Propagation studies of metastable intermolecular composites (MIC)

JOURNAL OF APPLIED PHYSICS 98, 064903 �2005�

Combustion velocities and propagation mechanisms of metastableinterstitial composites

B. S. Bockmon and M. L. Pantoyaa�

Department of Mechanical Engineering, Texas Tech University, Lubbock, Texas, 79409

S. F. Son, B. W. Asay, and J. T. MangLos Alamos National Laboratory, Los Alamos, New Mexico, 87545

�Received 4 January 2005; accepted 4 August 2005; published online 28 September 2005�

Combustion velocities were experimentally determined for nanocomposite thermite powderscomposed of aluminum �Al� fuel and molybdenum trioxide �MoO3� oxidizer under well-confinedconditions. Pressures were also measured to provide detailed information about the reactionmechanism. Samples of three different fuel particle sizes �44, 80, and 121 nm� were analyzed todetermine the influence of particle size on combustion velocity. Bulk powder density was variedfrom approximately 5% to 10% of the theoretical maximum density �TMD�. The combustionvelocities ranged from approximately 600 to 1000 m/s. Results indicate that combustion velocitiesincrease with decreasing particle size. Pressure measurements indicate that strong convectivemechanisms are integral in flame propagation. © 2005 American Institute of Physics.�DOI: 10.1063/1.2058175�

I. INTRODUCTION

A thermite reaction is a highly exothermic reaction be-tween a metal and another metal or metallic oxide. When thefuel-metal species is Al and the oxidizer is a metallic oxide�i.e., CuO3, Fe2O3, MoO3�, the reaction temperature and heatof combustion can be sufficiently high such that the compos-ite is ideal for energetic material applications.1 Thermitecomposites are of great interest in combustion studies, par-ticularly due to their varied ordnance applications includingprimers and other igniters.2 Wang et al.3 provide a thoroughreview of thermite reactions and discuss additional thermiteapplications.

Nanoscale Al particles are becoming more readily avail-able and provide an opportunity to enhance the reactivepower of a mixture by increasing the combustion velocity. Inthis article we consider spherical particles of nanoscale Aland define nanometer Al particles as a particle with a diam-eter of 100 nm or less. Particles with other geometries areconsidered nanoscale if one or more of their major dimen-sions are less than 100 nm.

Research has shown that increased combustion velocitiescan be achieved with smaller particle composites.4 For ex-ample, Rugunanan and Brown5 showed that pyrotechniccombustion velocities increased with decreasing particle size�in the micron range� of the constituents. Brown et al.6 var-ied the particle size of the fuel species in the Sb/KMnO4

system and found that reducing particle size from 14 to2 �m increased the burning rate from 2–8 to 2–28 mm/s.These results suggest that reducing the fuel particle size tothe nanoscale range may result in revolutionary combustionvelocity performance. This theory was further substantiated

a�Author to whom correspondence should be addressed; electronic mail:
[email protected]
0021-8979/2005/98�6�/064903/7/$22.50 98, 06490

Downloaded 31 Oct 2005 to 128.165.118.133. Redistribution subject to

by Shimizu and Saitou7 in their evaluation of the effect ofcontact points on reaction rate. They found that in theFe2O3–V2O5 system, increasing the number of contactpoints between fuel and oxidizer particles increased reactionrates. In another study, Aumann et al.4 examined the oxida-tion behavior of nanometer Al powders. They suggest that Alpowder mixtures with average particle sizes of 20–50 nmcan react 1000 times faster than conventional powdered ther-mites owing to reduced diffusion distances between indi-vidual reactant species.4 Both of these studies indicate that ahigher degree of intimacy in the mixing of fuel and oxidizerparticles results in improved performance of the energeticcomposite.

The rate at which energy is released �i.e., reactivepower� is related to the combustion velocity and depends ona number of factors including particle size distribution, de-gree of compaction of the mixture, and the degree of inter-mixing of fuel and oxidizer powders. In this study, all na-nometer Al powder samples appear to have log-normalrather than bimodal size distributions and all experimentswere performed with loose powder media, such that the de-gree of compaction was between 5% and 10% of the theo-retical maximum value. Intermixing of reactants wasachieved by suspending the particles in solution and usingultrasonic waves to break up agglomerates and produce awell-mixed composite.

