CORROSION BEHAVIOUR OF ALUMINIUM
ALLOY METAL MATRIX COMPOSITE JOINTS
OBTAINED BY FSW
F. Zucchi, V. Grassi, C. Monticelli, A. Frignani
“A. Daccò” Corrosion Study Centre, Department of Chemistry, University of Ferrara, Ferrara, Italy.
Abstract
It was studied the corrosion behaviour of two joints, obtained by Friction Stir Welding
(FSW), of aluminium alloy metal matrix composites (MMCs), reinforced with alumina
particles. W6A20A and W7A10A composite joints were examined. Corrosion tests
were performed in EXCO or 3.5% NaCl solutions. The corrosion rates of MMCs and
joints were compared. Corrosion tests were performed on both the nugget and the
thermally affected zones of the joints. Pitting potentials were measured on electrodes
prepared by isolating the different zones. Short-circuit currents were measured between
the nugget, the thermally affected zone and the base.The results indicated the different
behaviour of the two joints: the corrosive attack on the W7A10 joint was concentrated
on the nugget and in its adjacent zones both in EXCO and 3.5% sodium chloride solution;
in the case of W6A20A, the nugget remained protected while the adjacent zones and
the base material were corroded. Electrochemical measurements confirmed these results.
However, FSW, which produces welds without the formation of a fused zone, causes
structural modifications in zones near the nugget, due to the higher temperature produced
by the presence of the reinforcement.This negatively affects the corrosion resistance of
the composite, also in relation to the base alloy composition.
INTRODUCTION
The most commonly used metallic matrix composites (MMC) are those
with an Al alloy base matrix.These are chosen for their light weight, ductility,
low melting temperature and because their mechanical properties can be
improved by thermal treatments. Alloys of the 7XXX (Al-Zn) series, along
with those of the 2XXX (Al-Cu) series, are the most commonly used in
aeronautic field due to their high mechanical resistance and to weight ratio.
The reinforcement consists of ceramic materials, such as silicon carbide
(SiC), alumina (Al2O3), titanium diboride (TiB2), graphite.The most common
MMCs contain reinforcements in the form of particles, both for reasons of
cost and to improve isotropy.
To expand the use of MMC, the problem of their weldability must be
solved. The conventional welding methods with metal melting cause
21 - Metallurgical Science and Technology
Riassunto
E’ stato studiato il comportamento a corrosione di due
giunti, ottenuti con tecnica Friction Stir Welding (FSW), di
compositi a matrice metallica in lega di alluminio, rinforzati
con particelle di allumina. Sono stati esaminati giunti di
compositi W6A20A (matrice AA6061, rinforzo Al2O3, 20%
in volume) e W70A10A (matrice AA7005, rinforzo Al2O3,
10% in volume). Le prove di corrosione sono state eseguite
in soluzione EXCO ed in NaCl 3,5%. Sono state confrontate
le velocità di corrosione dei compositi e dei giunti. Si sono
eseguite prove di corrosione sul nocciolo e sulla zona
termicamente alterata della saldatura. Sono stati misurati i
potenziali di pitting su elettrodi preparati isolando le diverse
zone. Sono state misurate correnti di corto circuito fra il
nocciolo, la zona alterata termicamente e la base. I risultati
hanno messo in evidenza il diverso comportamento dei
due giunti: sul giunto W70A10 l’attacco corrosivo sia nella
soluzione EXCO che in cloruro di sodio 3,5% si concentra
sul nocciolo e sulle zone ad esso adiacenti; sul giunto
W6A20A il nocciolo rimane protetto mentre si corrodono
le zone ad esso adiacenti e la base. Le misure
elettrochimiche confermano questi risultati. La tecnica FSW,
che produce giunti senza la formazione di una zona fusa,
provoca comunque modifiche strutturali nelle zone
prossime al nocciolo, a causa della maggior temperatura
dovuta alla presenza del rinforzo, che influiscono
negativamente sulla resistenza a corrosione del composito,
anche in relazione alla composizione della lega base.
alterations in the distribution and features of the
reinforcement, with substantial modifications of
weld joint resistance. Therefore, these methods
are difficultly applicable.
Recently, there have been obtained MMC welds
by FSW [1,2]. The technique consists in welding
without material melting by exploiting the frictional
heat. During FSW, a rotating tool, consisting of a
conical part and a shoulder, is forced between two
plates and moves along a contact line.The frictional
heat created by the contact of the shoulder against
the workpieces produces a plastic zone on the
edges to be welded.
A rotating probe produces continuous hot work
transporting the plasticized metal from the leading
to the trailing edge of the shoulder, where
consolidation produces a continuous joint.
Nearly flaw-free weld joints are thus obtained.
Indeed, mechanical strength and fatigue behaviour
tests have demonstrated the excellent quality of
the weldings.
Changes in the corrosion behaviour of the nugget,
in the thermally and/or mechanically affected zones
adjacent to it with respect to the base alloy have
been found on different types of Al alloys welded
by this technique. Corrosion tests yield different
results. Thus, the altered microstructures around
the FSW AA5083 joints appear more resistant to
pitting and intergranular corrosion than the base
material [3]. Frenkel and Xia [4] studied the pitting
resistance of FSW AA5454 joints and found that
pitting resistance of the joint was higher than that
of the base alloy. Hu and Maletis [5] evaluated
corrosion and stress corrosion behaviour of FSW
AA2195 and AA2219 joints.
The welded joints showed a corrosion behaviour
better than that of the base alloy. Davenport et al.
[6-8] demonstrated that the nugget and the
thermally affected zones of the FSW AA2024 and
AA7010 joints were more susceptible to pitting
than the base material. Lumdsen et al. [9,10]
indicated that in FSW AA7050-T7651 and AA7075-
T6 joints, the nugget and its adjacent zones were
more sensitive to corrosion. Also the pitting
EXPERIMENTAL PART
The following composites were tested: AA6061
aluminium alloy reinforced with 20% (by volume)
Al2O3, and AA7005 aluminium alloy reinforced
with 10% Al2O3. Both alloys were subjected to
T6 thermal treatment: solubilisation, quenching and
aging.The chemical composition of the two alloys,
expressed as weight percentage, is shown in
Table 1.
The corrosion tests were performed on specimens
of the base material and on FSW joints, with an
exposed surface of approximately 20 cm2. In FSW
Table 1. - Chemical composition (%) of the materials used
MMC Si Fe Cu
W6A20A 0.65 0.15 0.18
W7A10A 0.25 0.24 0.08
22 - Metallurgical Science and Technology
potential of the nugget and thermally affected zone was respectively 75 and
50 mV lower than that of the base alloy. SSR tests in 0.6 M NaCl solution
showed that the joints were susceptible to stress corrosion.
Cracking appears on the interface between nugget and thermally affected
zone. Buchheit and Paglia [11] found that in FSW AA7075-T651 and AA7050-
T7451 alloy joints, the welding zones were more susceptible to corrosion
as compared to the base alloys. In the case of AA7075-T651 alloy, the
thermally affected zones were more susceptible to corrosion and stress
corrosion, while for the AA7050-T7451 alloy the region between the nugget
and the thermo-mechanically altered zone showed maximum susceptibility
to corrosion. In SSR tests, the FSW AA7075-T651 joint in general
demonstrated a better resistance to cracking than 7050- T7451 joint, with
reversed performance in comparison to that of the base alloys.
Pao et al. [12] found that the fatigue crack growth rate in FSW AA7050
joints in air was slightly higher than that of the base alloy. Both in air and in
3.5% NaCl, the fatigue crack advancement rate in the thermally affected
zone was significantly lower and Kth was significantly higher than that of
the base alloy. Gan and Meletis [13], examining the stress corrosion behaviour
of FSW AA2195 joints, did not find any reduction in ductility in SSR tests
under constant immersion in NaCl solutions. A certain degree of
susceptibility to SCC appeared on specimens which, prior to the SSR tests,
were subjected to alternating immersion in NaCl solutions. In this case,
cracking developed in the thermally affected zones.
The object of this paper is to examine the corrosion behaviour of joints
obtained by FSW of Al alloy based MMCs reinforced with Al2O3 particles.
As far as we know, no studies have been published on FSW aluminium alloy
MMC corrosion. The presence of reinforcement should increase the
temperature and cause higher thermal gradients, with greater influence on
alloy microstructures in the zones around the nugget.
joint specimens,the nugget was in central position. The tests for the
evaluation of the weight losses due to corrosion were performed by dipping
the specimens for 48 hours in an EXCO (Exfoliation Corrosion) solution
(4.0 M NaCl + 0.5 M KNO3 + 0.1 M HNO3 ) or for 21 days in 3.5% NaCl
solution, in both cases at room temperature.
Before immersion in the corrosive environment, the specimens were
prepared to obtained perfectly flat surfaces. Grinding was performed using
sandpaper with mesh progressively decreasing to N. 600. These specimen
were then washed in distilled water, degreased in acetone, weighed with an
accuracy of 0.01 mg, then immersed in the corrosive solution.At the end of
the test, the corrosion products formed during the immersion were
eliminated by 70% HNO3 pickling for a few minutes. The weight loss was
used to calculate the corrosion rate, expressed in mdd (mg/dm2 day).
Mn Mg Zn Ti Cr Zr Al
0.10 0.97 0.009 0.02 0.19 - to 100
0.43 1.33 4.59 0.03 0.13 0.13 to 100
The tests in EXCO solution were performed on untreated base material
and on welded specimens.
Coupons for weight loss tests and electrodes for pitting potential
determinations were obtained from the three characteristic zones of the
joints (nugget, thermally affected zone and base material), evidenced by
2.5% HF attack.
The electrodes were connected to a copper wire to ensures electrical
contact and then incorporated in an epoxy resin.
The anodic polarisation curves were recorded with a scanning rate of 0.1mV/
s, starting from the corrosion potential attained after one hour immersion
in the solution.
For both alloys, short-circuit tests were also performed in EXCO solution
between the nugget, the thermally affected zone
(TAZ) and the base material.
Finally, the base materials were subjected to four
different thermal treatments (375°C, 400°C, 450°C
and 500°C, respectively) for two minutes.
Then, on the specimens thus obtained, corrosion
tests in EXCO solution were carried out, as well
as pitting potentials determinations in 3.5% NaCl
solution.
All the tests were provided with photographic
documentation, macrostructural and
microstructural analysis by optical microscope.

