You are on page 1of 12

Eutectic Nucleation and Growth in Hypoeutectic Al-Si Alloys

at Different Strontium Levels


A.K. DAHLE, K. NOGITA, J.W. ZINDEL, S.D. McDONALD, and L.M. HOGAN

The effects of different levels of strontium on nucleation and growth of the eutectic in a commercial
hypoeutectic Al-Si foundry alloy have been investigated by optical microscopy and electron backscat-
tering diffraction (EBSD) mapping by scanning electron microscopy (SEM). The microstructural
evolution of each specimen during solidification was studied by a quenching technique at different
temperatures and Sr contents. By comparing the orientation of the aluminum in the eutectic to that
of the surrounding primary aluminum dendrites by EBSD, the eutectic formation mechanism could
be determined. The results of these studies show that the eutectic nucleation mode, and subsequent
growth mode, is strongly dependent on Sr level. Three distinctly different eutectic growth modes
were found, in isolation or sometimes together, but different for each Sr content. At very low Sr
contents, the eutectic nucleated and grew from the primary phase. Increasing the Sr level to between
70 and 110 ppm resulted in nucleation of independent eutectic grains with no relation to the primary
dendrites. At a Sr level of 500 ppm, the eutectic again nucleated on and grew from the primary phase
while a well-modified eutectic structure was still present. A slight dependency of eutectic growth
radially from the mold wall opposite the thermal gradient was observed in all specimens in the early
stages of eutectic solidification.

I. INTRODUCTION normally accompany the refined eutectic structure, although


CONTROL of microstructure formation during solidifi- the benefits associated with modification can sometimes
cation is becoming increasingly important for manufacturers be outweighed by altered porosity characteristics associated
with modification, particularly modification with stron-
of cast aluminum components. Not only does the microstruc-
ture determine the mechanical properties, but the evolution tium.[2] Because the formation of eutectic often is the final
stage of solidification, it can be expected to have a significant
of microstructure during the casting process also defines the
boundaries of the casting design, the castability, and the impact on the formation of casting defects, particularly
susceptibility to formation of defects, and therefore, consis- porosity, and castability. Recent work[3–6] has indicated that
tency and quality of the products. Significant development our knowledge and understanding of this crucial stage in
work has been performed to obtain means for monitoring, the solidification process is very limited and that there are
assuring, and controlling microstructure development. Al- three different possible eutectic nucleation and growth
Si alloys of hypoeutectic composition are the most widely modes depending on the solidification conditions. Figure 1 is
used aluminum foundry alloys today. Microstructure control an illustration of the three proposed eutectic growth modes.[3]
in these alloys is often obtained through further alloying These are nucleation at or adjacent to the wall and front
with, and/or in the presence of, other elements, notably Ti, growth opposite the thermal gradient, (Figure 1(a)), nucle-
Mg, Cu, and Fe. Grain refinement can be obtained by the ation of eutectic on primary dendrites, (Figure 1(b)), and
use of commercial master alloy additions, although these heterogeneous nucleation of eutectic on nucleant particles
are not as effective as in the wrought alloys.[1] in the interdendritic liquid, (Figure 1(c). The black regions
Commercial Al-Si foundry alloys normally contain about in Figure 1 indicate the Al-Si eutectic during solidification
50 to 90 vol pct eutectic. Controlling the eutectic silicon but do not show the coupled eutectic structure inside. If the
morphology by modification has been used extensively eutectic nucleates on the primary aluminum phase (mecha-
industrially since about the 1970s to improve the mechanical nism b), it can be expected that the crystallographic orienta-
properties of the castings. Modification, obtained by addition tion of the aluminum in the eutectic is related to that of the
of Na or Sr, or by quench modification, changes the morphol- dendrites. On the other hand, if the eutectic is nucleated
ogy of silicon from a platelike or lamellar structure to a fine separately in the interdendritic melt (mechanism c), it is
fibrous structure, while modification by addition of Sb only very likely that the aluminum and silicon still comprise a
refines the silicon platelets. Improved mechanical properties common solidification front during growth, i.e., coupled
eutectic growth, as divorced eutectic growth has not been
reported for the Al-Si system. Therefore, the aluminum
phase forms a low-energy interface with the silicon that leads
A.K. DAHLE, Senior Lecturer, K. NOGITA, Research Fellow, and S.D.
McDONALD, Postgraduate Student, are with the Department of Mining, the growth front, i.e., with a crystallographic orientation
Minerals and Materials Engineering, The University of Queensland, relationship with the silicon and not with the surrounding
Brisbane, QLD 4072, Australia. J.W. ZINDEL, Senior Technical Specialist, dendrites. If growth occurs from the wall (mechanism a), a
is with the Ford Research Laboratory, Ford Motor Company, Dearborn, similar eutectic growth front can be expected. The orienta-
MI 48121. L.M. HOGAN, formerly Honorary Research Fellow, Department
of Mining, Minerals and Materials Engineering, The University of Queens-
tion relationship between primary and eutectic aluminum
land, is deceased. would depend on the operating nucleation and growth mech-
Manuscript submitted June 20, 2000. anism of the eutectic. It would be of great benefit to be able

