Structure and properties of Al–Mg mechanical alloys

Mirko Schoenitz and Edward L. Dreizin

New Jersey Institute of Technology, Department of Mechanical Engineering, University Heights,
Newark, New Jersey 07102

(Received 29 January 2003; accepted 7 May 2003)

Mechanically alloys in the Al–Mg binary system in the range of 5–50 at.%
Mg were produced for prospective use as metallic additives for propellants
and explosives. Structure and composition of the alloys were characterized by
x-ray diffraction microscopy (XRD) and scanning electron microscopy.
The mechanical alloys consisted of a supersaturated solid solution of Mg in the
␣ aluminum phase, ␥ phase (Al12Mg17), and additional amorphous material.
The strongest supersaturation of Mg in the ␣ phase (20.8%) was observed for
bulk Mg concentrations up to 40%. At 30% Mg, the ␥ phase formed in quantities
detectable by XRD; it became the dominating phase for higher Mg concentrations.
No ␤ phase (Al3Mg2) was detected in the mechanical alloys. The observed Al
solid solution generally had a lower Mg concentration than the bulk composition.
Thermal stability and structural transitions were investigated by differential scanning
calorimetry. Several exothermic transitions, attributed to the crystallization of ␤ and
␥ phases were observed. The present work provides the experimental basis for the
development of detailed combustion and ignition models for these novel energetic
materials.

I. INTRODUCTION Recent observations suggest that ignition and combus-

Metals with high volumetric and gravimetric combus-
tion enthalpies are of interest as materials with high
energy densities. Aluminum, in particular, and aluminum-
based alloys have been investigated as additives in vari-
ous fuel formulations for propellants, explosives,
incendiaries, or pyrotechnics.1–3 Practical applications
are limited, however, since chiefly kinetic obstacles such
as long ignition delays and slow burning rates prevent the
theoretical combustion enthalpies from being fully ex-
ploited at the specific reaction rates required. Based on
recent research that linked internal phase transformations
to macroscopic events during combustion of metal par-
ticles, interest in metastable intermetallic materials has
developed.4
The present paper discusses supersaturated solid solu-
tions in the aluminum-magnesium binary system. Several
reports in the combustion literature indicate performance
advantages of Al–Mg alloys. Magnesium has been of
particular interest as an alloying component for its en-
hanced ignitability.5 Most experimental work however
was conducted on Al–50% Mg alloys, which exhibit
a significantly reduced specific combustion enthalpy
compared to pure Al. Materials for energetic applications
should therefore be tailored to balance the high specific
combustion enthalpy of the Al component against the
enhanced ignitability of the Mg component.
tion of aerosols of Al–Mg mechanical alloys are signifi-
cantly enhanced compared to equilibrium alloys or
blends of elemental powders. Metastable Al–Mg alloys
with Mg concentrations ranging from relatively low val-
ues of 5 to 50 at.% showed increased burning rates and
lower ignition temperatures.6,7
Detailed characterization of phase relations and the
degree of metastability in metastable Al–Mg alloys are
needed to attempt performance optimization of these ma-
terials as advanced energetics. This information is also
necessary to model processes of combustion of metallic
fuel particles, specifically to quantitatively predict igni-
tion behavior as a function of heating rate and of the
particles’ environment.8 The purpose of the present re-
search is to characterize mechanical alloys in the system
Al–Mg with the specific goal to optimize materials for
combustion applications and to address issues arising
for large-scale production.
II. BACKGROUND
Phases in the binary Al-Mg system relevant to the
present work include the ␣ Al, ␤ (Al3Mg2), and ␥
(Al12Mg17) phases. Under equilibrium conditions, the
concentration of magnesium in the ␣ aluminum structure
is 0.21 at.% at room temperature; maximum solubility at
450 °C has been reported as 16.23%,9 or 18.6%.10 The

