Effect of Conditions of Unidirectional Solidification on Microstructure

and Pore Morphology of Al-Mg-Si Alloys

Tae Bum Kim*, Shinsuke Suzuki and Hideo Nakajima

The Institute of Scientific and Industrial Research, Osaka University, Ibaraki 567-0047, Japan

Aluminum and Al-Mg-Si alloy ingots with pores were fabricated by unidirectional solidification through thermal decomposition ofCa(OH)2 powders. The porosity of aluminum and Al-Mg-Si alloy were 10–17% and 0.1–2%, respectively. While the pores with 250–400mmdiameter were observed in a grain or across several grains in the aluminum ingots, smaller pores with 50–300mm were observed in an eutecticregion between primary � dendrites in the Al-Mg-Si ingots. In the alloys with Mg(0.25–0.5mass%) and Si(0.2–0.4mass%), the unidirectionalpores were aligned between columnar dendrites grown in the unidirectional solidification. With higher Mg and Si contents, the equiaxed dendritezones with spherical pores were observed in a region with low temperature gradient. The results of thermal analysis showed that constitutionalsupercooling, which causes equiaxed dendrites, tends to occur with increase in Mg and Si contents and with low temperature gradient at thesolid-liquid interface. Under this condition, spherical pores were evolved, because the surrounding �-dendrites solidified isotropically.Therefore, it is concluded that the pore growth direction is affected by morphology of dendrites. [doi:10.2320/matertrans.M2009333]

(Received October 1, 2009; Accepted December 7, 2009; Published January 27, 2010)

Keywords: unidirectional solidification, porous aluminum, aluminum-magnesium-silicon alloy, A6061, columnar dendrite, equiaxed dendrite

1. Introduction

Recently, porous aluminum is expected to be applied tolight-weight structural materials with various functionalproperties. Among various kinds of porous aluminum,aluminum foam with a higher porosity than 90% is activelystudied.1) This kind of aluminum foam possesses isotropicspherical pores and is applied to sound-absorbing materialsand shock absorption materials. However, the aluminumfoam does not retain enough strength for structural materialdue to its high porosity; the spherical pores cause stressconcentration under a stress, which prevents improvement ofstrength even if the porosity decreases.

On the other hand lotus-type porous metals with longcylindrical gas pores aligned in one direction, exhibitsuperior mechanical properties to that of isotropic porousmetals.2–4) The specific strength of lotus-type porous metalsin the pore growth direction does not decrease because nostress concentration takes place even if the pores exist.4)

Therefore, lotus-type porous metals are considered to beeffective light-weight structural materials.

The lotus metals can be fabricated by the unidirectionalsolidification in the pressurized gas atmosphere such ashydrogen. During solidification of melt dissolving hydrogen,the pores are evolved by rejected gas from the solid and thenthe pores grow in the solidification direction. Mold castingand continuous casting technique are used as unidirectionalsolidification process.

It is well known that there are two kinds of supplyingmethods of hydrogen to the molten metal. In one methodhydrogen is supplied from a pressurized gas atmosphere.3,4)

In another method hydrogen is supplied through thermaldecomposition reaction of compounds including hydro-gen.5,6) The thermal decomposition method was developedby Nakajima and Ide in order to fabricate lotus-type porousmetals without explosive gas and high pressure chamber.5)

Until now several kinds of lotus aluminum alloys havealready been fabricated with hypo-eutectic Al-Si alloys7) andhypo-eutectic Al-Cu alloy8) by continuous casting techniquein hydrogen atmosphere of 0.1MPa. These results showedthat pores were formed in the eutectic region among severalprimary �-dendrites by the solubility gap of hydrogenbetween liquid and solid phases, and then grew in thedirection of growth of the columnar primary �-dendrites.Therefore the shape and the direction of the primary �-dendrite in the aluminum alloys are considered strongly toaffect the morphologies and the direction of pores, since theeutectic solidification is the final stage of the solidification.The influence of the surrounding �-dendrites on the mor-phologies and direction of pores is significant in an alloy witha small volume fraction of the eutectic phase, because thegrowth of pores is interrupted by the surrounding �-dendriteseasily. Further investigations of the effect of the micro-structure on the pore morphology are necessary for control-ling the pore morphology in various commercial aluminumalloys.

