The Role of Bismuth as Trace Element on the Solidification Path and Microstructure of Na‐Modified AlSi7Mg Alloys

The effects of Bi content as a trace element on the microstructure and the solidification path of A356.2 alloy have been investigated. The alloys containing different Bi levels (0, 20, and 200 ppm) have been modified by sodium. The experimental alloys have been thermally analyzed by using the two‐thermocouple method. Metallographic and image analysis techniques have been used to quantitatively examine the microstructural changes occurring at different Bi and Na concentrations. The results indicate how the presence of Bi as a trace element affects the eutectic structure. Upon increasing the Bi level, the nucleation and growth temperatures of eutectic Si raise, and the eutectic Si particles appear coarser in Na‐modified alloys. The EBSD analyses show that the crystallographic orientation between eutectic Al and surrounding primary Al dendrites becomes identical in Na‐modified Bi‐containing alloy. Furthermore, an irregular Bi + (Mg,Na)3Bi2 eutectic is formed prior to the precipitation of the eutectic Si, thus reducing the efficiency of Na addition to fully modify the eutectic Si.


Introduction
Aluminum-silicon alloys can be marked as primary foundry alloys, where bauxite ore is the most important raw material, but an alternative route is available using Al scrap and recycling. This is the secondary route for Al production that allows a reduction of about 95% of the required energy [1] and emits only 5% of the greenhouse gas. [2] The production of one ton of a so-called secondary Al alloy can save up to 14 000 kWh of energy and the average total exhaust emission is about 350 kg of CO 2 . [3] In general, secondary Al-Si alloys allow a good compromise between satisfying mechanical properties and reasonable material cost.
Despite a lower cost than primary alloys, various impurity elements coming from scrap or recycling processes can compromise the mechanical properties of the final product. The removal processes of the impurities are difficult and expensive. Dilution by mixing secondary Al alloys with commercial-purity aluminum seems the most practical, although uneconomic, solution for reducing the impurity level.
The main source of Bi contamination in the Al-Si alloys is associated with the Bicontaining free-cutting wrought Al alloys, such as EN AW-4104, 6012A, 6023, and others alloys, [4] which are progressively substituting the Pb-containing alloys. Over the years, attention has been paid to limiting and eventually eliminating lead and Pb-containing products because of their toxicity. Since 2008, Al alloys for machining purposes with a Pb content higher than 0.4 wt% have been forbidden by the European Union. [5] This has led to a steady increase of Bi content in secondary Al-Si alloys, as shown for instance in Figure 1 for secondary AlSi 9 Cu 3 (Fe) foundry alloys.
The positive effects of bismuth addition on the machinability of Al alloys are well known and some Bi-containing Al-Si casting alloys have been developed. [6] In general, Bi-rich phases having low melting temperatures provide self-lubrication and decrease the cutting force during machining operations. [7][8][9] Bismuth is also used as a chemical refiner for the eutectic silicon flakes in both hypoeutectic and hypereutectic Al-Si alloys resulting in higher tensile and wear resistance performances. [10][11][12][13][14][15] It is known how Bi can counteract, even as a trace element, the effect of further alloying additions or preliminary molten metal treatments affecting the final microstructure and tensile performance of the alloy. Some studies [16,17] stated how the presence of Bi in Sr-modified Al-Si foundry alloys increases the eutectic growth temperature, and a fibrous to the plate-like transition of eutectic Si occurs. During this transformation, the twin density in eutectic Si particles decreases due to Bi-Sr interactions, [18,19] which develop as pre-eutectic reactions leading to the precipitation of complex phases such as BiSr, Bi 2 Sr 3 , BiSr 3 , and Mg 2 Bi 2 Sr. This reduces the amount of available Sr for the eutectic Si modification. [16,[20][21][22] To mitigate the harmful effect of Bi in Sr-modified AlSi 7 Mg 0.4 alloys and to achieve a fully modified eutectic structure, Farahany et al. [23,24] suggested how the Sr/Bi mass ratio should be higher than 0.45.