J. Mater. Res., Vol. 18, No. 8, Aug 2003 © 2003 Materials Research Society 1827
 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys
compositional range of the ␤ phase is given as 38.5–
40.3% Mg in Ref. 11, while the equilibrium Mg concen-
tration of the ␥ phase ranges from 45% to 60% at its
greatest extent at about 440 °C.11
Metastable intermetallic materials are readily pro-
duced by mechanical alloying (MA). MA, a non-
equilibrium powder metallurgical method to prepare
highly metastable materials, was first devised by Benja-
min,12,13 and has since found wide application for the
production of structural (e.g., oxide dispersion hardened)
materials, or mechanically activated materials with in-
creased reactivity. An extensive review was given re-
cently by Suryanarayana.14
Structure and stability of a set of mechanical alloys in
the system Al–Mg were reported previously for compo-
sitions with 40, 60, and 80 at.% Mg15 and 30 and 50 at.%
Mg.16 Supersaturated ␣ phase and ␥ phase have been
found to form by MA, while in contrast, ␤ phase never
forms. Amorphous material was inferred from mass bal-
ance considerations.
Systematic information on phase makeup and the de-
gree of supersaturation of the ␣ phase specifically in
mechanical alloys with low Mg concentration is lacking.
Reliable thermodynamic characterization of subsolidus
reactions occurring in Al–Mg mechanical alloys on heat-
ing is also not available. The specific properties of the
prepared mechanical alloys generally depend on bulk
composition as well as on MA process parameters. Sev-
eral types of alloying apparatus are currently in use (e.g.,
shaker mills, planetary mills, attritor mills), yielding dif-
ferent results and suitable for alloy preparation on vary-
ing scales. At present, it is not clear whether the same
mechanical alloy can be produced using different types
of mills by appropriate adjustment of process parameters.
A thorough understanding of the relations of alloy prop-
erties and process parameters is therefore needed to iden-
tify alloying conditions required to obtain consistent
results with different devices.
III. EXPERIMENTAL
A. Mechanical alloying
A set of samples with bulk compositions of 5, 10, 20,
30, 40, and 50 at.% Mg were prepared in a SPEX 8000
shaker mill (Spex Certiprep, Metuchen, NJ). Mechanical
alloying is sensitive to a set of process parameters such as
the degree of filling of the milling vials, the size and size
distribution of the milling medium, and the presence of
process control agents (PCA) that balance cold welding
against fracturing and grinding. In the present study,
milling was carried out until the mechanical alloy
showed no structural or compositional changes with in-
creasing milling time. Optimum milling parameters re-
quired to achieve this stationary state in a minimum
amount of time were determined in a series of separate
1828 J. Mater. Res., Vol. 18, No. 8, Aug 2003
experiments before actual sample preparation was
begun. Starting materials were elemental powders of Al
(99.8% −100 mesh) and Mg (99% −325 mesh). For syn-
thesis, 10 g initially unmixed metal powder were milled
in zirconia vials under argon atmosphere with 2 wt.%
stearic acid [CH3–(CH2)16–COOH, 99%] added as PCA.
Zirconia balls of 10 mm diameter and a total mass of
50 g were used as the milling medium, giving a ball-to-
powder weight ratio (BPR) of 5. For comparison, a lim-
ited set of samples was milled with 2 g of powder, 26 g
of balls (BPR 13), and 4% PCA. Samples were milled
for a total time of 12 h each. The sample temperature rose
to 50–60 °C during milling. Moderate sample caking
posed a problem for milling effectiveness; milling for
extended periods of time also limits the useful lifetime of
the mill. Therefore, milling was carried out in intervals
of 180 min, after which the mill was allowed to cool to
room temperature, and caked samples, if present, were
manually broken apart in a glove box under protective Ar
atmosphere.
Milling progress was monitored for an individual alloy
(30% Mg) in a separate experiment by sampling the al-
loying product in 180-min intervals. Milling was carried
out on a mill with two stations, labeled A and B for
reference. During the first 15 h of milling, only the vial
on station A was sampled, while the vial on station B was
left unopened. Vial B was then sampled from 18 to 30 h,
and vial A was sampled again after 33 and 36 h. The final
sample, after 39 h of milling was taken from both vials.
To avoid systematic bias in the milling parameters, only
minute quantities (50 mg) were removed.
B. Sample characterization
Morphology and compositional homogeneity of the
mechanical alloys were examined on an Electroscan en-
vironmental scanning electron microscope (ESEM) op-
erated at 20 kV. The ESEM is equipped with a Kevex
energy dispersive x-ray spectrometer (EDX) that allows
standardless compositional analysis with a maximum
spatial resolution of approximately 0.1 ␮m.
Composition of the ␣ phase was determined for each
mechanical alloy by x-ray powder diffraction (XRD) on
a Bragg–Brentano type Philips X’pert MRD diffractom-
eter operated at 45 kV and 40 mA, using Cu K␣ radiation
(␭ ⳱ 1.5438 Å). Crystallite size and lattice strain/
compositional broadening of the supersaturated solid so-
lution were determined using the Williamson–Hall
method.17
Temperature dependent structural transformations
were observed by differential scanning calorimetry
(DSC) using a TA Q100 series calorimeter as well as
a Netzsch Simultaneous Thermal Analyzer STA409-PC
(Netzsch, Germany). For the Q100, samples were con-
tained in crimped aluminum pans, and the calorimeter
atmosphere was continuously flushed with a 50 ml/min
 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