Corrosion tests in EXCO solution Table 2. Weight losses in EXCO solution at room
temperature, of base and welded materials
Table 2 collects the corrosion rate values,
expressed in mdd, obtained from the weight losses
Aggressive Test time Material Vcorr
of W7A10A and W6A20A base material specimens,
not containing or containing the welding. environment (hours) (mdd)
The tests lasted 48 hours. W7A10A composite base 921
corrodes faster than W6A20A composite, i.e. 921 W7A10A
welding 1448
mdd in the former case and 588 mdd in the latter EXCO 48
case. base 588
By analysing the coupons containing the welding, W6A20A
welding 440
it was found that W7A10A weld joint showed a
considerably higher corrosion rate than the
corresponding base material, i.e. 1448 mdd for the
former and 921 mdd for the latter, respectively
(Table 2).
The difference appears even more evident if, as
documented in Figure 1, it is considered that the
corrosive attack is located on the nugget and in
the thermally affected zone only. Corrosion rates
higher than 2000 mdd were measured on
specimens taken from the nugget or thermally
affected zones (Table 3).
The behaviour of W6A20A joint is completely
different.
In this case the corrosive attack develops on the
base material and on the adjacent heat affected W7A10A W6A20A
zones.
From EXCO solution tests, it is evident a variation Fig. 1: Appearance of the FSW joints after soaking 48 hours in EXCO solution and
of corrosion resistance of the welded materials pickling
with respect to the base materials not subjected tests on the thermally treated composite specimens.
to FSW.The temperature increase during welding Lumsden et al [9] mapped the temperatures in the joint zone of 7075 alloy
indeed causes microstructural variations which subjected to FSW.
influence the corrosion behaviour of the two They found that in these joints, the zones most susceptible to corrosion
materials. The corrosion resistance modification were the thermally and the thermomechanically affected zones, which
of the two composites caused by a temperature reached temperatures from 375°C to 400°C.
increase was confirmed by performing corrosion Table 4 shows the corrosion rates determined on the base materials of the

23 - Metallurgical Science and Technology

two composites subjected for two minutes to Table 3. Weight losses in EXCO solution, at room
thermal treatment at different temperatures. temperature, of nugget and TAZ in W7A10A joint
The corrosion tests performed on W7A10A Aggressive Test time Material Zone Vcorr
thermally treated specimens indicate an environment (hours) (mdd)
intergranular type corrosion, caused by the
formation of copper-rich precipitates at grain Nugget 2413
boundaries. EXCO 48 W7A10A
TAZ 2092
The consequent copper depletion in the adjacent
matrix causes the lower resistance of the W7A10A
material.
The corrosion rate of the W7A10A composite Table 4. Weight losses in EXCO solution at room
increases as the thermal treatment temperature temperature for two base materials subjected to
increases. Furthermore, the corrosion rate value different thermal treatments
of the material treated at 450°C (2199 mdd) is Aggressive Test time Base Vcorr (mdd)
close that of the thermally affected zone of the environment (hours) Material 375°C 400°C 450°C 500°C
corresponding welded joint. W7A10A 1631 1657 2199 -
AA6061 MMC shows a very similar behaviour to EXCO 48
that above-described for AA7005 MMC, although, W6A20A 621 823 968 1207
in this case, a corrosion rate comparable to that
of joint TAZ is reached only after a thermal
treatment at 500°C. 450°C for AA7005 MMC and 500°C for AA6061 MMC, both higher than
It is evident that the thermal treatment those recorded by Lumsden [9].
temperatures which caused higher corrosion rate, This phenomenon is attributed to the presence of Al2O3 reinforcement, with
comparable to that of joint TAZs, are respectively a volumetric fraction of 10% in AA7005 MMC and of 20% in AA6061 MMC.

Tests in 3.5% NaCl solution

It must be observed that in this solution, pitting is

normally the corrosion attack and the low
corrosion rates determined from weight losses
give a very inaccurate indication of the entity of
the attack.
In fact, the corrosion rates range from 0.32 mdd
(FSW W6A20A) to 1.10 mdd (W7A10A base
material).
After 21 days of exposure, the welded W7A10A
W7A10A W6A20A
specimen showed that the corrosive attack was
Fig. 2: Appearance of FSW joints after soaking 21 days in 3.5% NaCl
concentrated on the nugget and in the immediately
solution and pickling
surrounding zones (Figure 2).
After pickling, a deep attack of the nugget was
observed, while the adjacent zone was unaffected.
The base material specimen shows only a strip of horizontal side and on the transversal sections of the specimen.
corrosion products arising from the hole drilled In 3.5% NaCl corrosive environment, considerably less aggressive than the
to hang up the specimen. EXCO solution, only in the case of W7A10A joints, corrosion was localised
After the same exposure time in the same solution, in the nugget zone.The pitting of the W6A20A joint did not appear, indeed,
pitting was clearly evident on the surface of particularly stimulated by the structural modifications occurring during
W6A20A specimen. welding; the corrosion points, not particularly deep, developed on the base
In this case, pits were distributed both on the material, while the nugget remained unaltered.

24 - Metallurgical Science and Technology

Pitting potential Similar or slightly more negative pitting potential
values were found on the electrodes made of the
Pitting potential values (Epit) of the two materials were obtained from the base material thermally treated from 375°C to
anodic polarisation curves performed in 3.5% NaCl. 450°C.
The electrodes used were taken from the zones of most concern: nugget, These measurements confirm the observations in
TAZ, TAZvn (thermally affected zone near the nugget), TAZvb (thermally 3.5% NaCl corrosion tests performed on the
affected zone near the base), base, and thermally treated base material. welded material in which corrosion was
In the case of W7A10A base material, Epit was -820 mV/ECS, while the concentrated in the central zone of the joint,
thermally affected zone and the nugget presented an Epit from -872 mV/ the one subjected to a highest heating
ECS (nugget) to -880 mV/ECS (TAZ) (Figure 3). temperature.

Fig. 3: Pitting potentials of the characteristic zones of the W7A10A joint and the base material after thermal treatment at different temperatures.

The Epit values of electrodes taken from the characteristics zones of
W6A20A joint and of the thermally treated material are considerably more
positive than the corresponding values measured on the first composite
(Figure 4).
This aspect confirms a higher resistance to pitting corrosion of this alloy in
this environment.
The pitting potentials decrease gradually from the nugget (-672 mV/ECS)
to the thermally affected zone (-688 mV/ECS), to
the base material (-730 mV/ECS).
The thermally treated specimens present an Epit
value which tends to increase as the treatment
temperature increases. The specimen treated at
500°C for two minutes shows a value of Epit close
to that of the thermally affected zone.

Fig. 4: Pitting potentials of the characteristic zones of the W6A20A joint and the base material after thermal treatment at different temperatures

Galvanic coupling measurements Figure 5 shows the corrosion currents related to

W7A10A composite.
From the galvanic coupling measurements in EXCO solution, the nugget- The behaviour of the nugget is anodic with respect
TAZ and nugget-base material short-circuit currents were obtained. The to the base material.
specimens were modelled to have exposed surfaces of equal area. In the nugget-base material galvanic coupling, after

25 - Metallurgical Science and Technology

a few minutes, the short-circuit current stabilises
at 2.5x10-3 A/cm2 and remains constant for the
entire duration of the test.
The currents between the nugget and the
thermally affected zone are considerably lower.The
nugget behaves as an anodic area with respect to
the TAZvb and for most of the time as cathodic
area when in contact with TAZvn.
The corrosion potential (Ecorr) measurements in the same solution confirm
the above-mentioned results.
Ecorr of the base material starts from -880 mV/ ECS, after ten minutes it
reaches -820 mV/ECS, then it remains constant until the end of the test.
For the nugget, the potential starts from -910 mV/ECS and stabilises at
-890 mV/ECS after ten minutes. The Ecorr values of the thermally affected

Fig. 5: Short-circuit currents and corrosion potentials of the characteristic zones of the W7A10A joint in EXCO solution

zones, initially equal to that of the nugget, increase
more slowly. Furthermore, at the end of the test,
Ecorr of the TAZvn was slightly more positive
than that of the nugget.
Figure 6 shows the trends of short-circuit currents
and potentials of the single zones for W6A20A
composite.After the first minutes in which the short-
circuit currents show an anodic behaviour of the
nugget, with respect to both the base material and the thermally affected zone,
an inversion is observed, in which the nugget behaves as a cathode.
The corrosion potentials confirm this trend.The nugget potential after the
first moments stabilises at approximately -750 mV/ECS, while the base
material and the thermally affected zone potentials nearly coincide and are
equal to approximately -780 mV/ECS.The short-circuit currents are reduced
and, at the end of the 24 hour test, they reach a value not greater than
2.5x10-4 A/cm2.

