Analysis of microstructure evolution and precise solid fractionevaluation of A356 aluminum alloy during partial re-meltingby a color etching method

Li Gao • Yohei Harada • Shinji Kumai

Received: 17 January 2012 / Accepted: 16 May 2012 / Published online: 30 May 2012

� Springer Science+Business Media, LLC 2012

Abstract Spheroidization of Al grains is required for the

production of semi-solid slurry either by a partial solidifi-

cation route or partial re-melting route. In this research,

A356 aluminum alloy was deformed and partially re-mel-

ted to semi-solid state. A segregation sensitive reagent

(Weck’s reagent) was used to reveal the inner micro-

structure of Al grains for the better understanding of the

microstructure evolution during partial re-melting. Optical

microstructure observation showed that the previously

compressed Al dendrites were actually ‘‘fractured’’ during

heat treatment and such ‘‘fractured’’ dendrites contributed

to the refinement and spheroidization of Al grains. Further

study of this phenomenon indicates that the ‘‘fractures’’ are

actually migrating high-angle grain boundaries, which was

related to the recrystallization that occurred during heat

treatment. Moreover, the growth layer of Al grains formed

during water quenching is clearly visualized after being

etched by Weck’s reagent. Consequently, precise evalua-

tion of solid fractions through image analysis was realized

by excluding growth layer when measuring the area of

solid phase.

Introduction

Semi-solid processes

For high-integrity aluminum alloy automotive component

fabrication, semi-solid process (SSP) has its advantages

over conventional casting processes using completely

melted metal, such as better mold filling ability (solid-front

fill), less air entrapment, and lower solidification shrinkage

[1]. Thus, many studies have been carried out on this

process since it was developed in the United States [2].

In aluminum alloy SSP, the existence of spheroidal Al

grains is required. Besides the partial re-melting of spray-

formed materials [3], generally speaking, the methods of

making semi-solid slurry containing spheroidal Al grains can

be mainly classified into two categories. One is agitation

either mechanically or electromagnetically [1] including

SEED process [4] and bubbling process [5]. The other one

includes strain-induced melt activation (SIMA) [6] and

recrystallization and partial melting (RAP) route [7], which

can be used to get spheroidal grains via partially re-melting

of previously deformed (rolling, extrusion, etc.) billet.

The difference between SIMA and RAP is that SIMA pro-

cess requires hot working above recrystallization tempera-

ture and cold working subsequently, while RAP process is

simpler and only needs to deform the material to a critical

extent by ‘‘warm working’’ between room temperature and

recrystallization temperature. For the alloy which is supplied

in worked state, both SIMA and RAP processes have an

advantage over agitating the alloy in liquid state because they

are simpler and need less equipment [8].

Recrystallization is thought to be the key mechanism

contributing to the grain refinement and spheroidization in

both the SIMA and RAP processes by their inventors [6, 7],

which is also widely considered by other researchers [9–14].

Needless to say, understanding of recrystallization behavior

during SIMA or RAP processes is of great significance.

However, most of the recognitions of recrystallization phe-

nomenon in SIMA or RAP processes are obtained without

visualization of deformed dendrites, which is commonly

used as the starting material in the above two processes.

Consequently, recrystallization behavior in previously

L. Gao � Y. Harada � S. Kumai (&)

Department of Materials Science and Engineering, Tokyo

Institute of Technology, Yokohama 226-8502, Japan

e-mail: kumai.s.aa@m.titech.ac.jp

123

J Mater Sci (2012) 47:6553–6564

DOI 10.1007/s10853-012-6585-x

deformed dendritic microstructure during heating and partial

re-melting is still not clear.

Evaluation of solid fraction

Pursuing of accurate solid fraction evaluation for a certain

semi-solid temperature has been persisted since SSP was

brought to researchers’ sight. Recently, several routes are

being used to measure or calculate the solid fraction. The

most essential one was Scheil’s model [15], where four

assumptions are required: (1) the interface is at equilibrium;

(2) there is no diffusion in the solid phase; (3) the composi-

tions are homogeneous through the entire liquid phase; and

(4) densities of the solid and liquid phases are equal. More

practically used methods are thermal analysis, thermody-

namic simulation, and quantitative metallography. Chen

et al. [16], Nafisi et al. [17], and Tzimas et al. [18] compared

the results from different methods and commented on each

method. However, in their studies, disagreement of results

yielded from different solid fraction measuring methods is

inevitable. Birol [19] focused on differential scanning calo-

rimetry (DSC) experiment. Results show that measured solid

fraction is influenced greatly by the scan (heating and cool-

ing) rate. In situ optical measurement of solid fraction was

carried out by Steinbach and Ratke [20]. The result is in

agreement with Scheil’s model. But the experimental set-up

is complicate and scatter of result can be caused due to the

scatter of intensity measured from the sample’s surface by a

CCD camera.

