Communication

An Evaluation of Primary Dendrite

Trunk Diameters in Directionally
Solidified Al-Si Alloys
S.N. TEWARI, R.N. GRUGEL, and D.R. POIRIER
The primary dendrite trunk diameters of Al-Si alloys
that were directionally solidied over a range of processing conditions have been evaluated. The empirical
data are analyzed with a model that is an extension of
one used to describe the ripening of dendrite arms. The
analysis, based primarily on an assessment of secondary
dendrite arm dissolution in the mushy zone, ts well
with the experimental data. It is suggested that the primary dendrite trunk diameter is a useful metric that
correlates well with the actual solidication processing
parameters, and complements the conventionally used
primary, secondary, and tertiary arm spacings.
DOI: 10.1007/s11661-014-2438-3
 The Minerals, Metals & Materials Society and ASM
International 2014

Typically, the microstructure of cast alloys consists of

an assemblage of primary dendrites with secondary and,
possibly, tertiary arms; the volume within this dendritic
network is often lled in by a terminal eutectic reaction
product. During steady-state directional solidication
(DS) these dendritic features establish unique spacing
relationships as a function of the alloy composition (Co),
the imposed growth velocity (V), and the temperature
gradient in the liquid ahead of the solidliquid interface
(GL). Furthermore, predictable changes in the spacings
result as Co, V, and GL are varied. Consequently,
innumerable experimental and theoretical papers have
reported on these spacing-processing relationships, particularly with regard to the primary dendrite arm
spacing, (k1) (e.g. see References 1 through 3.).
Despite their obvious appeal, these solidication
processing-dendritic spacing relationships must be
viewed with some caution. The primary dendrite spacing
takes considerable time to establish itself once solidication is initiated,[4] which can then be subject to
interactions from its neighbors.[5] Some bias is introduced from deciding between active and stagnant

S.N. TEWARI, Professor, is with the Department of Chemical &

Biomedical Engineering, Cleveland State University, Cleveland, OH
44115. R.N. GRUGEL, Metallurgical Engineer, is with the NASAMarshall Space Flight Center, Huntsville, AL 35812. Contact e-mail:
Richard.n.grugel@nasa.gov D.R. POIRIER, Professor, is with the
Department of Materials Science and Engineering, University of
Arizona, Tucson, AZ 85721.
Manuscript submitted March 31, 2014.
Article published online July 16, 2014
4758VOLUME 45A, OCTOBER 2014

branches when measuring secondary (k2) and tertiary

dendrite arm spacings (k3).[6] Finally, if there is a change
in one of the processing parameters then the corresponding adjustment in the primary spacing may be
considerably delayed until an appropriate modication,
such as a tertiary arm managing to advance enough to
become a primary arm, can occur.[7,8]
Another microstructural component of a primary
dendrite arm is its very trunk. Measurements of dendrite
trunk diameters from samples grown under DS also
show systematic changes as a function of Co, V, and
GL.[9,10] Such changes were also measured through the
mushy zone where the trunk diameter continually
increased until being frozen in by the terminal eutectic
reaction. Here the trunk diameter was found to increase
at a rapid rate from just behind the dendrite tip until the
(predicted) solute elds from neighboring dendrites
overlapped [5] after which the rate decreased considerably. This observation infers that the trunk diameter is
related to the often studied dendrite tip radius.[1116]
When coupled with observations that the dendrite tip
responds immediately to changes in processing parameters,[7,9,13] it follows that the dendrite trunk reacts
accordingly. The implication is that the trunk diameter
is a consequence of the tip behavior; it does not need a
time-dependent mechanism to adjust to processing
changes,[17] and it represents a microstructure-processing relationship better than dendrite arm spacing
measurements.
A cross-sectional micrograph on which the trunk
diameters are marked is seen in Figure 1.[9]
The diameter is determined by averaging the lengths
of a pair of the black diagonals, which are seen to be
very uniform and substantiated by the standard deviations in later gures. The trunk diameters also indicate
center positions of the dendrites from which the primary
dendrite arm spacing, k1, is determined. Figure 1 also
shows signicant variance in the distance between
nearest neighbor dendrites, a fact that complicates
spacing determinations.[18] Ambiguity in determining
the tertiary dendrite arm spacing, k3, is also obvious;
measuring secondary arm spacings is even more subjective, as can be seen in Figure 2. Figure 2 is a
longitudinal micrograph showing primary succinonitrile
dendrites growing at a constant rate through a constant
temperature gradient; it also depicts the regions later
discussed in detail. Here the primary dendrite trunk
diameter evolves from an initial size (/0) just past the tip
through the mushy zone to a nal diameter / after being
frozen in by a terminal monotectic reaction. Extension
of the parabolic tip shape leads to the initial trunk
diameter, /0. The nal diameter, /t, evolves in the
mushy zone due to dissolution of secondary arms, which
has positive curvature; that material is re-deposited on
the trunk surface where there is a negative curvature.
To quantify the dendrite trunk measurements, an
analytical analysis based primarily on Kirkwoods[20]
assessment of secondary dendrite arm dissolution was
undertaken. Figure 3 is a schematic dendrite section
where / represents the dendrite trunk diameter, L is the
length of an active secondary dendrite arm, h represents
the secondary arm spacing, r is the radius of the
METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 3Schematic representation of the primary dendrite trunk

