Oxidation and melting of aluminum nanopowders

Article abstract of DOI:10.1021/jp0614188

Recently, nanometer-sized aluminum powders became available commercially, and their use as potential additives to propellants, explosives, and pyrotechnics has attracted significant interest. It has been suggested that very low melting temperatures are ex

Full text of DOI:10.1021/jp0614188

13094
J. Phys. Chem. B 2006, 110, 13094-13099
Oxidation and Melting of Aluminum Nanopowders
†
†
†
‡
Mikhaylo A. Trunov, Swati M. Umbrajkar, Mirko Schoenitz, Joseph T. Mang, and
Edward L. Dreizin*,†
New Jersey Institute of Technology, Newark, New Jersey 07102, and Los Alamos National Laboratory,
Los Alamos, New Mexico 87545
ReceiVed: March 7, 2006; In Final Form: May 10, 2006
Recently, nanometer-sized aluminum powders became available commercially, and their use as potential
additives to propellants, explosives, and pyrotechnics has attracted significant interest. It has been suggested
that very low melting temperatures are expected for nanosized aluminum powders and that such low melting
temperatures could accelerate oxidation and trigger ignition much earlier than for regular, micron-sized
aluminum powders. The objective of this work was to investigate experimentally the melting and oxidation
behavior of nanosized aluminum powders. Powder samples with three different nominal sizes of 44, 80, and
121 nm were provided by Nanotechnologies Inc. The particle size distributions were measured using small-
angle X-ray scattering. Melting was studied by differential scanning calorimetry where the powders were
heated from room temperature to 750 °C in an argon environment. Thermogravimetric analysis was used to
measure the mass increase indicative of oxidation while the powders were heated in an oxygen-argon gas
mixture. The measured melting curves were compared to those computed using the experimental particle size
distributions and thermodynamic models describing the melting temperature and enthalpy as functions of the
particle size. The melting behavior predicted by different models correlated with the experimental observations
only qualitatively. Characteristic stepwise oxidation was observed for all studied nanopowders. The observed
oxidation behavior was well interpreted considering the recently established kinetics of oxidation of micron-
sized aluminum powders. No correlation was found between the melting and oxidation of aluminum
nanopowders.
Introduction
were provided by Nanotechnologies Inc., Austin, TX. For each
“
as received” powder, the particle size distributions (PSD) were
Nanosized aluminum powders have been produced and
17-19
1-3
determined using small-angle X-ray scattering (SAXS).
evaluated as reactive additives to propellants and explosives
as well as components of nanocomposite energetic materials.
4
,5
During the SAXS fitting process, polydispersity was taken into
account using a log-normal distribution. Two log-normal
distributed populations had to be used to fit the experimental
SAXS data for each analyzed nanopowder. While the shape of
the size distributions was assumed, the mean modal diameters
and widths of both log-normal distributions were allowed to
vary freely during the fitting process. The experimental SAXS
data with a best fit based on the double log-normal distribution
are presented in Figure 1.
Specific parameters determined from the SAXS measurements
describing the PSD are presented in Table 1. The normalized
frequency function, P(D), of aluminum nanoparticle distributions
that was used in the study to describe melting and oxidation of
aluminum nano-powders is expressed as
An accelerated reaction rate is generally due to an increase in
the specific surface of the powder. It has also been suggested
that the melting point depression reported to occur for metal
6
-11
nanoparticles
could result in an earlier aluminum ignition
(
triggered by accelerated oxidation) and therefore, in shorter
1
2,13
ignition delays.
However, experimental data on melting of
aluminum nanoparticles are scarce, and it was recently shown
that no correlation exists between melting and ignition or
oxidation for micron-sized aluminum powders.1
4-16
The main
objectives of this work are to characterize experimentally
melting and oxidation of the nanosized aluminum powders. The
effect of the particle sizes on melting and oxidation kinetics
was specifically addressed. Instead of a common practice of
representing each powder sample with a single average particle
size, the experimental size distributions for different nanopo-
wders were obtained and used directly for quantitative inter-
pretations of the measurements.