flow of nitrogen. Aluminum sample pans were not noticeable asymmetry, and a broad and weak peak ap-
crimped for the STA409-PC; the furnace was evacuated pears around 36.2°, the position of the strongest reflec-
to approximately 0.3 bar and backfilled with Ar (Mathe- tion of the ␥ phase (Al12Mg17). At Mg concentrations
son, ultrahigh purity) three times before each measure- higher than 30%, the main peaks of the strongly super-
ment. During measurements, the furnace was flushed saturated ␣ phase are no longer observed, while peaks of
with Ar at approximately 10 ml/min. DSC traces were the ␥ phase become more pronounced. The asymmetry
recorded between room temperature and 420 °C with of the peaks of the ␣ phase develops into a separate peak
heating rates ranging from 3 to 25 K/min. close to the position of the equilibrium ␣ phase.
The back-transformation of the mechanical alloys to The lattice parameter a0 of the ␣ phase after 12 h of
thermodynamically stable phase assemblages occurred milling was refined from the positions of all x-ray dif-
over a wide temperature range (Sec. IV. B). To relate fraction lines from (111) to (422). The results are shown
events observed by DSC to structural transitions, DSC
runs of Al0.7Mg0.3 and Al0.9Mg0.1 were interrupted at
intermediate temperatures, and XRD patterns of these
partially relaxed mechanical alloys were collected. The
effective quench rates achieved in these experiments
were relatively low—about 20 K/min.
For all alloys, a second heating cycle was carried out
after each measurement to serve as baseline correction
for the measured trace. Poor thermal contact within
loosely packed powders with small particle size as well
as particle coarsening and sintering at elevated tempera-
tures tend to give rise to spurious baseline effects. To
unambiguously distinguish genuine contributions of the
samples from baseline effects related to sample prepara-
tion, a sample was annealed in the DSC at an interme-
diate temperature, cooled to room temperature, and then
reheated in the DSC to the maximum temperature.
Particle size distributions of the bulk mechanical al-
loys were measured by low-angle laser light scattering
using a Beckman Coulter LS 230 Enhanced Laser Dif-
fraction particle size analyzer. Suspensions for particle
size analysis were prepared in ethylene glycol to prevent FIG. 1. Backscattered SEM image of a polished section of an
reaction of the mechanical alloys with water. Al0.7Mg0.3 mechanical alloy. Bright particles are ZrO2 detritus from
the milling vial.

IV. RESULTS
A. Structure and composition of
mechanical alloys
SEM analysis showed that all mechanical alloys pro-
duced in the present study are chemically homogeneous
on a scale of 艋0.1 ␮m, the maximum resolution of the
instrument. EDX analyses were only able to resolve
the bulk Mg composition of the respective mechanical
alloys regardless of the presence of multiple phases ob-
served by XRD. Figure 1 shows a polished section of a
mechanical alloy with composition Al0.7Mg0.3. Indi-
vidual grains are irregularly shaped; varying degrees of
porosity are observed.
Figure 2 shows a section of the diffraction patterns of
all mechanical alloys. Only peaks of the ␣ phase are
FIG. 2. X-ray diffraction patterns of Al–Mg mechanical alloys with
observed up to a magnesium concentration of 30%, shift- varying composition. Reference peak positions of the pure ␣ phase are
ing to lower angles with increasing Mg concentration. shown as vertical lines. The separately marked peaks coincide roughly
Above 20% Mg, the peaks of the ␣ phase develop a with diffraction lines of the ␥ phase (Al12Mg17).

J. Mater. Res., Vol. 18, No. 8, Aug 2003 1829

 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

in Table I and Fig. 3. The corresponding Mg concentra- the ␥ phase peaks have nearly disappeared. The peak
tion of the ␣ phase (xMg) was calculated according to positions of the (111) peak of the ␣ phase are shown in
previously reported linear relations that were established Fig. 5(A). To quantify peak asymmetry, a split Pearson
based on rapidly quenched alloys:18,19 VII function20 was fit to the (111) peak, and integral peak
widths (F) were determined for low-angle and high-angle
x␣Mg[mol%] ⳱ 2.234(␣0 − 4.049Å) ± 0.003 . (1)
portions of the peak, respectively. The peak asymmetry
The equilibrium temperature Teq corresponding to
each Mg concentration was calculated from the pub-
lished relation of the maximum equilibrium Mg concen-
tration of the ␣ phase versus temperature.9 The molar
fraction of ␣ phase in each mechanical alloy was calcu-
lated from compositional balance and is given as x␣ in
Table I. Also shown in Table I is the average crystallite
size t of the ␣ phase derived from experimental peak
widths, as well as the apparent lattice strain ⌬a0 /a0. This
value includes any peak broadening due to composi-
tional spread.
The structural development as a function of milling
time of a mechanical alloy with composition Al0.7Mg0.3
is shown in Figs. 4 and 5. Figure 4 shows the XRD
patterns of the alloy along with the pattern of the starting
powder blend. Figure 5 shows the development of the
position and the asymmetry of the (111) peak of the ␣ FIG. 3. Lattice parameter a0 of the ␣ phase as a function of bulk Mg
phase over time. Samples were taken from both stations concentration. The solid line gives the relation of a0 versus Mg content
of rapidly quenched alloys reported by Luo et al.19 The data points up
on a double-mill (A and B). From Fig. 4, it can be seen to 30% Mg were obtained from alloys milled with 2% PCA and a BPR
that after only 3 h of milling the amount of pure Mg has of 5, while the value at 40% Mg was obtained from an alloy milled
decreased below the detection limit, while peaks consis- with 4% PCA and a BPR of 13.
tent with the ␥ phase have appeared. The width of the
peaks of the ␣ phase has increased, but their position has
shifted only slightly. This peak shift is more significant
after 6 and 9 h, and the peaks of the ␥ phase have become
wider and weaker. After 12 h, no further change in the
position of the peaks of the ␣ phase is observed, and