DOI: 10.1002/adem.202201377
The effects of Bi content as a trace element on the microstructure and the solidification path of A356.2 alloy have been investigated. The alloys containing different Bi levels (0, 20, and 200 ppm) have been modified by sodium. The experimental alloys have been thermally analyzed by using the two-thermocouple method. Metallographic and image analysis techniques have been used to quantitatively examine the microstructural changes occurring at different Bi and Na concentrations. The results indicate how the presence of Bi as a trace element affects the eutectic structure. Upon increasing the Bi level, the nucleation and growth temperatures of eutectic Si raise, and the eutectic Si particles appear coarser in Na-modified alloys. The EBSD analyses show that the crystallographic orientation between eutectic Al and surrounding primary Al dendrites becomes identical in Na-modified Bi-containing alloy. Furthermore, an irregular Bi þ (Mg,Na) 3 Bi 2 eutectic is formed prior to the precipitation of the eutectic Si, thus reducing the efficiency of Na addition to fully modify the eutectic Si.
Sodium is the most effective chemical modifier in Al-Si alloys, and it is especially used for high Si-containing alloys or sandmolded castings. The Na addition facilitates multiple twinning reactions during eutectic Si growth and the formation of a high density of twins even at low amounts (<100 ppm). [25][26][27][28] Sodium modification is generally applied to improve mechanical properties, especially ductility, by changing the morphology of eutectic Si crystals from a coarse flake-like structure to fibrous morphology. [29,30] At Na levels greater than 180 ppm, some brittle intermetallic AlSiNa compounds precipitate during the solidification, and the coarsening of eutectic Si crystals is observed due to overmodification. [31][32][33][34] Even though separately added Na and Bi elements have been demonstrated to affect the solidification path and the microstructure of Al-Si alloys, studies on the effect of Na and Bi interactions in Al-Si alloys are limited. Furthermore, because the recycled Al-Si alloys may contain impurity elements such as Bi, a better understanding of the Na and Bi reactions seems to be necessary.
The aim of the present work is to study the influence of Bi content as an impurity element over Na-modification in an AlSi 7 Mg alloy. Attention has been paid to the possible Bi/Na interactions and changes in the solidification sequence, as well as their influence on the final microstructure of the alloy. Since the recycled Al alloys may contain an increasing level of Bi as an impurity, a better understanding of the Bi/Na interactions may be necessary during the preliminary melt treatment for the eutectic modification of the Al-Si-based alloys.

Materials and Processing
In the present study, a commercial-purity AlSi 7 Mg alloy (equivalent to the US designation A356.2) was used as a base alloy. The chemical composition of the alloy is listed in Table 1, and it was measured on separately poured samples by using a glow-discharge mass-spectrometer (GD-MS). This spectroscopy technique allows quantitative trace element analysis in the low-ppb range, as reported in several works. [35,36] The melt was prepared by charging about 3 kg of ingot pieces into a SiC crucible using an electrical furnace set at 730 AE 5°C. The melt was held in the furnace for 30 min to ensure the homogeneity of the bath and the surface was then skimmed to remove the dross. To investigate the interaction of Na modification with Bi impurity, the alloy was alloyed with different Na (nominally 0, 30, 55, and 75 ppm) and Bi (nominally 0, 20, and 200 ppm) levels.
To increase the Na content in the alloy, metallic sodium wrapped with aluminum foil was introduced into the molten bath; on the other side, weighted AlBi 9 master alloy in the form of waffle ingots was added, at the same time, to ensure that the Bi level reached the desired content. The Na and Bi combination in the different experimental alloys is listed in Table 2. To ensure the proper level of additives, the chemical composition of the melt was measured on samples separately poured at the beginning and the end of every set of castings.
The molten metal was then carefully poured into a BN-coated steel cup shown in Figure 2, which was preheated at 700 AE 5°C. After the addition of Bi and Na into the molten bath, the contact time between the Na/Bi additives and the melt was about 15 min, which was enough to completely dissolve the alloying elements into the bath and to avoid fading effects of Na.