diameter (denoted by /) and secondary arms.

local composition, and k is the solute partition coecient. Assuming that the material from the four dissolving secondary arms around the dendrite periphery is
re-deposited on the trunk surface over the height, h,
gives
p/h d/
dL
4pr2
2 dt
dt
Fig. 1Transverse micrograph of a directionally solidied Al6 wt pct Si alloy with the primary dendrite trunk diameters indicated. V=301 lm s1, GL=150 K cm1[9].

2

Assuming that h~/, and substituting Eq. [1] into Eq.

[2] yields,
/2

d/
Dl C
32
dt
ml CL 1  k

3

and integrating with /0 at time t=0 results in

/3 /30 96

Dl C
t:
ml CL 1  k

4

Equation [4] now represents the primary trunk diameter in terms of its initial diameter, /0.
Noting that the composition, CL, along the length
varies from the dendrite tip (approximately the bulk
composition, Co) to the composition at the eutectic, CE,
another representation of the coarsening of the trunk
diameter can be obtained. Here CL= Co+VGt/ml,
where V is growth speed and G is the thermal gradient in
the mushy zone. Substituting into Eq. [3] results in
/2

Fig. 2Longitudinal micrograph showing primary succinonitrile

dendrites growing from a succinonitrile-9 wt pct water mixture at a
constant rate through a constant temperature gradient.[19] Some terminology and the regions of interest are indicated.

secondary arm tip, and CL is the composition of the

liquid at the tip region.
Following Kirkwood,[20] the dissolution rate of a side
arm is:
dL
4Dl C

dt ml CL 1  kr2

1

Here Dl is the liquid diusion coecient, C is the

capillarity constant, ml is the liquidus slope, CL is the
METALLURGICAL AND MATERIALS TRANSACTIONS A

d/
Dl C
32
:
dt

ml 1  kCo 1 mVGt
l Co

5

Integrating Eq. [5] gives:

/3 96



Dl C
VGt
ln 1
:
VG1  k
ml Co

6

Esaka[21] conducted controlled DS experiments utilizing succinonitrile-acetone alloys to directly observe

the steady-state dendritic growth as a function of several
imposed growth velocities and temperature gradients.
He noted that the trunk diameter rapidly increased
behind the tip which, after formation of a number of
secondary branches, abruptly transitioned to an almost
constant diameter; this is denoted as the initial trunk
diameter, /0. Figure 4 plots his experimental data as a
ratio of the initial trunk diameter, /0, to the tip radius,
VOLUME 45A, OCTOBER 20144759

Fig. 4Plot of the initial trunk diameter to the tip radius as a function of material properties and phase diagram parameters as determined from Esakas[21] directional solidication experiments with
succinonitrile-acetone alloys.

