Author's personal copydreyzin/Papers-pdf/PECS-09-review.pdfmolecularcompounds,such asTNT,RDX,HMX, CL-20,etc.[1,2].The maximum heatof combustion of such materials is generallylimited

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Metal-based reactive nanomaterials

Edward L. Dreizin*

Department of Chemical Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA

a r t i c l e i n f o

Article history:Received 30 June 2008Accepted 26 September 2008Available online 30 November 2008

Keywords:NanoaluminumEnergetic materialsCompositeThermitesMetal combustionMaterials synthesis

a b s t r a c t

Recent developments in materials processing and characterization resulted in the discovery of a newtype of reactive materials containing nanoscaled metal components. The well-known high oxidationenergies of metallic fuels can now be released very rapidly because of the very high reactive interfaceareas in such metal-based reactive nanomaterials. Consequently, these materials are currently beingexamined for an entire range of applications in energetic formulations inappropriate for conventional,micron-sized metal fuels having relatively low reaction rates. New application areas, such as reactivestructural materials, are also being explored. Research remains active in manufacturing and character-ization of metal-based reactive nanomaterials including elemental metal nanopowders and variousnanocomposite material systems. Because of the nanometer scale of the individual particles, or phasedomains, and because of the very high enthalpy of reaction between components of the nanocompositematerials, the final phase compositions, morphology, and thermodynamic properties of the reactivenanocomposite materials may be different from those of their micron-scaled counterparts. Ignitionmechanisms in such materials can be governed by heterogeneous reactions that are insignificant formaterials with less developed reactive interface areas. New combustion regimes are being observed thatare affected by very short ignition delays combined with very high metal combustion temperatures.Current progress in this rapidly growing research area is reviewed and some potential directions for thefuture research are discussed.

� 2008 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1422. Manufacturing of reactive nanopowders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .143

2.1. Nanosized aluminum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1432.2. Additional nanopowders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

3. Preparation of reactive nanocomposite materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1463.1. Powder mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1463.2. Sol-gel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1463.3. Self-assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1473.4. Layered vapor deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1473.5. Arrested reactive milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

4. Materials properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1494.1. Particle size distributions, surface morphology, and active metal content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1494.2. Thermodynamics of nanoscaled components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

5. Exothermic heterogeneous reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1535.1. Aluminum oxidation in gaseous oxidizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.2. Reactions between condensed components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

6. Ignition studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1577. Combustion studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .160

7.1. Laboratory tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1607.2. Performance in practical applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

* Tel.: þ1 973 596 5751; fax: þ1 973 597 5855.E-mail address: [email protected]

Contents lists available at ScienceDirect

Progress in Energy and Combustion Science

journal homepage: www.elsevier .com/locate/pecs

0360-1285/$ – see front matter � 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.pecs.2008.09.001

Progress in Energy and Combustion Science 35 (2009) 141–167


8. Concluding remarks and future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .163Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

1. Introduction

Commonly used energetic materials are based on mono-molecular compounds, such as TNT, RDX, HMX, CL-20, etc. [1,2]. Themaximum heat of combustion of such materials is generally limitedby the enthalpy of formation of their reaction products, CO2 andH2O upon complete oxidation. The monomolecular energeticmaterials enable an exothermic reaction to occur very rapidly, withthe rate controlled primarily by the chemical kinetics processes forthe molecule decomposition [3,4]. On the other hand, the energydensities of such materials are relatively low. Higher combustionenergies and thus higher energy densities can be obtained fromcombusting metal fuels, such as Mg, Al, B, Ti, and others, as illus-trated in Fig. 1. Shown are the maximum gravimetric and volu-metric reaction enthalpies for selected monomolecular energeticcompounds and for some metals. The advantages of metal fuels areclear and become more significant when volumetric reactionenthalpies are compared. The main drawback of using such fuels isassociated with relatively low rates of energy release. Micron-sizedmetal particles ignite after a fairly long delay as compared to theinitiation of monomolecular energetic compounds. Such delays areusually controlled by relatively slow heterogeneous reactionsleading up to the self-sustaining combustion of the metal particles[5–8]. Furthermore, the rates of combustion of metal particles areoften not sufficiently high to fully utilize their energetic benefits inthe applications involving explosives, propellants, and pyrotech-nics. For micron-sized particles such rates are commonly limited bythe gas phase oxygen transport to the burning particle surface.

Combinations of conventional, micron-sized metal powderswith condensed oxidizers, such as relatively unstable metal oxidesin thermite compositions or ammonium perchlorate (AP) in solidpropellants, do not result in significant acceleration of the metalignition and combustion rates compared to the metal ignition ingaseous oxidizers. The heterogeneous processes controlling themetal ignition delays are usually associated with diffusion ofoxidizer and/or fuel through the protective layers of metal oxide.Such layers always form on the surface of the metal oxidizing ata low temperature (prior to its ignition) so that the concentration ofoxidizer outside the metal particle has only a limited effect on therate of the critical diffusion processes. In addition, decomposition ofthe oxidizer typically occurs much sooner than the metal particles

ignite, so that the igniting and burning metal particles are nearlyalways surrounded by a vapor phase oxidizer. The delayed ignitionoften causes further problems, making metal combustion lessefficient. For example, an issue of critical importance for metallizedsolid propellants is the agglomeration of unignited aluminumparticles [9,10]. Such particles, initially mixed with AP and binder,melt and agglomerate before they ignite. The resulting large sizeagglomerates may never ignite or they ignite after very long delays.Because of delayed ignition, such agglomerates often cannot burnduring the limited time they fly through the propulsion chamber.Thus, a large portion of the aluminum additive remains unburned,reducing dramatically both the efficiency of the propulsion systemand obtained specific impulse.

Similar issues also explain the very limited range of applicationof conventional thermites. Initiation of a metal–metal oxide redoxreaction is quite difficult for mixed micron-sized powders andrequires extended pre-heating of a relatively large or a well heat-insulated sample. Due to the high thermal conductivity of metal–metal oxide mixture, small, poorly insulated samples lose heat veryrapidly, and for such samples the initial heterogeneous reactionnever becomes self-sustaining.

An idealized metal-oxidizer system similar to the mono-molecular energetic compound can be described: a metastable,homogeneous metal-oxidizer solution in which the components arenot chemically bonded. Thus, the reaction rate would not be limitedby heterogeneous transport processes and can be dramaticallyaccelerated. It was observed that metastable metal–gas solutionsform naturally inside combusting metal particles [11,12]. Once suchcompounds form, they indeed react very rapidly resulting in micro-explosions and disruptive particle combustion [13–16]. However, itis anticipated that a relatively strong chemical bonding occurs insuch solutions which would limit their energy density and thusrespective practical applications.

It appears that a practically optimized metal-based energeticmaterial would have the reactive components mixed on a scale asfine as possible, as long as significant chemical bonding betweencomponents is prevented. This naturally leads to the idea of usingmaterials with high specific surface area, or materials divided downto the nanoscale in order to reduce ignition delays and acceleratecombustion of metals, as proposed in Ref. [17].

This review is focused on such metal-based reactive nano-materials, which became available recently as a result of activedevelopments in the fields of materials synthesis and character-ization. While the scale of mixing of metal and oxidizer in suchnanocomposite materials remains coarser than can be achieved ina true metal–oxidizer solution, the enthalpy of formation of thebulk nanocomposite can remain close to that of the individualcompounds. In addition to the metal–oxidizer systems, materialswith metal–metal and metal–metalloid components have also beendeveloped, with the common feature that the starting componentsare always capable of a highly exothermic reaction. The products ofthis initial reaction, intermetallic alloys or metal–metalloidcompounds, often continue oxidizing if the reaction is initiated inan oxidizing gaseous environment.

Reactive nanomaterials quickly attracted a lot of interest amongpotential users and many different compositions were recentlyprepared, characterized, and evaluated. Such applications aspercussion or electric primers, explosive additives, propellant ratemodifiers, and others became feasible. Fundamentally new and

Maximum combustion enthalpy

[kJ/g]

TNTHMXRDX

CL-20BAl

MgTiZr

100 20 30 40 50 0 30 60 90 120 150

[kJ/cm3]

Fig. 1. Maximum combustion enthalpies for selected monomolecular energeticcompounds and for selected metal fuels.

E.L. Dreizin / Progress in Energy and Combustion Science 35 (2009) 141–167142


exotic applications for metal-based energetic materials, such asMEMS energy sources [18], are also being discussed in conjunctionwith development of new metal-based reactive nanomaterials.

This review will discuss the methods used for preparation ofreactive metal-based nanomaterials, techniques developed fortheir laboratory testing, and some of the published resultsdescribing mechanisms of their ignition and combustion.

2. Manufacturing of reactive nanopowders

2.1. Nanosized aluminum

Nanosized aluminum powder or nanoaluminum (n-Al) is themost common component of metal-based reactive nanomaterials,while other nanopowders, e.g., boron, magnesium, or zirconiumhave also been considered [19]. The popularity of n-Al is under-standable because it is often considered as a potential replacementfor the conventional aluminum powders and flakes widely used inexplosives, propellants, and pyrotechnics. In fact, the rapid accel-eration of research in the area of reactive nanocomposite materialscan be readily traced to the development in the n-Al manufacturingand to the time it became available for experiments more thana decade ago.

The methods of manufacturing n-Al can be broadly classifiedinto those involving vapor phase condensation and liquid phasechemistry.

One of the first methods described in the literature wasproduction of n-Al by exploding electrically heated wires, pioneeredby Russian scientists [20–23]. This method continues to be devel-oped around the world [24–26] and a large portion of publicationsdescribing production of n-Al are related to this method. Theproduced powders were branded as Elex or Alex [27,28] and areavailable commercially. Aluminum nanoparticles are also producedfrom condensing aluminum vapor generated when a thin aluminumwire is vaporized by a strong electric current passing through it.Nanoparticles of many other metals and alloys were also obtained[29–31]. Characteristic transmission electron microscope (TEM)images of aluminum nanopowders obtained by exploding electri-cally heated wires are shown in Fig. 2. It was proposed that therequired current density should exceed 1010 A/m2 [25]. It was alsoproposed that the average size of the produced particles is inverselyproportional to the cube of the energy released into the wire [18,25].In various investigations, the characteristic wire diameter variedfrom tens to hundreds of microns. Specific electric circuits designedto produce such pulses were described in many publications[32–34]. A high-voltage source (5–30 kV) is commonly used toproduce a current pulse on the order of several thousands ofamperes. The pulse duration varies from nanoseconds [32], ach-ieved using customized electronic components, to several micro-seconds [24] produced using common L-C circuits. The underlyingphysics of the wire explosion remains the subject of current inves-tigations but is outside the scope of this paper. For preparation ofnanopowders, the wire explosion is commonly set up in an inert gasenvironment. The particles are collected on the walls of the explo-sion vessel. The sizes of the obtained particles typically vary overa fairly broad range – from 10–20 nm to microns. The effects ofpressure, gas environment, electric pulse characteristics, and otherexperimental parameters were studied [20,24–27]. In Ref. [24] itwas established that a higher pressure results in the formation ofcoarser particles. On the other hand, it was reported that anincreased pressure of the inert gas results in an increased yield ofaluminum nanoparticles [26]. Aluminum nitride was formed in theexperiments performed in nitrogen [26,35]. The modifications ofthis technique include automation of the wire feed into thedischarge region, setting up a flowing gas system, and altering thegas environment to passivate the aluminum nanoparticles [28,29].

In particular, treatment of the nanopowders in environments withlow partial pressure of oxygen (e.g., 0.01% of the total pressure)results in the formation of protective oxide coatings which effec-tively stop further oxidation of nanoparticles during their handlingand storage. Passivation is commonly accomplished as a separateprocessing step, in which the chamber filled by inert gas for powderproduction is evacuated and refilled with an oxidizing gas mixture.Other approaches have also been considered in which the powderswere passivated by fluoropolymers, stearic and oleic acids, andaluminum diboride [35,36]. The advantages of preparing n-Al fromelectro-exploded wires include the method’s relative simplicity andthe efficient use of electric energy. The main disadvantages are therelatively low production rate and difficulties in obtaining theproduct powders with a useful narrow size distribution.

Other techniques for the production of n-Al, including evap-oration of bulk aluminum samples or aerosolized micron-sizedpowders followed by controlled vapor condensation, have beendiscussed in the literature [37–44] and some of them werecommercialized. Commonly, condensation in a low pressure (lessthan 10 torr or 1.3 kPa) inert gas results in the formation ofnanoparticles, while higher pressures result in increased particlesizes. Condensation of metal vapors in lighter inert gases (e.g., Hevs Ar or Xe) results in respectively finer particle sizes [37]. Mostcommonly, a crucible containing bulk aluminum is heated ina flowing inert gas environment. Radiative heaters, inductionheaters, lasers, electric arcs, or special high temperature furnaceshave been used to vaporize the bulk aluminum sample. Ofparticular interest are devices using arc or induction plasma toevaporate precursor material and to further use the plasma flowto transfer the superheated vapor to the quenching zone [45,46].The Al vapor is carried out into a passivation section where theinert gas is mixed with a small amount of oxygen. Finally, thepassivated powder is collected thermophoretically. In Ref. [44],aluminum wire was fed into a vacuum chamber and evaporatedfrom a heated ceramic boat. The condensation occurred ina helium or argon gas stream at pressures in the range of2–16 torr. Many processing details remain proprietary and werenever published. Other examples of devices producing n-Al aredescribed in Refs. [28,37].

Particle growth from condensing vapor has been the subject ofmultiple theoretical studies, some of which specifically focus on Al,e.g., Ref. [47]. It was shown that nanoparticles form in a relativelynarrow temperature range. Finer particles are produced at highercooling rates. In addition, low evaporation temperature, low pres-sure or low metal vapor concentration are desired. Particle nucle-ation and growth were considered theoretically in Ref. [48] and theeffect of different gas environments was explored. It was suggestedbased on both experiments and calculations, that carrying out thesynthesis in helium results in much finer size particles compared tousing argon. It was further confirmed that higher pressures result incoarser particles. In another study, it was found that coalescence ofcolliding nanoparticles results in a substantial reduction of thesurface energy and resulting temperature increase of the productparticle [49]. This temperature increase affects the final dimensionsof the growing nanoparticles [50]. There are now numerous theo-retical papers on nanoparticle condensation from vapors, coveringa variety of metals and operating conditions; however, they areoutside the scope of this review.

An example of using modeling for the design of nanopowderproducing equipment is presented in Ref. [39]. Particle growth bynucleation and coagulation is considered theoretically. The problemis solved numerically. Numerical solutions offer the advantage ofexploring the effects of specific, system-dependent temperatureprofiles, gas concentrations, etc. For example, the effect of differenttemperature profiles on the product particle size distribution isreported in Ref. [39].

E.L. Dreizin / Progress in Energy and Combustion Science 35 (2009) 141–167 143


Modifications of the generic approach based on physical vaporcondensation include alloying or doping aluminum with variousadditives. For example, preparation of n-Al with added barium isdescribed in Ref. [40]. Such readily oxidizing additives reportedlyincrease the reactivity of the produced aluminum and enable it toreact at lower temperatures as compared to the pure powders.Similar to the n-Al produced by wire explosion, post-processingsteps are used to passivate the surface of the nanoparticles. Bothtreatment in low pressure gaseous oxygen environment andprocessing powders in various liquid solutions (such as oleic acidand phenyltrimethoxysilane [51]) have been reported.

Variations of the bulk aluminum heating technique werereported in Refs. [52,53] in which aluminum was heated in an arcdischarge or ablated by an Nd–YAG laser. The nanoparticlesproduced were introduced into cold argon environment for labo-ratory measurements which did not require substantial amounts ofmaterial. In laboratory scale measurements the particle size distri-butions were well controlled. However, currently, scaling up then-Al production using such heating techniques remains impractical.

Coating of magnetic metallic particles with carbon was reportedfor the arcs operated in a flow of methane [54,55]. In a modificationof the latter approach, aluminum was used as a consumable anode ofthe microarc discharge operated in natural gas to produce aluminumnanoparticles coated with a thin protective layer of carbon [56].Carbon-coated aluminum nanopowders were also obtained whenaluminum nanoparticles generated by arc- or laser ablation of analuminum target were quenched in an argon–ethylene flow [57].Single particle mass spectrometry measurements reported in

Ref. [57] showed that carbon coatings remain stable at the temper-atures exceeding 900 �C and thus are attractive for protection of n-Alpowders. Unfortunately, the reported production rates for carbon-coated n-Al powders were very low and not yet suitable for themanufacture of such powders.

Flame synthesis is among the most exotic vapor condensationtechniques reported for the preparation of metallic nanopowders[58–62]. The flame environment is adjusted so that the producedmetal particles do not oxidize immediately. In Refs. [59–62], themetal particles are immediately coated by salts also produced in theflame. While this coating is protective and can be readily removedby dissolution, it is not thin and, in fact, comprises a substantialvolume portion of the bulk product. Therefore, using such nano-particles embedded in a matrix of salt is impractical for mostenergetic material applications.

Wet chemistry techniques are attractive for the commercialsynthesis of aluminum nanopowder because of the inherent safetyof handling the reactive powder under liquid and the ability toreadily functionalize the particle surface. However, we are unawareof any aluminum nanopowders produced by wet chemistry tech-niques on the commercial or on a practically interesting scale. InRef. [63] nanosized aluminum powders were prepared by decom-posing alane-adducts in organic solvents under an inert atmo-sphere. Effective adduct species were reported to include trialkylamines, tetramethylethylene-diamine, dioxane, and other aromaticamines and ethers. Reportedly, highly uniform particles wereobtained with particle size selectable in the range of about65–500 nm by adjusting the catalyst concentration and by varying

Fig. 2. Typical TEM structure of the nano aluminium particle produced in different atmospheres: (I) helium, (II) argon, and (III) nitrogen: (a) 0.025 MPa, (b) 0.05 MPa, and(c) 0.1 MPa [26]. http://www.elsevier.com/wps/find/journaldescription.cws_home/505786/description#description.



the concentration of the adduct species. As typical for all reportedwet chemistry techniques, the methodology is based on careful andslow mixing of measured amounts of the starting solutionsfollowed by continuous stirring and drying the product. Sucha methodology is not well suited for scaled up production andsubstantial modifications are necessary to obtain practical quanti-ties of the desired reactive nanopowders.

Despite these limitations of the wet chemistry approach,surface passivation remains an important safety and handlingissue and current research is focused on preparing high quality andwell-passivated powders [64–66]. For example, surface layers oftransition metals were formed on Al nanoparticles to preventthem from oxidation in surrounding air [64,67]. In Ref. [64],aluminum nanopowder was synthesized by thermal decomposi-tion of an alane solution in the presence of a titanium catalystunder an inert atmosphere. In Ref. [67], the nanopowder wasformed upon mixing and drying of aluminum dissolved in NaOHco-mixed with a nickel salt solution. In both cases, aluminumnanoparticles served as a reducing agent for the transition metalcomplexes, so that reduced metal films were produced on thealuminum surface. In a different passivation approach, aluminumnanopowders coated with non-metallic self-assembled mono-layers (SAMs) were prepared [65,66]. Nanoscale Al particles wereprepared in a solution by catalytic decomposition of the Al-methylpyrrolidine alane adduct by titanium(IV) isopropoxide; followedby in situ coating using organic SAMs (e.g., diethyl ether solutionsof perfluorotetradecanoic acid, perfluorononanoic acid or per-fluoroundecanoic acid). The obtained composite particles areshown in Fig. 3. The protective layer is clearly fairly thick resultingin the overall reduction of the energy density of such materials.

Recently, a sonochemical approach was adapted to synthesizealuminum nanopowders [68]. In this technique, a solution isprepared in which acoustic cavitation is induced. The cavitationresults in the formation of small regions heated to about 5000 K,which cool very rapidly, at the rates up to 1010 K/s. Volatilecompounds trapped in these regions decompose and aggregate toform particles with characteristic dimensions of about 10–60 nm.The early efforts aimed at production of reactive n-Al appear to bepromising because of the potential capabilities to control theproduct properties and scale the production up.

New approaches to passivating the surface of n-Al continue tobe developed and substantial progress is expected in this area inthe near future. For example, encapsulation of aluminum nano-powders in polystyrene was recently described in ref. [69] and was

shown to be effective in preventing aluminum oxidation. Anapproach similar to that proposed in Ref. [70] resulting in theformation of atomic layer deposition of refractory metal on surfaceof metallic nanoparticles may also be of interest in the future.

2.2. Additional nanopowders

Highly exothermic intermetallic, metal–metalloid, and thermitereactions are commonly exploited in metal-based reactive nano-materials. Additional nanoscale components include metals otherthan aluminum (Mg, Zr, Ti, Ni, etc.), boron (a metalloid), and metaloxides. Recently, nanosized silicon powders and nanoporous siliconwafers were considered for energetic applications [71]. Poroussilicon wafers are typically prepared by electrochemical etching ofSi wafers in different electrolytes [72]. Because of multiple appli-cations in electronics and electro-optical devices, synthesis of Sinanoparticles was addressed by many researchers in the past. Someof the related techniques are described in Refs. [73–75]. Use ofmicrowave plasma reactors for production of silicon nanoparticlesshould be mentioned in particular because of substantial practicalbenefits and potential for the scaled up production [76].