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—949


Table I. Chemical Composition of 319 Alloys
(Weight Percent)
Alloy Si Cu Mg Fe Mn Ti Sr
Unmodified 7.06 3.37 0.18 0.61 0.34 0.11 ⬍0.002
70 ppm Sr 7.06 3.13 0.23 0.54 0.31 0.11 0.007
110 ppm Sr 7.06 3.13 0.20 0.60 0.30 0.11 0.011
500 ppm Sr 7.06 3.13 0.20 0.60 0.30 0.11 0.050

(a)

(b)

Fig. 2—Sketch of single-use cylindrical sample cup. All dimensions are


(c) in millimeters.
Fig. 1—Possible eutectic growth morphologies in the Al-Si system: (a)
front growth opposite the thermal gradient, (b) nucleation and growth
on the primary aluminum dendrites, and (c) independent heterogeneous Approximately 20 kg of the alloy was melted and held in
nucleation of eutectic grains in interdendritic spaces. The eutectic containing an electric resistance furnace at 720 ⬚C.
Al and Si growing side-by-side is illustrated in black.
Samples were prepared from melts with four levels of Sr.
The as-received alloy contained approximately 0.007 wt pct
to control eutectic nucleation and growth and suitable means Sr (70 ppm Sr modification). A lance was used to inject a
of control need to be identified. commercial flux into the melt to lower the Sr level to below
Furthermore, the effects of the different growth modes on 0.002 wt pct (unmodified sample) using a nitrogen carrier
casting defect formation are of particular practical gas. Two additional Sr levels were produced by adding
importance. AlSr10 master alloy to yield Sr levels of 110 and 500 ppm.
The present work was conducted to gain further under- Custom made sample cups were designed to hold a ther-
standing about the effect of eutectic modification by different mocouple in the center of the sample in order to monitor
levels of Sr on nucleation and growth of the eutectic in a the sample temperature during both the slow cooling and
commercial hypoeutectic Al-Si alloy. The macrostructural quench segments of the experiments. Figure 2 shows a sketch
evolution of the eutectic during solidification was character- of the single-use sample cup. The cups consisted of a stain-
ized by optical and scanning electron microscopy (SEM) less steel tube with an outer diameter of 22 mm and wall
and crystallographic orientation mapping by electron back- thickness of 0.9 mm, and with a 0.85-mm-thick cap welded
scattering diffraction (EBSD) of samples quenched at differ- to the bottom of the tube to fully seal the bottom end. Cup
ent stages during the eutectic arrest. exteriors were coated with boron nitride to minimize the
amount of aluminum and dross adhering to the outside of
the cup. Samples were taken by submerging the cup into
II. EXPERIMENTAL the melt after allowing it to reach the melt temperature. The
The composition of the base alloy, which is a variation sample was then placed in an insulating sleeve above a
of Aluminum Association alloy 319, is given in Table I. quench tank and allowed to cool until the desired quench

950—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A


temperature was reached, at which time the sample was
plunged into an ice-brine quench bath. The cooling rate of
the liquid at 610 ⬚C, just prior to nucleation of the solid,
was approximately 1.7 ⬚C/s, and the total solidification time
was approximately 300 seconds. The samples were then
sectioned perpendicular to the axis of the cylinder, 15 mm
from the bottom, and prepared for metallographic analysis.
Microstructures were studied in an optical microscope.
The samples were etched in a 10 g NaOH, 5 g K3Fe(CN)6,
and 60 mL H2O solution for 15 to 20 seconds after final
polishing to distinguish clearly between primary dendrites,
Al-Si eutectic, and quench liquid. The etchant preferentially
attacks the quenched liquid, which therefore appears black/
dark in the micrographs. Analysis of the microstructural
constituents and crystallographic orientation measurements
from primary dendrites and aluminum in the eutectic were
performed in a PHILIPS* XL30 SEM with an EBSD
*PHILIPS is a trademark of Philips Electronic Optics Ltd., Mahwah, NJ.

system attached. Sample preparation for EBSD mapping


was by an ion-milling procedure and further details on the
sample preparation and measurement technique have been
presented by Nogita and Dahle.[6] Several eutectic regions
were studied in each sample and representative results are
presented in this article.