Due to the growth of pre-existing grains when being

cooled from semi-solid state, overestimation of solid

fractions obtained by quantitative metallography from a

quenched specimen has been confirmed [16–18]. In order

to overcome this drawback, Wannasin et al. [21] used a

segregation sensitive reagent (Weck’s reagent) to etch the

rapidly quenched Al–Si–Cu–Fe aluminum alloy (JIS ADC

10) and showed different color between pre-existing a-Al

grain and growth layer formed during quenching, therefore,

the growth layers could be excluded when measuring solid

fraction via quantitative metallography. However, they just

measured a very limited range of solid fraction (from 0 %

to about 20 %), and detailed explanation of the growth

layer’s different color was not given in their paper.

Weck’s reagent for aluminum alloys

In 1980s, Weck and Leistner developed a color etchant that

could detect segregations in cast aluminum alloys [22]. The

chemical compositions are 4 g KMnO4, 1 g NaOH, and

100 mL distilled water. In the examples given in their

research, microsegregation in dendritic a-Al grains of

as-cast Al–Si specimens is revealed by different colors

between central part and peripheral part. Suarez-Pena and

co-authors explained the mechanism of this etching process

in detail [23] by referring other studies concerning alumi-

num’s chemical activity [24, 25] and permanganate con-

version coating on 2024 Al alloy [26].

(1) Aluminum is a very active metal. The initial product

of the corrosion of aluminum in an aqueous environ-

ment is the hydroxide aluminum, namely Al(OH)3.

Then, Al(OH)3 creates a hydrated oxide (Al2O3�H2O).

(2) When the sample is immersed in Weck’s reagent,

based on an alkaline solution of potassium perman-

ganate (pH value is 13 approximately), the oxide

layer will be dissolved. Therefore, the fresh alumi-

num is exposed.

(3) The freshly exposed aluminum reacts with KMnO4

thus a colored film (MnO2) is formed on the surface:

Al þ MnO�4 þ 2 H2O! Al OHð Þ�4 þMnO2

For a branch of Al alloy dendrite, due to the different

electrochemical potentials of edge and center regions

resulting from segregation, a difference of color intensities

is caused [23].

In this present research, we tried to use Weck’s reagent to

visualize the inner microstructure of spheroidal Al grains

obtained by RAP process. Consequently, a more intuition-

istic understanding of microstructure evolution was realized.

We also precisely evaluated the solid fractions for a large

semi-solid temperature range with a detailed instruction of

the method.

Experimental

Material and plastic deformation

The material used in this research is A356 Direct-chill cast

aluminum alloy supplied in cylindrical shape (U105 9

450 mm) by Kyushu Mitsui Aluminium Co., Ltd. Table 1

lists the main compositions of the as-received material. Ti

and Sr were added to refine the a-Al phase and eutectic phase,

respectively.

In order to induce plastic deformation to the billets, they

were compressed without pre-heating. To avoid buckling,

as shown in Fig. 1, the billet was cut into small ones (U105 9 150 mm). Compression was carried out axially by a

500-ton compressing machine. The compressing rate is

Table 1 Main chemical compositions of the materials used in this

research (wt%)

Si Mg Fe Ti Sr Mn Al

6.9 0.39 0.10 0.14 0.025 \0.10 Bal.

6554 J Mater Sci (2012) 47:6553–6564

123

0.2 mm/s. The billet’s temperature was increased to about

44 �C when the target height (100 mm, 33 % reduction)

was reached.

Partial re-melting and water quenching

The set-up inside electric furnace is exhibited in Fig. 2a.

Two thermocouples (A and B) were installed to the fur-

nace, thermocouple A is used to inspect the air temperature

near the specimen, while thermocouple B is inserted to the

specimen, which was machined in the shape shown in

Fig. 2b. The shape of the specimen was specially designed

in order to make the inserting and withdrawing of the

specimen easier.

Two groups of partial re-melting experiments were

performed. For group 1, the temperature measured by

thermocouple A (define as TA) was always adjusted to

600 �C before putting the specimen in. Therefore, if the

holding time is long enough, the temperature of specimen

(defined as TB) will be equal to TA = 600 �C. Heating was

interrupted at various stages when TB is on the way to

600 �C, and specimens were withdrawn from the furnace

and quickly water quenched. TB was recorded during

heating and partial re-melting.

