Heat release measurements on micron and nano-scale aluminum powders

Article abstract of DOI:10.1016/j.tca.2009.01.016

Aluminum nanoparticles are a focus of active research largely due to their attractive oxidation energetics and fast reaction kinetics compared with micron-scale or larger particles. A large number of different aluminum powders are currently available and

Full text of DOI:10.1016/j.tca.2009.01.016

Thermochimica Acta 488 (2009) 1–9
Contents lists available at ScienceDirect
Thermochimica Acta
journal homepage: www.elsevier.com/locate/tca
Heat release measurements on micron and nano-scale aluminum powders
Alexander B. Morgan∗, J. Douglas Wolf, Elena A. Guliants, K.A. Shiral Fernando, William K. Lewis∗
University of Dayton Research Institute, 300 College Park, Dayton, OH 45469, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Aluminum nanoparticles are a focus of active research largely due to their attractive oxidation energetics
and fast reaction kinetics compared with micron-scale or larger particles. A large number of different
aluminum powders are currently available and the present study reports the results from a side-by-side
comparison of several representative types of aluminum nanoparticles. Conventional oxide-passivated
micron-scale and nano-scale powders, as well as organically passivated nanoparticles were studied using
pyrolysis combustion flow calorimetry (PCFC) at temperatures up to 700 ◦C. The reaction onset temper-
ature and extent of oxidation were observed to depend on particle size. A multi-step oxidation process
was observed, with the individual steps becoming increasingly pronounced at smaller particle sizes. The
results from PCFC testing gave useful insight into the oxidation behavior of these materials, especially the
organic passivated particles, but, under these oxidation conditions the materials were not fully oxidized.
© 2009 Elsevier B.V. All rights reserved.
Received 9 December 2008
Received in revised form 7 January 2009
Accepted 12 January 2009
Available online 22 January 2009
Keywords:
Oxygen consumption calorimetry
Nanoparticles
Pyrolysis
1. Introduction
organic heat release and metal heat release, especially if they occur
simultaneously. In the present investigation we wish to study the
Nanoscale aluminum particles have become a recent focus of
active research due to their high heat of reaction and fast reac-
tion kinetics compared to micron-scale or larger particles [1–6].
Because such particles are highly reactive, they are typically pas-
sivated by a naturally occurring aluminum oxide shell. While the
oxide shell represents only a small fraction of the particle mass
for micron and larger particles, for nanoparticles the contribution
of oxide to particle mass can become quite large. Ultimately, this
limits the energy that can be obtained by oxidation of the nanopar-
ticle. To solve this problem, alternative methods such as passivation
using an organic layer are currently being explored [7–14]. To date,
a number of different types of aluminum powders are available
either commercially or from active research groups. The purpose
of the current study is to perform a side-by-side comparison of
the oxidation of several representative types of aluminum particles,
including normal oxide-passivated commercial samples, as well as
those passivated by an organic shell.
Oxidation processes involving aluminum particles have been
the subject of several detailed investigations [2,5,15]. These stud-
ies have utilized methods such as differential scanning calorimetry
(DSC), and thermogravimetric analysis (TGA), which are widely
used and well-understood techniques. However, for the organi-
cally passivated samples mentioned above, these techniques suffer
from the drawback that it can be difficult to distinguish between
oxidation of conventional aluminum particles as well as those pro-
tected by organic coatings. As a result, it is highly desirable to utilize
a technique that can separate the contributions to the overall oxi-
dation process arising from reaction of the metallic and organic
components of the latter. To achieve this separation, we have used
pyrolysis combustion flow calorimetry (PCFC) to study the oxida-
tion of the nanoparticle samples.
Pyrolysis combustion flow calorimetry is a technique that mea-
sures heat release by monitoring oxygen consumption as the
sample reacts/combusts. The heat release associated with the
observed oxygen depletion is then calculated from a relation appro-
priate to the material undergoing oxidation, as we will discuss in
detail below. One advantage to this technique is the small sample
size required (5–50 mg) which is ideal for research efforts when ini-
tial work only yields small amounts of nanoparticles. Additionally,
when the sample is reacted in an air atmosphere, this technique can
isolate oxide-forming reactions from nitride-forming reactions. But
the key advantage of this method is that the combustion of volatile
and non-volatile components of a sample can be measured sepa-
rately. Briefly, the technique works as follows. A sample is exposed
to fast heating under an inert gas flow (100% N2), pyrolyzing any
organic products, and the volatile decomposition products are then
pushed into a high temperature combustion furnace where they are
mixed with a synthetic air blend (80% N2 and 20% O2). An oxygen
sensor downstream records the oxygen depletion resulting from
combustion reactions. In this mode of operation, only species that
are volatile at the pyrolysis temperature may contribute to the O2
depletion signal. Alternatively, the sample is heated in the pres-
ence of synthetic air (80% N2 and 20% O2), and the gas flow is then
∗
Corresponding author.