III. RESULTS
A. Final Microstructures Fig. 3—Microstructure at the end of solidification (unquenched): (a)
unmodified, (b) 70 ppm Sr, (c) 110 ppm Sr, and (d ) 500 ppm Sr.
Figure 3 shows samples that were allowed to solidify
completely without interruption by quenching. A fully equi-
axed dendritic microstructure was found in all samples, inde-
pendent of the level of Sr modification. A coarse lamellar contain growing eutectic, but at this early stage of eutectic
silicon structure is observed in the unmodified alloys, solidification, it was mostly located close to the walls, as
whereas the alloys containing 70, 110, and 500 ppm Sr show discussed later. Several small regions of eutectic growth can
the typical fine fibrous eutectic structure. Although 500 ppm be observed in the micrographs of the sample quenched 45
Sr can be considered a very high level of strontium addition, seconds after the start of the eutectic reaction (Figure 5(b)).
the sample did not show any signs of overmodification, such In most cases, the regions of eutectic growth are associated
as the presence of primary silicon particles or overmodifica- with the tips of the dendrite arms, and rarely does the eutectic
tion bands. It is worthwhile emphasizing that eutectic bound- seem to initiate and grow at their bases. The eutectic has
aries often can be observed, and these are particularly evolved further in the sample quenched 76 seconds after the
distinguishable at higher silicon levels. These boundaries start of the eutectic arrest. It is clear that the eutectic has
are characterized by a high degree of disorder and often formed close to the tips of the dendrite arms and that it tends
ternary eutectic precipitates. However, it is not possible to to grow into the liquid with a very irregular solid/liquid
identify the eutectic growth mode by visually inspecting the interface. Branching of the silicon platelets can be observed.
microstructure solidified without interruption. Figure 6 shows optical micrographs of samples containing
70 ppm Sr, quenched at different stages during eutectic
growth. Only a small amount of eutectic is present 31 sec-
B. Optical Micrographs of Quenched Samples
onds after the start of the eutectic arrest and most of the
Figure 4 shows cooling curves and derivative curves for sample consists of only quenched liquid in the interdendritic
the four alloys. Three major phase diagram reactions can areas (Figure 6(a)). Some small eutectic regions can be
be clearly observed in the cooling curves; primary phase discerned growing close to the tips of the dendrites. The
solidification, binary Al-Si eutectic, and ternary Al-Si-Cu solid/liquid interface is quite compact, i.e., relatively smooth,
eutectic reaction. All samples were quenched during the and the eutectic has a grainlike appearance. The connection
binary Al-Si eutectic reaction. with the primary phase appears less than for the unmodified
Optical micrographs of the unmodified alloy quenched at alloy. Eutectic grain boundaries are clearly observed in the
different stages during the eutectic arrest are shown in Fig- sample quenched 95 seconds after the start of the eutectic
ures 5(a) to (c). Very little eutectic can be observed in the arrest (Figure 6(b)). What appears as several independent
sample quenched 19 seconds after the start of eutectic solidi- eutectic grains can be observed, and the black regions are
fication (Figure 5(a)). Only quenched liquid, which appears the etched quenched liquid. The solid/liquid interface is very
gray, and some large iron containing precipitates can be smooth and the grains have a slight globular appearance.
found in addition to the primary dendrites. This sample did The eutectic has developed further in the sample quenched

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—951


(a) (b)

(c) (d )
Fig. 4—Cooling curve and derivative curve of (a) unmodified alloy and alloys modified by (b) 70 ppm Sr, (c) 110 ppm Sr, and (d ) 500 ppm Sr. The
starting temperature and time for the eutectic reaction is indicated on each figure. Quench times quoted in this article are from the start time indicated as
teut for the eutectic arrest.

134 seconds after the eutectic arrest, and areas of impinge- surface of the sample are observed in all cases. The unmodi-
ment of eutectic grains can be observed, displaying a slightly fied sample contains a thin layer of exuded eutectic liquid
irregular appearance of the silicon particles compared to that on the surface that is not found in the modified samples.
inside each eutectic grain. The eutectic grain boundaries can
be observed to pass through the central stem of some
dendrites.
C. Electron Backscattering Diffraction
Quenched samples of the alloy containing 110 ppm Sr are
shown in Figure 7. The evolution of eutectic solidification is Secondary electron images from the SEM with corres-
very similar to that observed in Figure 6. The eutectic appears ponding EBSD orientation maps are shown for all alloys,
to solidify first in intergranular areas, and eutectic grain i.e., unmodified alloy and alloys containing 70, 110, and
boundaries can be observed to cut through primary dendrites 500 ppm Sr, in Figures 10 through 13, respectively. The
in the sample quenched 84 seconds after the start of the color of the mapping pixels is identical when the difference
eutectic arrest (Figure 7(c)). in crystallographic orientation of adjacent regions is less
Figure 8 shows optical micrographs of quenched samples than 5 deg, i.e., identical color indicates identical orientation.
containing 500 ppm Sr. The eutectic appears to nucleate and The unmodified alloy contains relatively large silicon plate-
start growing at the tips of the dendrite arms in the sample lets (Figure 10(a)), and the corresponding orientation map
quenched 33 seconds after the start of the eutectic arrest (Figure 10(b)) shows that all the aluminum in the eutectic
(Figure 8(a)). The eutectic grows into the intergranular areas has identical orientation and is, therefore, assigned the same
upon further solidification rather than solidifying the intra- blue color. All primary phase in the pictured region has the
dendritic regions within the dendrite envelopes (Figure 8(b) same color and, therefore, belongs to the same dendrite. The
and (c)). It therefore, again, seems that the intergranular mapping image also contains small lamellar-like regions
regions are more preferred for eutectic growth. with different colors. These correspond to silicon particles
Figure 9 shows optical micrographs from the wall region according to the secondary electron image and this was also
of the unmodified alloy and alloys modified by 110 and 500 confirmed by simultaneous energy-dispersive X-ray (EDX)
ppm Sr, respectively. Some dependency of eutectic volume measurements. Regions displayed in black are areas where
fraction with position from the wall can be observed in all the system did not record sufficient crystallographic data
the samples, but open liquid channels penetrating to the (diffraction data) to allow determination of crystallographic