Thermal Analysis
Computer-aided thermal analysis is a useful technique to monitor the solidification path of an alloy and to determine   the characteristic temperatures of the alloy during solidification. Figure 2 shows the schematic illustration of the thermal analysis setup, which was based on the two-thermocouple method as described by Bäckerud et al. [37] . After liquid pouring, the K-type thermocouples (Ø1 mm), which were fixed at the lid of the cup and covered with tight steel tubes, were inserted into the melt. The thermocouples were located along the central axis and adjacent to the wall of the cup at a depth of 30 mm from the bottom, as shown in Figure 2. The thermocouples were calibrated against the melting point of pure aluminum, assuming a melting point of 660.0°C, and used during the whole experimental campaign. The temperature and time were collected using a high-speed data acquisition system with a sampling rate of 0.1 s À1 , analogto-digital converter accuracy of 0.1°C, and connected to a personal computer. The thermal analysis setup was then allowed to cool in air. The cooling rate (the slope of the temperature-time cooling curve in the temperature range between the formation of the primary α-Al phase and the eutectic temperature) was constant and equal to 0.2 AE 0.1°C s À1 .
To determine the characteristic temperatures of the Al-Si eutectic reaction, the cooling curves and the corresponding derivative curves (dT/dt) referred to as the central thermocouple were plotted for each alloy. Phase reactions were realized by the change in the slope of the cooling curve which was easily identified by analyzing the first derivative of the curve. [38] The derivative of the cooling curve represents the rate of cooling of the solidifying metal. The precipitation of a new phase releases latent heat, which slows the rate of cooling, and the derivative curve increases.
Therefore, the characteristic temperatures related to the eutectic formation were collected. The eutectic nucleation temperature (T N,eu ), the minimum eutectic temperature (T min,eu ), the eutectic growth temperature (T G,eu ), and the recalescence undercooling of eutectic (ΔT R,eu ) were analyzed to estimate the eutectic modification efficiency and the possible Na/Bi interaction. [39,40] At least three thermal analysis runs were carried out for each experimental alloy, and it was observed that the variation of the characteristic temperatures was less than 1°C.

Microstructural Investigations
Samples for microstructural investigations were drawn close to the tip of the central thermocouple to correlate the microstructural characteristics with the results obtained from the thermal analyses. Samples were prepared to a 3 μm finish with diamond paste and, finally, polished with a commercial 0.04 μm fine silica slurry following standard metallographic techniques.
Microstructural investigations were performed using an optical microscope and a field-emission gun scanning electron microscope (FEG-SEM) equipped with an energy-dispersive spectrometer (EDS) and an electron backscattered diffraction (EBSD) unit, and quantitatively analyzed using an image analyzer.
The microstructural investigations were focused on the dimensional and morphological variations of the eutectic Si. Various microstructural parameters were investigated and measured, such as the particle average size and the roundness of eutectic Si crystals. Size is here defined as the equivalent circle diameter (d), while the roundness (r) was estimated according to the relation r = p 2 /(4πA), where p and A are the perimeter and the area of the particle, respectively.
A statistical average was obtained by analyzing at least 1000 eutectic Si particles from each sample. The secondary intermetallic phases, such as Mg 2 Si and Fe-bearing compounds, were excluded from the measurements.
Samples for EBSD investigations were sensitively prepared by mechanical polishing and then etched in a solution of 70% H 3 PO 4 , 25% H 2 SO 4, and 5% HNO 3 at 80°C for 30 s to improve the surface quality. The EBSD orientation mapping was performed in FEG-SEM operating at 20 kV on samples tilted 70°a t a working distance and step size of 25 mm and 0.5 μm, respectively. To process instantaneously the diffraction data as well as analyze the orientation imaging maps, EDAX Orientation Imaging Microscopy software was used.
Specimens for transmission electron microscope (TEM) investigations were mechanically prepared according to the standard metallography practice. When the thickness of the sample was reduced to %30 μm, the disc sample of TEM (Ø3 mm) was punched out. Ion-milling was performed for thinning of the disc sample up to the electron transparency. Investigations were carried out by TEM operated at 200 kV. Figure 3 shows the solidification path of the Al-Si eutectic reaction in the cooling curves of the experimental alloys containing different amounts of Na and Bi. It can also be observed how the cooling rate is constant in all the experimental alloys. The unmodified Bi-free alloy did not show undercooling in the solidification path of the eutectic (see Figure 3a). Upon increasing the www.advancedsciencenews.com www.aem-journal.com Na content, the eutectic reaction was progressively depressed showing recalescence undercooling too.