E-mail addresses: alexander.morgan@udri.udayton.edu (A.B. Morgan),
william.lewis@udri.udayton.edu (W.K. Lewis).
0040-6031/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2009.01.016

2
A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
routed through the furnace to the oxygen sensor as before. In this
mode, both volatile and non-volatile portions of the sample con-
tribute to the oxygen depletion. PCFC is a standardized technique,
ASTM D7309-07, and is sometimes referred to as micro-combustion
calorimetry (MCC). Additional references on the PCFC and some of
the other work that has been undertaken with this instrument are
included at the end of this article in the reference section [16–22]. In
this study, the PCFC technique was used to measure the heat release
of five aluminum powder samples (three nanoscale, one micron
sized, and one nanoscale coated with oleic acid) and an oleic acid
control sample using both modes of operation.
2. Experimental procedures
The five aluminum powder samples used in this study are
described as follows:
•
Micron-Al: 100 ␮m size powder, grey color. Purchased from Strem
Inc.
•
Nano Al-100A nanoparticles: 100 nm size powder, dark grey color.
Purchased from Aldrich.
Fig. 1. Schematic representation of pyrolysis combustion flow calorimetry (PCFC)
instrument.
•
Nano Al-100B nanoparticles: 100 nm size powder, dark grey color.
Purchased from Alfa Aesar.
•
Nano Al-10 nanoparticles: 10–20 nm size powder, dark grey color.
Purchased from MTI.
Nano Al-OA nanoparticles: aluminum metal nanoparticles
coated with oleic acid shell. Synthesized via sonochemistry at
UDRI/WPAFB [10].
the entire sample is exposed to oxidizing conditions and the result-
ing signal should contain contributions from reaction of aluminum
metal as well as any organic material.
•
As mentioned above, in the PCFC technique heat release is cal-
culated from the observed oxygen consumption. Thornton [23] has
shown that for most organics, the net heat of combustion per gram
of oxygen consumed is remarkably constant at 13.1 0.6 kJ/g O2.
This relation, known as Thornton’s rule [24], is used to calculate
heat release from the observed oxygen consumption for organic
samples. For the oxidation of aluminum, we instead use the heat
of reaction associated with formation of Al2O3, namely 34.9 kJ/g
O2 consumed [25], to calculate the heat release from aluminum
containing samples.
All commercial aluminum powders studied in this project were
used as received. The oleic acid control sample was tested under
identical conditions to the aluminum powder samples and was also
used as received. All samples were tested in triplicate, with the
exception of the oleic acid sample, which was tested in quadrupli-
cate due to some observed scatter during testing.
All samples were weighed using
a Mettler-Toledo UM-3
microbalance and small metal spatula. The oleic acid sample had
to be metered out using a micropipetter since it was a liquid at
room temperature. These materials were subjected to TEM, XRD for
nanoparticle analysis and PCFC for heat release/oxygen consump-
tion measurements.
Transmission electron microscopy studies (TEM) images were
obtained using a Hitachi H-7600 instrument. First the nanoparticle
powder samples were sonicated in hexane to disperse and then
these solutions were spotted onto a 300 mesh carbon coated grid.
All the TEM images were taken at 100 kV accelerated voltage.
X-ray powder diffraction (XRD) analysis was performed on
a Bruker D8-Advanced equipped with a Cu␣ source and Sol-X
detector. For analysis solid powder samples were placed on a zero-
background holder. In typical analysis, the scans were run over a
2Â range from 5◦ to 85◦, with an angle increment of 0.05◦ and the
collection time of 5 s/angle increment.
3. Results and discussion
All aluminum powder samples were tested with PCFC on a
MCC-1 unit (Govmark Organization Inc., Farmingdale, NY), shown
schematically in Fig. 1. Samples were tested as per ASTM D7309-07.
A 3–4 mg sample was placed in the Specimen Chamber and heated
at 1 ◦C/s to 700 ◦C. Two methods from ASTM D7309-07 were used
to study the samples. In method A, the pyrolyzer gas flow was set to
100% N2. Volatile decomposition species produced during heating
are pushed into a 900 ◦C combustor, where the atmosphere is set to
a 20% O2/80% N2 mixture, which is a composition very close to that
of air. Oxygen depletion resulting from combustion reactions are
monitored using an oxygen sensor downstream. Since the volatil-
ity of aluminum is low below 700 ◦C [25], we would not expect it to
reach the combustor, hence use of method A should give no mea-
sured heat release for aluminum metal, but will be able to measure
the heat release from combustion of the oleic acid coating on the
Nano Al-OA sample, as well as any heat release generated by pyroly-
sis of our oleic acid control sample. In method B, both pyrolyzer and
combustor gas composition set to a 20% O2/80% N2 mixture. Oxy-
gen depletion is measured as in method A. In this configuration,
3.1. Nanoparticle characterization
Before collecting heat release data on the aluminum nanopow-
ders, they were characterized in terms of their particle size,
chemical composition, crystal structure, and morphology. X-ray
diffraction analysis was used to identify the composition of the
samples and the particles’ crystallinity while transmission electron
microscopy (TEM) was instrumental in evaluating the size of the
particles, their size distribution, and the overall sample’s morphol-
ogy.