952—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 5—Optical micrographs of unmodified samples quenched at various Fig. 6—Optical micrographs of the alloy modified by 70 ppm Sr and
times: (a) 19 s, (b) 45 s, and (c) 76 s, after the commencement of eutec- quenched at different times after the start of the eutectic arrest: (a) 31 s,
tic solidification. (b) 95 s, and (c) 134 s.

orientation. From this result, it is clear that the crystallo-


graphic orientations of the aluminum in the eutectic and orientation map for the alloy containing 70 ppm Sr is shown
the surrounding primary dendrite arms are identical in the in Figure 11. A modified eutectic with small fibrous silicon
unmodified alloy. particles can be observed. The EBSD result of the same area
A secondary electron image with corresponding EBSD indicates that the investigated eutectic region is surrounded

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—953


Fig. 7—Optical micrographs of sample containing 110 ppm Sr and Fig. 8—Optical micrographs of alloy containing 500 ppm Sr quenched at
quenched at different times after the start of the eutectic arrest: (a) 33 s, different times after the start of the eutectic arrest: (a) 33 s, (b) 68 s, and
(b) 62 s, and (c) 84 s. (c) 76 s.

by dendrite arms that belong to two different equiaxed den- the eutectic has not nucleated on the primary aluminum
drites. A wide range of different orientations can be observed dendrites, as there is no systematic orientation relationship
within the eutectic region, representing a structure consisting between them. It is interesting to note that a different result
of very small grains. This indicates that the aluminum in is found within the small eutectic region between secondary

954—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A


Fig. 10—Unmodified specimen: (a) SEM secondary electron image and
(b) EBSD orientation map of the same area. Crystallographic orientations
of aluminum in eutectic and surrounding primary dendrites are identical.

Figure 12 shows the secondary electron image and corres-


ponding EBSD orientation map from the specimen con-
taining 110 ppm Sr. A well-modified eutectic structure is
readily observed. The eutectic region is surrounded by sev-
eral dendrite arms. A range of different orientations of alumi-
num is observed within the eutectic, some replicating the
orientation of the adjacent dendrite arm. However, several
different orientations not similar to the dendrite arms are
also found. Black regions in Figure 12 are the result of
insufficient orientation information, usually correlating with
silicon particles, which often cause a shaded/deformed dif-
fraction pattern that the system cannot index, or no diffrac-
tion pattern at all.
Figure 13 shows the secondary electron image and corres-
ponding EBSD orientation map for the alloy containing 500
ppm Sr. The eutectic does not display any signs of overmodi-
Fig. 9—Optical micrographs of near the wall region of quenched samples: fication, but rather a refined, well-modified silicon structure.
(a) unmodified (76 s after start of eutectic arrest), (b) 110 ppm Sr (33 s),
and (c) 500 ppm Sr (33 s).
Figure 13(b) shows that two dendrites enclose the eutectic
region, and the orientation mapping data indicate that most
of the eutectic aluminum originated from the surrounding
dendrite arms on the bottom left-hand side of Figure 11(b) dendrites, growing from each dendrite to form a grain bound-
and also in small regions adjacent to the dendrite arms in ary in the center of the eutectic region. Some small areas
the top part of the eutectic region. Here, the orientation of of different orientation are observed and also several black,
eutectic aluminum is identical to the surrounding primary nonindexed silicon particles. A relatively large eutectic
dendrite arms. region with a flake silicon structure and different aluminum