Fig. 5Primary dendrite trunk diameters in solidied, hypoeutectic,

Al-Si alloys as a function of the imposed growth velocity for a temperature gradient of 88.9 K cm1.

rt, as a function of material properties and phase

diagram parameters.
Consequently /0/rt is determined to be 6.591.3 with
a 95 pct condence level. Assuming that the parabolic
shape of the dendrite tip is maintained until time t0,
which would correspond to the initial dendrite trunk
diameter /0, geometrical evaluation gives t0 = 22rt/V.
Eq. [6] then becomes:
9
8
VGt =
<
1

ml Co
Dl C
 /30 :
ln 
7
/3 96
VG1  k : 1 VGt0 ;
ml Co

The steady-state value of the primary dendrite trunk

diameter can now be obtained by substituting the alloy
freezing time, t = ml(CE  Co)/VG, in the above relationship. In order to obtain the time, t0, we use the tip
radius, rt, predicted from the Hunt-Lu numerical
model.[16]
Figures 5 and 6 are plots of previously reported[9]
mean primary dendrite trunk diameters from directionally solidied hypoeutectic aluminum-silicon alloys as a
function of the imposed growth velocity; the error bars
denote one standard deviation. The curves in Figure 5
were calculated from Eq. [7] and correspond, as listed, to
a given composition noting that the temperature gradient, G, was determined through the mushy zone as
88.9 K cm1. The data in Figure 6 represent the mean
trunk diameters as a function of growth velocity for two
sets of Al-6 wt pct Si alloys solidied through two
dierent temperature gradients, 88.9 and 29.6 K cm1;
the curves were again calculated using Eq. [7].
The model calculations compare well with the experimental data. Very good agreement between the model
and data is seen in Figure 6 for the 6 wt pct Si alloys
solidied through two dierent temperature gradients.
Regarding Figure 5, there is also good t between the
data and the model for the 6, 8, 10, and 12 wt pct Si
alloys. The calculation, however, falls below the experimental data for the 4 wt pct Si alloys; note here that the
4760VOLUME 45A, OCTOBER 2014

Fig. 6Plot of the primary dendrite trunk diameters as a function

of growth velocity for Al-6 wt pct Si alloys solidied through temperature gradients of 88.9 and 29.6 K cm1.

dierences between the experiments and the model

decrease as the growth velocity increases. The extent of
scatter, as indicated by the error-band width, also
decreases as the growth velocity increases.
Depending on the composition, imposed temperature
gradient, and growth rate either a steady-state cellular
or dendritic structure establishes itself. Typical processing parameters that promote cellular growth are low
alloy compositions, low growth rates, and high temperature gradients. The interfaces of the 4 wt pct Si alloys,
while not strictly cellular, are in the processing range
where only rudimentary, and stagnant, secondary dendrite arms develop.[13] A possible exception is the
4 wt pct alloy that was solidied at 301 lm s1, in
which case, the tip is small enough to initiate some side
branches behind it that lead to rapid trunk growth
before neighboring solute elds are encountered. The
present model is valid for a dendrite morphology having
secondary branches. It utilizes the tip radius predicted
by Hunt-Lu model[16] for dendrites. For the same
growth conditions a larger value of tip radius would
METALLURGICAL AND MATERIALS TRANSACTIONS A

be predicted by the Hunt-Lu model for cells. Somboonsuk et al. [22] have shown that the tip radius increases as
the growth velocity decreases and the cell-dendrite
transition is approached. Given that the trunk diameter
calculations presented here are based on Kirkwoods
model, deviation should be expected in the absence of
dissolving secondary arms. Using a larger initial tip
radius would yield a larger /0 and bring the calculated
trunk diameters for 4 wt pct Si alloys closer to the
experimental measurements. Since the microstructure
tends to be more cellular and less branched at low
growth speeds, often containing both morphologies in
the same sample cross-section, it would also result in an
increased scatter in the measured trunk diameter, as seen
in Figures 5 and 6. In addition, similar to hypoeutectic
Al-Cu alloys,[23] hypoeutectic Al-Si alloys also have a
tendency to develop convection-induced dendrite steepling[24] during DS, especially at low thermal gradient
and low growth speeds. This results in a non-uniform
distribution of primary dendrites across the sample
cross-section. This would also cause increased scatter in
the measured trunk diameter at low growth speeds and
low thermal gradient seen in Figures 5 and 6. In
summary the models agreement with the data validates
using the trunk diameter as a representative microstructural measure of the processing parameters during
controlled DS, particularly when the growth is clearly
dendritic.

