- Email: [email protected]
Refinement of eutectic silicon phase of aluminum A356 alloy using high-intensity ultrasonic vibration
Scripta Materialia 54 (2006) 893–896 www.actamat-journals.com
Refinement of eutectic silicon phase of aluminum A356 alloy using high-intensity ultrasonic vibration X. Jian a
a,*
, T.T. Meek a, Q. Han
b
Materials Science and Engineering Department, University of Tennessee, Knoxville, TN 37996, USA b Oak Ridge National Laboratory, Oak Ridge, TN 37831-6083, USA Received 7 September 2005; received in revised form 24 October 2005; accepted 1 November 2005 Available online 29 November 2005
Abstract The eutectic silicon in A356 alloy can be refined and modified using either chemical, quench, or superheating modification. We observed, for the first time, that the eutectic silicon can also be significantly refined using high-intensity ultrasonic vibration. Rosette-like eutectic silicon is formed during solidification of specimen treated with high-intensity ultrasonic vibration. 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Eutectic solidification; Aluminum alloys; Microstructure; Grain refining; Ultrasonic vibration
1. Introduction Aluminum A356 alloy is one of the widely used casting aluminum alloys because of its good mechanical strength, ductility, hardness, fatigue strength, pressure tightness, fluidity, and machinability [1]. The A356 alloy contains about 50 vol.% eutectic phases. The final microstructure is largely determined by the eutectic reaction. Due to its diamond cubic crystal structure which predominantly grows in h1 1 2i directions on (1 1 1) planes, silicon is a faceted phase with strongly anisotropic growth thus it is difficult to change the growth direction [2]. In unmodified A356 alloy, the main eutectic reaction occurs at 574 C as a binary reaction, which results in coarse irregular plate-like silicon. The modification of eutectic silicon is of general interest since fine eutectic silicon along with fine primary aluminum grains improve mechanical properties and ductility [3]. Three well-known eutectic modification methods have been developed thus far, namely (1) chemical modification
*
Corresponding author. Tel.: +1 865 574 9956; fax: +1 865 574 4357. E-mail address: [email protected] (X. Jian).
[3–8], which produces fine fibrous silicon structure through the addition with trace-levels of several elements, such as sodium, antimony, potassium, calcium, strontium, and barium; (2) quench modification [2,9], which results in silicon forming an exceedingly fine form when compared to the unquenched silicon, through high cooling rates and rapid solidification in the growth rate ranging from 400 lm/s to 1000 lm/s; and (3) superheating modification, which requires the presence of magnesium in the alloy [10] to obtain fibrous silicon by heating up the melt from the usual pouring temperature (for example, 680 C) to a temperature of 850–900 C and then holding for about 15–30 min, then quickly cooling to the pouring temperature before casting. Chemical modification has been widely used in industry for producing A356 alloy with fine and fibrous eutectic silicon phase, and thus improved mechanical properties. In a recent project investigating the effect of high-intensity ultrasonic vibration on the microstructure of A356 alloy, we found that high-intensity ultrasonic vibration has a significant effect on refining both the primary aluminum phase and the eutectic silicon phase [11,12]. This article describes experimental results on modification of the eutectic silicon phase using ultrasonic vibration.