The objective of this study is to examine the combustionbehavior of nanocomposite powder mixtures of Al and mo-lybdenum trioxide �MoO3� as a function of aluminum par-ticle size. Specifically, combustion velocity and pressuremeasurements were analyzed for loose powder mixtures thatare confined in cylindrical tubing. The mechanisms respon-
sible for flame propagation are also discussed.
© 2005 American Institute of Physics3-1

AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

http://dx.doi.org/10.1063/1.2058175

http://dx.doi.org/10.1063/1.2058175

http://dx.doi.org/10.1063/1.2058175

064903-2 Bockmon et al. J. Appl. Phys. 98, 064903 �2005�

II. EXPERIMENT

A. Sample materials

Three Al samples were mixed individually with a singleMoO3 sample to produce three separate mixtures ofAl/MoO3 that differ by the Al average particle diameter. TheMoO3 was held constant through all tests. The Al particlecharacteristics and optimized fuel to oxidizer ratios are listedin Table I. The average Al particle diameter reported in TableI was reported by the Al suppliers and determined using agas adsorption analyzer and Brauner-Emmitt-Teller �BET�theory.8 Pure Al particles are pyrophoric and therefore pas-sivated by an oxide shell of Al2O3. The thickness of thisshell was also reported by the suppliers and determined fromweight-gain measurements using a thermal gravimetric ana-lyzer �TGA�. Even micron-scale Al particles have an oxideshell; however, as Al particle size decreases the percentageof Al2O3 relative to Al increases �Table I�. The term “activeAl content” refers to the portion of the Al powder that isnot in the form of Al2O3. As the particle size increases, theactive Al content also increases because the oxide shellthickness consumes a smaller percentage of the total particlecomposition.

A variable held constant in this study is the fuel-oxidizercomposition. Typically, the combustion velocity of a mixtureis a strong function of the fuel to oxidizer composition. Aprevious work with Al and MoO3 thermites showed that aslightly fuel-rich mixture produces the highest combustionvelocity.9 Granier and Pantoya9 explain that mixture compo-sition is often reported in terms of an equivalence ratio andfurther explain how to calculate an equivalence ratio basedon the active Al content of a nanoscale �or micronscale� Alpowder and the corresponding metal oxidizer. They showedthat an equivalence ratio of 1.2 produced the highest com-bustion velocity for compressed pellets.9 Figure 1 is consis-tent with their results and shows combustion velocities forvarious equivalence ratios ranging from stoichiometric�equivalence ratio=1� to fuel rich �equivalence ratio�1� formedium-density, loose powder mixtures of 80 nm Al com-bined with MoO3 �sample B, Table I�.

Figure 2 displays micrographs of three Al/MoO3

samples. Each micrograph was taken at the same magnifica-tion to provide visual comparison of the particle sizes andthe nature of the interaction between the Al �spherical� and

TABLE I. Aluminum particle characteristics and mixAl for samples A and B was supplied by NanotechnoC was supplied by Technanogies, Inc. �Irvine, CA�.

Al particlediameter

�nm�BET

Al particlediameter

�nm�SAXS

Oxithi

�

SampleA

44 31

SampleB

80 45

SampleC

121 48.6

MoO3 �crystalline sheet� structures. The MoO3 particles pro-


vide an example of how particles with a length of near 1 �mcan still be classified as nanoscale based on the 20 nm sheetthickness. The MoO3 particles will likely remain in this ge-ometry until sublimation begins at roughly 700 °C.

Figure 2 shows that the nanometer scale of the Al par-ticles facilitates increased contact between fuel and oxidizerand allows more homogeneity within the mixture. This im-age shows that the number of contact points between fueland oxidizer significantly increases compared with similarmicron-sized particle mixtures. This phenomenon is mostpronounced for Fig. 2�a� �i.e., sample A� because there aremore particles per unit mass of Al. At the relatively lowdensities shown in the micrographs ��5% theoretical maxi-mum density �TMD��, there is significant agglomeration ofthe Al particles and considerable void space between the fueland oxidizer. These voids create inconsistent diffusion dis-tances between the fuel and oxidizer particles which, itwill be shown later, may lead to inconsistent combustionvelocities.