Figure 3 shows the heating curves of the seven speci-

mens used in group 1. All the specimens had a similar

approach to the target temperature (600 �C), so their

microstructures can represent the microstructure evolution

during heating and partial re-melting. Three regions are

labeled for the heating process. Specimens are named from

Specimen 1 to Specimen 7 corresponding to points from

left to right in Fig. 3. A specimen which was not com-

pressed (Specimen 6) was also heated to 600 �C to com-

pare with the specimen that was compressed before

heating.

For group 2, TA was set to different temperatures

ranging from 575 to 615 �C, with an interval of 5 �C. The

as-compressed specimens were heated to TA and then

isothermally held for 10 min (at 575, 580, 585, 590, and

Fig. 1 Picture of the A356 aluminum billets used in this research, the

as-received billet was cut before compression in order to avoid

buckling

Fig. 2 a Experimental set-up inside furnace. b The specially designed specimen used for the partial re-melting experiment

200

250

300

350

400

450

500

550

600

650

0 200 400 600 800 1000 1200 1400 1600

tem

pera

ture

, TB

/

time/s

specimens compressed before heatingspecimen not compressed before heating

A B C

Fig. 3 Heating curves of specimens in group 1

J Mater Sci (2012) 47:6553–6564 6555

123

595 �C) or 5 min (at 600, 605, 610, and 615 �C), after

which the specimens were quickly water quenched.

Microstructure observation

Radial cross-sections of as-received and as-compressed

billet were selected for microstructure observation. For

as-heated specimens, the cross-sections that use to be radial

in the billet were observed, except for the specimens which

changed their shape severely after water quenching. Sec-

tioned specimens were polished using standard metallo-

graphic techniques, finished using Struers OPS colloidal

silica. Specimens were etched by Weck’s reagent by

immersion for approximately 12 s (±2 s) at room tempera-

ture. Microstructure was observed by optical microscope

(OM) afterward.

Solid fraction evaluation

Etched by Weck’s reagent, specimens were subjected to

optical microstructure observation. In order to measure the

solid fractions precisely, growth layer formed during water

quenching was excluded when measuring the area of solid

phase. For some specimens, dendritic a-Al grains were also

formed during quenching (see ‘‘Microstructure of sphero-

idized grain etched by Weck’s reagent’’ section). In these

cases, it is easy to distinguish them because of their small

sizes compared to spheroidal Al grains. They were also

excluded when measuring the area of solid phase.

For those specimens which were isothermally held at

lower temperatures (with higher solid fraction), since the

distribution of spheroidal Al grains is homogeneous even

after being put into water, only one micrograph taken with

1009 magnification (with an area about 1.35 mm2) was

used for each specimen. On the contrary, specimens held at

higher temperature changed their shape severely which

leads to an inhomogeneous distribution of spheroidal Al

grains. For these specimens, a broader area is needed for

better representing the Al grains distribution. In these

cases, we examined at least 11 micrographs (1009 mag-

nification) taken thoroughly from the etched surface. In

order to avoid repeating measurement, micrographs were

carefully cut and merged into one big panoramic micro-

graph. The area representing the Al grain at semi-solid

state was painted red, since red is different from any color

displayed in the rest area. The solid faction is calculated by

the ratio of red pixels to the whole image’s pixels.

For comparison, solid fraction was also measured before

etching from as-polished specimens, with the exclusion of

fine a-Al grains formed during water quenching.

Results

Microstructure of as-received and as-compressed billet

Figure 4 shows both the microstructures observed after

polishing and etching by Weck’s reagent. As shown in

100µm

100µm 20µm

20µm

(a)

(c) (d)

(b) Fig. 4 Microstructures

of as-received specimen

and as-compressed specimen.

a As-received, as-polished;

b as-received, etched by Weck’s

reagent; c as-compressed,

as-polished; d as-compressed,

etched by Weck’s reagent

6556 J Mater Sci (2012) 47:6553–6564

123

Fig. 4a, the as-received billet has a typical dendritic

microstructure. After being etched by Weck’s reagent,

shown in Fig. 4b, the different colors inside dendrites can

be seen. Figure 4c shows the microstructure after com-

pression. a-Al grains are deformed and elongated. After

being etched by Weck’s reagent, seen in Fig. 4d, similar

information with what is shown in Fig. 4b can be obtained.

Internal fracture was not found after compression.