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—955


Fig. 11—Sample modified by 70 ppm Sr: (a) SEM secondary electron
image and (b) EBSD orientation map of the same area. A wide variation
in orientations of the aluminum in the eutectic is observed.
Fig. 12—Specimen modified by 110 ppm Sr: (a) SEM secondary electron
image and (b) EBSD orientation map of the same area. A wide distribution
of orientation of the aluminum in the eutectic is observed.
orientation (shown in grey) is observed on the right-hand
side of the eutectic region.
It should be noted that the EBSD patterns from aluminum aluminum.[10] Silicon crystals grow strongly anisotropic
and silicon are very similar since aluminum has a face- along certain crystallographic directions, and they, therefore,
centered cubic atomic structure and silicon has a diamond become bounded by the more slowly growing facets, i.e.,
cubic structure. In the case of the present measurements, the {111} habit planes.[8] The Al-Si eutectic is an irregular eutec-
EBSD program was set for indexing of diffraction patterns tic. Due to a difficulty in changing the growth direction of
from aluminum. Simultaneous analysis by EDX for compo- silicon, branching and termination of growth occurs so that
sition measurements and indexing the EBSD map using there is not a unique lamellar spacing in these alloys. Because
the silicon crystal structure, or further transmission electron silicon can change its growth direction by twinning, there
microscopy (TEM) observation, is therefore required to clar- may not necessarily always be a specific orientation relation-
ify the orientation relationship between aluminum and sili- ship between aluminum and silicon, since it has been sug-
con within the eutectic. gested that aluminum may not be able to change its growth
direction sufficiently fast to adapt to the rapid twinning of
silicon during eutectic growth.[8] Repeated epitaxial nucle-
IV. DISCUSSION ation of the aluminum on silicon during eutectic growth has
The eutectic in Al-Si alloys is more complex than in been reported by Kobayashi et al.,[11] Shamsuzzoha and
other aluminum alloys because silicon is a faceted phase. Hogan,[12] and Hogan and Song.[13]
Directional solidification studies have shown that the Al-Si Much work has been performed to understand the effect
eutectic is a coupled eutectic, i.e., with eutectic aluminum of strontium and sodium in providing modification of the
and silicon growing in a quasisteady-state condition, side- silicon phase, and altered growth characteristics of eutectic
by-side with a low-energy interface.[7,8,9] The growth front silicon, by the twin-plane re-entrant edge mechanism, are
of the unmodified Al-Si eutectic has been shown to be very now the generally accepted explanation. The effect of stron-
jagged and led by silicon, which has the smallest constitu- tium is to increase the twin density by so-called impurity
tional undercooling and, therefore, projects ahead of the induced twinning, and it also yields a smoother eutectic

956—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A


lamellae in unmodified alloys. Silicon is the “leading phase”
during growth of the Al-Si eutectic, and according to the
theory proposed by Ohno,[18] this is strong evidence that
supports silicon being an efficient nucleant for aluminum,
whereas aluminum is a much less efficient nucleant for
silicon. The assertion is, therefore, that the formation of the
eutectic in hypoeutectic Al-Si alloys strongly depends on
the nucleant(s) for silicon, although direct evidence to sup-
port this statement seems to be lacking.
The observed evolution of the eutectic in the quenched
samples of the unmodified alloy in Figure 5, combined with
the EBSD orientation map in Figure 10, shows clearly and
unequivocally that the eutectic nucleates on or adjacent to
the primary aluminum dendrites and that eutectic aluminum
grows with the same orientation as that of the surrounding
primary dendrite. This corresponds to the mechanism illus-
trated in Figure 1(b). It is very interesting to note that the
results indicate that nucleation and growth of the eutectic
is concentrated on the tips of the dendrites in the early stages
of the eutectic arrest. Figure 14(a) presents a micrograph
from an unmodified sample quenched just after the eutectic
arrest, which lends further support to the theory that the
eutectic nucleates on the dendrite tips. This result, therefore,
clearly indicates that aluminum can act as a nucleant for the
eutectic, although it is not clear whether silicon is nucleated
directly on the aluminum or on particles just ahead of the
interface. In light of the results of Crosley and Mondolfo,[17]
who showed that aluminum is a relatively good nucleant for
silicon, the preceding result is not surprising and may indi-
cate that silicon nucleated on the aluminum dendrite. Eutec-
tic aluminum certainly did. Dahle et al.[3] have also shown,
by crystallographic studies, that the unmodified eutectic in
hypoeutectic Al-Si alloys nucleates on the dendrites. Inspec-
tion of the micrographs of quenched samples published by
Kim and Heine,[19] Flood and Hunt,[20] Denton and Spittle,[21]
and Hamed et al.[22] provides further indirect evidence of
Fig. 13—Specimen modified by 500 ppm Sr: (a) SEM secondary electron
image and (b) EBSD orientation map of the same area. Crystallograhic eutectic nucleation on primary aluminum in unmodified
orientations of aluminum in eutectic and surrounding primary dendrites hypoeutectic Al-Si alloys.
are identical. The tendency for the eutectic to grow into the intergranular
liquid, rather than toward the center of the equiaxed dendrite
in the unmodified alloy, is also very interesting. One possible
growth front during solidification.[11] Modification is also explanation for this effect is that it can be related to a higher
accompanied by a depression in the eutectic temperature degree of mixing and more unrestricted diffusion compared
and a shift of the eutectic point to higher silicon to regions closer to the center of the equiaxed dendrites,
concentrations.[11,14] such as between secondary dendrite arms.
Despite the significant amount of research on solidifica- When the alloy contains strontium, the eutectic silicon
tion of the Al-Si eutectic, the nucleation stage has not is transformed to a fibrous appearance, and the quenched
received a similar amount of focus. It has been concluded microstructures for 70 and 110 ppm Sr (Figures 6 and 7)
that the nucleation of silicon controls the solidification of indicate that there has also been a transition in the nucleation
the eutectic,[8,11,15,16] i.e., the eutectic does not start to grow of the eutectic. Instead of being connected to the dendrites,
until the first silicon has formed. However, it is worth empha- as for the unmodified alloy, the eutectic is now more distrib-
sizing that the information available about nucleation is both uted and fills the large intergranular areas. Eutectic grain
confusing and contradictory. It is generally believed that boundaries that intersect the dendrites are observed and there
silicon does not nucleate on aluminum.[16] However, the seems to be no linkage between the evolution of the eutectic
pioneering work by Crosley and Mondolfo[17] showed that and the presence of the equiaxed dendrites. The EBSD orien-
aluminum can nucleate silicon at a few degrees undercooling tation maps for these two Sr contents in Figures 11 and
and silicon does not nucleate aluminum. To the contrary, 12 show that there is no longer an orientation relationship
again, efficient nucleation of aluminum on silicon is evi- between the eutectic aluminum and the surrounding den-
denced by the formation of an aluminum halo around pri- drites. The eutectic growth mode at 70 and 110 ppm Sr is
mary silicon crystals in hypereutectic alloys (while the according to the illustration in Figure 1(c). Figure 14(b)
reverse does not readily occur), and also by the reported shows a micrograph of a Sr-modified sample quenched just
epitaxial renucleation of aluminum on growing silicon after the start of the eutectic arrest, which nicely indicates

