Reactions of Aluminum with Halogen Containing Oxides

by

Cory Farley, B.S.M.E, M.S.B.M.E.

A Dissertation

In

MECHANICAL ENGINEERING

Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

Approved

Michelle Pantoya Chair of Committee

Kevin Long

Gordon Christopher

Jordan Berg

Sukalyan Bhattacharya

Dominick Casadonte Interim Dean of the Graduate School

May 2013

Copyright 2013, Cory Farley

Texas Tech University, Cory Farley, May 2013

ACKNOWLEDGEMENTS “For I know the plans I have for you,” declares the Lord, “plans to prosper you and not to harm you, plans to give you hope and a future.” – Jeremiah 29:11. Rarely do we take a direct path to the place God wants us to be, but the destination is worth it.

Dr Michelle Pantoya, thank you for the patience you have shown to me over the years as I stumbled my way through. From the time I showed up at your office unannounced and unexpected and you recruited me to the Combustion Lab, you have been a rock on which I could rely. When I was confused on my research, lost in my classes, or panicking on my thesis you were there to guide me back on my path. I only hope that I can portray your professionalism and success as I move forward in my career.

To my family, I left for school a long time ago as a scared little kid. However, I have always had your support to do anything, and this was no different. Dad, thank you for instilling a work ethic. Mimi and Papa, thank you for the financial help. I may have eaten ramen noodles for a stretch, but I have no student loans, and there is tremendous peace of mind knowing I will be moving forward debt free.

To my beloved wife Stephanie and my son Eli, thank you for providing sanctuary. My home has always been a place of serenity where I am free to love and be loved. Stephanie, you are my soulmate with whom there are no words to describe how powerful my feelings are for you and with Eli, you are my proof that God’s plan for me does indeed give me “hope and a future”

Also, I would like to acknowledge Santanu Chaudhuri and Martin Losada at ISP/Applied Sciences Laboratory, Washington State University, WA for their efforts running the quantum mechanical modeling seen in Chapter III.

ii

Texas Tech University, Cory Farley, May 2013

TABLE OF CONTENTS ACKNOWLEDGEMENTS ...... II LIST OF TABLES ...... IV LIST OF FIGURES ...... V ABSTRACT ...... VI CHAPTER I: BACKGROUND ...... 1 Ignition ...... 1 Propagation ...... 5 Flame Speed Measurement ...... 12 Conclusion ...... 14 CHAPTER II: REACTION KINETICS OF NANOMETRIC ALUMINUM AND IODINE PENTOXIDE ...... 15 Introduction ...... 15 Experimental Method...... 16 Results and Discussion ...... 17 CHAPTER III: REACTION KINETICS OF NANO-ENERGETICS WITH HALOGEN CONTAINING OXIDES ...... 23 Introduction ...... 23 Methods...... 24 Near Equilibrium Experiments ...... 25 Non-Equilibrium Experiments ...... 26 Ab Initio Quantum Chemical Calculations ...... 29 Solid State Calculations ...... 30 Gas-phase Calculations...... 31 Results ...... 33 Discussion ...... 39 CHAPTER IV: THE CONTRIBUTION OF ATMOSPHERIC OXYGEN TO ENERGY PROPAGATION IN REACTIVE MATERIALS ...... 43 Introduction ...... 43 Methods...... 46 Results ...... 48 Discussion ...... 50 Iodine pentoxide ...... 51 Copper oxide ...... 52 Calcium Iodate...... 52 Iron oxide ...... 53 CONCLUSIONS ...... 55 BIBLIOGRAPHY ...... 57

iii

Texas Tech University, Cory Farley, May 2013

LIST OF TABLES 1. Powder characteristics ...... 16 2. Reactant average particle diameter, supplier and anticipated products from Al reaction with respective oxidizer ...... 25 3. Summary of experimental results...... 36 4. Decomposition of iodine oxide clusters in gas phase for B3LYP, PBE0 and ACM functionals at aug-cc-pVTZ (aVTZ) and aug-cc-pVQZ (aVQZ) levels of theory. The values in the bracket are calculated using spin-orbit (SO) corrections...... 37 5. The reaction enthalpy and free energy calculated for Al reactions with iodine oxide fragments calculated at aug-cc-pVTZ (aVTZ) and aug-cc- pVQZ (aVQZ) ...... 38 6. Gas production from REAL Code analyses ...... 48 7. Summary of flame speed results in units of m/s...... 49

iv Texas Tech University, Cory Farley, May 2013

LIST OF FIGURES 1. Energy changes as an exothermic reaction proceeds ...... 2 2. Concept sketch of temperature profile across reaction zone of the heterogeneous composite...... 6

3. Heat flow and mass loss curves of I2O5; experiments performed in an Ar environment at a heating rate of 10°C/min...... 17

4. Heat flow and mass loss curves for Al2O3 and I2O5, experiments performed in an Ar environment at a heating rate of 10°C/min...... 18

5. Heat flow and mass loss curves for micron scale Al and I2O5, experiments performed in an Ar environment at a heating rate of 10°C/min. Note the lack of any exothermic reaction ...... 20

6. Nanometer scale Al powder and I2O5 with an equivalence ratio of 0.4, experiments performed in an Ar environment at a heating rate of 10°C/min...... 21 7. Schematic of bomb calorimetry experiment...... 27 8. Schematic for open tray flame speed measurements...... 27 9. Heat flow measurements for Al+I2O5 scanned at 2, 5 and 10 K/min heating rates...... 33 10. Peak temperatures for Al+I2O5 plotted using the Type B-1.95 Peak Method. The resulting Ea is 175.6 kJ/mol ...... 33 11. Graph showing Al+I2O5 Parr bomb calorimetry result displaying dT ...... 34

12. Open tray still frame images A) Al+I2O5 taken at 33µs intervals, B) Al+Ca(IO3)2 at 17ms intervals, C) Al+AgIO3 at 40 µs intervals, D) Al+Fe2O3 at 54ms intervals...... 35 13. Position of flame front as a function of time for one test for Al+I2O5 with a flame speed of 898 m/s...... 35 14. Top view schematic for open tray flame speed experiments...... 48

15. A) Al+Ca(IO3)2 at 6.9ms intervals. B) Al+Fe2O3 at 6.9ms intervals. C) Al+I2O5 at 44µs intervals. D) Al+CuO at 121µs intervals...... 49 16. Flowchart of fast convective reactions ...... 51 17. Percent increase in flame speed as a function of sensitivity factor ...... 54

v

Texas Tech University, Cory Farley, May 2013

ABSTRACT Due to increasing threats of biological attacks, new methods for the neutralization of spore forming bacteria are currently being examined. Thermites may be an effective method to produce high temperature reactions, and some compositions such as aluminum

(Al) and iodine pentoxide (I2O5) also have biocidal properties. This study examines the thermal degradation behavior of I2O5 mixed with micron and nanometer scale aluminum (Al) particles. Differential scanning calorimetry (DSC) and thermo-gravimetric (TG) analyses were performed in an argon environment on both particle scales revealing a non- reaction for micron Al and a complex multistep reaction for the nanometer scale Al.

Results show that upon I2O5 decomposition, iodine ion adsorption into the alumina shell passivating Al particles is the rate controlling step of the Al-I2O5 reaction. This pre- ignition reaction is unique to nano-Al mixtures and attributed to the significantly higher specific surface area of the nanometric Al particles which provide increased sites for I- sorption. A similar pre-ignition reaction had previously been observed with fluoride ions and the alumina shell passivating Al particles.

Composite energetic materials comprised of nanoparticle fuel and oxidizer can exhibit high flame propagation speeds on the order of 1000m/s when burning in an unconfined environment. In particular, halogen based formulations such as aluminum and iodine pentoxide have received significant attention due to both high flame speeds and biocidal properties. Studies have attributed high flame speeds to convective influences within a reaction driving the heat forward in a pressure wave accelerating ignition of unburned powders. This study examines factors contributing to convective flows such as gas and heat generation and their relationship to the measured flame speed as well as fundamental chemical kinetics influencing the observed flame speeds. The goal is to understand parameters directly related to high flame speeds in halogen containing composites. Results show a direct correlation between apparent activation energy and flame speed indicating that flame speed is directly influenced by chemical kinetics. For

vi

Texas Tech University, Cory Farley, May 2013

this reason, the intermediate chemistry associated with Al and iodine species was examined to identify chemical influences accelerating flame speeds. Ab initio quantum chemical calculations of gas-phase reactions resolve key exothermic intermediate reactions contributing toward the kinetics of the fastest burning mixtures. Condensed phase density functional theory calculations of Al2O3/I2O5 interfaces resolved adsorption/desorption properties.

This study examines the effect of atmospheric oxygen concentration (4 or 93% oxygen) on energy propagation of nanometric aluminum with copper oxide

(Al+CuO), iron oxide (Al+Fe2O3), calcium iodate (Al+Ca(IO3)2), and iodine pentoxide

(I2O5). In all cases energy propagation was examined in terms of flame speed and higher in the high oxygen environments. However, the convectively driven mixtures showed a smaller percent increase in flame speeds. This behavior is attributed to the increased availability of oxygen as a result of solid oxidizer thermal decomposition at lower temperatures. The slower Al+Fe2O3 reaction showed greater increases in flame speed attributed to early stage reactions involving atmospheric oxygen that promote oxide decomposition and faster flame speeds. A predictor based on solid oxidizer thermal decomposition and flame speed was developed to evaluate the sensitivity of a reaction to atmospheric oxygen concentration.

vii

Texas Tech University, Cory Farley, May 2013

CHAPTER I: BACKGROUND During the scientific revolution of the 17th century, Johann Becher in his Physica Subterranea (1667), claimed that the minerals of the earth were made of “terra fluida” (fluid earth), terra phingus (fatty earth), and terra lapidea (stony earth) (Bowler and Morus 2005). According to Becher, the terra phingus content made a material flammable. In the early 17th century, Georg Ernst Stahl continued Becher’s work and renamed terra phingus “phlogiston.” Combustion occurred when a material’s phlogiston was released, and a substance such as ash, could not burn because it no longer contained phlogiston. The theory of phlogiston would hold until the 1770’s with the discovery of oxygen. The discovery of oxygen can be attributed to either Carl Scheele and his “fire air” (1772) , Joseph Priestly and his “dephlogisticated air” (1775) or Antoine-Laurent Lavoisier who coined the term “oxygène” (1775). Earlier in 1772, Lavoisier had claimed that combustion was a combination of burning substance and air. Later upon reading about Priestly’s dephlogisticated air, Lavoisier refined his theory and argued that it was the dephlogisticated air, which he labeled “oxygène” that contributed to combustion, and the modern concept of combustion was born (Bowler and Morus 2005).

According to Merriam Webster Dictionary, combustion is defined as “a usually rapid chemical process (as oxidation) that produces heat and usually light.”(“Combustion”) For combustion to take place, the reactants must come together with enough energy to initiate just as any chemical reaction. In terms of combustion, the reactants are usually a fuel (methane, carbon, aluminum) and an oxidizer (oxygen gas, metal oxide, halogen gas). Once initiated, if the energy transferred to the surrounding reactants reaches a certain level, then initiation and transfer take place again and again, and the combustion propagates. The continued propagation of the chemical reaction results in a flame that was studied for burn rate and heat production in this paper.