Fig. 6: Short-circuit currents and corrosion potentials of the characteristic zones of the W6A20A joint in EXCO solution

DISCUSSION Similar results were obtained in 3.5% NaCl tests. The thermally affected
zones and the nugget are those which show lower corrosion resistance.
The tests show the considerably different The electrochemical measurements (pitting potential, short-circuit currents,
corrosion behaviour in FSW joints obtained from corrosion potential) confirm the corrosion test results.
the two MMCs. Since no studies have been found on the corrosion behaviour of FSW joints
In W7A10A joints, corrosion in the EXCO solution of aluminium alloy MMC, the results could not be compared. In agreement
is concentrated in the nugget and in the adjacent with [9-11], concerning the corrosion behaviour of FSW 7XXX series alloys,
zone. The attack is enhanced by the contact with we can affirm that in this type of alloy, FSW causes decaying in the corrosion
the base material, which is more noble and works resistance of the joint. This decrease can be attributed to the increase of
as a cathode. precipitates in the zone adjacent to the nugget, with consequent variation
26 - Metallurgical Science and Technology
 BASE NUGGET
MATERIAL

TAZ BASE
MATERIAL
(TT 450°C)

Fig. 7: Optical microscope micrographs of the base material, the nugget, the TAZ of the W7A10A joint and the base material after thermal
treatment at 450°C

of the composition in the grain boundaries, thus making them more
susceptible to intergranular corrosion and pitting.
The tests performed on the thermally treated W7A10A base material
validate this conclusion.
Indeed, the W7A10A specimens, treated for two minutes at 450°C, show a
corrosion rate in EXCO similar to that of the thermally affected zones of the
joints, approximately double that of the W7A10A base material.Also the pitting
potentials of the thermally treated material are similar to those of the welded
zones and lower by around 60 mV than those of the base alloy.
The micrographs in Figure 7 show the most interesting structure of the
W7A10A specimens, i.e., the base material, the
nugget, the TAZ and the 450 °C thermally treated
base material.
It can be observed a considerable increase in the
size of the grains in the TAZ, due to heating, with
formation of a nearly continuous series of
precipitates at the grain boundary. A similar
structure was reproduced by the thermal
treatment at 450°C for two minutes.
On the W6A20A joints, with an alloy notoriously

BASE NUGGET
MATERIAL

TAZ BASE
MATERIAL
(TT 500°C)

Fig. 8: Optical microscope micrographs of the base material, the TAZ of the W6A20A joint and the base material after thermal treatment at 500°C