Thermal Analysis
In unmodified Bi-containing alloys, the eutectic reaction appeared without recalescence undercooling; further Na additions depressed the eutectic path, as shown in Figure 2b,c. In the alloys containing 20 ppm Bi, the cooling curves were effectively dropped by Na modification, as previously observed in the Bi-free alloy. Besides, the solidification path showed a recalescence undercooling after Na addition (Figure 2b). The additions of 55 or 75 ppm Na into the 200 ppm Bi-containing alloy depressed the eutectic solidification curves, while the eutectic reaction was slightly affected after the addition of 30 ppm Na. The recalescence undercooling increased with increasing Na level in the 200 ppm Bi-containing alloy. Table 3 summarizes the characteristic eutectic temperatures of the experimental alloys determined by thermal analysis. The eutectic nucleation temperature, T N,eu , of the base alloy (Na 0 Bi 0 ) was 577.1°C, while the minimum and growth eutectic temperatures were 575.0°C. In general, low undercooling and no recalescence are observed in commercial-purity Al-Si alloys such as the AlSi 7 Mg alloy used in the present work. Phosphorus, which is commonly present as an impurity in commercial-purity Al alloys, has been found to promote the nucleation of eutectic Si by forming pre-eutectic AlP compounds; these particles then act as  www.advancedsciencenews.com www.aem-journal.com a heterogeneous nucleation site for eutectic Si during alloy solidification. [35] The T G,eu gradually decreased by Na addition and it reached 567.7°C after 75 ppm Na addition in the Na 75 Bi 0 alloy; on the other side, the T N,eu was depressed to 572.2°C. The Na modification caused a remarkable depression of the growth eutectic temperature and an increase in the recalescence. The existence of a low and a high level (20 and 200 ppm, nominally) of Bi impurities is ineffective on the characteristic eutectic temperatures of the AlSi 7 Mg alloy (see Table 3). After Na modification, all the characteristic eutectic temperatures in the 20 ppm Bicontaining alloy gradually decreased and the eutectic plateau was depressed. Bi-free and 20 ppm Bi-containing alloys showed similar characteristic eutectic temperatures as the Na level increased.
In the 200 ppm Bi-containing alloys, T N,eu was equal to 577.2°C, and T G,eu decreased less than 1°C after 30 ppm Na modification. Among all, Na 75 Bi 200 alloy showed the lowest nucleation and growth temperatures, i.e. 574.8 and 569.8°C respectively.

Microstructure
The optical micrographs showing the eutectic structure of unmodified and 55 ppm Na-modified alloys with different Bi contents are displayed in Figure 4; while Figure 5 summarizes the average size and roundness values of the eutectic Si particles in the different experimental alloys. While the unmodified Bifree alloy showed coarse and plate-like eutectic Si particles, the addition of 55 ppm Na fully modified the eutectic structure and the silicon morphology changed from plate-like (Figure 3a) to a complete fibrous shape (Figure 3d). The size of the eutectic Si particles decreased from 8.9 to 1.3 μm by increasing the Na level up to 55 ppm; at the same time, the roundness decreased from 3.9 to 1.8, as shown in Figure 5. It is also observed how the scattering of size and roundness in Bi-free alloy decreases by increasing the Na addition until the level of 55 ppm, which indicates homogenized microstructure.   Furthermore, no modifications were observed in the Fe-rich and Mg 2 Si phases after Bi and Na additions.
The eutectic Si crystals were not affected by low (20 ppm) or high (220 ppm) Bi levels if compared to the base AlSi 7 Mg alloy (Figure 4b,c). The morphology remained the same; the size of the eutectic Si measured 9.3 and 9.1 μm in 20 and 200 ppm Bi-containing alloys, respectively. The addition of 55 ppm Na modified the alloy containing a low level of Bi impurity and produced fine and fibrous eutectic silicon, as shown in Figure 4e. Hereby, the size and roundness values of Na 55 Bi 20 alloy were almost similar to those measured in the modified base alloy (Na 55 Bi 0 ). On the other hand, the addition of 55 ppm Na into 200 ppm Bi-containing alloy partially modified the eutectic structure (Figure 4f ). A complete modification was provided by adding 75 ppm Na, which was higher than the required amount in Bi-free and 20 ppm Bi-containing alloys (Figure 6c). Therefore, the size and roundness values decreased to 2.3 μm and 1.9, respectively, in the Na 75 Bi 200 alloy. Furthermore, the scattering of size and roundness was greater for all the levels of Na addition when the Bi content increased up to 200 ppm.