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—957


growth front and, therefore, less renucleation of eutectic
aluminum, but the result clearly indicates a stronger tendency
for some eutectic nucleation on the dendrite arms with 110
ppm Sr.
A closer inspection of the micrographs presented in Fig-
ures 6 and 7, particularly for the longer quench times, indi-
cates that small aluminum “particles” can be observed in
locations close to the center of a large fraction of all eutectic
grains. These centrally located aluminum particles are quite
round in shape and are significantly smaller than the second-
ary dendrite arms, which may indicate that they are separate
and formed separately. They are likely to be rodlike rather
than spherical. The presence of these particles raises the
question as to whether they acted as nucleation sites for the
eutectic grains, or were perhaps the first stage of eutectic
solidification before the coupled eutectic started growing.
Further studies are required to investigate and confirm
(a) this effect.
When the alloy is modified by 500 ppm Sr, the quenched
microstructures indicate that the eutectic is again associated
with the equiaxed dendrites, and again nucleating close to
their tips, i.e., different from the samples containing 70 and
110 ppm Sr. The EBSD orientation data (Figure 13(b)) shows
clearly that the eutectic has grown from the dendrites,
according to the illustration in Figure 1(b), and identical to
what has been found in the unmodified alloy. Although a
fully modified structure is observed, the nucleation mecha-
nism and macroscopic growth mode has changed. This is
very interesting, as it shows that the nucleation and macro-
scopic evolution of the eutectic is independent of the appear-
ance of eutectic silicon, i.e., flake or fibrous. A strontium
level of 500 ppm is much larger than that used in commercial
operations (50 to 300 ppm), although overmodification (with
the presence of primary silicon particles and overmodifica-
tion bands) is generally not associated with Sr. Large stron-
tium-containing phases can often be found at high Sr
(b) concentrations.
It is clear from the previous discussion that strontium
Fig. 14—Samples quenched at the beginning of eutectic solidification
showing (a) nucleation on the tips of the dendrites in an unmodified speci- exerts a significant influence on the nucleation and growth
men and (b) independent nucleation in the interdendritic liquid in a Sr- of the Al-Si eutectic. The effect of strontium on the growth
modified specimen. of silicon is well documented and will not be discussed
further in the present article. However, the nucleation behav-
ior of the eutectic grains requires further discussion. The
the growth of a eutectic grain in the center of an interdendritic nucleation behavior and the resulting macroscopic distribu-
channel, totally independent of the primary phase. Indepen- tion of eutectic within the mushy zone can be expected to
dent nucleation of eutectic grains has also been reported in have a significant influence on interdendritic permeability
the work by Laslaz,[16] Dahle et al.,[3] and Wang et al.[5] and, therefore, to significantly affect and control the amount
Further support can be found in reinterpreting the quenched and distribution of porosity. The significance of eutectic
microstructures published by Denton and Spittle[21] and growth mode on porosity formation has been studied and
Hamed et al.[22] discussed further by McDonald et al.[2] and Dahle et al.[23]
Further inspection of the EBSD orientation maps for the The modification effect, i.e., altering the silicon morphol-
samples containing 70 and 100 ppm Sr in Figures 11(b) and ogy from flakes to fibers, of elements such as Na and Sr
12(b), respectively, indicate that the sizes of the isoorienta- was initially related to altered nucleation characteristics. The
tion areas for eutectic aluminum are very different, and much AIP particles were found to be very efficient nucleants for
larger at 110 ppm Sr. The significance of these isoorientation silicon, and Crosley and Mondolfo[17] suggested that the
areas is not quite clear. The eutectic interface is normally modification effect of Na, which is similar to Sr, was that
quite smooth upon Sr modification, as also confirmed by it neutralized active nucleants for silicon, such as the AIP
the quenched microstructures in Figures 6 and 7. However, particles. Phosphorus is well known to be a good nucleant
it is possible that silicon is still the leading phase during for primary silicon in hypereutectic Al-Si alloys and is used
growth of the Al-Si eutectic and that renucleation of alumi- as a grain refiner in these alloys. However, if the observed
num occurs. Larger isoorientation areas at the higher level effect of Sr on eutectic nucleation was a result of interaction
of Sr modification may therefore be the result of a smoother with dispersed particles in the melt, it is very difficult to