2
2
(ln(D) - ln(〈D〉 ))
1
k
P(D) )
Wk
exp -
(1)
∑
(
2
)
k)1
2 ln (σ )
D ln(σk)
x
2π
k
Experimental Section
where D is the diameter of the aluminum core, which is a
parameter measured by SAXS directly, the index k represents
two log-normal distributions, Wk is a weight factor for each of
those log-normal distributions, σk are the geometric standard
deviations, and 〈D〉k are the geometric mean diameters. The
aluminum powder manufacturers report the oxide layer thick-
Particle Size Distributions. Three aluminum nanopowders
with the respective BET particle sizes of 44, 80, and 121 nm
*
To whom correspondence should be addressed. E-mail: dreizin@njit.edu.
Tel: (973) 596-5751. Fax: (973) 642-4282. Address: Department of
Mechanical Engineering, New Jersey Institute of Technology, Newark, NJ
0
7102-1972.
2
0
†
ness, hox, in the range of 1.8-3.5 nm. It was assumed that
“as received” powders had an oxide thickness of 1.8 nm, so
New Jersey Institute of Technology.
Los Alamos National Laboratory.
‡
1
0.1021/jp0614188 CCC: $33.50 © 2006 American Chemical Society
Published on Web 06/10/2006

Oxidation and Melting of Aluminum Nanopowders
J. Phys. Chem. B, Vol. 110, No. 26, 2006 13095
Figure 1. Experimental SAXS intensity for three different aluminum nanopowders as a function of the scattering vector. Symbols with error bars
represent experimental data; solid lines are the modeled intensities based on linear superposition of two log-normal particle size distributions.
Figure 2. Volume based PSD for different aluminum nanopowders.
The vertical axis shows percent of a differential of cumulative volume
distribution for aluminum core divided by the logarithm of the particle
size. Therefore, the area under each normalized PSD curves equals
1
00%.
Figure 3. Transmission electron microscopy image of an Al nano-
powder with BET particle size of 44 µm.
TABLE 1: Particle Size Distributions of the Al
Nanopowders Obtained from SAXS
a limited number of particles are available from such images.
This limited number of particles is not statistically representative
and cannot be used for assessment of the powder size distribu-
tion.
Thermal Analysis. Melting and oxidation of the powders
were studied using a Netzsch Simultaneous Thermal Analyzer
STA409 PC. The instrument was calibrated for temperature with
the melting points of a set of metal standards resulting in a
temperature accuracy of (1 °C.
log-normal distribution params and weight factors
geometric geometric
BET particle size, mean diameter, standard deviation,
weight
factor
µm (powder ID)
〈D〉 , nm
σ, nm
4
4
18.8
1.80
1.71
1.75
1.47
1.72
1.48
0.999 42
0.000 58
0.999 936
0.000 064
0.997 937
0.002 603
1
3
1
29.2
33.4
28
38.4
96.6
8
0
1
21
To describe melting, differential scanning calorimetry (DSC)
was performed in argon, heating powders up to 750 °C with a
rate of 5 °C/min. The 35-45 mg samples were heated in
alumina and platinum sample holders. To establish reproducible
initial conditions for the DSC measurements, each sample was
heated from room temperature to 300 °C at 20 °C/min and kept
at 300 °C for about 30 min prior to the main heating program.
During heating and melting, sintering can occur affecting the
powder sample’s geometry, location in the DSC sample holder,
and the heat transfer in the DSC measuring head. To minimize
such artifacts, three experiments were performed. In the first
experiment, the powder was loosely packed in the sample holder.