At the highest Na level (75 ppm), traces of over-modification were observed by optical microscopy throughout the microstructure in the Bi-free and 20 ppm Bi-containing alloys, as shown in Figure 6a,b. In general, the density of the over-modification bands in the microstructure of Na 75 Bi 0 alloy was greater than Na 75 Bi 20 alloy. These bands consisting of coarse Si crystals in the eutectic structure are the main indicator of Na overmodification. [22,31] Due to over-modification, the size and roundness of eutectic Si slightly increased in Bi-free and 20 ppm Bi-containing alloys, as shown in Figure 5. As aforementioned, the addition of 75 ppm Na led to a full-modified microstructure in the 200 ppm Bi-containing alloy and no traces of over-modification were observed (Figure 6c). Therefore, to fully modify a 200 ppm Bi-containing AlSi 7 Mg alloy, about 35% more of Na content is necessary. This level is, however, feasible in the industrial reality and does not significantly compromise the standard foundry practice.

Intermetallic Bi-Rich Compounds
Intermetallic Bi-rich compounds were observed in the experimental Bi-containing alloys and characterized in detail (Figure 7). These particles were distributed mainly in the interdendritic regions and along grain boundaries. Due to the slow cooling rate during solidification (%0.2°C s À1 ) and very low Bi solubility in α-Al solid solution (0.006 wt% Bi at 271.4°C), these particles were in the size range from some μm to some tens of μm and showed mainly a blocky morphology. Upon increasing the Bi content, the size and area fraction of Bi-rich particles increased. In the alloy containing 20 ppm Bi (Na 0 Bi 20 ), the size and area fraction of the Bi-rich particles were 1.2 AE 0.7 μm and about 0.006%, respectively. When the Bi content reached 200 ppm in Na 0 Bi 200 alloy, the size and area friction of Bi-rich phases increased to 4.9 AE 2.4 μm and about 0.021%, respectively.
A backscattered electron FEG-SEM image of a bright Bi-bearing particle in an unmodified Bi-containing alloy (Na 0 Bi 200 ) and the corresponding EDS composition maps are shown in Figure 8. Among analyzed elements, only Mg and Bi were evident within the investigated particle. Figure 9 shows a backscattered electron FEG-SEM image of a bright Bi-bearing particle in 200 pmm Bi-containing alloy after 70 ppm Na modification (Na 70 Bi 200 ) as revealed by EDS composition maps. While Al and Si were not evident within the investigated particle, the presence of Bi and Na was clearly visible. Besides being distributed within the α-Al  www.advancedsciencenews.com www.aem-journal.com matrix, Mg appeared to be slightly concentrated in the bright particle.
Deeper TEM investigations of Bi-bearing compounds revealed a complex structure, in the form of an irregular eutectic lamellar structure (Figure 10a), consisting of Bi and (Mg,Na) 3 Bi 2 phases as revealed by the SAED ring pattern (Figure 10b). Because Mg and Na are interchangeable to each other, the increase of an element level caused a gradual decrease of the other one, while the corresponding phase stoichiometry was kept mostly constant and close to that of Mg 3 Bi 2 phase. This leads to a slight variation in the lattice parameter of the (Mg,Na) 3 Bi 2 phase, which is referred to the different atomic contents and radius of Mg and Na in the unit cell. The eutectic spacing between lamellae was almost measured to be 11 AE 2 nm.
Kurz and Fisher [41] stated how the geometrical arrangement of eutectic systems varies with volume fraction and entropy of fusion of the phases in the eutectic cell. In case one phase in the eutectic system has a volume fraction lower than 0.28, the phase tends to the formation of fibers. If the volume fraction of one of the phases is between 0.28 and 0.50, as in the present case, the phase becomes lamellar.