In this study, Al-Mg-Si alloys with various Mg andSi contents were unidirectionally solidified by moldcasting technique. The thermal decomposition method wasapplied to the mold casting method so that small pores wereformed in the sample. Al-Mg-Si alloys were prepared bydiluting A6061, which is widely used structural material.This paper reports the effect of the microstructures on thepore morphologies and discusses the model of pore forma-tion.

2. Experimental Procedure

2.1 Mold casting of Al-Mg-Si alloy with thermal decom-position reaction

Al-Mg-Si alloys with various Mg and Si contents wereprepared by diluting A6061 by 25, 33 and 50% with purealuminum, which are called for convenience 1/4, 1/3 and1/2A6061, respectively.*Corresponding author, E-mail: kim23@sanken.osaka-u.ac.jp

Materials Transactions, Vol. 51, No. 3 (2010) pp. 496 to 502#2010 The Japan Institute of Metals

Figure 1 shows a schematic drawing of the mold castingapparatus which consists of melting and casting part in achamber. A crucible was set in an induction heating coilin the upper part of the apparatus. The metal blocks in thecrucible were melted by induction heating and the melt waspoured into a mold through a hole at the bottom of thecrucible by opening the stopper. The mold made of a thinstainless steel sheet of 0.1mm in thickness was 25mm indiameter and 90mm in height. The mold was set on a copperchiller cooled by circulating water so that the melt wasunidirectionally solidified in the mold.

In this study, Ca(OH)2 was selected as the compound forsupplying hydrogen into the melt. Kim et al. fabricated lotusaluminum using various kinds of compounds.6) According totheir results, Ca(OH)2 provides higher porosity than TiH2 andNaHCO3.

The Ca(OH)2 powder of 0.2 g and aluminum powder of0.2 g were mixed homogeneously and wrapped in analuminum foil in order to disperse Ca(OH)2 in the melt.The wrapped powder was set on the copper chiller.

Pure aluminum (99.99%), 1/2A6061, 1/3A6061,1/4A6061 or A6061 of about 80 g was melted in a graphitecrucible by induction heating in vacuum (about 10 Pa) andpoured into the mold at 1023K. The melt was unidirection-ally solidified in the mold.

The samples were cut in the directions parallel andperpendicular to the solidification direction by a wireelectrical discharge machine (model A320D, Sodic Co.

Ltd., Yokohama, Japan). The average porosity and porediameters were measured using an image analyzer WinRoof(Mitani Co. Ltd., Tokyo, Japan). The cross-sections ofthe samples were polished and etched in a solution(distilled water : 32% Hydrochloric acid : 65% Nitric acid :

40% Fluorinated acid ¼ 1 : 1 : 1 : 0:4). The macro andmicrostructures were observed by an optical microscope(Keyence, VHX-200, Tokyo, Japan).

The Si contents of the cast samples were analyzed bymethod of weight of silicon dioxide following JIS H1352.The Mg contents were analyzed by ICP-MS (InductivelyCoupled Plasma-Mass Spectrometry, Shimadzu ICPV-1017).The Ca contents were analyzed by ICP-AES (InductivelyCoupled Plasma-Atomic Emission Spectrometer, ShimadzuICPS-8000). These analyses were carried out in KansaiDivision of Sumitomo Metal Technology, INC.

2.2 Temperature measurement during solidificationand thermal analysis of the cast samples

The temperature change during the solidification wasmeasured by K-type thermocouples set at the positions of 10,20 and 30mm from the copper chiller on the bottom, asshown in Fig. 1(b). The other conditions were almost sameas the mold casting described in 2.1. The difference wasthat mold casting was done without Ca(OH)2 powder in theair, because of difficulty of setting the thermocouples in thevacuum chamber. However, the microstructures of fabricatedsample in the air were confirmed to be similar to thatfabricated in vacuum. Therefore, the measurement results oftemperature change in the air can be applied to the experi-ments in vacuum.