Fig. 2 shows representative TEM images taken for two different
Nano Al-100 samples. Although the average particle size appears to
be similar for both samples, Nano Al-100A demonstrates a broader
size distribution. Both samples contain approximately spherical
particles with diameters in the range of ∼60–150 nm for Nano Al-
100B and ∼30–250 nm for Nano Al-100A, respectively. However,
compared to the obviously large Micron-Al sample in Fig. 3(left),
these samples can clearly be considered nanoscale in size. The
Nano Al-10 sample (Fig. 3, right) does indeed have some small

A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
3
Fig. 2. TEM images of Nano Al-100A (60,000× mag., left) and Nano Al-100B (50,000× mag., right).
10 nm sized nanoparticles present, but it also has larger particles
and a predominance of rod-shaped nanostructures in compari-
son to the spherical shapes of the other aluminum nanopowder
samples. The aluminum-oleic acid core-shell nanoparticles (Fig. 4)
are the smallest, showing an average size of ∼30 nm and the
size distribution of ∼10–70 nm (Fig. 4) but agglomerates of larger
sizes can be seen in the micrograph. Only the Micron-Al sample
appears to be relatively aggregate free; the smaller particles in
the rest of the samples readily aggregate during or after synthe-
sis.
The results of XRD analysis of one of the commercial samples
and our Al oleic acid core-shell nanoparticles is represented by the
spectra obtained for Nano Al-100B and Nano Al-OA in Fig. 5. Each
spectrum is characterized by five narrow features in the vicinity of
38◦, 45◦, 65◦, 78◦ and 82◦, respectively. The peaks in this spectrum
are well matched with face-centered-cubic aluminum (0) [26]. The
evident high intensity and narrowness of the peaks in both spec-
tra imply a high degree of crystallinity. The broad peak centered at
22◦ of Nano Al-OA sample most likely originates from oleic acid
[10]. In addition to these peaks, minor peaks appeared in some
Fig. 3. TEM images of Micron-Al (50,000× mag., left) and Nano Al-10 (40,000× mag., right).

4
A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
3.2. Nanoparticle oxygen consumption calorimetry studies
Results from the PCFC experiments on each of the materials
are collected in Table 1. The data in the table covers the following
measurements:
•
Char yield: This is obtained by measuring the sample mass before
and after pyrolysis.
•
Char notes: Description of the sample residues collected from
each test.
•
Total heat release (HR) for combustion of organic: Calculated from
the observed oxygen depletion using Thornton’s rule (13.1 kJ/g
O2) [23,24]. The precision of the oxygen consumption measure-
ments is expected to be in line with those listed in the ASTM D
7309-07 standard for the PCFC technique, and is estimated to be
6%.
•
% Sample consumed (via organic combustion): Estimate of the
percentage of the sample reacted obtained from the ratio of the
measured total HR (kJ/g) to the theoretical HR expected assuming
all of the sample undergoes complete combustion.
•
Total HR for oxidation of Al metal: Calculated from observed oxy-
gen depletion using heat of reaction of Al metal to form Al2O3
(34.9 kJ/g O2) [25]. We estimate that the heat release values listed
in the table are accurate to approximately 10%.
•
% Sample consumed by Al oxidation: Estimate of the percentage of
sample reacted obtained from the ratio of the measured total HR
(kJ/g) to the theoretical HR expected assuming the entire sample
mass undergoes complete oxidation to form Al2O3. Theoretical
heat release for this reaction is 31 kJ/g Al burned [25].
Fig. 4. TEM image of Nano Al-OA (50,000× mag.).
•
of the commercial aluminum samples which do not match with
the pattern for any well-characterized phase of aluminum oxide.
The lack of well defined oxide signals suggests that any oxide on
these aluminum particles is either of too low a concentration to be
detected, or, is amorphous in nature and does not yield a coherent
XRD signal [2]. These XRD data suggest that the aluminum samples
used in this study contain significant amounts of zero-valent FCC
aluminum.
% Sample consumed calculated by mass change: Estimate of the
percentage of the sample reacted obtained from the measured
char yield, assuming that the observed mass increase is the result
stoichiometric oxidation of Al to form Al2O3.
Before discussing the aluminum nanoparticle data, it should be
mentioned that this technique was originally developed to mea-
sure combustion of organic samples [20] and with that in mind, we
Fig. 5. XRD data of Nano Al-100B (left) and Nano Al-OA (right).

A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
5
Table 1
Char yield, calculated heat release, and extent of reaction (average of 3–4 runs) for aluminum nanoparticles.
