Uploaded by Ahmed EL Gazzar

Ryen2006 Article StrengtheningMechanismsInSolid

Strengthening Mechanisms in Solid Solution
Aluminum Alloys
A number of commercial and high-purity non–heat-treatable aluminum alloys are investigated in this
work. It is found that both magnesium and manganese in solid solution give a nearly linear concentration dependence of the strength at a given strain for commercial alloys. This deviates from highpurity AlMg binary alloys, where a parabolic concentration dependence is found. Mn in solid solution
is found to give a considerably higher strengthening effect per atom than Mg, both in terms of yield
stress and initial work hardening rate. This strengthening effect is stronger comparing commercial
grades to high-purity alloys. This enhanced strengthening is believed to be a synergy or clustering
effect caused by interaction between Mn atoms and trace elements, probably silicon, in solid solution.
NON–HEAT-TREATABLE aluminum alloys constitute
a class of alloys that owe their strength mainly to elements
in solid solution and the grain size, and in some cases also
to particles. Heat treatment of such an alloy will generally
not produce any strengthening precipitates as in the heattreatable alloys. The strength may in fact decrease during
heat treatment due to the removal of solute atoms (an
exception is the precipitation of dispersoids in Al-Mn
alloys). The alloy systems belonging to this class are the
AA1xxx system (commercially pure with small amounts of
mainly Fe and Si), the AA3xxx system (with manganese
addition), and the AA5xxx system (with magnesium addition). Non–heat-treatable alloys are used in a wide range
of applications where a low to medium strength, a good
formability, and a good corrosion resistance are desirable.
Beverage cans and automotive panels are two examples out
of many.
The introduction of foreign atoms into a crystal lattice
invariably increases the strength of the material. A huge
amount of work was done on the solid solution hardening
in the 1950s and 1960s. Dorn et al. performed a systematic
piece of work where they explored the effects of several
alloying elements on the properties of binary aluminium
alloys.[1,2] A classic paper on solution hardening is the
one by Fleischer and Hibbard,[3] while a more recent review
is given by Haasen.[4]
The solid solution hardening is a result of an interaction
between the mobile dislocations and the solute atoms. A
number of mechanisms have been suggested.[4,5] However,
the most relevant mechanisms for substitutional alloying of
aluminium are the elastic interactions due to (1) the size
ØYVIND RYEN and BJØRN HOLMEDAL, Postdoctoral Students,
OSCAR NIJS, Diploma Student, and ERIK NES, Professor, are with the
Department of Materials Science and Engineering, Norwegian University
of Science and Technology, Trondheim, Norway. Contact e-mail: [email protected]
material.ntnu.no EMMA SJÖLANDER, Diploma Student, is with the
Department of Physics, Royal Institute of Technology, Stockholm, Sweden.
HANS-ERIK EKSTRÖM, Manager, is with Sapa Technology, Finspong,
Manuscript submitted August 12, 2005.
misfit, where the size of the solute atom differs from the
size of the matrix atoms and creates a strain field around the
atom, and (2) the modulus misfit, where the difference in
binding force between the solute atoms and the matrix
atoms results in a hard or soft ‘‘spot’’ in the matrix. However the interaction works, the presence of solute atoms
increases the initial yield stress and reduces the dynamic
recovery rate of dislocations. This results in a higher dislocation density and a higher work hardening rate but also
a different dislocation structure, which is dealt with in
another paper.[6]
Several authors have reported correlations between the
flow stress and the alloy concentration of the type:
s 5 spure 1 Hcn
where spure is the flow stress of a pure metal and H and
n are constants. There seems to be general agreement that
for the yield stress of pure fcc substitutional solid solutions,
n is in the range 0.5 to 0.75.[3,4]
Al-Mg alloys are among the most widely used, and most
extensively studied, solid solution hardening alloys. As it is
possible to solve up to about 6 wt pct Mg in aluminum at
room temperature, high yield strengths and very high work
hardening rates can be achieved. At the same time the
elongation is not much affected by the Mg addition, making
them suitable for many applications. The results by Sherby
et al.[2] on high-purity Al-Mg alloys at room temperature
give a value of n 5 0.75 in Eq. [1], irrespective of strain, a
number that corresponds to the theoretical treatment of
Haasen.[4] The H value is about 15 MPa/(at. pct Mg)n at
the yield point.