The liquidus and solidus lines of the fabricated sampleswere analyzed by TG-DTA (Thermo Gravimetric-Differ-ential Thermal Analysis; Bruker, TG-DTA2000SA). Thesamples were heated at 2K/min in argon flow.

3. Results and Discussion

3.1 Microstructure, porosity and pore diameterTable 1 shows the results of ICP analyses of the fabricated

samples. It was confirmed that the contents of Si and Mgwere proportional to the dilution ratio of the original A6061.Not so large amount of Ca was detected. It is consideredthat the decomposed Ca from Ca(OH)2 powder does notsignificantly affect the composition of the alloys.

Figure 2(a) shows macrostructures of the cross-sections ofthe samples in the direction parallel to the solidificationdirection. The columnar dendrite zone was observed in purealuminum from the bottom to the top of the sample.However, the area of columnar dendrite zone reduced in

(a)

(b)

Fig. 1 Schematic set up of fabrication apparatus using the mold casting

technique (a) and setting of thermocouples inside the mold (b).

Table 1 Contents of Mg, Si and Ca in the cast samples.

(mass%)

Mg Si Ca

A6061 0.96 0.78 <0:01

1/2A6061 0.48 0.35 <0:01

1/3A6061 0.34 0.25 <0:01

1/4A6061 0.22 0.18 <0:01

Effect of Conditions of Unidirectional Solidification on Microstructure and Pore Morphology of Al-Mg-Si Alloys 497

the top of the sample with increasing Si and Mg contents andequiaxed dendrite zone was observed instead of the columnardendrite. An equiaxed dendrite zone was observed in thewhole sample of A6061.

The average porosities of the samples of pure aluminum,1/4A6061, 1/3A6061, 1/2A6061 and A6061, were 13%, 8%(including of large pores), 2%, 1% and 8%, respectively.

Figure 2(b) shows the cross-sectional views of the samplesin the direction perpendicular to the solidification directionat distances of 10, 20 and 30mm from the bottom. The purealuminum ingot had a lot of pores with diameter of about500 mm. The Al-Mg-Si samples had pores smaller than 1mmin diameter except large pores observed in 1/4A6061. A lotof pores smaller than several hundreds mm were observed byoptical microscope.

In the process of this study, hydrogen is consideredto generate by the following two-step reactions. First, thethermal decomposition reaction of Ca(OH)2 powder startedabout 823K.9)

Ca(OH)2 ! CaOþ H2O(g) ð1Þ

Then the decomposed moisture reacts with the aluminummelt.

2Alþ 3H2O(g) ! 6Hþ Al2O3 ð2Þ

The large pores observed in 1/4A6061 at distances of10mm from the bottom are considered to be trapped bubblesof H2 gas

10) during the solidification, because the large poresgrew across several grains. The conditions for the generationof this kind of large pore have not yet been clear.

On the other hand, small pores are considered to be formedby hydrogen extruded from the solid by the solubility gap ofhydrogen between liquid and solid aluminum alloys.

Figure 3 shows the microstructures on the cross-sectionsin the solidification direction at 10, 20 and 30mm fromthe bottom of the samples. While the pores were observedinside the grains in the pure aluminum, the pores (in black)were observed in the eutectic phase (grey) between the �phases (white) in the alloys. The directional thin pores wereobserved to be parallel to the unidirectionally solidifiedprimary � phases in the range up to 30mm away from thebottom in 1/4A6061 and 1/3A6061. However, a clearboundary from columnar to equiaxed dendrite region wasobserved at the position of 30mm from the bottom of the1/2A6061 sample. The pore morphologies changed at theboundary from directional pores to isotropic pores inirregular shapes. These results show that directional poresform in the columnar dendrite zones and isotropic pores formin the equiaxed dendrite zones. These results agree well withthe reported results on arc welding of A6061.11)