In the second test, the powder was pressed into a pellet. Finally,
in the third experiment, the powder was mixed with alumina
powder (Inframat Advanced Materials LLC, Farmington, CT,
particle size ≈ 150 nm) in the 1:1 mass ratio, and the mixture
was loosely packed in the sample holder. The results of all three
types of experiments were generally consistent, while the most
reproducible measurements were obtained with the mixed Al/
Al2O3 powders. These measurements are presented below.
Oxidation was characterized using thermogravimetric analysis
(TGA). Samples of 25-40 mg contained in alumina crucibles
were heated in an oxygen/argon environment (oxygen flow rate
10 mL/min, argon flow rate 50 mL/min) from room temperature
to 1350 °C. Heating to higher temperatures, performed in a
that the particle diameters measured by SAXS and representing
the metal core were corrected accordingly. Powders stored in
the laboratory for the duration of this project (about 6 months)
continued to slowly oxidize as was determined from the
thermogravimetric measurements performed over this period of
time with the same powder batches. This growth of oxide
thickness and respective changes in the overall particle diameters
were accounted for while interpreting the oxidation measure-
ments as described below.
The volume-based distributions for the aluminum core
diameters, as determined by SAXS and described by eq 1 and
in Table 1, are presented in Figure 2. PSD for each powder is
skewed to the right side because of the presence of relatively
large particles represented by the second log-normal mode (see
Table 1). These large particles produce a noticeable contribution
when such volumetric processes as melting and oxidation are
considered.
The presence of relatively large particles within the nanopo-
wders used in this research is clearly seen from the transmission
electron microscopy image shown in Figure 3. The image shows
a particle with a diameter of about 150 nm in the nanopowder
with the BET size of 44 nm. Transmission electron microscopy
images serve here for demonstration purposes only because only

13096 J. Phys. Chem. B, Vol. 110, No. 26, 2006
Trunov et al.
limited number of experiments, did not result in any further
weight increase. Two heating rates of 1 and 5 °C/min were used
in the study.
by two different processes: a mass change m˘ mox due to the
tr
direct aluminum oxidation and a mass change m˘ due to
polymorphic phase transformations. Because of the accepted
sequence of the polymorphic phase transitions: amorphous f
14
Oxidation Modeling
γ f R, the specific masses of the amorphous, γ-, and R-oxides,
referred to by the subscripts am, γ, and R, respectively, are
determined for each instant of time (or for each temperature)
from the following equations:
Recently, oxidation of micron-sized aluminum powders was
described considering formation of different polymorphs of
1
4-16
Al2O3.
At different oxidation stages, the oxidation rates
are limited by the diffusion resistance of different crystal-
lographic modifications of alumina. The oxidation model
described in detail elsewhere16 was modified in this work to
account for the effect of particle size distribution on the diffusion
and oxidation rates. After this modification, the model was used
to predict the TGA curves expected for each powder sample
while the PSD for each powder was obtained from SAXS
measurements. All particles were assumed to be spherical.
Major Processes during Aluminum Oxidation. Initially,
aluminum particles are covered by a thin layer of “natural”
amorphous oxide, and it should be noted that recent research
ox
am
tr
F V˙
)
∫
[ m˘ - m˘
]P(D) dD
(2.1)
(2.2)
(2.3)
am am
amfγ
ox
tr
tr
γfR
F V˙ )
∫
[ m˘ γ + m˘
- m˘ ]P(D) dD
γ
γ
amfγ
ox
tr
γfR
F V˙ )
∫
[ m˘ R + m˘ ]P(D) dD
R
R
2
µAl
ox
ox
ox
F V˙ ) -
∫
[ m˘ + m˘ γ + m˘ R ]P(D) dD
(2.4)
Al Al
am
µ
Al O
2
3
where V˙ is the rate of volume change, F is density, and all the
variables in the square brackets are functions of particle
diameter. The temperature of aluminum particles, T, affecting
the rates of mass change, m˘ i, increases from room temperature
at a constant rate used in the oxidation TGA experiments. The
modeled TGA weight change was calculated by integrating eqs
21
using prompt gamma neutron activation analysis and high-
resolution transmission electron microscopy22 showed that the
amorphous oxide coatings on aluminum nanopowders include
impurities of hydroxide, trapped water molecules, and boron.