Ignition The first step for combustion is the initiation of the chemical reaction achieved when a certain energy threshold for the reactants has been met. Svante Arrhenius studied this energy threshold and coined the term “activation energy,” which is visualized in

Figure 1 where denotes activation energy, heat of combustion.

1

Texas Tech University, Cory Farley, May 2013

Figure 1. Energy changes as an exothermic reaction proceeds

Arrhenius went on to quantify the relationship between and reaction rate, ( )as seen below in Eq. 1( ),

( ) ( ) (1)

where is a pre-exponential factor, is the gas constant, and is absolute temperature.

As seen in Figure 1, when the activation threshold is met, the reaction progresses. Combustion reactions are exothermic by definition, therefore the ensuing energy release is greater than the activation energy and heat is released into the surroundings. If the energy transferred to the neighboring reactants is greater than the activation energy, then the chemical reaction initiates, and the combustion of the particles propagates along the powder (Turns 2012).

The activation and initiation of a combustion reaction is called ignition. Glassman describes two types of ignition: thermal and spark ignition (Glassman and Yetter 2008). Thermal ignition can be active (i.e. a match) or passive resulting in a spontaneous combustion with no active trigger. For example, a large pile of leaves can spontaneously ignite as heat from the decaying process builds up. Simply put, as long as

iv Texas Tech University, Cory Farley, May 2013

the rate of heat generation is greater than the rate of heat dissipation, the ensuing energy buildup could ignite a combustion reaction. In 1935, Semenov modeled this behavior by combining an energy balance with the Arrhenius form of the chemical rate equation to develop a time constant comparison of the generation and dissipation energy terms as seen in Eq (2).

( )

( ) (2)

where and are the time constants for the generation and dissipation respectively, is the thermal energy release of the chemical reaction, and are the ideal gas constant and specific heats, is activation energy, and is the number .

For a more controlled ignition, a trigger is often used. For example, gasoline is typically ignited with a spark in an automobile internal combustion engine. Spark ignition is popular because it is reliable and safe due to the fact that a pilot light is not required to initiate combustion. Most discussion on spark ignition centers on the term “minimum ignition energy”. Minimum ignition energy describes the smallest energy impulse that can ignite a given mixture under given circumstances. In order to determine the minimum ignition energy required by a spark to initiate combustion, the spark itself is often modeled as a sphere of hot gas. Using the quench criteria that the heat generated by the reaction (or in this case spark) must be approximately equivalent to the rate of heat loss of a volume via conduction, a minimum energy input for a spark, and hence, a gap and voltage can be calculated. To meet the quenching criterion, Eq.(3) is introduced.

̇ ̇ (3)

In Eq. (3), ̇ is the rate of heat generation, is the spherical volume and ̇ refers to the heat loss rate via conduction. The heat generated can be related to the ̇ consumption of fuel ( ̇ ) by ̇ where is the heat of combustion. The heat lost to conduction can be modeled by Fourier’s law for an axisymmetric sphere as seen in Eq (4)

iv Texas Tech University, Cory Farley, May 2013

̇ ( ) (4)

for a material with thermal conductivity and minimal spherical radius corresponding to the sphere representing minimal ignition energy. Substituting

and ( ) ( ) for a given boundary temperature ( )

resulting in ignition in Eq. (4) allows for the determination of as seen in Eq (5).

√ √ (5)

where is thermal diffusivity, and is the latent flame speed and is the laminar flame thickness. Equation (5) indicates that the critical radius is roughly equivalent to the laminar flame thickness. From the minimum radius ( ), we can calculate a minimum mass ( ) and thus, a minimum energy ( ).

( ) (6)

Recall that for a spark, the energy is

( ) (7)

where is the capacitance in farads, and are the voltages after and before the spark.

The ignition of binderless solid particles results from a complex series of events adapted from a description by Kuo: (Kuo 1986)

 Thermal energy transfer to particles via conduction, convection, radiation

 Inert heating within the solid particles

 Oxidizer phase changes and decomposition

 Subsurface chemical reactions

 Diffusion of fuel/oxidizer species

 Heterogeneous reactions between gas/condensed phases

iv Texas Tech University, Cory Farley, May 2013

 Gas-Phase reactions

 Abrupt temperature increase

 Light emission

 Development of a flame

According to a review in 1966 by Price, ignition theories are classified into three major groups, 1) gas-phase, 2) heterogeneous, and 3) solid-phase (Goyne et al. 1966). In the case of aluminum combustion, most reactions are believed to fall in the latter two groups with nanometer scale aluminum particles having been shown to react in the solid phase as described by (change in thermal energy) = (conduction) + (heat generation of chemical reactions) + (energy accumulation due to in depth radiation absorption).

Reaction rates for these reactions are dependent on the chemical reaction rate constant, ( ), and how strongly the oxidizer adsorbs to the surface of the fuel (Turns 2012). If the oxidizer is weakly adsorbed, then the reaction rate, R, is proportional to the concentration of oxidizer [O].

( )[ ] (8)

Secondly, if the oxidizer is strongly adsorbed, then will depend mainly on ( )

( ) (9)

The remaining scenario involves a case where the combustion products compete with the oxidizer for adsorption sites on the surface of the fuel.

[ ] (10 ( ) [ ] )

Propagation After the flame has been ignited, the chemical reaction generates enough energy to drive the surrounding reactants to their activation energy and ignition. The “ignition and energy spread” cycle repeats with the surrounding reactants. The cyclic repetition of 1) activation, 2) ignition and 3) energy transfer describes the concept of flame propagation, which is displayed graphically in Figure 2.

iv Texas Tech University, Cory Farley, May 2013

Figure 2. Concept sketch of temperature profile across reaction zone of the heterogeneous composite. Classically, the study of flame propagation has been focused on three theories as summarized by Kuo.(Kuo 1986) He divided the theories into three major catergories: thermal, comprehensive, and diffusive. The theories cited from 1883 to 1954 are listed below with their corresponding citations.

 Thermal Theory

 Mallard and Le Chatelier’s Development

 The Damkohler theory

 The theory of Bartholome

 The theory of Emmons, Harr and Strong

 The theory of Bechert

 Comprehensive Theory

 The theory of Lewis and von Elbe

 The theory of Zel’dovich, Frank-Kamenetsky, and Semenov

iv Texas Tech University, Cory Farley, May 2013

 The theory of Boys and Corner

 The theory of Hirschfelder and Curtiss

 The theory of von Karman and Penner

 Diffusion Theory

 The theory of Tanford and Pease

 The theory of Van Tiggelen

 The theory of Manson

 The theory of Gaylord and Wolfhard

The purpose of laminar flame theories such as those listed above is to determine laminar flame speed ( ). In 1883, Mallard and Le Chatelier proposed the first thermal theories for flame propagation. Based on the temperature profile in Figure 2, they proposed that the preheating was driven by heat conduction from the combustion zone of the neighboring material. The flame speed is determined via energy balance and is found to be proportional to the exponential term for the reaction rate equation (Eq. (1) ) as seen in Eq. (11) below.

√ (11)

is activation energy, is the molar ideal gas constant and is the temperature at which the reaction takes place.

Later, Zeldovich, Frank-Kamenetskii, and Semenov would introduce the concept of mass diffusion to the thermal theory of Mallard and Le Chatelier. In the new comprehensive theory, mass diffusivity is assuming to be equal to thermal diffusivity ( ). This assumption leads to a combination of continuity and energy equations that when solved leads to the flame speed equation seen in Eq. (12).

√ (12) ( ̅ )

iv Texas Tech University, Cory Farley, May 2013

where is the specific heat at the flame temperature, and ̅ is the average specific heat between the initial and flame temperatures. are the pre-exponential term and activation energy from the Ahhrenius equation. The term refers to the initial reactant density while refers to the initial overall density.

In our modern age of computing, fewer assumptions need to be made since numerical solutions can be resolved with much less effort. Modern software such as CHEMKIN uses a scheme by Kee et al solves for mass burn rate as an eigenvalue problem and determines the flame speed from the mass consumption combined with the unburned gas density.

To understand the mechanics of solid combustion, a brief discussion on the combustion of carbon can help shed some light into the reaction kinetics. Unlike aluminum, carbon is not encased in any sort of protective shell; therefore, when burned in air, the surface of the carbon particle comes into direct contact with oxygen. Due to this intimate contact, the burning of a solid carbon particle can be analogously modeled to a burning fuel droplet. A straightforward discussion of the early part of this derivation is not readily available, therefore I included it here.

First, examine a first law balance for the surface of the particle in a one- dimensional Cartesian system. Eq. (13)

∑ ̇ ∑ ̇ ∑ ̇ ( )

(13)

∑ ̇ ( )

where ̇ ̇ are the heat and work boundary transfer rates for the control volume, ( ) are the enthalpy, kinetic and potential energy transfers associated with a given mass flow rate ̇ , and is the total energy of the control volume.

Assuming:

 Burning is steady state

 Potential and kinetic energies are negligible

 ̇ is negligible

iv Texas Tech University, Cory Farley, May 2013

 Bulk mass ( ̇) is a steady, single flow

Equation (13) simplifies to

(14 [ ̇ ] [ ̇ ] [ ̇ ] [ ̇ ] ) where ( ) denotes a flux over a given area .

Dividing Eq. (14) by ( ) yields

̇ ̇ ̇ ̇ (15

)

Then, take the limit as ,

̇ (16 ̇ )

From here, we need to take a detailed look at the heat transfer, ̇ . Until now, we have been treating the mass flow, ̇ , through the control volume as if it was a single component, steady flow. However, this is a reacting mixture and each species in the flow is changing as combustion takes place inside the control volume. In order to account for this, it is common practice to include the deviations of each species diffusion rates into the heat transfer term. Therefore, the heat transfer term includes both Fourier conduction and the enthalpies of species deviating from the bulk flow when diffusing through the carbon particle as seen below in Eq. 17,

(17 ̇ ∑ ̇ ) where ̇ is the mass diffusion deviating from bulk values with corresponding enthalpy, and is the thermal conductivity.

Recall from Fick’s law of mass diffusion,

(18 ̇ )

Substituting Eq. (18) into Eq. (17) yields

iv Texas Tech University, Cory Farley, May 2013

(19 ̇ ∑ ) assuming all the species have a similar diffusivity, .

From here, following Turns (Turns 2012), it can be shown that the Shvab-Zeldovich energy equation for spherical geometry is

(20 ∫ [ ( ∫ )] ∑ ̇ )

In order to get to this equation, the following assumptions have been made:

 Steady State

 No kinetic or potential energy effects

 No control volume work (i.e. shaft or piston)

 No radiation



Assuming a quasi-steady burning, ̇ is a constant with respect to therefore

( ) (21)

Recall that is bulk flow velocity. Thus, with constant properties, Eq. (20) can be rewritten as

̇ ( ) (22)

with the assumption of no reactions (evaporation only, ∑ ̇ ) at the surface of the droplet. From here, it is straightforward to follow Turns (Turns 2012) to arrive at the diffusion droplet lifetime equation

(23) where is total burn time, is initial diameter and is an evaporation constant calculated from material properties and temperature differences.

iv Texas Tech University, Cory Farley, May 2013

Now, burn time can be examined for a kinetically driven reaction. The flux of fuel at the droplet surface can be described as

̇ ̇ (24)

where the subscript denotes the particle surface.

