Materials Transactions, Vol. 43, No. 7 (2002) pp.

1711 to 1715

2002
c The Japan Institute of Metals

A Mechanism of Porosity Distribution in A356 Aluminum Alloy Castings

Kun-Dar Li and Edward Chang ∗
Department of Materials Science and Engineering, National Cheng Kung University, Tainan, Taiwan 701, R.O. China

The problem of porosity and shrinkage defects in metal casting is complex. Over the years there have been debates on the mechanisms
responsible for their formation. In this study A356 aluminum alloy with different hydrogen contents in the melts were cast in a permanent
mold and the porosity content and thermal parameters were measured. A simple mechanism was proposed to explain the porosity distribution
in the casting. In the mechanism both the roles of interdendritic feeding resistance by Darcy’s law and kinetic hydrogen diffusion into the pore
are involved. By differentiating one factor in the model, and comparing the prediction with the experimental data, the study suggests that both
factors should be taken into consideration to satisfactorily explain the porosity distribution in the castings.

(Received March 22, 2002; Accepted May 16, 2002)

Keywords: aluminum, porosity, Darcy’s law, hydrogen, thermal parameters

1. Introduction tion using the equilibrium lever rule:18)

[Hl ] = [H0 ]/[ f l + (1 − f l )kH ] (2)
The problem of porosity and shrinkage defects in metal
casting is complex. Despite many efforts to elucidate the where [Hl ] is the hydrogen content in the liquid, [H0 ] is
problem, the mechanisms of the phenomenon remain un- the initial hydrogen content in liquid (ml/100 g Al), f l is the
solved.1–8) Some researchers believe the problem is caused by weight fraction of liquid, and kH is the equilibrium partition
the resistance to the interdendritic liquid flowing from riser to ratio of hydrogen between solid and liquid. In eq. (2) a com-
feed the solidifying liquid at the advancing liquid/solid inter- plete diffusion of hydrogen from liquid into the gas pores has
face. This resistance causes a local pressure drop of the liquid been assumed. The predicted porosity content by thermody-
by Darcy’s law.9) One advocator of the theory was Lecomte- namics was considered too high since the time involved in
Beckers10) who, like others,11–13) described the magnitude of diffusion of hydrogen to the pores was neglected.14)
mushy zone pressure drop index (∆P) in the interdendritic The purpose of the paper is to present the important fac-
liquid in directional solidification or equiaxed dendritic cast- tor of time in the redistribution of hydrogen during the solid-
ing by the following expression: ification of aluminum alloys for the mechanism of porosity


 formation. The mechanism intends to explain the porosity
24π µβ  n̂τ 3 ∆T 2 d f s
∆P = (1) distribution in A356 aluminum alloy castings as affected by
ρl g G dt hydrogen contents.
where µ is the viscosity of liquid, β  = β/(1 − β) (β is the
solidification contraction), n̂ is the number of interdendritic 2. Experimental Procedure
channels, τ is the tortuosity, ρl is the density of liquid, g is
the acceleration of gravity, ∆T is the solidification range, G is Figure 1 shows the JIS SKD61 permanent mold for cast-
the thermal gradient, and d f s /dt is the average solidification ing a φ25 mm × L150 mm sample with φ50 mm × L50 mm
rate ( f s is the solid fraction and t is the time). riser. The materials of A356 ingot with 0.021 mass% Sr were
The pressure drop index based on Darcy’s law within the melted in a graphite crucible placed inside an electric resis-
interdendritic liquid was employed to estimate the formation tance furnace. For varying the hydrogen content, the melt was
of porosity in aluminum alloys11, 13) and nickel-base super- either untreated, degassed with moisture-free Ar, or using wet
alloy castings.10) Equation (1) suggests that internal sound- A356 alloy for the melt. The hydrogen contents in the melts
ness in a casting is favored by a short solidification range; were controlled at 0.10, 0.15, and 0.20 ± 0.005 ml/100 g Al,
a low dendrite number density, tortuosity and solidification detected by an Alscan hydrogen quantitative analyzer. The
rate; and a high fluidity and thermal gradient. melts were poured at 720◦ C. Thermal measurements were
For some directional solidified castings, however, Darcy’s used to obtain the thermal variables, e.g. thermal gradient (G),
law does not appear to be a controlling factor in porosity for- solidus velocity (Vs ) and local solidification time (tf ), in the
mation.14–16) The pressure drop associated with liquid flow is casting. The porosity content was calculated in accordance
much smaller than the prevailing atmospheric pressure and with the formula (Ds − D)/Ds × 100% where Ds is the theo-
could not be responsible for the shrinkage porosity. Poirier et retical density of the standard specimen and D is the density
al.17) ignored the effect of interdendritic fluid flow on poros- of the test pieces measured by the Archimedes’ method.
ity. Formation of porosity in Al–Cu alloys was treated ther-
modynamically by analyzing the equilibrium partition of hy- 3. Experimental Results
drogen content at the liquid/solid interface during solidifica-
The porosity distributions in A356 castings vary with hy-
∗ Corresponding
drogen contents and locations in the castings (Fig. 2). The
author: E-mail address: [email protected]
1712 K.-D. Li and E. Chang