TABLE I. Lattice parameter of the highly supersaturated ␣ phase in

Al–Mg mechanical alloys and its calculated Mg concentration. Syn-
thesis conditions were 10 g powder load, BPR 5, and 2% PCA, except
for Al0.6Mg0.4 where 2 g powder, BPR 13, and 4% PCA were used.

xMg a0 xMg Teq x␣ t ⌬a0 /a0

(mol%) (Å)a (mol%)b (°C)c (mol%)d (nm)e (%)f
0 4.0493 (4) 0 100 365 (+) 0.09 (3)
5 4.0601 (1) 2.4 (1) 211 95.3 46 (8) 0.42 (4)
10 4.0763 (1) 6.0 (1) 304 92.4 41 (7) 0.46 (4)
20 4.1002 (6) 11.4 (2) 390 81.7 114 (+) 1.0 (1)
30 4.122 (1) 16.3 (3) 450 67.6 40 (19) 1.1 (1)
40 4.142 (1) 20.8 (3) ⭈⭈⭈ 49.3 51 (16) 1.2 (1)
50g 4.1724 27.6 27.8
a
The error in the last digit is shown in parentheses.
b
Composition of the ␣ phase calculated according to the data reported in
Refs. 18,19.
c
Corresponding equilibrium temperature according to Ref. 9.
d
Fraction of ␣ phase in the mixture assuming that bulk composition FIG. 4. X-ray diffraction patterns of the Al0.7Mg0.3 mechanical alloy
is balanced by ␣ phase (xMg calculated from a0) and ␥ phase (xMg ⳱ as a function of milling time. The triangle marks peaks of the ␥ phase.
58.6 mol%) only. The XRD pattern of the raw powder blend is shown at the bottom.
e
Average crystallite size; uncertainties greater than 50% are marked as (+). Milling was carried out in a double-mill with two milling stations,
f
Apparent lattice strain. labeled A and B for reference. The milling vial from which the re-
g
Extrapolated value. spective samples were taken is indicated next to the milling time.

1830 J. Mater. Res., Vol. 18, No. 8, Aug 2003

 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

was then expressed as the ratio of Flow /(Fhigh + Flow). compositions Al0.8Mg0.2 and Al0.7Mg0.3. The initial en-
Accordingly, values greater or less than 0.5 correspond dothermic peak is followed immediately by a broad exo-
to peaks skewed to the right or left, respectively. This thermic peak with a maximum around 170 °C for alloys
value is shown in Fig. 5(B). While the peaks of the ␣ with less than 40% Mg. The magnitude of this exother-
phase from vial A are nearly symmetric at 9–15 h, the mic peak roughly correlates with the magnitude of the
respective peaks from vial B after 18 h are split into a preceding endothermic peak. At temperatures above 250 °C,
strong peak at slightly lower angles and a weaker peak a series of exothermic peaks is observed. Peak tempera-
closer to the position of pure aluminum. Although tures and tentative structural assignment for all observed
this peak split becomes less pronounced as milling peaks (Sec. V) are shown in Table II.
progresses, the peak asymmetry remains. At the end of
this experiment after milling for 39 h, it can also be seen
that the peaks from vial A develop asymmetry later on
during the time that vial A was left unopened.
Mechanical alloys produced under different milling
conditions are compared in Fig. 6. An alloy with com-
position Al0.6Mg0.4 was milled for 12 h with a BPR of 13
and 4 wt.% PCA (compare to Fig. 2). The result differs
strongly from the alloy milled with a BPR of 5 and
2% PCA. The higher BPR promotes the formation
of a strongly supersaturated ␣ phase, indicated by the
large peak shift. A lower BPR favors the formation
of ␥ phase while compositional balance is maintained by
␣ phase with low Mg concentration.

B. Thermal stability
A set of baseline-corrected DSC traces for mechanical
alloys recorded at a heating rate of 15 K/min is shown
in Fig. 7. All mechanical alloys show an endothermic
peak near 100 °C. This peak is particularly strong for
FIG. 6. Comparison of Al0.6Mg0.4 mechanical alloys prepared under
different milling conditions. Both alloys were milled for 12 h. Milling
with a higher BPR results in a more strongly supersaturated ␣ phase,
while a lower BPR results in the formation of ␥ phase (separately
marked peaks).