Sample (atmosphere)
Char yield Char notes
(wt.%)
Total HR
% Sample consumed
Total HR (Al
oxidation) (kJ/g)
% Sample consumed
Al oxidation From mass change
(organic) (kJ/g) (organic combustion)
Neat oleic acid (nitrogen)
Neat oleic acid (air)
Micron-Al (nitrogen)
Micron-Al (air)
Nano Al-100A (nitrogen) 103
Nano Al-100A (air) 135
Nano Al-100B (nitrogen) 108
Nano Al-100B (air)
Nano Al-10 (nitrogen)
Nano Al-10 (air)
0.2
3.6
100
111
Little brown film around bottom
Black film all over pan
Grey powder to silvery beads
Brownish ash, no white
Grey powder to black powder
Light grey powder, some white ash
Grey powder to black powder
Light grey powder, some white ash
Grey powder to black powder
Light grey powder, some white ash
Black powder unchanged
32
32
–
–
–
–
–
–
–
–
8.3
9.2
–
87
87
–
–
–
–
–
–
–
–
–
0
1
0
7.4
0
9.8
0
6.2
–
24.5
1.9
–
–
0
3
0
24
0
32
0
20
–
79
6
–
–
–
13
–
39
–
42
–
26
–
–
–
137
115
123
61
69
69
–
Nano Al-OA (nitrogen)
Nano Al-OA (air) peak #1
Nano Al-OA (air) peak #2
23
25
–
Dark grey, some white ash
Dark grey, some white ash
tested oleic acid by itself under air and nitrogen pyrolysis conditions
to verify proper operation of the instrument. Further, the combus-
tion behavior of the oleic acid is directly related to the organically
coated aluminum nanopowders discussed below since oleic acid
was used to coat the nanoparticle’s aluminum core. From Table 1,
the total heat release of oleic acid is measured by PCFC to be 32 kJ/g,
and this value is reasonably close to the reported heat of combus-
tion for oleic acid of 39.4 kJ/g [27,28]. It is noteworthy that this value
is independent of initial atmosphere (nitrogen or air) and that the
shape of the heat release curves, seen in Fig. 6, are very similar
in both cases with the peak heat release value occurring at prac-
tically the same temperatures. Interestingly, there is more residue
left behind under air conditions (3.6% char yield in air versus 0.2%
in nitrogen) which suggests some incomplete combustion under
these conditions. Still, even with the small amount of incomplete
combustion we can state that the instrument is working properly,
pyrolyzing most of the sample and yielding reasonable heat release
values.
ditions, we observe oxidation in Fig. 7(left) around 620 ◦C, similar to
results reported by Trunov et al. [2]. From the oxygen consumption,
we determine that we only measured 1.0 kJ/g of total heat release
for the Micron-Al sample, compared with the theoretical value of
31 kJ/g Al [25], which implies incomplete oxidation of the sample.
Using the measured heat release, we can estimate that only 3% of
the metal was oxidized. Incomplete combustion is supported by the
measured weight increase in the char yield, as there was only an
11 wt.% increase. Complete stoichiometric reaction would increase
the sample mass by 89 wt.%, so using the weight gain numbers we
can estimate a 13% oxidation of the Micron-Al sample. Although
these estimates are a bit different, they agree that the vast major-
ity of the sample remains unreacted. Further confirmation that the
sample was not fully oxidized is found in the fact that the Micron-Al
sample at the end of the test was brown in color, whereas Al2O3
is white in color. Of course, this result is not surprising since it
has been shown previously that oxidation is incomplete at these
temperatures [2].
When studying the micron-sized aluminum powder, we see
from Table 1 that under nitrogen conditions there is no heat release
observed (no oxygen consumed). This makes sense because Al is
non-volatile at the temperatures studied and cannot be carried to
the 900 ◦C combustion furnace. Under the 80% N2/20% O2 (air) con-
For the Nano Al-100A and Al-100B samples, we again observe
(Table 1) no heat release under nitrogen pyrolysis conditions but
do observe some weight gain in the recovered sample, likely
due to the formation of AlN [29,30]. Since there is more sur-
face area available with this material compared to the Micron-Al
Fig. 6. Heat release rate curves, oleic acid pyrolyzed under N2 (left) or 80% N2/20% O2 (right) atmospheres. Four runs are plotted for each.

6
A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
Fig. 7. Heat release rate curves, Micron Al (left), and Nano Al-10 (right). Three runs are plotted for each.
sample, it makes sense that weight gain due to AlN formation
would be observable under these conditions while no significant
weight gain occurred with the Micron-Al sample. Recall that the
PCFC method does not detect heat release caused by the forma-
tion of AlN, since no oxygen is consumed. On the other hand,
when these samples are heated in air, we observe similar heat
release curves (Fig. 8) between the Al-100A and Al-100B samples
and peak oxidation/heat release at ∼580 ◦C, with more intense
heat release values than that seen with the Micron-Al sample.
The lower onset temperature of heat release and higher inten-
sity of reaction are in agreement with results seen elsewhere for
nanoscale Al particles [5]. Interestingly, we find a shoulder on the
heat release signal on the low temperature side. As we will see,
this feature will become even more pronounced at smaller particle
sizes.