Fig. 3 Variation of local solidification time, thermal gradient and solidus

velocity with distance from the free end of the casting.

time will be available for the liquid metal to feed a solidifying

site.22, 23) Both are adverse effects. It appears that the correla-
Fig. 1 Schematic illustration of the experimental setup.
tion between porosity content (Fig. 2) and thermal parameters
(Fig. 3) along the casting length is not satisfactory.

4. Discussion

In this study, we propose that the volume content of poros-

ity (Vp (%)) at a location in the casting can be calculated in
accordance with the ideal gas law:
Vp (%) = K · [Hp ]/Pg (3)
where K is a proportional constant in vol%·atm/ml/100 g Al
and [Hp ] is the hydrogen content (ml/100 g Al) diffused into
the pore at a certain location in the casting. When pore forms
in the interdendrites, the gas pressure in the pore (Pg , atm)
can be represented as:24)
4σ
Pg = P0 + ρl gh + − 24π τ 3 µβ 
C · tf0.286

1.572 

∆T 0.428 Vs α
× ln (4)
C2 G 0.428 f l (x)
Fig. 2 Variation of porosity content with distance from the free end of the where P0 is the atmospheric pressure, ρl gh is the hydro-
castings with different hydrogen contents. static pressure of riser, σ is the surface tension of inter-
dendritic liquid, C = 35.1 µm/min0.286 ,24) ∆T = 38 +
18.655e−0.244tf ◦ C,24) α is the liquid volume fraction at the end
porosity content is lowest near the free end, and it gradually of mass feeding, f l (x) is the liquid volume fraction, and x is
increases to a maximum near the mid-length of the casting. the distance from the liquid/solid interface.13) The third term
Further in the direction toward riser, the porosity content low- on the right side of eq. (4) is the surface tension effect of the
ers slightly and then increases again gradually. The general liquid to overcome when a pore forms, and the fourth term on
trend of the results has been agreed by other studies.19, 20) the right is the pressure drop from the interdendritic fluid flow
The variation of thermal parameters with location in the based on Darcy’s law.
castings is exhibited in Fig. 3. The figure shows the local It has been observed and recognized that pores form in a so-
solidification time increases toward the riser, which indicates lidifying alloy by the process of nucleation and growth.14, 15)
prevailing of directional solidification. The thermal gradient We further assume that this kinetic process is one of the more
is highest at the free end and it reduces gradually toward the general phase transformations in materials.25) At initiation of
riser. The solidus velocity, as also indicated in the figure, in- solidification, i.e. time is small, both the numbers of pore and
creases slightly and then decreases toward the riser. A lower the pore size are small, and hence the kinetics of reaction
thermal gradient will result in a longer interdendritic feed- should be slow.25) It is well known that as time increases the
ing length,21) while a high solidus velocity means that less number of pore and pore size increases, which warrants an
 A Mechanism of Porosity Distribution in A356 Aluminum Alloy Castings 1713