While the flux of oxidizer can be described by

̇ ̇ [ ] (25)

where is the mass stoichiometric index, and is the reaction rate constant and [ ] is the concentration of oxidizer on the surface. However, [ ] is not readily known. However, if the reaction is kinetically limited, it can be assumed that the concentration on the surface is equal to the free stream concentration for oxidizer, or

[ ] [ ] (26) Combining Eqs. (24), (25), and (26) yields

̇ ̇ [ ] [ ] (27) which simplifies to

[ ] (28)

Separate the variables then integrate and solve for time.

(29) [ ] Therefore, in kinetic limited reactions, the burn time is proportional to the droplet diameter or

(30)

iv Texas Tech University, Cory Farley, May 2013

Flame Speed Measurement Flame speed is typically defined for gas phase reactions as “the velocity at which the unburned gases move through the combustion wave in a direction normal to the wave surface.” (Glassman and Yetter 2008) For example, in a flame tube, the flow rate that holds the flame stationary can be directly related to the flame speed. However, in multiphase media with varying particle geometries the concept of flame speed becomes less clear. Flame speeds for multiphase combustion processing involving solid phase reactants, flame speed will be defined here as “the velocity at which the flame front consumes and propagates through the unburned particles.”

Burn rate experiments for solid particle combustion can either consist of measuring the burn time for a single particle or burn rate for powders. Early experiments for calculating the burn time of solid particles involved dropping them into a flat flame, and in 1963, Friedman and Macek conducted this experiment with aluminum particles (Friedman and Maček 1963). The light traces from the burning particles were recorded and used to estimate burn times that correlated to power for aluminum particles from 10 to 74 µm. Recall, that for purely diffusion driven reactions, the burn time should be . For purely kinetic reactions, the correlation would be . So, a reaction of would have a mixture of diffusion and kinetic elements in the combustion. More recently, burn time experiments for single particles has been conducted in shock tubes. Beckstead, in 2005, collected data from a variety of experiments for particles ranging from 20 to 800µm and found a correlation of (Beckstead 2005). Lynch et al correlated 3-11µ1m particles at (Lynch, Krier, and Glumac 2009). Finally in 2011,

Gill et al used a CO2 laser to ignite 10-14µm particles and correlated them to (Gill, Badiola, and Dreizin 2010). The reaction kinetics involved causing the burn times to fall out of the range from kinetic to diffusion controlled are not fully understood at this time.

The powder flame speed measurements typically fall into three categories: aerosol, open tray and closed tube. Aerosols containing solid particles are extensively studied in an effort to understand dust explosions. Solid particle aerosols behave similarly to the premixed flame previously discussed (Proust and Veyssiere 1988). Bunsen burner style experiments with a constant flow are difficult to execute due to the difficulty in

iv Texas Tech University, Cory Farley, May 2013

mixing the solid particles into the flow. Instead, the powder is typically dispersed throughout a volume and then ignited. For volumes such as an open duct, the flame position is measured with optical high speed imaging. For spherical closed volumes, the flame speed can be related to pressure traces from pressure transducers within the vessel (Pu et al. 2007) or derived from optical techniques (Bradley, Chen, and Swithenbank 1989).

For combustion of masses of non-aerosolized powders, flame speed measurements are typically divided into open and closed measurements. For closed tube measurements, powder is loaded into a tube in such a way as to maintain a constant overall bulk density and ignited by a spark, hot wire or other stimuli (M. Weismiller et al. 2009; Bockmon et al. 2005). Optical readings can be taken by either high speed imaging or photodiodes. High speed cameras give more information to help determine the nature of the combustion and make accurate flame speed estimates. However, in many cases, flame speeds are so high ( ) that a camera with the resolution and frame rate required to get reliable information may prove cost prohibitive (Martirosyan, Wang, and Luss 2009). Photodiodes give similar information at a much lower cost. A photodiode works by converting light into an electrical signal. A device such as an oscilloscope triggers on a certain voltage met when a threshold of light signifying combustion strikes the diode. The drawback to photodiodes is that only a few set positions are monitored so only average speed data may be inferred during analysis.

Instead of photodiodes, pressure transducers can be used to help locate the passing flame front. Heat from the flame and gaseous products increases the pressure as the flame progresses. In 2007, Bockmon et al, designed an experimental setup that used both optical and pressure information by combining photodiodes and pressure transducers. In their experiments, flame speed exhibited a similar trend with pressure history (Bockmon et al. 2005).

Here, high speed video imaging was used for the flame speed measurements in an open tray. Loose powders were loaded into an open tray and ignited. The ensuing videos were analyzed and calibrated. Flame speed was determined by combining the distance information with the frame rate as a timing device. Drawbacks to

iv Texas Tech University, Cory Farley, May 2013

this method, aside from the camera cost, include spatial effects if the camera is not perfectly perpendicular to the sample. Also, with an optical camera, the luminescent front is assumed to travel with the flame at the same velocity. For highly convective flows, the hot product gases may be pushed ahead of the flame at a temperature capable of radiating visible light. The hot gases may increase the luminescence to a point where oversaturation leads to systemic error in flame speed calculations.

Conclusion Flame speeds of micron aluminum particles have been widely studied. With the onset of nanometer scale aluminum particles, a new regime in solid combustion has developed. Flame speeds can be 100 times faster just by reducing the size of the fuel particle (J. Granier and Pantoya 2004). Many theories have been developed in an attempt to explain this phenomenon including melt dispersion (Levitas et al. 2006), shrinking core (Park et al. 2005), and reactive sintering (Sullivan et al. 2012). However, none of these competing theories can completely explain the reaction kinetics for nanoparticle combustion. This work examines flame speeds in relation to basic reaction parameters in order to develop an understanding as to which reaction parameters are most important.

Aluminum interactions with halogen containing oxides have only been studied recently. Flame speeds of 2000m/s have been recorded (Martirosyan, Wang, and

Luss 2009) and product gases (I2, Cl2, F2) have potential in biocidal applications (Sullivan et al. 2011; B. R. Clark and Pantoya 2010). Zhang et al showed that mechanically allowing crystalline iodine to Al increased the reactivity of the powder (S. Zhang et al. 2012). Aluminum particles are passivated with a shell of aluminum oxide

(Al2O3 ). The Al2O3 shell has been combined with iodine to catalyze various reactions in organic chemistry and to oxidize methane into methanol (Gang et al. 2004). Flourine has been shown to react with the Al2O3 shell before Al ignites (D.T. Osborne 2006). The reactions of halogens with the Al2O3 shell leads to the hypothesis that the increased reactivity is due to interactions between the halogen and the alumina shell. An objective of this study is to examine the halogen-Al2O3 interaction in order to assess the effect on flame speed.

iv Texas Tech University, Cory Farley, May 2013

CHAPTER II: REACTION KINETICS OF NANOMETRIC ALUMINUM AND IODINE PENTOXIDE

Introduction The increase of organized terrorist cells around the world poses a growing threat to the United States and many other countries. For these terrorist cells, chemical and biological weapons make highly effective terror weapons against civilians and weapons of intimidation against soldiers (Harigel 2000). While large scale chemical weapon production requires a large chemical plant, biological weapons can be produced in basements and hospitals around the world (Harigel 2000). Of the organisms that could cause enough disease and death to cripple a region, anthrax poses one of the greatest threats (Inglesby 1999). Bioweapon attacks from agents such as anthrax would be difficult to predict, detect or prevent (Inglesby 1999). Therefore, complete elimination of the bacterial spore while in a storage bunker can effectively prevent great loss of life and psychological trauma induced from undergoing a terror attack. Popular methods for the destruction of spore forming bacteria such as anthrax involve either ultraviolet radiation (Blatchley et al. 2005) or an oxidation agent such as peroxide (Inglesby 1999). An assault on a bunker storing anthrax containers does not lend itself to a prolonged ultraviolet radiation exposure. Oxidation of anthrax spores is a slow process with necessary exposure times of up to an hour for effective neutralization (Blatchley et al. 2005).

Thermites consist of a mixture of Al and a metal oxide which produces a highly exothermic reaction when ignited (Babaitsev, Arzhevitov, and Presnakova 2007). With flame temperatures over 2000 K, thermites may act as a quick, effective sterilization tool when prolonged exposure to a neutralization agent is not a viable option.

A common oxidizer is iron (III) oxide, Fe2O3 (Fan et al. 2006); however, for bacterial sterilization I2O5 was selected due to iodine’s bactericidal properties (Taylor, Fina, and Lambert 1970). Iodine pentoxide is produced by heating iodic acid to 200 oC in a stream of dry air (Holleman and Wiberg 2001). The remaining powder is stable and produces a thermite reaction when combined with Al fuel. When I2O5 and Al react, the products are alumina (Al2O3) and iodine.

(31)

iv Texas Tech University, Cory Farley, May 2013

This study examines the kinetics for nanometer and micron scale Al particles o reacting with I2O5 in a thermal equilibrium experiment from 25 to 1000 C and in an inert argon (Ar) environment.

Experimental Method

Four powder samples were prepared consisting of nanometer scale I2O5, 80 nm Al, 15µm Al, and 40 nm Al2O3. The powder characteristics provided by the suppliers are listed in Table 1.

Table 1. Powder characteristics

Material Particle Size Manufacturer Purity (%) Al 15µm Alfa Aesar 99 Al 80nm Nanotechnologies 80 Al2O3 40nm Nanotechnologies 99.9 I2O5 Sigma Aldrich 99

Preliminary analysis showed that all the Al is not consumed at the slow heating o rates (10 C/min) investigated. In order to resolve the reaction kinetics, the Al/ I2O5 powders were mixed at varying equivalence ratios () from 0.4 to 1.2 as is defined by Eq. (32),

( ⁄ )

 (32)

( ⁄ )

where mf and mo are the mass of the fuel and oxidizer, respectively. The Al particles have an Al2O3 passivation shell accounting for the material’s impurity listed in

Table 1. The Al2O3/I2O5 powder was mixed at a 1:3.27 mass ratio (1:1 molar).

Samples for each of the powders were prepared by suspending the powders in 60cc of hexane and sonicating the mixture with a Misonix model S3000 for 70 seconds. In order to prevent damaging the oxide shell passivating the Al particles, the sonicator was programmed to cyclically mix for 10 seconds, and then to stop, allowing the mixture to cool for 10 seconds. The solutions were then placed in a glass tray under a fume hood to allow the hexane to evaporate. The powder mixtures were then reclaimed for further experimentation.

iv Texas Tech University, Cory Farley, May 2013

The thermal decomposition of each sample was studied with a Netzsch STA 409 differential scanning calorimeter and thermo-gravimetric analyzer (DSC/TGA). The system was programmed to heat the samples at a rate of 10 oCmin-1 from room temperature to 1000 oC. Sample masses of 7 mg were loaded into the sample crucible, and the DSC column was evacuated to less than 0.01 Pa using a Pfeiffer model TMU 071 P turbo molecular drag pump. The column was then backfilled with an argon atmosphere before a 50 mLmin-1 flow of argon was applied to the furnace for the rest of the heating cycle.

Results and Discussion

The results of the DSC/TG analysis for pure I2O5 can be found in Figure 3.