While these total heat release values (average values of 7.4 and
9.8 kJ/g for A and B, respectively) are larger than the Micron-Al
sample, they are still significantly smaller than the either the the-
oretical value (31 kJ/g Al) or recent experimental value [5] (23 kJ/g
Al), which again implies that incomplete oxidation occurred under
these experimental conditions. The estimates of the extent of oxida-
tion using heat release again differ somewhat from those calculated
from the mass increase, but both agree that the majority of the alu-
minum metal is not oxidized under these conditions. We would
anticipate that the estimates obtained using the mass increase
would be larger than those from the heat release (calculated from
the oxygen depletion), since formation of AlN contributes only to
the former. Actually, this may be advantageous since the differ-
ence between the two measurements allows us to separate the
contributions to the overall oxidation of aluminum particles in air
Fig. 8. Heat release rate curves, Nano Al-100A (left) and Nano Al-100B (right). Three runs are plotted for each.

A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
7
resulting from formation of Al2O3 and AlN. Finally, we find that
the sample appearance after the test conditions is also consistent
with incomplete oxidation of the particles, in that a grey sample
with white coloration was noted rather than a completely white
sample.
As mentioned above, the fraction of the particle mass con-
tributed by the oxide shell begins to become significant when
the particle size is decreased to the nanoscale. Although a pre-
existing oxide layer is present on these particles, this fact cannot
explain quantitatively that the observed heat release was indicative
of incomplete aluminum oxidation. Specifically, if we calculate the
amount of Al in a 100 nm particle with an assumed 3 nm Al2O3 shell
(using literature bulk densities for Al and Al2O3) [25] we determine
that these starting particles were ∼80 wt.% Al, and ∼20 wt.% Al2O3.
Even taking this into account, the fact that the measured extent of
reaction is significantly less than 80% still implies that oxidation of
the active aluminum content is incomplete.
We now examine our final sample, Nano Al-OA, the 30 nm alu-
minum nanoparticles protected by an oleic acid coating. When
considering the heat release for these samples one must keep in
mind that this particle has mixed metal and organic components
which may react with each other as well as with oxygen, and there-
fore the behavior is expected to be more complex than for the
above mentioned aluminum samples. It is helpful to first exam-
ine the heat release of the organic portion of these particles under
nitrogen pyrolysis. Since the aluminum will not be volatile under
these temperatures, it will not reach the combustor, hence all of
the signal should arise from the oleic acid coating. In Fig. 9(left),
we find that under nitrogen two peaks of heat release are observed
for the Nano Al-OA sample (Fig. 9, left). This may suggest that the
oleic acid bound to the aluminum breaks up into two molecules.
We speculate that there could be two types of oleic acid on the
aluminum nanoparticle surface (chemisorbed and physisorbed).
The physisorbed oleic acid would account for the first peak at
150–350 ◦C, and then the chemisorbed oleic acid would pyrolyze
between 350 and 500 ◦C. An alternate hypothesis that the first peak
would be from a bond breakage at the cis-alkene on oleic acid, and
the second peak would be from the rest of the oleic acid fragment
still bound to the aluminum surface. Further work will clearly be
needed to clarify this issue.
The char yield under these pyrolysis conditions was 61%, indi-
cating that the nanoparticles are at least 39% organic in content, but
we must take this as a lower boundary for two reasons. First, the
aluminum may pick up weight from reaction with nitrogen (AlN
formation) as the organic protecting layer is removed. Secondly,
we cannot rule out the possibility that some sort of char has been
left behind under these conditions, particularly since the sample is
black in appearance at the end of the test, and, aluminum oxide is
known to be a coking (charring) catalyst [31–34]. Since we know
that the sample is at least 39% organic, we would expect to see
approximately 40% of the heat release observed for oleic acid, which
should give a value of 12.5 kJ/g. Instead we observe only 8.3 kJ/g,
which would correspond to combustion of only 23% of the sample
mass, or about half of the organic content. This may indicate that
some of the organic coating formed by the sonochemical process
to make this sample contains sufficient oxygen content to allow
the aluminum sample to self-oxidize under pyrolysis conditions,
thereby lowering the oxygen consumption detected in the appa-
ratus. On the other hand, formation of a carbon-rich char layer
would trap some of the potential heat release (from oleic acid) in the
condensed phase and would thus prevent it from being pyrolyzed
and carried to the 900 ◦C combustion furnace. At this point, these
explanations are still speculative. Nevertheless, these data provide
a valuable comparison to the results obtained in air.