increasing kinetic reaction. If very sufficient time is allowed,

all the hydrogen content in liquid except the solubility of hy-
drogen in solid (C s0 ) should be diffused into the pore ther-
modynamically. That is the supersaturated dissolved hydro-
gen in the melt becomes exhausted as mentioned by Poirier et
al.17) Under the limiting condition, hence [Hp ] will be equal
to ([H0 ]−C s0 ) as tf approaches infinity. For otherwise the time
of diffusion is limited, the fraction of hydrogen ( f ), defined
as [Hp ]/([H0 ] − C s0 ), that can be diffused into the pore should
be determined mathematically as:

n 
[Hp ] tf
f = = 1 − exp − (5)
([H0 ] − C s )
0 tτ
The concept and mathematics of the above formulation,
though simple, is consistent with the rigorous Avrami’s equa-
tion.26) The argument is on the basis that the kinetics of nucle-
ation and growth of the pore in the solidifying aluminum al-
loy is merely another kinetic phase precipitation in alloys and Fig. 4 The viscosity of A356 alloy as a function of temperature. Data from
materials.27) Where in eq. (5) n is an exponential constant, tf Ref. 29).
and tτ are the local solidification time and the relaxation time,
respectively. n and tτ are associated with the characteristics
of nucleation and growth of pore. The relaxation time (tτ )
is a measure of how fast the hydrogen diffuses into the pore;
when tf = tτ , the hydrogen should transport a quantity equal
to (1 − 1/e) of ([H0 ] − C s0 ), or 0.63 ([H0 ] − C s0 ). Larger
values of tτ would mean a slow kinetic of hydrogen diffusion
into the pore. If circumstances allow a long period of diffu-
sion, the kinetics of the hydrogen precipitation would level off
due to the exhaustion of hydrogen being diffused into the gas
pore, and the hydrogen content inside the pore ([Hp ]) should
approachs ([H0 ] − C s0 ).
Association of eq. (5) with eq. (3), the volume percent of
porosity can be described as:

n 
Vp (%)Pg tf
= 1 − exp − (6)
K · ([H0 ] − C s0 ) tτ
The physical properties of A356 alloy are required in eval-
uation of eq. (6). Emadi et al. studied the effect of Sr
Fig. 5 Relation between the index of porosity fraction (F) and local solid-
modification on surface tension of A356 alloy, and found ification time.
the addition of Sr to A356 alloy decreased the surface ten-
sion of the liquid by about 19 pct to 0.64 N/m.28) The vis-
cosity of A356 melt at eutectic temperature (565◦ C) is re- drogen contents, while the quantities of tτ are 11.42, 13.35
gressed as 4.7 × 10−3 kgm−1 s−1 (Fig. 4) from the data of and 24.28 s for the hydrogen contents at 0.20, 0.15 and
Ref. 29). The solubility of hydrogen in the solid (C s0 ) was 0.10 ml/100 g Al, respectively. The results imply that the
proposed as 0.021 ml/100 g Al,24) which is in general agree- kinetics of nucleation and growth of pore formation occurs
ment with 0.0267 and 0.025 ml/100 g Al by other studies.17, 30) more rapidly in the case of the melt containing more dissolved
In addition, K in eq. (3) is a regression constant about gas, which corresponds to a small relaxation time. This find-
12 vol%·atm/ml/100 g Al.24) ing has been supported by Fang and Granger31) and Sigworth
For simplicity, the term on the left of eq. (6) is defined as an and Wang.14) In eq. (4), it is shown that Pg varies with the
index of porosity fraction (F). The relationship between the location in a casting, the range of which being 1.35–1.58 atm
index of porosity fraction (F) and the local solidification time in the study. If taking an average of Pg ∼ = 1.43 into eq. (6),
is shown in Fig. 5. From the figure, it is obvious that differ- then Vp (%) can be depicted as a function of tf . The plot of
ent initial hydrogen contents show the different kinetic rates the predicted variation of the volume percent of porosity con-
of nucleation and growth of pore, but all of them appear to tent with the local solidification time at different hydrogen
follow the Avrami’s equation, which might serve to indirectly contents per eq. (6) with the averaged Pg , together with the
support the derivation. experimental data, are illustrated in Fig. 7. From the figure,
In eq. (6), the quantities of n and tτ can be determined semi- the kinetic model appears to fit the experimental results ex-
empirically from the fitting curve of the plot of ln(− ln(1 − cept the deviations of porosity contents between prediction
F)) vs. ln tf as shown in Fig. 6. According to the result, and measurements corresponding to the local solidification
the values of n are regressed to 0.56 for three different hy- time of about 60 s. The higher experimental data of poros-
1714 K.-D. Li and E. Chang