Figure 4 shows that the pore aspect ratio calculated usingthe average pore diameter and the pore length at distance of10, 20 and 30mm from the bottom. The pores grew in thesolidification direction in pure aluminum because of noobstacles of the precipitates. The length of pores in thealuminum alloys, however, became shorter with increasingconcentration of alloying elements. The pores in A6061 werealmost spherical. These results were attributed to an increasein the semi-solid mushy zone region in front of the solidphase with increasing Si and Mg.

(a)

(b)

Fig. 2 Cross sectional views of the cast sample parallel (a) and perpendicular (b) to the solidification direction (arrow). The cross sectional

views perpendicular to the solidification direction at distances of 10, 20 and 30mm from the bottom are shown in (b).

498 T. B. Kim, S. Suzuki and H. Nakajima

Figure 5(a) shows the pore diameters measured on thecross-section at the position of 10, 20 and 30mm away fromthe bottom of each sample. The large pores formed bybubbling were eliminated from the measurements. Thepore diameters in the columnar dendrite zone of 1/4, 1/3and 1/2 A6061 were about 200 mm, which was similar tothe distances between the primary � phases. On the otherhand, the pore diameters in the equaixed dendrite zone of1/2A6061 and the A6061 were larger than that in thecolumnar dendrite zone.

Figure 5(b) show the porosity estimated as the areafraction of the pores in the cross-sectional area measuredby the image analyzer. Although the porosity seemsextremely higher at 10mm from the bottom of 1/4A6061sample than the other samples in Fig. 2(b), the increase ofthe porosity was caused by the large pores, which wereformed by bubbling. After elimination of the data of the largepores, the porosity of 1/4A6061 was similar to that of otherAl-Mg-Si alloys. The pore diameter and the porosity in theequiaxed zone increased with an increase in distance fromthe bottom.

According to the previous studies, addition of 1mass% ofmagnesium increases the solubility of hydrogen in aluminumalloys by about 1� 10�5 mass% H,12,13) while addition of1mass% of silicon decreases in the solubility of hydrogenby about 2� 10�6 mass% H.13,14) These data show that thesolubility change by the alloying elements of silicon andmagnesium is considered not to affect the porosity signifi-cantly in this study. Therefore, the ratio of hydrogen, which

Fig. 3 Microstructures of the cast samples on the cross section in the parallel direction to the solidification direction (arrow) at distance of

10, 20 and 30mm from the bottom.

0 100

2

4

6

8 1/4A6061 1/3A6061 1/2A6061 A6061

Distance from bottom of mold / mm

Por

e as

pect

rat

io,

. d−1

20 30 40

Fig. 4 Pore aspect ratio calculated using the average pore diameter d and

the pore length ‘ at distance of 10, 20 and 30mm from the bottom.

(b)

0 10 30 50

0

5

10

15

20 Pure Al 1/4A6061 1/3A6061 1/2A6061 A6061

Por

osity

/%

Distance from bottom of mold /mm

(a)

0 10 30 500

200

400

600

800

1000P

ore

diam

eter

/µm

Distance from bottom of mold /mm

Pure Al 1/4A60611/3A6061 1/2A6061 A6061

20 40

20 40

Fig. 5 Pore diameter (a) and porosity (b) plotted against distance from

bottom of mold. The error bars show the standard deviations.

Effect of Conditions of Unidirectional Solidification on Microstructure and Pore Morphology of Al-Mg-Si Alloys 499

contributed to the formation of pore, was discussed based onthe hydrogen solubility in pure aluminum at the melting pointin hydrogen atmosphere of 0.1MPa.