Manganese is first of all added to aluminum to control
the recovery and recrystallization, and thus the final grain
size and texture in the annealed sheet. Together with other
elements of low solubility (iron, silicon), Mn creates coarse
constituent particles and small dispersoids that serve as
nucleation sites for recrystallization and obstacles against
grain boundary migration, respectively. But Mn and Fe and
Si also have a considerable effect on the strength when
VOLUME 37A, JUNE 2006—1999
present in solid solution. These effects are studied to a
much less extent than the effects of Mg. One obvious reason
for this is the low solubility of these elements in aluminum,
making it difficult to produce true solid solutions. However,
under certain processing conditions, supersaturated solid
solutions may be produced.
The reported results on the solution hardening effect of
Mn, Fe, and Si appear to be somewhat spread and inconsistent. According to Altenpohl,[7] Mn has a weaker effect
on both the yield strength and the tensile strength than an
equal amount of Mg (in wt pct). In contrast, several studies
report a strong effect of Mn on the undeformed strength but
a relatively weak effect on the work hardening.[8–11] The
same papers observe the opposite for Si: a small influence
on the yield strength but a large effect on the work hardening and ultimate tensile strength. Mahon and Marshall[9]
also found that very small amounts (up to 0.01 at. pct) of Fe
in solid solution give a considerable increase in strength.
In this work, a number of commercial and high-purity
alloys of various concentrations of Mg and Mn are investigated. The purpose is to explore the solid solution effects
of the main alloying elements, but also the effects from
trace impurities in solid solution such as Fe and Si.
Seven alloys of commercial-purity grades and five of
high-purity grades were investigated in this work (Table I).
All of them had a coarse, equiaxed grain structure. The
grain sizes were determined by linear intercept on the basis
of EBSD mapping with a step size much smaller than the
grain size. AA1050 and AA1200 represent commercially pure alloy variants with different levels of Fe and
Si content. The three Al-MgX alloys (where X 5 0.5, 1, and
3 wt pct Mg) were designed keeping the Fe and Si content
as close to the composition of AA1050 as possible. The
AA1050 alloy can thus be regarded as a Mg-free reference
material for the Al-MgX alloys. In addition, the high-Mg
alloy AA5182 was investigated, giving a wide span in Mg
concentrations. To study the effect of Mn, an AA3103 alloy
and five high-purity grades were included in the tests. All
the commercial alloys were delivered in the DC-cast condition, except for AA5182, which was homogenized and
hot rolled into a 25-mm transfer slab gauge. The DC-cast
variants had a random texture, and also the hot-rolled
AA5182 alloy had a nearly random texture with a weak
cube component (about two times random). The AA1050
and Al-Mg X alloys were given an industrial homogenization procedure in an air-circulating furnace (6 hours at
550 °C, slowly heated and cooled). This treatment has no
effect on the grain size.
After casting of AA3103, the amount of Mn in supersaturated solid solution Mn(ss) was calculated from electrical resistivity measurements to be 0.40 at. pct (0.81 wt pct),
with the rest present in intermetallic phases. Four different
homogenization procedures were carried out in an air-circulating furnace (Table II).[11,12,13] After such long homogenization times the smallest particles will dissolve, while the
solved Mn atoms will diffuse to the primary particles.
A fast air cooling ensures no reprecipitation of dispersoids.
By using this procedure it was possible to achieve in total
five alloys with different solute content but without small
dispersoids. After casting the particles were relatively small
and plate-shaped, mostly Al6 (Mn-Fe) and some a-Al(MnFe)Si. The homogenization treatment led to a spheroidization of the coarse constituent particles. The size distribution
of these was found to be approximately similar for the
conditions A, B, C, and D.[12] A phase transition from
Al6 (Mn-Fe) to a-Al(Mn-Fe)Si was observed but not completed during these homogenization treatments. Due to the
high common initial homogenization temperature (645 °C),
some grain growth occurred, resulting in a common grain
size in the as-homogenized variants of about 125 microns.
The AA1200 alloy was given the A-type homogenization
treatment to obtain a Mn-free reference material for the
various AA3103 homogenized conditions. This homogenization treatment resulted in only marginal changes in the
grain size compared to the as-cast condition.
