Metal particle combustion and nanotechnology

Transcription

1 Available online at Proceedings of the Combustion Institute 32 (2009) Proceedings of the Combustion Institute Metal particle combustion and nanotechnology Richard A. Yetter a, *, Grant A. Risha b, Steven F. Son c a The Pennsylvania State University, University Park, PA, USA b The Pennsylvania State University Altoona, Altoona, PA, USA c Purdue University, West Lafayette, IN, USA Abstract Metal combustion has received renewed interest largely as a result of the ability to produce and characterize metallic nanoparticles. Much of the highly desirable traits of nanosized metal powders in combustion systems have been attributed to their high specific surface area (high reactivity) and potential ability to store energy in surfaces. In addition, nanosized powders are known to display increased catalytic activity, superparamagnetic behavior, superplasticity, lower melting temperatures, lower sintering temperatures, and higher theoretical densities compared to micron and larger sized materials. The lower melting temperatures can result in lower ignition temperatures of metals. The combustion rates of materials with nanopowders have been observed to increase significantly over similar materials with micron sized particles. A lower limit in size of nanoenergetic metallic powders in some cases may result from the presence of their passivating oxide coating. Consequently, coatings, self-assembled monolayers (SAMs), and the development of composite materials that limit the volume of non-energetic material in the powders have been under development in recent years. After a brief review of the classifications of metal combustion based on thermodynamic considerations and the different types of combustion regimes of metal particles (diffusion vs. kinetic control), an overview of the combustion of aluminum nanoparticles, their applications, and their synthesis and assembly is presented. Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Metal combustion; Nanotechnology; Nanoparticle; Synthesis; Assembly 1. Introduction Metal combustion has received renewed interest largely as a result of the ability to now routinely synthesize and characterize metallic nanoparticles, and of the longer-term potential for nanotechnology to allow for an unprecedented level of control over the structure of reactive/energetic materials on length scales from nanometers to meters. * Corresponding author. addresses: rayetter@psu.edu (R.A. Yetter), gar108@psu.edu (G.A. Risha), sson@purdue.edu (S.F. Son). Nanoscale materials are known to exhibit significantly different physical, chemical, electrical, and optical properties compared to their properties at the macroscale. The nanoparticles of interest have length scales commonly between 1 and 100 nm. For comparison, the length scale of a hydrogen atom is the order of 0.1 nm. A spherical particle having a diameter of a few nanometers contains only thousands of atoms. Therefore, the ratio of surface atoms to bulk atoms increases dramatically as the diameter of the particle decreases. To illustrate this point, Fig. 1 shows the surface to bulk atom ratio for a spherical iron crystal as a function of particle size [1]. Because surface atoms /$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi: /j.proci

2 1820 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) % of atoms in bulk/on surface bulk atoms surface atoms particle size (nm) Fig. 1. Surface to bulk atom ratio for spherical iron crystals from [1]. have a lower coordination, the electrical and thermo-physical properties are vastly different than the bulk atoms. When the surface to bulk atom ratio becomes significant, the bulk material can begin to exhibit the properties of the surface atoms. For example, gold is a well-known inert material. However, if gold particles are reduced in size to diameters 1 5 nm they show excellent catalytic properties [2]. Furthermore, properties such as melting point, freezing point, and heat of fusion change drastically when diameters are below 10 nm [3 5]. The effect of particle size on melting temperature and heat of fusion for tin nanoparticles is illustrated in Fig. 2 [5]. Absorption cross-sections vary considerably with particle size as evident from nanoscale aluminum particles having a black color versus a light grey color for micron aluminum. One important characteristic of nanoscale materials is the increase in specific surface area of the material, which allows for increased reactivity [6,7]. In addition, nanosized powders are known to display increased catalytic activity [8], superparamagnetic behavior [9], superplasticity [10], lower melting temperatures [11,12], lower sintering temperatures [13], and higher theoretical densities compared to micron and larger sized materials. The excess energy of surface atoms contributes to many of the extraordinary characteristics of nanoparticles [14]. In the nanotechnology community, there has been tremendous progress in the molecular sciences toward the total command of chemistry at all length scales (supramolecular chemistry). This progress has been inspired primarily by advances in structural determination of biological systems [15], for instance the chromosome, where meterlong individual DNA molecules are intricately wound around protein spools to fit into micronlong cells, and the abalone shell, which consists of millions of intricate biological layers that provide scaffolds for the assembly of hard inorganic layers. Similar advancements in assembly of molecular and nanoscale elements have been made in the pharmaceutical [16,17] and microelectronics [18] fields as well. These developments make it clear that in the foreseeable future it will be possible to synthesize any desired macroscopic structure with precise location of every atom [19 21]. For example, self-assembly of a binary system of particles into an ordered array has the potential to create macroscale structures with interesting mechanical [22,23], optical [24], and electrical properties [25 27]. Early particle selfassembly [23,26,28 35] was motivated by finding organized SiO 2 nanoparticles in the opal gem (Fig. 3), the property responsible for its unusual optical properties [31 33]. Researchers found that these organized structures were driven by entropy, thus allowing only certain lattice structures to form under precise conditions [35,36]. More recent work has focused on binary systems of particles that can assemble due to forces other than entropy, such as electrostatics [20,37 40]. Recently, binary systems of nanoparticles have been arranged into various lattice structures as shown in Fig. 4. The ability to form many different crystal structures is attributed to the particles having electrostatically charged surfaces. The two constituents are oppositely charged, thus having an attraction for each other and not to particles of the same species. Lattice structures similar to NaCl, AlB 2, and diamond have been created by self-assembly of nanoparticles. While the combustion and energetic materials communities have lagged in the usage of nanotechnology, it is clear that soon many areas of combustion will be influenced by nanotechnology melting temperature (K) particle diameter (nm) ΔH m (J/g) particle diameter (nm) Fig. 2. Melting point and normalized heat of fusion for tin nanoparticles as a function of particle diameter [5].