The hydrogen solubility of solid aluminum is 0.036 cc/100 g–Al and the hydrogen solubility of liquid aluminumis 0.69 cc/100 g–Al in hydrogen atmosphere of 0.1MPa.Namely, the hydrogen solubility gap between solid andliquid is 0.65 cc/100 g–Al under the 0.1MPa hydrogenatmosphere.15) When the hydrogen solubility in the solid-phase and liquid state can be calculated by a sievert’s law,the solubility gap between solid phase and liquid states is0:65� ð10 Pa=105 PaÞ1=2 ¼ 6:5� 10�3 cc/100 g–Al under ahydrogen pressure of 10 Pa.

On the other hand, the amount of hydrogen, whichcontributed to the pore formation, was possible to becalculated by the volume of pores and the equation of gasstate. Here an aluminum sample with porosity of 3% isdiscussed, according to Fig. 5(b). The volume of pores inthe aluminium sample of 100 g is 1:1� 10�6 m3, whichcorresponds to hydrogen content of 1:4� 10�3 cc/100 g–Alunder the standard condition according to Boyle’s law.The pressure is assumed to be a sum of the hydrostaticpressure of the melt of 400 Pa and the atmospheric pressureof 10 Pa. Although the hydrostatic pressure varies by thevertical position of the melt, the value at 30mm under theuppermost part of the melt, which is considered to be theaverage value.

Under the assumption that the hydrogen dissolved in themelt at the equilibrium value of 6:9� 10�3 cc/100 g–Alunder a hydrogen pressure of 10 Pa, about 20% of thedissolved hydrogen contributed to the pore formation. It wasreported that about 50% of hydrogen excluded from solidphase by solubility gap during solidification contributes tothe pore formation in experiments of fabrication of lotuscopper by mold casting under hydrogen atmosphere.16) Therest of the excluded hydrogen is considered to escape into theatmosphere. The hydrogen escape ratio of hydrogen fromaluminum is higher than that from copper, because thesolubility of hydrogen in liquid aluminum is much lower thanthat in liquid copper.

3.2 Pore morphology and solidification conditionsSince the morphologies and direction of pores are

influenced by that of the surrounding primary �-dendrites,the pores are considered to form in the eutectic region, whichsolidified in the final stage of the solidification.

Although the volume fraction of the eutectic phase wassmall in the alloys used in this study, the influence of thesurrounding �-dendrites on the morphologies and direction ofpores is significant. The growth of pores is interrupted by thesurrounding primary �-dendrites, when the volume fractionof eutectic phase is small according to the results offabrication of lotus-type porous Al-Si alloys.7) In Al-4mass%Si and Al-8mass% Si with a small volume fraction ofeutectic phases, the morphologies and direction of pores wereinfluenced by the surrounding primary �-dendrites. On theother hand, in Al-Si alloys with a Si-content of 12, 14, and18mass% with a large volume fraction of eutectic phases, thepores grew in the solidification direction and were in acylindrical shape.

The effects of the solidification conditions on the mor-phology of the pores formed in the eutectic region arediscussed. Figure 6 shows the cooling curve during unidirec-tional solidification at distance of 10, 20 and 30mm from thebottom of the samples. The liquidus temperature TL and thesolidus temperature TS were also shown in Fig. 6. Theliquidus temperature TL and the solidus temperature TS ofeach alloy were measured by the TG-DTA as shown inFig. 7. Both the liquidus and solidus temperatures decreasedwith increasing Mg and Si contents and the semi-solid region(mushy zone) between the liquidus and solidus temperaturesincreased.

Figure 8 shows the time of start of solidification afterpouring of each alloy, which is indicated as an arrow in Fig. 6in Pure Al, against the distance from the bottom of the

Fig. 6 Cooling curves during solidification of each alloy at the distance of

10mm (dotted line), 20mm (dashed line) and 30mm (solid line) from the

bottom of the mold. The arrow shows the time of start of solidification

after pouring.

Fig. 7 TG-DTA curves of alloys of 1/4A6061, 1/3A6061, 1/2A6061,

A6061.