Of the five high-purity Al-Mn variants, three contain an
addition of 0.02 wt pct Fe (Table I) to study the effect of
trace amounts of iron reported by Mahon and Marshall.[8]
Table I. Nominal Composition (in wt pct) and Grain Size of the Materials Investigated
Grain Size (mm)
to 1000
to 1000
to 1000
to 1000
to 1000
*Supplied by Pechiney, now Alcan-Voreppe, France. Other commercial alloys supplied by Hydro Aluminium, Sunndal, Norway.
The contents of the various trace elements (including Cu) in the commercial alloys were typically on the level 0.005 to 0.01 wt pct or less. The high-purity
alloys were prepared by Sapa, Finspong, Sweden.
2000—VOLUME 37A, JUNE 2006
Table II.
Homogenization Treatment and Solute Levels, Mn(ss), in the AA3103 Alloy Variants
Homogenization Before Quenching
Mn(ss), At. Pct
100 °C/h to 645 °C, held 24 h
As A + 20 °C/h to 607 °C, held 24 h
As B + 20 °C/h to 553 °C, held 24 h
As C + 20 °C/h to 500 °C, held 24 h, 20 °C/h to 453 °C, held 34 h
These alloys were prepared using 99.999 purity aluminum
supplied by Hydro Aluminium Deutschland. Melting was
done in an aluminum oxide crucible prevent impurities
from getting into the liquid metal. The melt was heated
to 840 °C and casted into block molds (6 3 16 3 2 cm).
The ingots were heat-treated for 24 hours at 630 °C to
obtain a homogenous distribution of the alloying elements.
The ingots were then cold rolled in a laboratory mill down
to 1 mm in thickness. Tensile specimens were machined
from the cold-rolled sheets and heat-treated at 630 °C for
2 hours and cooled in air to obtain a recrystallized structure
and to dissolve the alloying elements. Because of the metal
purity and the high annealing temperature, the grain size
approached the mm range.
Tensile testing of the as-cast and homogenized variants
was performed using cylindrical specimens measuring 5 or
6 mm in diameter and a parallel length of 30 mm. An MTS
880 servohydraulic testing machine was run under a constant ramp rate, giving a true initial strain rate of about
10$3 s$1. The AA5182 alloy was tested by Hydro Aluminium
Deutschland[14] using flat specimens with cross-section
25 3 7.7 mm2 and length 80 mm and a strain rate of
6 " 10$3 s$1. The tensile specimens of the high-purity alloys
were 12.5 mm wide and the gauge section was 75 mm long.
Fig. 1—True stress–strain curves from tensile testing of Al-Mg alloys. A
curve for the AA1050 alloy is included for comparison.
A. Magnesium in Solid Solution
True stress–strain curves from tensile tests of the
Mg-containing alloys, as well as AA1050, are shown in
Figure 1. As expected, the yield stress, the rate of work
hardening, and the ultimate tensile strength increase with
Mg additions. It is also evident from the curves of AA1050
and Al-Mg0.5 that small additions of Mg reduce the ductility, whereas a further increase in Mg content beyond
0.5 pct has only a minor influence. The flow curves of
the Al-Mg alloys are partly serrated, indicating a dynamic
strain ageing effect that increases with the Mg concentration.
The high-Mg AA5182 alloy also shows a yield point elongation due to Lüders banding.
Based on the stress–strain curves, the hardening from Mg
atoms in solid solution is estimated. For such a comparison
between alloys to be made, other factors that may influence
the strength must contribute equally. For instance, the grain
size effects are relatively large in Al-Mg alloys and increase
with Mg content.[15] However, this is due to the artificially
high yield strengths caused by the Lüders bands, and even
though the grain sizes in the present alloys are somewhat
different, they are all in a range where grain size effects
are expected to be small (.50 mm, Table I) and Lüdering
Fig. 2—Flow stress at various strain levels as a function of the Mg content
(at. pct).