B. Small-angle x-ray scattering „SAXS…

Particle size distribution was determined using small-angle x-ray scattering �SAXS� and assuming the shape of thedistribution was log-normal. During the fitting process, thewidth and peak positions were allowed to vary freely. Thechoice of the log-normal distribution was made based uponthe work of Granquist and Buhrman.10 While the log-normalshape is strictly valid for most nanoscale particles, Granquist

atios corresponding to Al and MoO3. The nanometers, Inc. �Austin, Texas�. The nanometer Al for sample

ells

ActiveAl

content�%�

Massratio

Al:MoO3

Equivalence.ratio ��

50 42.5/57.5 1.3

62.1 45/55 1.35

79.7 40/60 1.40

FIG. 1. Velocity as a function of equivalence ratio for sample B, medium

ture rlogie

de shcknesnm�

2

2

1.8

density.



and Buhrman10 state that for the larger mean particle sizes�where crystal habits begin to appear�, some variation fromthe log-normal shape can be expected in the wings of thedata. This appears to be an insignificant effect except in thecase of the 121 nm Al, where scanning electron microscopy�SEM� micrographs show particles�1 �m in diameter andgood fits for the small-angle scattering data are difficult toobtain. Several other distribution shapes, including the log-hyperbolic and the skew-Laplace distributions, have been ex-plored with the intention of improving our knowledge of thewings of the distribution. Stable results could not be obtainedwith the log-hyperbolic distributions, while reasonable fitswere obtained with the skew-Laplace distribution. However,the log-normal distribution was chosen because it providedthe best and most consistent results. The average Al particlediameter based on SAXS measurements is provided in

FIG. 2. SEM Micrographs of Al/MoO3 composites: �a� sample A; �b�sample B; �c� sample C.

Table I.


C. Apparatus

Powder confinement was provided by open-ended trans-parent acrylic tubing that allowed for visual inspection of thecombustion front. The acrylic tubing had an inner and outerdiameter of 0.3175 and 0.635 cm, respectively. Each tubewas 10.16 cm long and filled with powder. The volume ofthe cavity in the tube was 0.8044 cm3. This tube size wassufficiently large for efficient visual inspection yet smallenough for sample sizes that were within self-imposed safetylimits. All experiments were performed in an ambient airenvironment. Preliminary tests were performed to verify thattube size did not significantly affect combustion velocity. Forsix tubes with inner diameters ranging from 1 to �6 mm,combustion velocity variations, were found to be insignifi-cant. Each experiment resulted in expulsion and dispersionof product particulates from the apparatus such that collec-tion of product residue was difficult and not accomplished.

Acrylic was chosen as the tubing material based on itsmechanical strength advantages over similar glass tubing,which allowed the reaction to proceed completely throughthe tube before its destruction. The ends of the tubing wereleft open to prevent the stagnation of gases ahead of thereaction front. Visual examination of the tubing fragmentssubsequent to the experiment revealed little thermal damage,but there remains the remote possibility for small quantitiesof vaporized acrylic to interact with the main reactions.

For each sample, three bulk densities were studied. Thelow-density samples were prepared by pouring loose powderinto the tubes until the entire tube volume was filled �thisdensity is commonly referred to as the “poured density”�.Medium-density tests were prepared similarly, except eachtube was vibrated for approximately 5 s to compact the par-ticles and allow more powder to fill the tubes. The high-density tests were subjected to constant vibration and filledcontinuously until no additional compaction could be de-tected. The TMD for each mixture as a weighted average ofthe pure solid densities of the three reactants �Al, MoO3, andAl2O3� is 4 g/cm3. The bulk densities for these experimentsranged from 5% to 10% TMD.