FIG. 5. Position and asymmetry of the (111) peak of the ␣ phase in the
XRD pattern of an Al0.7Mg0.3 mechanical alloy as a function of mill-
ing time. Asymmetry values above 0.5 denote left-skewed peaks, val-
ues less than 0.5 denote right-skewed peaks. Milling was carried out in
a double-mill with two milling stations, labeled A and B for reference.
FIG. 7. DSC traces of Al–Mg mechanical alloys with varying com-
position. Baseline correction was carried out by subtracting the signal
of a second heating from the initial scan.

J. Mater. Res., Vol. 18, No. 8, Aug 2003 1831

 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

All mechanical alloys show a pronounced exothermic immediately following exothermic peak increase until
background above 150 °C. In Fig. 8, two scans of 12 h of milling and remain relatively constant thereafter.
Al0.8Mg0.2 mechanical alloy are compared. For refer- The exothermic peak at approximately 260 °C is stron-
ence, one sample was heated over the entire temperature gest at short milling times, decreasing as milling
range. The second sample was annealed at 160 °C for
1 h prior to heating—close to the peak temperature of the
first exothermic peak. As this sample is heated, initially
only minor baseline drift is observed. Above 200 °C,
however, the exothermic shift of the signal rapidly ap-
proaches that of the reference sample until the two scans
coincide within the reproducibility of the measurement.
Thus, the exothermic bias is caused by transformations
within the mechanical alloy; it is not an artifact intro-
duced by sample geometry.
Figure 9 shows XRD patterns of samples of the
Al0.7Mg0.3 mechanical alloy as milled, and heated at
15 K/min to 100, 150, 200, 300, and 400 °C, respec-
tively. The temperature resolution in this set of experi-
ments is low; however, the diffraction patterns are still
instructive. The peaks of the ␣ phase start to shift back to
equilibrium positions at temperatures as low as 100 °C.
The ␣ phase has essentially compositionally equilibrated
at 200 °C, although the peaks continue to become nar-
rower at higher temperatures. Raised background in the
vicinity of 36.2° at a temperature of 100 °C develops to
FIG. 9. XRD patterns of the Al0.7Mg0.3 mechanical alloy after being
weak diffraction peaks consistent with peaks of the ␥ phase heated at 15 K/min to the indicated temperatures in the scanning
at 150 and 200 °C. Peaks consistent with the ␤ phase ap- calorimeter. With increasing temperature, the peaks of the ␣ phase
pear at temperatures above 200 °C. The same set of successively shift towards higher angles as Mg is released from the
experiments was performed on mechanical alloys with structure. A set of weaker peaks indicates, first, the crystallization of
compositions Al0.95Mg0.05 and Al0.9Mg0.1. In these experi- ␥ phase between 150 and 200 °C, and finally the formation of the ␤
phase above 200 °C.
ments, the only observable effect was a slight decrease in
lattice parameter, and thus the Mg concentration, of the
␣ phase. No other phase could be detected by XRD.
The development of the DSC signals as a function
of milling time for an Al0.7Mg0.3 mechanical alloy is
shown in Fig. 10. The initial endothermic peak and the

FIG. 8. Comparison of DSC signals from mechanical alloys with

composition Al0.8Mg0.2. The solid curve represents the fresh mechani-
cal alloy, while the dashed curve represents an alloy that was annealed
for 1 h at 160 °C (near the peak temperature of the first exothermic
transition) before scanning from 100 to 420 °C. FIG. 10. DSC traces of Al0.7Mg0.3 as a function of milling time.

1832 J. Mater. Res., Vol. 18, No. 8, Aug 2003

 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

progresses. Figure 11 shows the heat effect and onset and
peak temperatures of this peak as a function of milling
time. Comparable to the differences between the separate
milling vials that are seen in the diffraction patterns
(Figs. 4 and 5), small differences are observed here as
well. After an initial decrease between 3 and 9 h, onset
and peak temperatures increase slightly as milling
progresses further. The heat effect generally decreases
with time, and although there is a marked offset between
separate vials, the decreasing trend continues for both
vials equally. Contrary to the results from XRD (Fig. 5),
no saturation is observed at the longest milling times;
onset and peak temperatures and heat effect appear to
continue to change after 36 h.
From DSC scans at heating rates varying from 3 to
25 K/min, apparent activation energies for all transitions
in the temperature interval 250–350 °C have been deter-
mined by the method after Kissinger (e.g., Ref. 21) ac-
cording to the following relation:
⌽ EA
ln 2 = − +C . (2)
T max kBNATmax
⌽ is the heating rate in K/s, Tmax is the temperature at the
maximum of the respective DSC peak, EA is the activa-
tion energy, kB is the Boltzmann constant, and NA
Avogadro’s number. The constant C depends on any spe-
cific kinetic model—no such model was applied in this
consideration, and thus C was not further interpreted.
Both EA and C are determined from linear plots of ln(⌽/
T 2max) versus 1/Tmax. Results are shown in Table III.
C. Particle sizes
The results of the particle size determination for the
mechanical alloys with 10–50% are shown in Fig. 12.
The size distributions of the mechanical alloys are gen-
erally similar, no systematic change with composition is
observed.