3 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) Fig. 3. Naturally self-assembled SiO 2 particles of two different sizes found in the opal gem [32]. Fig. 4. Binary nanoparticle superlattices (BNSLs) formed by electrostatic self-assembly. SEM and TEM (insets) image of 5.0 Ag and Au BNSL (left) with its diamond counterpart (right) [37]. as a result of future fuels, propellants, pyrotechnics, explosives, and reactive materials having nanoscale features (ingredients). The fabrication/ synthesis of many of these future materials will involve molecular engineering approaches. Research on nanoparticles provides an important component to understanding the behavior of materials at small length scales. Future energetic/reactive materials will involve more than just nanoparticles, e.g., nanoscale films and rods are two other nanostructures currently being investigated for energetic materials. However, nanoparticles serve as a convenient starting point for many fundamental studies. The combustion of metals has long been of interest to the combustion community because of their high energy densities. Metals are commonly used in solid-propellant rockets and are currently being studied for underwater propulsion using seawater as the oxidizer. Metals may be important fuels for the establishment of a lunar mission base and the exploration of Mars. High-temperature metal combustion is important to self-propagating high-temperature synthesis (SHS) of materials and to the production of metal oxide and nitride particles, as well as to spectacular displays of pyrotechnics. Metal cutting and welding can also be considered high-temperature combustion processes. Metal fires, and particularly metal dusts, are extremely dangerous, and have also led to explosions. A phenomenological understanding of metal combustion has been known for over forty years [41 49]. For oxygen-containing environments, in which the final product is a refractory metal oxide, early studies [50 52] recognized (i) the importance of the volatility of the metal relative to the volatility of the metal oxide and (ii) the relationship between the energy required to gasify the metal or metal oxide and the overall energy available from the oxidation reaction. For a given metal/oxygen system, the magnitudes of these energies and the metal and metal oxide vaporization dissociation or volatilization temperatures have been used to classify the metal combustion process. The two commonly described processes based on the energy required for gasification of the pure metal vs. the energy available from oxidation are (1) the metal is volatile, readily vaporizes, and the oxidation reaction occurs in the gasphase, and (2) the metal is nonvolatile and the oxidation process begins by heterogeneous surface reactions. For metals with non-volatile metal oxides that have heats of gasification exceeding the chemical energy released during the reaction, the requirement that the condensed oxide be present limits the maximum flame temperature to the vaporization dissociation or volatilization temperature of the oxide. Since phase-transition temperatures vary with the chemical composition of the atmosphere, the resulting combustion mechanism can be strongly influenced by the oxidizer type and the environmental pressure. Intersolubility of the metal and its products, as well as some reactants, is also relevant to the combustion behavior. For volatile metals, certain solubility combinations are known to lead to disruption and breakup of the solid or liquid metal. For nonvolatile metals, purely condensed-phase combustion may result (as in SHS). In addition, products (sometimes protective, sometimes not) may coat and build up on the surface or within the metal. This paper briefly reviews the classifications of metal combustion based on thermodynamic considerations and the different types of combustion regimes of metallic particles based on their size. The emphasis of the paper is then placed on the combustion and use of nanoparticles (particularly aluminum nanoparticles) in different combustion applications and on their synthesis and assembly. As mentioned above, much of the highly desirable traits of nanosized metal powders in combustion systems can be attributed to their high specific surface area (high reactivity) [6,53] and potential ability to store energy in surfaces [14]. As will be shown later, the lower melting temperatures of nanoparticles can result in lower ignition temperatures of metals. For example, ignition temperatures of nanoscale aluminum (nal) particles have been observed to be as low as 1000 K (versus ignition temperatures closer to the melting temperature of alumina, common of micron-sized particles) [54,55]. The combustion rates of materials with nanopowders have been observed to

4 1822 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) increase significantly over similar materials with micron-sized particles. For example, self-propagating high-temperature synthesis (SHS) reactions with nanopowders can support fast deflagrations and detonations (with combustion speeds of over 1000 m/s), which are several orders of magnitude greater than the propagation speeds of SHS reactions with micron- and larger-sized particles. A lower limit in size of nanoenergetic metallic powders in some applications may result from the presence of their passivating oxide coating. For example, Al particles typically have an oxide coating with a limiting thickness of about 3 nm at room temperature. With a 100-nm diameter particle having a 3-nm-thick coating, the energy loss per unit volume due to the presence of the oxide layer is 10%. A particle with a diameter of 10 nm with the same oxide layer thickness would have an energy loss per unit volume of approximately 60%. Consequently, coatings, self-assembled monolayers (SAMs), and the development of composite materials that limit the volume of non-energetic material in the powders have been under development in recent years. This field of research is now developing into what is referred to as nanoenergetics [56,57]. Because of the brevity of this paper, the topics are presented in many cases to introduce the reader to the subject, and consequently, an in depth review of this fast growing subject cannot be covered here. The reader is referred to the many references provided for more detailed discussion and other related research topics. 2. Metal combustion classification The combustion of metals in oxygen is typically classified by the way the metal is oxidized to its smallest suboxide. This process can either occur with the metal and oxidizer in the gas-phase (a vapor phase reaction) or with the metal as a condensed phase (a heterogeneous reaction). Because of the highly refractory nature of metal oxides, the flame temperature of many metal oxygen systems is limited by and therefore cannot exceed the vaporization dissociation or volatilization temperature of the metal oxide product. The limiting flame temperature results from the fact that the heat of vaporization dissociation of the metal oxide formed is greater than the energy available to raise the temperature of the condensed-phase oxide above its boiling point. Glassman [41] writes this condition as: DH vap-dissoc > Q R H 0 T;vol H ¼ DH avail where Q R is the heat of reaction of the metal at the reference temperature 298 K, ðh 0 T;vol H 0 298Þ is the enthalpy required to raise the product to its volatilization temperature at the pressure of concern, and DH vap-dissoc is the heat of vaporization dissociation of the metal oxide. In 1958, Von Grosse and Conway [50] introduced the concept that the flame temperature of a metal was limited to the boiling point of the oxide (or dissociative gasification into species other than the original metal oxides). Glassman [51,52] recognized the importance of dissociation and of the overall energetics in metal combustion systems to propose a method to classify the combustion process. For a condensed-phase fuel droplet to burn in the vapor phase, the gas-phase temperature must exceed the boiling point of the fuel droplet. Outwardly diffusing fuel species then react with oxidizer in the gas-phase, and thermal diffusion to the droplet surface sustains fuel evaporation. Most metals have high boiling point temperatures, and thus, for a metal to burn in the vapor phase, the oxide vaporization dissociation or volatilization temperature must be greater than the boiling point temperature of the metal. If the oxide s vaporization dissociation temperature is less than the boiling point of the fuel, combustion must proceed heterogeneously on the particle surface. This concept has become known as Glassman s criterion for the vapor phase combustion of metals. Metals that will burn in the vapor phase in oxygen can then be determined by comparing the metal s boiling point temperature to the temperature at which the metal product oxide is decomposed or dissociated to gas-phase molecules. When the species formed are all gas-phase species, this has been referred to as the volatilization temperature. These temperatures may be compared for different metal oxygen systems as shown in Table 1 [41]. From the data of Table 1 and Glassman s criterion, the oxidation of a large diameter aluminum particle will proceed with a detached diffusion flame because the boiling point temperature of aluminum (T bp = 2791 K) is significantly below the decomposition temperature of aluminum oxide (T vol = 4000 K). Accordingly, Be, Cr, Fe, Hf, Li, Mg, and Ti should also have the ability to burn as vaporphase diffusion flames at 1 atm in pure O 2. In contrast, B, Si, and Zr would be expected to burn heterogeneously. In the case of boron, although there is sufficient energy to vaporize the oxide, there is an insufficient amount of energy available to raise the temperature of the metal to its boiling point and change its phase. For Cr, Fe, Hf, and Ti, any heat loss from the reaction zone can change the mode of combustion since the metal boiling point temperature is within 400 K of the metal oxide volatilization temperature. In the case of Hf, if the flame temperature drops below the boiling point temperature of the metal, then both the metal and metal oxide would be non-volatile and oxidation would occur on or within the particle. Depending upon the reaction mechanism, a gas-phase intermediate need not exist.