After the tubes were filled with powder, they were in-serted into an acrylic block instrumented with fiber-opticphotodetectors and piezocrystal pressure sensors to facilitatethe combustion velocity and pressure measurements �Fig. 3�.The six pressure sensors �PCB 113A22� were installed at1 cm spacing on one side of the tube and optical fibers�Thorlabs M21L01� leading to photodetectors �ThorlabsDET210� which opposed each of the pressure sensors on theother side of the tube, again at 1 cm intervals. The resulting5 cm instrumented section was located in the center of thelength of the tube to allow the reaction to reach steady state.

As the tubes were inserted into the instrumented block,predrilled 1 mm ports in the tubing walls were aligned withthe pressure sensors. These ports were filled with standardclear vacuum grease to thermally insulate the piezocrystalsensors from the high reaction temperatures. This precautionprevents nonlinearities in the electrical response of the pi-ezocrystals that can arise from temperature changes in the
sensor. Effects from the grease on the dynamic response of


the sensors were assumed to be negligible. The full accep-tance cone of the fiber-optic cables �46° � is effectively re-duced to less than 10° by the mounting configuration in theblock. An 8-channel, PCB 482A20 signal conditioner trans-mitted the pressure signals to two Tektronix TDS460A4-channel digital oscilloscopes.

All experiments were performed inside a stainless-steelvessel to contain flying fragments from the small explosions.A 3-cm-thick acrylic-viewing window provided visual accessto the experiment. A Phantom V high-speed camera recordedimages at 62 500 frames/s �fps�. This frame rate allowed a16 �s time resolution between images. A bare, 51-gaugeelectric match head provided the ignition source for thereaction.

FIG. 4. Sequence of still frame images captured for sample B medium
density. Images are roughly 20 �s apart.

III. RESULTS

Figure 4 shows a typical sequence of still frame imagestaken from the high-speed camera that correspond to flamepropagation in sample B medium density �Table I�. The firstsensor location can be seen in image B �note �a��, and theothers become visible in the following frames. Figures 5 and6 show representative data recorded from the pressure sen-sors and photodetectors, respectively, for the reaction illus-trated in Fig. 4. Maximum pressure, the pressurization rate,and the duration of the pressure pulse are recorded from Fig.5. The light intensity histories depicted in Fig. 6 are used todetermine the average combustion velocity.

Combustion velocity is calculated from the slope of a setof position versus time data from each experiment �Fig. 7�.Flame position is known from the location of the opticalsensors. Corresponding times are recorded for each lightpulse to reach a specific intensity. The initial rises in lightintensity are the result of light refracting forward from the

FIG. 3. Acrylic mounting block: �a� photograph of as-sembly; �b� schematic of data collection system.

FIG. 5. Representative 6-channel pressure history for sample B medium
density.


approaching reaction front. These effects are minimized bythe mounting configuration in the acrylic block. Figure 7 is arepresentative position versus time plot illustrating that thedata symbols lie very close to the linear curve fit. The slopeprovides the combustion velocity approximation with a con-fidence interval of 99% based on the R2 value of the curvefit.

Table II summarizes the average values for velocity andpressure. The values reported were averaged over at leastthree repeatability experiments and the standard deviationassociated with each set of measurements is also shown.Each of the three Al/MoO3 samples was divided into threesubcategories by bulk density. The pressure is the approxi-mate maximum pressure reached inside the tube. The actualrecorded pressure is approximated from the maximum sus-tained pressure �i.e., pressure spikes were ignored�. Alsoincluded are the average mass consumption rates for eachexperiment.

Statistical analysis

The uncertainty in these measurements is based on thestatistical significance of the average velocities shown inTable II. In situations with very small sample sizes, the con-fidence interval of the measurements can be determined us-ing the “student’s t distribution” �Eq. �1�� in place of the“standard normal” distribution curve,11

FIG. 6. Representative 6-channel light intensity history for sample B me-dium density.

FIG. 7. Representative “position vs time” plots used to estimate combustion
velocity sample B medium density.