FIG. 11. Heat effect, onset and peak temperatures of ␤⬘ formation in
the Al0.7Mg0.3 mechanical alloy. Milling was carried out in a double-
mill with two milling stations, labeled A and B for reference (compare
Figs. 4 and 5).
V. DISCUSSION
The results reported here expand the experimental
coverage of mechanical alloys in the Al–Mg system to
low Mg concentrations. At Mg concentrations of 40 at.%
and above, current results are generally consistent with
previously published findings. Calka et al. (1993) re-
ported the extended solid solubility of Mg in the ␣ phase
with 18% Mg for an Al0.7Mg0.3 mechanical alloy, and
45% Mg for an Al0.5Mg0.5 MA.15 The latter value was
derived by deconvoluting strongly overlapping XRD
peaks of ␣ and ␥ phases, and is likely to be biased. Zhang
et al. (1994) reported ␣ phase with 23% Mg in an
Al0.4Mg0.6 MA.16 The highest Mg concentration ob-
served in this work is 20.8% in Al0.4Mg0.6. As Figs. 5
and 6 show, the degree of supersaturation of the ␣ phase
achievable by MA depends critically on milling condi-
tions. Since the degree of supersaturation of the ␣ phase
is not uniquely determined by thermodynamic con-
straints, some compositional variation must be assumed,

TABLE II. Peak temperatures (in °C) of the transitions in Al–Mg mechanical alloys observed by DSC at a heating rate of 15 K/min. The notations
(+) and (−) indicate exothermic and endothermic effects, respectively.

Al0.95Mg0.05 Al0.9Mg0.1 Al0.8Mg0.2 Al0.7Mg0.3 Al0.6Mg0.4 Al0.5Mg0.5 Assignment

96.0 (−) 94.4 (−) 107.7 (−) 109.0 (−) 99.2 (−) 92.7 (−) GP zone dissolution
169.6 (+) 172.4 (+) 171.4 (+) 164.4 (+) ␥ formation
271.9 (+) 265.6 (+) 267.2 (+) 268.4 (+) ␤⬘ formation
375.0 (+) 334.6 (+) 316.6 (+) 312.5 (+) 307.9 (+) ␤⬘ → ␤
348.9 (−) ␤ dissolution
402.3 (+) 399.0 (+) 394.0 (+) ␥⬘ formation