5 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) Table 1 Various Properties of Metal and Metal Oxides (from [41]) Metal T bp (K) Oxide T vol (K) DH f,298 (kj/mol) DH vol (kj/mol) H Tvol H DH vol (kj/mol) Al 2791 Al 2 O B 4139 B 2 O Be 2741 BeO Cr 2952 Cr 2 O Fe 3133 FeO Hf 4876 HfO Li 1620 Li 2 O Mg 1366 MgO Si 3173 SiO Ti 3631 Ti 3 O Zr 4703 ZrO T vol = volatilization temperature (or stoichiometric combustion temperature creating compound under ambient conditions T = 298 K, P = 1 atm).t bp = metal boiling point at 1 atm. If the form of the oxygen reactant is varied, then the available enthalpy will change as a result of a change in the heat of reaction, and thus, the mode of combustion may change as well. The condition for vapor phase combustion versus heterogeneous combustion may also be influenced by pressure by its affect on the flame temperature (T vol or T d )as well as by its affect on the vaporization temperature of the metal reactant (T b ). For aluminum combustion in pure oxygen, combustion for all practical conditions occurs in the vapor phase. In air, this transition would be expected to occur near 200 atm as shown in Fig. 5 where for pressures greater than 200 atm, the vaporization temperature of pure aluminum exceeds the adiabatic flame temperature. As some reactant vaporization will occur at temperatures below the boiling point temperature of pure Al when inert and product species are present near the particle surface, the results of Fig. 5 will vary for an actual burning particle. Figure 5 also shows that when aluminum burns with either CO 2 or H 2 O (with the reactants initially at near ambient conditions), the combustion process will burn heterogeneously at considerably lower Temperature (K) Al+0.75NH 4 ClO 4 2Al+1.5(O Ar) Al vaporization 2Al+1.5O 2 2Al+3CO 2, Reactants at 298K 2Al+3H 2 O Reactants at 298K 2Al+3CO 2, Reactants at 1500K Pressure (atm) Fig. 5. Boiling temperatures of Al, and adiabatic flame temperatures of various stoichiometric Al-oxidizer-inert systems, as a function of pressure. pressures than with O 2 due to the lower heats of reaction and consequently lower flame temperatures. Williams [43] has summarized several of the dominant criteria in classifying metal combustion (Table 2). The three rows of this table contain three criteria for an overall classification. The first discriminator is the energy equation given above, which determines whether the available energy exceeds the energy required to heat and volatilize the final metal oxide. The second discriminator is also an energy statement that determines if the available energy exceeds the energy required to heat and vaporize the metal itself. With sufficient available energy, the combination of a volatile product and volatile metal indicates that the metal will burn much like a hydrocarbon droplet. A volatile product resulting from a non-volatile metal during reaction produces combustion phenomenology similar to how carbon particles burn. The clean surface combustion stage of boron-particle combustion is another example of this mode of combustion. A non-volatile metal-oxide product and volatile metal are typified by magnesium and aluminum particle combustion in air at 1 atm, while examples of a nonvolatile metal oxide with a nonvolatile metal are hafnium and zirconium combustion in air at 1 atm. As a third discriminator, intersolubility of the metal and its product is also relevant to the combustion behavior. For volatile metals, certain solubility combinations are known to lead to disruption and breakup of the original particle. For nonvolatile metals, purely condensed phase combustion may result (as in SHS). In addition, product coatings may build up on the surface or within the metal. Thermodynamic properties of the metal and metal oxide affect the combustion mode, but they can also affect the ignition behavior of metals. In addition to the volatilization temperatures of the metal and metal oxide, the relationship of the respective melting temperatures to each other and to the volatilization temperatures must be

6 1824 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) Table 2 Classification of metal particle combustion (from [43]) I Volatile product Nonvolatile product II Volatile metal Nonvolatile metal Volatile metal Nonvolatile metal Surface combustion Gas-phase combustion Surface or condensed phase combustion Gas-phase combustion III Soluble Nonsoluble Soluble Nonsoluble Soluble Nonsoluble Soluble Nonsoluble Product coating makes ignition difficult Metal may diffuse through growing product layer, purely condensed phase combustion possible Disruption strongly favored if product returns to metal If product returns to metal it may dilute it and cause disruption No product penetration into metal Product may build up in metal during burning No flux of product to metal Product may dilute metal during burning and cause disruption if its boiling point exceeds that of metal considered. For example, in aluminum combustion, particle ignition has typically been associated with the melting of the initial oxide layer that protects the metal, while in flame spread across aluminum surfaces, the melting of the aluminum substrate, which expands and cracks the oxide surface, is typically associated with the ignition temperature at the flame front. Structural phase changes of the metal that are energetic have also been suggested to be pertinent to ignition of metals. As mentioned in the introduction, only when the diameter of a particle goes below approximately 10 nm will the energetics of the surface significantly affect the thermodynamic properties. 3. Metal particle combustion regimes From the thermodynamic analysis of the previous section, it is obvious that metal combustion can occur either heterogeneously at the particle surface or homogeneously in the surrounding gaseous environment. The formation of the final product can also be a heterogeneous process or a homogeneous process. The combustion of metal particles introduces a length scale into the problem and hence time scales for mass and energy transport. Transport time scales may be compared to chemical time scales to further define the combustion mode, which ultimately controls macroscopic features such as burning rates and ignition delays. In a kinetically-controlled regime, the reaction rate is slow compared to the rates of mass and energy transport so that spatial non-uniformities are eliminated. When reactions are fast, the spatial non-uniformities of temperature and composition fail to be eliminated in the available combustion times. As a consequence, gradients of temperature and species are established in space. Reactants diffuse into the flame zone whereas products diffuse away from the flame zone. Such unmixed combustion is diffusion-controlled. For diffusion control and a Lewis number of unity (Le = a/d), the mass consumption rate of a particle per unit surface area in a quiescent environment is [41,42] _m 4pr 2 p ¼ qd r p lnð1 þ BÞ where B is the mass transfer number, q is the gas density, D is the gas mass diffusivity, a is the gas thermal diffusivity, and r p is the particle radius. For a vaporizing particle, B is based on the coupling function between the energy and oxidizer species equations, B Oq ¼ iy O;1H þ c p ðt 1 T s Þ L v where i is the mass stoichiometric fuel oxidant ratio, H is the heat of reaction of the fuel per unit