Etching the deformed specimen by Weck’s reagent is

very difficult. Usually only a limited area of the specimen

can be colored with a satisfying quality. It should be

pointed out that the result of etching varies with etching

time and environment temperature. Hence, comparison of

color between different specimens is meaningless.

Microstructure of spheroidized grain etched by Weck’s

reagent

Two representative micrographs of spheroidized grains

etched by Weck’s reagent are exhibited in Fig. 5. Two

specimens are both compressed prior to heating. The inner

microstructure of the spheroidal grains is clearly revealed.

In Fig. 5a, corresponding to the Specimen 5 in group 1

quenched when TB reached 581 �C, the original dendrites

(labeled 1), spheroidized grain at semi-solid state (labeled

2 and encircled by dotted curve) and the growth layer

formed during water quenching (labeled 3, light yellow

region) as well as the eutectic structure solidified from

intragranular liquid (entrapped liquid) and intergranular

liquid (labeled as 4 and 5, respectively) are able to be

visualized. The fact that region 3 is the growth layer

formed during quenching will be further evidenced in

‘‘Microstructure evolution of previously compressed

specimens during partial re-melting’’ section (Fig. 10). In

addition, it is notable that even the growth layer’s thickness

of one single grain is not constant, such as the grain located

in the center.

However, if specimen is quenched from a higher tem-

perature after long time isothermal holding, for instance,

600 �C for 5 min (specimen in group 2), as shown in

Fig. 5b, fine a-Al grains are formed among spheroidal

grains during quenching. Moreover, much thicker and

unstable growth of spheroidal Al grains occurs, and the

morphology of growth layer revealed by coloring is more

complex. With consideration of the fact that Al grains are

well spheroidized after being heated and partially re-melted

for a long time, the solid–liquid interface at semi-solid state

should be the smooth curves marked with arrows in

Fig. 5b. Also in Fig. 5b, due to the high temperature and

long holding time, diffusion in the solid phase leads to the

homogenization of inner microstructure. Therefore, origi-

nal dendrites cannot be visualized.

Comparison of semi-solid microstructure

between previously strain-induced specimen

and strain-free specimen

Semi-solid microstructures of Specimens 6 and 7 are

shown in Fig. 6. As indicated in Fig. 3, they have similar

heating curves. Figure 6a and b is corresponding to dif-

ferent regions of Specimen 6 which was strain free before

heating. While Fig. 6c and d is corresponding to different

regions of Specimen 7 which was compressed before

heating. It is obvious that either Specimen 6 or Specimen 7

has an inhomogeneous distribution of Al grains, and the

region containing fewer Al grains tend to form more Al

dendritic grains during quenching. Comparing Specimen 6

with Specimen 7, one observes that the specimen which

was compressed before heating forms finer and spheroidal

Al grains when partially re-melted. On the contrary,

20µm

1

4

2 3 5

(a)

40µm liquid-solid interface at semi-solid state

(b)

Spec. 5 in group 1

600 for 5 minutes

Fig. 5 Representative microstructure of spheroidal grains after water

quenching. a Specimen 5 in group 1, quenched when TB reached

581 �C; b a specimen in group 2, quenched after holding at 600 �C

for 5 min. Note the different magnifications of a and b

J Mater Sci (2012) 47:6553–6564 6557

123

spheroidization cannot be realized in strain-free specimen.

Moreover, less amount of liquid is entrapped in the pre-

viously compressed specimen.

Additionally, because of the high temperature and long

heating time (about 1400 s indicated in Fig. 3), solid state

diffusion enhances homogenization inside Al grains. That

is why in both figures, original dendrites cannot be seen as

clearly as in Fig. 5a.

Microstructure evolution of previously compressed

specimens during partial re-melting

Figure 7 shows the as-polished microstructure evolution of

previously compressed specimens during partial re-melting.

The six micrographs are corresponding to Specimens 1–5

and Specimen 7 in Fig. 3, respectively. One observes that

Al dendritic grains evolve gradually into spheroidal grains.

The Al dendrites coarsening behavior is similar to the

‘‘ripening’’ of dendrites in solidification processes. There-

fore, the terminology ‘‘ripening’’ will be used in this

research to describe the coarsening behavior of Al grains.

Microstructures with higher magnification are shown in

Fig. 8, focusing on the intergranular morphological change.