J. Mater. Res., Vol. 18, No. 8, Aug 2003 1833

 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys
and the above values represent therefore (volume) aver-
ages. XRD peak widths and particularly peak asymmetry
(Figs. 4 and 5) illustrate this point; crystallite size/lattice
strain analysis using XRD peak broadening is unable
to distinguish between lattice strain and small-scale
compositional spread. The direct effect milling param-
eters have on the resulting phases and their composi-
tional variation is not immediately obvious. Sample
caking was observed in nearly all experiments that
were carried out with a powder load of 10 g, a BPR of
5, and 2% PCA. No caking was observed during
synthesis of an Al0.6Mg0.4 mechanical alloy with 2 g
powder load, a BPR of 13, and 4% PCA. Caking is most
directly influenced by the amount of process control
agent. One aspect of caking is the drastic reduction of
milling efficiency—no further alloying occurs once the
entire powder load solidly coats the interior surfaces
of the milling vials. In addition, the increase in peak
asymmetry of the alloy in vial A after it had been left
unopened between 15 and 33 h as shown in Fig. 5
strongly suggests that if a mechanical alloy with a well-
established supersaturated ␣ phase is allowed to form
large agglomerates, the ␣ phase starts to disproportionate
into a low-Mg and a high-Mg solid solution as milling
progresses. Eventually, this trend would result in an as-
semblage of ␣ phase with near equilibrium Mg concen-
tration and ␥ phase, partially canceling the increased
FIG. 12. Particle size distributions of Al–Mg mechanical alloys.
1834 J. Mater. Res., Vol. 18, No. 8, Aug 2003
metastability of the mechanical alloy, thought to be re-
sponsible for enhanced performance of the Al–Mg me-
chanical alloys as energetic materials.
The present results confirm previous reports that
␥ phase is an early, intermediate product during mechani-
cal alloying of Al-rich compositions which continues to
dissolve into the ␣ phase as milling progresses.15,16 The
relationship between bulk Mg concentration and Mg con-
centration of the ␣ phase observed in the present work
further suggests that the steady state after prolonged mill-
ing consists of a systematic compositional balance of ␣
and ␥ phases (Fig. 3, Table I). In contrast, the ␤ phase is
never observed. However, ␤ phase and its precursors are
known to form during aging of supersaturated ␣ phase.9
It is therefore suggested that precursors of the ␤ phase
are present in the mechanical alloys although not detect-
able by XRD.
This hypothesis is supported by the results of thermal
analysis. Endothermic peaks at temperatures less than
100 °C on heating Al–Mg alloys with supersaturated ␣
phase have been attributed to the breakdown of Guinier-
Preston (GP) zones.9,22 These zones are two-dimensional
precipitates within the ␣ phase that form on dislocations
during aging at ambient conditions and act as nucleation
points for ␤ phase. The observed endothermic peak at
low temperatures (Fig. 7, Table II) therefore qualitative-
ly indicates their presence. At higher temperatures, the
formation of ␥ phase was observed to be accompanied
by an exothermic peak at about 150 °C in mechanical
alloys.16 This is consistent with the observation of
␥ phase in an Al0.7Mg0.3 MA heated to 200 °C in the pres-
ent work (Fig. 9). The interpretation of this DSC peak as
formation of ␤ phase, as suggested in Ref. 15, cannot be
supported based on the present results. Peak assignment
above 250 °C is more tentative. A number of exothermic
peaks between 250 and 300 °C have been observed, both,
in quenched alloys9,22 and mechanical alloys.15,16 From a
structural analysis on quenched alloys,9,22 two exother-
mic events in the temperature ranges 180–290 °C and
280–330 °C have been attributed to the formation of ␤⬘
and ␤ phase, respectively. This is generally consistent
with the observation of ␤ phase in Al0.7Mg0.3 MA heated
to 300 °C in the present work (Fig. 9). The exothermic
TABLE III. Activation energies for exothermic transitions in the 250–
350 °C temperature interval.
xMg (mol%) Tmax (°C)a EA (kJ/mol) Constant
10 334.6 154.7 −16.75
20 271.9 254.9 −42.77
20 316.6 250.8 −37.04
30 265.6 238.1 −39.97
30 312.5 214.6 −29.41
40 267.2 229.6 −37.33
50 268.4 211.4 −32.94
a
Heating rate of 15 K/min; for comparison with Table II.
 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys
peaks above 300 °C observed in mechanical alloys with
low Mg concentration (Al0.95Mg0.05 and Al0.9Mg0.1)
have not been previously reported; it is however sug-
gested that they are caused by ␤ phase precipitation as
well, even though no ␤ phase could be detected in alloys
annealed at and recovered from intermediate tempera-
tures. It is likely that the newly formed ␤ phase was
either not abundant enough to be detected by XRD, or
has redissolved into the ␣ phase during the relatively
slow cooling in the DSC apparatus (approximately
20 K/min). The peaks observed around 400 °C in alloys
with 30%, 40%, and 50% Mg (Fig. 7, Table II) could not
be unambiguously assigned to structural transitions.
From the equilibrium phase relations,11 a relation to the
formation of ␥⬘ phase could be surmised.
The exothermically biased background seen to varying
degrees in all samples (Figs. 7 and 8) has not been de-
scribed for quenched alloys, nor has it been systemati-
cally described for mechanical alloys. One possible
explanation is the gradual relaxation of mechanical
stress accumulated during the MA process, as well as
loss of interfacial (grain boundary) area due to coarsen-
ing. Accelerated grain growth has been observed to occur
in heavily strained Al-3%Mg alloy at temperatures above
about 100 °C,23 supporting this interpretation.
Activation energies determined in this work for exo-
thermic reactions associated with ␤⬘/␤ formation in
Al–Mg mechanical alloys (Table III) are generally
larger than the activation energy of 78 kJ/mol reported
for ␤ formation in rapidly quenched alloys.5 The appar-
ent activation energy decreases with increasing Mg
concentration. The value obtained for the exothermic
transition in the Al0.9Mg0.1 mechanical alloy is likely
influenced by the overlap of the exothermic event
(334.6 °C) with the following endothermic event
(348.9 °C, Fig. 7, Table III) and therefore should be
regarded with caution.
VI. SUMMARY AND CONCLUSIONS
A number of mechanical alloys in the Al–Mg binary
system have been prepared in a shaker mill. The me-
chanical alloys consist of supersaturated ␣ phase and
␥ phase. Additional amorphous material is likely to be
present, although it has not been detected by x-ray dif-
fraction. The maximum degree of supersaturation of
the ␣ phase achievable is sensitive to milling conditions.
Higher milling energy (increased BPR) combined with
higher levels of PCA to prevent caking leads to stronger
supersaturation. The highest Mg concentration of the
␣ phase observed in the present work is 20.8 at.% in
an Al0.6Mg0.4 mechanical alloy. Average particle sizes
of the mechanical alloys vary between approximately
25 and 35 ␮m. No significant dependence of the
particle size distribution on alloy composition has been
J. Mater. Res., Vol. 18, No. 8, Aug 2003
observed, indicative of similar mechanical properties of
mechanical alloys in the covered compositional range
(Al0.95Mg0.05–Al0.5Mg0.5).
On heating, the mechanical alloys transform to equi-
librium phase assemblages via a series of distinguishable
steps. Individual transitions that were identified by XRD
of intermediate products are the crystallization of ␥ phase
near 170 °C, and the precipitation of ␤ phase in the range
of 250–350 °C. The activation energy for the ␤⬘/␤ phase
formation has been determined as a function of alloy
composition. The values of approximately 250 kJ/mol
are roughly three times as large as activation energies
reported for ␤ formation from rapidly quenched Al–Mg
alloys, suggesting different transformation kinetics in
mechanical and quenched alloys. Further peaks have
been observed by DSC at temperatures near 400 °C in
mechanical alloys with more than 30 at.% Mg but have
not been assigned to structural transitions. In addition to
structural transitions at defined temperatures, the relaxa-
tion of the mechanical alloys is associated with a spread-
out exothermic effect starting at about 100 °C and
continuing over the entire temperature range investigated
(Tmax ⳱ 420 °C). The magnitude of this effect is on the
order of 2.5 kJ/mol (100 J/g). The heat effects of
the identified DSC peaks are being considered in ignition
models describing the behavior of Al–Mg mechanical
alloys in energetic formulations.8 Possible influence of
the observed phase transformations on oxidation kinetics
will also be addressed. The overall exothermic character
of the relaxation of the mechanical alloys is expected to
be at least partially responsible for the previously re-
ported accelerated ignition and combustion of aerosols of
Al-Mg mechanical alloys.
ACKNOWLEDGMENTS
This work was primarily supported by Dr. Judah Gold-
wasser of Office of Naval Research, Grant No. N00014-
00-1-0446. Additional support was provided by New
Jersey Commission on Science and Technology, Award
No. 01-2042-007-24.
REFERENCES
1. M.L. Chan, R. Reed, and D.A. Ciaramitaro, Prog. Astronautics
Aeronautics 185, 185 (2000).
2. A. Gany and D.W. Netzer, Int. J. Turbo Jet Engines 2, 157 (1985).
3. B. Palaszewski and J.S. Zakany, Metallized Gelled Propellants:
Oxygen/RP-1/Aluminum Rocket Combustion Experiments, 31st
Joint Propulsion Conference Proceedings, San Diego, CA, AIAA-
95-2435 (1995).
4. E.L.Dreizin, Prog. Energy Combust. Sci. 26, 57 (2000).
5. T.A. Roberts, R.K. Burton, and H. Krier, Combust. Flame 92, 125
(1993).
6. Y.L. Shoshin, R.S. Mudryy, and E.L. Dreizin, Combust. Flame
128, 259 (2002).
7. M. Schoenitz, E.L. Dreizin, and E. Shtessel, J. Prop. Power (in
press, April 2003).
1835
 M. Schoenitz et al.: Structure and properties of Al–Mg mechanical alloys

