Rp0.2) appeared to increase with the increased fraction of Mg17Al12.At a same strain rate, the sample with a high offset yield point is the one with a high amount of intermetallic (Figure 2a,b).It is worth mentioning that the influence of the strain rate on  1c,d.The maps illustrate a lack of texture, and the assumption of isotropic behavior in tensile testing could thus be made.Similar to SDAS, the higher solidification rate leads to finer grain sizes (see Table 1).
Mg 17 Al 12 precipitated in a different morphology in the inter-dendritic regions.The size of the Mg 17 Al 12 particles was measured.For the high solidification rate (6 mm/s), the microstructural scale for Mg 17 Al 12 ranged from 0.8 µm to 16.5 µm.For the low solidification rate (0.3 mm/s), this range was from 1.3 µm to 24.2 µm.The shape factor (a value of one corresponds to a perfect sphere) of Mg 17 Al 12 for different solidification speeds appeared to be independent of the solidification rate (0.60 ± 0.2), but the area fraction showed a significant variation.The wide range of the shape factor suggests that Mg 17 Al 12 has a strongly elongated shape and is present at the intergranular regions.It was interesting to note that an increased solidification rate resulted in a decreased fraction of Mg 17 Al 12 (see Table 1).
Figure 2a,b shows typical true stress-strain curves for different SDASs and fractions of Mg 17 Al 12 at room temperature and 190 • C, respectively.The chosen curves illustrate the most critical phenomena observed.The offset yield strength (R p0.2 ) appeared to increase with the increased fraction of Mg 17 Al 12 .At a same strain rate, the sample with a high offset yield point is the one with a high amount of intermetallic (Figure 2a,b).It is worth mentioning that the influence of the strain rate on tensile properties was already studied and the results suggested that there is an interaction between the fraction of Mg 17 Al 12 and the strain rate [14].At a high strain rate, the effect of the Mg 17 Al 12 fraction on yield strength is dominant, while no significant effect was observed at a lower fraction of Mg 17 Al 12 .
Metals 2016, 6, 251 5 of 8 tensile properties was already studied and the results suggested that there is an interaction between the fraction of Mg17Al12 and the strain rate [14].At a high strain rate, the effect of the Mg17Al12 fraction on yield strength is dominant, while no significant effect was observed at a lower fraction of Mg17Al12.It is important to note that hardening appears higher for a lower fraction of Mg17Al12 and a smaller SDAS.This is in agreement with the findings of Cáceres et al. [12].Furthermore, a higher fraction of Mg17Al12 appears to reduce the elongation to failure which also was found by Zhang et al.
[25].The ultimate tensile strength was mainly determined by temperature.
The average value of Rp0.2 obtained for tensile tests at room temperature and 190 °C is plotted against the area fraction of Mg17Al12 in Figure 3a.It is seen that an increased fraction of Mg17Al12 caused an increase of the offset yield strength.Furthermore, Figure 3a shows that two different groups of behavior could be observed both at room temperature and at 190 °C.It appears as both a change in slope and as a jump in the actual level of the dependence of Rp0.2 with the fraction of Mg17Al12.This change of behavior appears as the fraction changes from approximately 9% up to 11%.The possibility of having a network of Mg17Al12 was analyzed earlier through the thermal expansion of AZ91D [26].Also, Nagasekhar et al. [9,11] showed the presence of a network of interconnected intermetallic through 3D reconstructed images.The change in the behavior of the offset yield point may, as such, be related to the formation of a more or less fully connected network that was achieved at approximately 11% of Mg17Al12 fraction.During a tensile test, initial plastic deformation can only occur when the Mg17Al12 network is deformed.Hence, the formation of an interconnected array of Mg17Al12 leads to increase of the offset yield strength.It is also suggested that below 9% of the Mg17Al12 fraction, this network did not form and as a consequence showed a lower offset yield strength.This effect was valid for both room temperature and at 190 °C.It is important to note that hardening appears higher for a lower fraction of Mg 17 Al 12 and a smaller SDAS.This is in agreement with the findings of Cáceres et [12].Furthermore, a higher fraction of Mg 17 Al 12 appears to reduce the elongation to failure which also was found by Zhang et al. [25].The ultimate tensile strength was mainly determined by temperature.