Under air, we observe different heat release behavior for the
Nano Al-OA sample, shown in Fig. 9(right). In the same temperature
region observed under nitrogen pyrolysis conditions (150–550 ◦C,
heat release on left axis), we only observe one broad peak of heat
release (peak #1 in Table 1) in air rather than the two seen under
nitrogen. However, we also observe a much sharper heat release
event around 600 ◦C (peak #2 in Table 1, heat release on right axis
from 550 to 700 ◦C). In Fig. 10, we show an expanded view of peak #2
and find that this signal is actually a doublet, very similar in shape to
that observed in the 10 nm sample discussed above. The observed
temperature range for this feature for these 30 nm particles is only
slightly lower than that for the 100 nm samples above. Since this
doublet is similar in shape and onset temperature to that observed
for other nano-aluminum samples, it seems clear that this signal is
due to Al oxidation. It is interesting that at temperatures higher than
peak #1, we see little activity until the onset of Al oxidation near
600 ◦C. This, combined with the fact that peak #2 is so similar to the
signals observed for conventional nanoparticles implies that fol-
lowing removal of the organic passivation layer, the nanoparticles
Turning our attention to the results for the Nano Al-10 powder
found in Table 1, we again observe no heat release/oxygen con-
sumption under nitrogen, but do observe weight gain at the end
of the experiment, even more so than that observed for the Nano
Al-100A and Nano Al-100B powders. This makes sense in that the
Nano Al-10 powder, which is on average 10–20 nm in size, has even
more surface area than the 100 nm aluminum powders. Again, any
heat release that occurs from AlN formation will not be detected by
PCFC. When the sample is heated in air (Fig. 7, right) we observe a
pronounced two peak curve with peak heat release events at 515
and 535 ◦C. This behavior seems reminiscent of the results for the
100 nm powders, where we saw a single peak with a prominent
shoulder at lower temperatures. The fact that we now see two
peaks in the present case implies that the oxidation mechanism
that generated the shoulder in the 100 nm case has become even
more important here. This observation is consistent with the multi-
step oxidation mechanism proposed by Trunov et al. [2], where
the first oxidation step involves diffusion-controlled growth of an
amorphous oxide layer, followed by a second step characterized by
a phase transition to ␥-Al2O3. Since ␥-Al2O3 is more dense than
the amorphous phase, platelets form and expose some of the alu-
minum core. The exposed metal then rapidly oxidizes. Although for
micron-sized particles the initial diffusion-controlled reaction was
observed to be slow compared to the oxidation rate observed imme-
diately following the phase transition [2], these results suggest that
the two oxidation steps may become increasingly competitive as
the particle size is decreased.
Referring to Table 1, we find that the measured heat release for
the Nano Al-10 powder is again lower than the theoretical value,
implying incomplete oxidation/combustion. This is supported by
the extent of reaction estimated from the measured char yield. Here
again we find that the estimate of the extent of reaction obtained
from the mass increase is larger than that from the heat release,
since formation of AlN can contribute to the former. Interestingly,
the amount of mass gain and heat release are lower than those
observed for the 100 nm Al powders, which is the opposite trend
of increasing mass gain and heat release that we had observed in
going from micron-sized to 100 nm aluminum powders. However,
this can be explained by considering the contribution of the pre-
existing oxide shell. A 10 nm diameter aluminum particle with an
assumed 3 nm oxide shell would be only ∼20 wt.% aluminum and
80 wt.% oxide (again assuming bulk densities) [25]. So when looking
at our estimates of oxidation by mass gain and heat release (Table 1),
this number is in good agreement with the observed extent of reac-
tion, which points to nearly complete combustion of the remaining
active aluminum for this sample. This is noteworthy considering
that the majority of the sample was already oxidized prior to heat-
ing due to the passivating oxide shell that spontaneously forms in
air.

8
A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
Fig. 9. Heat release rate curves, Nano Al-OA in nitrogen (left) and Nano Al-OA in air (right). At right, two axes are shown; one corresponding to organic combustion according
to Thornton’s rule and the other corresponding to oxidation of Al to form Al2O3 (see text). Three runs are plotted for each.
become oxide-passivated. At higher temperatures, they then follow
the same oxidation mechanisms discussed above for conventional
particles.
At this point the relative contributions to peak #1 from combus-
tion of the organic shell and from oxidation of exposed aluminum
once the organic shell is removed are unclear. In Table 1, we list
two heat release values for peak #1 for the Nano Al-OA sample; an
organic heat release value that assumes the detected oxygen con-
sumption is entirely due to organic combustion, and an Al oxidation
value that assumes peak #1 is solely a result of Al2O3 formation.
Although at present we cannot precisely determine the relative
contribution of each, we can see that the oxygen consumption cor-
responding to peak #1 is only ∼10% larger than that measured for
the sample under nitrogen pyrolysis conditions, suggesting that
even in air the majority of this peak comes from the organic com-
ponent of the nanoparticle.
Returning to peak #2, we note that the heat release associated
with peak #2 is only 1.9 kJ/g, and can be explained by oxidizing a
quantity of Al metal that represents just 6% of the sample mass.