8. M.A. Trunov and E.L. Dreizin, Ignition of Metastable Metal
Based High Energy Density Fuels. Submitted for presentation at
the Ninth International Workshop on Combustion and Propulsion,
La Spezia, Italy 14–18 September 2003.
9. M.J. Starink and A-M. Zahra, Acta Mater. 46, 3381 (1998).
10. H. Okamoto, J. Phase Equilibria 19, 598 (1998).
11. J.L. Murray, Bull. Alloy Phase Diagrams 3, 60 (1982).
12. J.S. Benjamin, Scientific American 234, 40 (1976).
13. J.S. Benjamin, Metal Powder Report 45, 122 (1990).
14. C. Suryanarayana, Prog. Mater. Sci. 46, 1 (2001).
15. A. Calka, W. Kaczmarek, and J.S. Williams, J. Mater. Sci. 28, 15
(1993).
16. D.L. Zhang, T.B. Massalski, and M.R. Paruchuri, Metall. Mater.
Trans. 25A, 73 (1994).
17. G.K. Williamson and W.H. Hall, Acta Metall. 1, 22 (1953).
18. C.E. Ellwood, J. Inst. Metals 80, 605 (1952).
19. H.L. Luo, C.C. Chao, and P. Duwez, Trans. Metall. Soc. AIME
230, 1488 (1964).
20. N.L. Johnson, S. Kotz, and N. Balakrishnan, Continuous Univari-
ate Distributions, 2nd ed. (Wiley, New York, 1994).
21. E.J. Mittermeijer, J. Mater. Sci. 27, 3977 (1992).
22. M.J. Starink and A-M. Zahra, Philos. Mag. A 76, 701 (1997).
23. Z. Horita, D.J. Smith, M. Furukawa, M. Nemoto, R. Valiev, and
T.G. Langdon, Mater. Characterization 37, 285 (1996).

1836 J. Mater. Res., Vol. 18, No. 8, Aug 2003