In contrast, the eutectic interface morphology can be affected by the entropy of fusion. When both phases in the eutectic show low entropy of fusion, the eutectic forms a regular structure due  If both phases are facetted in the eutectic cell, phases show smooth interface growth, and irregular lamellar structure forms branching or converging by one of the phases, especially when the solidification rate is low. [42] Whether a phase is facetted or non-facetted can be determined according to [41] α ¼ where α is the Jackson's factor, ΔS f is the entropy of fusion, L f is the latent heat of fusion, R is the general gas constant, and T is the melting temperature of the phase. If α is greater than 2, the phase is facetted. On the contrary, phase is non-facetted when the α is lower than 2. In the present work, α values of 2.50 and 6.19 were calculated for Bi and (Mg,Na) 3 Bi 2 phases, respectively, suggesting a facetted/facetted irregular structure.

EBSD Investigation
The crystallographic orientation relationship between the primary α-Al dendrites and the eutectic aluminum phase in the Bi-free alloy modified by 30 ppm Na (Na 30 Bi 0 ) and in the 200 ppm Bi-containing alloy modified by 30 ppm Na (Na 30 Bi 200 ) were analyzed using EBSD technique. The secondary electron images and the corresponding EBSD orientation maps for the eutectic regions are given in Figure 11. The color of the mapping pixels is the same when the difference in the crystallographic orientation of the surrounding areas is less than 2°. Furthermore, the misorientation angle graphs and the pole figure of the investigated alloys are shown in Figure 12. It needs to be mentioned that face-centered-cubic aluminum and diamondcubic silicon have a similarity in their diffraction patterns. Eutectic silicon is also indexed even though the data collection software of EBSD is set up for the indexing of aluminum diffraction patterns.
The eutectic structure of Na 30 Bi 0 alloy was fully modified, and fine-fibrous eutectic Si particles can be observed in Figure 11a. The corresponding EBSD orientation mapping in Figure 11b shows how the investigated eutectic region was surrounded by dendrite arms that belong to two different dendrites. The colors of the eutectic aluminum were completely different from the colors of the dendrite arms. This indicates eutectic aluminum has a wide variation of orientations different from the orientation of primary dendrites. Figure 11c shows the microstructure of Na 30 Bi 200 alloy where flake eutectic Si crystals are visible. In the corresponding EBSD orientation map (Figure 11d), a fraction (%28%) of eutectic aluminum showed the same crystallographic orientation with the surrounding primary α-Al dendrite arms. Besides, there were eutectic Al regions with different orientations which were caused by still active silicon modifier elements during the eutectic precipitation although the existence of 200 ppm Bi decreased the activity of Na in the melt.
The misorientation graphs of Na 30 Bi 0 and Na 30 Bi 200 alloys in Figure 13 show a slight difference in the crystal orientations of the Al phase generally scattering at higher angles. The Bi-free alloy (Na 30 Bi 0 ) shows a higher relative frequency of low-angle boundaries of Al grains with respect to the Bi-containing alloy (Na 30 Bi 200 ). The EBSD results seem to approach the values for complete random distribution, i.e., a MacKenzie distribution. The pole figure in Figure 12a indicates how the crystal orientation of the Al phase in Na 30 Bi 0 alloy tends to be scattered, while the pole figure of Na 30 Bi 200 alloy shows a concentrated crystal orientation of Al (Figure 12b).