Even if we assume that 10% of peak #1 is due to Al oxidation as
mentioned above, the heat release can still be explained by com-
bustion of only ∼14% of the sample mass. This seems odd, since this
extent of oxidation is somewhat lower than that observed for larger
100 nm particles (refer to Table 1). It is similar to the extent of oxi-
dation observed for 10 nm particles, but there ∼80% of the sample
aluminum was already consumed by the naturally occurring oxide
shell. This observation may be an indication that during combus-
tion of the organic layer, a sort of char is left behind which continues
to protect the aluminum to some degree. Alternatively, this may be
evidence that the Nano Al-OA particles still possess some oxide.
Indeed, this possibility has been suggested elsewhere [10]. What-
ever the reason, these results are clearly consistent with incomplete
oxidation of the sample, a finding that has been confirmed using
XRD on samples after heating to 700 ◦C [10].
4. Conclusions
The PCFC method was used to measure the heat release of
five aluminum samples through the use of oxygen consumption
calorimetry, and some general trends were observed.
1. For conventional oxide-passivated aluminum particles, we find
(in agreement with other researchers) [5] that the onset temper-
ature for reaction decreases with particle size. For micron-sized
particles, reaction is observed between ∼600 and 650 ◦C. By the
Fig. 10. Expanded view of heat release peak #2 for Nano Al-OA in air. Three runs are
plotted.

A.B. Morgan et al. / Thermochimica Acta 488 (2009) 1–9
9
time the particle size has decreased to 10 nm, the threshold has
been lowered to ∼500–550 ◦C.
hope to begin chemical analysis of the materials left after pyroly-
sis events to better characterize the chemical reactions that have
occurred and further explore understand how nanoparticle chem-
istry affects the formation rates of alumina and aluminum nitride.
Further, we plan to analyze these samples under Ar gas rather than
N2 so that we can better quantify the AlN formation observed in the
current results.
2. The extent of reaction for temperatures under 700 ◦C is observed
to increase as the particle size is decreased. Combustion of
micron-size particles is far from complete, but the extent of
oxidation increases for 100 nm samples, and all of the active alu-
minum content of smaller 10 nm particles is fully oxidized or
nearly so. However, for such small sizes we note the majority of
the aluminum is already “consumed” by the naturally occurring
oxide layer.
Acknowledgments
3. As aluminum particles are heated, we detect oxygen consump-
tion occurring in two distinct peaks. Following previous work [2],
we assign these to a slower diffusion-controlled oxidation step
and more rapid oxidation step following a phase change in the
Al2O3 layer. In micron-scale particles, only one peak is readily
visible, presumably from the faster second step. As the particle
size is decreased to the nanoscale, the diffusion-controlled step
becomes more prominent, and for 10 nm particles, the two steps
become competitive.
4. For the aluminum nanoparticles passivated by oleic acid (30 nm
average size), the observed oxidation behavior is more complex
than that observed for conventional nanoparticles. At temper-
atures between 200 and 500 ◦C the organic coating combusts.
Following removal of the organic passivation layer, the nanopar-
ticles appear to become oxide-passivated (or char-passivated)
and subsequently behave like conventional aluminum nanopar-
ticles at higher temperatures.
The authors would like to thank Mary Galaska for collecting the
PCFC data, and the Defense Threats Reduction Agency (DTRA) for
funding under grant #HDTRA-07-1-0026. Also, we would like to
thank Chris Bunker of AFRL/RZPF (Fuels Branch) for technical dis-
cussions and support on use of XRD and TEM instrumentation to
characterize the nanoparticles.
References
[1] P.W. Cooper, Explosives Engineering, Wiley–VCH, New York, 1996.
[2] M.A. Trunov, M. Schoenitz, X. Zhu, E.L. Dreizin, Combust. Flame 140 (2005)
310–318.
[3] L.P.H. Jeurgens, W.G. Sloof, F.D. Tichelaar, E.J. Mittemeijer, J. Appl. Phys. 92 (2002)
1649–1656.
[4] L.P.H. Jeurgens, W.G. Sloof, F.D. Tichelaar, E.J. Mittemeijer, Thin Solid Film 418
(2002) 89–101.
[5] J. Sun, M.L. Pantoya, S.L. Simon, Thermochim. Acta 444 (2006) 117–127.
[6] A.P. Il’in, A.A. Gromov, G.V. Yablunovskii, Combust. Explo. Shock Waves 37
(2001) 418–422.
[7] R.J. Jouet, A.D. Warren, D.M. Rosenberg, V.J. Bellitto, K. Park, M.R. Zachariah,
Chem. Mater. 17 (2005) 2987–2996.