generally does not occur. An exception is AA5182, and
therefore the initial yield strength of this alloy is omitted
from the analyses. An estimate for the expected variation in
the flow stress due to the variation in the grain size becomes
about 2 to 3 MPa or a few percent only for the Al-Mg alloys
VOLUME 37A, JUNE 2006—2001
listed in Table I (calculated on the bases of a Hall-Petch k #
0.08 MPam1/2). For an analysis of and further references
to the effect of grain size on the initial yield strength and
subsequent work hardening of aluminum and aluminum
alloys, see the recent work by Nes et al.[16] Furthermore,
the alloys have random initial textures, and the secondphase particle structures are assumed to be roughly similar,
at least in AA1050 and Al-MgX. AA5182 has a Mn content
of 0.27 wt pct that may give a different particle contribution
and cause some Mn in solid solution, so this comparison
must be considered with caution. A final assumption is that
all the Mg is in solid solution (i.e., that the primary particles
contain no Mg). This is supported by a work by Lloyd,[17]
where it was found that the particles in commercial
AA5xxx alloys contained Fe, Mn, and Cr in addition to
Al, but no Mg.
In Figure 2 the stress at various strain levels is plotted
against the concentration of Mg (in at. pct). The results fit
very well into the relationship given in Eq. [1] with an
exponent close to unity. The curves in the figure have been
fitted selecting only the as-cast alloy data (i.e., with Mg
content in the range 0 to 3.3 at. pct), since these alloys
are identical except for the Mg content. The values of the
fitted parameters spure, n, and H are given in Table III. The
exponent n is about 0.9 except for the lowest strain level,
where it has a level slightly above 1. The data for the
AA5182 alloy (5 at. pct) are about 5 to 10 MPa above these
best-fit lines, which is to be expected due to the 0.27 wt pct
Mn content in this alloy (see Section III–B). AA1050 is
here used as the pure metal (i.e., the values of spure should
correspond to the flow stress values of this alloy). At the
yield point e 5 0.002, this value of H simply reflects the
solute effect on the strength. For higher strains, however,
H reflects a combined effect of solutes and dislocations.
Therefore, the change in H seen in Table III indicates the
work hardening potential of this alloy system. As mentioned above, binary high-purity solid solutions generally
exhibit n values around 0.5 to 0.75, and n 5 0.75 can be
deduced from Sherby et al.[2] The present results, however,
are in better accordance with other observations on alloys
of commercial purity. Burger et al.[15] found that H 5 14.5
MPa/at. pct Mg at the yield strength, nearly the same as
found in this work. Based on this, it is suggested that by
increasing the impurity level in Al-Mg alloys, the hardening exponent n increases from about 2/3 in pure alloys to
about 1 in commercial alloys. The reason for this may
relate to a combined hardening from Mg with traces of
Fe, Si, and other elements in solution. As Eq. [1] then
changes from a parabolic to a linear relationship, there must
be a relative increasing effect of Mg as the concentration
increases. This can be explained in terms of a synergy or
clustering effect, as will be discussed in Section III–B.
B. Manganese in Solid Solution
Stress–strain curves of the five conditions of AA3103
(Table I) are shown in Figure 3(a). Here they are compared
to the Mn-free reference alloy AA1200, tested in the homogenized condition. It must be borne in mind, however, that
the strength of a commercial Al-Mn alloy, like the AA3103
alloy in this work, originates from a combined effect of
several factors: solid solution, dispersoids, and primary particles. Isolating one of these contributions is never easy, as
a thermo-mechanical treatment will change the distribution
of Mn atoms in several ways. The homogenization procedure of the present AA3103 alloy resulted in four different
solute levels, but also changes in the particle size and
volume fraction. Observations show that the variation in
particle structure is small, and consequently the particle
contribution is approximately equal for the four homogenized alloy variants A, B, C, and D.[12] Thus, the solute content is considered as the major difference between these
conditions. The as-cast condition, on the other hand, must
be treated carefully because its particle structure differs
from the homogenized conditions. Accordingly, to isolate
the strengthening effect due to Mn in solid solution only,
this investigation also includes testing of high-purity-grade
alloys. These alloys cover the same range in solid solution
contents as the commercial variants, and stress–strain
curves are shown in Figure 3(b). This figure also includes,
for comparison, the flow curve for the high-purity
The true stress at various strain levels is plotted in Figure
4(a) for the five conditions of AA3103, together with the
homogenized versions of the AA1200, and in Figure 4(b)
for the five high-purity alloy variants. It can be seen from
Figure 4(a) that the data points for the as-cast variant do not
fit in well with the other results for this alloy, which are
well fitted by a linear trend line. The as-cast values are
shifted systematically upward by about 10 MPa, an effect
primarily attributed to the much finer distribution of plateshaped constituent particles in this variant compared to the
homogenized conditions. Nearly linear trend lines gave the
best fit also for the high-purity alloys, for the two lowest
strain levels (Figure 4(b)). The values of the parameters in
Eq. [1] are given in Table 3 for the commercial alloy variants
Table III. Parameters in Eq. [1] as Determined from Figures 2 and 4
Mn (n 5 1)
High-Purity Mn
(exp) MPa
(exp) MPa
(exp) MPa
(At. Pct)$n
(At. Pct)$1
(At. Pct)n
The experimental flow stresses of 99.999 aluminium, AA1200, and AA1050 are shown for comparison.