27 - Metallurgical Science and Technology

more corrosion resistant than 7XXX alloys, the
FSW process does not decrease so drastically the
corrosion resistance, although the mechanical
features of the joint are subjected to the same
variations (hardness, fatigue strength, etc.) as the
W7A10A joint.The micrographs in Figure 8 show
the microstructure variations found in the TAZ
and in the specimen thermally treated at 500°C
as compared to the base alloy.
TAZ
Fig. 9: Optical microscope micrographs of the TAZ and of the base material of the W6A20A joint after thermal treatment
at 500°C and exposure to EXCO solution
CONCLUSIONS
FSW is used to weld materials not weldable by
other techniques. Both joints present a fatigue
strength comparable to that of the base materials
providing the welding technique is applied with
correct parameters.
Resistance to intergranular corrosion and to pitting
of the two MMCs welded by FSW is entirely
different: lower in the W7A10A welds, better in
the W6A20A welds.
REFERENCES
[1] M. PONTE, G. GAMBARO, A.V.
STROMBECK, L.M. MARZOLI and J.F. DOS
SANTOS, Proc. 30° Convegno Nazionale
AIM,Vicenza, November 2004, CD-ROM.
[2] L. CESCHINI, A. CASAGRANDE, G.
MINAK,A. MORRI and F.TARTERINI, Proc.
30° Convegno Nazionale AIM, Vicenza,
November 2004, CD-ROM.
[3] F. ZUCCHI, G.TRABANELLI,V. GRASSI, Mat.
Corros., 52 (2001), p.853-859.
[4] G.S. FRANKEL and Z. XIA, Corrosion 55
(1999), p.139.
[5] WUSHENG HU and E.I. MELETIS, Materials
Science Forum 331- 337 (2000), p.1683.
[6] R. AMBAT, M. JARIYABOON, S.W.
28 - Metallurgical Science and Technology
No modifications in the grain size were observed, but there was an increase
of precipitates in the grain boundary zone of the thermally treated material.
The nugget has grains of smaller size.The EXCO solution attack, as shown
in the micrographs in Figure 9, is essentially intergranular. It can be
hypothesized that the lower corrosion rate of the nugget.
Is linked to a complete recrystallisation, with formation of very small crystals.
The extended grain boundary area prevents the formation of a continuous
network of precipitates in the grain boundary zone, so that the material
results much less susceptible to intergranular corrosion.
BASE
MATERIAL
(TT 500°C)
The intergranular attack and the pits are concentrated in the nugget and in
its adjacent zones in the W7A10A joint.
In the W6A20A joint, the nugget, entirely re-crystallised with very fine grains,
corrodes at a lower rate than the base material and the both thermally and
thermo-mechanically altered zones.
In both cases, it is evident the need to thermally treat the welded material
to improve the corrosion behaviour.
ACKNOWLEDGEMENTS
We thank Linda Pellegrini for contributing to the experimental part.
WILLIAMS, D.A. PRICE, A. WESCOTT and A.J. DAVENPORT, Proc.
Conference Aluminium Surface Science and Technology, Bonn (2003).
[7] A.J. DAVENPORT, R. AMBAT, M. YARIYABOON, P.C. MORGAN, D.
PRICE, A. WESCOTT and S. WILLIAMS, Proc. Conference UMIST,
Manchester 2003.
[8] A.J. DAVENPORT, R. AMBAT, M. YARIYABOON, B.C. CONNOLLY,
S.W.WILLIAMS, D.A. PRICE,A.WESCOTT and P.C. MORGAN, Proc.
Electrochemical Society 2004, 2003-23, p.403-412.
[9] J.B. LUMSDEN, M.W. MAHONEY, G. POLLOCK and C.G. RHODES,
Corrosion 55, (1999), p. 1127-1135.
[10] J.B. LUMSDEN, M.W. MAHONEY, G. POLLOCK and C.G. RHODES,
Corrosion 59, (2003), p.212-219.
[11] R.G. BUCHHEIT and C.S. PAGLIA, Proc. Electrochemical Society 2003,
2002-24.
[12] P.S. PAO, S.J. JILL, C.R. FENG and K.K. SANKARAN, Scripta Materialia
45, (2001), p.605-612.
[13] L. GAN and E.I. MELETIS, J. of Mechanical Behaviour of Materials 14,
(2003), p.163-172.
 FOURIER THERMAL ANALYSIS APPLIED
TO SODIUM EUTECTIC MODIFICATION
OF AN ALSI7 ALLOY
F. Piasentini**, F. Bonollo*, A. Tiziani*
* DTG - Department of Management and Engineering - University of Padova
** ENDIF - Engineering Department - University of Ferrara
Abstract
The so-called “Fourier Thermal Analysis” (or FTA) is an evolution of the “integral thermal
analysis”, which is actually used as a process control in Aluminium and cast iron foundries.
It has been developed since late 80’s in order to investigate nucleation and growing
kinetics of the various phases in multi-component alloys. FTA is based on the evaluation
of the thermal gradient in one-dimensional thermal field that arises in a cylindrical
solidifying specimen.
During the last twenty years, both the traditional thermal analysis and FTA have been
applied to the experimental determination of the solid fraction during solidification, in
order to assess results from numerical simulation. Nevertheless, FTA has not been
applied to foundry process control or optimisation.
Eutectic modification is extensively used in low-pressure permanent mould processes,
in order to improve tensile properties and toughness of Al-Si alloys.The effectiveness of
the treatment is subjected to the presence a minimum amount of modifying elements,
such as Sr, Na or other elements. Traditional thermal analysis is useful in determining
modification level of the alloy, then to control the modification treatment.
Aim of this work is to verify the potentiality of gradient-based thermal analysis method,
such as FTA, in eutectic modification investigation.
An A356-type hypoeutectic Al-Si alloy has been modified with metallic sodium at four
different modification levels. Two-thermocouple thermal analysis curves have been
recorded, in order to perform FTA analysis. Fraction solid versus time (fs(t)) and
temperature (fs(T)) have been determined at different modification levels.
Microstructural characterization has been made using automatic image analysis.Average
values of dimension and roundness of eutectic Si have been compared to thermal analysis
results.
A significant correlation between the so-called “eutectic depression” and silicon
morphology has been observed. More relevant is the behaviour of the fraction solid
curves, from which it is possible to note a significant delay in the start of eutectic
reaction. This delay has been observed not only in time but also in temperature and
fraction solid domain.
11 - Metallurgical Science and Technology
Riassunto
La “Fourier Thermal Analysis” (FTA) è una tecnica avanzata
di analisi termica, sviluppata negli anni ’80-’90 per
determinare il calore latente rilasciato in fase di
solidificazione, allo scopo di indagare le cinetiche di
nucleazione ed accrescimento delle varie fasi coinvolte. I
risultati delle indagini condotte con questa metodologia
hanno contributo allo sviluppo dei codici di calcolo per la
modellazione dei fenomeni di solidificazione di leghe da
fonderia, come le ghise e le leghe di alluminio. Non è
attualmente impiegata nell’ambito della fonderia come
strumento di controllo di processo o come metodo per la
caratterizzazione dei fenomeni di solidificazione.
D’altra parte neanche l’analisi termica “tradizionale” viene
utilizzata in modo diffuso nella fonderia di Al-Si, al contrario
della fonderia di ghisa.
Scopo di questo lavoro è di verificare le potenzialità della
FTA nello studio della modifica eutettica di una lega Al-Si,
e più in generale nella dinamica della solidificazione delle
leghe Al-Si. La modifica eutettica viene impiegata allo scopo
di incrementare le caratteristiche di resilienza e resistenza
a fatica delle leghe Al-Si, l’efficacia del trattamento è
vincolata alla presenza di una percentuale minima di
elemento modificante. L’analisi termica può essere
impiegata per verificare il livello di modifica della lega, e
quindi l’efficacia del trattamento.
é stata presa in considerazione una lega della classe A356,
modificata con il Na, e sono state confrontate le curve
fs(t) e fs(T) ottenute con diversi livelli di modifica, allo scopo
di correlare le caratteristiche delle curve di frazione di
fase solida rispetto al tempo ed alla temperatura con le
caratteristiche microstrutturali della lega.
Sono stati riscontrati effetti della modifica sull’evoluzione
nel tempo delle varie fasi, compresa la fase primaria, ricca
di alluminio.
INTRODUCTION
In a foundry process, the possibility of monitoring
in detail the evolution of solidification is
fundamental, in order to have the correct
understanding of the final microstructure and
metallurgical quality of the casting. Thus, the so-
called “cooling curves” (even if a more correct
definition is that of “solidification and cooling
curves”) are assuming an increased relevance and
diffusion, both from literature and industrial
application viewpoints.
A liquid metal or alloy, placed into a mould,
decreases its temperature, due to the heat transfer
phenomena (conduction, convection, radiation).
The amount of these phenomena is dependent
on various factors, including the geometry of the
mould, the insulation conditions, the possible
motion of the liquid, the presence of cooling
devices, and so on [1-7].
Generally speaking, a temperature vs. time diagram
allows the description of the thermal evolution of
the metal or of the alloy. In the liquid state, the
analytical description of the phenomenon is given
by the equation
dT
A . q = V . ρ . cp . (1)
dt
where V and A are, respectively, the volume and
the surface of the casting, ρ and cp are, respectively,
the density and the specific heat of the material
under investigation, q is the density of the thermal
flow going from the metal to the mould.The term
q can be evaluated, for instance, by the Fourier
equation, once known the starting and contour
conditions of the system. The variables T and t
are, respectively, the metal temperature and the
time; thus dT/dt (i.e. the derivative of temperature
with respect to time) is the cooling rate of the
metal [1,4-5].
Under the hypotheses of small variations in ρ and
cp with temperature and of q = constant, the
cooling rate is also constant, with the cooling curve
assuming a linearly decreasing aspect.
The liquid to solid transition is associated to the
release of the latent heat. Consequently, equation
(1) must be properly modified and becomes:
dT dVS
A . q = V . ρ . cp . − ∆H . (2)
dt dt
12 - Metallurgical Science and Technology
where, apart from the already defined symbols, DH is the solidification
volumetric latent heat and VS represents the volume of already solidified
metal [1,4-5].
At the end of solidification, temperature decreases again, with a different
slope, due to the change in r and cp values from the liquid to the solid state.
These general considerations must be integrated including the role of
nucleation (heterogeneous nucleation is considered) and growth
mechanisms, whose effect depend specifically on the processing conditions.
In particular, undercooling must be considered, being the real driving force
of the whole solidification phenomenon [1]. Nucleation and growth rate
can be described with specific reference to undercooling degree. Applying
the continuous nucleation model, it can be written
N = µ . ∆Tn
1
(3)
and
∂N ∂T
= − n . µ1 . ∆T(n-1) . (4)
∂t ∂t
where N is the number of nuclei formed in the volume unit, DT is the
undercooling, n and m1 are the so-called nucleation parameters.
For the growth rate (Vaccr), it can be written
Vaccr = µaccr . ∆T
2
(5)
where µaccr = growth constant [1,6].
A “real” solidification curve is thus characterised by the presence of
undercooling, associated with the progressive development of the latent
heat, leading to the recalescence phenomenon, which is well described by
the first derivative of the cooling curve (i.e. the cooling rate).
The parameters of the above equations, correlating undercooling with
nucleation and growth rate, are strictly dependent on composition and
technological aspects. In other terms, the evolution of solidification is strongly
related with the specific properties of the cast alloy and on the foundry
processing conditions adopted.
The acquisition and the analysis of the cooling and solidification curve of a
cast alloy have demonstrated their industrial relevance for alloys presenting
eutectic solidification, such as the Fe-C-Si (cast irons) and Al-Si systems [8-
14], both in terms of process control and of microstructure understanding
[11,13].
A recent development of this approach has been introduced for analysing
the kinetics of nucleation and growth of eutectic systems. It is the “Fourier
Thermal Analysis” approach, requiring the acquisition and the subsequent
elaboration of at least two T(t) curves, as well as cylindrical or spherical
crucibles [15-16].
In this paper, such an approach is developed and applied on Al-Si foundry
alloys, to monitor their solidification conditions and to describe the effects
of the modification stage.
FOURIER THERMAL ANALYSIS
Theoretical aspects
The slope of the curve T(t) of a generic alloy, solidifying in a thermal analysis
cup, is due to two contributions: the rate of volumetric heat exchanged to
the environment (HEXC, [Jm-3s-1]) and the rate of volumetric latent heat
released during solidification phenomena (HGEN, [Jm-3s-1]). The slope of
solidification curve is given by equation (6), where Cv is the volumetric
heat capacity [Jm-3K-1] of the system.
HEXC(t,T) + HGEN(t,T,fs)
dT
= [Ks-1]
dt CV(t,TfS)
HGEN depends on the phase transformations that occurs during solidification,
so it is influenced by the melt treatment (grain refining, eutectic modification)
and also by the rate of heat extraction from the melt (the late influencing
the depth of undercooling in solidification transformations).
HLOSS is related to the conditions of heat exchange between the specimen
and the external environment. In the so-called “Newtonian Thermal Analysis”
[8,14-15], the alloy is supposed to be uniform in temperature, so that the
Fig. 1: thermal diffusivity as obtained from (4) (in black) and estimated inside the
solidification interval (in red)
Fig. 2: first derivative of T(t) curve for the central thermocouple (black) and baseline
for latent heat determination (ZF, green).
13 - Metallurgical Science and Technology
heat exchanged is determined only by the heat
exchange at the interface between metal and the
external environment.
In “Fourier Thermal Analysis”, instead, the heat
exchanged is supposed to be related to
temperature gradient inside the alloy. The slope
of solidification curve is then related also to the
thermal diffusivity a [m2s-1] of the alloy (a quantity
related also to the heat conductivity of the material
[15-16]) and the Laplace operator 2T [Km-2], as
(6) reported by equation 7.
HGEN
∂T
= α 2T + Cv (7)
∂t
In the case of a cylindrical cup, temperature
gradient can be estimated by means of a two-point
temperature measurement [15-16], each located
to a certain radius Ri from the centre:
2
4(T2 - T2 )
T = 2 2 2 1 [°C/m2] (8)
R 2- R 1
So that the rate of volumetric heat generated can
be obtained as:
∂T
HGEN = Cv -α 2T [J/s] (9)
∂t
In this equation the term α 2Tgives the so-called
“base line” for the heat measurements, i.e. the
theoretical cooling curve of a material not having
phase changes.
If absence of phase changes (in liquid and solid
state), thermal diffusivity a can be expressed as
the ratio between the slope of T(t) curve in the
centre of the specimen and the Laplace operator.
(∂T/∂t)
α= 2 (10)
T
From this equation, having a bigger diffusivity gives
a bigger capacity of the material of releasing heat
when subjected to the same thermal gradient.
Once kwon thermal diffusivity in solid and liquid
state, diffusivity during solidification can be derived
using a simple linear correlation [15]. An example
of this approach is given by figure 1.
From the product of diffusivity and thermal
gradient, a base line (or “zero line”) Z F can
calculated and used to determine the heat released
during solidification, according to (9).
In figure 2 the zero curve ZF is plotted against the
slope of solidification curve, or dT/dt.
Practical application of FTA
The experimental set up for FTA requires only
two additional features: a symmetrical cup
(cylindrical or spherical) and a second
thermocouple.
Once measured the two time-temperature curves
(Tint(t) e Text(t)), their elaboration consists in this
calculation steps:
1. first derivative in the inner point
2. thermal gradient
SODIUM EUTECTIC
MODIFICATION OF AN AL-SI
ALLOY
Eutectic modification is one of the more effective
techniques for microstructure and mechanical
properties optimisation in Al-Si foundry alloys.
It consists in the addition of specific elements, or
“modifying” elements (such as Na, Sr, Ca), in order
to modify the structure of eutectic Si, from lamellar
to fibrous and spherical.This leads to a substantial
improvement in mechanical properties such as
toughness and tensile strength. Other elements
have been reported to have modifying properties
(i.e. Ba,Y,Yb) [18-26].
Speaking about Na modification, one of the critical
aspects in the use of this element is the short fading
time, after which the modifying effect disappears.
Being a “temporary modification”, it has also some
advantages, because:
- it does not affect significantly properties like
fluidity and viscosity
- modified alloys can be recycled without
additional treatments
EXPERIMENTAL SET UP
The present study has been carried out using an
alloy belonging to the EN AB 42100 class (an A356
equivalent). Chemical composition of the
investigated alloy is given in table 1.
A steel thermal analysis sampling cup, preheated
before sampling, has been used. Thermocouples
are K-type, outer diameter 1.5 mm, shielded with
stainless steel (AISI 316).An additional protection
Table 1. Chemical composition of the investigated alloy
Alloy type\elements Si Mg
AlSi7MgTi 7,06 0,448
14 - Metallurgical Science and Technology
3. thermal diffusivity in liquid and solid state
4. thermal diffusivity during solidification
5. base line ZF
6. heat released during solidification
7. fraction solid with respect to time and temperature
One of the critical aspects is the thermal diffusivity calculation. Some authors
suggest evaluating diffusivity in function of fraction solid, using and iterative
procedure, in which diffusivity is initially supposed to be linear function
with time [15-17].
- it does not form compounds or intermetallic phases
It has been recently pointed out the up-to-dateness of sodium modification,
having care that the overall process is controlled by means of suitable process
control, such as thermal analysis [23]
In practice, sodium fading effect makes necessary a process control on the
modification level of the melt, in order to develop a constant quality in the
casting.
A traditional thermal analysis system can be useful in order to investigate
on the effect of sodium and other modifying elements on the solidification
mechanisms and on the final properties of the modified alloys. Traditional
thermal analysis uses features of the T(t) curves such as eutectic depression
and recalescence, and can be used in conjunction with chemical analysis in
order to characterize the melt and its modification level [12-14,17]
Fourier Thermal Analysis (FTA) has been developed in order to evaluate
heat released during solidification in different cooling conditions, so to
determine the dependence fraction of solid with time and temperature.
The potentiality of the application of this technique to the study of eutectic
modification has not yet been verified in literature, and it’s the main task of
this work.
The knowledge of the dependence of fraction solid with different
solidification condition (heat exchange, melt treatment) is also one of the
crucial points in numerical modelling of microstructure evolution.As a matter
of fact, FTA has been used by different authors to characterize the various
foundry alloys behaviour [2,6-7].
for thermocouple has been used, consisting in stainless steel tube (AISI
316), inner diameter 1.7 mm, thickness 0.1 mm. The distance between the
inner thermocouple (T1) and the outer is 20 mm. Sampling frequency: 10
Hz.
The various step of the application of FTA are listed in the following points:
1. alloy melting and overheating (750°) in muffle furnace, addition of
metallic Na, thermal analysis crucible preheating (750°C);
2. sampling of specimen M01, 5 minutes after modification
3. sampling of specimen M02, 25 minutes after modification
4. sampling of specimen M03, 45 minutes after modification
Cu Mn Fe Zn Ti Al
0,036 0,007 0,13 0,002 0,125 Bal
5. sampling of specimen M04, 65 minutes after modification
Traditional thermal analysis and FTA have then been performed on the
recorded T(t) curves. Results of the analysis have been compared to the
main microstructural features of specimens:
- secondary dendrite arm spacing (SDAS) of a-phase
Fig. 3: T(t) curves at different modification level, corresponding to
specimens M01-M04.
Fig. 4: microstructure of specimen M03 (200x).
Fig. 5: position of the nine areas for metallographic investigation in the thermal
analysis specimen.
15 - Metallurgical Science and Technology
- average area, roundness and aspect ratio of
eutectic silicon
Details on the microstructure evaluation as well
as on thermal analysis are reported in the next
paragraphs.
RESULTS AND DISCUSSION
Traditional and Fourier thermal analysis have been
applied to the specimen taken at various fading
time. Results from these analyses have been
compared to specimen microstructures.
Solidification curves T(t) of the various specimens
are reported in figure 3. Red curves refer to inner
thermocouples, black curves to the outer ones.
Arrows show the effect of Na fading on eutectic
depression. The minimum temperature reached
during eutectic transformation gradually rises with
fading time.This tendency is coherent with results
reported by other authors [22, 23].
Curves have then been elaborated according to
the procedure presented in previous paragraphs,
in order to obtain the dependence of fraction of
solid with time and temperature fs(t) e fs(T), and
to evaluate the effect of Na modification on the
shape of these curves.
Microstructural investigation
Microstructure of specimen named M01ÖM04 has
been characterized by means of Automatic Image
Analysis, in order to quantify alloy modification
level in terms of average roundness and dimensions
of eutectic Si. Other phases have been also
identified, as reported in figure 4.
For each specimen, a total of 9 different areas have
been investigated, as reported in figure 5. In each
of these areas, 9 fields taken at a 500 magnification
have been photographed and analysed. This
procedure has been performed by means of an
automatic image analysis system (LEICA QWIN),
including the field individuation, data recording and
analysis.
In figure 6-9, examples of microstructures taken
on different specimens are presented. At higher
fading times, eutectic Si grows in dimension and
its shapes moves from fibrous to lamellar.
Geometric parameters of the eutectic silicon that
have been investigated are: area, aspect ratio
(length/width), roundness (a parameter defined
inside the image analysis software, linked to
perimeter and area). Both aspect ratio and roundness are useful in describing
the morphology of eutectic silicon: when they are close to unit, the eutectic
is close to spherical shape. On the contrary, if their values are far from the
unit, the shape of eutectic is lamellar.
Average values for the various specimens are listed in table 2.
Fading of Na modification, that occurs during time, leads to the reduction
in modification level. This causes the drop in roundness of eutectic silicon,
and the contemporaneous increase in its size.