In addition, single crystalline layers were observed on the surface
and in localized areas of the otherwise amorphous oxide. Such
impurities as hydroxide are expected to become more significant
for aged particles. However, addressing the effect of such
impurities on the aluminum oxidation was beyond the scope of
this paper. These effects, together with effects of specific surface
morphology (e.g., nonspherical and agglomerated particles) and
bulk impurities (e.g., iron, copper, and potassium detected by
2
.1-2.4 over time numerically.
The rates of mass change due to direct oxidation were
described by the model of diffusion for the rate-limiting species.
To simplify the boundary conditions necessary to compute the
diffusion mass flows, the model allows the direct oxidative
growth of only one alumina polymorph at any given time.
Specifically, at any given time only the alumina polymorph with
the highest diffusion resistance, which is tracked during the
calculations, is allowed to grow because of the direct oxidation.
The phase transformations can occur simultaneously and may
contribute to the growth of any coexisting alumina polymorph.
Thus, if several alumina polymorphs coexist, each particle is
surrounded by concentric shells of the respective oxide poly-
morphs. The spherical symmetry allows one to recalculate
volumes of alumina and aluminum core for each specific particle
size using equations similar to 2.1-2.4, but written for an
individual particle, without the integration over PSD.
22
X-ray photoelectron spectroscopy ) will certainly reduce the
accuracy of the aluminum oxidation model16 that is also used
in the present paper.
Neglecting the impurities present in the initial amorphous
oxide layer it is simply assumed that at low temperatures its
thickness increases. At elevated temperatures, the amorphous
alumina transforms into a denser γ-Al2O3 polymorph. This phase
transformation can reduce the thickness and the diffusion
resistance of the oxide layer. For thin oxide layers observed at
high heating rates, this phase change can also result in local
discontinuities in the oxide coverage.1
4,16
Thus, the rate of
Diffusion-Limited Oxidation. For each polymorph, the rate
of direct, diffusion-limited oxidation for a spherical oxide layer
oxidation accelerates rapidly until a continuous, polycrystalline
layer of the γ-Al2O3 is produced. Note that the decomposition
of any hydroxide phase would lead to a similar volume
reduction, accelerating oxidation as well. As the temperature
continues increasing, the layer of the γ-Al2O3 grows until
crystallites of an even denser R-oxide start forming. Formation
of a higher density oxide polymorph once again results in a
reduction in the overall oxide thickness and the diffusion
resistance of the oxide layer. An accelerated oxidation continues
until the oxide layer transforms into a continuous, polycrystalline
film. According to this model,14,16 five processes need to be
analyzed to describe the oxidation quantitatively. In the order
of their occurrence at the increasing temperatures, these
processes are (1) growth of amorphous oxide, (2) the amorphous
to γ-alumina phase change, (3) growth of γ-alumina, (4) the γ
to R-alumina phase change, and (5) growth of R-alumina.