In accordance with Loue and Suery’s research [27], during

heating in Region A (indicated in Fig. 3), eutectic Si is

coarsened which is corresponding to Fig. 8a and b. Through

in situ optical microstructure observation, Sato et al. [14]

confirmed that such Si particle coarsening phenomenon

occurs during heating. After all the Si particles are coars-

ened, the eutectic structure begins to melt (Region B,

indicated in Fig. 3), and the formed liquid penetrates along

ripened grains’ boundaries and finally solidifies into fine

eutectic structure again during quenching, as shown in

arrowed areas of Fig. 8c and d. At this stage (Region B),

indicated in Fig. 3, the temperature is almost kept at a

constant value, which is defined as Teutectic. Temperature

increases again when all the eutectic melts and at this very

moment, as shown in Fig. 8e as well as Fig. 7e, the desired

fine spheroidal microstructure is obtained. The subsequent

heating until 600 �C (Region C, indicated in Fig. 3) leads to

coarsening of Al grains, as indicated by Fig. 7f.

Figure 9 shows microstructures of Specimens 3 and 4

etched by Weck’s reagent, which were quenched when

partial re-melting is in progress. One can see that the rip-

ening dendritic morphology takes up the most area of the

micrograph. A limited number of smaller spheroidal grains

are also observed as marked by red circles, which are

thought to be peripheral cross-sections of ripening den-

drites. It is also possible that they are incipient grains

formed due to recrystallization. Similar phenomenon has

Spec. 7 Spec. 7

Spec. 6

100µm

Spec. 6

100µm

100µm 100µm

(b)(a)

(d)(c)

Fig. 6 Comparison of partial-remelted microstructure between previously compressed specimen and strain-free specimen. a, b Specimen 6 in

Fig. 3; c, d Specimen 7 in Fig. 3

6558 J Mater Sci (2012) 47:6553–6564

123

been reported for 7075 aluminum alloy [9]. But obviously

in the present research, they don’t have a considerable

contribution to the refinement of Al grains.

Microstructures of Specimens 1–5 obtained after etching

by Weck’s reagent are shown in Fig. 10, the inner micro-

structures are visualized. The contrast of coarsened Si

particles with their background (appear light blue, indi-

cated by arrows) is reduced but they are still distinguish-

able. A new phenomenon was discovered that contributes

to the refinement of Al grains. As mentioned already in

‘‘Microstructure of as-received and as-compressed billet’’

section, internal fracture was not found in the as-com-

pressed specimen, as indicated by Fig. 4d. However, as

shown in Fig. 10a and b, seemingly, when the temperature

is increased, revealed by Weck’s reagent, ‘‘fragmentation’’

of dendrite’s branches is found. Figure 10c–e shows that

such fragmentation of dendrites produces more candidates

with simpler shape for spheroidal Al grains, thus decreases

the average grain size. At the same time, liquid penetrates

into the fractures and further spheroidized the Al grains.

This refining effect is somewhat similar to the refinement

of primary Al grains caused by stirring during rheocasting

described in [1, 2]. With more and more liquid formed, it

becomes easier for ripened grains to move, so it is more

difficult to find such phenomenon in specimens with lower

solid fraction. But we still luckily captured one micrograph

(a)

(f)(e)

(d) (c)

(b)

Spec. 1

Spec. 7 Spec. 5

Spec. 4 Spec. 3

Spec. 2

100µm

100µm

100µm

100µm

100µm

100µm

Fig. 7 Microstructure (as-polished, 2009 magnification) evolution during partial re-melting. a–e Specimen 1–5 in Fig. 3, respectively;

f Specimen 7

J Mater Sci (2012) 47:6553–6564 6559

123

(Fig. 10f) of Specimen 4 showing two Al grains totally

separated by liquid. Moreover, it is evident that the number

of those ‘‘fractures’’ decreases when the heat treatment

continues. This phenomenon will be further discussed in

‘‘Refinement and spheroidization of Al grains’’ section.

In ‘‘Microstructure of spheroidized grain etched by

Weck’s reagent’’ section, growth layer formed during

quenching was described as peripheral light yellow region

of spheroidal Al grains, which can be evidenced here. In

Fig. 10c and d, which are corresponding to the period of

the melting of eutectic structure shown in both Figs. 8b and

10b, such light yellow structure is formed in the area free

of unmelted Si particles. Therefore, it confirms that the

spheroidal Al grain grows toward liquid phase during

quenching.

Evaluation of solid fractions

According to Kliauga and Ferrante [28], the melting of

A356 aluminum alloy starts at a temperature lower than the

Al–Si eutectic melting temperature, resulting in Mg-rich

liquid. After that, melting of Al–Si eutectic will take place.

In the present research, we measured the solid fractions for

temperatures equals to Teutectic or higher, at which Al–Si

eutectic will be completely melted if holding time is long

enough.