7 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) mass, c p is the specific heat, T 1 is the gas temperature far from the surface, T s is the surface temperature and L v is the latent heat of vaporization. For a particle with heterogeneous surface reactions, B is obtained from the coupling function of the fuel oxidizer species equations, B OF ¼ ðiy O;1 þ Y F;S Þ : ð1 Y F;S Þ Since there is no volatility of fuel, Y F,S = 0 and B OF = iy O,1, and the consumption rate per unit particle surface area reduces to, _m ¼ qd ln ð1 þ iy 4pr 2 O;1 Þ: p r p The combustion times are then q p d 2 0 t b;diff ¼ 8qD lnð1 þ BÞ and q p d 2 0 t b;diff ¼ 8qD lnð1 þ iy O;1 Þ ; the latter time specific for a particle with a surface reaction. Here, d 0 is the initial particle diameter, q p is the particle density, and qd is the product of the gas density and the diffusivity. For kinetically-controlled combustion (assuming that the diffusion rate of oxidizer to the surface is much faster than the reaction rate at the particle surface), the oxidizer mole fraction at the surface, X O,s, is approximately equal to X O,1. Therefore, the mass consumption rate of the particle per unit surface area is _m MW 4pr 2 p kpx O;1 p where X O is the oxidizer mole fraction and k is the surface reaction rate with the oxidizer. The combustion time for kinetic control is then q p d 0 t b;kin ¼ : 2MW p kpx O;1 Thus, t b in a kinetically-controlled regime is proportional to d 1 and in a diffusion-controlled regime to d 2. Moreover, t b is found to be inversely proportional to pressure under kinetically-controlled combustion, and in contrast, independent of pressure under diffusionally-controlled combustion (since D / P 1 ). The dominant combustion mechanism may be determined through a Damkohler number (Da) defined as Da ¼ t b;diff ¼ MW pkpd 0 X O;1 t b;kin 4qD lnð1 þ iy O;1 Þ Assuming Da = 1 defines the transition between diffusion and kinetic controlled regimes, an inverse relationship exists between the particle diameter and the system pressure at fixed Da. The equation also shows that large particles at high pressure likely experience diffusion-controlled combustion, and small particles at low pressures often lead to kinetically-controlled combustion. Another length scale of importance to the combustion of particles is the mean free path of the surrounding gas-phase. A comparison of this length scale to the particle diameter defines whether continuum conditions exist (i.e., whether the particle may be distinguished separately from the gas molecules). The Knudsen number is defined as, Kn ¼ 2k d p where k is the mean free path of the gas-phase. The mean free path for like molecules is given by 1 k ¼ pffiffi 2 pr2 N where r is the molecular diameter of the molecule and N is the gas concentration. From kinetic theory, 2l k ¼ 1=2 P 8MW prt where l is the gas viscosity and MW is the molecular weight of the gas. The condition Kn = 1 may also be used to roughly determine limiting regimes of particle combustion, which are effected by temperature and pressure through the mean free path. At atmospheric pressure, particles of dimensions 100-nm and smaller are characterized by Knudsen numbers greater than unity for the entire temperature range from room temperature to combustion flame temperatures, indicating that they can no longer be considered as macroscopic particles in a continuum gas. In the free molecular regime (Kn > 1), nanopowders will in many ways behave similar to large molecules. Consequently, the reactivity of nanoparticles will generally be defined by kinetic rates and not transport rates of reactants (or energy) to the particle surface or products from the surface (however, as discussed below, transport rates within nanoparticles must also be considered as rate-limiting). As a result, considerable interest presently exists in the application of nanometersized metal particles to combustion, where the Knudsen limit is attained for all temperatures. 4. Nanoscale aluminum particle combustion and ignition Considerable research has been performed on the combustion and ignition of micron sized aluminum particles. These studies have been recently

8 1826 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) reviewed by Beckstead [58,59] and Yetter and Dryer [49]. Early phenomenological models of aluminum combustion were proposed by Brzustowski and Glassman [60]. These models have been followed by analytical models [61,62], more detailed models with various levels of chemistry submodels [63,64], and most recently with molecular dynamics models of very small particles [65]. As the size of the particle is decreased, Biot numbers become small and Fourier numbers become large indicating that particle temperatures will quickly equilibrate to the surrounding temperature and become uniform throughout the particle. Hence, the particle temperature will be driven largely by the surrounding environmental temperature. If the environmental temperature is above the vaporization temperature of the metal, the particle will burn through a vapor phase mechanism, although a detached flame will not occur because of the fast transport rates in the surrounding environment. If the environmental temperature is above the melting temperature of the oxide, but not the vaporization temperature of the metal, then the oxide may form a cap on the surface because of the surface tension difference between the molten oxide and metal. Oxidation will occur heterogeneously at the molten metal surface. At still lower environmental temperatures, transport of metal and oxygen through the solid oxide shell with subsequent reaction will dominate. In Fig. 6, burning times of nanosized aluminum particles, as well as micron particles are shown [66]. The measurements of Parr et al. [56] were obtained by injecting nal into the gases of a hydrogen/oxygen/argon micro-diffusion flame burner. The mixture ratio was typically stoichiometric so that the aluminum would burn in a high temperature steam environment. The temperature of the environment was varied by diluting with argon. As can be seen from the figure, a slope change exists at about 10 lm, indicating the possible transition from a diffusively-controlled combustion process to a kinetically-controlled combustion process. From shock tube measurements, the existence of a transition in oxygen mixtures has also been shown to occur near particle sizes of 10 lm at a pressure of 10 atm by Bazyn et al. [67]. Above 20 lm, combustion temperatures were above the boiling point temperature. However, as the particle diameter was decreased to 10 lm, combustion temperatures became close to the boiling point temperature of Al. Their results also showed that nanoparticle combustion temperatures were below the aluminum boiling temperature for ambient temperatures below the vaporization temperature. Combustion temperatures that exceeded the melting temperature of the oxide appeared to deviate higher from the ambient than those below the melting temperature. The results of Parr et al. for particle sizes smaller than 10 lm show a strong temperature dependence, which was also found with the shock tube measurements using oxygen [68]. For micron-sized particles, the burning time is only a weak function of the environmental gas temperature [58]. A strong pressure dependence was found for nanoparticles [67], also indicative of a kinetically-controlled process. The diameter exponent in the burning time relationship t b = bd n is less than one for both the data of Parr et al. and Bazyn et al. However, uncertainty in the size distribution and the amount of particle agglomeration in the experiments prevents one from determining the extent of deviation from unity. In the experiments with nanoparticles, only small amounts of AlO were found in the surrounding gas phase suggesting that a significant fraction of the overall reaction is occurring on or within the particles or that gas-phase reactions of AlO, like those of aluminum vapor, are fast and therefore suppress the AlO concentrations. Particle burning rates of nanoscale aluminum have also been deduced from burning velocities of aerosol flames [69]. These studies show a dependence on diameter close to or below unity as well. Also shown in the plots are the oxidation times from non-flame studies of aluminum nanoparticle oxidation at lower temperatures. The burning times at lower temperatures are considerably larger than those at higher temperatures. Park et al. [70] and Rai et al. [71] conducted flow reactor oxidation studies on a high-temperature particle laden air flow. After a defined reaction time the particles were sampled into a particle mass spectrometer where the composition was determined, from which the percent conversion and reaction rate were determined. From their oxidation studies, they concluded that the oxidation process occurred in the condensed phase, in two regimes, and was diffusion limited by the transport of reactant across the oxide layer [72]. For temperatures below the melting Burning Time [ms] Wilson and Williams Wong and Turns [2] Prentice [4] Olsen and Beckstead [1] Hartman [3] Friedman and Macek [5,6] Davis [8] Parr et al. [9] (T=1500 K) Parr et al. [9] (T=2000 K) Park et al. [10] (T=1173 K) Eisenreich et al. [11] (T=900 K) Diameter [μm] Fig. 6. Particle burning times as a function of diameter [66]. References for the data in the figure can be found in [66].