General Model Formulation. The mass of each alumina
polymorph and the mass of metallic aluminum are computed
as a function of time and temperature. To account for the effect
of specific PSD, the computations are made for each particle
size and then integrated over the PSD. According to the model,
changes in the mass of each oxide polymorph can be caused
of outer diameter, D , is described using an Arrhenius type
expression
i
-
1
1
2
Ei
1
1
Di
ox
m˘ i ) C exp -
-
(3)
i
(
)(
)
RT D
i-1
where the subscript i indicates the specific oxide polymorph,
that is, amorphous, γ, or R alumina, subscript i-1 indicates
the underlying substrate or “parent” material, which could be
aluminum, amorphous, or γ alumina, respectively, C and Ei
i
are the parameters describing the diffusion of rate-limiting
species in different alumina polymorphs, and R is the ideal gas
constant. The values of C , and E were the same as determined
i
i
recently for the oxidation of micron-sized aluminum particles.16
Diffusion in Thin Oxide Layers. Additional assumptions
were necessary to describe the oxidation kinetics when the oxide
thickness was comparable to the size of individual γ- and
R-oxide crystallites. When such crystallites are initially formed,
the discontinuities in the oxide coverage are produced.1 The
initial diffusion resistance of the newly formed γ- and R-oxide
crystallites was neglected. A critical thickness for each of the
4,16

Oxidation and Melting of Aluminum Nanopowders
J. Phys. Chem. B, Vol. 110, No. 26, 2006 13097
new growing layers of the γ- and R-oxide polymorphs was
identified at which these layers stop behaving as individual
crystallites and start producing diffusion resistance typical of
respective continuous polycrystalline layers. These transition
oxide thicknesses, introduced previously to describe oxidation
of the micron-sized aluminum powders,16 were also used here.
According to the model proposed in ref 16, the diffusion
resistance of a new growing oxide layer increases linearly with
the increase in the oxide thickness until a continuous, poly-
crystalline oxide coverage is formed. Further details are available
in ref 16.
Figure 4. Calculated initial thickness of the amorphous oxide layer
as a function of the particle size for an aluminum nanopowder (BET
diameter 80 nm) stored in the laboratory for a period of time.
Polymorphic Phase Changes in Al2O3. The rate of mass
tr
increase, m˘ (i-1)fi, of the oxide phase i transformed from the
oxide phase i-1 is described as
corresponding to the largest fully oxidized particles. Extended
aging will result in a slow increase of the oxide thickness for
the decreasing branch of the curve (particles that are not fully
oxidized) and a shift of the peak toward greater particle sizes.
It is clear that the size-dependent “aging” effect is substantial
only for the particles that are smaller than ∼20 nm while it can
be neglected for coarser particles.
tr
2
i-1
m˘ (
) πD ‚Fi-1V(i-1)fi
(4)
i-1)fi
where V is the velocity of the interface motion. The velocity is
described using a phenomenological expression proposed in ref
6, in which the thermal kinetics are supplemented by the effect
of the thickness of the parent layer of transition alumina, hi-1
1
1
)
/2(Di-1 - Di-2)
Melting Models
Melting Temperature and Entropy Dependence on Par-
ticle Size. The effect of particle size on its melting temperature
and latent heat of melting for nanosized metal particles has been
extensively discussed in the literature. A model commonly
quoted in recent papers dealing with aluminum nanopowders
ν(i-1)fi
)
K(i-1)fihi-1
E(i-1)fi
F(i-1)fiT 1 - exp -
exp -
) (5)
{
(
)}
(
RT
RT
13
used in energetic materials, for example, was developed about
The activation energies E for each specific phase transition, as
well as parameters F and K, were determined earlier by
processing the TGA data for the micron-sized particle.
6
five decades ago. That model describes the melting point, Tm,
1
6
as a function of the particle diameter, Dp, and the oxide shell
thickness hox, as
Initial Conditions. The initial volumes (and thus thicknesses)
of both γ and R phases were equal to zero. The initial volumes
of the amorphous oxide and aluminum core were considered
as a function of PSD using the SAXS measurements and a size-
dependent correction required to account for “aging” or oxida-
tion of the powders stored in the laboratory. The overall mass
of the initial oxide for the “aged” particles was determined from
the total mass increase measured from the respective TGA
curves acquired in the oxygen/argon environment. Thus, a
specific correction was introduced for interpretation of each
specific TGA experiment. The oxidation of powders during their
storage was analyzed assuming a uniform, 1.8 nm, initial oxide
thickness for “as received” particles of all sizes and using eqs
4
σsl
T ) T 1 -
(6)
m
b
[
]
H F (D - 2h )
b
Al
p
ox
where Tb is the melting temperature of bulk aluminum, Hb is
the enthalpy of fusion of bulk aluminum, and σsl is the interfacial
surface tension between the solid and the liquid. The difference
in the molar volumes between solid and liquid aluminum was
neglected.