Production of most metal nanopowders is accomplished usingone or more of the same techniques described above for aluminum.Such powerful energy sources as lasers or arcs make it possible tovaporize practically any metal and thus enable vapor condensationbased techniques. Reduction of various metallic complexes insolution represents another common approach for generatingmetal nanopowders. In both cases, the cost of the nanopowders isrelatively high and the production rates are limited. Commerciallyavailable boron powders have not been marketed as nanopowders,but they typically have primary particles with sizes of the order of100 nm. Typical SEM images of a commercial boron powder areshown in Fig. 4. The fine primary particles are strongly agglomer-ated while the specific surface of the material is similar to that ofa spherical nanosized powder. The surface features of individualparticles are very fine, but the particle shapes are irregular. In orderto achieve effective mixing of such a powder with other compo-nents, advanced mixing techniques are required.

Metal oxides are often available commercially in various sizeranges and thus need not be always produced specifically forpreparation of reactive nanocomposite materials. At the same time,development of optimized and customized technologies for prep-aration of nanosized oxide powders remains an active research area.For example, detailed procedures for preparing nanosized MoO3 aredescribed in Refs. [77,78]. The process is based on sublimation of

Fig. 4. Boron (SB-95) powder by SB Boron Corp. Fine, micron-sized primary particlesare strongly agglomerated; individual particles have highly developed nanometer scalesurface features.

Fig. 3. SEM image of Al–C13F27COOH composite [65]. The perfluoroorganic carboxylatelayer, instead of the usual aluminum oxide layer, protects aluminum from oxidation inambient air. http://pubs.acs.org/journals/cmatex/index.html.



a precursor material followed by the controlled condensation ofMoO3. The condensation occurs in a stream of oxygen-rich gas toenable complete oxidation of the condensing molecules. A variant ofthis approach is described in Ref. [79] where a thin, electricallyheated metal filament is used as the source of metal vapor.

Many nanosized oxide powders were readily produced bychemical vapor deposition-based techniques. In particular, variousflame synthesis approaches [58,80,81] are particularly well suitedfor producing nanoparticles of various oxides with controlledproperties. Usually, metalorganic precursors are injected into a fuel/air mixture. The precursor materials decompose and the producedmetals oxidize in a well-controlled flame yielding a uniformnanosized oxide powder product. More recently, the formation ofnarrow-sized and specifically shaped nano-oxide particles or nano-fibers was achieved by flame synthesis performed over specificallyprepared substrates introduced in well-controlled flat flames [82].Alternately, external electric fields can be used to control theformation of oxide nanoparticles and their agglomeration [83].

Microwave plasma reactors are also widely used for preparationof nanosized oxide and composite particles [84,85]. The synthesis isbased on gas phase reactions in a non-equilibrium plasma. Inaddition to preparing a wide range of nanoparticles, microwaveplasma processing enables preparation of coated nanoparticles[86]. Particles leaving the plasma zone are electrically charged sothat their agglomeration is suppressed and high quality coatingbecomes possible. Recently, multiple plasma-based processes forsynthesis of nanopowders were reviewed in Ref. [87].

Metal oxide nanoparticles customized for reactive nano-composite materials were developed in Ref. [88]. Strong oxidizernanoparticles (potassium permanganate) were coated with a layerof a relatively mild oxidizer (iron oxide). The composite oxidizernanoparticles were synthesized by a new aerosol approach in whichthe nonwetting interaction between iron oxide and molten potas-sium permanganate aids the phase segregation of a nanocompositedroplet into a core–shell structure. The iron oxide coating thicknesswas varied to tune the reactivity of the product. Such core–shellengineered nanoparticles are promising for adjustment of thereaction rates in the nanocomposite energetic materials.

Nanoscaled oxides are also produced for applications in ener-getic materials using the sol-gel technique [89–91], which can alsoproduce nanocomposite materials. This method is discussed inmore detail in the following section.

A wet chemistry approach is generally useful for preparation ofnanosized oxide particles suitable for energetic formulations. Forexample, a process for the preparation of nanosized tungsten oxideparticles is described in Ref. [92]. In that process, ammoniumparatungstate was dissolved in acid and the product, tungstic acid,was precipitated by addition of distilled water. The product con-sisted of 7-nm thick platelets of hydrated WO3 with the plate lengthof about 100 nm. Such particles are useful for preparation ofnanocomposite thermites as discussed below.

3. Preparation of reactive nanocomposite materials

3.1. Powder mixing

Given the availability of n-Al powders, most metal-based reac-tive nanomaterials are prepared by mixing such powders withadditional components capable of highly exothermic reactions withaluminum. Literature data are available for many mixes of n-Al withnanosized metal oxide powders; including Fe2O3, MoO3, CuO,Bi2O3, and others. These thermite compositions made of nanosizedcomponents were termed MIC (for metastable intermolecularcomposites or metastable interstitial composites) or super-thermites [92–97]. A superthermite is commonly prepared byultrasonic mixing the nanosized components in a bath of hexane

[98–101], isopropyl alcohol [100,101], or another liquid carrier,which is subsequently evaporated. The mixing is typically per-formed using high intensity ultrasonic actuators or commerciallyavailable ultrasonic cell disruptors. For liquids that can oxidizealuminum, such as isopropanol, the exposure time to the solvent isrestricted. Assessment of the quality of mechanical mixing ofnanopowders represents a major challenge. Such an assessment ishighly desirable as a quality control measure for the preparednanocomposite mixtures. A recent review [102] showed that mostcurrent techniques are based on imaging of a relatively smallfraction of the prepared mixed sample, so that a large number ofmeasurements are required for a reliable assessment. Imagespresented in Fig. 5 show the variety of morphologies for nano-composite MIC materials obtained by powder mixing [100]. Asclearly visible in Fig. 5, the issue of mixing quality becomesespecially critical when the morphologies of nanopowders beingmixed are very different, for example for the Al–MoO3 case shownin Fig. 5a. Often, nano-oxide particles are not rounded or equiaxialbut instead are shaped as rods or flakes. Also, aluminum flakes withnanometer thickness are often used and mixed with different sizeor shape oxide nanoparticles. A representative imaging of thesample is especially challenging because particles with largefootprint (e.g., flakes) can shield smaller particles of the othercomponents.

While relatively successful in laboratory evaluations, ultrasonicmixing of nanopowders is a process that is difficult to scale up.Processing of larger sample batches inevitably leads to a lowerquality of mixing.

Recently, thermite nanocomposites were produced by mixingthe starting Al and Fe2O3 nanopowders using rapid expansion ofa supercritical dispersion (RESD) [103]. A better degree of mixingwas achieved compared to conventional ultrasonic mixing. Inaddition to providing a higher quality of mixing, the RESD process isbetter suited for continuous operation than ultrasonic mixing.The main concern in developing and scaling up RESD for mixingreactive nanopowders appears to be the safety of operation. Inparticular, the reactions between the supercritical, or expandingCO2, and nanoaluminum need to be prevented. In addition, themixed nanocomposite is generated by RESD in an aerosolized formand thus there is the possibility of an aerosol explosion.

3.2. Sol-gel

An alternative to mechanical mixing of nanopowders wasproposed in a series of papers [89–91,99,104] describing sol-gelprocessing in which a nanocomposite structure is obtained withaluminum (or other metal) nanoparticles residing in the pores ofa matrix made of the oxidizer material. Hydrated salts of metalsserve as precursors and propylene oxide serves as a gelation agent.The process is carried out preferably in a polar protic solvent andmonolithic gels are produced. In order to prepare reactive nano-composites, metal powders are added just before the gelation whilethe solution is being stirred. The final step of removing the pore fluidfrom the system is accomplished either by controlled slow evapo-ration or by supercritical extraction with CO2. Respectively, xerogelsor aerogels are produced. To produce aerogel, prior to supercriticalextraction, the liquid in the pores is replaced by CO2 by a series offlush and drain cycles. Further functionalization of the oxide matrixis achieved using various silane additives [105]. This approach isattractive as it naturally generates a very intimate mixing betweenthe components. The disadvantages include: high porosity of thefinal composite makes it undesirable for some applications,restrictions on the types of materials that can be gelled, and a diffi-culty in scaling up the production rate – chiefly because of the needto introduce the metallic nanopowders in the stirred solution ata very specific stage of the gel preparation process.



3.3. Self-assembly

Recently, self-assembly approaches were considered to preparereactive nanocomposite materials starting with nanosizedaluminum powder and functionalized nanosized oxide particles[106–109]. To produce ordered assemblies, metal particles werearranged around the exterior surface area of oxide nanorods or inthe ordered pore structure of the mesoporous oxidizer particles incomposites. For example, the self-assembly in an Al–CuO systemwas achieved by initial functionalization of the CuO nanorods byapplying a monofunctional polymer, poly(4)-vinyl pyridine (P4VP).The nanorods were prepared for these experiments using thesurfactant-templating method. Al nanoparticles adhere to func-tionalized nanorods and these ‘‘decorated’’ nanorods becomeordered within the material [107,108]. Such ordered structures arereported to produce higher flame speeds in small-scale laboratoryevaluation tests compared to nanocomposite materials with thesame compositions conventionally mixed using ultrasonicatedsuspension. A conceptually similar approach was used to prepareself-assembled Al–Fe2O3 nanocomposite thermites [109]. A porousFe2O3 was synthesized using a micelles template-assisted sol-gelsynthesis route using surfactants. The addition of surfactantsduring sol-gel synthesis generates an ordered pore structure. Areference sample of porous Fe2O3 was also prepared by the sameprocess but without the use of surfactants, so that the pore struc-ture was not ordered. Both oxidizers were mixed with n-Al and thecombustion rates were compared to each other. The orderedstructure produced a higher flame speed. The ordered nano-composites are attractive as offering a better control over thematerial properties and potentially higher reaction rates in prac-tical applications. The shortcomings of this approach include thehigh costs of custom-made oxides, the presence of functionalizingagents which generally reduce the energy density of the energeticformulation, and the inherently high porosity of the producedmaterials.

3.4. Layered vapor deposition

A distinct class of nanocomposite reactive materials is repre-sented by reactive nanofoils. Well-controlled, nanosized layers ofmaterials capable of highly exothermic reaction are coated on top of

each other using vacuum deposition. One of the first descriptions ofmultilayered Al–Ni films with the individual layer thickness variedfrom 60 to 300 nm is given in Ref. [110]. The films were produced ata pressure of 10�7 torr by alternate electron beam evaporation of Aland Ni onto a photoresist coated glass slide. The photoresist wasdissolved to obtain a free-standing film. The films were reported toreact rapidly and sustain Self-propagating High temperatureSynthesis (SHS) reaction producing mixed phases of Al3Ni, Al3Ni2,and Al with a flame speed of about 4 m/s. This approach wassystematically developed in a patent by Barbee and Weihs [111],followed by extensive research [112–117]. Developed originally forjoining applications [112–114], reactive nanofoils attract more andmore interest as energetic components. Most work has beenreported for Ni–Al nanofoils, for which many reaction details werestudied, including the effect of partial annealing resulting in anincreased thickness of the partially reacted layer between bilayersof Al and Ni [115]. Multilayer Al–Ni foils were prepared usingmagnetron sputtering, by rotating a water-cooled brass substrateover fixed Al and Ni guns. The number of produced bilayersexceeded 4000. The thickness of individual bilayers varied from 25to 80 nm. Substantial efforts were also made to prepare and char-acterize nanofoils of Al–CuO thermite [116,117] CuO–Al multilayerfoils were similarly prepared by magnetron sputtering performedin argon at 5 mTorr. To avoid the reaction between CuO and Al, thesputter guns were shielded to contain the plasma in a small volumeabove each target. The substrate carousel was also water-cooled tominimize mixing and reacting of the layers during deposition. Thesputter-deposited CuOx had the structure of the mineral para-melaconite, Cu4O3. Each thermite bilayer was 1 mm and the total foilthickness was 14 mm. Electron microscopy images of the preparedthermite nanofoils are shown in Fig. 6.

3.5. Arrested reactive milling

Current literature describes only one ‘‘top-down’’ approach forpreparing reactive nanocomposite materials where the nanoscaledstructure is obtained by refining coarser starting materials. Thenanocomposites are produced using a technique similar tomechanical alloying, called Arrested Reactive Milling (ARM)[118–124]. The starting materials are mixtures of regular metal,metalloid and/or oxide powders. The sizes of starting powders are

Fig. 5. Scanning electron microscopy images of different thermite compositions prepared by mechanical mixing of starting nano-sized components in a liquid solvent: a) Al–MoO3;b) Al–Bi2O3; c) Al–WO3; and d) Al–CuO [100]. http://www.aiaa.org/content.cfm?pageid=322&lupubid¼24.



not critical and using very fine or nanosized powders as startingmaterials is, in fact, undesirable. In order to produce reactivenanocomposites, starting components are selected among mate-rials capable of reacting exothermically. Metals and metal oxides(thermites) represent one popular class of related compositions.Boron and metals such as titanium, zirconium, or hafnium formingrespective borides represent another class of useful compositions.When powders of such materials are mixed and ball milled, theexothermic reaction can be initiated mechanically. Once initiated,the reaction becomes self-sustaining. The reaction usually proceedsvery rapidly resulting in substantial increases in both the pressureand temperature within the milling vessel. Reactive nano-composites are produced when the milling process is interrupted(or arrested, hence ARM) just before the self-sustaining reaction ismechanically triggered. For small-scale samples, preliminaryexperiments are used to establish the time when the self-sustainingreaction is triggered. For larger scale samples, the milling conditionsand the time of milling are predicted using numerical modeling ofenergy transfer between the milling media and the powder [125].

ARM leads to the formation of fully dense, micron-sizedcomposite particles with nanoscaled structural features. Eachparticle is a three-dimensional composite of starting materials asopposed to homogenized or chemically bonded compounds thereof.The milling time at which the reaction is mechanically triggeredeffectively sets a limit to the spatial scale on which the componentsare mixed. This time limit can be influenced by the specific millingparameters chosen – the powder batch size, the mass ratio ofpowder sample to that of the milling media, the processingtemperature, and the use of process control agents. Collisionsbetween the milling media subject the milled powder to transientpressures of up to 5 GPa [126], individual particles have thereforenear theoretical maximum density (TMD).

ARM processing is very flexible and versatile. It does not have themany limitations of chemical or vacuum condensation techniques,

which can be used only with selected compositions. It has beenobserved that essentially any combination of reactive materials canbe processed by ARM to prepare a nanocomposite reactive powder.Table 1 lists all compositions prepared by ARM to date includingmultiple thermites and metal–metalloid systems. The process isreadily scalable and inexpensive. One of the important limitationsof the ARM processing is the inevitable presence of a fraction ofreacted material in the prepared nanocomposite powder. Suchreacted material forms in relatively small quantities during theprocessing when a reaction between the components is locallytriggered mechanically but is not self-sustained within the sample.As a result, the small quantities of the reaction products are redis-tributed and homogenized within the sample that is continued tobe ball milled. The presence of the partial reaction products can beminimized by adjusting the milling conditions and parameters[123,124]. Another important issue that needs to be consideredwhen preparing reactive nanocomposite materials by ARM is thesafety of the ball mill operation, specifically, the self-sustainingreaction between the reactive components must be prevented toavoid the damage to the processing equipment and facility.

A characteristic cross-section of a particle of an Al–CuO nano-composite prepared by ARM is shown in Fig. 7. The inclusions ofCuO in Al matrix vary in size from 1 mm down to less than 50 nm. Itis important to note that inclusions of one component (e.g., CuOfor the example shown in Fig. 7) are fully embedded into thematrix of the other component (e.g., Al). Therefore, the entire Al/CuO interface area will participate in the exothermic heteroge-neous reaction upon thermal (or other) initiation of such a mate-rial. This would not be the case for a material prepared as a mixtureof nanopowders in which the reactive particles only have directcontact over a relatively small portion of the total surface area ofeach nanosized component. Note, however, that the propagationmechanisms for the reactions in fully dense nanocompositesprepared by ARM and in the highly porous mixtures of nano-powders with identical bulk chemical compositions may beentirely different. The pores may play a critical role in promotingthe pressure driven reaction propagation, while a slower, thermalreaction propagation mechanism is expected for the fully densenanocomposites.

Reactive milling has also been investigated in Russia to produceenergetic compositions refined on the nanoscale termed Mechan-ically Activated Energy Composites (MAEC) [127,128,129]. Most ofthe effort focused on metal-Teflon compounds [127,128]. Al, Mg, Ti,

Table 1Reactive nanomaterials prepared at New Jersey Institute of Technology by ARM todate.

Nanocomposite thermites

Fuel Oxidizer

Fe2O3 MoO3 CuO Bi2O3 WO3 SrO2 NaNO3

Al x xa xa x x x xb

Mg x x xAl0.5Mg0.5 xMgH2 x xSi x x xZr x x x x2Bþ Tic xb

2Bþ Zrc xb

Reactive metal–metalloid compositesB Reactive metals: Ti, Zr, Hf

Nanostructured Al-based alloysAl Alloying components: W, Hf, Mg, MgH2, Ti, Li, Zr, C

a Metal-rich nanocomposites also have been synthesized.b Metal-lean nanocomposites also have been synthesized.c Nanocomposite powder used as component for compound nanocomposite.

Fig. 6. Elemental distribution images of as-deposited foil. The substrate side of the foilis the left side of the Al layer. The right side of the Al layer is the top side. (a) Brightfield image. (b) Cu map (L2,3 edge, 931 eV) (c) O map (K edge, 535 eV). (d) Al map(K edge, 1560 eV) [117]. http://jap.aip.org/.



and Zr powders were used while major experiments focused onAl- and Mg-based composites [129]. Similar to the ARM approach,the milling conditions in a vibratory mill are selected to avoid orminimize the reaction between the components while achievingmaximum degree of homogenization between the startingmaterials.

4. Materials properties

Knowledge of material properties is essential for understandingtheir reaction mechanisms, for prediction of their performance inenergetic formulations, and for designing practical systems andcomponents employing such materials. Some of the materialproperties commonly used for micron-sized reactive powders andstructures will need to be redefined for materials with nanoscalefeatures. For example, many reaction features for conventionalmetal fuels are affected by such thermodynamic properties asmelting and boiling point of the metals and their respective oxides.However, for material domains of nanometer dimension, thestructure and phase stability are strongly affected by the surfaceenergy. As a result, melting can occur at different temperatures fordomains (particles or crystallites) of different sizes and may not beobserved at all if the domains are sufficiently small, e.g., see Refs.[130–133]. The enthalpy of formation of respective nanodomainsand their combustion enthalpies will also be altered compared tothe coarser components with the same elemental compositions.Effects of surface energy on the boiling temperature and on theequilibrium vapor pressure as a function of temperature can also besignificant [134]. Similarly to thermodynamic properties, mechan-ical properties of nanosized particles or surface layers coating suchparticles cannot be described using conventional characteristics forbulk materials [135]. In general, understanding and describingproperties of the nanomaterials is an important and exciting area ofthe current research, which has applications well beyond thematerials systems discussed in this article.

4.1. Particle size distributions, surface morphology, and activemetal content

The sizes of the nanodomains and the nanoparticles, are ofprimary importance in determining the rates of heterogeneousreactions commonly leading to ignition of such materials. Similarly,the particle size distributions and related specific surface values of

nanocomposite materials affect their aging characteristics, i.e.,changes in the oxide layer thickness, partial reaction betweencomponents, etc. Particle sizes need to be known to designappropriate material handling and mixing techniques. Because ofthe polydisperse nature of nanopowders and nanodomainsobserved in reactive nanocomposites, the quantitative descriptionsof the respective particle size distributions are difficult to obtain.Agglomeration that can be very significant for nanopowders makesquantitative description of the particle distributions even moredifficult. It should be noted that the combustion performance maybe quite different for the powders with the same specified primediameters or even with similar size distributions for the primaryparticles if one of the powders is much more agglomerated thananother. The agglomeration of nanopowders is unavoidable [136],while its detailed mechanisms and their correlation with otherpowder properties are poorly understood.

The problem of quantifying the particle size distribution for thenanocomposite materials becomes even more challenging consid-ering that at least two nanoscaled phases are present. For thesimplest case of aluminum nanopowder, the second phase is thepassivating oxide (or any other) surface layer. For the more complexsuperthermites, intermetallic, or metal–metalloid composites, addi-tional phases include various oxides, metals, alloys, and metalloids.