958—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A


understand the reversion of the nucleation mechanism at mechanism dominates in the early part of the eutectic solidi-
500 ppm Sr. The quenched microstructures of the unmodified fication. The initial growth from the wall may be a result
alloy indicate that nucleation occurs at the tips of the den- of a CET, and the local isotherms and thermal gradients in
drites. This effect could be the result of the phenomenon of the sample in the initial stages of the eutectic arrest would
particle pushing where nonwetted particles are pushed ahead therefore be important. Further studies of the effect of ther-
of the evolving dendrites, as well as nucleation directly on mal gradient (during eutectic solidification) and growth rate
the interface. If heterogeneous nucleant particles pushed are warranted to determine their effect in controlling the
ahead of the interface are activated, their point of activation nucleation behavior of the eutectic in hypoeutectic Al-Si
can be expected when the constitutional undercooling for alloys.
the nucleating phase is sufficient to cause nucleation of the The nucleation behavior of the eutectic between dendrite
eutectic and when the melt composition is sufficiently close arms in hypoeutectic Al-Si alloys raises an interesting point.
to the eutectic. Our crystallographic orientation measurements often show
It seems probable that the formation of the eutectic in identical orientation of the eutectic aluminum and the sur-
hypoeutectic Al-Si alloys is also governed by a columnar- rounding secondary dendrite arms. These areas often contain
equiaxed transition (CET), where nucleation on the dendrites a single relatively large silicon flake that is visible between
(Figure 1(b)) can be considered “columnar” and independent a pair of arms, the remainder being aluminum. It seems
eutectic grains (Figure 1(c)) “equiaxed.” The change from probable that in such cases the silicon nucleated first from
the melt (necessarily Si-rich because of Si rejection from
radial growth from the wall (Figure 1(a)) to evenly distrib-
the dendrite). Growth of the silicon flake would consume
uted over the cross section is also likely to be the result of
the excess silicon, leaving growth of the dendrite to resume
a CET. Hunt[24] has proposed a mathematical model describ-
to solidify the remaining melt and initiate two-phase eutectic
ing the CET. At low thermal gradients and high growth growth, somewhat similar to a divorced eutectic growth
rates, equiaxed grains are expected. The driving force for mechanism. Further studies to support this theory should
the transition is constitutional undercooling and the potency be considered.
of the nucleants present. The effect of strontium can be
twofold. It is likely to be segregated into the liquid ahead
V. CONCLUSIONS
of the primary dendrites, therefore potentially increasing the
constitutional undercooling. More likely, however, is that it The macrostructural evolution and solidification of eutec-
affects the nucleants present in the melt according to the tic in hypoeutectic Al-7Si-3Cu-0.2Mg (319) alloys with dif-
theory proposed by Crosley and Mondolfo.[17] The reversion ferent levels of strontium was characterized by optical
of eutectic nucleation for 500 ppm Sr is very surprising and microscopy, SEM, and EBSD mapping analysis of samples
may be an indication that the effect of Sr is related to the quenched during solidification.
presence of some strontium containing particles, or other The results show that the eutectic nucleates on the primary
inclusions introduced with the strontium addition, that act phase in the unmodified alloy. Nucleation seems to occur
at the tips of the dendrites and the eutectic has a tendency
as nucleants for the eutectic. Although active at 70 and 110
to evolve into intergranular regions rather than filling the
ppm Sr, these particles may have become too large to be
dendrite envelopes.
efficient nucleants at higher Sr contents, or they may be of
With 70 and 110 ppm Sr, eutectic grains are nucleated
different composition and/or crystal structure. More detailed in the intergranular regions independently of the equiaxed
studies of the effect of Sr on nucleant population is required aluminum dendrites. A well-modified eutectic silicon struc-
to clarify the mechanism by which strontium affects the ture is also observed.
macroscopic distribution of solidifying eutectic in hypoeu- At 500 ppm Sr, the eutectic is still well modified, but is
tectic Al-Si alloys. The observed difference in eutectic nucle- found to nucleate on the dendrites as in the unmodified alloy.
ation and growth at 500 ppm, compared to 70 and 110 This shows that nucleation of eutectic grains is independent
ppm Sr, correlates very well with the observations made of the modification of the silicon phase promoted by stron-
by DasGupta et al.[25] who found that the eutectic arrest tium additions.
temperature increased with Sr additions above 100 ppm. Some tendency for the eutectic to evolve radially inward
They also found that the normal connected network of silicon from the mold wall in the early stages of the eutectic arrest
fibers in a seaweed structure vanished at Sr levels exceeding was observed. However, a relatively uniform distribution of
100 ppm, although the eutectic was still well modified, as growing eutectic was found later during eutectic solidi-
in the present work. DasGupta et al.[25] also found that there fication.
was a change in fracture mode, from ductile to more brittle, It seems likely that strontium affects the number and
when the level of strontium exceeded 100 ppm. potency of the nucleants for silicon available in the interden-
Figure 9 shows that the eutectic does show a tendency to dritic liquid. Because a reversion in nucleation mode is
radiate inward from the mold wall, similar to the mode observed at 500 ppm Sr, it is possible that Sr containing
illustrated in Figure 1, in the initial stages of the eutectic particles promote independent nucleation at low Sr levels,
arrest for all samples. However, open channels penetrating while these particles have become too large or have changed
to the surface of the samples were observed in all samples potency at higher Sr levels.
and these remained open for a significant part of the eutectic
arrest. Eutectic growth was observed with a relatively uni- ACKNOWLEDGMENTS
form distribution in samples quenched further into the eutec- This project is funded by a large grant from the Australian
tic arrest. This result, therefore, indicates that the eutectic Research Council (ARC) and is also supported by an award
did not solely grow radially from the wall, but this growth from the Kazato Research Foundation, Japan.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 32A, APRIL 2001—959