The average value of R p0.2 obtained for tensile tests at room temperature and 190 • C is plotted against the area fraction of Mg 17 Al 12 in Figure 3a.It is seen that an increased fraction of Mg 17 Al 12 caused an increase of the offset yield strength.Furthermore, Figure 3a shows that two different groups of behavior could be observed both at room temperature and at 190 • C. It appears as both a change in slope and as a jump in the actual level of the dependence of R p0.2 with the fraction of Mg 17 Al 12 .This change of behavior appears as the fraction changes from approximately 9% up to 11%.The possibility of having a network of Mg 17 Al 12 was analyzed earlier through the thermal expansion of AZ91D [26].Also, Nagasekhar et al. [9,11] showed the presence of a network of interconnected intermetallic through 3D reconstructed images.The change in the behavior of the offset yield point may, as such, be related to the formation of a more or less fully connected network that was achieved at approximately 11% of Mg 17 Al 12 fraction.During a tensile test, initial plastic deformation can only occur when the Mg 17 Al 12 network is deformed.Hence, the formation of an interconnected array of Mg 17 Al 12 leads to increase of the offset yield strength.It is also suggested that below 9% of the Mg 17 Al 12 fraction, this network did not form and as a consequence showed a lower offset yield strength.This effect was valid for both room temperature and at 190 • C. In Figure 3b, the fraction of Mg17Al12 is plotted versus grain size.In Figure 3b, different behaviors can be identified.One group shows a weak dependence with the primary grain size, with fractions ranging typically from 7% to 9%.The other group shows a strong dependence with the grain size, with typical fractions just above 11%.The intersection in slope between these two groups occurs at around 9% of the intermetallic fraction.The outlier showing a low fraction Mg17Al12 at a large grain size also has a low Rp0.2 (107 MPa), supporting the hypothesis of connectivity and its influence on material yielding.
Analyzing the elongation to failure, in Figure 3c, a clear reduction is seen with an increasing Mg17Al12 fraction.Above 10% Mg17Al12 the elongation to failure appears to be in the order of 4% with a fair scatter.It can thus be concluded that the yield strength and elongation to failure depended on the fraction of Mg17Al12.
The ultimate tensile strength did mainly depend on temperature and no statistically significant dependence on grain size, SDAS or Mg17Al12 fraction could be established.As mentioned above, a finer SDAS appeared to display a higher rate of hardening.In order to analyze this behavior, an In Figure 3b, the fraction of Mg 17 Al 12 is plotted versus grain size.In Figure 3b, two different behaviors can be identified.One group shows a weak dependence with the primary grain size, with fractions ranging typically from 7% to 9%.The other group shows a strong dependence with the grain size, with typical fractions just above 11%.The intersection in slope between these two groups occurs at around 9% of the intermetallic fraction.The outlier showing a low fraction Mg 17 Al 12 at a large grain size also has a low R p0.2 (107 MPa), supporting the hypothesis of connectivity and its influence on material yielding.
Analyzing the elongation to failure, in Figure 3c, a clear reduction is seen with an increasing Mg 17 Al 12 fraction.Above 10% Mg 17 Al 12 the elongation to failure appears to be in the order of 4% with a fair scatter.It can thus be concluded that the yield strength and elongation to failure depended on the fraction of Mg 17 Al 12 .
The ultimate tensile strength did mainly depend on temperature and no statistically significant dependence on grain size, SDAS or Mg 17 Al 12 fraction could be established.As mentioned above, a finer SDAS appeared to display a higher rate of hardening.In order to analyze this behavior, an apparent toughness index was calculated as R m × ε f and plotted against SDAS, in Figure 4.The apparent toughness showed an inverse dependence on SDAS.It should also be noted that the temperature did not shift the influence and all measured values fell on the same line.