Figure 3. Heat flow and mass loss curves of I2O5; experiments performed in an Ar environment at a heating rate of 10°C/min.

On the left axis, the TG % represents the percent mass change occurring at a given temperature, and on the right , the DSC scale represents the energy change in mW/mg. Common examples of weight loss are the release of bonded gases, drying of the sample, and a chemical reaction producing gases. Iodine pentoxide bonds with water to become iodic acid, HIO3 (Holleman and Wiberg 2001). This reaction becomes reversible and the acid decomposes at 200 oC as shown by

iv Texas Tech University, Cory Farley, May 2013

(33)

This decomposition is illustrated in Figure 3 where the bonded water is released causing the slight mass loss and related endotherm. The I2O5 decomposition was observed at an onset temperature of 390 °C where the I2O5 fully decomposes into gas corresponding to a 100% mass loss. It is noted that I2O5 melting and decomposition are simultaneous (Holleman and Wiberg 2001).

(34)

The I2O5/ Al2O3 reaction is displayed in Figure 4.

Figure 4. Heat flow and mass loss curves for Al2O3 and I2O5, experiments performed in an Ar environment at a heating rate of 10°C/min.

By combining I2O5 and Al2O3, the kinetics between I2O5 and the Al2O3 shell encapsulating the Al particles were isolated. The main differences between Figure 3 and Figure 4 are the presence of an inflection in the mass loss curve and the presence of a third endotherm with an onset temperature of 460 °C. The inflection point in the mass loss (Figure 4) may correspond to a slight exothermic behavior causing the heat flow curve to return to the baseline slighter faster than in Figure 3. The inflection and

iv Texas Tech University, Cory Farley, May 2013

exotherm may be described by Al2O3 binding to either the I2 or O2 gas when these oxides disassociate. Since Al2O3 is inert to oxygen at this temperature, the reaction must be due to the Al2O3 binding to the iodine when freed from disassociating I2O5. This is an interesting observation and similar to a study by Osborne and Pantoya (Dustin T. Osborne and Pantoya 2007) that showed fluoride ions from Teflon decomposition bond to the alumina shell passivating Al particles. Sarbak (Sarbak 1997) showed the interaction between fluoride ions and Al2O3 may be facilitated by hydroxyls that are bonded to portion of the Al2O3 surface. Both of these studies showed the fluoride ion sorption into alumina was an exothermic reaction (Dustin T. Osborne and Pantoya 2007; Sarbak 1997). Toyohara et al (Toyohara et al. 2002) studied the iodine sorption mechanism into mixed solid alumina cement. They similarly found iodine ions replaced hydroxyls and bonded to alumina. Fluorine and iodine are halogens and this study (as well as others (Dustin T. Osborne and Pantoya 2007; Sarbak 1997; Toyohara et al. 2002)) suggests both have similar sorption behaviors when interacting with alumina.

At 550oC mass loss in Figure 4 continues again accompanied by a third endotherm. This second stage of mass loss and related endotherm may correspond to

Al2O3 phase changes from amorphous Al2O3 to - Al2O3 (M Trunov et al. 2005). The alumina phase change may trigger a release of iodine gas resulting in a subsequent mass loss.

Figure 5 shows the analysis of the reaction between micron scale Al and I2O5.

iv Texas Tech University, Cory Farley, May 2013

Figure 5. Heat flow and mass loss curves for micron scale Al and I2O5, experiments performed in an Ar environment at a heating rate of 10°C/min. Note the lack of any exothermic reaction

iv Texas Tech University, Cory Farley, May 2013

Passivated micron Al has been shown to be thermally nonreactive until it approached its melting temperature of 660oC (J. J. Granier and Pantoya 2004). Also, the lower specific surface area of the larger particles reduces the Al2O3 shell/I2O5 contact area limiting the sorption of iodine into the amorphous Al2O3. This specific surface area affect was also observed for the sorption of fluoride ions on varying specific surface area Al particles (Dikici et al. 2009). A combination of the lower specific surface area of the o Al particles coupled with the 400 C degradation of I2O5 and the slow heating rate gives the argon gas flowing through the cylinder 22 minutes to purge the oxygen and iodine gases from the system resulting in the lack of any reaction.

Nanometer scale Al, however, has been shown to be reactive in its solid state at much lower temperatures (Dustin T. Osborne and Pantoya 2007; J. J. Granier and

Pantoya 2004). This increased reactivity allows the Al to react with the decomposed I2O5 as shown in Figure 6 resulting in a complex multistep process.

Figure 6. Nanometer scale Al powder and I2O5 with an equivalence ratio of 0.4, experiments performed in an Ar environment at a heating rate of 10°C/min.

iv Texas Tech University, Cory Farley, May 2013

Similar to Figure 3, an endotherm appears at 200 oC resulting from iodic acid o releasing water and forming I2O5. Also, the endotherm at 400 C corresponding to the decomposition of I2O5 can be observed but is somewhat masked by the exothermic reaction of Al and oxygen with an onset temperature of 309 °C shown in Eq (35).

(35) Examination of the TG curve reveals a similar region to Figure 4 over 400 to 550 o C implying an interaction between the Al2O3 passivation shell and the iodine gas. Examination of the DSC curve over this temperature range reveals that the endotherm for the iodine release at 400 °C to the alumina phase change at 550 °C is partially masked by an exothermic Al-I-O reaction.

Finally, an endotherm can be seen at 660oC where unreacted Al melts.

iv Texas Tech University, Cory Farley, May 2013

CHAPTER III: REACTION KINETICS OF NANO-ENERGETICS WITH HALOGEN CONTAINING OXIDES

Introduction Halogenated oxidizers have received significant attention recently for their potential application in neutralizing spore forming bacteria via the production of halogen gases (B. R. Clark and Pantoya 2010; Sullivan et al. 2011; S. Zhang et al. 2012). Due to the electronegativity of common halogens: fluorine (4.0), chlorine (3.0), iodine (2.5) and oxygen (3.5) (Brown, Lemay, and Bursten 2000), reactions containing oxygen with either iodine or chlorine have the potential for the aluminum to react with oxygen leaving the halogen species as a product. Fluorine, while potentially effective for neutralizing bacterial spores, is simply too reactive to be utilized in this application because it bonds too quickly with surrounding gasses or aluminum leaving the spore unaffected. Therefore, iodine and chlorine containing oxides were selected for this analyses including

I2O5, AgIO3, KClO4, Ca(IO3)2 each combined with aluminum (Al) fuel. The goal is to analyze the relationship between reaction parameters for halogen containing formulations and advance our understanding of nano-energetic material composite reactivity.

It has been well documented that energetic materials undergo a transition in flame speed as particle size decreases (J. Granier and Pantoya 2004; Lynch, Krier, and Glumac 2009; Moore, Son, and Asay 2004; Perry et al. 2004; Prentice, Pantoya, and Clapsaddle 2005; Tillotson et al. 2001). With materials on a nanometer scale, flame speeds have been documented on the order of 2000m/s for Al+I2O5 (Martirosyan, Wang, and Luss 2009). While many theories exist as to why flame speeds are so high with certain formulations, a commonly accepted hypothesis is that mixtures such as Al+MoO3 and Al+I2O5 produce high flame speeds because they produce significant heat and gas that propel energy forward in a pressure wave convecting rapidly to the yet unburned reactants (Asay et al. 2004). However, in an unconfined or open-tray burn, convective effects are reduced without any confinement directing pressure to move in the direction of the unburned powder. For example, in this study, the flame speed for Al+I2O5 in an unconfined burn was found to be 916m/s as opposed to confined flame speed measurements of 2000 m/s made by Martirosyan (Martirosyan, Wang, and Luss 2009). In addition to the role of

iv Texas Tech University, Cory Farley, May 2013

confinement, there is a need for more research investigating reaction parameters and their effect on flame speed.

When considering combustion of solid powders from a fundamental perspective, an energy threshold, defined here as apparent activation energy, must be achieved in order for a reaction to initiate. Once ignited and in order for the reaction to self- propagate, the heat generated by the reaction, called heat of combustion, must be sufficient in that enough energy is transferred to the adjacent particles to elevate them to a state beyond their activation energy. In 2006, Trunov et al. showed that the smaller particle size resulted in increased reactivity as well as a reduced melting point in Al particles (MA Trunov 2006). As particle size is reduced from micrometer to nanometer scales, the apparent activation energy decreases to the point of igniting the aluminum reaction even in the solid phase at temperatures well below Al melting point at 660oC (Sanders et al. 2007). Since the transition from micron to nanoscale particles results in both an increase in flame speed as well as a decrease in apparent activation energy, one could hypothesize a correlation between the two. Also, these macroscopic combustion behaviors are linked to molecular reactions at the interfaces between grain boundaries within the system. Linking the macroscopic observations with molecular level kinetics will enable a more complete understanding of the overall reaction dynamics. Therefore, an objective of this study is to examine this correlation by experimentally comparing flame speed to apparent activation energy, gas production, and heat of combustion and further modeling molecular dynamic interactions between grain boundaries to resolve fundamental kinetics governing the overall reaction mechanism. This molecular dynamic analysis is limited to the kinetics of the fastest burning mixtures. The combined experimental/theoretical picture presented in this work shows the complexity of Al – halogen oxide reactions. At the same time, the results reveal some hallmark features of reactions involving Al fuel that includes alumina abstracting oxide species and forming intermediates that increase the exothermic yield.

Methods The Al powder has an 80nm average particle diameter and supplied by Novacentrix (formerly Nanotechnologies), Austin, TX. Particles are passivated with a 4

iv Texas Tech University, Cory Farley, May 2013

nm thick alumina shell such that the Al content is 75 % by volume. For the oxidizers,

I2O5 and KClO4 were supplied by Sigma Aldrich, St Louis MO, and AgIO3 and Ca(IO3)2 by Alfa Aesar, Ward Hill, MA. See Table 2 for powder information.

Table 2. Reactant average particle diameter, supplier and anticipated products from Al reaction with respective oxidizer Powder Diameter Supplier Anticipated Products

Al 80nm Novacentrix -

I2O5 355µm Sigma Aldrich Al2O3+I2

Ca(IO3)2 15µm Alfa Aesar Al2O3+CaI2

AgIO3 100nm Alfa Aesar Al2O3+AgI

KClO4 50µm Sigma Aldrich Al2O3+KCl

All reactions were balanced based on oxygen transfer assuming 100% of the oxygen transferred to the aluminum. Samples for each of the powders were prepared by suspending the powders in 60 cc of hexane and sonicating the mixture with a Misonix model S3000 for 70s. In order to prevent damaging the alumina shell passivating the Al particles, the sonicator was programmed to cyclically mix in intervals of 10s allowing the mixture to cool for 10s. The solutions were then placed in a glass tray under a fume hood to allow the hexane to evaporate. The powder samples were then reclaimed for further experimentation. In all experiments, loose powder was used, such that the powder samples are 5% of the theoretical maximum density (TMD).