Fig. 6: specimen M01-5’. Fig. 7: specimen M02-25’.

Fig. 8: specimen M03-45’. Fig. 9: specimen M04-65’.

Table 2. Geometric parameters of eutectic silicon vs. sodium fading time

M01-5 min M02-25 min M03-45 min M04-65 min
Area [mm2] 10.8 13.1 35.9 77.6
Roundness 3.3 3.6 5.8 8.4
Aspect ratio 2.5 2.4 3.2 4.3

Microstructural characterization has been between secondary dendrite arms of Al-rich a phase.The SDAS (Secondary
completed with the measurements of the spacing Dendrite Arm Spacing) has been evaluated in 81 different fields. Results from

16 - Metallurgical Science and Technology

 Table 3. SDAS variation with fading time. these measurements are listed in table 3.
Specimen time [min] SDAS [mm] s [mm] The variation of roundness and aspect ratio with
fading time is presented in figure 10. The average
M01 5 64.0 5.0 values of both these features have a strong increase
M02 25 68.0 5.5 between 25 and 45 minutes, showing a decrease
in modification level. After 45 minutes from the
M03 45 68.1 7.2 introduction of Na in the melt, the modification
M04 65 68.1 4.1 level reaches a critical point, from which the
properties of the material are supposed to show
a significant decrease.

Rate of heat released versus time

With FTA it is possible to evaluate the rate of
heat released during time (or HGEN), so that fraction
of solid can be calculated as the fraction of heat
released until that time on the total of heat
released.
A visualization of the various stages of the
solidification processes is then visible, as reported
in figure 11, referred to 5 minutes after
modification occurred. Three different areas are
visible, the first one related to the primary a phase,
the second one belonging to Al-Si eutectic and
Fig. 10: variation of eutectic silicon roundness and aspect ratio
for different fading time.
the third one to the secondary phases and
intermetallic compounds with other alloying
elements (like Mg2Si and Fe-Mn-Mg rich phases).
The area related to the last solidifying phases
seems to be more evident at the highest fading
times, as reported in figure 12-13.The contribution
on heat generated gradually tends to get mixed
up with the contribution of eutectic silicon.

Fig. 11: latent heat released vs. time, 5 min. after modification, specimen M01.

Fig. 12: latent heat released vs. time, 25 min. after Fig. 13: latent heat released vs. time, 65 min. after
modification, specimen M02. modification, specimen M04.
17 - Metallurgical Science and Technology
Rate of heat released versus
solid fraction
Latent heat released can be compared to fraction
of solid, in order to free the results from time
dependence. In this way it’s possible to compare results obtained in different
cooling conditions.
From figures 14-15 it is evident the shift in the initial point for eutectic
reactions, which is moving towards lower fraction of solid.

Fig. 14: latent heat released vs. solid fraction, 5 min. after modification, Fig. 15: latent heat released vs. solid fraction, 65 min. after modification,
specimen M01. specimen M04.

Fraction of solid versus

temperature
Fraction of solid fs, as measured by means of FTA,
can be plotted against the temperature measured
in the centre of the thermal analysis cup.The effect
of modification on transformation kinetics is then
visible in the curve fs(T).
Figure 16 collects the various fs(T) curves for
specimens M01-M04. Starting point for eutectic
reaction is identified with the minimum point of
fs(T) curves.
Not only modification shifts this point to lower
temperatures (as shown also from traditional
thermal analysis), but also seems to shift the
fraction of solid of this point to higher values. Fig. 16: solid fraction vs. temperature, at various fading time.
In other terms, eutectic reaction starts at lower
temperatures, when a higher quantity of Aluminium
rich a-phase has been already formed.