2002—VOLUME 37A, JUNE 2006
Fig. 3—(a) True stress-strain curves from tensile testing of the AA3103alloys variants and the AA1200 alloy. (b) Stress-strain curves for the high
purity AlMn alloys.
Fig. 4—Flow stress at various strain levels as a function of the amount of
Mn in solid solution (at. pct) (a) for the various conditions of the AA3103
alloy and (b) for high-purity Al-Mn alloys.
and for the two lowest strain levels for the high-purity
grades. For the latter alloys it was not meaningful to fit
the result to Eq. [1] for the higher strain levels because of
extensive dynamic recovery in the purity aluminium. It
follows from the results in Figure 4 and Table III that the
solute strengthening rate is about 50 to 60 pct stronger in
the commercial alloy variants compared to the high-purity
grades. The flow stress levels, however, for the two variants
in Figure 4 are widely different. It was somewhat unexpected that the strengthening due to an increasing amount
of Mn is nearly linear (n 5 0.9) in the high-purity alloys,
since a parabolic effect with n 5 0.75 was found for the
high-purity Al-Mg grades studied by Sherby et al.[2] However, better statistics in terms of more data points for the
Al-Mn case would be required to make conclusions as to
the precise value of this n exponent as illustrated by the
broken line in Figure 4(b), fitted using n 5 0.75.
The hardening coefficient H of the AA3103 alloy is seen
to be considerably larger than for the Al-Mg alloys (Figure
5 and Table III). The slope of the yield stress vs concentration line for the Al-Mn alloy variants in Figure 5 is a
factor of about 4 to 5 larger than for the Mg variants. A corresponding comparison between the results from the Al-Mn
high-purity grades with similar plots constructed on the
basis of results reported by Sherby et al.[2] for high-purity
Al-Mg alloys also revealed a stronger strengthening effect
due to Mn than Mg. The difference, however, was a factor
of 2 smaller than for the commercial grades. Another
important difference is that while the H parameter for the
commercial-purity Al-Mg alloys increased strongly with
strain over the entire strain range, this variation is much
less for AA3103 beyond a strain of 0.025. It is clear from
the present results that Mn in solid solution has a much
VOLUME 37A, JUNE 2006—2003
Fig. 5—Yield strength (Rp0.2) as a function of Mg or Mn in solid solution,
demonstrating the large difference between these two alloy systems.
Fig. 6—Work hardening rate ds/de as a function of solute content for
various strain levels for the AA3103 variants, and at e 5 0.01 for the
high-purity AlMn alloys.
stronger effect on the initial yield stress than Mg. It was
suggested by Sanders et al.[9] and others[8,10] that Mn has a
lower effect than Mg on work hardening. However, for the
commercial grades, it is here evident that the opposite is
true, as illustrated in Figure 6, which shows that the initial
hardening rate (ds/de)e 5 0.002 as a function of solute concentration is about a factor of 3 larger for the AA3103 alloy
variants than for the Al-Mg alloys. With increasing strain,
the work hardening rates for the two alloy systems merge,
reaching the same levels at a strain of 0.1. It follows that the
initial higher hardening rate in the Al-Mn alloy is being
balanced by a correspondingly higher dynamic recovery
rate compared to the AlMg alloys. No similar strong initial
work hardening effect was found for the high-purity alloys,
2004—VOLUME 37A, JUNE 2006
as shown in Figure 6. In comparing work hardening rates
between commercial and high-purity grades, the presence
of the constituent particles in the former will play a role,
but not a significant one. The major difference, for instance,
in comparing work hardening in the high-purity alloys to
that in AA1200 and AA3103 is due to the enhanced
dynamic recovery rate in the high-purity material. It is
difficult to find any other explanation for this difference
than that it must be caused by the suppression of dynamic
recovery in the commercial alloys due to the additional
presence of about 0.15 at. pct of trace atoms (mostly Si)
in solid solution.