REFERENCES 12. M. Shamuzzoha and L.M. Hogan: J. Cryst. Growth, 1986, vol. 76,
pp. 429-39.
1. Y.C. Lee, A.K. Dahle, D.H. StJohn, and J.E.C. Hutt: Mater. Sci. Eng.
13. L.M. Hogan and H. Song: Acta Metall., 1987, vol. 35, pp. 677-80.
A, 1999, vol. A259, pp. 43-52.
14. M.D. Hanna, S.-Z. Lu, and A. Hellawell: Metall. Trans. A, 1984, vol.
2. S.D. McDonald, K. Nogita, A.K. Dahle, J.A. Taylor, and D.H. StJohn:
15A, pp. 459-69.
AFS Trans., 2000, vol. 108, pp. 463-70.
15. M.G. Day: J. Inst. Met., 1970, vol. 98, p. 57.
3. A.K. Dahle, J. Hjelen, and L. Arnberg: Proc. 4th Decennial Int. Conf.
on Solidification Processing (SP97), J. Beech and H. Jones, eds., 16. G. Laslaz: Proc. 4th Int. Conf. on Molten Aluminum Processing, Ameri-
Sheffield, United Kingdom, 1997, pp. 527-30. can Foundrymen’s Society, Des Plaines, IL, 1995, pp. 459-80.
4. A.K. Dahle, J. Hjelen, and L. Arnberg: Electron Microscopy 1998, 17. P.B. Crosley and L.F. Mondolfo: AFS Trans., 1966, vol. 74, pp.
Proc. ICEM 14, Institute of Physics Publishing. Materials Science 1, 53-64.
Bristol, United Kingdom, 1998, vol. 2, pp. 135-36. 18. A. Ohno: The Solidification of Metals, Chijin Shokan Co. Ltd.,
5. Q.G. Wang, D. Apelian, L. Arnberg, S. Gulbrandsen-Dahl, and J. Tokyo, 1976.
Hjelen: AFS Trans., 1999, vol. 107, pp. 249-56. 19. C.B. Kim and R.W. Heine: J. Inst. Met., 1963–64, vol. 92, pp. 367-76.
6. K. Nogita and A.K. Dahle: Materials Characterization, in press. 20. S.C. Flood and J.D. Hunt: Met. Sci., 1981, vol. 15, pp. 287-94.
7. D.C. Jenkinson and L.M. Hogan: J. Cryst. Growth, 1975, vol. 28, pp. 21. J.R. Denton and J.A. Spittle: Mater. Sci. Technol., 1985, vol. 1, pp.
171-87. 305-11.
8. S.-Z. Lu and A. Hellawell: Metall. Trans. A, 1987, vol. 18A, pp. 22. Q.S. Hamed, R. Elliott, and P.S. Cooper: in Light Metals 1992, E.R.
1721-33. Cutshall, ed., TMS, Warrendale, PA, 1991, pp. 1391-97.
9. M. Shamuzzoha, L.M. Hogan, and J.T. Berry: AFS Trans., 1992, vol. 23. A.K. Dahle, J.A. Taylor, and D.A. Graham: Aluminum Trans., 2000,
100, pp. 619-29. vol. 3 (1), pp. 17-30.
10. S.-Z. Lu and A. Hellawell: JOM, 1995, Feb., pp. 38-40. 24. J.D. Hunt: Mater. Sci. Eng., 1984, vol. 65, pp. 75-83.
11. K. Kobayashi, P.H. Shingu, and R. Ozaki: Proc. Int. Conf. Solidification 25. R. Dasgupta, C.G. Brown, and S. Marek: AFS Trans., 1988, vol. 96,
and Casting, Metals Society, London, 1979, pp. 101-05. pp. 297-310.

960—VOLUME 32A, APRIL 2001 METALLURGICAL AND MATERIALS TRANSACTIONS A

You might also like