Various particle shapes introduce yet another layer ofcomplexity in the quantitative particle size distribution measure-ments. Most particle sizing techniques implicitly assume that theparticles are spherical in shape. Depending on the preparationtechnique, this assumption may or may not hold true for differentaluminum nanopowders. The assumption of the spherical particleshape is most likely to be incorrect for oxide and metalloid nano-particles. An approach for describing particle dimensions andshapes for non-spherical and highly agglomerated particles basedon analysis of the powder fractal dimensions has been developedand used extensively for many agglomerated powders and aerosols[137]. Most commonly, textural and density (or structural) fractaldimensions of the aerosol particles are considered [138,139]. Thisapproach may prove to be useful for characterization of many non-spherical and agglomerated reactive nanopowders.

Substantial efforts were made to characterize aluminum nano-powders which serve as the most popular component of reactivenanomaterials. With advances in electron microscopy, imaging ofthe nanoparticle samples became a common practice. RespectiveSEM or TEM images can be readily used for straightforward,although labor-intensive, size classification of the observed parti-cles, e.g., Ref. [56]. In order to be representative, such measurementsmust consider a large number of particles. It is also desired that theparticle sizing be performed in different locations of the sampleprepared for the electron microscopy. In addition to the directmeasurement of the particle size distribution, electron microscopyprovides very important information about the particle shapes, thethickness of the oxide layer (high resolution TEM), and about theparticle surface morphology. This information is important forinterpreting the results of various other particle sizing techniquesbased on light scattering, surface areas by gas absorption, etc.

The most common characterization of the particle size isobtained from the specific surface measurements. Usually, a BET(Brunauer–Emmett–Teller) measurement of gas adsorption isemployed, e.g., Refs. [51,140,141] to obtain a single number char-acterizing the dimension of the specific nanopowder. The ‘‘BETdiameter’’ is often introduced by assuming that the measuredspecific surface of a powder is equivalent to that produced bymonodisperse spherical particles. For many applications thisrepresents the single most important characteristic of a nano-material. A similar, ‘‘one number’’ assessment of the particlesize can be obtained for nanopowders and for fully dense nano-composite materials using broadening of the X-ray diffraction line

Fig. 7. Cross-section of a nanocomposite reactive particle with bulk composition8Alþ 3CuO (metal rich thermite composition). The image is taken with backscatteredelectrons and shows phase contrast between CuO (light inclusions) and Al (darkmatrix).



and the well-known Scherrer formula [142]. Modifications of thisapproach have also been discussed in the literature to obtaina more accurate assessment of the nanodomain dimension [143].However, for many energetic materials, the width and shape of theparticle size distribution are as important as any specific weightedaverage diameter, so that additional measurements are needed toobtain reliable particle size distributions.

Few commercial devices are available for sizing nanopowders.Submicron particle sizes can be quantified using low-angle laserlight scattering and several commercial instruments utilizing thistechnique are available. These instruments are generally designedfor size characterization of micron-sized powders and their rangeof measurements is extended to include particles as small as 40 nmwith specific algorithms for processing the measured scatteredlaser emission. The main advantage of this type of measurement isthat a powder with relatively broad particle size distribution can becharacterized. However, for accurate measurements the opticalproperties of the material surfaces need to be known. Such prop-erties are not well established for many materials, especially, for thenanosized particles, for which, as noted above, the material prop-erties are expected to substantially differ from those of therespective bulk materials. Another group of devices uses photoncorrelation spectroscopy, which determines particle size bymeasuring the rate of fluctuations in laser light intensity scatteredby particles as they diffuse through a fluid. The measurement needsto be processed assuming a specific shape of the size distributionfunction, so that generally the mean particle size and the width ofthe particle size distribution can be quantified. For successfulmeasurement, the powder needs to have a relatively narrow sizedistribution and the measurements become less and less useful asthe width of the particle size distribution increases.

Currently, new approaches are being actively developed forcharacterization of the particle sizes of reactive nanomaterials. Inone related approach, the particle size distributions are determinedusing thermo-gravimetric analysis (TGA) [144]. TGA measurementenables identification of the active metal content, e.g., the amountof un-oxidized aluminum for the Al nanopowder. In addition, thepresence of volatile impurities can be detected and particle sizedistributions can be obtained. Interpreting the TGA curve, byconsidering the effect of particle oxidation of a pre-selected particlesize, produces a measure of the primary particle size in the powdertested, similar to the gas absorption measurement. Comparison ofTGA, BET, and SEM measurements of particle sizes, with the XRDanalysis yielding a crystallite size, is presented in Ref. [144] for 12different nanoaluminum samples. The results generally comparewell to one another. However, for samples with the coarser parti-cles implied by the TGA and SEM measurements, XRD did not showan appreciable increase in the crystallite sizes. In order to obtainparticle size distributions, a curve-fit procedure is used to representthe experimental TGA curve as a superposition of ‘‘basic curves’’that ideally represent the mono-modal powder fractions. The basiccurves are selected based on preliminary information about thespecific nanopowder, e.g., obtained from SEM images. The mainassumptions made were the shape of the particles (spherical), thethickness of the initial oxide layer, and the oxide density. Note thatdifferent alumina polymorphs have substantially different densities[145], so using an incorrect density for the starting ‘‘natural’’alumina layer can cause a large error in the final particle sizedistribution. The TGA-based method was shown to be well suitedfor monitoring samples for the presence of larger, 0.5–5 mmdiameter particles but ineffective for quantitative size distributionsof powders containing particles smaller than 100 nm [144]. Themain problem identified in Ref. [144] was the inconsistency inoxidation of various nanopowders at lower temperatures. Thisinconsistency was attributed to possible impurities in some of thenanopowders. In addition, particle coalescence during or after

melting can occur resulting in the change of the particle sizes andrespective differences in the oxidation behavior. Particles withdimensions greater than 10 mm impose another limitation becausesuch particles may remain incompletely oxidized even at thehighest temperatures (typically, 1500 �C) achieved by TGA.

Another particle size measuring approach, suitable for manynanopowders or nanocomposite materials, involves small-angleX-ray and/or neutron scattering (SAXS, SANS) [146,147,148]. In Refs.[147,148], the results of SAXS and SANS measurements were shownto compare well to the results of electron microscopy, TGA, BET, andX-ray diffraction measurements. An example of such a comparisonfor SEM, SANS and SAXS results is shown in Fig. 8 [147]. A detaileddiscussion of the underlying theoretical approach, as presented inRef. [147], shows that specific information about the particle shapeand morphology is needed in order to meaningfully process thescattering measurements. This information can be gained fromdetailed electron microscopy studies complementing the X-ray andneutron scattering measurements. Because X-rays and neutronsinteract with matter differently, it is possible not only to quantifythe particle size distributions, but also characterize the features ofthe composite particles. For example, the thickness of the oxidelayers was evaluated with good accuracy based on the processedscattered intensity data [147]. In order to transform the scatteringresults into a size distribution, a specific model for the size distri-bution function must be assumed. In many cases, as in Refs.[146,148], the assumption of the lognormal distribution is welljustified.

One of the more unconventional approaches used to size classifyreactive nanopowders relied on a differential mobility analyzer [65]well suited for measuring particle sizes for very fine airborneparticles. Similar to the discussed above TGA technique, theassumptions about the thickness and density of the initial oxidelayer (or impurities) affect the output substantially. Also well suitedfor the nanosized airborne particles is a technique described inRef. [53] and using quantitative laser-induced breakdown spec-troscopy (LIBS). Qualitative LIBS was shown to be effective in

Fig. 8. Comparison of the particle size distributions determined by SEM, SAXS andSANS analysis of a commercially available nano-aluminum powder produced byTechnanogy (sample ID: TN-Al-40) [147]. http://www.mrs.org/s_mrs/sec_jmr.asp?CID¼30&DID¼48&SID¼1.



finding the degree of oxidation of n-Al particles processed atdifferent temperatures.

As mentioned earlier, generic SEM and TEM imaging techniquesare invaluable for characterization of particle sizes, shapes, andsurface morphology. High resolution TEM is successfully used tocharacterize the oxide coatings present on aluminum particlesexposed to oxidizing environments. The information about thick-ness, crystallinity, microstructure, and homogeneity of the oxidelayer is of critical importance for understanding the mechanisms ofoxidation of metal nanoparticles. In turn, the mechanisms ofoxidation affect both ignition and aging kinetics of many nano-composite reactive materials. Most of the reports agree that thethickness of the oxide layer on the surface of aluminum particles isessentially independent of the particle size and is generally in the2.5–3 nm range. Excellent high resolution TEM images of the oxidefilm are presented in many published papers, including Refs.[36,149,150] with some of the images from Ref. [150] reproduced inFig. 9.

It is currently well established that the natural oxide layer onaluminum particles is amorphous. However, the images presentedin Fig. 9 also show that the oxide coating is not homogeneous andincludes precursors of the growing g-Al2O3 crystallite sheets. Asshown in Fig. 9, these mismatching crystalline sheets can beexfoliated from the main amorphous film.

Characterization of composite materials is more difficultcompared to elemental metal nanoparticles and much less prog-ress has been made. For the materials prepared using startingnanosized powders, e.g., by mechanical mixing or sol-gel synthesis,the measure of the reactive surface is commonly obtained from thestarting particle sizes. For the fully dense, nanocompositematerials, scanning electron microscopy of the cross-sectionedmaterials and XRD measurements of the crystallite sizes producethe data about the size distributions of the nanodomains, e.g.,Ref. [123].

Finding the active (or metallic) Al content in powders is closelyconnected to the correct identification of the particle size, asalready mentioned above. High resolution TEM images showing thethickness of the alumina layer combined with the data on theparticle size distribution can be used to estimate either the volu-metric or gravimetric fraction of pure Al in the powder. The activealuminum content can also be determined from TGA results bymeasuring the degree of powder oxidation upon heating. The aboveapproaches all rely on assumptions about the density, structure,

and homogeneity of the initial oxide layer. Yet, it has been reportedthat such layers can be porous, partially hydrated, somewhatnonuniform in thickness, or can contain some adsorbed gases[150,151], all of which would affect the accuracy of the calculatedavailable active metal content. The situation is complicated furtherfor materials that are processed to produce specific passivatinglayers. Techniques that enable finding the active aluminum contentfor such cases include quantitative measurements of the gasevolved upon base hydrolysis [144], chemical analysis usinginduced coupled plasma (ICP) emission to determine the Al/Al2O3

ratio [149], or various wet chemistry techniques, such as perman-ganate [152] or cerimetric (cerium titration) [153] methods. Anassessment of the active aluminum content can also be made usingoxygen bomb calorimetry [35].

4.2. Thermodynamics of nanoscaled components

At the spatial scale of nanoparticles and nanodomains, ther-modynamic properties of materials are altered. This includeschanges in the melting point and latent heat of melting, which areof particular significance for reactive nanomaterials. In addition,phase transformations, such as polymorphic transitions betweendifferent aluminum oxide phases, are also affected by the reduceddimensions of the oxide films. Properties of other oxides or metalsin other nanomaterials are affected similarly, but discussed muchless in the literature.

The changes in the melting point for materials reduced tonanosized particles or inclusions are extensively discussed inliterature, e.g., see a recent review [154]. This review of recentresults showed that the Gibbs–Thomson model widely used toexplain the depression of the melting point, Tm, is deficient at thenanoscale and further work needs to be done to fully capture sizeeffects on melting behavior. In this review, the issue of depressedmelting point is discussed in the context of its particular impor-tance for reactive nanoparticles because melting is expected tosubstantially affect both ignition and combustion behaviors.Melting and oxidation of aluminum nanoparticles were recentlysystematically addressed in Refs. [146,149,155] using differentialscanning calorimetry. Similar methodologies for data processingwere used in Refs. [146,149] coming from different researchgroups obtaining similar results. The main advantage of theapproach proposed in Ref. [146] is that the actual particle sizedistribution for each specific powder sample was obtained and

Fig. 9. High resolution electron micrograph of an aluminum nanoparticle showing layered structure in oxide/hydroxide surface layer and the aluminum/oxide boundary. The oxidefilm is generally amorphous with inclusions of discontinuous crystalline layers (image in the center). Exfoliation of single molecular thickness laminar sheets from an oxide coatingis illustrated in the image on the right [150]. http://www.marinkas.com/.



used instead of the more common weighted average particle sizeor bulk value of the powder. The discussion below follows closelythat of Ref. [146]. The experimental particle size distributionswere obtained using SAXS [147,148]. Two lognormal distributionswere used to fit the experimental SAXS data for each analyzednanopowder. While the shape of the size distributions wasassumed, the mean modal diameters and widths of bothlognormal distributions were varied to obtain the best fit with theSAXS measurement. The experimental particle size distributionfunctions were used to predict the differential scanning calorim-etry (DSC) melting curves for different powder samples accordingto different models proposed for the melting depression asa function of particle size. These melting point models will bediscussed briefly below.

A model describing the melting point depression and referencedin several recent papers dealing with aluminum nanopowders usedin combustion systems, e.g., Refs. [156–158], was developeddecades ago by Reiss and Wilson [159]. The model describes themelting point, Tm, as a function of the particle diameter, Dp, and theoxide film thickness hox, as:

Tm ¼ Tb

�1� 4ssl

HbrAl�Dp � 2hox

��

(1)

where Tb is the melting temperature of bulk aluminum (933.47 K or660.32 �C), Hb is the enthalpy of fusion of bulk aluminum, and ssl isthe interfacial surface tension between the solid and the liquid. Thedifference in the molar volumes between solid and liquidaluminum was neglected.

More recently, a theoretical model of melting for nanocrystal-line metal powder was developed by Jiang et al. [159–162]. Themelting point was suggested to depend on the diameter ofaluminum nano-crystals, D, often assumed to be equal to thediameter of the aluminum core for the oxide-coated aluminumnanoparticles, as:

Tm ¼ Tbexp�� 2Hb

3RTb

1ðD=6lÞ � 1

�(2)

where R is the universal gas constant and l is the length of the Al–Alatomic bond. It was further suggested that for nanoparticles, thelatent heat of melting, Hm, depends on the particle diameter as:

Hm ¼ Hbexp�� 2Hb

3RTb

1ðD=6lÞ � 1

� �1� 1ðD=6lÞ � 1

�(3)

Processing the experimental results reported by Eckert et al.[163] for composite materials containing nanosized aluminuminclusions produces phenomenological dependencies for both themelting point (K) and latent heat of melting (kJ/mol) as functions ofthe aluminum core diameter (nm):

Tm ¼ 977:4� 1920D

(4)

Hm ¼ 14:705� 177:49D

(5)

The experimental data presented in Ref. [163] limit the appli-cability of Eqs. (4) and (5) to particles with diameters in the range of12 nm<D< 43 nm. For larger particles, Tm¼ Tb and Hm¼Hb. Forparticles with the metal core smaller than 12 nm, the latent heat ofmelting is reported to be negligible [163]. Free aluminum nano-particles with sizes less than 20 nm are not currently available, butthe approach described in Ref. [155] enables direct experimentalstudy of melting in such particles obtained by controlled oxidationof coarser starting powders.

The three different melting models introduced above, were usedwith the experimental PSD to quantitatively predict the DSC signalsexpected for melting of different commercial nanopowder samples.Each calculation was compared to a specific DSC run. Additionalcorrections for possible sample aging were made as described indetail in Ref. [146]. Eqs. (1), (2), and (4) give the functionaldependence of the melting temperature on the particle diameter.These equations were converted into equations that give thefunctional dependence of the aluminum core diameter on themelting temperature, D ¼ DðTmÞ. The converted equations, andtheir temperature derivatives dD=dT , were used to predict the DSCsignal directly. The DSC signal, _Q , predicted for each specifictemperature, T, and each specific heating rate, b, was calculated as:

_Q ¼ MsPðDÞp6

D3HmðDÞdDdT

b (6)

where P(D) is the normalized frequency function, of aluminumnanoparticle distributions obtained from SAXS measurements andMs is the normalization parameter for P(D) accounting for thealuminum metal mass in the analyzed sample.

Comparisons of the experimental DSC melting curves with thepredicted DSC curves from these models, are presented in Fig. 10.There is no close agreement between the experimental DSC curveand any of melting model predictions; however, the overall shapeof the melting endotherm was predicted by all the models. Inparticular, it is interesting that the experimental endotherms haveat least two peaks and this overall shape is generally predictedconsidering the bimodal size distributions obtained from SAXS.According to Eqs. (4) and (5) [163], only a fraction of the powder ispredicted to melt below Tb¼ 660.32 �C. The rest of the powder, asnoted in Fig. 10, is expected to melt at a constant temperature of Tb,so that respective calculations for the melting endotherms couldnot be performed and are not shown in Fig. 10.

It was found in Ref. [146] that the shapes of the predicted curvesare very sensitive to the specific type of the particle size distribu-tion, e.g., bimodal vs single mode lognormal distribution. Thus, oneneeds to be especially careful interpreting the DSC results withoutthe information on the specific shape of the particles size

44 nm

80 nm

121 nm

Experiment

Calculations

Ref. [163]: 73.3% of Al volume melts

at bulk Tm


at bulk Tm

-1.0

-0.5

0.0

-1.0

-0.5

0.0

DS

C S

ig

nal (H

eat F

lo

w). m

W/m

g


at bulk Tm

Reiss and Wilson, [159]Jiang et al. [160-162]Eckert et al. [163]

500 550 650 700 750

Temperature,°C600

-1.0

-0.5

0.0

Fig. 10. Experimental and computed DSC curves showing melting of aluminumnanopowders [146]. Each plot is labeled with nominal particle size. http://acsinfo.acs.org/journals/jpcbfk/index.html.



distribution function. The melting curves can further depend on thespecific surface morphologies of the used nanoparticles that couldaffect the sizes of melting nanodomains. Independent experiments[146,149,155] confirm that the melting of aluminum nanopowdersstarts at a lower temperature than the melting point of bulkaluminum; however, existing models describe the effect of the particlesize on the melting point depression only qualitatively. Detailedanalysis of the particle size distributions and surface morphology isneeded for quantitative verification of any related models.

The thermodynamic parameters of oxides or other nanoscalematerial components used in reactive nanocomposite materialshave not been discussed in the literature dealing with reactivematerials and their applications. At the same time, reducing thecrystallite or particle size to the nanoscale certainly affects thethermodynamic properties and the stability of oxides, hydroxides,and other related compounds, e.g., see reviews [164,165]. The lackof attention to such effects in the community dealing with reactiveand energetic materials is most likely due to fact that the reactionenthalpies in the systems of interest (e.g., thermites) are muchgreater than the anticipated effects of fine particle sizes on theenthalpy of formation or surface energy. On the other hand, oftenthe reaction rates in nanocomposite materials are limited by thediffusion of fuel and/or oxidizer through the growing or decom-posing oxide layers. Thus, the properties of such layers determinethe diffusion and respective reaction rates. A relevant example isthe recently developed aluminum ignition model in which the rateof aluminum oxidation is calculated as a function of oxide layerthickness and, most significantly, the polymorphic modifications inthe aluminum oxide layer present on the particle surface [6,7].Polymorphic transformations occur in alumina as a function of bothtemperature and thickness of the oxide film [166], while the ratesof diffusion through different alumina polymorphs differ substan-tially. In addition, polymorphic phase changes in alumina areaccompanied by substantial density change, so that the thicknessand continuity of the oxide layer and thus the oxidation rate can bedramatically affected by such transitions, even though the transi-tion energy is negligible compared to that of aluminum oxidationheat release. In another relevant case, many important thermitereactions occur through formation of multiple intermediate phases.Likewise, for the 2AlþMoO3 / Al2O3þMo reaction to occur in thesolid state (at relatively low temperatures), based on the Mo–Ophase diagram [167], molybdenum oxide is likely to form a numberof oxides, MoO3 / Mo9O26 / Mo4O11 / MoO2 before beingreduced to metallic Mo. When the initial MoO3 particle or inclusionis of the nano-dimension, the stability of different produced sub-oxides can be affected and, respectively, the sequence of the phasechanges can be altered compared to that implied by the knownphase diagrams developed for bulk materials. Therefore, the reac-tion mechanism in such thermite systems may depend on thedimensions of the initial oxide particles and layers. Detailed studiesof the stability of and phase transitions in various nanoscaledoxides used in reactive nanocomposite materials are thus neededfor the mechanistic understanding and modeling of the heteroge-neous reactions in such materials.

5. Exothermic heterogeneous reactions

The exothermic heterogeneous reactions are of primary interestas both driving ignition and affecting aging of reactive nanomaterials.Among such reactions, oxidation of nanoaluminum is the mostfundamental and occurs in all such materials containing aluminum.This oxidation process was recently studied extensively and itsmechanisms continue to be the subject of considerable debate. Itwill be discussed first, followed by a discussion of more complex,relevant solid–solid reactions.