1359-6462/$ - see front matter 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2005.11.004
894
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
2. Experimental conditions Commercial aluminum A356 alloy (Al, 7.0 wt.% Si, 0.4 wt.% Mg, 0.1 wt.%, Fe) was used in this study. The liquidus temperature of the alloy is 614 C and the solidus temperature is 554 C. The primary aluminum dendrites start to form at 614 C and the binary Al–Si eutectic at 574 C. The tertiary eutectics and complex intermetallics form at a late stage of solidification [13]. An ultrasonic system was used to treat the liquid metal after it was poured into a Cu permanent mold. The details of the ultrasonic system can be found in Ref. [12]. It consisted of a 20 kHz ultrasonic generator, a transducer made of piezoelectric lead zirconate titanate crystals (PZT), an ultrasonic horn, and a radiator. The horn and the radiator were made of titanium Ti–6Al–4V alloy. The power output of the unit was variable within a maximum of 1500 W by adjusting the output acoustic amplitude from 24.3 lm to 81 lm, or 30% to 100% of the unitÕs upper limit. The ultrasonic radiator was placed at the bottom of a copper mold which held up to 250 g molten aluminum. The alloy was melted and held in a furnace for 30 min at 700 ± 5 C, about 86 C higher than its liquidus temperature, to allow the complete dissolution of silicon particles. The molten alloy was then poured into the mold at a pouring temperature of 630 C. Ultrasonic vibration with an amplitude of 56.7 lm was started right before the melt was poured into the copper mold. For comparison reasons, samples were also made without ultrasonic vibration. These untreated castings were hemispherical in shape with a diameter of 5 cm and weight of 200 g. The as-polished samples were characterized using an optical microscope with the capability for quantitative metallographic analysis. They were then lightly etched using a 0.5% HF–water solution and further examined using a Hitachi S-4700 scanning electron microscopy (SEM) with light element energy dispersive spectroscopy (EDS) X-ray detectors. Finally the samples were deep etched in a 0.5% HF–water solution for 2 h in order to reveal the three-dimensional morphology of the eutectic silicon phase. 3. Experimental results and discussion The optical micrographs (100·) of the samples from castings made with and without high-intensity ultrasonic vibration are shown in Fig. 1. The casting sample made without ultrasonic vibration exhibited coarse acicular eutectic silicon dispersed among the fully developed primary aluminum dendrites. The eutectic silicon was about 100 lm in length. In addition, one branch of a primary dendrite shown in the middle of Fig. 1(a) was about 800 lm in length, indicating that the grain size was in the range of a few millimeters since one equiaxed grain usually contained six primary dendrite arms. In contrast, the ultrasonically treated A356 alloy displayed a microstructure of continuous very fine Al–Si eutectic interspersed with fine
Fig. 1. Eutectic modification observed by optical metallurgical microscope at low magnification, (a) without ultrasonic treatment and (b) with ultrasonic treatment.
globular primary aluminum grains (Fig. 1(b)). The shape and size of individual eutectic silicon grains were not distinguishable at such a magnification. The formation of spherical primary aluminum phase has been reported elsewhere [12]. Both samples were characterized with SEM at a high magnification of 2000·, as shown in Fig. 2. The coarse acicular eutectic silicon observed on the untreated A356 alloy agreed with that of low magnification observation. While with ultrasonic treatment, very fine eutectic silicon was observed, which was finer than that in strontium-modified A356 alloy of a similar casting condition [14]. The EDS X-ray analysis verified that those particles shown in the eutectic areas were mainly elemental silicon, rather than intermetallic phases or inclusions. Results of quantitative metallographic analysis of silicon morphology are illustrated in Fig. 3. Without ultrasonic treatment, the average length of eutectic silicon was about 26 lm, and the average width 2.7 lm. The aspect ratio was slightly less than 10. While with ultrasonic treatment, the average length and width were respectively about 2 lm and 0.6 lm, with an aspect ratio of slightly less than 3. A comparison of the aspect ratio of the untreated (about
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
895
Fig. 4. Three-dimensional eutectic morphology observed using SEM at 2000· magnification, (a) without ultrasonic treatment and (b) with ultrasonic treatment. Fig. 2. Eutectic modification observed by SEM at 2000· magnification, (a) without ultrasonic vibration and (b) with ultrasonic vibration.
Fig. 3. Silicon morphological analyses of A356 alloy without and with ultrasonic treatment.
9.8) and ultrasonically treated (about 2.8) A356 alloy suggests that the silicon morphology in ultrasonically modified A356 alloy is not just an exceedingly fine form of the silicon in the unmodified alloy.