Near Equilibrium Experiments Both thermal equilibrium and non-equilibrium tests were performed and the results compared to thermal equilibrium models (i.e. REAL code modeling software) in order to gauge how the model predicts actual energy output and relative efficiencies of each reaction. A Neztsch STA 409 differential scanning calorimeter and thermogravimetric analyzer (DSC/TGA) was used for thermal equilibrium testing. An 8mg sample was loaded into a platinum crucible with an alumina liner and placed on a sample carrier inside an oven and heated at rates of 2, 5, and 10K/min until the reaction was complete. The crucible lid is vented in order to prevent pressurization within the

iv Texas Tech University, Cory Farley, May 2013

device. Within the DSC/TGA, the sample crucible is compared to an empty reference crucible in order to obtain the net energy and mass changes. Also, the sample carrier is mounted on a microscale (i.e. TGA) allowing for mass change measurements that relay phase change (i.e. gas production) information as a function of equilibrium temperature. The slope of the DSC curve changes when the reaction within the DSC/TGA produces enough energy to become noticeable within the natural noise of the machine. The temperature at which the slope changes is defined here as the onset temperature. The area under the DSC curve corresponds with the net exothermic behavior. Apparent activation energy (Ea) can be determined from the DSC measurements using the Ahhrenius equation in Eq. (1) that is described as the Type B-1.95 Peak Method described by M. Starink (Starink 2004), the procedure for which can be found in ASTM 698E (ASTM E698 2011).

In this method, the reaction rate is estimated as the ratio of heating rate (B) to temperature at which the reaction is at its peak (Tp) and R is the universal gas constant. The pre-exponential factor, A, is a constant also expressed as C in Eq. (2).

( ) (36)

By taking the natural log of both sides, can be calculated by plotting the

( ) vs for each of the heating rates and finding the slope of the best fit line.

( ) (37)

Non-Equilibrium Experiments For the non-equilibrium tests, a Parr model 1341 bomb calorimeter was used with a model 1108 oxygen combustion bomb chamber as seen in Figure 7. Using REAL code (Timtec L.L.C.) values for theoretical heats of combustion based on the heat of formation for the reactants and products, powder mass was determined in order to obtain a 0.5 to 1K elevation of a 2kg water bath above the calibrated energy inputs of a stirrer and other environmental factors. A thermocouple within the water bath was sampled at a rate of 10 Hz for 40 minutes. Using a 60% value between the lines extrapolated before and after

iv Texas Tech University, Cory Farley, May 2013

the reaction on a temperature as a function of time plot, a change in temperature was obtained in order to find the energy associated with heating the steel bomb chamber and water.

Figure 7. Schematic of bomb calorimetry experiment.

For the flame speed experiments, an average of 500 mg of loose powder was loaded into a 0.5 by 13cm notch and distributed evenly with no packing. By taking the volume of the notch and the mass of powder added, %TMD was calculated as an average of 4.5%. Nickel-chromium wire, Ni80-020, from Omega Engineering, Inc was placed at one end of the notch and heated at 7V by a variac transformer. Once the powder ignited, flame speeds were recorded via a Phantom 7 high speed camera at frame rates ranging from 900 to 90000 frames per second as seen in Figure 8. National Instruments Vision Builder AI 2010 software was used to locate the flame front in the high speed images and when combined with the frame rate, allows for the calculation of flame speed.

Figure 8. Schematic for open tray flame speed measurements.

iv Texas Tech University, Cory Farley, May 2013

iv Texas Tech University, Cory Farley, May 2013

Ab Initio Quantum Chemical Calculations

All simulations were performed for the Al – I2O5 system. This reaction can be described by chemical events at the Al grain boundaries. The role of grain boundaries in initiation of chemical reactions is complex. However, if the events are broken down in solid-state and gas-phase reactions, the state of initial and final phases of reactants are important aspects of the process. In order to use first-principles calculated thermokinetics to describe the complex series of elementary reactions, the following assumptions are applied. First, Al particles are covered with an Al2O3 oxide layer. This oxide layer is often amorphous but modeling is performed using stable -Al2O3 as the ground state surface. In general, the first-principles calculated reactions show the upper limit for kinetic barriers for such reactions. Rate observed is higher as Al2O3 layers containing defects and disorders are more reactive than ideal surface morphology used in the current work.(Wang et al. 1999; Zhai and Chen 2011; L. Zhang et al. 1995) Thermal excitations at the Al/Al2O3 grain boundaries in presence of oxidizers are a primary focus when studying aluminized reactive composites. The working hypothesis of the simulations presented in this work assumes that upon activation of reactions at the

Al/Al2O3/I2O5 interfaces, exothermic chemical reactions simultaneously weaken the oxide layer and increase local temperatures. This mechanothermal load present during the initiation stages can lead to melting of core Al, rupture of oxide layer, acceleration due to faster Al-I2O5 reactions and related species, and eventual ignition or detonation depending on the reaction conditions.

An objective of this analyses is to explore the Al2O3-I2O5 interfaces using first- principles calculated initial adsorption (exothermic) and possible I-transfer (activated endothermic) processes at the interface using condensed-phase periodic density functional theory (DFT) calculations. The gas-phase thermokinetics of the reactions are determined by three primary groups of coupled reactions: (a) Al released undergoes exothermic oxidation in air producing Al2O3 (b) I2O5 thermally decomposes at temperatures produced by oxidation reactions and (c) Al reacts with I2O5 decomposition fragments. Such reactions are studied for their reaction enthalpies using gas-phase quantum chemical theories at various levels including spin-orbit DFT (SO-DFT) level theories needed for I2O5 decomposition reactions.

iv Texas Tech University, Cory Farley, May 2013

In the present work we describe DFT and SO-DFT quantum chemical calculations on a selection of iodine oxide molecules and their reactions with aluminum which are important initial steps in the mechanistic understanding of Al/I2O5 thermite reactions. First, the elementary steps and the associated thermochemistry for the dissociation of the - I2O2, I2O3, I2O4, and I2O5 oxides. Then , reactions between the [Al13] cluster and iodine oxide dissociation fragments will be discussed. Our recent theoretical study (Losada and Chaudhuri 2010) of the finite size effects on aluminum / polytetrafluoroethylene (PTFE) - reactions showed that the [Al13] cluster provides a suitable model for some aspects of the surface of bulk Al or nanoscale Al in reactive composites. This system is the smallest one in which an Al atom can have a coordination of 12 of a close-packed bulk crystal and the surface atoms mimic the triangular Al arrangement on an Al(111) surface termination. In - addition, [Al13] has been proved, both experimentally and theoretically, (Hettich 1989; Rao and Jena 1999) to be a stable system. Its stability arises from closures of its geometry and electronic shells. With heavy element iodine present in determining the chemical reactivity in Al/I2O5 composites, it is necessary to include relativistic effects when calculating the thermochemistry of these systems.

Solid State Calculations

An -Al2O3 layer oriented in (100) direction was used for studying I2O5 adsorption. An ultrasoft pseudopotential and revised Perdew Burke Ernzerhof (RPBE) exchange correlation functional as implemented in DFT code CASTEP was used for all calculations. (S. J. Clark et al. 2005; Segall et al. 2002) The slab thickness varied until convergence was achieved at kinetic energy cut-off of (300 eV) and a surface layer consisted of 160 atoms in a 9.83 x 9.83 Å surface in an approximately 16 Å thick slab. Due to this large size of the unit cell needed to accommodate the precision in surface energy and relatively large I2O5 molecule, -point calculations were sufficient to obtain convergence with respect to known surface energy values for Al2O3. An I2O5 molecule was first optimized using gas-phase DFT calculations (described below). The optimized

I2O5 molecule was allowed to relax in a same size period box for RPBE calculated total energy for a gas-phase molecule (EIO). The I2O5 molecule was allowed to adsorb on

Al2O3 (100). The following energy difference was used to estimate absorption energy:

iv Texas Tech University, Cory Farley, May 2013

Eads=ESurface+EIO –ESurface+IO. The I2O5 adsorbed on Al2O3 surface using a strong I-O-Al bridge bond.



Gas-phase Calculations. In selecting a suitable and effective methodology for the treatment of the molecular systems of I2O5 and fragments considered, which contain the heavy iodine atom, oxygen and aluminum, the choice of appropriate basis set is crucial for the accuracy of the calculations. Thus, correlation consistent basis sets were used. More specifically, the diffuse-augmented sets from double- to quadruple-zeta were used for oxygen, aug-cc-pVnZ (n = D, T, Q), (Dunning 1989; Kendall, Jr, and Harrison 1992) while those based on accurate relativistic small-core pseudopotentials/correlation consistent basis sets developed by Peterson et al.(Peterson et al. 2006) were taken for iodine, aug-cc-pVnZ-PP (n = D, T, Q). In the following, the combination of the aug-cc- pVnZ on oxygen and aug-cc-pVnZ-PP on iodine is abbreviated as aVnZ.

The B3LYP, (C. Lee, Yang, and Parr 1988; Becke 1993) PBE0, (Adamo and Barone 1999) ACM, (Becke 1993) M05, (Zhao and Truhlar 2004) and M06-L (Zhao and Truhlar 2006) functionals were used. Computational studies of the enthalpies of formation of OXO(S. Y. Lee 2004) (X = Cl, Br, and I) showed that the B3LYP functional delivers good results when compared with CCSD(T) correlated calculations. As for the PBE0 and ACM functionals, the investigation of Lee and co-workers showed that these functionals performed reasonably well for molecular geometries and vibrational frequencies in iodine-containing molecules.(Kim, Ihee, and Lee 2010) Moreover, the performance of M05 and M06-L for iodine oxides has not been assessed for these oxide molecules. For these reasons, we chose these five DFT functionals for the present study. Structure optimizations and harmonic frequencies were calculated at the corresponding levels for the aVDZ and aVTZ basis. Zero-point energy (ZPE) and enthalpy corrections for the aVQZ level were evaluated using the harmonic frequencies calculated at the aVTZ level for each functional.

In addition to the scalar relativistic effects which are taken into account by using the relativistic small-core pseudopotential/correlation consistent basis sets, a second

iv Texas Tech University, Cory Farley, May 2013

relativistic correction, spin-orbit (SO), was applied to all the iodine oxide molecules in this work. The effects due to SO were recovered by use of the SO-DFT method as implemented in the NWChem program.(Valiev et al. 2010) SO effects were calculated using the aug-cc-pVnZ-PP SO (n = D, T, Q) basis set on iodine and the corresponding aug-cc-pVnZ on oxygen for each DFT functional considered. We used uncontracted basis for the SO-DFT calculations to get a better description of the SO effects, as pointed out by Lee and co-workers.(Kim, Ihee, and Lee 2010)

Heats of formation were calculated by combining our computed atomization energy values with the known enthalpies of formation ( fH) at 0 K for oxygen and iodine, fH (O) = 58.98 kcal/mol and fH (I) = 25.61 kcal/mol.(Grant et al. 2010) Enthalpies of formation at 298 K were obtained by following the procedures outlined by Curtis et al.(Curtiss et al. 1997)

For the calculations of the aluminum-iodine oxide polyatomics, we used the - [Al13] cluster as the model system. Geometries were optimized using the all-electron aug-cc-pVTZ basis for oxygen and the aug-cc-pVTZ-PP for iodine in conjunction with the 6-311G basis set for aluminum. This level of theory was previously used for the - description of the [Al13] -Teflon combustion reactions.(Losada and Chaudhuri 2010) Because of the computational expense of geometry optimization and frequency calculations with aug-cc-pVQZ basis sets for these large systems, the aug-cc-pVTZ geometries were used in single point aug-cc-pVQZ calculations. Harmonic frequencies were calculated using the aug-cc-pVTZ on oxygen, aug-cc-pVTZ-PP on iodine, and 6- 311G basis set on aluminum. No symmetry constraints were imposed in the optimization of all the molecular species. All calculations for the aluminum-iodine oxide polyatomics were carried out with the Gaussian 09 suite of programs and the SO-DFT calculations with the NWChem program.

iv Texas Tech University, Cory Farley, May 2013

Results

Figure 9. Heat flow measurements for Al+I2O5 scanned at 2, 5 and 10 K/min heating rates.