5.1. Aluminum oxidation in gaseous oxidizers

The primary experimental technique used for studies of oxida-tion of n-Al in oxygen-based, vapor phase oxidizers is thermalanalysis. TGA curves show the sample weight increase as a result ofoxidation and simultaneous DSC measurements can be used toquantify the corresponding heat effects. TGA and DSC studies ofoxidation of various nanosized aluminum powders are reported inmany papers, including Refs. [144,146,149,168–170] and others. Theexperimental results are generally consistent between themselvesand show that for particles of about 100 nm and finer, the majorityof oxidation occurs at temperatures below the bulk aluminummelting point. This result was obtained with various aluminumnanopowders, using constant heating rates (most authors) andisothermal experiments [169]. Examples of combined DSC curvesmeasured in inert environments and TGA curves measured for thesame powders in oxygenated gas are shown in Fig. 11 [146] (note:all measurements are at the heating rate of 5 �C/min). The firstwell-defined oxidation step occurs between 500 and 600 �C. Fornanosized aluminum powders, this step results in a substantialweight increase. The onsets of the melting endotherms for allpowders shown in Fig. 11 are clearly observed at a highertemperature. Interestingly, the first oxidation step for the micron-sized aluminum powders occurs at the same temperatures as fornanopowders, resulting, however, in a very small mass increase,e.g., Refs. [6,7,144]. Furthermore, the oxidation kinetics identified inRefs. [6,7] for different steps observed for micron-sized aluminumpowders explains successfully the oxidation of nanosized powders[146].

The TGA results presented above are generally inconsistent witha set of data obtained using single particle mass spectrometry(SPMS) [52,171]. Unlike TGA experiments, SPMS measurements areperformed on individual nanoparticles. The nanoparticles areeither generated within the experimental setup by laser or arcablation of an aluminum target or from commercial powders fedinto the apparatus from a methanol suspension. The particles arepassed through a furnace with a section heated to a targettemperature and then fed into the SPMS analyzer. Characteristi-cally, the heated section is 30 cm long and the typical exposuretime is 1 s. In the analyzer, the particles are fed into a powerful laserbeam ensuring the complete ablation and ionization of all particles.The produced ions are sampled with a linear time of flight massspectrometer. Thus, the abundances of aluminum and oxygenatoms are obtained and the differences between particles withdifferent degrees of oxidation are identified.

Onset of meltingendotherm (~ 570 °C)

TGA (oxidation)

DSC (melting)

40 nm80 nm120 nmH

eat F

lo

w, m

W/m

g (exo

=u

p)

0

20

40

60

80

100

Mass ch

an

ge, Δ

m %

400 600 800 1000 1200Temperature, °C

Fig. 11. DSC and TGA curves measured for three aluminum nanopowders in argon andoxygen, respectively. Both sets of curves were acquired at a heating rate of 5 �C/min[146]. http://acsinfo.acs.org/journals/jpcbfk/index.html.



The SPMS results detect substantial oxidation only above themelting point of aluminum. These results were found to be inquantitative agreement with the hot-stage TEM observations alsoreported in Ref. [171]. The TEM imaging showed pronouncedchanges in morphology and evidence of aluminum leaking outfrom the oxide shell once the aluminum melting point wasexceeded. The authors of Refs. [52,171] argue that their results aremore accurate than TGA measurements (which they also per-formed obtaining results consistent with other published TGAexperiments [52]). The main proposed reason for the improvedaccuracy is that single, non-interacting particles are studied in theSPMS experimental configuration while particle interaction withinthe TGA sample could cause heating of the sample above thecontrolled TGA furnace temperature.

Review of multiple references describing similar TGAmeasurements for oxidation of a broad range of aluminum powderssuggests that the interaction of the particles in TGA experimentsresulting in a measurable sample overheating causing an acceler-ated oxidation is unlikely. Consistent TGA observations have beenreported for different powder samples treated in different instru-ments and using different sample holders and sample masses.Furthermore, the same reaction rates were observed in isothermalexperiments [169]. Note also that the overall oxidation sequenceincluding low temperature oxidation steps distinguished clearly inthe TGA experiments was observed for powders with a broad rangeof particle sizes, including 10–14 mm spherical powders [6,7], forwhich self-heating due to oxidation at low temperatures isexpected to be negligible based on their relatively small specificsurface area.

The reduced rates of oxidation implied by the SPMS experi-ments could have been affected by the following two issues thatneed to be addressed before definitive conclusions are drawn. Thefirst issue has to do with the sensitivity of the SPMS measurementand its capability to quantify the amount of oxygen present inaluminum particles coated by natural, 2–3 nm thick oxide layers. Itis stated in Ref. [52] that the method should be sensitive enough todetect a 1–2 nm oxide coating on aluminum nanoparticles. Despitethat statement, no oxide was detected for a commercial aluminumnanopowder while such powders always contain 2–3 nm naturaloxide layers. A clear demonstration of the sensitivity of the SPMStechnique is therefore needed that would indeed show the pres-ence of the 2–3 nm oxide coating present on commercial nano-particles, for which the presence of the oxide coating has beenestablished by TEM or another appropriate technique. A calibrationof the SPMS oxygen concentration measurement backed by anindependent measurement would certainly be beneficial. Thesecond issue is that in the experiments described in Refs. [52,171]the particles were subjected to a temperature ramp before enteringthe furnace section heated to the target temperature. Because theentire particle feed system was filled with an oxidizing gas, theheating ramp could substantially passivate (oxidize) the particlesresulting in the formation of a thicker and thus more protectiveoxide layer than exists on the starting particles. Based on thereported experimental conditions, it is estimated that the particlesapproaching the test section at the pre-set temperature, weresubjected to the elevated temperatures for a time period close to1 s. Thus, the formation of a protective oxide film was inevitable forthe nanoparticles produced in situ before these particles enteredthe test section pre-heated to the target temperature. For thecommercial nanoparticles with the initial protective oxide layer,the oxidation in the heating ramp section of the apparatus wouldbe comparable to that occurring in the calibrated test section. Thisissue is critical for the particles generated in situ by arc or laserablation, for which the passivation of the bare aluminum surfaceduring the heating ramp can occur very rapidly. Thus, it is expectedthat both commercial and in situ prepared nanoparticles could have

been similarly partially passivated prior to entering the furnacesection. Therefore, the measured reaction rates in both casescharacterize aluminum nanoparticles with initial passivating oxidecoating.

Aluminum oxidation has been studied using a differentialmobility analyzer to measure the density of aluminum particlessubjected to different temperatures in the oxidizing environment[172]. The oxidized particles were also examined using a TEM. Itwas found that particles exposed to a temperature exceeding1000 �C become hollow. Based on those observations, it wasproposed that at low temperatures, the rate of oxidation iscontrolled by inward diffusion of oxygen; above the aluminummelting point, the rate is controlled by aluminum outward diffu-sion. Thus, the oxidation was proposed to occur in two distinctregimes, responsible for the slow and fast oxidation occurringbefore and after aluminum melting, respectively. It was suggestedthat the conventional thermal analysis techniques are character-izing slow processes occurring in the time scale of minutes whilethe SPMS measurements presented in Refs. [52,171] probed a muchfaster process completed within 1 s [172]. The difference in timescales involved with thermal analysis and SPMS measurements iscertainly real and substantial. In fact, the time scales characteristicof the most practical applications of related energetic nano-materials are much shorter than those used in both techniques.Therefore, straightforward extrapolation of either type of results tothe time scales of interest is not expected to be meaningful. Instead,it is necessary to identify the involved reaction mechanisms andtheir kinetics so that meaningful reaction models can be developedand applied for the appropriate time scales. Generally, the longerthe time scales involved with a specific experimental technique, theeasier it is to overlook a short duration process. At the same time,such short duration processes can only be meaningfully detectedand described by studies of the same material systems at thesystematically varying time scales.

An increase in the aluminum oxidation rate at higher temper-atures [172] is certainly observed in all experiments and is inagreement with the concept of the oxidation reaction ratecontrolled by mass transfer processes. However, the distinct tran-sition from one oxidation regime to another upon aluminummelting is not detected in many TGA measurements that show noincrease in the oxidation rate upon aluminum melting. A short-lived change in the oxidation mechanism detectable by SPMS butnot by TGA could be associated with a rapid change in the oxidationrate (e.g., caused by discontinuities or cracks in the passivatingoxide layer) followed by a particle self-heating and, therefore,sustained higher oxidation rates. In TGA experiment, the self-heating of the sample is effectively suppressed that would mini-mize the effect of a short duration process. Further work is neededto better understand and quantify processes that affect oxidation ofdifferent size aluminum particles during aluminum melting.

A systematic study of aluminum oxidation using TGA andanalysis of the samples pre-oxidized to and recovered from specifictemperatures was presented in Refs. [6,7,146]. As noted above,these TGA results are in agreement with many other TGAmeasurements reported elsewhere. The results in Ref. [6] wereprocessed to establish a model [7] describing oxidation ofaluminum powders for both micron and nanosized particles [146].From XRD analysis of the samples recovered from specific oxidationtemperature conditions, it was established that different Al2O3

polymorphs are observed during different oxidation stages. Adiagram in Fig. 12 shows a characteristic thermo-gravimetricanalysis (TGA) curve of oxidizing aluminum powder (with micron-sized particles) and the sequence of changes in the alumina scalegrowing on the particle surface. The entire oxidation process isdivided into four stages and specific processes occurring duringeach stage are illustrated schematically. The natural amorphous



alumina layer covering the particle initially grows slowly during thelow temperature oxidation – Stage I. The energy of the oxide–metalinterface stabilizes the amorphous oxide at low temperatures andonly up to a critical thickness of about 5 nm [167,173]. When thecritical thickness is approached or when the temperature becomessufficiently high, the amorphous oxide transforms into g-alumina.The density of g-alumina exceeds that of amorphous alumina [145],and the smallest observed g-alumina crystallites have a size ofabout 5 nm [174]. Thus, if prior to the phase change the thickness ofthe amorphous layer was less than 5 nm, the newly formed g-Al2O3

crystallites no longer form a continuous coverage for the aluminumsurface. As a result, the rate of oxidation increases rapidly at thebeginning of stage II as shown in Fig. 12. For nanoparticles, this canresult in the complete or nearly complete oxidation (see Fig. 11). Forlarger particles, the openings in the oxide coating heal while onlya small fraction of the metal is oxidized. Upon healing, the rate ofoxidation decreases.

Based on the TGA trace shown in Fig. 12, no detectable massincrease occurs upon aluminum melting. Eventually, a regularpolycrystalline layer of g-Al2O3 forms by the end of stage II. Thegrowth of g-Al2O3 continues in stage III for which the oxidationrate was reported to be limited by the inward grain boundarydiffusion of oxygen anions [175,176]. Growth of the g-Al2O3 layercan be accompanied by phase transformations into other transi-tion polymorphs, such as d-Al2O3 and q-Al2O3, which havedensities very close to that of g-Al2O3 [145]. Such transitions arenot expected to affect the oxidation rate significantly. Otherprocesses are largely irrelevant for powders or materials withaluminum nanodomains, which are oxidized completely wellbefore the alumina layer becomes unstable. However, forcompleteness, the processes occurring at higher temperatures arereviewed briefly. Stage III ends when the increased temperaturedestabilizes the transition between alumina polymorphs. Theparticles stabilize at elevated temperatures and still densera-alumina polymorph starts forming by the end of stage III. Thus,strictly speaking, stage III can be further broken down into threeseparate sub-stages: growth of g-Al2O3, transformations to d- andq-alumina polymorphs, and transformation to a denser a-Al2O3

polymorph. Stage IV is considered to start when the oxide scale iscompletely transformed to a-alumina. When the initial a-Al2O3

crystallites form at the end of stage III, the thickness of the g-Al2O3

layer decreases, and the oxidation rate increases momentarily.When most of the oxide layer is transformed to the coarse anddenser a-Al2O3 crystallites, resulting in continuous polycrystallinecoverage, grain boundary diffusion processes slow down and theoxidation rate decreases rapidly.

In Ref. [7] the kinetic relations for the polymorphic phasechanges occurring in alumina films coating aluminum powderswere determined from TGA experiments. Thus, a quantitativedescription of aluminum oxidation in gaseous oxygen for particlesand domains of different sizes was obtained.

5.2. Reactions between condensed components

Three main types of related reactive systems include thermite,intermetallic, and metal–metalloid compounds. Among these threeclasses of composites, reactions in thermites are of most interest forenergetic materials applications and they have attracted the mostattention in the research community. Reactions mechanisms inhighly exothermic metal–metalloid systems (e.g., B–Ti or B–Zrnanocomposites) are among the least studied. Therefore, thediscussion below focuses primarily on the reactions in thermites.

The sequence of processes occurring in aluminum oxide andaffecting aluminum oxidation in gaseous oxygen described above isexpected to remain generally valid for aluminum reactions withvarious oxidizers, including metal oxides in nanocomposite ther-mites. Clearly, the kinetic parameters of individual processes areexpected to be affected. The stability ranges of different aluminapolymorphs can also change as a result of presence of other metalsor oxides. Formation of alloys or ternary oxides can result in evenmore complex and multistage reactions. For thermites, reduction ofmetal oxides most often does not occur in one step; for example, asnoted above, it is reasonable to expect that MoO3 reduction to Mooccurs through formation of Mo9O26, Mo4O11, and MoO2 accordingto the Mo–O phase diagram [167]. Formation of each individualsuboxide is expected to alter the reaction kinetics. Current researchis primarily aimed at quantitative experimental characterization ofthese more complex reaction mechanisms. It is expected that laterit will become possible to identify which processes and reactionsare rate controlling under which conditions, so that simplified,accurate practical reaction models can be developed.

Thermal analysis remains the most important tool in studyingreactions in thermite-type nanocomposites and in nanocompositematerials employing other types of exothermic solid–solid reac-tions. While the nano-dimensions themselves may not necessarilyresult in new reaction mechanisms, the exothermic processes thatoccur very slowly and remain undetected for coarser materialscan become dominant ignition triggers for nanocomposite mate-rials with the identical bulk compositions. In fact, DSC and DTAmeasurements for many nanocomposite materials detect reactiononsets at low temperatures and will be useful in characterizingthe respective reaction kinetics. Despite many advantages, as noted

Fig. 12. Change in mass of the aluminum powder oxidizing in a thermal analyzer. Different stages of oxidation are indicated and the respective changes in the growing aluminascale are shown schematically [7]. http://www.tandf.co.uk/journals/titles/13647830.asp.



above, the thermal analysis techniques are poorly suited fordetecting short-lived processes that might result in self-heating forindividual particles or small samples. It should also be emphasizedthat the straightforward extrapolation of the kinetic behaviorobserved in thermo-analytical experiments to much shorted timescales is not meaningful. Instead, reaction mechanisms identifiedfrom detailed thermo-analytical studies combined with otherexperimental techniques should be elucidated and used to predictthe reactions occurring in practical applications.

Reactions in aluminum–copper oxide nanofoils, produced bymagnetron sputtering, were studied in a series of experimentspresented in Refs. [116,117]. Differential thermal analysis was usedin addition to TEM and Auger profiling of the partially reactedsamples. The preparation of multilayered thermite nanofoilsinvolves elevated temperatures so formation of pure CuO was notpossible. It was reported that the sputter-deposited copper oxidehad the structure of the mineral paramelaconite, Cu4O3. The Al/Cu4O3 molar ratio in the foils is approximately 2.5, so that based onthe final products of Al2O3 and Cu, the system is somewhataluminum-rich. Furthermore, aluminum was partially oxidized andcopper oxide was partially reduced during the deposition so thatthat Al and Cu4O3 bilayers with combined thickness of about 1 mmwere separated by approximately 100 nm thick interface layers inwhich the concentrations of components were continuouslychanging, as was confirmed by the Auger profiles of Al, Cu, and O[116]. A narrow region of the interface was identified as anamorphous or nanocrystalline Al2O3.

DTA analysis showed that the exothermic reaction in suchnanofoils occurred in two steps, with the peaks around 625 �C(890 K) and 835 �C (1110 K). Based on the DTA trace presented inRefs. [116,117], and reproduced here as a dashed line in Fig. 13,the onset of an exothermic reaction can be assigned to about500 K. The authors mention a broad exothermic shoulder begin-ning at about 470 �C (about 745 K) in addition to the two largepeaks mentioned above. Based on detailed XRD and Auger studiesof the samples quenched before the first exothermic peak, theauthors concluded that most of the paramelaconite was trans-formed into a mixture of CuO and Cu2O. Based on further anal-yses of the partially reacted foils, likely rate-determiningprocesses for each of the two reaction steps were proposed. In

the first exothermic reaction, the lateral growth of Al2O3 nucleiwas proposed to control the reaction rate. The reaction endedwhen a continuous Al2O3 layer was formed. Note the similarity ofthe temperature ranges for the end of this reaction step describedin Ref. [117] and the end of aluminum oxidation step II illustratedin Fig. 12, which was also assigned to the formation of a contin-uous g-Al2O3 layer [6,7]. The second exothermic peak for theAl–CuO nanofoils was proposed to be controlled by either diffu-sion of O through the Al2O3 or by thickening of the Cu product bya nonuniform reduction in copper oxide serving as the oxygensource. This interpretation is again consistent with that discussedin Refs. [6,7] for aluminum, where the respective oxidation stepwas proposed to be controlled by the diffusion-controlled growthof g-Al2O3.

Reactions in a similar Al–CuO nanocomposite material preparedby ARM were studied in Ref. [177]. The bulk composition wasstoichiometric, i.e., 2Alþ 3CuO and detailed DSC studies accom-panied by XRD analysis of the samples oxidized to differenttemperatures were reported. A DSC trace from Ref. [177] is directlycomparable to that presented in Refs. [116,117] – shown in Fig. 13.The onset of exothermic reactions for the ARM-prepared materialoccurs at a noticeably lower temperature and the low temperatureshoulder is resolved to include at least two broad peaks. Only one ofthe two higher temperature peaks observed for nanofoils [116,117]is reported for the nanocomposite powders [177], for which thetemperature scan was stopped at 1013 K (740 �C). In addition, two-weak endothermic peaks are observed between 800 and 900 K,corresponding to the eutectics with CuAl2 and Cu9Al4, respectively.The presence of intermetallic phases in the reaction products wasconfirmed by XRD. Both the better resolved low temperatureshoulder and the formation of intermetallic phases in the Al–Cusystem (requiring the presence of metallic Cu), suggest a substan-tially greater reaction rate at low temperatures for materialsprepared by ARM as compared to the nanofoils. Because of thehigher rate, a greater fraction of the copper oxide is reduced at thesame temperatures for the ARM-prepared materials. The metalliccopper formed is then capable of reacting with aluminum andproducing the observed intermetallic phases. This enhancement ofreactivity for the ARM-prepared materials compared to respectivenanofoils can be explained by two main factors. First, as notedabove, the magnetron sputtered copper oxide was partially reactingwith aluminum during its deposition and the initial phase of thedeposited copper oxide was partially reduced. Second, the spatialscale of mixing between the metal and the copper oxide in thenanofoils was likely coarser than in the three-dimensional nano-composite particles produced by ARM.

Studies of the partially reacted samples recovered from differenttemperatures presented in Ref. [177] were inadequate to proposea specific reaction mechanism corresponding to the complex DSCpattern. It was clear that the reduction of CuO occurred throughformation of Cu2O followed by the formation of Al–Cu alloys andmetallic copper. Produced aluminum oxide phases were poorlycrystalline and only weak peaks of g-Al2O3 were detected in theXRD patterns of the partially reacted samples. Thus, instead ofproposing a specific reaction mechanism, the reaction was simplypresented as a superposition of at least four overlapping individualreactions. For each individual contribution, the generic reactiontype was selected and activation energy was estimated specified bymatching experimental DSC results collected at different heatingrates. Additional work is needed to further clarify the reactionmechanisms in this system.

Similarly to reactions for Al–CuO thermites, reactions forAl–MoO3 thermites were studied using thermal analysis for nano-composite materials prepared by different techniques. DSC curvesfor stoichiometric 2AlþMoO3 prepared by ARM [178] and fora mixture prepared by ultrasonication of nanosized Al and MoO3

300 400 500 600 700 800 900 1000 1100Temperature, K

1200

Heat flo

w, m

W/m

g; E

xo

- u

p

Nanofoil, DTA signal [116]

Nanocomposite powder produced by ARMDSC signal [177]

Fig. 13. TGA and DSC traces for reacting aluminum–copper oxide nanocompositematerials prepared as multilayer nanofoils [116,117] and as fully dense nanocompositepowders [177], respectively. Both traces recorded in flowing argon at 40 K/min.



powders at the equivalence ratio of 1.2 [170] are shown in Fig. 14.The equivalence ratio for the ultrasonicated nanopowders wasselected based on earlier experiments [158] showing the maximumreactivity of the respective nano-thermite prepared at that equiv-alence ratio.