Fig. 4 shows the comparison of the three-dimensional morphology of eutectic silicon observed by SEM on the deep-etched samples. Fig. 4(a) shows the eutectic silicon phase in the sample not subject to ultrasonic treatment. The silicon phase exhibits a typical coarse plate-like form. Most likely these silicon plates grew epitaxially from the surrounding primary aluminum dendrites. This result is in accordance with previous reports for unmodified A356 alloys [15]. Fig. 4(b) shows the silicon morphology in a sample subject to ultrasonic vibration. There are two distinct features of the silicon phase in the ultrasonically processed sample. Firstly the silicon phase is rosette-like, rather than plate-like. The center of a rosette could be the center of an eutectic cell/grain, indicating the nucleation of the eutectic phases could be independent to the surrounding primary aluminum dendrites. Another feature is that the eutectic phase spacing is much smaller in the ultrasonically processed sample than that in the sample not subject to ultrasonic vibration. These two features suggest that the modified eutectic silicon using ultrasonic vibration is not simply a finer form of the unmodified A356 alloy. The mechanism by which the finer eutectic silicon phase forms under ultrasonic treatment is not clear at the moment. The eutectic spacing is much smaller in the ultrasonically processed specimen than that in the untreated
896
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
specimen. Small eutectic spacing usually results from high growth rates of these eutectic phases [16]. However, since ultrasonic energy about 600 W was injected into a specimen, the cooling rates in the ultrasonically processed specimen were in fact slower than the one not subject to ultrasonic vibration. On the other hand, the eutectic grains/cells as outlined in Fig. 4(b) were smaller than that in the specimen not subject to ultrasonic vibration. This is again not in accordance with previous experimental evidence that techniques that increase nucleation lead to the coarsening of the eutectics by decreasing their growth rates [17]. A number of unique phenomena that occur during ultrasonic processing of alloy may contribute to the refinement of the eutectic silicon phase. One phenomenon is acoustically induced convection in the liquid pools. This vigorous convection tends to dislodge AlP particles at the existing dendrite interfaces and disperse them uniformly into the remaining liquid [18,19]. Phosphorous in the form of AlP particles is often quoted as a good nucleant for silicon. The removal of these particles from the existing dendrite interfaces prevents the nucleation of silicon particles on the existing dendrites. Acoustically induced convection can also affect nucleation of the silicon phases by altering the constitutional supercooling at the front of the growing eutectic grains [20]. Another phenomenon is the ultrasonically induced pressure variations in the remaining liquid pools. Further research is needed to investigate the effect of these two phenomena on the nucleation and growth of the eutectics in order to fully understand the mechanisms of silicon modification under ultrasonic vibration. 4. Conclusions Ultrasonic vibration at 20 kHz was introduced into A356 alloy as it was cast into a metal mold at the temperature of 630 C. The results showed that: (1) The morphology of eutectic silicon was modified from a coarse acicular plate-like form when no ultrasonic vibration was used, to a finely dispersed rosettelike form when ultrasonic treatment was employed. (2) Ultrasonic treatment reduced the size of eutectic silicon from 26 lm to 2 lm in length, which is over
an order of magnitude, and the width from 2.7 lm to 0.6 lm. The aspect ratio was also reduced by ultrasonic treatment from slightly less than 10 to slightly less than 3.
Acknowledgements This work was supported by the US Department of Energy, Office of Energy Efficiency and Renewable Energy, Industrial Technologies Program, Industrial Materials for the Future (IMF), under Contract No. DE-PS0702ID14270 with UT-Battelle, LLC. The authors would like to thank Larry Allard and Larry Walker for their suggestions and help in the SEM characterization and EDX analysis. The authors also thank Edward C. Hatfield for his help in the metal casting with ultrasonic processing. The authors also wished to thank Tom Geer for valuable assistance in the preparation and polishing of the samples. References [1] Gruzleski JE, Closset BM. The treatment of liquid aluminum–silicon alloys. Schaumburg, IL: American FoundrymenÕs Society; 1990. p. 17. [2] Lu SZ, Hellawell A. J Cryst Growth 1985;73:316–28. [3] Hanna MD, Lu SZ, Hellawell A. Metall Trans A 1984;15A:459. [4] Heiberg G, Nogita K, Dahle AK, Arnberg L. Acta Mater 2002;50:2537. [5] Nogita K, Dahle AK. Scripta Mater 2003;48:307. [6] Jenkinson DC, Hogan LM. J Cryst Growth 1975;28:171. [7] Dahle AK et al. Metall Mater Trans A 2001;32:949. [8] Nogita K et al. Mater Trans 2001;42:1981. [9] Lu SZ, Hellawell A. Metall Trans A 1987;18A:1721. [10] Jie W, Chen Z, Reif W, Muller K. Metall Mater Trans A 2003;34A:799. [11] Jian X, Xu C, Meek T, Han Q. AFS Transaction 2005, paper 05-085. [12] Jian X, Xu H, Meek T, Han Q. Mater Lett 2005;59(2–3):190. [13] Han Q, Viswanathan S. Mater Sci Eng A 2004;364(1–2):48. [14] Nogita K, Dahle AK. Mater Charact 2001;46:305. [15] Nogita K, McDonald SD, Dahle AK. Mater Forum 2004;28:945. [16] Flood SC, Hunt JD. Met Sci 1981;15:287. [17] McDonald SD, Nogits K, Dahle AK. Acta Mater 2004;52:4273. [18] Han Q, Hunt JD. J Cryst Growth 1995;152:221. [19] Han Q, Hunt JD. ISIJ Int 1995;35:693. [20] Heiberg G, Gandin Ch-A, Goerner H, Arnburg L. Metall Mater Trans 2004;35A:2981.