More recently, a theoretical model of melting for nanocrys-
talline metal powder was developed by Jiang et al.
been proposed that the melting point depends on the diameter
of aluminum nanocrystals, assumed in our study to be equal to
the diameter of particle aluminum core, as
8-10
It has
2.1, 2.4, and 3 while allowing the particle’s temperature to
remain at a constant room-temperature level. The growing
thickness of the amorphous oxide layer was computed as a
function of time until the total mass of the oxide layer matched
the oxide mass inferred from the TGA measurements. The oxide
growth rate was assumed to be limited by the diffusion of the
reacting species at the storage temperature. An example of the
resulting distribution of the oxide thickness over the particle
diameter is shown in Figure 4. A similar distribution of the oxide
thickness as a function of the particle diameter is expected for
any aluminum nanopowder.
2
Hb
1
T ) T exp -
(7)
m
b
3
RT D
b
(
- 1
)
(
)
6
l
where l is the length of the Al-Al atomic bond. Furthermore,
it was suggested that for nanoparticles, the latent heat of melting,
Hm, depends on the particle diameter according to
The calculations in which the diffusion rates depend on
particle size (cf. eq 3) predict that the oxide thickness increases
faster for the finer particles. This explains the decreased oxide
thickness as a function of the particle diameter for particles
greater than 10 nm. On the other hand, particles less than 10
nm (in the specific case shown in Figure 4) are already fully
oxidized. For the fully oxidized particles, the thickness of the
oxide layer is naturally limited by the particle diameter, so that
a sharp peak is observed in Figure 4 for the particle diameter
2
Hb
1
1
H ) H exp -
1 -
(8)
m
b
3RT D
D
6l
b
(
- 1
)
[
- 1
]
6l
Finally, processing the experimental results reported by Eckert
1
1
et al. suggests phenomenological dependencies for both the
melting point (in K) and latent heat of melting (in kJ/mol) as
functions of the aluminum core diameter (in nm):

13098 J. Phys. Chem. B, Vol. 110, No. 26, 2006
Trunov et al.
1
920
D
Tm ) 977.4 -
(9)
1
77.49
D
H ) 14.705 -
(10)
m
The experimental data presented in ref 11 limit applicability of
the trends given by eqs 9 and 10 for the particles with diameters
in the range of 12.07 nm < D < 43.24 nm. For larger particles,
Tm ) Tb and Hm ) Hb. For particles with the metal core smaller
than 12.07 nm, the latent heat of melting is reported to be
1
1
negligible.
Modeling of Experimental DSC Signal. Three different
melting models introduced above were used together with the
experimental PSD to quantitatively predict the DSC signals
expected while different powder samples melt. Each calculation
was compared to a specific DSC run. The correction for the
initial oxide thickness as a result of aging for each specific DSC
run was made in the respective calculations using the procedure
described above and the results of the concurrent TGA measure-
ments. Equations 6, 7, and 9 represent functional dependence
of the melting temperature on the particle diameter. Each of
these equations was converted into a respective dependence of
the aluminum core diameter as a function of the melting
temperature, D ) D(Tm). The converted functional dependencies
and their temperature derivatives dD/dT were used to predict
the DSC signal directly. The DSC experiments were performed
at a constant heating rate, â, so that the DSC signal, Q˙ , detected
at each specific temperature, T, was calculated as
Figure 5. DSC and TGA curves measured for three aluminum
nanopowders in argon and oxygen, respectively. Both sets of curves
were acquired at a heating rate of 5 °C/min.
π
6
3
dD
dT
Q˙ ) M P(D) D H (D)
â
(11)
s
m
where Ms is the parameter accounting for frequency function
normalization, proportional to the mass of the aluminum metal
in the analyzed nanopowder sample.