Conclusions
An attempt to shed light on the influence of the non-equilibrium microstructure on mechanical properties of AZ91D was made through tensile testing of material produced through controlled solidification.The following conclusions can be drawn from the present study: 1.The offset yield strength (Rp0.2) of AZ91D at a temperature range from Room Temperature (RT) to 190 °C was strongly related to the fraction of Mg17Al12 with an abrupt increase occurring between 9% and 11%. 2. The fraction of Mg17Al12 showed a strong relation with the primary grain size.Larger primary grains showed higher fractions to be precipitated.At a precipitated fraction above 9%, the dependence on grain size became very strong.3. The ductility (elongation to failure (εf)) of AZ91D strongly depended on the fraction of Mg17Al12.
Ductility was reduced with an increased fraction of Mg17Al12 up to approximately 9%, after which a further ductility decrease was inconclusive.4. The hardening rate in the tensile tests appeared to be higher for smaller SDAS while no such relation could be seen for the ultimate tensile strength.5.The apparent toughness showed a strong inverse relation to the SDAS.This relationship was temperature-independent.

Conclusions
An attempt to shed light on the influence of the non-equilibrium microstructure on mechanical properties of AZ91D was made through tensile testing of material produced through controlled solidification.The following conclusions can be drawn from the present study: 1 The offset yield strength (R p0.2 ) of AZ91D at a temperature range from Room Temperature (RT) to 190 • C was strongly related to the fraction of Mg 17 Al 12 with an abrupt increase occurring between 9% and 11%.
The fraction of Mg 17 Al 12 showed a strong relation with the primary grain size.Larger primary grains showed higher fractions to be precipitated.At a precipitated fraction above 9%, the dependence on grain size became very strong.3 The ductility (elongation to failure (ε f )) of AZ91D strongly depended on the fraction of Mg 17 Al 12 .
Ductility was reduced with an increased fraction of Mg 17 Al 12 up to approximately 9%, after which a further ductility decrease was inconclusive.4 The hardening rate in the tensile tests appeared to be higher for smaller SDAS while no such relation could be seen for the ultimate tensile strength.5 The apparent toughness showed a strong inverse relation to the SDAS.This relationship was temperature-independent.

Figure 1 .
Figure 1.(a) Optical micrograph of samples drawn at 6 mm/s (SDAS of 4.2 ± 1.2 μm) and (b) samples drawn at 0.3 mm/s (SDAS of 23.9 ± 4.4 µ m); (c) Illustration of IPF maps for samples drawn at 6 mm/s (grain size of 93 µ m ± 4.4 µ m) and (d) sample drawn at 0.3 mm/s (grain size of 254 µ m ± 3.7 µ m).Nonindexed regions (white regions) around the grain boundaries represent the Mg17Al12 phases.Nonindexed reigns inside the grains indicate defects and/or deformed area.

Figure 2 .
Figure 2. Stress-strain curves for specimens deformed at (a) room temperature, and (b) 190 °C, and a strain rate of 0.0001 1/s.

Figure 2 .
Figure 2. Stress-strain curves for specimens deformed at (a) room temperature, and (b) 190 • C, and a strain rate of 0.0001 1/s.

Figure 3 .
Figure 3. Dependence of (a) Rp0.2 and (c) elongation to fracture with fraction of Mg17Al12 at room temperature and 190 °C and linear regression plot; (b) Fraction of Mg17Al12 versus grain size.

Figure 3 .
Figure 3. Dependence of (a) R p0.2 and (c) elongation to fracture with fraction of Mg 17 Al 12 at room temperature and 190 • C and linear regression plot; (b) Fraction of Mg 17 Al 12 versus grain size.
an inverse dependence on SDAS.It should also be noted that the temperature did not shift the influence and all measured values fell on the same line.

Figure 4 .
Figure 4. Apparent toughness as a function of SDAS at both room temperature and at 190 °C.

Figure 4 .
Figure 4. Apparent toughness as a function of SDAS at both room temperature and at 190 • C.