Since the Mn atoms have a much stronger effect on the
yield strength than the Mg atoms in both the high-purity
and commercial grades, one might expect these atoms to
have highly different properties in the Al lattice—for
instance, a larger size misfit or modulus misfit. However,
in this comparison it is the Mg atoms that have the largest
size misfit to the Al lattice,[18] and accordingly one should
anticipate a smaller interaction force between dislocations
and Mn atoms than Mg atoms. However, neither Dorn
et al.[1] or Doherty and McBride[19] found any correlation
between strength and atomic misfit in their studies of many
alloy systems. The modulus misfit, relating to the binding
strength between the solute atoms and the aluminum
matrix, will be different for Mn and Mg atoms. This may
explain the observed difference between the high-purity
grades, but not the even stronger effect of Mn compared
to Mg in the commercial alloys. There are good reasons to
believe that another effect must be present that can explain
these strong hardening effects due of Mn in solid solution
in the commercial alloy systems.
An explanation may lie in a synergy or clustering effect.
The idea behind this is that in the commercial grades,
(trace) solute atoms (Fe, Si, Ti, Cu, and others) cluster
together with the Mn atoms and create harder spots in the
alloy than the discrete solute atoms do. Nonrandom distributions of solutes in alloys are mentioned by Haasen,[4] but
very little attention has been paid to this phenomenon in
non–heat-treatable alloys, which is easy to understand, considering the small dimensions of such clusters and the difficulty of detecting them. Clustering of solute atoms in the
early stages of precipitation in, for instance, Al-Mg-Si
alloys is a widely known phenomenon,[20] and the observation of a relatively constant contribution to the stress by
increasing the Mn content (see Figure 3(a)) corresponds
well with the strengthening from clusters in heat-treatable
alloys resulting from room-temperature aging.
The strongly enhanced initial work hardening rate (in
AA3103) caused by small amounts of Mn in solid solution
also points in the direction of a synergy effect. The work
hardening rate is controlled by the dislocation slip length
(i.e., the average distance a mobile dislocation travels from
the source to becoming stored in the substructure). This slip
length will be affected by the presence of atoms in solid
solution, as shown in the paper by Ryen et al.[6] on the work
hardening of Al-Mg alloys. However, the Mn effect on work
hardening revealed by Figure 6, which is three times stronger than that due to Mg, is difficult to understand unless the
migrating dislocations were exposed to Mn-induced pinning sites impeding the migration of dislocations even
stronger than individual atoms do.
It follows from the results in Figure 3 that both the initial
strength and the work hardening rate of the high-purity
Al-0.5 at. pct Mn alloy are lower than those for the commercial-purity AA1200 alloy. This implies that small
amount of elements in solid solution in AA1200 (about
0.15 at. pct of Fe, Si 1 other trace elements, of which Si
accounts for more than half) had a stronger effect on
strength and work hardening than the 0.5 at. pct of Mn in
the binary Al-Mn alloy. An additional trace amount of iron
did not cause any synergy effect on either strength or work
hardening rate in the high-purity Al-Mn alloys, as shown by
Figures 3(b) and 4(b). Iron is in chemical terms similar to
manganese and probably will affect strength and work
hardening in a similar way. A natural conclusion becomes
that the strengthening of the AA1200 compared to the highpurity Al-Mn grades is the result of clustering between
trace elements in solid solution. An interesting speculation
becomes synergy effects or clustering between Fe (and
possibly other trace elements) and Si due to the high diffusivity of silicon in aluminum and its dominating presence
in solid solution in commercial non–heat-treatable alloys,
compared to other trace elements. Similarly, in the 3xxx
series alloys, the formation of Mn-containing solute clusters is assumed to be aided by the presence of most notably
Si (which is also suggested by Kenawy et al.[21]). It is
obvious that the clusters are too small to be detected in
electron microscopes, even with high-resolution instruments.