Comparison of the DSC traces shown in Fig. 14 to each othershows that the exothermic reaction starts at lower temperaturesand occurs more actively for the materials prepared by ARM. Inboth cases, the exothermic features are broad and individual peaksare difficult to separate. A closer inspection of the traces shown inFig. 14 indicates that the DSC trace for the ARM-produced materialcould be constructed by overlapping the trace observed for themixed nanopowders, with an additional, very broad exothermicfeature starting at about 500 K and monotonously increasing up toabout 900 K. Indeed, without specifying the reaction mechanism,the DSC traces for ARM-prepared materials were proposed to bemodeled as a superposition of four overlapping reactions. Three ofthese reactions described individual smaller peaks, similar to thosedetectable from the DSC trace for mixed nanopowders. Each ofthese three peaks was described as a first order reaction, with theactivation energies of 209, 211, and 373 kJ/mol, respectively [178].The fourth, broad feature, not detectable for the mixed nano-powder samples, was approximated by a low activation energy(90 kJ/mol) Jander type reaction commonly describing 3-dimen-sional diffusion. Note that the activation energy for the strongestpeak observed in Ref. [170] was about 240 kJ/mol. Based on the DSCtraces recorded at the same heating rate and shown in Fig. 14, forthe ARM-prepared materials this peak is located between the thirdand fourth identified peaks, for which respective activationenergies were 211 and 373 kJ/mol, respectively [178].

Research aimed at identifying the mechanisms of these andother exothermic reactions in nanocomposite reactive materials isongoing and it is clear that the respective mechanisms are complexand involve a number of overlapping exothermic processes. Twofundamentally different approaches are possible to describe suchreactions. One approach is based on understanding the individualprocesses occurring in the reacting materials. Such understandingis only possible based on complex investigations involving variousmaterial probing techniques combined with thermal analysis. The

only example of this approach is reported in Refs. [116,117] forthermite nanofoils. However, much more work is needed in orderto truly understand the steps and processes involved in reactions inAl–CuO or other reactive systems.

Other published studies pursued a less fundamental approach inwhich the thermo-analytical measurements were simply inter-preted as one or many overlapping processes-effectively using‘‘curve-matching.’’ While effective and relatively simple, thissecond approach is only useful for quantitative description of thevery same systems studied in the DSC experiments. A deviation inthe sample preparation conditions, stoichiometry, or usinga different method for preparation of the sample with the identicalcomposition, can make the sequence of overlapping processesidentified earlier irrelevant. Only development of the mechanismsof the occurring processes, combined with detailed informationabout the initial sample morphology and particle or domain sizedistribution, can be translated into a model transferable from onematerial modification to another. Developing such mechanisticmodels is, therefore, a high priority objective for future research. Inaddition to techniques commonly employed, it is expected thatadvanced materials characterization techniques, such as synchro-tron radiation based diffraction and absorption methods, will needto be used to analyze nanocomposite materials recovered fromdifferent reaction stages, or even in real time, during their reactionoccurring under well-controlled conditions.

Reactions in intermetallic systems have been studied for manyyears and good reviews of the current approaches and under-standing can be found in the literature, e.g., Refs. [179–181]. Theresearch is focused on understanding of the combined processes ofdiffusion, formation of solid solutions, and the formation of theintermetallic compounds. One of the most reactive intermetallicsystems is Al–Ni and despite extensive previous studies [181], thereaction mechanisms in this system remain the subject of debate.This reaction was recently studied for nanofoil materials [182]. Themechanism of the Al–Ni reaction is complex and includes forma-tion of multiple intermediate phases, so that detailed studiescombining thermal analysis with recovery and characterization ofsamples heated to intermediate temperatures, as done in Ref. [182],are extremely useful. The simple morphology of the nanofoilscombined with a large reactive interface area enabled researchersto observe and study reaction steps that are difficult to characterizefor other types of composite materials. Different phases (e.g.,Al9Ni2, Al3Ni) were observed to form as the first intermediatereaction product in the bilayers of different thicknesses, which mayreflect the effect of nanoscale dimensions on the thermodynamicstabilities of different compounds – mentioned above. Furtherstudies of this reaction, validating and expanding the simplifiedmodels suggested in Ref. [182], will be useful for understanding themechanism of intermetallic product formation.

6. Ignition studies

The exothermic heterogeneous reactions reviewed in theprevious section are expected to be responsible for ignition ofreactive nanocomposite materials under practically useful condi-tions. It is therefore expected that detailed reaction models basedon thermo-analytical measurements combined with analysis ofchanges in the material morphology, structure, and composition,will be capable of predicting the ignition kinetics in such materials.However, given the complexity of processes involved, independentexperimental validations of such models are necessary. The salientfeature of ignition is that the materials are heated rapidly withpowerful heat sources, such as a detonation front produced by anignition primer or as an expanding fire ball generated by a highexplosive. Respective heating rates are of the order of 106–107 K/s.Thus, experimental validations of the ignition models must

Heat F

lo

w, m

W/m

g; E

xo

- u

p

400 500 600 700 800 900 1000

Temperature, K

Nanocomposite powder produced by ARM [178]

Mixed nanopowders [170]

Fig. 14. DSC traces for reacting aluminum–molybdenum oxide nanocomposite mate-rials prepared as mixed nanopowders [170] and as fully dense nanocomposite powders[178], respectively. Both traces recorded in flowing argon at 5 K/min.



generate comparable heating rates while enabling controllable andreproducible conditions and accurate ignition detection.

In addition to reproducing the high heating rate, the entirenanocomposite should be heated and ignited in the ignitionexperiment, as opposed to selective heating (and possible volatili-zation or decomposition) of one component, which could result indifferent reaction kinetics. The uniform heating of compositematerial is readily achieved in thermo-analytical experiments withvery low heating rates, but may present a challenge when theheating rates increase. Finally, for materials prepared as mixednanopowders, the sample porosity can vary widely betweenexperiments which could also substantially change the ignitionbehavior.

Lasers offer the capability of readily adjustable and well-controlled energy transfer to reactive materials with the range ofpower necessary to achieve the heating rates of interest. Lasers arenow commonly used for laboratory studies of material ignition, e.g.,Refs. [183–188]. Often, however, a number of issues make theinterpretation of laser ignition experiments difficult. Most signifi-cant issues for the nanocomposite materials are the consistency inthe sample preparation and differences in efficiency of absorptionof the laser energy by different material components (e.g., metals vsoxides or particles of different sizes present in the material). Theefficiency of absorption of the near-infra red and visible spectra bynanoparticles of Al and B embedded in nitrocellulose or Teflonoxidizers was studied in Ref. [189]. It was found that absorptionstrength is greater for Al nanoparticles as compared to bulk Al. Adifferent experimental approach was used to assess the scatteringand absorption efficiency of n-Al and nanosized MoO3 powders inRef. [190]. It was found that close-packed nanopowder of MoO3

scatters most of the incident light while about 2/3 of the incidentlight are absorbed in a similarly packed n-Al. The experimentalapproach used in Ref. [190] required preparation of a 1-D slab witha moderate optical thickness, which proved to be difficult for n-Al.Additional measurements quantifying optical properties areneeded for a broader range of nanocomposite energetic materialsand incorporation of such data for interpretation of the laser igni-tion experiments is desirable.

In some laser ignition experiments, nanosecond or even shorterlaser pulses are used resulting in significantly higher heating ratesoften exceeding 109 K/s [191–194]. Such experiments result ina completely different initiation mechanism and are well suited forstudying kinetics of the ensuing vapor phase reactions, which are,in most cases, irrelevant for heterogeneously driven ignition.

An experimental study on laser ignition of Al–MoO3 nano-composite thermites prepared by ultrasonic mixing of respectivenanopowders was reported in Ref. [158]. A 50 W CO2 laser was usedto ignite pellets pressed to 38% TMD prepared from different sizealuminum powders and different equivalence ratios. Ignition delaywas measured as the time between the laser onset and the firstdetection of optical emission by the pellet. The results of experi-ments with nearly stoichiometric thermites prepared usingaluminum particles with different sizes are illustrated in Fig. 15. Theignition delays were unaffected by the particle size for nano-powders. Compared to the 100 nm powder, the ignition delaysappeared to increase roughly by an order of magnitude with anorder of magnitude increase in the aluminum particle sizes. Theeffect of equivalence ratio was studied using three differentaluminum nanopowders (nominal sizes of 30, 40, and 108 nm). Theresults did not show substantial differences in the ignition delayswhich were around several tens of milliseconds for the range ofequivalence ratios of 0.5–2. Ignition delays increased noticeablywhen the samples were Al rich (equivalence ratios greater than 2)and when micron-sized Al powders replaced the n-Al.

The results presented in Ref. [158] are interesting, but they donot seem to provide the data conducive for direct validation of any

specific ignition model. The difficulties in the data interpretationpresented in Ref. [158] are instructive. The millisecond ignitiondelays could have been described theoretically by modeling theheat transfer processes of the pellet heated by the laser; however,such a description requires the value of the bulk thermal conduc-tivity of the pellet, which is very difficult to either predict ormeasure. This difficulty is common for all reactive nanocompositematerials. The absence of a clear effect of the size of aluminumnanoparticles on the ignition delay most likely results from thedifferences in the mixing quality of the thermite compositionsprepared with different nanopowders. It is, for example, reasonableto expect that the finer nanoparticles were mixed less uniformlywith MoO3 than coarser ones, so that the effect of reduced particlesize was effectively negated. Mixing uniformity is determined bythe sample preparation method, and focused efforts will benecessary to evaluate it quantitatively. The specific details of CO2

laser radiation interaction with either nanoaluminum or nano-MoO3 are also poorly known and it is unclear what the differencesmight be without a specific study. Finally, in addition to opticalmeasurements of the igniting sample emission, temperaturemeasurement of the sample was attempted in Ref. [158] usingthermocouples. The reported temperature traces suggested igni-tion of thermites prepared with nanopowders at around 100 �Cwhile the same thermites prepared with the micron-sized powderswere reported to ignite at 610 �C. As reported earlier in the samearticle, ignition for both types of samples was detected opticallyusing a high speed video camera. Without a very sensitive IRdetector, however, optical emission is extremely hard to detect at100 �C and even at 610 �C. These low temperatures are alsoinconsistent with any other measurements on ignition of materialscontaining Al or nano-Al as fuels. The authors of Ref. [158] admitthat the temperature curves presented are inaccurate and do notaccount for transient effects and other experimental errors butsuggest that such data can be used for qualitative comparisonbetween different size aluminum powders. At this time, however,such qualitative comparisons are less useful while quantitativemeasurements of ignition delays are needed to validate specificignition mechanisms.

An attempt to obtain a more quantitative characterization ofignition kinetics for nanocomposite reactive materials using laserignition was made in Ref. [195]. The nanocomposite systemconsidered was Al–Ni and the material was prepared by ultrasonicmixing of the respective nanopowders. Cylindrical pellets with55–60% TMD were then pressed and ignited using a CO2 laser beam.The variation of the measured ignition temperature as a function ofthe heating rate was obtained. Activation energies were recovered

Fig. 15. Ignition delays measured for laser ignition of Al–MoO3 thermite powdersprepared with aluminum particles of different sizes [158]. http://www.elsevier.com/wps/find/journaldescription.cws_home/505736/description#description.



using isoconversion data processing. A dramatic difference in theignition activation energy for micron-sized and nanosized Alpowders was reported. Thermocouple measurements similar tothose in Ref. [158] were used to identify the ignition temperature.Even though the heating rates in Ref. [195] were rather low, on theorder of 1–10 K/min, the error in the thermocouple measurementstill needs to be quantified in order for the reported ignitionkinetics to be meaningful. Unfortunately, no error analysis for thethermocouple measurements was provided. Because such errorswere certainly affected by the heating rate, the reliability of theactivation energies reported in Ref. [195] is low. A good summary ofactivation energies relevant for the Al–Ni reactions, including thoseoccurring in the nanofoils, is presented in Ref. [182], which does notinclude the unusually low activation energies for nanocompositesreported in Ref. [195]. Generally, a decrease in the particle size isnot expected to reduce dramatically the activation energy for anyheterogeneous reaction; instead, it is expected to substantiallyincrease the value of the pre-exponent factor in the Arrhenius-typereaction kinetic description. Surprisingly, the values of pre-expo-nent factors, also determined in Ref. [195], are of the same order ofmagnitude for all micron-sized and nanosized powders used toprepare composite samples. This could indicate a problem with thedata processing, and/or again point out at the effect of the errorassociated with thermocouple-based identification of the ignitiontemperature. Finally, the heating rates achieved in Ref. [195] aretypical for thermal analysis experiments and much higher rates aregenerally desired in laser ignition configurations, which are inter-esting for validation of practically useful ignition kinetics models.

Heterogeneous shock tube experiments offer another configu-ration in which well-defined and high heating rates can be achieved[196]. In such experiments, a powder-like sample is placed orinjected near the wall the ‘‘driven’’ section of a shock tube and theincident and reflected shock waves pass over the sample quicklyand heat it in two sequential steps, closely following each other. Thisway, the sample is nearly instantaneously introduced into a hot gasand its temperature history can be relatively easily calculated witha convective heating model. The sample ignition and combustionare monitored optically and the ignition delay is quantifiedconsidering the well-known timing of the reflected shock. Thistechnique is not suitable for the nanocomposite samples preparedas mixed, low density nanopowders because such materials disin-tegrate after being dispersed by the shock wave. On the other hand,this technique is well suited to study ignition kinetics for nanosizedaluminum powders as well for the fully dense nanocompositepowders prepared by ARM. The results of measurements presentedin Ref. [196] are shown in Fig. 16. Ignition delays measured inRef. [196] for nanosized aluminum powders and nanocompositethermites were of the order of a microsecond, while milliseconddelays were observed for a micron-sized aluminum powder(cf. inset in Fig. 16.) The main combustion peaks in the powderemission were also shifted in time. The most rapid combustion wasobserved for the Al–MoO3 nanocomposite thermite, followed bynearly coinciding peaks for n-Al (ALEX) and Al–Fe2O3 nano-composite thermite. The combustion peak for the micron-sized Alpowder was observed after the longest delay.

Unfortunately, the measurements of ignition kinetics arestrongly affected by the particle size distribution in the sampletested and by possible particle agglomeration (especially significantfor nanosized powders). Thus, in order to directly validate a quan-titative ignition model, detailed characterization of the particle sizedistribution is necessary and the particle size distribution effectneeds to be explicitly addressed in both experimental validationsand respective calculations.

Experimental studies of the ignition of powder-like samplescoated on an electrically heated filament were described in Refs.[121,123,177,197,198]. The filament temperature was measured in

real time using an infrared pyrometer focused on an uncoatedfilament surface adjacent to the powder coating. The ignitioninstant was detected optically using a second photo sensor focusedon the powder coating. The method provides well-controlledheating rates between 102 and 105 K/s. A detailed analysis of theheat transfer between the filament and the thin layer of the powdercoating was presented in Ref. [197] and ignition kinetics for variousnanocomposite samples was reported in Refs. [121,123,177]. Igni-tion temperatures for nanocomposite materials were found to varyin a general range of 600–1000 K [177,178].

The kinetics of material ignition determined from experimentsperformed with different heating rates were compared with thekinetics of various exothermic events observed with the samematerials from a thermal analysis (DSC), performed over a differentrange of heating rates. An example of such comparison for Al–MoO3

thermites is shown in Fig. 17. The data shown in the coordinatescorresponding to the isoconversion processing: the logarithm ofthe ratio of the temperature square over the heating rate (T2/b) is

Fig. 16. Results of experiments on ignition of different powders in heterogeneousshock tube experiments [196]. Emission intensity at 486 nm is shown vs time for fourenergetic materials in 30%O2/70%N2. The ambient temperature is 2250 K and theambient pressure is 3.0 atm. The inset shows a magnified temporal region soon afterendwall reflection. http://www.tandf.co.uk/journals/titles/00102202.asp.

600 4005008001000T,

apparent peak

peak onsets

thermal analysis

filament ignition

0.0010 0.0015 0.0020 0.0025

T-1

, K-1

0

5

10

15

20

ln

(T

2/β

)

peak 1peak 2peak 3

peak 4

Incr

easi

ng h

eatin

g ra

te

Fig. 17. Isoconversion analysis of DSC and ignition data for nanocomposite powderswith bulk composition 2AlþMoO3 prepared by ARM [178].



plotted against the inverse temperature (1/T). For clarity, a top axisis added showing the actual temperatures corresponding to thehorizontal coordinate. The DSC curve processed is similar to the topcurve shown in Fig. 14 and includes a number of exothermicfeatures. Four of the most prominent features were processed frommultiple heating rate DSC experiments to obtain a family of linescorresponding to the low heating rate range in Fig. 17. The heatedfilament ignition experiments, performed at much higher heatingrates are shown in the same plot. One immediate conclusion fromcomparing the filament ignition and DSC inferred kinetic trends isthat the strongest peak (represented by open circles and labeled‘‘peak 4’’in Fig. 17) observed in the DSC signal at higher tempera-tures does not really affect ignition of these powders, which occursat lower temperatures despite much higher heating rates. Thus, theprocesses occurring at lower temperatures must be responsible fortriggering the material ignition. Weaker exothermic events areindeed observed in the DSC curves at lower temperatures(cf. Fig. 14). None of the trends obtained from DSC for the lowtemperature events can be directly correlated with the kinetictrend observed for the filament ignition experiments, however. Themost likely explanation is that the low temperature exothermicfeatures observed in the DSC experiments do not represent indi-vidual reaction steps but rather are compounded events. The directextrapolation of kinetic trends for such compounded eventsobserved in the narrow range of heating rates available for DSCmeasurements to the much higher heating rates (as can be impliedby Fig. 17) is poorly justified. Thus, the low temperature exothermicevents need to better characterized in order to predict ignition ofsuch materials. An initial analysis and modeling approach wereintroduced in Refs. [177,178] based on a model of multiple over-lapping processes that occur during the sample heating. Iso-conversion processing may not be very useful for analysis of suchreactions involving steps that overlap and substantially affect oneanother [199].

7. Combustion studies

7.1. Laboratory tests

In the early work on reactive nanocomposite materials, theprimary motivation for most of the laboratory combustion experi-ments was to establish a clear and reproducible difference betweencombustion of such materials and that of reference materials usingthe same elemental or molecular compounds which were notmixed on the nanoscale. This led to a large number of semi-quali-tative measures of reactivity, primarily involving one or anothervariant of an open tray burn experiment, e.g., Refs.[92,99,105,120,141,200,201]. A sample of material, typicallya powder, was placed in an arbitrarily selected open sample holder,often an elongated channel, and ignited using a pilot flame, hotwire, piezoelectric igniter, laser, or other appropriate device. Theensuing flame was typically visualized using high speed video. Acharacteristic photograph of such an experiment presented inRef. [89] is shown in Fig. 18. Samples with reactive components ofdifferent sizes, mixed on different scales, mixed using differenttechniques, or prepared using different synthesis approaches, wereexamined. Optical temperature measurements were also attemp-ted [200,201]. Generally, qualitative differences in combustionwere observed with the apparent flame speeds for nanocompositematerials several orders of magnitude greater than the speedsmeasured for similar materials mixed on the micron or coarserscales. From these studies, flame speeds for the nanocompositematerials, with the same or similar compositions, vary over a broadrange from w0.1 to 103 m/s and appear to be most affected by thepacking density of the sample. Unlike conventional explosives, forreactive nanomaterials, a higher sample density always results in

a lower visible flame speed. The visible flame speed was observedto be strongly affected by the size distributions and/or types of fuel[98,201,202] and oxidizer [99,141] particles, method and uniformityof mixing nanosized fuel and oxidizer powders [51,141], and scale ofmixing achieved in the fully dense nanocomposites prepared byARM [120].

As the initial qualitative features of combustion of nano-composite materials became established, open tray experimentsbecame more instrumented to obtain more quantitative andmeaningful combustion characteristics. Photodiodes with colli-mated inputs were used to register flame arrival instant and obtaina more accurate measurement of the flame speed [38,140,202].Pressure sensors were also used to obtain the pressure signatures ofthe propagating flames [100]. Poor reproducibility between theflame speed measurements in open tray experiments was oftenobserved and it was suggested that the setup could be improved byinstalling a periodic sequence of baffles with small openings alongthe channel in which the sample is placed and ignited [38,97]. Thebaffles were meant to minimize the effect of removal or suspensionof unignited free powder as a result of the strong convective flowsproduced by the propagating flames. Improved reproducibility wasindeed achieved; however, the measured flame velocitiesdecreased substantially and the mechanism of flame propagationremained unclear.