Characteristic Eutectic Temperatures
Phosphorus always exists as an impurity element in commercially pure Al-Si alloys, and it precipitates as a pre-eutectic AlP compound which is the common nucleant for eutectic Si when the level of P is greater than %4 ppm in the alloy. The www.advancedsciencenews.com www.aem-journal.com precipitation temperature of the AlP phase increases with the increasing P content in the Al-Si-P system. [43] Sodium is commonly added to Al-Si cast alloys to modify the eutectic silicon; the characteristic eutectic temperatures are generally depressed after Na modification. [22,30] As observed in Figure 3 and Table 3, the Na modification depressed the eutectic nucleation and growth temperatures by poisoning mechanisms of AlP particles, which are potent nucleation sites for the eutectic Si in AlSi 7 Mg alloys. [44] The eutectic modification can be even observed in the experimental Bi-containing alloys with some  www.advancedsciencenews.com www.aem-journal.com differences with respect to the base AlSi 7 Mg alloy. To evaluate the Sr or Na modification level of a commercial Al-Si-based alloy, it is well known how the recalescence undercooling (ΔT R,eu ) is strongly correlated with the microstructural changes. In the Bi-free AlSi 7 Mg alloy, the ΔT R,eu value increased after the addition of 30 ppm Na, which partially modified the eutectic structure; then, it decreased when the microstructure was fully modified after 55 ppm Na addition. The further addition of Na increased ΔT R,eu when the alloy was over-modified. The same trend of the recalescence undercooling was observed in the AlSi 7 Mg alloy containing 20 ppm Bi impurity. On the other side, in the 200 ppm Bi-containing alloy, the partially modified eutectic structure was observed only after 55 ppm Na addition, when the ΔT R,eu was high. Further Na addition, which fully modified the eutectic, decreased the recalescence undercooling. It can be noted that the ΔT R,eu reaches a high value by increasing the level of Na before it decreases when the Na content is suitable for a full modification.

Al-Si Eutectic Nucleation
With regard to the binary Al-Bi phase diagram, previous studies [45,46] explained the precipitation of Bi-bearing phases during the solidification process. Al and Bi form a monotectic binary phase system with a monotectic point at 3.4 wt% Bi at 657°C. Thus, when the melt (L) is cooled in the hypermonotectic composition range, it decomposes into two immiscible melts, i.e., L Al-rich and L Bi-rich . Bi-rich melt droplets nucleate and grow by means of diffusion in Al-rich melt until the temperature of the system reaches 657°C, which is the monotectic temperature. After the monotectic reaction (L Al-rich ! α-Al þ L Bi-rich ), Bi-rich droplets tend to cluster and coarsen in the range between monotectic and eutectic temperatures. On the other hand, considering the Mg-Bi binary system, the Mg 3 Bi 2 phase can precipitate as β-Mg 3 Bi 2 or α-Mg 3 Bi 2 above or below 703°C, respectively. In case β-Mg 3 Bi 2 precipitates in magnesium-enriched Bi-rich melt droplets, it can be a stable intermetallic phase due to a higher melting temperature than the surrounding melt at 730°C, which is the casting temperature in the current work. The prediction of Scheil calculations through CALPHAD modeling also confirmed that Mg 3 Bi 2 can precipitate before Al-Bi eutectic temperature with the Bi segregation. The nucleation temperature of the Mg 3 Bi 2 phase is estimated at about 557°C in the multi-component AlSi 7.5 Mg 0.25 Bi 0.002 system ( Figure 13). When the amount of Bi is enhanced up to 200 and 7500 ppm, the onset temperature of the Mg 3 Bi 2 phase increases to about 570 and 578°C, respectively. On the other side, the Al-Si eutectic reaction is expected to start at 575°C in the Bi-free alloy and slightly decrease after the progressive addition of Bi (see Figure 13). Bismuth has no solubility in the α-Al phase at these temperatures. Therefore, where Bi is segregated, the Mg 3 Bi 2 compound can precipitate at higher temperatures.

Al-Si Eutectic Growth
The existence of Bi impurity is ineffective on the morphology and the size of eutectic silicon flakes in unmodified AlSi 7 Mg alloys. Even though the eutectic morphology can be fully modified by adding 55 ppm Na in 20 ppm Bi-containing alloy, greater amounts of Na need to be introduced in the molten bath when the level of Bi impurity increases. When no over-modification was observed in the Na 75 Bi 200 alloy, the eutectic structure of Na 75 Bi 0 and Na 75 Bi 20 alloys were over-modified (see Figure 6).