Refinement of eutectic silicon phase of aluminum A356 alloy using high-intensity ultrasonic vibration X. Jian a
a,*
, T.T. Meek a, Q. Han
b
Materials Science and Engineering Department, University of Tennessee, Knoxville, TN 37996, USA b Oak Ridge National Laboratory, Oak Ridge, TN 37831-6083, USA Received 7 September 2005; received in revised form 24 October 2005; accepted 1 November 2005 Available online 29 November 2005
Abstract The eutectic silicon in A356 alloy can be refined and modified using either chemical, quench, or superheating modification. We observed, for the first time, that the eutectic silicon can also be significantly refined using high-intensity ultrasonic vibration. Rosette-like eutectic silicon is formed during solidification of specimen treated with high-intensity ultrasonic vibration. 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Eutectic solidification; Aluminum alloys; Microstructure; Grain refining; Ultrasonic vibration
1. Introduction Aluminum A356 alloy is one of the widely used casting aluminum alloys because of its good mechanical strength, ductility, hardness, fatigue strength, pressure tightness, fluidity, and machinability [1]. The A356 alloy contains about 50 vol.% eutectic phases. The final microstructure is largely determined by the eutectic reaction. Due to its diamond cubic crystal structure which predominantly grows in h1 1 2i directions on (1 1 1) planes, silicon is a faceted phase with strongly anisotropic growth thus it is difficult to change the growth direction [2]. In unmodified A356 alloy, the main eutectic reaction occurs at 574 C as a binary reaction, which results in coarse irregular plate-like silicon. The modification of eutectic silicon is of general interest since fine eutectic silicon along with fine primary aluminum grains improve mechanical properties and ductility [3]. Three well-known eutectic modification methods have been developed thus far, namely (1) chemical modification
*
Corresponding author. Tel.: +1 865 574 9956; fax: +1 865 574 4357. E-mail address: [email protected] (X. Jian).
[3–8], which produces fine fibrous silicon structure through the addition with trace-levels of several elements, such as sodium, antimony, potassium, calcium, strontium, and barium; (2) quench modification [2,9], which results in silicon forming an exceedingly fine form when compared to the unquenched silicon, through high cooling rates and rapid solidification in the growth rate ranging from 400 lm/s to 1000 lm/s; and (3) superheating modification, which requires the presence of magnesium in the alloy [10] to obtain fibrous silicon by heating up the melt from the usual pouring temperature (for example, 680 C) to a temperature of 850–900 C and then holding for about 15–30 min, then quickly cooling to the pouring temperature before casting. Chemical modification has been widely used in industry for producing A356 alloy with fine and fibrous eutectic silicon phase, and thus improved mechanical properties. In a recent project investigating the effect of high-intensity ultrasonic vibration on the microstructure of A356 alloy, we found that high-intensity ultrasonic vibration has a significant effect on refining both the primary aluminum phase and the eutectic silicon phase [11,12]. This article describes experimental results on modification of the eutectic silicon phase using ultrasonic vibration.
1359-6462/$ - see front matter 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2005.11.004
894
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
2. Experimental conditions Commercial aluminum A356 alloy (Al, 7.0 wt.% Si, 0.4 wt.% Mg, 0.1 wt.