18 - Metallurgical Science and Technology

 Table 4. variation of eutectic start with fading Data regarding the shift of the eutectic start point
time. are listed in table 4.
The presence of sodium in the melt seems to
Specimen Fading time fs at start of T at start of
influence also the primary solidification, not only
[min] eutectic reaction eutectic reaction
the eutectic formation.
[°C]
M01 5’ 0.66 558.0
M02 25’ 0.55 558.5
M03 45’ 0.52 560.1
M04 65’ 0.51 565.0

CONCLUSIONS

The so-called Fourier Thermal Analysis (FTA) has been used to monitor
the effect of modification on the evolution of solid fraction during
solidification of an A356 Al-Si alloy.
The latent heat released during solidification has been measured, and from
that the solid fraction has been obtained during the various stages of
solidification.The evolution of solid fraction in temperature (the fs(T) curves)
has been found to be the most effective in evaluating the effect of melt
treatment on the solidification phenomena.
Solid fraction curves fs(T) have been obtained at different modification levels,
showing significant effects of the sodium modification level on the starting
point of formation of eutectic phases.
These results confirm the feasibility of using FTA as an experimental method
for evaluating fs(T) curves in different melt treatment conditions, in particular
regarding eutectic modification. These curves can be used also as input
parameters for numerical modelling of microstructure formation in foundry
alloys.

ACKNOWLEDGMENTS

Authors want to thank professor Ingvar Svensson and dr. Attila Diószegi
(Jonköping University, SE) for the theoretical support; Roberto Losco,
Giacomo Mazzacavallo, Enrico Della Rovere and Alice Dal Martello for their
irreplaceable contribution in the test realization and in the microstructural
characterization of the specimens.

19 - Metallurgical Science and Technology

REFERENCES
[1] M.C. Flemings: “Solidification Processing”,
Mc Graw Hill, New York (1974).
[2] M. Rappaz, D.M. Stefanescu: “Modelling of
microstructural evolution”, ASM Metals
Handbook, 9th ed., vol. 15: Casting (1988),
ASM - Metals Park, Ohio, pp 883-891.
[3] R.Trivedi,W. Kurz,“ Dendritic Growth”, Int.
Materials Reviews, 19 (1994), pp 49-74.
[4] R.B. Bird, W.E. Stewart, E.N. Lightfoot,
“Transport Phenomena”, J. Wiley & Sons,
New York (1966).
[5] R.S. Brodkey, H.C. Hershey, “Transport
Phenomena:A Unified Approach”, McGraw-
Hill Int.Book Co., New York (1988).
[6] D.M. Stefanescu: “Critical review of the
second generation of solidification models
for castings: macro transport -
transformation kinetics codes”, Proc. Conf.
“Modelling of Casting, Welding and Advanced
Solidification Processes VI”,TMS (1993), pp 3-
20.
[7] M. Rappaz: “Solidification processes:
constitutive equations and
microstructures”, Proc. Conf. “Modelling of
Casting, Welding and Advanced Solidification
Processes VI”, TMS (1993), pp 67-91.
[8] M. J. Oliveira, L. F. Malheiros, C.A. Silva
Ribeiro:“Evaluation of the heat of solidification
of cast irons from continuous cooling curves”,
Journal of Materials Processing Technology, 92-
93 (1999), pp 25-30.
[9] R. Sillen: “Monitoring and optimisation of
melting and treatment processes for cast
iron alloys by means of advanced thermal
analysis”, Proc. Australasian Foundr y
Conference (2000).
[10] F. Bonollo, A. Tiziani, “Le curve di
raffreddamento e la solidificazione delle
ghise – Le basi teoriche”, Industria Fusoria, 6
(2001), pp 47-57, Assofond – Milano.
[11] F. Piasentini, F. Bonollo,A. Pennetta,A.Tiziani,
“The use of thermal analysis for process
control in cast iron foundry”, Proc. 2nd Int.
Conf. “New Developments in Metallurgical
Process Technology”, Riva del Garda (19-21
settembre 2004); paper 159; Associazione
Italiana di Metallurgia – Milano
[12] M. Djurdjevic, H. Jiang, J. Sokolowski, “On
line prediction of aluminium-silicon
modification level using thermal analysis “,
Materials Characterization, 46 (2001), pp 31-
38.
20 - Metallurgical Science and Technology
[13] F. Piasentini, A. Tiziani, F. Bonollo, “L’analisi termica come strumento
di controllo di processo nella fonderia della ghisa e dell’alluminio”,
Proc. Conf.“30°Convegno Nazionale AIM”, Vicenza (17-19 novembre
2004), paper 134; CD-ROM edited from Associazione Italiana di
Metallurgia - Milano.
[14] J. C.Escobedo, M. A. Ramírez, D. A. Cortés, “Computer-aided cooling
curve analysis (CA-CCA), applied to an AlSi system”, Light Metals,
1997, TMS Annual Meeting
[15] A. Diószegi, I. Svensson, “A comparison of Fourier vs. Newtonian
thermal analysis and its influence on the inverse kinetic growth
calculation”, On microstructure formation and mechanical properties in
grey cast iron, PhD thesis, 2004, Lynköping University
[16] E. Fras, W. Kapturkiewicz, A. Burbielko, H.F. Lopez, “Numerical
Simulation and Fourier Thermal Analysis of Solidification Kinetics in
High-Carbon Alloys” Metallurgical and materials transactions B, vol.
28B, 1997, pg 115-123
[17] C. Baez., “Fourier thermal analysis of the solidification kinetics in
A356/SiCp cast composite” Journal of Materials Processing Technology,
Volumes 153-154, 10 November 2004, pages 531-536
[18] H. Makhlouf, H.V. Guthy, “The aluminum silicon eutectic reaction:
mechanism and crystallography”, Journal of light metals, 3 (2002), pp.
199-218
[19] B. Closset,“Microstructure Modification by Using Strontium in Various
Casting Processes”, Proc. Conf.“Alluminio e Magnesio: nuove tecnologie
di fonderia”, Milano (1995);Associazione Italiana di Metallurgia - Milano.
[20] J. E. Grusleski, B.M. Closset, “The treatment of liquid Aluminum-
Silicon alloys”, (1990),The American Foundrymen’s society Inc, USA
1990.
[21] A. K. Dahle, K. Nogita. J.W. Zindel, S.D. Mc Donald, L.M. Hogan, “Eutectic
Nucleation and Growth in Hypoeutectic Al-Si alloys at Different
Strontium Levels”, Metallurgical and Materials Transactions, 32A, 4
(2001), pp 949-960.
[22] A. Knuutinen, K. Nogita, S.D. McDonald,A.K. Dahle, 2001,“Modification
of Al–Si alloys with Ba, Ca, Y and Yb”, Journal of Light Metals, 1, pp
229–240.
[23] A. Manente, P. Ferro, E. Della Rovere, “Aspetti innovativi nello studio
della modifica eutettica al sodio di leghe alluminio-silicio”, La
Metallurgia Italiana, 6 (2003), pp 27-35.
[24] N.Fatahalla, M.Hafiz, M.Abdulkhalek,“Effect of microstructure on the
mechanical properties and fracture of commercial hypoeutectic Al-
Si alloy modified with Na, Sb and Sr”, Journal of Materials Science, 34
(1999), pp 3555-3564.
[25] K.Nogita, S.D.McDonald, J.W.Zindel,A.K.Dahle,“Eutectic solidification
mode in sodium modified Al-7mass%Si-3,5mass%Cu-0,2mass%Mg
casting alloys”, Materials Transactions, v 42, n 9, 2001, p 1981-1986
[26] K.Nogita, A.K.Dahle, “Eutectic growth mode in strontium, antimony
and phosphorus modified hypoeutectic Al-Si foundry alloys”, Materials
Transactions, v 42, n 3, 2001, p 393-396
 SOLIDIFICATION BEHAVIOUR AND
MICROSTRUCTURE OF A360-SICP
CAST COMPOSITES
P. Bassani*, B. Previtali**, A. Tuissi*, M. Vedani**, G. Vimercati**, S. Arnaboldi**
* CNR IENI, Lecco, Italy
** Politecnico di Milano, Milan, Italy
Abstract
In the present investigation, the solidification microstructure of Al based composites
was studied. The materials investigated were a commercial A360 aluminium alloy and
the corresponding composites reinforced with 10 and 20 vol. % SiC particles.The materials
were cast under different conditions for a first set of laboratory tests, aimed at studying
the solidification behaviour and the related microstructures. The reinforcement
distribution in the cast composites revealed to be governed by the pushing action exerted
by the matrix solidification front during casting, giving rise to clustering effects of the
SiC at dendrite boundaries, especially in sand castings. Copper mould castings,
experiencing reasonably high cooling rates during solidification, showed a very fine
microstructure within which the segregational effects of the reinforcement was of limited
relevance. SiC particles also revealed to be a site for heterogeneous nucleation of coarse
Si crystals in the composites
3 - Metallurgical Science and Technology
Riassunto
Nel presente studio si sono volute analizzare le
microstrutture di solidificazione di materiali compositi a
matrice in lega di alluminio. I materiali utilizzati sono stati
una lega commerciale A360 di alluminio e i corrispondenti
compositi rinforzati rispettivamente con il 10% e il 20% in
volume di particelle di SiC. I materiali sono stati colati in
differenti condizioni per una prima serie di analisi di
laboratorio, con lo scopo di studiare il comportamento di
solidificazione e le relative microstrutture. La distribuzione
del rinforzo nei getti in composito si è rilevata essere
governata da fenomeni di “pushing” esercitati dal fronte di
solidificazione durante la colata, generando, in particolar
modo nei getti in sabbia, fenomeni di segregazione del SiC
ai bordi delle dendriti. Colate in lingottiera in rame, che
come noto garantiscono elevate velocità di raffreddamento
in fase di solidificazione, hanno permesso di ottenere
microstrutture molto fini, con fenomeni di segregazione
del rinforzo di limitata rilevanza. Inoltre nei compositi le
particelle di SiC si sono rilevate essere in grado di
promuovere una nucleazione eterogenea di grossolani
cristalli di silicio primario.
INTRODUCTION
Metal matrix composites (MMCs) are a promising
group of materials, produced by combining a
metallic alloy matrix, usually an aluminium alloy,
with a hard and stiff ceramic reinforcement. Their
properties can be tailored by a proper selection of
matrix properties and by changing the composition,
shape, size and volume fraction of the reinforcing
phase. Interesting application fields exist in the
automotive and aerospace sectors, electronic
packaging as well as in sport equipments [1,2].
Discontinuously reinforced MMCs, such as
particulate reinforced aluminium alloys, offer the
additional advantage of being amenable to a large
part of the conventional metalworking techniques
usually adopted for the corresponding
unreinforced alloys. However, due to the effects
of the abrasive ceramic particles in the composite,
machining undoubtedly remains one of the less
attractive routes for obtaining complex parts. For
this reason, cast composite parts, whose final shape
is directly obtained by solidification, are considered
of strategically importance for a rapid uptake of
MATERIALS AND
EXPERIMENTAL PROCEDURES
The materials studied in the present investigations
are an A360 alloy and two composites obtained
by the addition of 10 and 20% (nominal content)
of SiC particles.The composition of the materials
investigated is listed in Table 1.The materials were
provided in the form of commercial foundry ingots
weighting about 12 kg.
A first set of castings was produced by a laboratory
vacuum induction melting furnace (Balzers VSG10)
equipped with a ZrO2 (65%)/SiO2 (35%) crucible.
For electromagnetic coupling, the melting crucible
was inserted in an external graphite crucible
(Figure 1). Melting operations were conducted
Table 1. Chemical composition (mass %) of the A360 alloy and of the matrix
of the composites investigated
Alloy Si Fe Cu
A360 9,23 0,49 0,079
A360-10%SiC 10,8 0,90 0,004
A360-20%SiC 9,94 0,89 0,007
4 - Metallurgical Science and Technology
MMCs in advanced industrial applications [3].
The solidification behaviour of particulate reinforced MMCs is strongly
influenced by the interaction of the solid reinforcement phase with the
solidification front. Depending on physical properties of ceramic
reinforcement and liquid matrix, and on solidification conditions (e.g.:
temperature gradient and solidification rate), particle pushing ahead of the
solidification front rather than reinforcement engulfment in the solidified
metal matrix can occur. These possibilities in turn significantly affect the
degree of reinforcement distribution of the cast composite since, when the
particles are rejected in the liquid phase, they will concentrate in the
interdendritic regions of the structure, leading to harmful segregation effects
[1,4-8].
From a technological perspective, there are other peculiar aspects that
deserve special attention when comparing the foundry procedures of Al
based MMCs with respect to those of the unreinforced alloys. Melting is
usually performed in conventional furnaces based on gas combustion,
resistance or induction heating. However, particle settling must be prevented
by periodical stirring of the melt.Temperature should be strictly controlled
at any time in order to avoid unwanted reaction between SiC and Al that
would occur on overheating [9-12]. Skimming as well as degassing to increase
melt cleanliness are not usually accomplished in MMC melts due to the
tendency of reinforcement accumulation at melt surface that would result.
Therefore, the starting ingots and all the equipments to be used in contact
with the composite melt have to be carefully cleaned and devoid of humidity.
under controlled Argon atmosphere at a pressure of 650 mbar and the
materials were cast in a square-section (70x70 mm) water-cooled copper
mould.A thermocouple was positioned at the centre of the mould in order
to record the ingot temperature during solidification and cooling.To evaluate
the effectiveness of electromagnetic stirring of the liquid Al alloy and the
suspension of the solid SiC particles, no mechanical stirring of the melt was
performed.
A second set of castings was then produced with an industrial gas furnace
at a foundry plant. For this case, the melt was periodically stirred both
during melting and just before pouring to prevent particle settling. For a set
of castings, modification of the alloy was performed by using Na-based
additives, according to conventional foundry practice.The molten alloy and
composites were then poured in dry sand moulds having two different
shapes. Simple cylindrical shaped castings of 50 mm in diameter and 80 mm
in height were produced as samples for the laboratory analyses. Also in
these circumstances, the temperature was measured by a thermocouple
Mn Mg Cr Zn Ti Ni Sr Al
0,30 0,34 0,017 0,063 0,03 0,001 - bal.
0,76 0,63 0,002 0,005 0,09 0,005 0,010 bal.
0,62 0,63 - 0,013 0,08 - - bal.
 Fig. 1: Experimental set-up for laboratory scale castings. Insight of the vacuum chamber (a) and schematic section
of the water-cooled copper mould (b).