Both Na-modified and unmodified Bi-containing alloys exhibit Bi-bearing intermetallic phases. As shown in the binary Mg-Bi system, when the melt is cooled below 260°C, an eutectic structure is formed, i.e., Bi þ α-Mg 3 Bi 2 . This is consistent with the analyzed intermetallic in Figure 8 and the results presented in ref. [46]. Bi-and Na-bearing phase, as investigated by EDS mapping analyses in Figure 9 is shown as bright particle with a round shape close to eutectic silicon particles. EDS mapping analyses confirm the segregation of Mg, Bi, and Na. The SAED ring pattern in Figure 10b identifies the Bi-and Na-bearing bright phase as Bi þ Mg 3 Bi 2 eutectic phase. In the same manner, Na also segregates into the Bi-rich melt droplets in Na-modified Bi-containing alloy and it substitutes with Mg in L þ β-Mg 3 Bi 2 or L þ α-Mg 3 Bi 2 region forming (Mg,Na) 3 Bi 2 phase. The TEM image in Figure 10a demonstrates the faceted/faceted irregular Bi þ (Mg,Na) 3 Bi 2 eutectic structure. Thus, the amount of effective Na for eutectic modification in the melt contributes to the formation of Bi þ (Mg,Na) 3 Bi 2 prior to the precipitation of eutectic Si. Na element cannot neutralize the effect of AlP compounds. Therefore, a fibrous-to-flake transition of the eutectic Si particles develops when Bi is present as an impurity in Al-Si alloys.
Several studies [36,47,48] have demonstrated three different modes of nucleation and growth of the eutectic in Al-Si-based alloys: 1) nucleation of eutectic grains on nucleant particles in the vicinity of the primary α-Al dendrites; 2) heterogeneous nucleation of eutectic on nuclear particles in the interdendritic liquid; and 3) nucleation at or adjacent to the wall and front growth opposite the thermal gradient (see Figure 14).
The different mechanisms depend on the chemical composition, i.e., the modifier's type and the cooling rate during www.advancedsciencenews.com www.aem-journal.com solidification. In particular, while mode (i) is typical of unmodified commercial purity Al-Si alloys, the eutectic grains are nucleated in the intergranular liquid when Sr or Sb is added (mode ii). The last mechanism (mode iii) of eutectic solidification is typically observed in Na-modified alloys. [49] Here, the nucleation of eutectic grains begins close to the mold wall, and they grow opposite to the thermal gradient with a well-defined growth front on a macro scale, even though independent nucleation of eutectic grains in the interdendritic regions occurs on a micro-scale. These growth mechanisms of the eutectic are generally characterized by crystallographic orientation measurements with EBSD. When growth mechanisms (ii) and (iii) apply, a wide distribution of orientations of the eutectic Al phase is generally observed in EBSD orientation maps. On the other side, if the eutectic growth happens as in mode (i), the eutectic Al phase and the primary α-Al dendrites show similar crystallographic orientation in EBSD maps. Figure 11b shows the independently nucleated eutectic grains with representative different colors in the EBSD map, which indicate no systematic orientation relationship between primary Al dendrites and eutectic Al due to the heterogeneous nucleation of eutectic grains in the interdendritic liquid in Na modified Bi-free alloys. Here, the eutectic Si leads the growth front due to the lower constitutional undercooling that Si needs for nucleation and subsequent growth to proceed. [36] The crystallographic relation of Al in the eutectic is then determined by Si rather than by the surrounding Al dendrites. On the other side, in Na-modified alloys with Bi impurity, a fraction of crystallographic orientations between eutectic Al and surrounding primary Al dendrite arms are identical (see Figure 11d), indicating a change in the eutectic growth from mode (iii) to (i) due to Bi. In this situation, the eutectic Si nucleates first, and the growth then proceeds until the melt is depleted of Si and the eutectic Al starts to solidify by growing from the primary Al dendrites. [50]

Conclusions
The effect of the low and high levels of Bi impurities on the microstructure and Bi-bearing phases in unmodified and Na-modified A356.2 alloy have been investigated. Thermal analyses and extensively applied microstructure investigations enlightened the formation of Bi-bearing phases and counteraction between Bi and Na resulting in fibrous to flake transition. In conclusion, we found that: 1) Bi impurity is ineffective on the characteristic temperatures and the morphology of the eutectic Si in unmodified AlSi 7 Mg alloys; 2) Bi þ (Mg, Na) 3 Bi 2 eutectic structure is formed prior to the precipitation of eutectic Si in Na modified Bi-containing alloys, hence, the amount of effective Na for modification of eutectic Si decreases. More Na addition is required for a fully modified eutectic structure; and 3) The independent nucleation of eutectic Al in the Na-modified alloy is prevented by Bi and the number of misorientation is reduced.