The authors appreciate and acknowledge the support provided by NASA Grant NNX08AN49G, and
the NASA Microgravity Materials Science program.
One of the authors (RNG) wishes to acknowledge
Professor W. Kurz of the Ecole Polytechnique Federal

METALLURGICAL AND MATERIALS TRANSACTIONS A

de Lausanne for the time in his laboratory which generated the data shown in Figures 5 and 6.
REFERENCES
1. M. McLean: Directionally Solidied Materials for High Temperature Service, Book 296, Metals Society, London, 1983, pp. 1139.
2. R. Elliott: Eutectic Solidication Processing: Crystalline and Glassy
Alloys, Butterworths, London, 1983, pp. 23275.
3. W. Kurz and D.J. Fisher: Fundamentals of Solidication, Trans
Tech Publications, Switzerland, 1992, pp. 6389.
4. R. Trivedi and K. Somboonsuk: Mat. Sci. Eng., 1984, vol. 65, pp.
6574.
5. D.J. Allen and J.D. Hunt: Metall. Trans. A, 1979, vol. 10A, pp.
138997.
6. R.N. Grugel: J. Mat. Sci., 1993, vol. 28, pp. 67783.
7. K.A. Jackson, J.D. Hunt, D.R. Uhlmann, and T.P. Steward, III:
Trans. Met. Soc. AIME, 1966, vol. 236, pp. 14956.
8. S.-Z. Lu and J.D. Hunt: J. Cryst. Growth, 1992, vol. 123, pp. 1734.
9. R.N. Grugel: Mater. Charact., 1992, vol. 28, pp. 21319.
10. R.N. Grugel: Metall. Mater. Trans. A, 1995, vol. 26A, pp. 49699.
11. R. Trivedi: J. Cryst. Growth, 1980, vol. 49, pp. 21032.
12. W. Kurz and D.J. Fisher: Acta Metall., 1981, vol. 29, pp. 1120.
13. R. Trivedi: Metall. Trans. A, 1984, vol. 15A, pp. 97782.
14. S.N. Tewari: Mat. Sci. Eng., 1990, vol. A130, pp. 21929.
15. J.D. Hunt and S.-Z. Lu: Mat. Sci. Eng., 1993, vol. A173, pp. 79
83.
16. J.D. Hunt and S.-Z. Lu: Metall. Mater. Trans. A, 1996, vol. 27A,
pp. 61122.
17. R.A. Pratt and R.N. Grugel: Mater. Charact., 1993, vol. 3, pp.
22531.
18. M.A. Tschopp, J.D. Miller, A.L. Oppedal, and K.N. Solanki:
Metall. Mater. Trans. A, 2014, vol. 45A, pp. 42637.
19. R.N. Grugel: Unpublished research, Michigan Technological University, Houghton, 1981.
20. D.H. Kirkwood: Mat. Sci. Eng., 1985, vol. 73, pp. L1L4.
21. H. Esaka: Ph.D. Thesis, No. 615, Swiss Federal Institute of
Technology, Lausanne, Switzerland, 1986.
22. K. Somboonsuk, J.T. Mason, and R. Trivedi: Metall. Trans., 1984,
vol. 15A pp. 96775.
23. M. Burden, D. Hebditch, and J.D. Hunt: J. Cryst. Growth, 1973,
vol. 20, pp. 12124.
24. M.A. Lauer, L. Johnson, M. Ghods, D.R. Poirier, S.N. Tewari,
and R.G. Erdmann: Unpublished research, The University of
Arizona, Tucson, Arizona 85721.

VOLUME 45A, OCTOBER 20144761