9 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) point of aluminum, a slow oxidation regime occurs in which the primary mechanism of oxidation is diffusion of oxygen through the oxide shell. Above the melting point of aluminum, a faster oxidation process occurs with diffusion of both aluminum and oxygen across the oxide shell producing hollow particles. The research of Eisenreich et al. [73] was conducted in a thermogravimetric analyzer at temperatures up to approximately 900 K. They also concluded that the oxidation process occurred via a shrinking core model, and provided estimates of both the kinetic rate at the reacting surface and the diffusion rate. Although the model generally predicted a diffusion-controlled mechanism, the magnitude of the overall rates and hence burning time was not consistent with the predictions of [70,71]. In fact, Park et al. [70] also performed TGA experiments and found the oxidation process occurred with higher activation energy, but the rates were always slower than obtained from the flow reactor studies. For nanoparticles, the processes of ignition and combustion become more difficult to distinguish. Figure 7 shows the experimentally observed aluminum-particle ignition (or onset of reaction) temperature as a function of particle diameter in oxygen-containing environments [74]. The data plotted were obtained under different conditions and particle size distributions; therefore, some caution is required in comparing the data from different studies. However, the general trends show that for larger particles (>100 lm), most experimental studies [75 81] indicate that ignition is achieved at a temperature near the melting point of aluminum oxide (i.e., 2350 K). The general interpretation has been that the particle is covered by an impervious oxide shell, and aluminum does not significantly react until the oxide shell melts or breaks up near its melting temperature under the effect of aluminum thermal expansion. Aluminum particles with diameters of lm, however, could be ignited over a wide range of temperatures from 1300 to 2300 K. For nanosized particles, reaction has been reported to occur at temperatures below the melting point of aluminum, as low as 900 K [56,82]. Trunov et al. [83] conducted experiments of thermogravimetric analysis (TGA) for oxidizing aluminum powders. They suggested that aluminum oxidation and polymorphic phase transformations of the alumina shell are responsible for these diverse ignition (or onset of reaction) temperatures. In their model, ignition occurs with melting of the oxide coating, as typically assumed for micron particles. The oxidation process leading to ignition is exothermic and occurs through particle oxide growth and phase transformations of the oxide. The oxidation process starts with the growth of the natural amorphous alumina layer, which is controlled by outward diffusion of Al cations. The energy of the oxide metal interface stabilizes the oxide only up to a critical thickness of about 5 nm. When the temperature is sufficiently high, the amorphous oxide transforms to the c phase. If prior to the phase change, the thickness of amorphous layer is less than 5 nm, c-alumina no longer forms a continuous covering of the surface due to its higher density. The resulting pores or openings allow direct oxidation and gradually heal. The growth of c-alumina continues and oxidation is limited by inward grain boundary diffusion of oxygen anions. Transformations of c-alumina to h and d phases occur with the eventual formation of dense a phase alumina. Lower ignition temperatures can also be affected by the presence of water in the environment and the formation of stable aluminum oxyhydrides, and therefore the oxide coating may be weakened prior to its melting temperature. It has also been proposed that with a growing oxide film at a specific temperature, mechanical stresses at the metal metal oxide interface occur that result from differences in linear expansion coefficients and bulk densities of the metal and oxide [84]. After attaining the ultimate strength of the oxide film, these stresses disrupt the protective properties of the oxide coating and form cracks. Recent molecular dynamics simulations suggest that large internal pressure gradients are present inside nanoparticles, with the aluminum core under a positive pressure and the alumina oxide shell under a negative pressure [85], consistent with the idea of oxide cracks or fragmentation. Initiation experiments of energetic materials with nal of different oxide layer thicknesses have shown that the rate (or strength) of the resulting reaction of the particles depends upon the heating rate and the thickness of the oxide coat [86]. Under slow heating rates (such as in a deflagration), there is no significant difference in the reaction rate between particles with Ignition Temperature, K Parr et al. Bulian et al. Assovskiy et al. Yusasa et al. Brossard et al. Ermakov et al Particle Diameter, μm Derevyaga et al. Merzhanov et al. Friedman et al. Trunov et al. Curve Fit Fig. 7. Particle ignition temperatures as a function of diameter [74]. References for the data in the figure can be found in [74].

10 1828 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) thin and thick oxide coatings. However, under high heating rates (such as in a detonation), the particles with a thicker oxide shell produce a stronger reaction that is explained by the thicker oxide shell confining the hot Al for a longer period of time allowing the internal pressure to become larger. Levitas et al. [87,88] have proposed a melt dispersion mechanism for nanoparticles that are rapidly heated, e.g., during fast reactions of nanothermites. In this mechanism, the volume change due to melting of the aluminum core induces pressures of GPa and causes spallation. The subsequent unloading wave creates high tensile pressures resulting in dispersion of aluminum clusters yielding a situation where oxidation can be greatly accelerated. When nanoparticles approach the size of 10 nm, the changes in thermophysical properties can become significant. Melting temperatures of both the metal and oxide shell will be affected [89]. The changes in the properties of the oxide shell are often overlooked, and may also be important in the mechanisms of combustion and ignition. Clearly, the mechanisms of nanoparticle ignition and combustion depend upon the physical environment. Although these mechanisms are clearly not understood yet, the properties of nanoparticles introduce interesting behaviors into the reaction dynamics. At diameters of a nanometer and less, metallic clusters are formed. Aluminum clusters of less than 100 atoms have been shown to exhibit interesting properties that change significantly as one atom is removed or added to the structure [90,91]. For example, aluminum ion clusters with 13 atoms are very stable whereas those with one atom more or less are not. Clusters of boron also have interesting characteristics and can form a number of structures at the nanoscale such as nanotubes [92]. It is likely in the near future metallic clusters in fuels and energetic materials will receive considerable attention. 5. Applications of nanoparticles in combustion systems 5.1. Nanofluids Nanofluids are fluids in which nanoparticles are dispersed at very low concentrations (Choi et al. [93]). Nanofluids were first studied after they were found to have significantly higher thermal conductivities than the same fluid without nanoparticles. Conductivity enhancements of greater than 10% are common with low particle concentrations, and an enhancement as large as 150% has been found using engine oil with 1% (vol.) carbon nanotubes (CNTs) dispersed within the oil [94]. Thermal conductivities of nanofluids have also shown a strong temperature and particle size dependence, increasing with temperature and decreasing with particle or agglomeration size. The temperature dependence is significant, indicating that nanofluids are sensitive to temperature increases or hot spots, and respond will correspondingly by increasing conductivity [95]. In addition to enhanced thermal conductivities [96 98], nanofluids have also been shown to exhibit enhanced mass diffusivity [99], radiative heat transfer [100,101], and ion transfer. The dispersion stability of particles within the fluid is improved using nanoparticles while other physical properties, such as viscosity and density, are not greatly affected since concentration and diameters of particles in nanofluids are small. Microscale particles settle rapidly after dispersion (a problem encountered with slurries in many previous combustion studies); nanoscale particles, due to their small mass, diameter, and specific surface area, can be suspended for weeks and even months under certain conditions. Overall, settling rates may be simply described using the Stokes Einstein theory [102]. Most of the research published to date involving nanofluids has been on thermal transport properties, mainly heat conduction, although two-phase heat transfer (both pool and convective boiling) has been studied more recently. A great deal of effort has been made to understand the mechanisms which create large thermal transport enhancements. Choi et al. have offered several good reviews of experimental and theoretical investigations involving the thermal transport of nanofluids in the last few years [103,95,104,105]. Although nanoparticles disperse fairly easily, they agglomerate easily as well. Once nanoparticles agglomerate, their effective diameter and settling rates increase dramatically. In the majority of studies involving nanofluids, physical dispersion methods are used, including high-shear mixing and ultrasonication. Physical methods are easy to employ, but long term stability may not be obtained at this time without use of a chemical stabilization method to form a colloid. Chemical stabilization is used to overcome van der Waals forces between particles that lead to agglomeration. Chemical stabilization methods that have been used to date are electrostatic and steric dispersion, as well as surface modification through the use of self-assembled monolayers (SAMs) [103,106,107]. These methods offer increased dispersion stabilization over physical methods, although they all have drawbacks. Studies involving the combustion of nanofluid type mixtures are limited, particularly those using reactive fuel particles. Tyagi et al. [108] have studied the addition of small quantities of aluminum and aluminum oxide nanoparticles to diesel fuel to improve the ignition properties. Droplet ignition experiments were carried out using particle sizes of 15 and 50 nm and particle volume fractions of 0%, 0.1%, and 0.5%. Over the range of