Fig. 7 The measured (symbols) and predicted (curves) porosity content as

a function of local solidification time in A356 alloy castings with different
hydrogen contents. In drawing the predicted curves of porosity content, an
average of Pg ∼= 1.43 in everywhere of the castings was assumed.

ity content than the prediction arises because Pg at the above

location of the castings should have been about 1.35 atm in-
stead of 1.43 atm. An examination of Fig. 3 will find that
the location that solidifies at about 60 s in the castings corre-
sponds to a maximum solidus velocity. Thus the difficulties of
interdendritic fluid feeding caused by this thermal parameter
appears to result in the interdendritic fluid pressure drop based
on Darcy’s law as well as the maxima as shown in Fig. 2.
In the predicted curves of Fig. 7 the variation of Pg has
been purposely neglected to differentiate the sole effect of dis-
solved gas redistribution without the interference of the factor
of interdendritic feeding resistance. The aim was to elucidate
a complex long debates among the two schools of thinking
in this field of researches in the last 50 years:14) which fac-
tors of the interdendritic feeding resistance by Darcy’s law9)
or the solute redistribution that are responsible for the shrink-
age porosity.17) For the advocators who favor the role of hy-
drogen redistribution, the mechanism of which is unsloved.17)
Thermodynamic model predicted a too high porosity content,
since the kinetics of diffusion of dissolved gas has been ne-
glected.14) Our present mechanism involving the roles of in-
terdendritic feeding and kinetics of hydrogen redistribution
appears to support that both mechanisms are exercising. If
allowing variation of Pg per (4) to vary in eq. (6), the pre-
dicted volume percent of porosity content along the length of
the castings with different hydrogen contents, together with
the experimental data, are displayed in Fig. 8. In this figure,
both of the interdendritic feeding resistance and the kinetics
of hydrogen redistribution are taken into consideration to de-
scribe the mechanism of porosity formation. It appears that
Fig. 6 Relation between ln(− ln(1 − F)) and ln tf in A356 alloy castings
a satisfactory quantitative porosity content at any location of
with different hydrogen contents: (a) 0.2 ml/100 g Al, (b) 0.15 ml/100 g the castings with different hydrogen contents can be predicted
Al, and (c) 0.1 ml/100 g Al. by the proposed model.
 A Mechanism of Porosity Distribution in A356 Aluminum Alloy Castings 1715

differentiating one factor in the proposed model, we are able

to prove that both roles mentioned above should exert an ef-
fect in order to satisfactorily explain the porosity distribution
in A356 aluminum alloy castings.