Further modifications of the experimental methodology haveincluded using cylindrical tubes in which the sample was packedand ignited [100,203]. Such experiments are often referred to asa ‘‘confined sample burn.’’ Usually the tubes are equipped withmultiple side openings for pressure transducers and fiber opticscables feeding the flame emission signal to photodiodes for kineticsand emission analyses. As the ends of the tubes are typically open,such confinement of the combustion materials is only partial.Recently, flame propagation in microchannels was reported fornanocomposite thermite [204] and nanoaluminum/water systems[205]. A sequence of the high speed images illustrating combustionof an Al–MoO3 thermite prepared by mixing respective nano-powders in a narrow, 2 mm tube is shown in Fig. 19 [204]. Ina related study, electrical conductivity was measured for a burninga nanocomposite sample partially confined between two brasselectrodes [206]. Electrical contacts were used to measure theflame propagation speed in nanocomposite aluminum–Teflon�

materials prepared by mechanical milling [127–129] and placed inchannels at relatively low densities (typically under 50% TMD). Themeasured speeds of flame propagation varied in the range of700–1280 m/s, while independent measurements established thatthe speed of sound in a similarly prepared sample is only about100 m/s. Thus, the feasibility of detonation in a nanoscale Al–Teflon

Fig. 18. Photo of thermal ignition of energetic nanocomposite [89]. http://www.aiaa.org/content.cfm?pageid¼322&lupubid¼24.



mixture was reported. As in the other similar measurements, themeasured flame propagation speed was observed to decrease withthe increase in mixture density [129].

Despite a relatively large number of reports, observations basedon the apparent flame speed for nanocomposite materials arerather limited. It was reported that the dilution of the nano-composite materials with inert additives [141] results in anappreciable reduction of the flame speed. It was also observed thatpreparing off-stoichiometric compounds can result in an increasedflame speed for slightly metal rich cases, e.g., for Al–CuO thermite[97] or for Al–MoO3 thermite [158,203]. On the other hand, it hasbeen reported that an increased flame speed can be maintainedwhen a nanosized aluminum powder is blended with a micron-sized powder. The flame speed for a nanocomposite with the fuelprepared as a blend of micron and nanosized powders matchedthat of the nanocomposite prepared with pure aluminum nano-powder when about 60% of nanopowder was added [140]. As notedabove, it was observed that the sample packing density affects theapparent flame speed dramatically; for example a difference by twoorders of magnitude was observed for loose and ‘‘packed’’ samples,with respective densities of 5–10% and 35–55% TMD [100]. Inter-estingly, it was also reported in Ref. [100] that the effect of packingdensity on the pressure produced by the burning nanocompositethermites was opposite to the effect of packing density on theapparent flame propagation velocity.

Interpretations of the apparent flame propagation measure-ments, often showing supersonic values of the flame speeds for thenanocomposite materials, are difficult. The convective flow patternsare better defined for the partially confined samples, but the thermalproperties of the loosely packed or even pressed nanocompositesamples critical for description of the flame propagation remainpoorly quantified. These properties are most likely changingdramatically as the propagating flame heats up the sample and/or asthe sample is compressed by the pressure wave generated by theflame front. The magnitude of such changes is likely comparable to

that observed for the electrical properties of the nanocompositemixtures affected by the flame propagation [206]. Simplified heattransfer models for flame propagation in confined and unconfinednanocomposite materials are offered in Refs. [97,207,208]. Thesemodels operate with lumped time scales and bulk burn rates and donot include the inherent reaction kinetics and varied transportproperties of the reacting materials. Such models are attractive forpractical users but the multiple simplifying assumptions made arenot currently justified. Therefore, while it is possible to select theadjustable parameters to fit a selected experimental data set, it isunreasonable to expect that the same parameters will be useful inpredicting combustion behavior for materials with altered proper-ties or burning in a different configuration.

At the present stage of development of reactive nanocompositematerials, useful flame propagation models should take advantageof the available reaction kinetics measurements discussed above. Inaddition, changes in the thermodynamic and transport properties ofthe reacting materials need to be considered. For example, anenhanced heat transfer is expected when gasified reduced metalformed in a thermite reaction, such as Mo (or Cu and Bi) condenseson unignited or burning particles [204]. Only after such details areincorporated into more complete models, can a simplified version ofthe flame propagation model be offered in which the simplifyingassumptions can be justified. In another recent report, it wassuggested [100] that the flame propagation mechanism can becompared to convective detonation [209,210] for flame propagationin a porous medium with a thin surface layer of explosive. It appearspresently that further work is needed to understand such flamepropagation mechanisms and whether convective detonationindeed describes the phenomena observed for combustion ofnanocomposite materials. Without such understanding, futureexperiments using open burn or partially confined burn configura-tions do not appear to be well justified and are not expected to yielduseful information about the properties and reaction mechanisms inreactive nanocomposite materials.

Another experimental technique aimed at the assessment of thecombustion performance of reactive nanocomposite materials isbased on a pressure measurement in a confined chamber or pres-sure cell [51,88,94,101,120,140]. Standard vessels for oxygen bombcalorimetry [94,140] as well as smaller [51,88,101] and larger [120]vessels have been used and combined with different igniters andpressure transducers. In this approach, the experiment can beconsidered as nearly adiabatic and the pressure increase can berelated to the energy produced in combustion [94]. One substantialdifficulty in interpreting the experimental data is the production oftransient, or semi-stable gas species. Examples of this kind ofsystem are: Bi vapor in combustion of thermites using Bi2O3 as anoxidizer [51,101], production of hydrogen and water using hydratedoxides as oxidizers [94], production of nitrogen if nitrates are usedas oxidizers [121], etc.

Because combustion may not follow the equilibrium thermo-dynamic predictions, thermodynamic calculations can only serve asan initial guideline for interpreting the experimental results.Despite such difficulties, it appears that comparison of differentmaterials, or materials prepared from the same compounds butwith different particle sizes or morphologies, can be very mean-ingful based on such pressure measurements. Both maximumpressures achieved and the rates of pressure rise are of interest. Inaddition, the final pressure in the chamber after the reaction iscompleted and the products are cooled, is of interest as indicative ofthe final make-up of the gaseous reaction products. Finally,condensed reaction products can be readily recovered and analyzedyielding information important for development of the reactionmechanism. Sets of related measurements with reactive nano-composite materials prepared by ARM are presented in Refs.[124,211]. Because these materials comprise fully dense micron-sized

Fig. 19. Combustion of nanoscale Al/MoO3 thermite in a microchannel [204]: imagesequence of an experiment performed in a 2-mm diameter tube. The time betweenimages is 13.8 ms. The expansion of the rubber tube used to hold the spark igniter andsome blowby of products from the ignition event are visible at the top of the image.http://www.aiaa.org/content.cfm?pageid¼322&lupubid¼24.



particles that retain their morphology and structure upon beingdispersed in a gaseous oxidizer, the composites ignited and burneduniformly when aerosolized within the explosion vessel [124,211].Thus, the pressure measurements were interpreted in terms of theflame propagation speed, in addition to the straightforward inter-pretation of the maximum achieved pressure as an indicator of theoverall energy release.

Unfortunately, a survey of the currently published reports showsthat the details available in the literature are insufficient formeaningful comparison of different pressure measurements orother experimental data for this type of experiments reported bydifferent research groups. It is suggested that future experiments inconfined vessel experiments specify the cell volume, mass of thepowder load, and the initial and final pressures obtained. Further-more, the type of gas present in the pressure cell and the pressureincrease obtained from the igniter itself (without the nano-composite powder) would be useful for meaningful comparisons ofthe data from different investigators.

Measurements describing combustion of individual reactivenanocomposite particles prepared by ARM were recently reported[212]. The powder-like material was incorporated into a liquid fuel(decane) and the produced slurry was aerosolized using anultrasonic nozzle. The aerosol jet was burned in a lifted laminarflame configuration and the combustion of nanocomposite parti-cles was studied optically. It was observed that micron-sizedaluminum particles could not be ignited in this configuration;however, the same or coarser size nanocomposite reactive parti-cles with bulk composition 2Bþ Ti ignited and burned completely.The ignited particles were also observed to disintegrate duringtheir combustion and continue burning as smaller fragments, asillustrated in Fig. 20. This type of combustion is very attractive forpractical applications where rapid burn rates are desired forparticles that are coarser and easier to work with thannanopowders.

Heterogeneous shock tube measurements were mentionedearlier as useful for finding ignition delays [196]. The samemeasurements are also useful for characterization of the ensuingcombustion of the aerosolized material. If the material can survivethe initial interaction with the incident and reflected shock waves

without being disintegrated, the information obtained fromdetailed optical measurements can be used to determine the burnrates, combustion temperatures, and identify some of the productspecies formed, all of which is critically important for developingmeaningful combustion models. The issues of the material survivalare effectively removed when combustion of n-Al in gaseousoxidizers is studied, as in Ref. [213] using the heterogeneous shocktube technique. The results are very interesting and showsubstantial differences between combustion of n-Al and relativelywell-characterized combustion of micron-sized Al particles. It wasfound that the burn time of n-Al particles decreases rapidly withthe increase in the ambient gas temperature. Furthermore,substantial reduction of the combustion time was also observed forn-Al at increased pressures. These combustion features are indic-ative of a kinetic burning regime for n-Al.

Comparison of combustion features of micron- and nanosized Alpowders was also presented in Ref. [214]. Experiments employeda premixed Bunsen-type flame with Al-laden flow fed from theBurner’s nozzle. Bimodal nano- and micron-sized Al particles wereused to produce laminar flames for which the speed and opticalstructure were studied and interpreted theoretically.

Finally, detailed kinetics of gas phase combustion reactions issuccessfully addressed in a few well-instrumented laser ignitionexperiments employing extremely high heating rates, as mentionedabove [191–194].

7.2. Performance in practical applications

Development of new reactive nanomaterials is driven by theirpotential applications in propellants, explosives, and pyrotechnics.Experimental validations of the performance of related practicalformulations are, therefore, very important for justifying furtherresearch and for guiding the material development efforts. Becausereactive nanomaterials became available only recently, the pub-lished results describing their performance are relatively scarce.Some of the first demonstrations of the effect of replacing micron-sized aluminum powder with its nanosized analog for metallizedsolid propellants were presented in Ref. [19]. Recently, moreexperimental studies, e.g., Refs. [215–219] and reviews [220,221]were published. It is generally established that the burn rate andpressure exponent of the propellant formulations with n-Al can beimproved compared to the standard metallized formulations usingmicron-sized powders. The improvement in the overall perfor-mance, however, depends on a broad range of the propellant char-acteristics, including oxygen balance of the formulation, particle sizedistributions of other components, e.g., AP, the type of oxidizer used,for example AP vs RDX and by other parameters, as discussed indetail in a recent review [221].

Applications of reactive nanomaterials in various pyrotechnicdevices have also been investigated recently [222,223]. Specifically,nanocomposite thermites were considered as components for lead-free primers and substantial potential for this application hasbeen recently demonstrated [223]. Combustion of nanocompositethermite was found to produce gas pressures that are inadequate forignition primers; however, the combination of nanocompositematerials with conventional or insensitive energetic formulations wasfound to present a promising approach for future practical devices.

Applications of reactive nanomaterials as additives to liquidfuels [212], to enhanced blast explosives [224–226], and othercompositions are also being actively explored and substantialprogress is anticipated in the near future. New application areas ofreactive nanomaterials are also being explored. Such reactivematerials are either expected to replace high explosives in specificapplications [227], or replace inert structural components of theweapons systems [228].

Fig. 20. An image showing a micro-explosion of a nanocomposite 2Bþ Ti particlefollowing its ignition [212]. The particle moves upward at about 50 cm/s and the image istaken with an open shutter camera. http://www.aiaa.org/content.cfm?pageid¼322&lupubid¼24.



8. Concluding remarks and future research

In the last 5–10 years, research has been very active in devel-opment and evaluation of many different types of reactive nano-materials. Several approaches, including those based on mixednanopowders, multilayered nanofoils, and three-dimensionalmicron-sized nanocomposite powders have been extensivelyinvestigated for a broad range of material compositions. It isanticipated that future research will focus on the development ofscalable and cost-effective manufacturing approaches in which therequired uniform and nanometer scale mixing between reactivecomponents can be maintained. Clearly, further development ofnew approaches, such as self-assembly will continue. Issues dealingwith handling and aging of reactive nanomaterials were addressedonly initially but more extensive work in that direction is antici-pated. In addition to conventional stability and processing charac-teristics, health effects of nanomaterials will need to be addressed,in particular for the materials using individual nanoparticles ascomponents. It should also be noted that most of the reactivecompositions described in the current literature are binary andoptimization of the materials characteristics is expected whenternary or more complex compositions are used. A scientificallysound approach for designing such complex reactive nanomaterialswill need to be developed to guide the future efforts.

Advances in the materials characterization techniques, such ashigh resolution TEM have been critically important for rapiddevelopment in the area of reactive nanomaterials. Contemporarymaterials characterization techniques are expected to remainimportant in furthering our understanding of reactive nano-material properties. Quantitative characterizations of the materialsmorphology and mixing uniformity are difficult and newapproaches tailored for reactive nanomaterials and utilizing thestate of the art materials characterization equipment, need to bedeveloped in the near future.

Some of the most important and least understood processes inreactive nanomaterials occur along the reactive interfaces, whichrepresent a substantial portion of the overall volume of suchcompositions. These processes are critical for understanding bothignition and aging. The reactive interfaces can be as simple aslayers of amorphous aluminum oxide separating aluminum fromgaseous oxidizer and as complicated as multilayer structuresforming in thermite or intermetallic compositions that can includeindividual metals, oxides, suboxides and alloys. Their morphologycan be as simple as planar in reactive nanofoils or as complicatedas multiple three-dimensional surfaces formed in the micron-sized nanocomposite particles produced by ARM. It is important tounderstand what the structure and formation mechanisms forsuch interfaces are, how these interfacial layers are different forthe materials with the same nominal compositions but producedusing different technique, how such layers evolve at elevatedtemperatures, and how heterogeneous reactions starting at verylow temperatures ultimately lead to ignition in reactive nano-materials. It is expected that a combination of experimental andtheoretical efforts will be needed to address these issues. Inparticular, the development of more sophisticated models,coupled with dramatic improvements in computational capabil-ities, is expected to enable molecular dynamics models to accu-rately predict crystallographic phase formation at such interfaces,their transport properties, and phase stability ranges. In addition,mechanical properties of nanoscaled oxide layers and relatedcomposite structures, such as in oxide-coated aluminum nano-particles will be predicted. Such predictions will be very importantfor understanding reaction mechanisms of the reactive nano-materials. Short-lived phenomena, e.g., oxide cracking uponaluminum melting, that can be easily overlooked by traditionalmaterial studies, e.g., using thermal analysis, would be possible to

predict theoretically. It will also be important to validate suchpredictions experimentally. Experimental studies of very thin andpoorly formed crystalline interfaces are very challenging andadvanced techniques, such as synchrotron radiation based X-raydiffraction and absorption measurements are expected to be usedfor such validations.

Finally, substantial advances are needed and expected in theignition and combustion testing of the reactive nanomaterials.Despite a large number of the related papers, current character-ization of these processes remains semi-quantitative. Well-documented, reproducible measurements aimed specifically atmeasuring reaction rates, energies, and rate-limiting mechanismsare needed. For such measurements, the effect of material density(as related to TMD) needs to be addressed systematically and poresizes and morphology need to be well quantified for the poroussamples. Time resolved pressure measurements for the fullyconfined samples combined with time resolved, optical spectros-copy appear to be quite promising. Experiments aimed to charac-terize the differences between the pressure driven (shock) andthermal ignition of reactive nanomaterials are also of great interest.The behavior of individual nanoscaled components duringcombustion, which can be accompanied with gas release anddisruption of the initial nanoscaled structure, is poorly understoodbut very important for development of adequate flame propagationmodels. Most importantly, the results of the experimental effortsneed to be presented in a way enabling one to compare materialsprepared using different manufacturing approaches to one another.Once such measurements and comparisons are available, devel-opment of related ignition and combustion models is expected tobecome more consistent between research groups and give moreconfidence in the predictions of such models.

Acknowledgement

This work was primarily funded by Defense Threat ReductionAgency (DTRA) and interest, encouragement, and support ofDrs. W. Wilson and S. Peiris of DTRA are very much appreciated.Additional funding was provided by RDECOM ARDEC, Picatinny(Mr. P. Redner). Multiple contributions from the graduate studentsand research staff members of the NJIT group are gratefullyacknowledged. Discussions with Drs. M. Schoenitz and M. A. Trunovare acknowledged in particular.

References

[1] Agrawal JP. Recent trends in high energy materials. Progress in Energy andCombustion Science 1998;24:1–30.

[2] Pagoria PF, Lee GS, Mitchell AR, Schmidt RD. A review of energetic materialssynthesis. Thermochimica Acta 2002;384:187–204.

[3] Tarver CM. Chemical kinetic modeling of HMX and TATB laser ignition tests.Journal of Energetic Materials 2004;22(2):93–107.

[4] Brill TB, James KJ. Kinetics and mechanisms of thermal decomposition ofnitroaromatic explosives. Chemical Reviews 1993;93(8):2667–92 (Wash-ington, DC, United States).

[5] Rozenband VI, Vaganova NI. A strength model of heterogeneous ignition ofmetal particles. Combustion and Flame 1992;88(1):113–8.

[6] Trunov MA, Schoenitz M, Zhu X, Dreizin EL. Effect of polymorphic phasetransformations in Al2O3 film on oxidation kinetics of aluminum powders.Combustion and Flame 2005;140(4):310–8.

[7] Trunov MA, Schoenitz M, Dreizin EL. Effect of polymorphic phase trans-formations in alumina layer on ignition of aluminium particles. CombustionTheory and Modelling 2006;10(4):603–23.

[8] Kazakov YV, Medvedev AE, Fedorov AV, Fomin VM. Mathematical modeling ofignition in dusty gases. Archivum Combustionis 1987;7(1–2):7–17.

[9] Price EW. Effect of multidimensional flamelets in composite propellantcombustion. Journal of Propulsion and Power 1995;11(4):717–28.

[10] Babuk VA, Vasilyev VA. Model of aluminum agglomerate evolution incombustion products of solid rocket propellant. Journal of Propulsion andPower 2002;18(4):814–23.

[11] Dreizin EL. Phase changes in metal combustion. Progress in Energy andCombustion Science 2000;26(1):57–78.



[12] Dreizin EL. Effect of phase changes on metal particle combustion processes.Combustion Explosion and Shock Waves 2003;39(6):92–6.

[13] Dreizin EL, Suslov AV, Trunov MA. General trends in metal particle hetero-geneous combustion. Combustion Science and Technology 1993;90:79–99.

[14] Dreizin EL, Suslov AV, Trunov MA. Temperature jumps in free metalparticle combustion. Combustion Science and Technology 1992;87:45–58.

[15] Molodetsky IE, Dreizin EL, Law CK. Evolution of particle temperature andinternal composition for zirconium burning in air. In: Proceedings of thecombustion institute (twenty-sixth symposium (Int’l) on combustion).Pittsburgh: The Combustion Institute; 1997. p. 1919–27.

[16] Molodetsky IE, Dreizin EL, Vicenzi EP, Law CK. Phases of titanium combustionin air. Combustion and Flame 1998;112:522–32.

[17] Danen WC, Martin JA. Energetic composites. US Patent 5,266,132; November30, 1993.

[18] Rossi C, Zhang K, Esteve D, Alphonse P, Tailhades P, Vahlas C. Nanoenergeticmaterials for MEMS: a review. Journal of Microelectromechanical Systems2007;16(4):919–31.

[19] Kuo KK, Risha GA, Evans BJ, Boyer E. Potential usage of energetic nano-sizedpowders for combustion and rocket propulsion. In: Materials researchsociety symposium proceedings, 800 (synthesis, characterization and prop-erties of energetic/reactive nanomaterials). Materials Research Society; 2003.p. 3–14.

[20] Zelinskii VY, Yavorovskii NA, Proskurovskaya LT, Davydovich VI. Structuralstate of aluminum particles prepared by electric explosion. Fizika i KhimiyaObrabotki Materialov 1984;1:57–9 (in Russian).

[21] Yavorovsky NA. Method of production of highly dispersed powders of inor-ganic materials. Patent of Russian Federation No. 2048277; 1995.

[22] Yavorovsky NA, Davydovich VI, Bil’ BA. Apparatus for production of highlydispersed powders of inorganic materials by electric explosion and reactorfor explosion of metallic component. Patent of Russian Federation No.2048278; 1995.

[23] Ivanov YF, Osmonoliev MN, Sedoi VS, Arkhipov VA, Bondarchuk SS,Vorozhtsov AB, et al. Productions of ultra-fine powders and their use in highenergetic compositions. Propellants, Explosives, Pyrotechnics 2003;28(6):319–33.

[24] Jiang W, Yatsui K. Pulsed wire discharge for nanosize powder synthesis. IEEETransactions on Plasma Science 1998;26(5):1498–501.

[25] Kwon Y-S, Jung Y-H, Yavorovsky NA, Illyn AP, Kim J-S. Ultra-fine powder bywire explosion method. Scripta Materialia 2001;44:2247–51.

[26] Sarathi R, Sindhu TK, Chakravarthy SR. Generation of nano aluminiumpowder through wire explosion process and its characterization. MaterialsCharacterization 2007;58:148–55.

[27] Ivanov GV, Tepper F. ‘‘Activated’’ aluminum as a stored energy source forpropellants. In: Kuo KK, editor. Challenges in propellants and combustion 100years after nobel. New York: Begell House; 1997. p. 636–45.