Si

Penetrating liquid

Penetrating liquid

Spec. 1

Spec. 7 Spec. 5

Spec. 4 Spec. 3

Spec. 2

(a)

(f) (e)

(d) (c)

(b)

20µm

20µm 20µm

20µm 20µm

20µm

Fig. 8 Microstructure (as-

polished, 91000 magnification)

evolution during partial re-

melting. a–e Specimen 1–5 in

Fig. 3, respectively;

f Specimen 7

6560 J Mater Sci (2012) 47:6553–6564

123

As mentioned in ‘‘Solid fraction evaluation’’ section, by

excluding the growth layer and fine dendritic a-Al grains

formed during quenching, one can accurately measure the

solid fraction. Further discussion on the microstructure

changes during quenching is provided in ‘‘Grain growth

during water quenching’’ section. Evaluated solid fractions

are shown in Fig. 11. Results are obtained from both

etched and as-polished specimens. Deviation between the

two sets of results exists. Because in both cases, fine

dendritic a-Al grains are excluded from the solid phase,

such deviation should be caused by growth layer formed

during water quenching. The deviation for high tempera-

tures is more serious. This is not surprising because thicker

growth layer is formed at high temperature as mentioned in

‘‘Microstructure of spheroidized grain etched by Weck’s

reagent’’ section.

Discussion

Grain growth during water quenching

As previously mentioned in Microstructure of spheroidized

grain etched by Weck’s reagent’’ section, by comparing

Fig. 5a and b, one can see that quenching from higher

temperature leads to the unstable growth layer. This phe-

nomenon is in accordance with the research done by

Martinez et al. [29], which concludes that with a same

cooling rate, lower solid fraction (higher temperature)

tends to cause unstable growth.

However, it is also notable that thicker growth layers are

formed in those specimens that quenched from higher

temperature. Explanation for this phenomenon is as fol-

lows: for the alloys with a partition coefficient k \ 1, such

as Al–Si system, overlapping of solute fields at semi-solid

state decreases the undercooling ahead of solid–liquid

interface. The theoretical model for Al–4.5 wt %Cu

described in [29] indicates a sharp decrease of grain growth

velocity due to the decrease of undercooling caused by

solute fields overlapping. That means the growth of grain

during quenching is probably stopped by solute field

overlapping in our case, too. For specimens heated at

higher temperature, the lower solid fraction causes the

larger average distance between two grains compared with

it at lower temperature. Therefore, thicker growth layer can

be formed before the onset of solute fields overlapping.

This explanation can be further evidenced by the different

thickness of growth layer of the grain located at the center

of the Fig. 5a, where the thickness of growth layer changes

coherently with the change of distance to the neighbor

grains. However, such discipline is weakened in Fig. 5b,

due to the interference of newly formed a-Al grains.

Refinement and spheroidization of Al grains

Fragmentation of dendrites in solidification process by

agitation is thought to be an important factor that leads to a

refined and non-dendritic microstructure [1]. Margarido and

Robert also suggest that fractured dendrites are responsible

for grain refinement and spheroidization in RAP process

[30]. However, the micrograph (in black and white but the

etching method was not mentioned) of ‘‘fragmented den-

drite’’ given as the evidence in their paper is more like

cross-section cutting the parallel branches that are still

connected to the main trunk of a dendrite, which is called

pseudo-individual and isolated particles proposed by Nafisi

et al. [31]. Furthermore, the present research has shown that

in most cases, evidence of fractured dendrites can only be

found before the fragments are totally separated by liquid,

namely, before the boundary is completely wetted by liquid

Spec. 3

Spec. 4

(a)

(b)

100µm

100µm

Fig. 9 Microstructure of Specimens 3 and 4 etched by Weck’s

reagent. The small grains marked by circles are thought to be

peripheral cross-sections of ripened dendrites, but also possible that

they are newly formed grains due to recrystallization

J Mater Sci (2012) 47:6553–6564 6561

123

(Fig. 10). Therefore, micrographs in Fig. 10 can be regar-

ded as the direct evidence of the theory that deformed

dendrites fracturing during heating and partial re-melting

contributes to refinement and spheroidization of Al grains at

semi-solid state. As for the decreasing in the number of

‘‘fractures’’ mentioned in ‘‘Comparison of semi-solid

microstructure between previously strain-induced specimen

and strain-free specimen’’ section, two different explana-

tions can be made. One is that as partial re-melting con-

tinues, sintering occurs. The other one is that those

‘‘fractures’’ are actually migrating grain boundaries. The

dendrites fracturing cannot be observed in as-polished

condition, which indicates a higher possibility that the

‘‘fractures’’ are actually migrating grain boundaries.