Figure 10 shows ( ) as a function of , the resulting slope is Ea .

Figure 10. Peak temperatures for Al+I2O5 plotted using the Type B-1.95 Peak Method. The resulting Ea is 175.6 kJ/mol

iv Texas Tech University, Cory Farley, May 2013

The activation energy can be converted to a mass basis since applied energy is commonly considered in the same physical sense. For this reason, Ea for each material is converted to kJ/g in . Uncertainty for this method in calculating Ea is largely based on repeatability in finding the peak temperatures in the DSC, and for our device, Ea values have been within 15%.

The results for a typical bomb calorimetry experiment are shown in Figure 11.

Figure 11. Graph showing Al+I2O5 Parr bomb calorimetry result displaying dT

In order to calculate the heat of combustion, Hc = CpdT was used where Hc is the heat of combustion, Cp is the specific heat of the water bath, and dT is the change in temperature as seen in Figure 5. The specific heat was modeled to include a mass percentage of the specific heats of the steel container (3.838 kg, 473 J/kgK) and the water bath (2.00kg, 4184J/kgK) for a final value of 10.18kJ/K. Also, Hc was found for each mixture and was found to be 3.68±0.22 kJ/g for Al+I2O5, 2.64±0.20 for Al+Ca(IO3)2,

2.67±0.30 for Al+AgIO3, and 3.62±0.16 for Al+KClO4. The average values are shown in Table 3.

Representative still frame images of the flame propagating are shown in Figure 12.

iv Texas Tech University, Cory Farley, May 2013

Figure 12. Open tray still frame images A) Al+I2O5 taken at 33µs intervals, B) Al+Ca(IO3)2 at 17ms intervals, C) Al+AgIO3 at 40 µs intervals, D) Al+Fe2O3 at 54ms intervals.

The resulting analysis from the luminescent front signaling flame propagation is displayed below in Figure 13.

Figure 13. Position of flame front as a function of time for one test for Al+I2O5 with a flame speed of 898 m/s.

For the fast class of powders, Al+I2O5 and Al+AgIO3, the average flame speed was found to be 916±53 and 926±29 m/s, respectively. For the slow class of powders,

iv Texas Tech University, Cory Farley, May 2013

Al+KClO4 and Al+Ca(IO3)2, the average flame speed was found to be 0.77±0.09 and 2.50±0.92 m/s, respectively. The results for the average flame speed for each mixture have been tabulated in Table 3.

Table 3 displays the compilation for the experimental results over the three different testing protocols for the four reaction chemistries examined. The heat of combustion from a thermal equilibrium code (REAL), an equilibrium diagnostic (DSC), and a non-equilibrium method (Bomb) are displayed as well as mass loss from the TGA, apparent activation energy (Ea), onset temperature, and average flame speed.

Table 3. Summary of experimental results. Onset Average TGA Mass Heat of Combustion, E Reactants a Temperature Speed Loss (%) (kJ/g) (kJ/g) (oC) (m/s) REAL DSC Bomb

10Al+3I2O5 95 6.14 1.13 3.68 1.79 310 916

4Al+Ca(IO3)2 75 7.81 0.18 2.64 2.78 511 2.5

2Al+AgIO3 35 5.02 1.49 2.67 1.84 309 926

8Al+3KClO4 0 10.63 2.31 3.62 3.62 514 0.8

The results from theoretical calculations are summarized in Table 4 and Table 5. The thermal decomposition enthalpies for different iodine clusters calculated at B3LYP and PBE0 level are provided in Table 4. The reactions are all exothermic when O2 is produced from an iodine oxide cluster. So, even if iodine oxide’s role is limited to producing I2 and O2, Table 5 shows that such steps also add to the highly exothermic nature of the Al/I2O5 reaction.

iv Texas Tech University, Cory Farley, May 2013

Table 4. Decomposition of iodine oxide clusters in gas phase for B3LYP, PBE0 and ACM functionals at aug-cc-pVTZ (aVTZ) and aug-cc-pVQZ (aVQZ) levels of theory. The values in the bracket are calculated using spin-orbit (SO) corrections.

     

→

→

→

→

→

→

→

→

→

→

iv Texas Tech University, Cory Farley, May 2013

As concentration of gas-phase I2O5 is limited in high temperatures, reactions with smaller fragments are shown in Table 5.

Table 5. The reaction enthalpy and free energy calculated for Al reactions with iodine oxide fragments calculated at aug-cc-pVTZ (aVTZ) and aug-cc-pVQZ (aVQZ)

   

→

→

→

→

→

→

→

→

→

→

→

→

iv Texas Tech University, Cory Farley, May 2013

→

→

→

→

→

→

Discussion One of the first points to notice is that the heats of combustion are much lower in equilibrium experiments than in non-equilibrium and both of these measurements are significantly lower than predicted by the modeling software. For the equilibrium study (DSC), the difference can be explained by the reactants or intermediates gasifying and leaving the system prior to reaction, thus not contributing to the heat production. The off gassing cannot be calibrated by comparison to a known standard because different reactions will experience phase changes at different paces. Therefore, non-equilibrium tests provide a more accurate indication of the actual heat of combustion for a given reaction.

From Table 3 there is not a direct correlation to flame speed for mass loss or heat of combustion. By using the thermo-equilibrium TGA experiments, the gas generation for a reaction is approximated to an order of scale. If a convective flow was actually driving the unconfined reaction to produce the higher flame speeds, one would expect to see high mass loss and a large heat of combustion. However, the only factor which corresponds to a fast burning powder is apparent activation energy.

iv Texas Tech University, Cory Farley, May 2013

If activation energy as defined by the Arrhenius equation was the primary controller for flame speed, an exponential relationship would be expected between the reaction rate and Ea. For the four mixtures chosen for this study, changes in Ea result in order of magnitude changes in flame speed. Other factors may be contributing to this relationship, but the dramatic changes in flame speed correlating to small changes in Ea would imply that the chemical kinetics of the reaction are the primary drivers for flame speed in open tray burns instead of a physical limiting agent such as diffusion rates, mass transfer, etc. For this reason, the remainder of this section is focused on the intermediate chemistry influencing the high flame speeds observed in Table 3.

Halogens have been shown to both bond to alumina (Farley and Pantoya 2010; Saikia et al. 2009; Watson, Pantoya, and Levitas 2008) and accelerate aluminum oxidation (Gang et al. 2004). For example, Osborne et al. described a pre-ignition (PIR) reaction caused by the fluorination of the alumina shell (Dustin T. Osborne and Pantoya 2007). They describe an interaction of fluorine with the hydroxyl ions boned to the alumina surface and theorized that the PIR acted to accelerate the reaction by assisting in the decomposition of Teflon. Iodine has also been shown to react with hydroxyl ions on the surface of alumina (Flockhart, Liew, and Pink 1974), the surface reaction itself may contribute to the formation of species that may accelerate the reaction such as I- and IO- , F- (Flockhart, Liew, and Pink 1974; Pagni, Kabalka, and Boothe 1988). In 2010, Farley et al. displayed a bonding of iodine with alumina in a thermo-equilibrium setting between o 400 and 500 C by reacting Al2O3 with I2O5 (Farley and Pantoya 2010). In the case of reaction involving halogenated oxides, the existence of free, unbonded halogen gas could lead to the halogen bonding to alumina and accelerating the reaction similar to previous observations (Farley and Pantoya 2010; Saikia et al. 2009; Watson, Pantoya, and Levitas 2008; Gang et al. 2004). The effect may contribute to reduced activation energy and increased reaction kinetics as seen in the faster burning mixtures. In order to determine the likelihood of the existence of unbonded halogen gases, a comparative analysis involving the heat of formation of halides was performed.

The I2O5 adsorption on Al2O3 is strong with bridge bonds forming between oxygen from I2O5 and surface Al leading to two I-O-Al bonds. It is possible that the initial composite may not have the ability to overcome the kinetic barrier to form these

iv Texas Tech University, Cory Farley, May 2013

bonds. However, the adsorption energy (Eads) is highly exothermic and calculated to be -

422.99 kJ/mol for a single I2O5 molecule, or nearly -211.495 kJ/mol per I-O-Al bond formed. This estimate is important as related I2O5 decomposition fragments such as IO3 and IO2 will also form single I-O-Al bridge bonds for adsorption. In order to compare the adsorption energy and provide context to this result, adsorption of a water molecule on

Al2O3 is exothermic by -140 kJ/mol on -Al2O3. It is noted that most oxide surfaces are covered in water in atmospheric conditions. However, I2O5 has a much stronger adsorption enthalpy and therefore can replace water provided it has enough thermal energy. At around 200oC this process may start at the interface and proceeds to replace water. The reactions can be due to I2O5 or HIO3 (hydrated intermediate) and will provide a similar exothermic response in Al/I2O3 .

On transition from solid-phase reactions to gas-phase reactions, different gas- phase species can be present. If we consider that the I2O5 and Al channels are decoupled except some chemisorption of I2O5 at the Al2O3/I2O5 boundaries, the majority of the exothermic yield will occur upon thermal decomposition of I2O5; consider the following steps: I2O5 → IO2 + IO3 → I2O3 + O2. The O2 thus produced can subsequently react with

Al. As Al+O2→Al2O3 reaction happens at much faster rate, the I2O5 can act as accelerant.

The first important result for Al reacting with iodine oxide species as evident from Table 5 is that most of these reactions are highly exothermic. Secondly, Al can act as an effective intermediate to aid decomposition of iodine oxide species and move the oxidation reaction forward. At the same time, steps in which Al-I species are produced are less exothermic than Al-O species formation. Therefore, Al acts as an intermediate that aids the production of I and I2 in the gas-phase reaction.

The trends in reactions from Table 4 and Table 5 indicate that the Al/I2O5 has the following behaviors:

(1) I2O5 provides multiple exothermic channels that contribute to the next exothermicity;

(2) many of the steps of thermal decomposition produce O2 that sustains the combustion process and increases reaction temperature; and,

iv Texas Tech University, Cory Farley, May 2013

(3) Al-O species are highly stable and a method is shown in which Al-IO or Al-O species converts to Al2O3 and produces more I2 as the gas-phase product.

These chemical kinetic steps may be the controlling factors accelerating the Al-

I2O5 and similarly the Al-AgIO3 reactions but cannot be applied to the Al-Ca(IO3)2 and

Al-KClO4 reactions because their onset temperatures are in excess of 200 degrees greater than for the Al-I2O5 and Al-AgIO3 oxidizers.

iv Texas Tech University, Cory Farley, May 2013

CHAPTER IV: THE CONTRIBUTION OF ATMOSPHERIC OXYGEN TO ENERGY PROPAGATION IN REACTIVE MATERIALS .