[28] Tepper F. Nanosize powders produced by electro-explosion of wire and theirpotential applications. Powder Metallurgy 2000;43(4):320–2.

[29] Tepper F. Electro-explosion of wire produces nanosize metals. Metal PowderReport 1998;53(6):31–3.

[30] Kwon YS, Ilyin AP, Nazarenko OB. Electric explosion of wires in multicom-ponent reactionary liquid ambiences as method for producing nanopowderof complex composition. In: Proceedings – ninth Russian–Korean interna-tional symposium on science and technology, KORUS-2005 1, art. no.1507688; 2005. p. 211–3.

[31] Wang Q, Yang H, Shi J, Zou G. Preparation and characterization of nano-crystalline powders of Cu–Zn alloy by wire electrical explosion method.Materials Science and Engineering 2001;A307:190–4.

[32] Romanova VM, Pikuz SA, Ter-Oganesyan AE, Mingaleev AR, Shelkovenko TA.Nanosecond electrical explosion of micron diameter wire. CzechoslovakJournal of Physics 2006;56(Suppl. B):B349–56.

[33] Sedoi VS, Mesyats GA, Oreshkin VI, Valevich VV, Chemezova LI. The currentdensity and the specific energy input in fast electrical explosion. IEEETransactions on Plasma Science 1999;27(4):845–50.

[34] Chemezova LI, Mesyats GA, Sedoi VS, Semin BN, Valevich VV. The integral ofspecific current action and the specific energy input under fast electricalexplosion. In: IEEE eighteenth interntional symposium on discharges andelectrical insulation in vacuum; 1998. p. 48–51. Eindhoven.

[35] Kwon Y-S, Gromov AA, Ilyin AP, Rim G-H. Passivation process for superfinealuminum powders obtained by electrical explosion of wires. AppliedSurface Science 2003;211:57–67.

[36] Gromov AA, Forter-Barth U, Teipel U. Aluminum nanopowders produced byelectrical explosion of wires and passivated by non-inert coatings: charac-terisation and reactivity with air and water. Powder Technology 2006;164:111–5.

[37] Granqvist CG, Buhrman RA. Ultrafine metal particles. Journal of AppliedPhysics 1976;47(5):2200–19.

[38] Puszynski JA. Formation, characterization, and reactivity of nanoenergeticpowders. In: Proceedings of the twenty-ninth international pyrotechnicsseminar. Defense Science & Technology Organization, Pyrotechnics Group;2002. p. 191–202.

[39] Schefflan R, Kovenklioglu S, Kalyon D, Mezger M, Leng M. Formation ofaluminum nanoparticles upon condensation from vapor phase for energeticapplications. Journal of Energetic Materials 2006;24(2):141–56.

[40] Pivkina A, Ivanov D, Frolov Y, Mudretsova S, Nickolskaya A, Schoonman J.Plasma synthesized nano-aluminum powders: structure, thermal properties

and combustion behavior. Journal of Thermal Analysis and Calorimetry2006;86(3):733–8.

[41] Munz RJ, Addona T, da Cruz A-C. Application of transferred arcs to the productionof nanoparticles. Pure and Applied Chemistry 1999;71(10):1889–97.

[42] Phillips J, Perry WL, Kroenke WJ. Manufacture of spherical metal nano-particles by plasma heating of precursor aerosol for vapor condensation. USPatent 6689192 B1 20040210; 2004.

[43] Champion Y. Evaporation and condensation for metallic nanopowders.Annales de Chimie (Cachan, France) 2006;31(3):281–94.

[44] Miziolek AW. Nanoenergetics: an emerging technology area of nationalimportance. The AMPTIAC Newsletter 2002;6(1):43–8.

[45] Boulos MI, Jurewicz J, Guo J. Induction plasma synthesis of nanopowders. USPatent Application 20070029291; 2007.

[46] Choi SD, Byun GS, Yang CJ, Son YG, Park EB, Lee YJ, et al. Method for preparingaluminum nanopowders by electromagnetic wave plasma device. Repub.Korean Kongkae Taeho Kongbo 2008.

[47] Panda S, Pratsinis SE. Modeling the synthesis of aluminum particles byevaporation–condensation in an aerosol flow reactor. NanoStructuredMaterials 1995;5(7/8):755–67.

[48] Puszynski JA, Jayaraman S, Bichay M, Carpenter G, Carpenter P. Formationand reactivity of nanosized aluminum powders. In: Proceedings of thefourth world congress on powder technology; 2002. Sydney, Australia,Paper 164.

[49] Lehtinen KEJ, Zachariah MR. Energy accumulation in nanoparticle collisionand coalescence processes. Aerosol Science 2002;33:357–68.

[50] Mukherjee D, Sonwane CG, Zachariah MR. Kinetic Monte Carlo simulation ofthe effect of coalescence energy release on the size and shape evolution ofnanoparticles grown as an aerosol. Journal of Chemical Physics2003;119:3391–404.

[51] Puszynski JA, Bulian CJ, Swiatkiewicz JJ. The effect of nanopowder attributeson reaction mechanism and ignition sensitivity of nanothermites. MaterialsResearch Society Symposium Proceedings 2006;896:147–58.

[52] Park K, Lee D, Rai A, Mukherjee D, Zachariah MR. Size-resolved kineticmeasurements of aluminum nanoparticle oxidation with single particlemass spectrometry. Journal of Physical Chemistry B 2005;109(15):7290–9.

[53] Mukherjee D, Rai A, Zachariah MR. Quantitative laser-induced breakdownspectroscopy for aerosols via internal calibration: application to the oxidativecoating of aluminum nanoparticles. Aerosol Science 2006;37:677–95.

[54] Dong XL, Zhang ZD, Jin SR, Kim BK. Carbon-coated Fe–Co(C) nanocapsulesprepared by arc discharge in methane. Journal of Applied Physics1999;86(12):6701–6.

[55] Zhang ZD, Zheng JG, Skorvanek I, Wen GH, Kovac J, Wang FW, et al. Shell/corestructure and magnetic properties of carbon-coated Fe–Co(C) nanocapsules.Journal of Physics: Condensed Matter 2001;13(9):1921–9.

[56] Ermoline A, Schoenitz M, Dreizin E, Yao N. Production of carbon-coatedaluminium nanopowders in pulsed microarc discharge. Nanotechnology2002;13:638–43.

[57] Park K, Rai A, Zachariah MR. Characterizing the coating and size-resolvedoxidative stability of carbon-coated aluminum nanoparticles by single-particlemass-spectrometry. Journal of Nanoparticle Research 2006;8:455–64.

[58] Rosner DE. Flame synthesis of valuable nanoparticles: recent progress/current needs in areas of rate laws, population dynamics, and characteriza-tion. Industrial/Engineering Chem-Research 2005;44:6045–55.

[59] Calcote HF, Felder W. A new gas-phase combustion synthesis process for puremetals, alloys, and ceramics. In: Twenty-fourth symposium (international) oncombustion. Pittsburgh: The Combustion Institute; 1992. p. 1869–76.

[60] Wooldridge MS. Gas-phase combustion synthesis of particles. Progress inEnergy and Combustion Science 1998;24(1):63–87.

[61] Steffens KL, Zachariah MR, DuFaux DP, Axelbaum RL. Optical and modelingstudies of sodium/halide reactions for the formation of titanium and boronnanoparticles. Chemistry of Materials 1996;8(8):1871–80.

[62] Axelbaum RL. Synthesis of stable metal and non-oxide ceramic nanoparticlesin sodium/halide flames. Powder Metallurgy 2000;43(4):323–5.

[63] Higa KT, Johnson CE, Hollins RA. Preparation of fine aluminum powders bysolution methods. United States Patent 6,179,899; 2001.

[64] Foley TJ, Johnson CE, Higa KT. Inhibition of oxide formation on aluminumnanoparticles by transition metal coating. Chemistry of Materials2005;17(16):4086–91.

[65] Jouet RJ, Warren AD, Rosenberg DM, Bellitto VJ, Park K, Zachariah MR. Surfacepassivation of bare aluminum nanoparticles using perfluoroalkyl carboxylicacids. Chemistry of Materials 2005;17(11):2987–96.

[66] Jouet RJ, Carney JR, Granholm RH, Sandusky HW, Warren AD. Preparationand reactivity analysis of novel perfluoroalkyl coated aluminium nano-composites. Materials Science and Technology 2006;22(4):422–9.

[67] Gao Y. Method for preparing nanoscale nickel-coated aluminum powder.Chinese Patent CN 101041180 A 20070926; 2007.

[68] Harruff BA, Smith MJ, Guliants EA, Bunker CE. Synthesis and characterizationof aluminum nanostructures prepared via the sonochemical method.Greenville, SC: South-Eastern Regional ACS Meeting; October, 2007.

[69] Zhang K, Fan J-H, Huang Y-H, Tan Y. Content and activity analysis ofaluminum powder in nano-aluminum/PS microcapsules. Hanneng Cailiao/Chinese Journal of Energetic Materials 2007;15(5):482–4.

[70] Wilson CA, Goldstein DN, McCormick JA, Weimer AW, George SM. Tungstenatomic layer deposition on cobalt nanoparticles. Journal of Vacuum Scienceand Technology A: Vacuum, Surfaces and Films 2008;26(3):430–7.



[71] Son S.F. Personal communication; 2008.[72] Sun W, Puzas JE, Sheu T-J, Liu X, Fauchet PM. Nano- to microscale porous

silicon as a cell interface for bone-tissue engineering. Advanced Materials2007;19(7):921–4.

[73] El-Shall MS, Abdelsayed V, Pithawalla YB, Alsharaeh E, Deevi SC. Vapor phasegrowth and assembly of metallic, intermetallic, carbon, and silicon nano-particle. Journal of Physical Chemistry B 2003;107(13):2882–6.

[74] Makimura T, Mizuta T, Murakami K. Laser ablation synthesis of hydrogenatedsilicon nanoparticles with green photoluminescence in the gas phase. Japa-nese Journal of Applied Physics, Part 2: Letters 2002;41(2A):L144–6.

[75] Dang H, Swihart MT. Computational modeling of silicon nanoparticlesynthesis. ECS Transactions 2007;2(7):255–65.

[76] Giesen B, Wiggers H, Kowalik A, Roth P. Formation of Si-nanoparticles ina microwave reactor: comparison between experiments and modeling.Journal of Nanoparticle Research 2005;7(1):29–41.

[77] Khan MH, Cole JA, Taube JA. Method and apparatus for producing nano-particles of molybdenum oxide. US Patent, International Publication NumberWO 02/38239 A1; 2002.

[78] Khan MH, Cole JA, Taube JA. Molybdenum oxide nanoparticles. US Patent0120950 A1; 2006.

[79] Mitra S, Sridharan K, Unnam J, Ghosh K. Synthesis of nanometal oxides andnanometals using hot-wire and thermal CVD. Thin Solid Films2008;516:798–802.

[80] Pratsinis SE. Flame aerosol synthesis of ceramic powders. Progress in Energyand Combustion Science 1998;24(3):197–219.

[81] Skandan G, Chen Y-J, Glumac N, Kear BH. Synthesis of oxide nanoparticles inlow pressure flames. Nanostructured Materials 1999;11(2):149–58.

[82] Xu F, Tse SD, Al-Sharab JF, Kear BH. Flame synthesis of aligned tungsten oxidenanowires. Applied Physics Letters 2006;88(24). art. no. 243115.

[83] Zhao H, Liu X, Tse SD. Control of nanoparticle size and agglomerationthrough electric-field-enhanced flame synthesis. Journal of NanoparticleResearch 2007:1–17.

[84] Vollath D, Szabo DV, Haußelt J. Synthesis and properties of ceramic nano-particles and nanocomposites. Journal of the European Ceramic Society1997;17(11):1317–24.

[85] Kleinwechter H, Janzen C, Knipping J, Wiggers H, Roth P. Formation andproperties of ZnO nano-particles from gas phase synthesis processes. Journalof Materials Science 2002;37(20):4349–60.

[86] Vollath D, Szabo DV. Coated nanoparticles: a new way to improved nano-composites. Journal of Nanoparticle Research 1999;1(2):235–42.

[87] Vollath D. Plasma synthesis of nanopowders. Journal of NanoparticleResearch, 2008;10(Suppl. 1):39–57.

[88] Prakash A, McCormick AV, Zachariah MR. Tuning the reactivity of energeticnanoparticles by creation of a core–shell nanostructure. Nano Letters2005;5(7):1357–60.

[89] Tillotson TM, Gash AE, Simpson RL, Hrubesh LW, Satcher Jr JH, Poco JF.Nanostructured energetic materials using sol-gel methodologies. Journal ofNon-Crystalline Solids 2001;285(1–3):338–45.

[90] Gash AE, Tillotson TM, Satcher Jr JH, Poco JF, Hrubesh LW, Simpson RL. Use ofepoxides in the sol-gel synthesis of porous iron(III) oxide monoliths fromFe(III) salts. Chemistry of Materials 2001;13(3):999–1007.

[91] Gash AE, Tillotson TM, Satcher Jr JH, Hrubesh LW, Simpson RL. New sol-gel synthetic route to transition and main-group metal oxide aerogelsusing inorganic salt precursors. Journal of Non-Crystalline Solids2001;285(1–3):22–8.

[92] Perry WL, Smith BL, Bulian CJ, Busse JR, Macomber CS, Dye RC, et al. Nano-scale tungsten oxides for metastable intermolecular composites. Propellants,Explosives, Pyrotechnics 2004;29(2):99–105.

[93] Danen WC, Jorgensen BS, Busse JR, Ferris MJ, Smith BL. Los Alamos nano-energetic metastable intermolecular composite (super thermite) program.Two hundred and twenty-first ACS National Meeting, San Diego, CA, UnitedStates, Abstracts of Papers; 2001.

[94] Perry WL, Tappan BC, Reardon BL, Sanders VE, Son SF. Energy release character-istics of the nanoscale aluminum–tungsten oxide hydrate metastable intermo-lecular composite. Journal of Applied Physics 2007;101(6):064313/1–064313/5.

[95] Son SF, Hiskey MA, Naud DL, Busse JR, Asay BW. Lead-free electric matches. In:Proceedings of the twenty-ninth international pyrotechnics seminar. DefenceScience & Technology Organisation, Pyrotechnics Group; 2002. p. 871–7.

[96] Prakash A, McCormick AV, Zachariah MR. Synthesis and reactivity of a super-reactive metastable intermolecular composite formulation of Al/KMnO4.Advanced Materials 2005;17(7):900–3.

[97] Walter KC, Pesiri DR, Wilson DE. Manufacturing and performance of nano-metric Al/MoO3 energetic materials. Journal of Propulsion and Power2007;23(4):645–50.

[98] Pantoya ML, Granier JJ. Combustion behavior of highly energetic thermites:nano versus micron composites. Propellants, Explosives, Pyrotechnics2005;30(1):53–62.

[99] Plantier KB, Pantoya ML, Gash AE. Combustion wave speeds of nano-composite Al/Fe2O3: the effects of Fe2O3 particle synthesis technique.Combustion and Flame 2005;140:299–309.

[100] Sanders VE, Asay BW, Foley TJ, Tappan BC, Pacheco AN, Son SF. Reactionpropagation of four nanoscale energetic composites (Al–MoO3, Al–WO3,Al–CuO, and Bi2O3). Journal of Propulsion and Power 2007;23(4):707–14.

[101] Puszynski JA, Bulian CJ, Swiatkiewicz JJ. Processing and ignition character-istics of aluminum–bismuth trioxide nanothermite system. Journal ofPropulsion and Power 2007;23(4):698–706.

[102] Wei D, Dave R, Pfeffer R. Mixing and characterization of nanosized powders:an assessment of different techniques. Journal of Nanoparticle Research2002;4(1–2):21–41.

[103] Marioth E, Kroeber H, Loebbecke S, Krause H. Comparison of nanoparticulatethermite mixtures formed by conventional and supercritical fluid processes.In: International annual conference of ICT, thirty-seventh (energetic mate-rials). Fraunhofer-Institut fuer Chemische Technologie; 2006. p. 113/1–113/7.

[104] Gash AE, Satcher Jr JH, Simpson RL, Clapsaddle BJ. Nanostructured energeticmaterials with sol-gel methods. Materials Research Society SymposiumProceedings 2003;800:55–66.

[105] Clapsaddle BJ, Zhao L, Gash AE, Satcher Jr JH, Shea KJ, Pantoya ML, et al.Synthesis and characterization of mixed metal oxide nanocomposite ener-getic materials. Materials Research Society Symposium Proceedings2003;800:91–6.

[106] Gangopadhyay S, Shende R, Subramanian S, Gangopadhyay K, Hasan S.Ordered nanoenergetic thermites as self-assembled nanostructures withtunable combustion properties. U.S. Pat. Appl. Publ. US 2007095445 A120070503; 2007.

[107] Subramaniam S, Hasan S, Bhattacharya S, Gao Y, Apperson S, Hossain M, et al.Self-assembled ordered energetic composites of CuO nanorods and nano-wells and Al nanoparticles with high burn rates. Materials Research SocietySymposium Proceedings 2006;896:9–14.

[108] Apperson SR, Shende V, Subramanian S, Tappmeyer D, Gangopadhyay S,Chen Z, et al. Generation of fast propagating combustion and shock waveswith copper oxide/aluminum nanothermite composites. Applied PhysicsLetters 2007;91. Article No. 243109.

[109] Mehendale B, Shende R, Subramanian S, Gangopadhyay S, Redner P,Kapoor D, et al. Nanoenergetic composite of mesoporous iron oxide andaluminum nanoparticles. Journal of Energetic Materials 2006;24:341–60.

[110] Ma E, Thompson CV, Clevenger LA, Tu KN. Self-propagating explosivereactions in Al/Ni multilayer thin films. Applied Physics Letter 1990;57(12):1262–4.

[111] Barbee TW, Weihs T. Ignitable heterogeneous stratified structure for thepropagation of an internal exothermic chemical reaction along an expandingwavefront and method of making the same. US Patent 5,538,795, 1996.

[112] Wang J, Besnoin E, Duckham A, Spey SJ, Reiss ME, Knio OM, et al. Room-temperature soldering with nanostructured foils. Applied Physics Letters2003;83(19):3987–9.

[113] Wang J, Besnoin E, Duckham A, Spey SJ, Reiss ME, Knio OM, et al. Joining ofstainless-steel specimens with nanostructured Al/Ni foils. Journal of AppliedPhysics 2004;95(1):248–56.

[114] Duckham A, Spey SJ, Wang J, Reiss ME, Weihs TP, Besnoin E, et al. Reactivenanostructured foil used as a heat source for joining titanium. Journal ofApplied Physics 2004;96(4):2336–42.

[115] Gavens AJ, Van Heerden D, Mann AB, Reiss ME, Weihs TP. Effect of inter-mixing on self-propagating exothermic reactions in Al/Ni nanolaminate foils.Journal of Applied Physics 2000;87(3):1255–63.

[116] Blobaum KJ, Reiss ME, Plitzko JM, Weihs TP. Deposition and characterizationof a self-propagating CuOx/Al thermite reaction in a multilayer foil geometry.Journal of Applied Physics 2003;94(5):2915–22.

[117] Blobaum KJ, Wagner AJ, Plitzko JM, Van Heerden D, Fairbrother DH, Weihs TP.Investigating the reaction path and growth kinetics in CuOx/Al multilayerfoils. Journal of Applied Physics 2003;94(5):2923–9.

[118] Dreizin EL, Schoenitz M. Nano-composite energetic powders prepared byarrested reactive milling. US Patent Application 20060053970; published inMarch 2006.

[119] Schoenitz M, Ward T, Dreizin EL. Preparation of energetic metastable nano-composite materials by arrested reactive milling. Materials Research SocietyProceeding 2004;800:AA2.6.1–2.6.6.

[120] Schoenitz M, Ward TS, Dreizin EL. Fully dense nano-composite energeticpowders prepared by arrested reactive milling. Proceedings of theCombustion Institute 2005;30:2071–8.

[121] Umbrajkar S, Trunov MA, Schoenitz M, Dreizin EL, Broad R. Arrested reactivemilling synthesis and characterization of sodium-nitrate based reactivecomposites. Propellants, Explosives, Pyrotechnics 2007;32(1):32–41.

[122] Dreizin EL, Schoenitz M, Shoshin YL, Trunov MA. Highly-energetic nano-composite powders produced by arrested reactive milling. In: Internationalannual conference of ICT 2005, thiry-sixth (Energetic Materials). p. 138/1–138/12.

[123] Umbrajkar SM, Schoenitz M, Dreizin EL. Control of structural refinement andcomposition in Al–MoO3 nanocomposites prepared by arrested reactivemilling. Propellants, Explosives, Pyrotechnics 2006;31(5):382–9.