%, Fe) was used in this study. The liquidus temperature of the alloy is 614 C and the solidus temperature is 554 C. The primary aluminum dendrites start to form at 614 C and the binary Al–Si eutectic at 574 C. The tertiary eutectics and complex intermetallics form at a late stage of solidification [13]. An ultrasonic system was used to treat the liquid metal after it was poured into a Cu permanent mold. The details of the ultrasonic system can be found in Ref. [12]. It consisted of a 20 kHz ultrasonic generator, a transducer made of piezoelectric lead zirconate titanate crystals (PZT), an ultrasonic horn, and a radiator. The horn and the radiator were made of titanium Ti–6Al–4V alloy. The power output of the unit was variable within a maximum of 1500 W by adjusting the output acoustic amplitude from 24.3 lm to 81 lm, or 30% to 100% of the unitÕs upper limit. The ultrasonic radiator was placed at the bottom of a copper mold which held up to 250 g molten aluminum. The alloy was melted and held in a furnace for 30 min at 700 ± 5 C, about 86 C higher than its liquidus temperature, to allow the complete dissolution of silicon particles. The molten alloy was then poured into the mold at a pouring temperature of 630 C. Ultrasonic vibration with an amplitude of 56.7 lm was started right before the melt was poured into the copper mold. For comparison reasons, samples were also made without ultrasonic vibration. These untreated castings were hemispherical in shape with a diameter of 5 cm and weight of 200 g. The as-polished samples were characterized using an optical microscope with the capability for quantitative metallographic analysis. They were then lightly etched using a 0.5% HF–water solution and further examined using a Hitachi S-4700 scanning electron microscopy (SEM) with light element energy dispersive spectroscopy (EDS) X-ray detectors. Finally the samples were deep etched in a 0.5% HF–water solution for 2 h in order to reveal the three-dimensional morphology of the eutectic silicon phase. 3. Experimental results and discussion The optical micrographs (100·) of the samples from castings made with and without high-intensity ultrasonic vibration are shown in Fig. 1. The casting sample made without ultrasonic vibration exhibited coarse acicular eutectic silicon dispersed among the fully developed primary aluminum dendrites. The eutectic silicon was about 100 lm in length. In addition, one branch of a primary dendrite shown in the middle of Fig. 1(a) was about 800 lm in length, indicating that the grain size was in the range of a few millimeters since one equiaxed grain usually contained six primary dendrite arms. In contrast, the ultrasonically treated A356 alloy displayed a microstructure of continuous very fine Al–Si eutectic interspersed with fine
Fig. 1. Eutectic modification observed by optical metallurgical microscope at low magnification, (a) without ultrasonic treatment and (b) with ultrasonic treatment.
globular primary aluminum grains (Fig. 1(b)). The shape and size of individual eutectic silicon grains were not distinguishable at such a magnification. The formation of spherical primary aluminum phase has been reported elsewhere [12]. Both samples were characterized with SEM at a high magnification of 2000·, as shown in Fig. 2. The coarse acicular eutectic silicon observed on the untreated A356 alloy agreed with that of low magnification observation. While with ultrasonic treatment, very fine eutectic silicon was observed, which was finer than that in strontium-modified A356 alloy of a similar casting condition [14]. The EDS X-ray analysis verified that those particles shown in the eutectic areas were mainly elemental silicon, rather than intermetallic phases or inclusions. Results of quantitative metallographic analysis of silicon morphology are illustrated in Fig. 3. Without ultrasonic treatment, the average length of eutectic silicon was about 26 lm, and the average width 2.7 lm. The aspect ratio was slightly less than 10. While with ultrasonic treatment, the average length and width were respectively about 2 lm and 0.6 lm, with an aspect ratio of slightly less than 3. A comparison of the aspect ratio of the untreated (about
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
895
Fig. 4. Three-dimensional eutectic morphology observed using SEM at 2000· magnification, (a) without ultrasonic treatment and (b) with ultrasonic treatment. Fig. 2. Eutectic modification observed by SEM at 2000· magnification, (a) without ultrasonic vibration and (b) with ultrasonic vibration.
Fig. 3. Silicon morphological analyses of A356 alloy without and with ultrasonic treatment.
9.8) and ultrasonically treated (about 2.8) A356 alloy suggests that the silicon morphology in ultrasonically modified A356 alloy is not just an exceedingly fine form of the silicon in the unmodified alloy.