placed at cylinder axis. In addition, a series of much larger castings, weighting of the A360 alloy. Optical and scanning electron
about 10 kg, were produced as depicted in Figure 2. Detail on the properties microscopy (SEM) as well as hardness and
of these latter castings will be presented in a future paper. microhardness measurements were adopted for
Samples cut from the sectioned castings were then investigated for the microstructural analyses presented in this
microstructural analyses to study the effects of the different solidification paper.
conditions and of the different amount of reinforcement on the structure

Fig. 2: MMC castings produced according to foundry practice. Pouring of the composite melt in the sand mould (a)
and MMC complex-shape casting (b).

5 - Metallurgical Science and Technology

RESULTS AND DISCUSSION

Cooling curves and cooling rates

In Figures 3 and 4 the cooling curves of the
materials investigated are reported for water-
cooled copper-mould castings and cylindrical sand
castings, respectively.
An estimation of the cooling rates experienced
by the composites on solidification was obtained
by calculating the first derivative of the
temperature-time curves.The cooling rate curves
are also reported in figures 3 and 4.
Interpretation of the cooling curves and their
derivatives can be obtained by knowledge of the
equilibrium solidification sequence and by other
published works [13,14]. It is well established that
the initially high cooling rate corresponds to liquid Fig. 3: Cooling curves and cooling rate curves of the copper permanent mould
castings.
cooling, solidification onset is marked by the
nucleation of primary a-Al phase from the liquid,
resulting in a large decrease of the solidification
rate with its main thermal effects corresponding
to the first downward peak in the cooling rate
curves.After a short period, the Al-Mg2Si-Si ternary
eutectic solidification occurs under quasi-
isothermal conditions, resulting in a plateau of the
cooling curve profiles.As a rough estimation, from
the recorded cooling curves, it is possible to state
that, at the first stages of solidification, the cooling
rate in the water cooled copper mould was of the
order of 200°C/min whereas the cooling rate in
the sand castings was of about 25°C/min.
Completion of cooling was reached after about
60 second in water cooled mould, while it took
about 10 minutes for sand mould.
Inspection of the cooling curves of the three
materials also suggests that the solidification Fig. 4: Cooling curves and cooling rate curves of the cylindrical sand castings.
behaviour of the matrix alloy is apparently not
affected by the presence of the SiC particles. As
expected, the length of the eutectic solidification
Microstructure
plateau is a direct function of the amount of alloy
poured into the moulds and therefore not strictly Preliminary analyses on the as received composite ingots were carried out
constant for the tests performed in this to identify the main phases in the unreinforced A360 alloy and in the
investigation. More accurate considerations would composite matrices. Irrespective of reinforcement content, the matrix alloy
suggest that the solidification time of the eutectic was formed by the α-Al solid solution and by four types of secondary
should also be proportional to the amount of phases. Electron probe microanalyses were carried out on unetched samples
matrix alloy (i.e. decreasing with increasing SiC to identify the nature of these constituents. Their chemical compositions
content) [14]. were analytically treated to account for the unwanted interaction of the X-
ray beam with the matrix surrounding the smallest phases [15]. Si platelets
and coarse primary Si crystals, Mg2Si, FeMg3Si6Al8, (FeMn)3Si2Al15 were
identified as the main intermetallics constituting the matrix alloy. These
phases could also be easily recognised in optical micrographs due to their
contrast in colour and typical shape, as shown in Figure 5a. The composite
structure additionally featured the presence of blocky-shaped, roughly
equiaxed SiC particles, whose average size was of 6 micrometers, as depicted
in Figure 5b.

6 - Metallurgical Science and Technology

 A) B)

Fig. 5: Representative micrographs of the as received A360 alloy (a) and of the A360-20%SiC composite (b). In figure 5a the labels
correspond to the following phases: (1) FeMg3Si6Al8; (2) (FeMn)3Si2Al15; (3) Si platelets; (4) Mg2Si.