Introduction Reactive materials are a class of energetic materials that combine Al fuel with a solid oxidizer such that the powder mixture is a composite energetic material. When the particle size of the fuel and oxidizer include one geometric dimension on the nanoscale, the composite may be referred to as a nano-composite reactive material. Various studies have examined the effect of atmosphere on aluminum (Al) combustion either as an aerosolized powder or single particles(Huang et al. 2009; Gill, Badiola, and Dreizin 2010; Beckstead, Liang, and Pudduppakkam 2005; Friedman and Maček 1963). However, there have been minimal studies examining the effect of environment on flame speeds of reactive materials. Yet, in practice, reactive materials are often used under atmospheric conditions such that the role of oxygen in the environment influencing the combustion should be well understood.

Weismiller and Malchi studied the effects of atmospheric pressure on the flame speeds of nanometric Al and copper oxide (CuO) mixtures(M. Weismiller et al. 2009). Their study observed the effect of adding argon, nitrogen or helium up to 15MPa and noted a shift from convective to conductive heat transfer modes at high pressures(M. Weismiller et al. 2009). By adding inert gas to the environment, the partial pressure of oxygen was reduced, thereby creating a low oxygen environment at an elevated pressure. They showed that in the low oxygen high pressure environment flame speeds decreases by two orders of magnitude. However, the change in flame speed for this study was attributed to the pressure change, not environmental gas. In another study, Asay et al. observed flame speeds of nanometric Al with molybdenum trioxide (MoO3) under vacuum conditions in a tube(Asay et al. 2004). A vacuum environment is, by definition, a low oxygen atmosphere, but at a low pressure, and flame speed was found to be unaffected. Results from these two studies pose questions regarding the role of oxygen in the environment on combustion at atmospheric conditions.

iv Texas Tech University, Cory Farley, May 2013

Two widely studied reactions are Al+CuO and Al with iron (III) oxide (Fe2O3). Studies have shown that Al+CuO is a gas generating, convectively driven, relatively fast burning mixture ( i.e., flame speeds greater than 100m/s)(Umbrajkar, Schoenitz, and Dreizin 2006; Sanders et al. 2007; Bockmon et al. 2005; Jian et al. 2013; Prakash, McCormick, and Zachariah 2005; M. R. Weismiller, Lee, and Yetter 2011). Theoretically, the gas production from this mixture could reach 0.54mol per 100g of powder burned(Fischer and Grubelich 1998). In contrast, Al+Fe2O3 is a conductively driven reaction with little gas production, e.g., 0.14mol per 100g of powder burned(Fischer and Grubelich 1998). Also, due to the conductive nature of the reaction, flame speeds are considerably slower for this formulation than the Al+CuO (i.e., on the order of 1m/s)(Crane, Pantoya, and Dunn 2010).

Jian et al.(Jian et al. 2013) conducted a study of various powders and correlated ignition temperatures to oxidizer decomposition temperatures. According to their study, the mixtures used here ignite when their oxidizer begins to decompose. For example,

CuO decomposes to Cu2O at 1000 K and Fe2O3 decomposes to Fe3O4 at 1400K(Jian et al. 2013). When CuO decomposes, 10% of its mass converts to free oxygen, O, while only

6% of the Fe2O3 results in free oxygen. Therefore, the decomposition of CuO results in more oxygen available to react with Al, and Umbrajkar et al, showed that oxygen, when freed from a decomposing oxide, prefers reacting with Al instead of forming

O2(Umbrajkar, Schoenitz, and Dreizin 2006). Also, ignition of nanometric Al in a gaseous oxygen environment occurs at 1000K.(Huang et al. 2009)

In addition to these two well-studied mixtures, Al+I2O5 and Al+Ca(IO3)2 are also included in this study. The Al+I2O5 reaction has recently sparked interest due to its highly convective nature (0.62 mol per 100g burned)(Fischer and Grubelich 1998) and extreme flame speeds of up to 2000m/s in a confined burn(Martirosyan, Wang, and Luss 2009). This mixture is rather unique in that the oxidizer interacts with the alumina shell causing the shell to become an active participant in the reaction instead of just being a diffusion barrier or heat sink(Farley and Pantoya 2010). Also since part of the combustion products are I2 gas, this mixture has also been studied as a biocidal agent(B. R. Clark and Pantoya

2010). The Al+Ca(IO3)2 reaction has been found to initiate upon oxidizer decomposition

iv Texas Tech University, Cory Farley, May 2013

at 825K and 1025K and Ca(IO3)2 has been found to decompose as follows in Eqs. (38) and (39). At about 825K, Ca(IO3)2 initially decomposes(K. Stern 1974).

( ) ( ) (38)

iv Texas Tech University, Cory Farley, May 2013

Then at 1050K the decomposition undergoes a second step.

( ) (39)

The objective of this study is to examine the influence of atmospheric oxygen on flame speeds for two convectively dominant reactions: Al+CuO and Al+I2O5, and two conductively dominant reactions: Al+Ca(IO3)2 and Al+Fe2O3. All reactions initiate when oxygen is released from the oxide and is available to react with the Al, however each oxide dissociates at different temperatures either before or after Al melting and when Al becomes reactive with atmospheric oxygen at 1000K(Huang et al. 2009). The objective will be accomplished through high speed imaging of reaction propagation defined as flame speed in controlled oxygen concentration environments.

Methods Passivated aluminum (Al) with an average diameter of 80nm and an active Al content of approximately 75wt% was supplied by Novacentrix (formerly Nanotechnologies), Austin, TX. These particles were used for all tests and mixed with either CuO or Fe2O3 supplied by Sigma Aldrich, St Louis MO or Alfa Aesar, Ward Hill,

MA, respectively. All particles except I2O5 have a spherical morphology. The average particle diameter of CuO is 50nm; Ca(IO3)2 is 15 microns; Fe2O3 is 40 microns; and, I2O5 has a flake morphology with the largest dimension no greater than 45 microns. All reactions were balanced based on oxygen transfer assuming 100% of the oxygen transferred to the Al.

Individual reactant powders were measured for equivalence ratios of 1 and 1.2, corresponding to stoichiometric and slightly fuel rich. The powders were then suspended in 60 cc of hexane and sonicated with a Misonix model S3000 for 70 s with 10 s rest intervals in order to prevent heating the sample. The solutions were then placed in a glass tray under a fume hood to allow the hexane to evaporate before the powders were reclaimed for further experimentation.

Loose powder was loaded into a 0.5 by 13cm notch and distributed evenly with no packing. By considering the volume of the notch and the mass of the loaded powder, the bulk density of the sample ranged from 4 to 8% of the theoretical maximum density (TMD) of the solid mixture meaning 92 to 96% of the volume in the notch is occupied by

iv Texas Tech University, Cory Farley, May 2013

air. The mass of the powder varied between 335-720 mg depending on the composite used to maintain a constant bulk density. For each test, three experiments were performed to establish repeatability which was found to be the largest source of uncertainty in the data.

The experimental setup shown in Figure 14 Error! Reference source not found.consists of a sealed combustion chamber with a blow-off valve to prevent pressurization. Two roughing pumps are connected to the sealed combustion chamber and one third of the atmosphere was vacuumed out. Either argon (Ar) or oxygen (O2) was pumped into the chamber depending on the experiment. This atmospheric cycle was repeated 4 times for the argon and 6 times for the oxygen as seen in Eqs. (40) and (41).

( ) (40)

( ) ( ) (41)

Nickel-chromium wire, Ni80-020, from Omega Engineering, Inc was placed at one end of the notch and heated at 7V by a Variac transformer to provide ignition for the reaction. Flame speeds were recorded via a Phantom 7 high speed camera at frame rates ranging from 900 to 90000 frames per second aligned perpendicular to the direction of flame propagation (Figure 14). National Instruments Vision Builder AI 2010 software was used to locate the flame front in the high speed images. Initially, a threshold filter was applied to each frame and the pixel location corresponding to the edge of bright light was recorded. The pixels were converted to millimeters via calibration and combined with the frame rate to calculate flame speed.

iv Texas Tech University, Cory Farley, May 2013

Figure 14. Top view schematic for open tray flame speed experiments.

Results Gas generation calculations for each reaction were computed using REAL code software. The calculations were performed for atmospheric pressure conditions and the gas production did not vary significantly with stoichiometry such that Table 6 reports the stoichiometric simulations for comparison.

Table 6. Gas production from REAL Code analyses

Al+CuO Al+I2O5 Al+Ca(IO3)2 Al+Fe2O3

24% 75% 71% 11%

A representative sequence of still frame images captured from the high speed recording for each mixture is shown in Figure 15. The position versus time data are linear for each speed measured indicating steady propagation.

iv Texas Tech University, Cory Farley, May 2013

Figure 15. A) Al+Ca(IO3)2 at 6.9ms intervals. B) Al+Fe2O3 at 6.9ms intervals. C)

Al+I2O5 at 44µs intervals. D) Al+CuO at 121µs intervals.

Table 7 displays the flame speeds for each reaction under the various conditions. Flame speeds are organized by equivalence ratio (ϕ) and atmosphere for the four separate testing conditions. In order to compare flame speeds for the variety of conditions, percent difference was used as defined in Eq. (42) where FSH is the flame speed under high O2 conditions and FSL is the flame speed for low O2 conditions.

| | (42)

Table 7. Summary of flame speed results in units of m/s.

Al+I2O5 Al+CuO Al+Ca(IO3)2 Al+Fe2O3

ϕ=1.0 ϕ=1.2 ϕ=1.0 ϕ=1.2 ϕ=1.0 ϕ=1.2 ϕ=1.0 ϕ=1.2

840±4 700±3 224± 408±1 2.9±0.0 5.4±0. High O2 4.5±0.7 3.4±0.1 0 5 4 5 4 2

770±2 670±6 204± 320±2 2.2±0.0 2.9±0.0 1.7±0.0 1.8±0. Low O2 0 0 9 5 5 1 4 1

% 9% 4% 10% 30% 30% 55% 100% 200% Increase

iv Texas Tech University, Cory Farley, May 2013

Discussion The flame speeds shown in Table 7 suggests combustion of solid fuel-oxidizer mixtures in an oxygen rich environment occur in different ways. Based on the literature(Umbrajkar, Schoenitz, and Dreizin 2006; J. H. Stern and Passchier 1962; Jian et al. 2013; Fischer and Grubelich 1998) and observations from Table 7, Figure 16 illustrates a flow chart describing the mechanisms promoting flame speed. For solid oxidizers with decomposition temperatures below 1000K, the first stage is the decomposition of the metal oxide. Next, stage two is marked by the ignition of Al with oxygen freed from the solid oxidizer. The heat from this reaction will increase temperatures and provide energy to dissociate atmospheric O2 to O. As Al particles reach 1000K, stage three is designated by reactions of Al with atmospheric oxygen. The heat generated from these reactions increases local temperatures to over 3000K(Dreizin et al. 2011), such that stage four occurs when the rest of the metal oxide decomposes and provides additional energy to dissociate atmospheric O2 to O. it is noted that the heat of formation for O from O2 is endothermic and 249 kJ/mol, while the heat of formation of

Al2O3 is -1670 kJ/mol(Cox, Wagman, and Medvedev 1987). In stage five, the remaining Al then reacts with the available oxygen from the solid decomposition and the atmosphere. Finally, with sufficient oxygen available, reduced metal products will then react with remaining atmospheric oxygen. All of the stages involved in producing the convectively dominant fastest flame speeds are shown graphically in Figure 16.