500 T. B. Kim, S. Suzuki and H. Nakajima

sample. The relation between the thickness of the solidifiedlayer x and the time from the start to the end of solidificationcan be described as

t ¼ x2=�2 þ t0 ð3Þ

where � and t0 are the solidification constant and the time ofthe end of the solidification at the bottom, respectively.17)

The plots in Fig. 8 were fitted by eq. (3) with fittingparameters of � and t0. Figure 9 shows the solidificationrates estimated by the differentials of the fitting curves ateach position. The temperature gradients G in liquid at theinterface to the mushy zone were calculated using thesolidification rates in Fig. 8 and the cooling rate, which is thegradient of the cooling curve just above TS in Fig. 10.

G ¼ ðdT=dtÞ=R ð4Þ

Hunt reported the solidification conditions for equiaxeddendrite zone and the columnar dendrite zone.18) A diagramof the microstructures, the pore morphologies, temperaturegradient and the solidification rate was shown in Fig. 11based on the Hunt’s model. The lower right side of thediagram is the conditions for equiaxed dendrite zone withisotopic pores and the upper left side of the diagram is theconditions for columnar dendrite zone with directional pores.

The columnar dendrite zone with directional pore wasobserved under the cooling conditions with a high temper-ature gradient and a slow solidification rate. Therefore, ahigher temperature gradient and a lower solidification rate arenecessary to form directional pores and dendrite.

Figure 12 shows schematic drawings of the effect of thetemperature distribution on the microstructure and poremorphology based on the model of columnar-to-equiaxedtransition.19,20) Since the redistribution of solute does notoccur in pure aluminum, the melting point is constant andconstitutional supercooling does not take place. Thereforedirectional pores are formed in pure aluminum (Fig. 12(a)).

Even in alloys, constitutional supercooling does not takeplace either with a narrow solidification temperature range ofthe alloy or with a large temperature gradient in front of thesolid-liquid interface (Fig. 12(b)).

If the temperature gradient in the mushy zone decreasesin an alloy with a wide solidification temperature range,primary � phase nucleates in the mushy zone due to

0 100

10

20

30

40

50

1/2A6061

1/3A6061

1/4A6061

Sol

idifi

catio

n tim

e / s

Distance from bottom of mold / mm

1/4A6061 1/3A6061 1/2A6061 A6061

A6061

20 30 40

Fig. 8 Time of start of solidification after pouring (Fig. 7) of each alloy

against the distance from the bottom of the mold.

0 100

1

2

3

4

Sol

idifi

catio

n ra

te /

mm

s-1

Distance from bottom of mold / mm

1/4A6061 1/3A6061 1/2A6061 A6061

20 30 40

Fig. 9 Solidification rate of each alloy with distance from bottom of mold.

0 10 20 30 401

2

3

4

5

Tem

pera

ture

gra

dien

t G /

K m

m-1

.

Distance from bottom mold / mm

1/4A6061 1/3A6061 1/2A6061 A6061

Fig. 10 Temperature gradient in the semi solid region at the solidification

front. The x-axis shows the position of the solidification front.

Fig. 11 Influence of velocity and temperature gradient on the micro-

structure. The numbers next to the plots show the distance from the bottom

of the mold in mm. Columnar dendrites were observed the region upper

left side from the dotted line. Equxied grains were observed the region

lower right side from the dotted line.

Effect of Conditions of Unidirectional Solidification on Microstructure and Pore Morphology of Al-Mg-Si Alloys 501

constitutional supercooling. At the same time the pores formby excluded gas from the solid phase in the mushy zone(Fig. 12(c)). The pores grow spherically in this kind of alloys,because the surrounding � phases are isotropic equiaxeddendrite. Furthermore, as the surrounding � phases do notsuppress the pore growth, the pores in the equiaxed zonegrow larger than that in the columnar zone.

Although the part at 20mm from the bottom of A6061 wascooled under the condition where columnar dendrites withdirectional pores were expected, equiaxed dendrites withisotropic irregular pores were observed, which are shown inbrackets shown in Fig. 11.