A) B)

C) D)

Fig. 6: Micrographs of the unmodified sand cast A360 alloy (a), of the A360-10%SiC composite (b and c) and of the A360-20%SiC composite (d).

Analyses on the cylindrical sand castings showed that the reinforcement
distribution was fairly homogeneous, at least on a macroscopic scale. From
a microstructural viewpoint, it was noticed that the reinforcement particles
were preferably segregated in the interdendritic regions (Figure 6), i.e. SiC
was pushed by the growing α-dendrites and gathered together with the
eutectic constituent at dendrite boundaries. High-
magnification views of the structure showed that
the SiC particles featured well defined edges and
a shape substantially comparable to those detected
in the as-received ingots, thus supporting the

7 - Metallurgical Science and Technology

 A) B)

Fig. 7: Comparison of the unmodified (a) and Na-modified (b) structure in the sand cast A360-10%SiC composite.

hypothesis of absence of significant matrix-
reinforcement reactions occurring during melting
and casting of the materials. In addition, it was
observed that coarse Si crystals were distributed
in the structure of the composites (arrowed dark
grey phases in Figure 6b). From inspection at higher
magnification, there was reasonable evidence that
these coarse crystals took their origins by
heterogeneous nucleation at SiC particles (light
phase containing several SiC particles in Figure 6c).
Similar conclusions concerning the nucleation of
the Si phase at SiC particles were already given in
literature for the eutectic Si platelets [2,13,16,17]
and for the primary Si crystals [14]. Of particular
relevance for the present investigation is the
observation that SiC particles can promote the formation of primary Si
crystals even in hypoeutectic alloys such as the A360 alloy matrix. The
nucleation role played by SiC is also confirmed by the fact that no primary
Si crystals were detected by careful observations of the unreinforced A360
alloy samples (see Figure 6a).
The effect of Na-based modifier in order to change the shape of the Si
platelets is depicted in figure 7. It is inferable that the modified eutectic
features tiny Si particles whose shape is nearly spherical or at least equiaxed,
as opposed to the more elongated shape of the unmodified samples (see
also figure 5b). It is well established that this morphology gives improved
ductility properties when compared to the unmodified eutectic [14,18,19].
Figure 8 depicts representative micrographs of the water-cooled copper-
mould cast composites. As already mentioned, these castings underwent a
rapid solidification leading to a much finer grain structure. For this condition,
the size of the matrix grains became of the same order of magnitude of the

A) B)

Fig. 8: Optical micrographs of the water-cooled copper mould castings (a) A360-10%SiC composite, (b) A360-20%SiC composite.

8 - Metallurgical Science and Technology

reinforcement particles. The tendency toward SiC clustering to dendrite
boundaries was thus less evident due to the limited segregation path allowed
to the reinforcement. For this reason, the dispersion of reinforcement in
the copper mould cast composites was remarkably improved. However,
lack of homogeneity was still noted in the A360-10%SiC composite in lower
magnification micrographs, where regions of reinforcement free matrix were
observed (Figure 8a). This effect was supposed to be related to uneven
reinforcement distribution in the liquid melt before
casting.
Finally, it is worth emphasizing that the only casting
defects detected consisted of small and isolated
microvoids related to shrinkage phenomena, a few
tenths of micrometer in size, often located within
SiC particle clusters.

Hardness From analysis of the hardness data it can be stated

that the largest effect is played by the
Table II summarizes the average Vickers hardness values of the materials
reinforcement content, leading to a monotonic
investigated together with the standard deviation of data.
increase of hardness when increasing the SiC
Table II. Vickers hardness values of the materials content. As expected, solidification rate also plays
investigated a remarkable role, especially in the unreinforced
alloy, while the effects of Na-modification revealed
Alloy Copper mould Sand Sand casting to be less defined, being of the same order of the
casting casting (Na modified) experimental data scatter.
A360 78,6 ± 4,5 63,6 ± 2,6 -
A360-10%SiC 105,6 ± 4,1 97,0 ± 9,1 93,3 ± 8,0
A360-20%SiC 125,3 ± 10,8 107,3 ± 14,9 109,3 ± 11,8

CONCLUSIONS 4. In the solidified composites, the SiC particles

were generally located at dendrite boundaries
From the investigations on the A360-alloy based particulate-reinforced due to the pushing action exerted by the
composites, the following conclusions could be drawn. solidification front.When the dendrite-branch
1. The composites could be satisfactorily cast in different shapes and by size became of the same order of magnitude
different processes, experiencing a large range of cooling rates. of the particle size (copper moulds), this
2. Induction melting generally allowed to achieve good homogeneity of effects was less relevant. However, local SiC-
the molten alloy due to electromagnetic stirring.This effect also revealed depleted regions related to uneven
to be suitable to retain a satisfactory dispersion of the reinforcement reinforcement distribution in the liquid
particles, preventing SiC settling due to difference in density with the composite were still noticed in the 10%SiC-
matrix alloy. reinforced composites.
3. Sand casting were successfully produced by mechanical stirring of the 5. The presence of SiC particles also revealed
melt. Due to larger dendrite size, the SiC distribution was less to promote the formation of coarse primary
homogeneous from a microstructural viewpoint. Only small casting Si crystals in the composite structure.
defects related to microshrinkage phenomena were detected in the 6. Na-based modification of the eutectic was
sand cast samples. used during a set of sand casting of the
composites. The effectiveness of the Si-
platelets modification was demonstrated.

ACKNOWLEDGEMENT

The present research was performed with the financial support of

Fondazione CARIPLO. The authors would also like to acknowledge the
skilful contribution of Fonderia F.lli Liotti SrL (Verbania, Italy) for their support
in preparing the foundry sand castings and Mr. M. Pini for the laboratory
copper-mould castings.

9 - Metallurgical Science and Technology

REFERENCES [11] Kobashi M., Choh T.The wettability and the reaction for SiC particle/
Al alloy system. J. Mater. Sci. 1993;28:684-690.
[1] Warren H.H., Darrell R.H. Metal matrix [12] Howe J.M. Bonding, structure and properties of metal/ceramic
composites.Adv. Mater. Proc. 2004;2:39-42. interfaces: Part1 Chemical bonding, chemical reaction and interfacial
[2] Surappa M.K. Microstructure evolution structure. Int. Mater. Rev. 1993;38(5):233-256.
during solidification of DRMMCs: state of [13] Braszczynski J., Zyska A.Analysis of the influence of ceramic particles
the art. J. Mater. Proc. Techn. 1997;63:325- on the solidification process of metal matrix composites. Mater. Sci.
333. Engng. 2000;A278:195-203.
[3] Klimowicz T.F. The large scale [14] Wu S., You Y., An P., Kanno T., Nakae H. Effect of modification and
commercialisation of aluminium-matrix ceramic particles on solidification behaviour of aluminium-matrix
composites. J. Metals 1994 ;46(11) :49-53. composites. J. Mater. Sci. 2002;37:1855-1860.
[4] Kim J.K., Rohatgi P.K. The effect of the [15] Quian M., Taylor J.A., Yao J.Y., Couper M.J., St John D.H.. A practical
diffusion of solute between the particle and method for identifying phase particles in aluminium alloys by electron
the interface on the particle pushing probe microanalysis. J. Light Met. 2001;1:187-193.
phenomena. Acta Mater. 1998;46(4):1115-
[16] Asthana R. Review Reinforced cast metals – Part I Solidification
1123.
microstructure. J. Mater. Sci. 1998;33:1679-1698.
[5] Juretzko F.R., Dhindaw B.K., Stefanescu D.M.,
[17] Wu Y., Liu H., Lavernia E.J. Solidification behaviour of Al-Si/SiC MMCs
Sen S., Curreri P.A. Particle engulfment and
during wedge-mold casting. Acta Metall. Mater. 1994;42(3):825-837.
pushing by solidification interfaces: part I –
ground experiments. Metall Mater Trans [18] Paray F., Gruzleski J.E. Effect of modification on aluminium matrix of
1998;29A(6):1691-169. Al-Si-Mg alloys. Mater. Sci. Techn. 1994;10(9):757-761.
[6] Mukherjee S., Stefanescu D.M. Liquid [19] Conley J.G., Huang J.,Asada J.,Akiba K. Modelling the effects of cooling
convection effects on the pushing- rate. Hydrogen content, grain refiner and modifier on microporosity
engulfment transition of insoluble particles formation in Al A356 alloys. Mater. Sci. Engng. 2000;A285:49-55.
by a solidifying interface: Part I – analytical
calculation of the lift forces. Metall. Mater.
Trans. 2004;35A(2):613-621.
[7] Sasikumar R., Kumar M. Redistribution of
particles during casting of composite melts:
effects of buoyancy and particle pushing.
Acta Metall. Mater. 1991;39(11):2503-2508.
[8] Kaptay G. Reduced critical solidification
front velocity of particle engulfment due to
an interface active solute in the liquid metal.
Metall. Mater. Trans. 2002;33A:1869-1873.
[9] Ocelik V.,Vreeling J.A., De Hosson J.T. EBSP
study of reaction zone in SiC/Al metal
matrix composites prepared by laser melt
injection. J. Mater. Sci. 2001;36:4845-4849.
[10] Yan M., Fan Z. Review – durability of
materials in molten aluminium alloys. J.
Mater. Sci.. 2001;36:285-295.

10 - Metallurgical Science and Technology