Results and Discussion
Experimental DSC curves showing the melting endotherms
for different powders are shown in Figure 5 together with the
TGA curves showing the weight increase for the same powders
as a result of oxidation.
Figure 6. Comparison of experimental TGA curves with results of
the oxidation modeling.
For all powders, the oxidation is observed to begin at the
temperatures substantially lower than the onset for the melting
endotherm (occurring at around 570 °C for all powders). Thus,
no correlation between melting and oxidation is observed,
There are, however, minor but systematic discrepancies
between the predicted and calculated curves, as shown in Figure
6. The calculated TGA curves are extremely sensitive to the
specific shape of the particle size distribution and particle surface
morphology. It is suggested that the disagreement between the
presented oxidation theory and experiments is within the ranges
expected on the basis of the accuracy with which the particle
size distributions are determined.
1
4-16
similarly to the results for the micron-sized Al powders.
The substantial difference in the slopes of the DSC curves before
and after melting is most likely explained by the change in the
powder morphology as a result of its melting. The change in
morphology affects the heat transfer within the DSC sample
holder and, therefore, the baseline of the measurement.
Comparisons of the experimental TGA curves with the curves
predicted by the recently developed aluminum oxidation model
are shown in Figure 6. The absolute measured mass increase
becomes smaller for the same particles used in different
experiments after increasingly longer storage periods, indicating
that oxidation did occur during the storage. At the same time,
the final mass increases for the calculated and experimental TGA
curves always coincide because of the discussed above correc-
tion for the initial oxide layers for the powders stored in the
laboratory during different periods of time. The calculations,
in general, reproduce well the shapes of the experimental TGA
curves. The shift between the TGA curves as a function of the
heating rate is also well described.
Comparisons of the experimental DSC melting curves with
the curves predicted to describe respective measurements by
different melting models of metal nanopowders are presented
in Figure 7. No close agreement between experiment and any
of the tested melting models was observed; however, the overall
shape of the melting endotherm was predicted by all the models.
In particular, it is interesting that the experimental endotherms
have at least two peaks, and this overall shape is generally
predicted considering the bimodal size distributions obtained
from SAXS (see Table 1). Note that according to eqs 9 and
1
1
10, only a fraction of the powder is predicted to melt before
Tb ) 660 °C is reached. The rest of the powder, as noted in
Figure 7, is expected to melt at a constant temperature of Tb,

Oxidation and Melting of Aluminum Nanopowders
J. Phys. Chem. B, Vol. 110, No. 26, 2006 13099
developed in the study was successful in describing the
experimental TGA curves. The same model was used to evaluate
the particle size dependent thickness of the oxide layer produced
on the aluminum nanopowders during their storage at room
temperature. Observed minor inconsistencies between predic-
tions of the proposed oxidation model and TGA measurements
can be related to the extreme sensitivity of the oxidation rates
to the specific shape of the particle size distribution that was
described in this work by a superposition of two log-normal
functions.
Acknowledgment. This work was supported in parts by
Office of Naval Research, Grant N00014-00-1-0446, Defense
Threat Reduction Agency, Award DAAE30-1-9-0080, and
TACOM-ARDEC, Picatinny, Award DAAE30-03-D-1015.
J.T.M.'s work was funded by the joint DoD/D.O.E. Munitions
Technology Development Program at Los Alamos National
Laboratory. Los Alamos National Laboratory is operated by the
University of California for the D.O.E. under Contract W-7405-
ENG-36.
References and Notes
(
1) Mench, M. M.; Kuo, K. K.; Yeh, C. L.; Lu, Y. C. Combust. Sci.
Figure 7. Experimental and computed DSC curves showing melting
of aluminum nanopowders.
Technol. 1998, 135, 269.
(2) Shafirovich, E.; Bocanegra, P. E.; Chauveau, C.; G o¨ kalp, I. Euro.