[124] Umbrajkar SM, Seshadri S, Schoenitz M, Hoffmann VK, Dreizin EL.Aluminum-rich Al–MoO3 nanocomposite powders prepared by arrestedreactive milling. Journal of Propulsion and Power 2008;24(2):192–8.

[125] Ward TS, Chen W, Schoenitz M, Dave RN, Dreizin EL. A study of mechanicalalloying processes using reactive milling and discrete element modeling.Acta Materialia 2005;53:2909–18.

[126] Suryanarayana C. Mechanical alloying and milling. Progress in MaterialsScience 2001;46(1–2):1–184.

[127] Dolgoborodov AY, Makhov MN, Streletskij AN, Kolbanev IV, Gogulya MF,Fortov VE. About opportunity for detonation in mechanically activatedcomposite material aluminum–fluoroplastic. Khimicheskaya Fizika2004;23(9):85–9.

[128] Dolgoborodov AY, Makhov MN, Kolbanev IV, Streletskii AN, Fortov VE.Detonation in an aluminum–Teflon mixture. JETP Letters 2005;81(7):311–4.



[129] Dolgoborodov AY, Streletskii AN, Makhov MN, Kolbanev IV, Fortov VE.Explosive compositions based on mechanically activated metal-oxidizermixtures. Khimicheskaya Fizika 2007;26(12):40–5.

[130] Tanaka T, Hara S. Thermodynamic evaluation of binary phase diagrams ofsmall particle systems. Zeitschrift fuer Metallkunde/Materials Research andAdvanced Techniques 2001;92(5):467–72.

[131] Goldstein AN, Echer CM, Alivisatos AP. Melting in semiconductor nano-crystals. Science 1992;256(5062):1425–7.

[132] Lai SL, Guo JY, Petrova V, Ramanath G, Allen LH. Size-dependent meltingproperties of small tin particles: nanocalorimetric measurements. PhysicalReview Letters 1996;77(1):99–102.

[133] Allen GL, Bayles RA, Gile WW, Jesser WA. Small particle melting of puremetals. Thin Solid Films 1986;144(2):297–308.

[134] Farrell HH, Van Siclen CD. Binding energy, vapor pressure, and melting pointof semiconductor nanoparticles. Journal of Vacuum Science & Technology, B:Microelectronics and Nanometer Structures – Processing, Measurement, andPhenomena 2007;25(4):1441–7.

[135] Misra A, Kung H. Deformation behavior of nanostructured metallic multi-layers. Advanced Engineering Materials 2001;3(4):217–22.

[136] Singhal A, Skandan G, Wang A, Glumac N, Kear BH, Hunt RD. On nanoparticleaggregation during vapor phase synthesis. Nanostructured Materials1999;11(4):545–52.

[137] Kindratenko VV, Van Espen PJM, Treiger BA, Van Grieken RE. Fractal dimen-sional classification of aerosol particles by computer-controlled scanningelectron microscopy. Environmental Science & Technology 1994;28:2197–202.

[138] Colbeck I, Atkinson B, Johart Y. The morphology and optical properties of sootproduced by different fuels. Journal of Aerosol Science 1997;28:715–23.

[139] Chakrabarty RK, Moosmuller H, Garro MA, Arnott WP, Walker J, Susott RA,et al. Light scattering and absorption by fractal-like carbonaceous chainaggregates: comparison of theories and experiment. Journal of GeophysicalResearch D: Atmospheres 2006;111(7). art. no. D07204.

[140] Moore K, Pantoya ML, Son SF. Combustion behaviors resulting from bimodalaluminum size distributions in thermites. Journal of Propulsion and Power2007;23(1):181–5.

[141] Prentice D, Pantoya ML, Clapsaddle BJ. Effect of nanocomposite synthesis onthe combustion performance of a ternary thermite. Journal of PhysicalChemistry B 2005;109:20180–5.

[142] Patterson AL. The Scherrer formula for X-ray particle size determination.Physical Review 1939;56(10):978–82.

[143] Hall BD, Zanchet D, Ugarte D. Estimating nanoparticle size from diffractionmeasurements. Journal of Applied Crystallography 2000;33(6):1335–41.

[144] Johnson CE, Fallis S, Chafin AP, Groshens TJ, Higa KT, Ismail IMK, et al.Characterization of nanometer- to micron-sized aluminum powders: sizedistribution from thermogravimetric analysis. Journal of Propulsion andPower 2007;23(4):669–82.

[145] Levin I, Brandon D. Metastable alumina polymorphs: crystal structures andtransitions sequences. Journal of the American Ceramic Society1998;81(8):1995–2012.

[146] Trunov MA, Umbrajkar SM, Schoenitz M, Mang JT, Dreizin EL. Oxidation andmelting of aluminum nanopowders. Journal of Physical Chemistry B2006;110(26):13094–9.

[147] Mang JT, Hjelm RP, Son SF, Peterson PD, Jorgensen BS. Characterization ofcomponents of nano-energetics by small-angle scattering techniques. Journalof Materials Research 2007;22(7):1907–20.

[148] Borchert H, Shevchenko EV, Robert A, Mekis I, Kornowski A, Grubel G,et al. Determination of nanocrystal sizes: a comparison of TEM, SAXS, andXRD studies of highly monodisperse CoPt3 particles. Langmuir 2005;21(5):1931–6.

[149] Rufino B, Boulc’h F, Coulet M-V, Lacroix G, Denoyel R. Influence of particlessize on thermal properties of aluminium powder. Acta Materialia2007;55:2815–27.

[150] Ramaswamy AL, Kaste P. A ‘‘Nanovision’’ of the physiochemical phenomenaoccurring in nanoparticles of aluminum. Energetic Materials 2005;23:1–25.

[151] Fedotova TD, Glotov OG, Zarko VE. Peculiarities of chemical analysis of ultrafine aluminum. In: Thirty-sixth international annual conference of ICTcombined with the thiry-second international pyrotechnics seminar,Karlsruhe, Germany, June 28–July 1, 2005. p. 147/1.

[152] Fedotova TD, Glotov OG, Zarko VE. Chemical analysis of aluminum asa propellant ingredient and determination of aluminum and aluminumnitride in condensed combustion products. Propellants, Explosives,Pyrotechnics 2000;25:325–32.

[153] Fedotova TD, Glotov OG, Zarko VE. Application of cerimetric methods fordetermining the metallic aluminum content in ultrafine aluminum powders.Propellants, Explosives, Pyrotechnics 2007;32(2):160–4.

[154] Alcoutlabi M, McKenna1 GB. Effects of confinement on material behaviour atthe nanometre size scale. Journal of Physics: Condensed Matter2005;17:R461–524.

[155] Sun J, Simon SL. The melting behavior of aluminum nanoparticles. Thermo-chimica Acta 2007;463(1–2):32–40.

[156] Hunt EM, Plantier KB, Pantoya ML. Nano-scale reactants in the self-propa-gating high-temperature synthesis of nickel aluminide. Acta Materialia2004;52(11):3183–91.

[157] Hunt EM, Pantoya ML, Jouet RJ. Combustion synthesis of metallic foams fromnanocomposite reactants. Intermetallics 2006;14(6):620–9.

[158] Granier JJ, Pantoya ML. Laser ignition of nanocomposite thermites.Combustion and Flame 2004;138(4):373–83.

[159] Reiss H, Wilson IB. The effect of surface on melting point. Journal of ColloidScience 1948;3:551–61.

[160] Jiang Q, Yang CC, Li JC. Melting enthalpy depression of nanocrystals. MaterialsLetters 2002;56:1019–21.

[161] Liang LH, Li JC, Jiang Q. Size-dependent melting depression and latticecontraction of Bi nanocrystals. Physica B 2003;334:49–53.

[162] Zhao M, Jiang Q. Melting and surface melting of low-dimensional in crystals.Solid State Communications 2004;130:37–9.

[163] Eckert J, Holzer JC, Ahn CC, Fu Z, Johnson WL. Melting behavior of nano-crystalline aluminum powders. Nanostructured Materials 1993;2:407–13.

[164] Navrotsky A. Thermochemistry of nanomaterials. Reviews in Mineralogy andGeochemistry 2001;44:73–103.

[165] Navrotsky A. Energetics of nanoparticle oxides: interplay between surfaceenergy and polymorphism. Geochemical Transactions 2003;4(6):34–7.

[166] Jeurgens LPH, Sloof WG, Tichelaar FD, Mittemeijer EJ. Thermodynamicstability of amorphous oxide films on metals: application to aluminum oxidefilms on aluminum substrates. Physical Review B 2000;62(7):4707–19.

[167] Levinsky YV. P-T-X handbook, pressure dependent phase diagrams of binaryalloys. ASM Intl; 1998.

[168] Mench MM, Kuo KK, Yeh CL, Lu YC. Comparison of thermal behavior ofregular and ultra-fine aluminum powders (Alex) made from plasma explo-sion process. Combustion Science and Technology 1998;135:269–92.

[169] Jones DEG, Brousseau P, Fouchard RC, Turcotte AM, Kwok QSM. Thermalcharacterization of passivated nanometer size aluminium powders. Journalof Thermal Analysis and Calorimetry 2000;61(3):805–18.

[170] Sun J, Pantoya ML, Simon SL. Dependence of size and size distribution onreactivity of aluminum nanoparticles in reactions with oxygen and MoO3.Thermochimica Acta 2006;444(2):117–27.

[171] Rai A, Lee D, Park K, Zachariah MR. Importance of phase change of aluminumin oxidation of aluminum nanoparticles. Journal Physical Chemistry B2004;108:14793–5.

[172] Rai A, Park K, Zhou L, Zachariah MR. Understanding the mechanism ofaluminium nanoparticle oxidation. Combustion Theory and Modelling2006;10:843–59.

[173] Jeurgens LPH, Sloof WG, Tichelaar FD, Mittemeijer EJ. Structure andmorphology of aluminum-oxide films formed by thermal oxidation ofaluminum. Thin Solid Films 2002;418:89–101.

[174] Dwivedi RK, Gowda G. Thermal stability of aluminum oxides prepared fromgel. Journal of Materials Science Letters 1985;4:331–43.

[175] Jeurgens LPH, Sloof WG, Tichelaar FD, Mittemeijer EJ. Growth kinetics andmechanisms of aluminum oxide films formed by thermal oxidation ofaluminum. Journal of Applied Physics 2002;92:1649–56.

[176] Riano OA, Wadsworth J, Sherby OD. Deformation of fine-grained alumina bygrain boundary sliding accommodated by slip. Acta Materialia2003;51:3617–34.

[177] Umbrajkar SM, Schoenitz M, Dreizin EL. Exothermic reactions in Al–CuOnanocomposites. Thermochimica Acta 2006;451:34–43.

[178] Schoenitz M, Umbrajkar S, Dreizin EL. Kinetic analysis of thermite reactionsin Al–MoO3 nanocomposites. Journal of Propulsion and Power2007;23(4):683–7.

[179] Mehrer H. Diffusion in intermetallics. Materials Transactions, JIM1996;37(6):1259–80.

[180] Laurila T, Vuorinen V, Kivilahti JK. Interfacial reactions between lead-freesolders and common base materials. Materials Science and Engineering R:Reports 2005;49(1–2):1–60.

[181] Morsi K. Review: reaction synthesis processing of Ni–Al intermetallic mate-rials. Materials Science and Engineering A 2001;299(1–2):1–15.

[182] Blobaum KJ, Van Heerden D, Gavens AJ, Weihs TP. Al/Ni formation reactions:characterization of the metastable Al9Ni2 phase and analysis of its formation.Acta Materialia 2003;51:3871–84.

[183] Dimitriou P, Hlavacek V, Valone SM, Behrens RG, Hansen GP, Margrave JL.Laser-induced ignition in solid-state combustion. AIChE J 1989;35(7):1085–96.

[184] Zakharov NS, Kazarin SA. Ignition and combustion of condensed materialsduring continuous carbon dioxide laser irradiation. Izvestiya Akademii NaukSSSR, Seriya Fizicheskaya 1991;55(6):1198–201.

[185] Fetherolf BL, Liiva PM, Hsieh WH, Kuo KK. Carbon dioxide laser ignitionbehavior of several pyrotechnic mixtures. In: Proceedings of the sixteenthinternational pyrotechnics seminar; 1991. p. 675–90.

[186] Kuo KK, Kim JU, Fetherolf BL, Torikai T. Preignition dynamics of RDX-basedenergetic materials under CO2 laser heating. Combustion and Flame1993;95(4):351–61.

[187] Ahmad SR, Russell DA, Leach CJ. Studies into laser ignition of unconfinedpropellants. Propellants, Explosives, Pyrotechnics 2001;26(5):235–45.

[188] Ali AN, Son SF, Asay BW, Decroix ME, Brewster MQ. High-irradiance laser igni-tion of explosives. Combustion Science and Technology 2003;175(8):1551–71.

[189] Yang Y, Wang S, Sun Z, Dlott DD. Near-infrared and visible absorptionspectroscopy of nano-energetic materials containing aluminum and boron.Propellants, Explosives, Pyrotechnics 2005;30(3):171–7.

[190] Begley SM, Brewster MQ. Radiative properties of MoO3 and Al nanopowdersfrom light-scattering measurements. Journal of Heat Transfer2007;129(5):624–33.

[191] Yang Y, Sun Z, Wang S, Dlott DD. Fast spectroscopy of laser-initiated nano-energetic materials. Journal of Physical Chemistry B 2003;107(19):4485–93.

[192] Wang S, Yang YYH, Dlott DD. Dynamical effects of the oxide layer inaluminum nanoenergetic materials. Propellants, Explosives, Pyrotechnics2005;30(2):148–55.



[193] Zamkov MA, Conner RW, Dlott DD. Ultrafast chemistry of nanoenergeticmaterials studied by time-resolved infrared spectroscopy: aluminum nano-particles in Teflon. Journal of Physical Chemistry C 2007;111(28):10278–84.

[194] Mileham ML, Kramer MP, Stiegman AE. Laser initiation processes in thermiteenergetic materials using laser desorption ionization time-of-flight massspectrometry. Journal of Physical Chemistry C 2007;111(45):16883–8.

[195] Hunt EM, Pantoya ML. Ignition dynamics and activation energies of metallicthermites: from nano- to micron-scale particulate composites. Journal ofApplied Physics 2005;98:34909-1–34909-8.

[196] Bazyn T, Glumac N, Krier H, Ward TS, Schoenitz M, Dreizin EL. Reflected shockignition and combustion of aluminum and nanocomposite thermitepowders. Combustion Science and Technology 2007;179:457–76.

[197] Ward TS, Trunov MA, Schoentiz M, Dreizin EL. Experimental methodologyand heat transfer model for identification of ignition kinetics of powderedfuels. International Journal of Heat and Mass Transfer 2006;49(25–26):4943–54.

[198] Shoshin YL, Trunov MA, Zhu X, Schoenitz M, Dreizin EL. Ignition ofaluminum-rich Al–Ti mechanical alloys in air. Combustion and Flame2006;144(4):688–97.

[199] Umbrajkar SM, Chen C-M, Schoenitz M, Dreizin EL. On limitations of iso-conversion data processing for reactions in Al-rich Al–MoO3 thermites.Thermocimica Acta, submitted for publication.

[200] Kwon Y-S, Gromov AA, Ilyin AP, Popenko EM, Rim G-H. The mechanism ofcombustion of superfine aluminum powders. Combustion and Flame2003;133:385–91.

[201] Moore DS, Son SF, Asay BW. Time-resolved spectral emission of deflagratingnano-Al and nano-MoO3 metastable interstitial composites. Propellants,Explosives, Pyrotechnics 2004;29(2):106–11.

[202] Puszynski JA. Reactivity of nanosize aluminum with metal oxides andwater vapor. Materials Research Society Symposium Proceedings 2003;800:223–32.

[203] Bockmon BS, Pantoya ML, Son SF, Asay BW, Mang JT. Combustion velocitiesand propagation mechanisms of metastable interstitial composites. Journalof Applied Physics 2005;98(6):1–7. art. no. 064903.

[204] Son SF, Asay BW, Foley TJ, Yetter RA, Wu MH, Risha GA. Combustion ofnanoscale Al–MoO3 thermite in microchannels. Journal of Propulsion andPower 2007;23(4):715–21.

[205] Risha GA, Son SF, Yetter RA, Yang V, Tappan BC. Combustion of nano-aluminum and liquid water. Proceedings of the Combustion Institute2007;31II:2029–36.

[206] Tasker DG, Asay BW, King JC, Sanders VE, Son SF. Dynamic measurements ofelectrical conductivity in metastable intermolecular composites. Journal ofApplied Physics 2006;99:023705.

[207] Wilson DE, Kim K. A simplified model for the combustion of Al–MoO3nanocomposite thermites. AIAA Paper 2003-4536; 2003.

[208] Wilson DE, Kim K. Combustion of consolidated and confined metastableintermolecular composites. AIAA Paper 2005-0275; 2005.

[209] Ershov AP. Convective detonation wave in a porous medium. Fizika Goreniyai Vzryva 1997;33(1):98–106.

[210] Ershov AP, Kupershtokh AL, Medvedev DA. Modeling of convective detona-tion waves in a porous medium by the method of lattice gases. FizikaGoreniya i Vzryva 2001;37(2):94–102.

[211] Trunov MA, Hoffmann VK, Schoenitz M, Dreizin EL. Combustion of boron–titanium nanocomposite powders in different environments. Journal ofPropulsion and Power 2008;24(2):184–91.

[212] Beloni E, Hoffmann VK, Dreizin EL. Combustion of decane-based slurries withmetallic fuel additives. Journal of Propulsion and Power 2008;24(6):1403–11.

[213] Bazyn T, Krier H, Glumac N. Combustion of nanoaluminum at elevatedpressure and temperature behind reflected shock waves. Combustion andFlame 2006;145(4):703–13.

[214] Huang Y, Risha GA, Yang V, Yetter RA. Combustion of bimodal nano/micron-sized aluminum particle dust in air. Proceedings of the Combustion Institute2007;31(II):2001–9.

[215] Dokhan A, Price EW, Seitzman JM, Sigman RK. The effects of bimodalaluminum with ultrafine aluminum on the burning rates of solid propellants.Proceedings of the Combustion Institute 2002;29(2):2939–45.

[216] De Luca LT, Galfetti L, Sevedrin F, Meda L, Marra G. Combustion of compositesolid propellants with nanodimensional aluminum. Fizika Goreniya i Vzryva2005;41(6):80–94.

[217] Galfetti L, De Luca LT, Severini F, Meda L, Marra G, Marchetti M, et al.Nanoparticles for solid rocket propulsion. Journal of Physics: CondensedMatter 2006;18(33):S1991–2005.

[218] Luman JR, Wehrman B, Kuo KK, Yetter RA, Masoud NM, Manning TG, et al.Development and characterization of high performance solid propellantscontaining nano-sized energetic ingredients. Proceedings of the CombustionInstitute 2007;31(Pt. 2):2089–96.

[219] Zhang M, Liang Y, Tang Q. Application of aluminum nanopowders in solidpropellant. Huojian Tuijin 2006;32(1):35–9.

[220] De Luca LT. Burning of aluminized solid rocket propellants: from micrometricto nanometric fuel size. Theory and practice of energetic materials. In:Proceedings of the seventh international autumn seminar on propellants,explosives and pyrotechnics, Xi’an, China; 2007. p. 277–89.

[221] Beckstead MW, Puduppakkam K, Thakre P, Yang V. Modeling of combustionand ignition of solid-propellant ingredients. Progress in Energy andCombustion Science 2007;33(6):497–551.

[222] Son SF. Performance and characterization of nanoenergetic materials atLos Alamos. Materials Research Society Symposium Proceedings 2003;800:161–72.

[223] Higa KT. Energetic nanocomposite lead-free electric primers. Journal ofPropulsion and Power 2007;23(4):722–7.

[224] Lu JP, Franson MD, Cliff MD. Evaluating the effect of ultrafine aluminum inmetalised explosives using plate dent tests. In: Thirty-fourth internationalannual conference of ICT; 2003. p. 128/1–128/14.

[225] Ward TS, Lormand B, Blomshield F, Bui DT. Evaluation of boron-nano-composite fuel additives in metallized propellants and thermobarics JANNAFPaper, San Diego, CA; 2006.

[226] Gogulya MF, Makhov MN, Brazhnikov MA, Dolgoborodov AY, Arkhipov VI,Zhigach AN, et al. Explosive characteristics of aluminized HMX-basednanocomposites. Combustion, Explosion and Shock Waves 2008;44(2):198–212.

[227] Kim SS, Elliott KC. Reactive material screening for mine countermeasureapplication JANNAF paper, Boston, MA; 2008.

[228] Ames RG. Energy release characteristics of impact-initiated energetic mate-rials. Materials Research Society Symposium Proceedings 2006;896:123–32.


Documents

Author's personal copydreyzin/Papers-pdf/PECS-09-review.pdfmolecularcompounds,such asTNT,RDX,HMX, CL-20,etc.[1,2].The maximum heatof combustion of such materials is generallylimited