11 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) C, it was observed that the ignition probability for the fuel mixtures that contained nanoparticles was significantly higher than that of pure diesel, although a strong correlation with particle size or concentration was not identified. Sabourin and Yetter [109] studied the burning rates of nanofluids, consisting of nitromethane and nanoscale high surface area metal oxides of silica and alumina. Increased nitromethane burning rates of greater than 50% were found with less than 1.0% (mass) of inert particle addition at 5.24 MPa. The dilute particle additions showed only small effects on equilibrium flame temperatures, surface tension, density, viscosity, and specific heat. The nanofluids displayed a lower burning rate equation pressure exponent, which decreased with increasing particle concentration. Burning rates were also found to increase with nanoparticle surface area and with nanoparticle concentration. Wickman et al. [110] have studied carboxylatoalumoxanes as combustion catalysts. In these nanocatalysts, the boehmite support particles are very small, between 20 and 200 nm, and can be dispersed very effectively in liquids. The functional group on the carboxylic acid, R, is adjusted to make the particles dispersible in a wide variety of liquids. For example, if the R group is polar, then the nanoscale boehmite particles are dispersible in polar solvents such as water. If R is a nonpolar paraffin, then the material is dispersible in organic materials such as jet fuels. Although alumoxanes have catalytic activity in a combustion environment, their catalytic properties can be greatly improved by adding other metals into the boehmite core. Over 35 different metals have been exchanged into the lattice, including metals that catalyze oxidation such as palladium, platinum, lanthanum, strontium, manganese, and cerium. Combustion experiments with hydrocarbon fuels have shown lower ignition temperatures, shorter ignition delays, and conversion of carbon to carbon dioxide at much lower temperatures with the addition of the nanocatalysts. Fuel borne catalyst in diesel fuels for controlling soot emissions has also received considerable interest. Numerous companies have looked at different nanoscale metal oxides that alter the soot structure during combustion and hence its oxidative reactivity, e.g., after capture in particle traps at low temperatures [111]. Nanoscale iron particles in jet fuels are being studied as a means to remove adsorbed oxygen to eliminate coking in fuel lines [112]. The iron nanoparticles (7 nm) are formed by a sonochemical method and coated with dioctyl sulfosuccinate sodium salt (AOT) or oleic acid. FTIR analysis shows that the iron particle surface is chemically bonded to the organic component [113]. At low temperatures, the core-shell nanoparticles are resistant to oxidation, the coating preventing the oxidation of the Fe 0 core to an oxide form (Fe x O y ). However, at higher temperatures (110 C) over a narrow temperature range, the coatings open or release from the surface, giving oxygen access to the iron core. In a fuel with adsorbed oxygen, the iron particles then react with the oxygen before coking reactions can occur. Interestingly, the iron oxide particles formed may then be used as a catalyst in the oxidation of the hydrocarbon at higher temperatures in the gas-phase to reduce ignition delays. Such capability suggests development of temperature activated catalyst and reactive particles Gels At much higher particle concentrations, gelled metallized propellants have long attracted attention from the space exploration and defense community in the search for improved propulsion systems in terms of safety and energy density [114]. Gelled propellants are recognized as highperformance propellants which share some of the advantageous properties of liquid and solid propellants. The gelling of liquid mono- and bipropellants reduces risk of leakages while maintaining their ability to be pumped and throttled, unlike solid propellants. Gelled propellants are generally less sensitive to impact, friction, and electrostatic discharge than solid propellants. Additionally, propellant cracks are a major concern in solid propellants, which is not an issue with gels. From a performance standpoint, gelled propellants offer high specific and density impulses, which are comparable to liquid systems, and performance may be further increased with more energetic materials such as metal particles. Gelled propellants are generally more difficult to break up into droplets however for combustion. Energy densities of propellants are increased with the addition of metals such as aluminum (Al), which in turn increases burning rates, due to the exothermic Al oxidation reaction. Micronsized particles, which are used in the majority of metalized propellants, demand longer combustion times for complete combustion and require higher ignition temperatures than smaller nanosized particles. Nanoparticles are desirable since they offer shortened ignition delay, decreased burn times, more complete combustion, higher specific surface area, and the ability to act as gelling agents for liquids, replacing traditionally used non-energetic gelling agents such as fumed silica (SiO 2 ). The history of experimental and theoretical research of metalized gelled propellants dates back decades, with most of the work devoted to aerospace applications [114,115]. Much work began in the 1950s and 1960s with the study of slurry fuels with various metals. In the 1970s hydrocarbons, hydrazine derivatives, and inhibited red fuming nitric acid (IRFNA) began to be studied. The majority of the work involving metalized gelled propellants in the last decade and a

12 1830 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) half has been devoted to bi-propellant systems [116,117], and very little of these studies have involved particles using nanoscale particles (d p < 100 nm) [118]. Because of the high specific surface area of nanoscale gellants, gelled cryogenic propellants with nanoscale gellants require 25 50% less mass of gellant compared to traditional micron-sized gellants. Several recent studies were conducted by various groups focusing on the use of nanosized aluminum as a gelling agent. Spray combustion of gelled RP-1 fuel with ultra fine aluminum (ALEX) was studied in a rocket motor with gaseous oxygen [119]. The weight percentage of ALEX added to the RP-1 was as high as 55%, although the mixture with 5 weight percent had the best C * efficiency. A similar study [120] with 16 weight percent nal showed good ignition and stability characteristics, and C * efficiencies as high as 97% at an O/F of unity. Metalized gelled nitromethane, using ALEX and 5 lm diameter particles, has also been studied in strand burner experiments [121]. Using 5% Aerosil (fumed silica) and ALEX particle loadings of 5% and 10%, burning rates of approximately 4 and 5 mm/s were found at the highest pressure studied (12 13 MPa), significantly higher than those of pure nitromethane. From theoretical calculations, it was determined that flame temperatures (confirmed from experiments) and specific impulse are increased with nal addition to gelled nitromethane, as well. Sabourin et al. [122] extended the work of Weiser et al. showing that nitromethane may be gelled solely by passivated nal particles. At even higher particle loadings, nal has been mixed with liquid water to form a thick paste and mixtures that appear as dry powders (where most of the water is absorbed onto the high surface area nanoparticles). Ivanov et al. [ ] first investigated ultrafine aluminum metal powders (UFP) in a mixture of water in the presence of a thickening agent, polyacrylamide (3%). The specific surface area of the particles ranged from 5 to 50 m 2 /g. In their experiments, they mixed UFP aluminum with distilled water and added the thickening agent at equivalence ratios of 0.67 and 1.0. The mixture would not ignite without including the polyacrylamide thickening agent in the mixture. The mixture was filled into 10 mm tubes and ignited with an electrical coil inside of a constant pressure vessel with argon as the atmospheric gas. At the maximum test pressure of 7 MPa, the maximum burning rate of the mixture was found to be approximately 1.5 cm/s. Shafirovich et al. [126] investigated the combustion behavior of 80-nm nal water mixtures also with a polyacrylamide gelling agent. They found that 80-nm nal H 2 O mixtures yielded a combustion efficiency of 50%. Risha et al. [127,128] investigated the combustion of nal and liquid water without the use of any additional gelling agent. Steady-state burning rates were obtained at room temperature (25 C) using a windowed vessel for a pressure range of MPa in an argon atmosphere with particles having a nominal diameter of 38 nm. The effects of particle diameter (50, 80, and 130 nm) and overall mixture equivalence ratio (0.5 < / < 1.25) on the burning rate were also studied at a pressure of 3.65 MPa. Chemical efficiencies were found to range from 27% to 99% depending upon particle size and sample preparation. Burning rates increased significantly with decreased particle size attaining rates as high as 8 cm/s for the 38 nm diameter particles above approximately 4 MPa. Burning rate pressure exponents of 0.47, 0.27, and 0.31 were determined for the 38, 80, and 130-nm diameter particle mixtures, respectively. The burning rates for a stoichiometric mixture with 38 nm diameter particles are shown in Fig. 8 and compared to burning rates of other energetic ingredients. Mixture packing density, which varied with particle size and produced larger interstitial spacing for smaller particle diameter mixtures, was determined to affect the burning rates at high pressure. Mixtures of nal and water with other compounds, such as hydrogen peroxide [129] and ammonia borane [130,131], have also been studied. Various methods have been considered to increase shelf life of the mixtures, such as coating the particles, changing the ph level of the water, or freezing the liquid water to ice [132]. Because of the high energy density of aluminum water mixtures, they have been considered as sources of hydrogen for power generation, space propulsion [ ], and underwater propulsion [135]. Two studies have recently considered nal-ice rocket motors [132,136] Solid propellants The addition of nanoparticles in solid propellants has been shown in many studies to significantly enhance burning rates of the propellant compared to the same propellant formulations with micron-sized particles. Experiments by Ivanov and Tepper [137] with ultra fine aluminum (ALEX) and ammonium perchlorate have indicated an increase of burning rate by a factor of 10 compared to burning rates when industrial grade (micron-sized) aluminum was used. In AP/ HTPB/Al-based solid propellants (18% Al), Mench et al. [138] studied the effect of replacing a portion of the aluminum (50%) with nal (ALEX) and observed the burning rates to increase by a factor of 2, achieving rates near 1.8 cm/s. In these experiments, the pressure exponent in the burning rate law increased slightly with the addition of the nal. However, ignition times were reduced with the addition of ALEX. Similar studies with AP-based solid propellants (solids loading of 88%) have been reported by Arm-