Space Agency, (Spec. Publ.), ESA SP 2004, 557, 71.
(
3) Brousseau, P.; Anderson, C. J. Propellants, Explos., Pyrotech. 2002,
7, 300.
4) Pivkina, A.; Ulyanova, P.; Frolov, Y.; Zavyalov, S.; Schoonman,
J. Propellants, Explos., Pyrotech. 2004, 29, 39.
so that respective calculations for the melting endotherms could
not be performed and are not shown in Figure 7.
2
(
It was found that the shapes of the predicted curves are very
sensitive to the specific type of the particle size distribution,
for example, bimodal versus single mode log-normal distribu-
tion. The melting curves can further depend on the specific
surface morphologies of the used nanoparticles that could affect
the sizes of melting nanodomains.
(5) Son, S. F. AdV. Mater. Processes 2005, 163, 11.
(6) Reiss, H.; Wilson, I. B. J. Colloid Sci. 1948, 3, 551.
(7) Revesz, A. J. Mater. Sci. 2005, 40, 1643.
(8) Jiang, Q.; Yang, C. C.; Li, J. C. Material Lett. 2002, 56, 1019.
9) Liang, L. H.; Li, J. C.; Jiang, Q. Physica B (Amsterdam) 2003,
34, 49.
(
3
(
10) Zhao, M.; Jiang, Q. Solid State Commun. 2004, 130, 37.
11) Eckert, J.; Holzer, J. C.; Ahn, C. C.; Fu, Z.; Johnson, W. L.
(
Conclusions
Nanostruct. Mater. 1993, 2, 407.
12) Rai, A.; Lee, D.; Park, K.; Zachariah, M. R. J. Phys. Chem. B 2004,
108, 14793.
(
Melting and oxidation of aluminum nanopowders are studied
experimentally and the results are compared with the predictions
of several models published to date. Experiments showed that
the oxidation and melting onsets of the heated aluminum
nanopowders do not correlate consistently with the conclusion
made earlier for the micron-sized aluminum powders. The
melting of nanopowders starts at a lower temperature than the
melting point of bulk aluminum; however, existing models
describe the effect of the particle size on the melting point
depression only qualitatively. Detailed analysis of the particle
size distributions and surface morphology is needed for further
quantitative verification of any related models. A recent
oxidation mechanism taking into account polymorphic phase
changes in the Al2O3 layer describes oxidation of the nanopo-
wders satisfactorily. The particle-size dependent oxidation model
(
(
13) Granier, J. J.; Pantoya, M. L. Combust. Flame 2004, 138, 373.
14) Trunov, M. A.; Schoenitz, M.; Dreizin, E. L. Combust. Flame 2005,
1
40, 310.
15) Trunov, M. A.; Schoenitz, M.; Dreizin, E. L. Propellants, Explos.,
Pyrotech. 2005, 30, 36.
16) Trunov, M. A.; Schoenitz, M.; Dreizin, E. L. Combust. Theory
Modell. 2006, in print.
17) Mang, J. T.; Skidmore, C. B.; Hjelm, R. P.; Howe, P. M. J. Mater.
Res. 2002, 15, 1199.
18) Peterson, P. D.; Mang, J. T.; Asay, B. W. J. Applied Phys. 2005,
7, 093507.
19) Mang, J. T.; Hjelm, R. P.; Peterson, P. D.; Jorgensen, B. S.; Son,
(
(
(
(
9
(
S. F. Proc. Int. Conf. Powder Metall. Part. Mater. 2004.
(20) Wilson, D. E.; Kim, K. Presented at 39th AIAA/ASME/SAE/ASEE
Joint Propulsion Conference, Reno, NV, 2003; AIAA-2003-4536.
(21) Ramaswamy, A. L.; Kaste, P.; Trevino, S. F. Journal of Energetic
Materials 2005, 22, 1.
(22) Ramaswamy, A. L.; Kaste, P. J. Energetic Mater. 2005, 23, 1.