REFERENCES

1) E. Niyama, T. Unchida, M. Morikawa and S. Saito: A Method of Shrink-

age Prediction and Its Application to Steel Casting Practice, 49th
International Foundry Congress, April (1982) p. 1.
2) S. Minakawa, I. V. Samarasekera and F. Weninberg: Metall. Trans. B
16B (1985) 823.
3) J. A. Spittle, M. Almeshhedani and S. G. Brown: Cast Metal 7 [1]
(1994) 51.
4) R. A. Entwistle, J. E. Gruzleski and P. M. Thomas: Solidification and
Casting of Metals, Proc. International Conference on Solidification
(The Metal Society, 1979) p. 345.
5) H. Huang and J. T. Berry: AFS Trans. 101 (1994) 669.
6) G. V. Kutumba Rao and V. Panchanathan: AFS Trans. 81 (1973) 110.
7) Y. W. Lee, E. Chang and C. F. Chieu: Metall. Trans. B 21 B (1990) 715.
8) J. A. Taylor: Cast Metals 8 [4] (1995) 225.
Fig. 8 Variation of measured (symbols) and predicted (curves) porosity 9) T. S. Piwanka and M. C. Flemings: Trans. Met. Soc. AIME 236 (1966)
content with distance from the free end of the castings with different hy- 1157.
drogen contents. 10) J. Lecomte-Beckers: Metall. Trans. A 19A (1988) 2341.
11) H. Shahani: Scandinavian Journal of Metallurgy 14 (1985) 306.
12) P. S. Mohanty, F. H. Samuel and J. E. Gruzleski: Metall. Trans. A 24A
5. Summary (1993) 1845.
13) S. T. Kao, E. Chang and Y. W. Lee: Mater. Trans., JIM 35 (1994) 632.
14) G. K. Sigworth and C. Wang: Metall. Trans. B 24B (1993) 349.
The problem of porosity and shrinkage in metal casting is 15) P. D. Lee and J. D. Hunt: Acta Mater. 45 [10] (1997) 4155.
complex since many thermal parameters as well as physical 16) R. C. Atwood, S. Sridhar, W. Zhang and P. D. Lee: Acta Mater. 48
(2000) 405.
properties are involved. Further complication of the problem
17) D. R. Poirier, K. Yeum and A. L. Maples: Metall. Trans. A 18A (1987)
is due to the behavior of fluid flow in the interdendritic space 1979.
and the rejection of dissolved gas into the pore kinetically 18) M. C. Flemings: Solidification Processing, International Edition
during solidification. Over the years, two main schools of (McGraw-Hill Book Co., 1974).
19) S. N. Tiwari, A. K. Gupta and S. L. Malhotra: The British Foundryman
thinking in this field have been in debate: which factors of the 79 [3] (1986) 129.
interdendritic feeding resistance by Darcy’s law or the solute 20) S. Saikawa, K. Nakai, Y. Sugiura and A. Kamio: Mater. Trans., JIM 40
redistribution that are responsible for the shrinkage porosity. (1999) 57.
In this study the A356 aluminum alloy with three hydro- 21) W. H. Johnson and K. J. Kaur: AFS Trans. 67 (1959) 532.
22) V. de L. Davies: AFS Cast Metals Res. J. 11 (1975) 33.
gen contents were cast in a permanent mold and the poros- 23) V. de L. Davies: Metal Society (1979) 357.
ity contents and thermal parameters in different locations of 24) K. D. Li, M. C. Cheng and E. Chang: AFS Trans. 109 (2001) 301.
the casting were measured. A simple mechanism for kinetic 25) J. W. Christian: The Theory of Transformation in Metals and Alloys
diffusion of hydrogen into the pore disregarding the thermo- Part I Equilibrium and General Kinetic Theory, 2nd Ed., (Pergamon
Press, Oxford, 1975).
dynamic model was proposed. The concept and mathematics 26) M. Avrami: J. Chem. Phys. 7 (1930) 1103.
involved in the mechanism is assumed to obey the Avrami’s 27) F. S. Ham: J. Phys. Chem. Solids 6 (1958) 335.
equation in the more general phase transformations. The re- 28) D. Emadi, J. E. Gruzleski and J. E. Toguri: Metall. Trans. B 24B (1993)
sults of regression of n and tτ imply that the kinetics of nucle- 1055.
29) Procast Build-in Database, Procast The Professional Casting Simula-
ation and growth of pore formation occurs more rapidly in the tion System (UES, Inc., 1988).
case of the melt containing more dissolved gas, which corre- 30) J. E. Gruzleski and B. M. Closset: The Treatment of Liquid Aluminum-
sponds to a small relaxation time. This finding has been sup- Silicon Alloys, (The American Foundrymen’s Society, Inc., 1990).
31) Q. T. Fang and D. A. Granger: AFS Trans. 97 (1989) 989.
ported by Fang and Granger31) and Sigworth and Wang.14) By