13 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) strong et al. [139] also showing burning rate performance enhancement with nanopowders. A change in aluminum particle size from one micron to 100 nm produced an increase in burning rate of about a factor of 10. In addition, a slightly smaller pressure exponent (0.66 vs. 0.8) was reported when micron aluminum particles were replaced with ALEX particles. Simonenko and Zarko [140] also observed a decrease in pressure exponent, along with an increase in burning rate, when ALEX was substituted for micron aluminum powder in AP-based propellants with HMX. Other studies have also shown propellant burning rates to increase in GAP/AN and HTPB/AP mixtures when ALEX was substituted for conventional micron aluminum [141,142]. Studies show that by using nanopowders high burning rates can be achieved [143,144]. Observations of the propellant burning process with nal show a nearly uniform bright region immediately above the propellant surface, in contrast to large agglomerates that leave the surface as discrete isolated particles burning individually some distance (centimeters) from the surface. The more uniform burning process and close proximity to the propellant surface are indicative of the lower ignition temperatures and shorter burning times of nanoparticles. Agglomeration generally appears to be reduced and consequently, the sizes of the condensed phase products remain smaller leading to reduced slag formation in motors. While the use of nanoparticles can significantly increase burning rates and combustion efficiency, their high surface area can also introduce other complications particularly during processing. For example, high surface areas make nanoparticles more susceptible to agglomeration and unwanted ignition during processing. In addition, they are more susceptible to long term environmental degradation. As noted by Simonenko and Zarko [140], propellants with only ALEX particles vs. a mixture of micron and nanoparticles can exhibit unstable combustion because of the non-uniformities in the distribution of the nanoparticles during the fabrication process producing local spots of high concentrations. The resulting propellants can be more brittle too, leading to increased possibility of cracks. To minimize these deleterious effects, nanocomposite particles of micron size (including oxidizer and nanofuels) are under development as well as modifications to the particle surfaces, e.g., passivation layers to stabilize against unwanted ignition, and reduce long-term aging and surface contamination Solid fuels Solid fuels, such as those used in hybrid rockets, have slower burning rates than solid propellants and hence lower mass consumption rates. Since thrust is directly proportional to mass flow rate through a nozzle, specialized grain designs or methods to enhance the burning rate are required for hybrid motors to produce the same thrust as a solid propellant motor. Nanoparticles have also been investigated to enhance burning rates of solid fuels. The use of nanoparticles allows for energy release closer to the fuel surface, thus increasing the energy feedback rate to the surface and the regression rate of the fuel. Chiaverini et al. [145,146] used a two dimensional motor to study the effects of various weight percentages of ALEX aluminum nanoparticles (4%, 12%, and 20%) in HTPB-based solid fuels. The addition of 20% ALEX enhanced mass burning rates up to 70% in comparison to pure HTPB fuel formulations. Risha et al. [147] evaluated the addition of various energetic nanoparticles in a long-grain center perforated hybrid rocket motor. ALEX particles coated with Viton-A had the highest percentage increase in fuel mass burning rate with an enhancement of 120% compared to pure HTPB, which was more than a factor of two increase in mass burning rate over the solid fuel with uncoated ALEX. Nanoscale boron particles (80 nm nominal diameter) showed about the same percentage increase in burning rate as the ALEX particles. It was believed that fluorine compounds produced from the dissociation of Viton- A contributed to the rapid ignition and combustion of the nanoparticles Thermites In the 1970s, the study of self propagating high-temperature synthesis (SHS) reactions, mostly solid-phase (heterogeneous) reactions of metals, was initiated [148] to form many refractory and composite materials such as carbides, borides, selenides, silicides, and sulfides from their constituent powders. The reactions are mainly of Burning Rate [cm/s] r b [cm/s] = 4.5*(P[MPa]) 0.47 ADN* HNF* CL-20* JA2 # HMX* * Altwood, 1999 # Kopicz, Pressure [MPa] Fig. 8. Burning rate of nal/h 2 O mixtures compared to other energetic materials [127].