� = x ± t��

2:n − 1�� s

�n� . �1�

In this equation, � is the true mean, x is the average valuefrom the sample, and t is tabulated as a function of �, theconfidence level, and the degrees of freedom r, or n−1. Thestandard deviation of the sample is estimated as s, and n isthe number of tests in the sample.

For the case of the largest standard deviation, �i.e., low-density sample B� it was found with 90% confidence that thetrue mean combustion velocity will lie within 11% of theaverage shown. In the case of the smallest deviation, �high-density sample C� it is predicted with 90% confidence thatthe true mean combustion velocity will lie within 2.5% of theaverage given. The confidence levels for all other series liebetween these two extremes.

IV. DISCUSSION

Table II shows that a decrease in the Al particle diameterincreases the combustion velocity of the composite in thehighest-density samples. The combustion velocity is in-versely proportional to the particle radius squared when

TABLE II. Velocity and pressure measurements for samples A, B, and C asa function of bulk density.

Density Sample A Sample B Sample C

Average velocity �m/s� 959.3 988.7 684.4Standard deviation 35.4 135.6 81.8

Low Pressure �MPa� 10.8 12.4 10.9Standard deviation 0.8 3.6 1.3Mass rate �g/s� 2308 2665 1794Average velocity �m/s� 916.0 988.3 784.7Standard deviation 59.4 56.9 57.0

Medium Pressure �MPa� 17.5 16.5 12.4Standard deviation 1.4 2.5 1.8Mass rate �g/s� 2805 3230 2522Average velocity �m/s� 950.7 948.7 765.3Standard deviation 46.1 37.0 22.5

High Pressure �MPa� 22.1 17.9 18.6Standard deviation 4.5 1.2 1.8Mass rate �g/s� 3600 3607 2856

FIG. 8. Combustion velocity as a function of particle diameter for high-density samples. Error bars represent maximum and minimum velocities for
at least four repeatability tests.


combustion is diffusionally controlled.12 Figure 8 showscombustion velocity as a function of Al particle size for thehigh-density �10% TMD� case. From Fig. 8, a critical diam-eter may exist for which smaller particle diameters do notresult in a further increase in combustion velocity. A limit tomaximizing velocity may be achieved that could result fromphysical or chemical properties of the Al particles. Figure 8suggests that there exists a critical particle diameter between80 and 110 nm, below which mass diffusion cannot be fur-ther enhanced by particle geometry and combustion velocityis independent of particle diameter. Note that an additionalseries of experiments with 110 nm Al was conducted toverify the trend observed with the 121 nm Al when this ma-terial became available. The results using the 110 nm Alcomposite agree with the observations made above.

There are multiple explanations for a combustion veloc-ity that becomes independent of particle diameter �Dp�:

�1� the reaction may undergo transition from diffusion tochemistry controlled;

�2� the geometry associated with the volume to surface arearatio decrease �i.e., on the order of Dp� may counterbal-ance the diffusion-controlled particle combustion veloc-ity �i.e., on the order of 1 /Dp�; or,

�3� although the equivalence ratio is held constant, as the Dp

decreases the Al2O3 content increases and could act as a“dead weight” inhibiting further increases in velocity.Al2O3 is an insulative material and thus acts as a thermalheat sink and resists diffusive heat transport.

In case �1� the rate-limiting step in the chemical reactionbetween Al and MoO3 is a good approximation for the reac-tion rate and therefore combustion velocity. This rate-limiting step is dictated by the chemistry of the reactants andis independent of the bulk density of the mixture. Table IIshows that the combustion velocity is a function of bulkdensity �and therefore not controlled by chemistry�. Thus, theindependence of the combustion velocity on particle size forthe smallest diameter samples may be attributed to adiffusion-controlled reaction that experiences the competingeffects explained in cases �2� and �3�. In summary, this worksuggests that a transition from diffusion- to chemistry-controlled burning is not likely for the Al particle diametersinvestigated here.