The data collected from the PCFC does give us some insight into
the reactions between Al nanoparticles and oxygen. Although we
are not able to push the samples to full oxidation in the conditions
used in the PCFC (700 ◦C), we are still able to observe and measure
amounts of oxidation in bulk aluminum powders. However, the data
suggests that under typical air mixtures the aluminum will react
with oxygen and nitrogen at elevated temperatures. The formation
of Al2O3 and AlN are both exothermic, something that would be
observed with DSC or oxygen bomb calorimetry. Of course, since
both reactions would occur simultaneously in an air atmosphere, a
separation strategy would need to be employed in order to quan-
tify the relative contributions. The PCFC data may indicate that
the maximum amount of Al2O3 that can be formed up to 700 ◦C
is that measured in this paper – with the difference of mass gain
and potential heat release coming from AlN rather than Al2O3. The
Nano Al-OA nanoparticle presents an interesting system in that it
gives off heat from hydrocarbon combustion (the oleic acid coating)
at very low temperatures (as early as 150 ◦C) and then again at ele-
vated temperatures once the aluminum activates. The complexity of
this system indicates that more work is needed to understand how
organic molecules and metals behave at their interface under com-
bustion conditions, and, how that influences combustion/oxidation
behavior.
[8] R.J. Jouet, R.H. Granholm, H.W. Sandusky, A.D. Warren, in: M.D. Furnish, M. Elert,
T.P. Russell, C.T. White (Eds.), Shock Compression of Condensed Matter – 2005,
Am. Inst. Phys., 2006.
[9] C.E. Bunker, J.K. Karnes, J. Am. Chem. Soc. 126 (2004) 10852–10853.
[10] K.A.S. Fernando, M.J. Smith, B.A. Harruff, W.K. Lewis, E.A. Guliants, C.E. Bunker,
J. Phys. Chem. C 113 (2009) 500-503.
[11] A.C. Templeton, W.P. Wuelfing, R.W. Murray, Acc. Chem. Res. 33 (2000) 27–36.
[12] R.W.J. Scott, O.M. Wilson, R.M. Crooks, J. Phys. Chem. B 109 (2005) 692–704.
[13] A.C. Templeton, M.J. Hostetler, C.T. Kraft, R.W. Murray, J. Am. Chem. Soc. 120
(1998) 1906.
[14] Y.-S. Shon, G.B. Dawson, M. Porter, R.W. Murray, Langmuir 18 (2002) 3880.
[15] M.A. Trunov, S.M. Umbrajkar, M. Schoenitz, J.T. Mang, E.L. Dreizin, J. Phys. Chem.
B 110 (2006) 13094.
[16] P.M. Hergenrother, C.M. Thompson, J.G. Smith, J.W. Connell, J.A. Hinkley, R.E.
Lyon, R. Moulton, Polymer 46 (2005) 5012–5024.
[17] J.L. Jurs, J.M. Tour, Polymer 44 (2003) 3709–3714.
[18] H. Zhang, R.J. Farris, P.R. Westmoreland, Polymer 36 (2003) 3944–3954.
[19] R.N. Walters, R.E. Lyon, J. Appl. Polym. Sci. 87 (2002) 548–563.
[20] R.E. Lyon, R.N. Walters, J. Anal. Appl. Pyrol. 71 (2004) 27–46.
[21] B. Schartel, K.H. Pawlowski, R.E. Lyon, Thermochim. Acta 462 (2007) 1–14.
[22] A.B. Morgan, M. Galaska, Polym. Adv. Technol. 19 (2008) 530–546.
[23] W.M. Thornton, Philos. Mag. 33 (1917) 196–203.
[24] E.H. Battley, in: R.B. Kemp (Ed.), Handbook of Thermal Analysis and Calorimetry,
vol. 4, Elsevier, Amsterdam, 1999, pp. 219–265.
[25] CRC Handbook of Chemistry and Physics, CRC Press, ISBN 978-1420066791.
[26] P.R. Watson, M.A. Van Hove, K. Hermann, J. Phys. Chem. Ref. Data 5 (1994)
1–1341.
[27] L. Keffler, J.H. McLean, J. Soc. Chem. Ind. (1935) 178–185.
[28] A.G. Emery, F.G. Benedict, Am. J. Physiol. 28 (1911) 301–307.
[29] N. Glumac, J. Appl. Phys. 98 (2005) 053301.
[30] R.K. Paul, K.-H. Lee, B.-T. Lee, H.-Y. Song, Mater. Chem. Phys. 112 (2008) 562–565.
[31] N. Cinausero, N. Azema, M. Cochez, et al., Polym. Adv. Technol. 19 (2008)
701–709.
[32] M. Soyama, K. Inoue, M. Iji, Polym. Adv. Technol. 18 (2007) 386–391.
[33] A. Laachaci, M. Chochez, E. Leroy, M. Ferriol, J.M. Lopez-Cuesta, Polym. Degrad.
Stab. 92 (2007) 61–69.
The challenge of quantifying nanoscale behavior on a bulk sam-
ple is not fully solved with the use of PCFC, but information is
gained which should be useful in the development of new nanoscale
aluminum powders, or new techniques which can combine tradi-
tional calorimetry with evolved gas and condensed phase analysis
to provide the quantification of nanoparticle performance which is
needed to move this field of research forward. In future work we
[34] F. Laoutid, L. Ferry, J.M. Lopez-Cuesta, A. Crespy, Polym. Degrad. Stab. 82 (2003)
357–363.