Recently, alloys with scandium, zirconium, and hafnium
have been investigated using 3D atom probe techniques,
finding evidence of clustering of solutes and a corresponding increase in strength.[22] A similar technique would be of
great interest for the alloys investigated in the present work.
The problem is to understand the mechanism behind this
synergy or cluster formation. Cold deformation of aluminum alloys will introduce a high density of dislocations
and subgrain boundaries. These lattice defects may serve
as high-diffusivity paths for alloying elements, and one can
imagine that if clustering of slowly diffusing elements
occurs, the process will be strongly enhanced by deformation. Therefore, some additional experiments were performed. The 1050-alloy and the 3103-B condition were
cold rolled to strains of e 5 3 (95 pct reduction) and then
isothermally annealed in an oil bath of 100 °C to 200 °C.
Hardness measurements were conducted after various
annealing times. Figure 7 shows that the hardness of
cold-rolled 3103 actually increases by approximately
10 pct during the first 15 minutes of annealing at 160 °C.
Thereafter, the strength decreases due to the recovery of
dislocations. A similar behavior is seen also at 100 °C
and 200 °C, with some difference in reaction time. As
shown in Figure 7, the cold-rolled 1050 alloy also has an
age-hardening potential. The hardness increases only 2 pct,
and softening occurs earlier, probably because of less solute
and less resistance against recovery than in 3103. The same
hardening phenomenon was also observed by Sæter[23] in
an Al-0.9Mn-0.14Fe alloy. He studied the samples in TEM,
and after finding nothing in the microstructure that could
cause the hardening, he suggested that some very small preprecipitate clusters might be present. This correlates well
with the above discussion, and the strengthening can be
explained by the formation of clusters. The picture resembles the ageing characteristics of age-hardenable alloys, but
Fig. 7—Change in hardness with isothermal annealing at 160 °C of 1050
and 3103 cold-rolled 95 pct. Hardness as cold rolled: 1050, 46.3 VHN;
3103, 64.6 VHN.
in the present case there is a balance between formation of
clusters and recovery of dislocations.
Finally, although arguments have been presented in support of a synergy or clustering mechanism for explaining
work hardening of commercial 3xxx alloys, no such clusters have so far been detected experimentally by direct
imaging. The main argument against such a mechanism
rests on the low diffusivity of Mn, even along dislocation
pipes, at room temperature. One should bear in mind, however, that the extrapolation of diffusivity values from higher
temperatures (where they are measured) over several hundred degrees toward diffusion in the ambient temperature
range is not straightforward. Besides that, the expected
drastic reduction in diffusivity will be accompanied by a
corresponding increase in supersaturation and an associated
increase in the chemical potential for cluster formation. In
our view, the most important argument given for a cluster
effect in commercial alloys is that this strengthening due
to manganese requires the simultaneous presence of Si (and
possibly other trace elements) in solid solution. It seems
very difficult to understand such a synergy effect between
different solute atoms in ways other than in terms of a
dislocation-core–stimulated interaction between solute elements where the reaction product must be some sort of
solute–atom cluster. However, more work is required to
properly understand this effect.
In this work it is seen that magnesium in solid solution
gives a near-linear concentration dependence of the strength
at a given strain for commercial alloys. A characteristic
effect caused by increasing the Mg content is a relatively
moderate effect on the initial yield stress and a strong effect
on the work hardening rate. Manganese in solid solution
also gives rise to a nearly linear concentration dependency.
In this case, however, the hardening contribution is much
stronger than in the Al-Mg case, both in high-purity and
commercial alloys. Another characteristic of the commercial
VOLUME 37A, JUNE 2006—2005
Al-Mn alloy variants is that the initial work hardening rate
increases at a much higher rate due to increases in the solid
solution content than that found for AlMg. These strong
Mn effects in commercial alloys are partly attributed to a
synergy effect between Mn and trace elements in solid
This research was carried out as part of the EC Fifth
Framework project VIRFORM, Contract No. G1RD-CT1999-00155. Funding by the European Community and
the industrial partners is gratefully acknowledged. Thanks
also to Professor J.D. Embury for many interesting discussions.