In general, Table II shows that the standard deviations invelocity associated with the high-density composites are sig-nificantly less than the lower-density particle composites.Specifically, for the 80- and 121-nm-diameter Al compositesthe variability in combustion velocity is reduced with in-creasing density. The 44 nm Al composites do not followthis trend, but instead show no significant change in combus-tion velocity variability as a function of density. Some of thecombustion velocity variabilities at lower densities may re-late to unpredictable settling of the powder that creates a freevolume above the powder and inside the tube.13

Table II shows increasing peak pressure with increasingbulk density. This effect may be a direct result of increasingthe mass of reactants confined within a constant volume. As
the reaction propagates, more gaseous intermediate products

are produced. The buildup of gaseous reactants, intermedi-ates, and products in the confined region of the tube in-creases the pressure in the tube.

Most derivations for velocity as a function of particlesize are based on the assumption that the mechanism respon-sible for reaction propagation is diffusion. Table II suggeststhat peak pressure and pressurization rates are sufficientlyhigh that convection may play a dominant role as the reac-tion mechanism in powder composites. When convectiveprocesses are significant, a pressure gradient propels the ad-vective transport of energy forward. The reactants and prod-ucts of pure Al/MoO3 composite materials are both solidsaccording to the balanced reaction in which �Hr�n repre-sents the heat of reaction and Taf represents the adiabaticflame temperature,1

2Al + MoO3 → Al2O3 + Mo,

�Hr�n = 4705 kJ/kg, Taf = 3253 K.

However, thermoequilibrium calculations show that 24% ofthe mass is gaseous at Taf,

1 and intermediates or productsmay be gaseous before they condense as they cool. Also, theinitial ambient air between particles that is heated to adia-batic flame temperatures could expand and behave as aworking fluid to transport heat forward. For example, MoO3

begins to sublime at 973 K and therefore, gas phase reactantsare likely present at temperatures well below the adiabaticflame temperature. In addition, analysis of high-speed pho-tographic data shows that for the burning of powders con-fined in tubes, particulates are ejected from the ends of thetube throughout the reaction.14 The fact that these materialsexpel gas and particulates may illustrate the significance ofconvective mechanisms driving the reaction forward. This,coupled with high-pressure �for a thermite reaction� mea-surements, further supports the idea that the convectivemechanism dominates the reaction propagation by drivingmolten Al and gaseous oxidizer into the unreacted material.

For comparison, binary composite thermites composedof micron-scale and larger particles exhibit combustion ve-locities on the order of 10 mm/s.6 Also, at 1 atm pressurehigh melting explosive �HMX� burns at �2 mm/s and deto-nates at 9000 m/s.15 The combustion velocity of the nano-Al/MoO3 composites studied here lies in between these twoenergetic material regimes at approximately 900 m/s. Thespeed of sound in air is approximately 345 m/s and the ve-locities observed in these experiments are approximately900 m/s. Therefore, assuming a gaseous medium �the mix-tures are 90% air by volume�, velocities approaching Mach 3are conceivable. In this velocity regime compressibility ef-fects may be important and further efforts are ongoing toresolve this issue. However, the pressure histories �Fig. 5� donot conclusively signify the contribution of compressibilityas a mechanism for reaction propagation. These results showthat nanoscale thermite composites exhibit combustion ve-locities significantly faster than traditional pyrotechnics andthermites and almost on the order of explosives. These re-sults also suggest that nanocomposite reactants may be tai-lored to control reactive power by controlling the Al particle
diameter and corresponding combustion velocity. The ability


to tailor the reactive power of a mixture to a specific appli-cation would be highly beneficial for both military and in-dustrial purposes.

The term “reactive power” describes the energy releaserate from a given reaction. Reactive power is defined by thefollowing relationship and is expressed in terms of kilowatts:

Power = ��Hcomb��Uav��bulk��Across section� . �2�

�Hcomb is the heat of combustion for the Al+MoO3 re-action, Uav is the average combustion velocity, �bulk is thebulk density of the powder, and Across section is the area per-pendicular to the direction of flame propagation, and in thiscase, is the cross-sectional area of the tube. Reactive powersfor these nanocomposites ranged from over 600 kW in theexperiments that combined lower densities and the slowercombustion velocities to over 1000 kW for the higher-density, high-speed experiments. These values are several or-ders of magnitude smaller than the large reactive power thatwould be released from a typical explosive. For example,composition B �59.5% rapid detonating explosive �RDX�,39.5% trinitrotoluene �TNT�, 1% wax� has a detonation ve-locity in excess of 9 km/s that contributes to its reactivepower of 6.72�108 kW. Conversely, micron-scale thermitecomposites exhibit very low reactive power due to combus-tion velocities on the order of 10 mm/s.