Atkinson et al. [9] observed recrystallization in 7075

aluminum alloy characterized by new grains formed

between deformed and elongated grains. However, in our

research, as well as some other researches using A356

aluminum alloy [28, 32], such phenomenon was not

observed. Kliauga and Ferrante observed low-angle grain

boundaries (LAGB) in extruded (at 400 �C) A356 alumi-

num alloy using electron back scatter diffraction (EBSD)

Si

Si

Si

Si

Spec. 1 Spec. 2

Spec. 3 Spec. 4

Spec. 5 Spec. 4

20µm

20µm

20µm

20µm

20µm

20µm

(a)

(f) (e)

(d) (c)

(b)

Fig. 10 ‘‘Fractured’’ dendrites observed in a–e Specimen 1–5, respectively; f Specimen 4. Note that in f, the spheroidal grains are completely

separated

6562 J Mater Sci (2012) 47:6553–6564

123

technique [28]. When the specimen was heated and par-

tially re-melted to about 560 �C, a decrease in the number

of LAGB was found. A recently published research done

by Moradi et al. observed the HAGB formation inside

strain-induced grain during heating [32]. The material

processing is quite similar to ours. As cast A356 alloy was

equal channel angular pressed (ECAP) at room temperature

and heated, followed by quenching. The as-deformed

specimen as well as heated specimen was subjected to

EBSD observation. In the as-deformed specimen, subgrains

characterized by LAGBs were found in Al grains. While

the formation of HAGB was found inside coarsening Al

grains quenched from 190 and 450 �C. They also illus-

trated the formation mechanism of HAGB schematically.

From the discussion above, one can conclude that the

‘‘fractures’’ in Fig. 10 are probably HAGBs formed during

heating. When the dendrites are ripening, simultaneously

those HAGBs are migrating inside dendrites and tend to

link with each other to become longer, resulting in the

decrease in the number of them.

Evaluation of solid fraction using Weck’s reagent

The solid fraction evaluation using Weck’s reagent is an

improved image analysis. The most important step in this

method is the area selection. Since a good contrast between

original Al grain and growth layer is obtained after etching,

the error resulted from area selection is thought to be

suppressed.

As mentioned in ‘‘Comparison of semi-solid micro-

structure between previously strain-induced specimen and

strain-free specimen’’ section, less liquid is entrapped in

previously deformed specimen during partial re-melting.

Valer et al. [33] classified the entrapped liquid into two

types. One is not totally entrapped by ripening dendrite but

still connected with the intergranular liquid. During

quenching, it is solidified in a similar manner with the

intergranular liquid. The other one is completely entrapped

by ripening dendrite and solidified into a very fine micro-

structure (finer than intergranular eutectic). In our research,

most of the intragranular eutectic in RAP processed spec-

imens has a very fine microstructure which indicates the

completely entrapping of the liquid. In semi-solid state, the

volume fraction of intergranular liquid has a considerable

influence on the viscosity, but those entrapped droplets are

isolated from intergranular liquid, thus it is rational to treat

the entrapped liquid as a part of the spheroidal grains when

measuring the solid fraction.

Conclusions

A356 aluminum alloy was RAP processed to produce semi-

solid slurry with spheroidal Al grains. Optical microstruc-

ture observation was carried out using Weck’s reagent to

visualize the inner microstructure of ripening Al grains

with attention paid on the growth layer formation during

water quenching and ripening behavior of Al dendritic

grains.

The thickness of growth layer formed during quenching

is influenced by the solid fraction at semi-solid state. At

higher temperature with low-solid fraction, the larger

average distance between Al grains provides more space

for grain growth before the solute fields are overlapped,

thus producing thicker layer on Al grains.

Al dendrites were found ‘‘fractured’’ during heating

before liquid was formed, which contributes to the refine-

ment and spheroidization of Al grains. This phenomenon

was compared with EBSD observation results of other

researches [28, 32], which leads to the conclusion that the

‘‘fractures’’ are actually HAGBs. The numbers of those

HAGBs are decreasing during heating, due to the connec-

tion of them resulted from migration inside dendrites.

Solid fractions were measured by image analysis for the

entire semi-solid temperature range where Al–Si eutectic is

completely re-melted. The growth layer and a-Al grains

formed during quenching were excluded, thus accurate and

trustable results were obtained. The entrapped droplets

were regarded as a part of solid Al grains when evaluating

the solid fraction because they are isolated from inter-

granular liquid.