4. Conclusion

Pores were observed in Al-Mg-Si alloys with various Mgand Si contents (diluted A6061) unidirectionally solidified bythe mold casting technique in vacuum through thermaldecomposition of Ca(OH)2 powders. The direction andmorphology of pores were influenced by that of thesurrounding primary �-dendrites. Therefore unidirectionalpores and spherical pores form under the solidificationconditions for columnar dendrites and equiaxed dendrites,respectively, which can be explained by the Hunt’s model.The results suggest that the low solidification rate and thehigh temperature gradient are necessary for formation ofdirectional pores.

Acknowledgements

This work was supported by the Global COE program(Centre of Excellent for Advanced Structural and FunctionalMaterials Design) from the Ministry of Education, Culture,Sports, Science and Technology and the Grants-in-Aid for

Young Scientists (B), Iketani Science and TechnologyFoundation. The pure aluminium ingots were supplied bythe Light Metal Educational Foundation, Inc.

REFERENCES

1) J. Banhart: Prog. Mater. Sci. 46 (2001) 559–632.

2) O. Knacke, H. Probst and J. Wernekinck: Z. Metallkd. 70 (1979)

1–6.

3) V. I. Shapovalov: MRS Bull. 19 (1994) 24–28.

4) H. Nakajima: Prog. Mater. Sci. 52 (2007) 1091–1173.

5) H. Nakajima and T. Ide: Metall. Mater. Trans. A 39A (2008) 390–

394.

6) S. Y. Kim, J. S. Park and H. Nakajima: Metall. Mater. Trans. A 40A

(2009) 937–942.

7) J. S. Park, S. K. Hyun, S. Suzuki and H. Nakajima: Metall. Mater.

Trans. A 40A (2009) 406–414.

8) S. Suzuki, T. B. Kim and H. Nakajima: J. Phys.: Conf. Ser. 165 (2009)

012068.

9) T. B. Kim, S. Suzuki and H. Nakajima: J. Phys.: Conf. Ser. 165 (2009)

012067.

10) T. A. Engh and P. Bakke: Second Australian Asian Pacific Course and

Conference on Aluminum Melt Treatment and Casting, (University of

Melbourne, Australia, 1991) pp. 1–5.

11) Y. M. Zhang, C. Pan and A. T. Male: Metall. Mater. Trans. A 31A

(2000) 2537–2543.

12) W. R. Opie and N. J. Grant: Trans. AIME 188 (1950) 1237–1241.

13) W. Baukloh and F. Oesterlen: Z. Metallkd. 30 (1938) 386–389.

14) P. N. Anyalebechi: Scr. Metall. Mater. 33 (1995) 1209–1216.

15) Y. Shinada and S. Nishi: J. Jpn. Inst. Light Met. 30 (1980) 317–323.

16) S. Yamamura, H. Shiota, K. Murakami and H. Nakajima: Mater. Sci.

Eng. A 318 (2001) 137–143.

17) C. W. Briggs and R. A. Gezelius: Trans. Am. Found. Assoc. 43 (1935)

274–302.

18) J. D. Hunt: Mater. Sci. Eng. 65 (1984) 75–83.

19) M. A. Martorano and V. B. Biscuola: Acta Mater. 57 (2009) 607–

615.

20) G. Heiberg, K. Nogita, A. K. Dahle and L. Arnberg: Acta Mater. 50

(2002) 2537–2546.

TL

Tactual

Distance0 X

Tem

pera

ture

(a)

Pore

Solid Liquid

Tem

pera

ture

Distance

Tactual

0 X

TL

(b)

Liquid

Pore

Columnar dendrite

(c)

Fig. 12 Schematic drawings of the relationship between the temperature distribution and the pore formation. (a) Pure metal, (b) alloy

without constitutional supercooling, (c) alloy with constitutional supercooling.

502 T. B. Kim, S. Suzuki and H. Nakajima