14 1832 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) the thermite (or intermetallic) type and have been reviewed by Wang et al. [149]. The study of superthermites (also called nanothermites or metastable intermolecular composite, MIC) from nanometric powders began at Los Alamos National Laboratory in the 1990s [150,151]. In this work, powder mixtures of Al and MoO 3 (or with other oxidizers) with average particle sizes of nm were found to react more than 1000 times faster than conventional powdered thermite owing to reduced diffusion distances between individual reactant species. Interesting, the reaction enthalpy of a stoichiometric mixture of Al and MoO 3 is 1.12 kcal/g, which is approximately 12% greater than TNT. Fischer and Grubelich provide thermodynamic analysis of many exothermic thermite and intermetallic mixtures [152]. A considerable amount of research has continued in the area of nanothermites since 2000 investigating the effects of particle size, oxidizer and fuel types, mixture ratio, packing density, pressure, and mixture sensitivity. An introduction to many of these studies can be found in [ ]. Many applications of nanothermites have been proposed including environmentally clean primers and detonators, chemical agent neutralization, improved rocket propellants and explosives, IR flares/decoys, thermal batteries, in-situ welding and soldering, nanoenergetics on a chip and others. Nanoscale aluminum has been studied the most as a fuel in nanothermite systems. Various oxidizers have been studied including MoO 3, CuO, WO 3,Fe 2 O 3, and Bi 2 O 3. The effect of decreasing particle size is generally to increase the combustion rate and lower the ignition temperature and energy. Eventually for very small nal mixtures, the combustion rates become independent of particle size because the passivation shell of Al 2 O 3 becomes a more appreciable component. In the case of nal particles at about 50 nm in average diameter, the amount of Al 2 O 3 reaches 43%. The added inert lowers reaction temperatures and increases diffusion barriers. Independent studies where nanoparticles of Al 2 O 3 have been added to nanoscale mixtures of Al and MoO 3 and CuO have also confirmed the effect of added diluents on propagation rates [156,157]. Both the size of the fuel and the size of the oxidizer particles are important. A decrease in either results in a faster reaction. Propagation rates are strongly affected by packing density of the thermite. For nanothermite mixtures, the lower the bulk density (i.e., the greater the porosity), the greater the propagation speed [158]. This trend is just the opposite for micron Al/MoO 3 thermite mixtures. At low packing densities for micron-sized mixtures, the reaction rates are low due to the larger length scales and void spaces, and heat losses represent a greater percent of the energy release. The propagation speeds are also a strong function of mixture ratio [159] and can exhibit different modes of propagation. Dutro et al. [161] have observed three modes of propagation for a nanoscale Al/MoO 3 thermite in burn tube tests; a steady high-speed propagation ( m/s), an oscillating and accelerating wave, and a steady slow speed wave (0.1 1 m/s). Mixture ratio limits of propagation were also determined. Correlations have been found between the optimum propagation speed of a nanothermite and the maximum pressure output [159]. In general, the propagation speed depends on the gas production, the thermodynamic state of the products, and the temperature of the reaction. In loosely packed powders, the propagation speed is likely supersonic with respect to the mixture sound speed. However, as noted above, the propagation speed decreases with density, which is opposite classical detonations, in which the detonation velocity increases with density. The propagation rates are also dependent on the materials processing and aging factors of the nanopowders (such as exposure to air and light) [160]. For example, the surface area of nanoscale MoO 3 has been reported to decrease twofold in days while aluminum metal content of nal particles can decrease by as much as 50% over 2 years [161]. The ageing of nanoscale ingredients and their sensitivity have resulted in numerous studies devoted to particle coatings [ ]. Ignition of thermites has been accomplished by chemical reaction, radiation from a heat source, electric currents and discharges, and mechanical impact. Nanothermites have been shown to exhibit significantly reduced ignition delays compared to micron-composite thermites by up to two orders of magnitude [158,168]. 6. Synthesis and assembly Nanoparticles have been synthesized by a number of different procedures using gas or liquid phase techniques, many of which have now been made commercial. In the future, the most significant gains will come from newly developed methodologies for passivating reactive particles and assembling them into reactive structures. Gas-phase condensation based techniques for production of nanosized metal particles have included exploding wire [137], plasma [169], and flame synthesis [ ]. Particles using these techniques can be produced continuously and therefore easily scaled. Wet chemistry techniques require batch processing, but generally permit more precise control over the end product, but precursor ingredients can sometimes be expensive. Nanoscale metallic particles are generally passivated with oxygen to achieve a stable oxide coating. At small length scales, this material represents

15 R.A. Yetter et al. / Proceedings of the Combustion Institute 32 (2009) a considerable amount of the mass and volume of the material. Considerable effort has been directed in recent years towards the development of chemistries for the synthesis of stabilized metal nanoparticles without the passivating oxide layer [173]. A central contributor to the progress made in this area is the development of powerful new methods that can be used to both functionalize and passivate the surfaces of metal clusters [174,175]. Organic self-assembling monolayers (SAMs) are densely packed organic thin films that spontaneously organize on a materials surface as a result of the chemisorption of a molecular amphiphile [176]. Because of their thermodynamically directed organization, SAMs serve as very effective barrier layers and have found wide use as materials modifying surface properties by rational molecular design [176,177]. Given that the organic layer comprising a SAM is only one molecule thick, it is the minimal structure one can envision using to preserve the energy content of a nanoscale material additive. Jouet et al. [178,179] have recently passivated bare aluminum nanoparticles using a perfluoroalkyl carboxylic acid (C 13 F 27 COOH), which binds to the surface of the Al particle. Because the aluminum particles were prepared using wet chemistry techniques and coated in solution, they are free of oxygen passivation. In shock reactivity and ignition tests, the SAM coated particles were found to have high energy content and fast reaction. Another example of surface passivation by SAMs was the nanoscale iron particle discussed earlier [113]. Fabrication of composite materials with nanoparticles has either involved powder mixing, where nanoparticles are first synthesized using the gas and liquid phase techniques mentioned above and then integrated into a fuel or energetic material, or non-conventional techniques, where the synthesis of the nanoparticle and nanocomposite structure (usually a fuel/oxidizer structure) are combined into a single process. For example, solution techniques such as sol-gel/aerogel processing [ ] have enabled production of particle-based composite systems. Chemical vapor deposition (CVD) and physical vapor deposition (PVD) [ ] have been used to form reactive structures with nanometer length scales, e.g., alternating films of reactive materials. Mechanical alloying and ball milling [ ] have been used to make particles, as well as micron-sized reactive composite particles. Spray processing of nanoscale materials is another alternative to producing large-scale bulk materials with nanoscale features [198]. More recently, soft lithography, nanopatterning and etching, self-assembly, and molecular engineering approaches have been applied to fabricate and assemble reactive materials at the nanoscale. Bottom up approaches for fabrication of reactive materials are still very limited. Nanostructured fuel/metal oxidizer nanocomposites have been fabricated by oppositely charging fuel and metal oxide particles in aerosols so that aggregation rates are enhanced for charged particles with opposite signs over those with the same sign [199]. The process was applied to form nanoscale Al/Fe 2 O 3 composite thermites, where burning rate tests showed a factor of ten increase over thermites produced by random Brownian coagulation. Copper oxide nanorods have been assembled with nal particles by coating the nanorod with poly(4-vinyl pyridine) polymer, washing, and drying the rods before mixing with nal particles [200,201]. The assembled nanothermite achieved burning rates as high as 2000 m/s, which is comparable to detonation velocities of some explosives. Recently, Malchi et al. [202] have used electrostatic self-assembly to create a nanoscale thermite system from nal and ncuo particles. Ligands with a positive or negative x-functionalization were attached to the surface of each nanoparticle to create a charged self-assembled monolayer (SAM). The ligand used to functionalize the surfaces of the aluminum particles was an x trimethylammonium (TMA) functionalized carboxylic acid, HOOCðCH 2 Þ 10 NMe 3 þ Cl. The ncuo surfaces were treated with the x-carboxylic acid functionalized thiol, or mercaptoundecanoic acid (MUA), HS(CH 2 ) 10 COOH. The positively charged nanoaluminum and negatively charged nanocupricoxide were each suspended in a separate solution and upon mixing the two constituents, many self assembled into microspheres with diameters on the order of 1 lm. Examples of the micron-sized composite particles formed are shown in Fig. 9. Gaining better understanding of these types of nanoscale manipulation and construction will allow for specific tailoring of the burning characteristics by varying the most fundamental of building blocks. Etching and nanopatterning techniques have recently been used to fabricate porous aluminum and silicon. The pores of these materials are filled using various techniques to form reactive composite materials. For example, electrochemical anodization of aluminum foils have been used to form nanoporous aluminum in which an array of iron Fig. 9. SEMs of nal and ncuo self-assembled nanothermite microspheres (SANTMs). The particles shown are a few micron in diameter [202].