1. J.E. Dorn, P. Pietrokowsky, and T.E. Tietz: J. Metals, 1950, vol. 188,
pp. 933-43.
2. O.D. Sherby, R.A. Anderson, and J.E. Dorn: J. Metals, 1951, vol. 189,
pp. 643-52.
3. R.L. Fleischer and W.R. Hibbard: in The Relation Between the Structure and Mechanical Properties of Metals, Her Majesty’s Stationary
Office, London, 1963, pp. 262-94.
4. P. Haasen: in Physical Metallurgy, 4th ed., R.W. Cahn and P. Haasen,
eds., Elsevier Science BV, 1996, pp. 2009-68.
5. G.E. Dieter: Mechanical Metallurgy, 2nd ed., McGraw-Hill Book Co.,
London, 1988, pp. 205-06.
6. Ø. Ryen, H.I. Laukli, B. Holmedal, and E. Nes: Metall. Mater. Trans.,
2006, vol. 37A, pp. 2007-14.
7. D. Altenpohl: Aluminium und Aluminiumlegierungen, SpringerVerlag, Berlin, 1965, pp. 689-90.
8. G.J. Mahon and G.J. Marshall: J. Metals., 1996, vol. 48, pp. 39-42.
2006—VOLUME 37A, JUNE 2006
9. R.E. Sanders, S.F. Baumann, and H.C. Stumpf: in Aluminum Alloys:
Their Physical and Mechanical Properties, vol. III, E.A. Starke and
T.H. Sanders, eds., Engineering Materials Advisory Services, Warley,
1986, pp. 1441-84.
10. J.P. Suni, R.T. Shuey, and R.D. Doherty: in Aluminum Alloys: Their
Physical and Mechanical Properties, vol. I, T.H. Sanders and E.A.
Starke, eds., Georgia Institute of Technology, Atlanta, 1994, pp.
11. K. Sjølstad: Ph.D. Thesis, Norwegian University of Science and Technology, Trondheim, 2003.
12. S. Tangen: Ph.D. Thesis, Norwegian University of Science and Technology, Trondheim, 2004.
13. S. Tangen, K. Sjølstad, E. Nes, T. Furu, and K. Marthinsen: in Aluminium Alloys: Their Physical and Mechanical Properties, P.J. Gregson
and S.J. Harris, eds., Trans Tech Publications, Switzerland, 2002, pp.
14. O. Engler: Hydro Aluminium Deutschland, Bonn, Germany, unpublished research, 2002.
15. G.B. Burger, A.K. Gupta, P.W. Jeffrey, and D.J. Lloyd: Mater.
Charact., 1995, vol. 35, pp. 23-39.
16. E. Nes, B. Holmedal, E. Evangelista, and K. Marthinsen: Mater. Sci.
Eng. A, 2005, vol. 410–411, pp. 178-82.
17. D.J. Lloyd: Metall. Trans. A, 1980, vol. 11A, pp. 1287-94.
18. G.H. Aylward and T.J.V. Findlay: SI Chemical Data, 2nd ed., John
Wiley & Sons, Milton, Australia, 1974, pp. 6-13.
19. R.D. Doherty and J. McBride: in Aluminum Alloys for Packaging,
J.G. Morris, H.D. Merchant, E.J. Westerman, and P.L. Morris, eds.,
Minerals, Metals and Materials Society, Warrendale, PA, 1993, pp. 34768.
20. A. Edwards, K. Stiller, G.L. Dunlop, and M.J. Couper: Acta Mater.,
1998, vol. 46, pp. 3893-904.
21. M.A. Kenawy, G. Graiss, G. Saad, and A. Fawzy: J. Phys. D Appl.
Phys., 1987, vol. 20, pp. 125-29.
22. B. Forbord, W. Lefebvre, F. Danoix, H. Hallem, and K. Marthinsen:
Scripta Mater., 2004, vol. 51, pp. 333-37.
23. J.A. Sæter: Ph.D. Thesis, Norwegian University of Science and Technology, Trondheim, 1997.