Acknowledgements We would like to thank Kyushu Mitsui

Aluminium Co., Ltd for supplying the billets and Nissan Motor Co.,

Ltd for compressing the billets. We also gratefully acknowledge

helpful discussions about Weck’s reagent with Prof. B. Suarez-Pena,

Prof. J. Asensio-Lozano in University of Oviedo, Spain, and Prof.

J. Wannasin in Prince of Songkla University, Thailand.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

570 580 590 600 610 620

solid

fra

ctio

n

temperature/

etched by Weck's reagent

as-polished

Fig. 11 Solid fractions measured before etching and after etching

J Mater Sci (2012) 47:6553–6564 6563

123

References

1. Flemings MC (1991) Metall Trans A 22A:957

2. Spencer DB, Mehrabian R, Flemings MC (1972) Metall Trans

3:1925

3. Kim H-S, Stone IC, Cantor B (2008) J Mater Sci 43:1292. doi:

10.1007/s10853-007-2151-3

4. Doutre D, Hay G, Wales P (2002) United States Patent No.

6428636

5. Wannasin J, Martinez RA, Flemings MC (2006) Scripta Mater

55:115

6. Young KP, Kyonka CP, Courtois JA (1983) United States Patent

No. 4415374

7. Kirkwood DH, Sellars CM, Elias-Boyed LG (1992) European

Patent No. 0305375 B1

8. Amir Bolouri, Mohammad Shahmiri, Kang CG (2012) J Mater

Sci 47:3544. doi:10.1007/s10853-011-6200-6

9. Atkinson HV, Burke K, Vaneetveld G (2008) Mater Sci Eng, A

490:266

10. Nakagawa K, Hoshino K, Otani T (2004) J Jpn Foundry Eng Soc

(in Japanese) 76:9

11. Amir Bolouri, Mohammad Shahmiri, Kang CG (2011) J Alloys

Compd 509:402

12. Jung BJ, Ichihata S, Sato T, Tezuka H, Kamio A, Saikawa S,

Nakai K (1997) J Jpn Inst Light Met (in Japanese) 47:626

13. Lapkowski W (1998) J Mater Proc Technol 80–81:463

14. Sato T, Ueno H, Tezuka H (2005) J Jpn Foundry Eng Soc

(in Japanese) 77:547

15. Flemings MC (1974) Solidification processing. McGraw-Hill,

New York

16. Chen SW, Huang CC (1996) Acta Mater 44:1955

17. Nafisi S, Emadi D, Ghomashchi R (2009) Mater Sci Eng, A

507:87

18. Tzimas E, Zavaliangos A (2000) J Mater Sci 35:5319. doi:10.1023/

A:1004890711322

19. Birol Y (2009) J Alloys Compd 486:173

20. Steinbach S, Ratke L (2004) Scripta Mater 50:1135

21. Wannasin J, Canyook R, Burapa R, Sikong L, Flemings MC

(2008) Scripta Mater 59:1091

22. Weck E, Leistner E (1986) Metallographic instructions for colour

etching by immersion, part iii: non-ferrous metals, cemented

carbides and ferrous metals, nickel-base and cobalt-base alloys.

DVS GmbH, Dusseldorf

23. Suarez-Pena B, Asensio-Lozano J, Vander-Voort GF (2010) Rev

Metal (in Spanish) 46:469

24. Godard HP, Jepson WP, Bothwell MR, Kane RL (1967) The

corrosion of light metals. John Wiley and Sons, New York

25. Pourbaix M (1966) Aluminium atlas of electrochemical equilib-

rium in aqueous solutions. Pergamon, Oxford

26. Kulinich SA, Akhtar AS, Wong PC, Wong KC, Mitchell KAR

(2007) Thin Solid Films 515:8386

27. Loue WR, Suery M (1995) Mater Sci Eng, A 203:1

28. Kliauga AM, Ferrante M (2005) Acta Mater 53:345

29. Martinez RA, Karma A, Flemings MC (2006) Metall Mater Trans

A 37A:2807

30. Margarido M, Robert MH (2003) J Proc Technol 133:149

31. Nafisi S, Ghomashchi R (2006) JOM 58:24

32. Moradi M, Nili-Ahmadabadi M, Poorganji B, Heidarian B, Parsa

MH, Furuhara T (2010) Mater Sci Eng, A 527:4113

33. Valer J, Meneses P, Saint-Antonin F, Suery M (1999) Mater Sci

Eng, A 272:342

6564 J Mater Sci (2012) 47:6553–6564

123