Fig. 4 shows the comparison of the three-dimensional morphology of eutectic silicon observed by SEM on the deep-etched samples. Fig. 4(a) shows the eutectic silicon phase in the sample not subject to ultrasonic treatment. The silicon phase exhibits a typical coarse plate-like form. Most likely these silicon plates grew epitaxially from the surrounding primary aluminum dendrites. This result is in accordance with previous reports for unmodified A356 alloys [15]. Fig. 4(b) shows the silicon morphology in a sample subject to ultrasonic vibration. There are two distinct features of the silicon phase in the ultrasonically processed sample. Firstly the silicon phase is rosette-like, rather than plate-like. The center of a rosette could be the center of an eutectic cell/grain, indicating the nucleation of the eutectic phases could be independent to the surrounding primary aluminum dendrites. Another feature is that the eutectic phase spacing is much smaller in the ultrasonically processed sample than that in the sample not subject to ultrasonic vibration. These two features suggest that the modified eutectic silicon using ultrasonic vibration is not simply a finer form of the unmodified A356 alloy. The mechanism by which the finer eutectic silicon phase forms under ultrasonic treatment is not clear at the moment. The eutectic spacing is much smaller in the ultrasonically processed specimen than that in the untreated
896
X. Jian et al. / Scripta Materialia 54 (2006) 893–896
specimen. Small eutectic spacing usually results from high growth rates of these eutectic phases [16]. However, since ultrasonic energy about 600 W was injected into a specimen, the cooling rates in the ultrasonically processed specimen were in fact slower than the one not subject to ultrasonic vibration. On the other hand, the eutectic grains/cells as outlined in Fig. 4(b) were smaller than that in the specimen not subject to ultrasonic vibration. This is again not in accordance with previous experimental evidence that techniques that increase nucleation lead to the coarsening of the eutectics by decreasing their growth rates [17]. A number of unique phenomena that occur during ultrasonic processing of alloy may contribute to the refinement of the eutectic silicon phase. One phenomenon is acoustically induced convection in the liquid pools. This vigorous convection tends to dislodge AlP particles at the existing dendrite interfaces and disperse them uniformly into the remaining liquid [18,19]. Phosphorous in the form of AlP particles is often quoted as a good nucleant for silicon. The removal of these particles from the existing dendrite interfaces prevents the nucleation of silicon particles on the existing dendrites. Acoustically induced convection can also affect nucleation of the silicon phases by altering the constitutional supercooling at the front of the growing eutectic grains [20]. Another phenomenon is the ultrasonically induced pressure variations in the remaining liquid pools. Further research is needed to investigate the effect of these two phenomena on the nucleation and growth of the eutectics in order to fully understand the mechanisms of silicon modification under ultrasonic vibration. 4. Conclusions Ultrasonic vibration at 20 kHz was introduced into A356 alloy as it was cast into a metal mold at the temperature of 630 C. The results showed that: (1) The morphology of eutectic silicon was modified from a coarse acicular plate-like form when no ultrasonic vibration was used, to a finely dispersed rosettelike form when ultrasonic treatment was employed. (2) Ultrasonic treatment reduced the size of eutectic silicon from 26 lm to 2 lm in length, which is over
an order of magnitude, and the width from 2.7 lm to 0.6 lm. The aspect ratio was also reduced by ultrasonic treatment from slightly less than 10 to slightly less than 3.
Acknowledgements This work was supported by the US Department of Energy, Office of Energy Efficiency and Renewable Energy, Industrial Technologies Program, Industrial Materials for the Future (IMF), under Contract No. DE-PS0702ID14270 with UT-Battelle, LLC. The authors would like to thank Larry Allard and Larry Walker for their suggestions and help in the SEM characterization and EDX analysis. The authors also thank Edward C. Hatfield for his help in the metal casting with ultrasonic processing. The authors also wished to thank Tom Geer for valuable assistance in the preparation and polishing of the samples. References [1] Gruzleski JE, Closset BM. The treatment of liquid aluminum–silicon alloys. Schaumburg, IL: American FoundrymenÕs Society; 1990. p. 17. [2] Lu SZ, Hellawell A. J Cryst Growth 1985;73:316–28. [3] Hanna MD, Lu SZ, Hellawell A. Metall Trans A 1984;15A:459. [4] Heiberg G, Nogita K, Dahle AK, Arnberg L. Acta Mater 2002;50:2537. [5] Nogita K, Dahle AK. Scripta Mater 2003;48:307. [6] Jenkinson DC, Hogan LM. J Cryst Growth 1975;28:171. [7] Dahle AK et al. Metall Mater Trans A 2001;32:949. [8] Nogita K et al. Mater Trans 2001;42:1981. [9] Lu SZ, Hellawell A. Metall Trans A 1987;18A:1721. [10] Jie W, Chen Z, Reif W, Muller K. Metall Mater Trans A 2003;34A:799. [11] Jian X, Xu C, Meek T, Han Q. AFS Transaction 2005, paper 05-085. [12] Jian X, Xu H, Meek T, Han Q. Mater Lett 2005;59(2–3):190. [13] Han Q, Viswanathan S. Mater Sci Eng A 2004;364(1–2):48. [14] Nogita K, Dahle AK. Mater Charact 2001;46:305. [15] Nogita K, McDonald SD, Dahle AK. Mater Forum 2004;28:945. [16] Flood SC, Hunt JD. Met Sci 1981;15:287. [17] McDonald SD, Nogits K, Dahle AK. Acta Mater 2004;52:4273. [18] Han Q, Hunt JD. J Cryst Growth 1995;152:221. [19] Han Q, Hunt JD. ISIJ Int 1995;35:693. [20] Heiberg G, Gandin Ch-A, Goerner H, Arnburg L. Metall Mater Trans 2004;35A:2981.