V. CONCLUSIONS

The effect of Al particle size on combustion velocity wasexamined for nanoscale Al particles reacting with MoO3 in aconfined loose powder configuration. The average combus-tion velocity was found to increase from approximately 750to 950 m/s when the particle size was decreased from 121 to80 nm. Below 80 nm, to 44 nm, significant changes in com-bustion velocity were not observed. The reaction becomesindependent of particle diameter below a specific critical di-ameter. This may result from excessive amounts of Al2O3,inherent in the nanometer Al particles, which acts as a ther-
mal heat sink inhibiting further increases in flame propaga-

tion. It was further observed that improved velocity repeat-ability was achieved by increasing the bulk density of thepowder.

Pressure histories and burning behavior also indicate thatconvection plays a significant role in the propagation of thereaction in powder media and is most likely the mechanismthat controls combustion velocity.

ACKNOWLEDGMENTS

The authors would like to thank the Los Alamos Na-tional Laboratory for financial support through the AdvancedEnergetics Initiative and the Defense Threat ReductionAgency �DTRA�. James Busse and Ed Roemer are gratefullyacknowledged for their advice and assistance with these ex-periments. One of the authors �M.P.� additionally acknowl-edges the U.S. Army Research Office �W911NF-04-1-0217�.

1S. H. Fischer and M. C. Grubelich, Proceedings of the 24th InternationalPyrotechnics Seminar, Monterey, CA, 27–31 July 1998 �unpublished�.

2J. J. Giles, Nature �London� 427, 580 �2004�.3L. L. Wang, Z. A. Munir, and Y. M. Maximov, J. Mater. Sci. 28, 3693�1993�.

4C. E. Aumann, G. L. Skofronick, and J. A. Martin, J. Vac. Sci. Technol. B13, 1178 �1995�.

5R. A. Rugunanan and M. E. Brown, Combust. Sci. Technol. 95, 61�1994�.

6M. E. Brown, S. J. Taylor, and M. J. Tribelhorn, Propellants, Explos.,Pyrotech. 23, 320 �1998�.

7A. Shimizu and J. Saitou, Solid State Ionics 38, 261 �1990�.8S. Brunauer, Ph. H. Emmett, and E. Teller, J. Am. Chem. Soc. 60, 309�1938�.

9J. J. Granier and M. L. Pantoya, Combust. Flame 138, 373 �2004�.10C. G. Granquist and R. A. Buhrman, J. Appl. Phys. 47, 2200 �1976�.11R. V. Hogg and J. Ledolter, in Applied Statistics for Engineers and Physi-

cal Scientists, 2nd ed. �Macmillan, New York, 1992�.12I. Glassman, in Combustion, 3rd ed. �Academic, San Diego, California,

1996�.13B. W. Asay, S. F. Son, J. R. Busse, and D. M. Oschwald, Propellants,

Explos., Pyrotech. 29, 216 �2004�.14M. L. Pantoya, S. F. Son, W. Danen, B. Jorgensen, B. W. Asay, J. Busse,

and J. Mang, Defense Applications of Nanomaterials, ACS SymposiumSeries 891, edited by A. W. Miziolek, S. P. Karna, J. M. Mauro, and R. A.Vaia �American Chemical Society, Washington, DC, 2005�, pp. 227–240.

15S. F. Son, Burn rate measurements of HMX, TATB, DHT, DAAF, andBTATz �Los Alamos National Laboratory�; H. L. Berghout, C. A. Bolme,D. E. Chavez, D. Naud, and M. A. Hiskey, Proc. Combust. Inst. 28, 919
�2000�.

Propagation studies of metastable intermolecular composites (MIC)

Documents