In a low oxygen environment, the oxide will still decompose such that oxygen is available to initiate combustion by 1000K. Stages three-five continue to occur, with less atmospheric oxygen participating, less convective influence and slightly reduced flame speeds. A slower conductively driven reaction such as Al+Fe2O3 follows the flow chart as well, but with higher decomposition temperatures, the mechanism limits stages one and two.

iv Texas Tech University, Cory Farley, May 2013

Stage 1: Decomposition of solid oxidizer below 1000K

Stage 2: Reaction of Al with dissociated oxygen

Stage 3: Reaction of Al with atmospheric O2 at 1000K

Stage 4: Further decomposition/reaction >1000K

Stage 5: Reaction of reduced metal products

Figure 16. Flowchart of fast convective reactions

Iodine pentoxide

Iodine pentoxide decomposes at 663K into IO3 and IO2(Farley and Pantoya 2010).

These two compounds bond exothermically to the passivating Al2O3 shell on the nAl particles. The IO3 and IO2 compounds then further disassociate at 730K into I and O gas reacting with Al to form Al2O3 and I2. At this point, a fully decomposed I2O5 molecule produces 100% total gas of which 24% is oxygen by mass. From this sequence of stages in Figure 16, nearly all of the oxygen from I2O5 becomes available to react in stage one. As a result of the low decomposition temperature and complete gas disassociation of the

I2O5 particle, this reaction is highly convective with large pressure increases and extremely high flame speeds (670-840 m/s) regardless of atmospheric oxygen concentration. Adding or removing atmospheric oxygen into this reaction affect heat generation by reacting with Al that is left unreacted. As a result, changing the atmospheric oxygen content from 4-93% increases the flame speed by only 9% for the stoichiometric mixture. It should be noted that the flame speeds for the fuel rich (ϕ=1.2) mixtures have slower flame speeds than the stoichiometric mixtures. This is possibly due

iv Texas Tech University, Cory Farley, May 2013

to a reduced amount of IO2 and IO3 radicals to bond to the alumina layer reducing the early onset of heat generation (stages one and two) and reducing the flame speed.

Copper oxide In view of the five stage approach to understanding atmospheric oxygen effects and flame speeds (Figure 16), while Al+I2O5 is largely a stage one and two reaction, Al+CuO utilizes all five stages with an emphasis on stages one-three. As previously discussed, the initial decomposition of CuO→Cu2O+O begins at 1000K, releasing 50% of the oxygen. This temperature also corresponds to Al particles initially reacting with atmospheric O2 and published ignition values for Al+CuO(Jian et al. 2013). In other words, Al+CuO in a sense, compresses stages one-three into a narrow temperature band at 1000K. Afterwards, Cu2O decomposes and reacts in stage four and Cu reacts with atmospheric O2 in stage five to help propagate the reaction. This reaction has long been considered convective producing a lot of gas (25% from thermal equilibrium code) with very high flame speeds (200-400 m/s). Since a large portion of this reaction begins at lower temperatures (stage three), the reaction is less affected by atmospheric oxygen (10-

30%) than Al+Fe2O3 and Al+Ca(IO3)2, but more affected than Al+I2O5. Increasing the content of oxygen surrounding the mixture increases the effects of stages three-five and provide more heat and product gas to convect the reaction forward. Higher equivalence ratios (ϕ=1.2) result in slightly higher gas production and higher flame speeds. Increasing oxygen concentration for ϕ=1.2 mixtures results in the largest increase in flame speed for this mixture due to the exaggerated effect of the initial stage since more Al is available to react to the O2.

Calcium Iodate As a result of the two stage decomposition as seen in Eqs. (38) and (39), this reaction follows the stages in a similar fashion to Al+CuO. Roughly 50% of the oxygen content is released right before the Al powder reacts at 1000K. Then, as the temperature increases, stage four occurs with the second stage of the decomposition. If this five stage process explaining the effect of atmospheric oxygen was all encompassing, then the flame speed for Al+CuO and Al+Ca(IO3)2 would be expected to be the same.

However, Ca(IO3)2 and Ca5(IO6)2 (Hf = 2.57 and 5.92 kJ/g)(K. Stern 1974) are much

iv Texas Tech University, Cory Farley, May 2013

more stable compounds than CuO and Cu2O (Hf = 1.95 and 1.17 kJ/g)(Sottery 1985). As a result, even though the decompositions occur at similar temperatures and follow similar steps, the heat absorbed in order to decompose both materials are drastically different. As a result flame speeds are reduced by the decrease in latent heating to the surrounding material. By increasing atmospheric oxygen, more heat can be generated from reacting Al to help compensate for the large amount of heat absorbed in the decomposition process. As a result, flame speeds are more sharply increased as compared to Al+CuO (30-55%).

Combining more Al (ϕ=1.2) and a high level of O2 results in the highest heat output and similarly highest flame speed.

Iron oxide

The decomposition of Fe2O3 occurs at 1400K such that the initial phase for

Al+Fe2O3 favors Al reacting with atmospheric oxygen to produce energy promoting

Fe2O3 decomposition. As a result, this reaction is largely a stage three-five reaction. The role of atmospheric oxygen is more critical with this mixture in this first phase of reaction because the solid oxidizer provides less oxygen for Al reaction. The difference in initial oxidizer decomposition between Al+CuO and Al+ Fe2O3 reactions may be key to understanding the dramatic increase in flame speed for Al+ Fe2O3. While Al+CuO has a large concentration of oxygen available at stage three, Al+Fe2O3 only has atmospheric oxygen available under half of the stages. When atmospheric oxygen is not present, the reaction does not initiate until stage four. However, with the presence of atmospheric oxygen, Al can react in stage three to accelerate decomposition of Fe2O3. None of the other reactions are as largely affected by atmospheric oxygen and as a result, Al+Fe2O3 is impacted the most with flame speed increases of 100%. Adding extra Al increases the initial reactivity under all conditions, but combined with a high O2 atmosphere, the increased emphasis of the stage three reaction increases flame speeds by 200%.

Assuming that the relationship between the temperatures of oxidizer decomposition and reaction with atmospheric oxygen are key to the sensitivity of a reaction to atmospheric oxygen, a non-dimensional sensitivity factor (S) is introduced. The sensitivity factor should also include information on reaction pressures that can affect the ability of atmospheric oxygen to come in contact with the Al particles and

iv Texas Tech University, Cory Farley, May 2013

ignite. Generally, fast reactions (flame speed (FS) >100m/s) are more convective with a greater pressure differential created by the high speed. In this way the sensitivity factor also includes convective effects by comparing the log10(FS) to the log10(100).

( ) ( ) (43) ( )

In Eq (43), Td is oxidizer decomposition temperature in Kelvin. The sensitivity factor includes a combination of the thermal decomposition sensitivity initiating the reaction and the convective sensitivity for the ability of atmospheric oxygen to contribute to the reaction once ignition has occurred. The revised sensitivity factors are shown in Figure 17 alongside the data symbol for the reaction. Figure 17 provides a relationship for the sensitivity of the reaction to influences of atmospheric oxygen that is consistent with the diverse mixtures studied here. Future work can further affirm this predictor and assess the role of atmospheric oxygen on flame speed.

120 Al+I2O5 Al+CuO 100 1.2, 100 Al+Ca(IO3)2

80 Al+Fe2O3

60

40 0.62, 30 Increase FS (%)in 20 -0.72, 9 -0.16, 10 0 -1 -0.5 0 0.5 1 1.5

Sensitivity Factor

Figure 17. Percent increase in flame speed as a function of sensitivity factor

iv Texas Tech University, Cory Farley, May 2013

CONCLUSIONS

Differential scanning calorimetry (DSC) and thermogravimetric (TG) analysis of I2O5, I2O5/Al and I2O5/Al2O3 mixtures in argon show a scale dependent reaction based on the size of the Al powder. While larger scale micron powders show little to no reactivity, nanometer scale passivated Al powders undergo a complex multistep reaction when mixed with I2O5. The reaction commences upon I2O5 - decomposition and is triggered by I adsorption into the Al2O3 passivation shell, further reactions between the iodine and oxygen gas and Al in the solid phase is then observed.

This study found no correlation between heat of combustion or mass loss and flame speed in open tray burns. Instead, apparent activation energy correlates with flame speeds by orders of magnitude suggesting flame speeds are attributed to diffusion limited but kinetically controlled reactions. In order to understand why the activation energy was lower for these reactions, studies of the free halogen gas were conducted showing intermediate chemistries contribute toward heightened levels of exothermicity. Surface chemistry associated with iodine interaction with the alumina shell is considered a first step in the reaction mechanism. Intermediate chemistries between iodine species and aluminum are the controlling accelerant mechanism for Al-I2O5 and Al-AgIO3 reactions.

In contrast iodine radicals are not as easily released in Al-Ca(IO3)2 and Al-KClO4 with onset temperatures in excess of 200 degrees higher for these mixtures.

Theoretical analyses show that almost all of the intermediate reactions between Al and iodine species are highly exothermic. Also, Al can act as an effective intermediate to aid decomposition of iodine oxide species and aid in the production of I and I2 in the gas phase. In summary the chemical kinetics responsible for the low activation energies and high flame speeds observed for Al-I2O5 and Al-AgIO3 are: (1) I2O5 and IO3 provide multiple exothermic channels that contribute to the next exothermicity; (2) many of the oxidizer thermal decomposition steps produce O2 that sustains the combustion process and increases reaction temperature; and, (3) Al-O species are highly stable and a method in which Al-IO or Al-O species converts to Al2O3 and produces more I2 as the gas-phase product has been presented. Results for these experimental and theoretical investigations elucidate the factors controlling reactivity of nano-Al energetic composites containing

iv Texas Tech University, Cory Farley, May 2013

halogenated oxide species. Theoretical chemical influences describe in terms of reaction kinetics explain the observed apparent activation energy and flame speed measurements.

Flame speeds were measured for four composite reactions. All reactions are highly gas generating except Al+Fe2O3. However, Al+Ca(IO3)2 has a flame speed of only 2.5 m/s implying a primarily conductively dominant reaction despite high gas generation. Results show increasing atmospheric oxygen has a much more dramatic effect on the flame speed of conductive reactions, Al+Fe2O3 with a 100-200% increase and

Al+Ca(IO3)2 with a 30-55% increase. For the convective reactions, Al+CuO increased by only 10-30%, and Al+I2O5 by 4-9%. This behavior is the basis for a model that describes a sensitivity factor based on solid oxidizer decomposition temperature and flame speed that accounts for oxygen availability and thus how the flame speed will be affected by the presence of atmospheric oxygen. Earlier onset of solid oxidizer decomposition temperatures and higher flame speeds cause reactions to be less sensitive to atmospheric oxygen concentration. In general, these results suggest atmospheric oxygen plays a greater role enhancing flame speeds for conductively dominant reactions. Since the boost to flame speed is limited by Al availability, adding more fuel increases flame speed and results show that increasing equivalence ratio from 1.0 to 1.2 produces accelerated flame speeds.

iv Texas Tech University, Cory Farley, May 2013

BIBLIOGRAPHY

Adamo, Carlo, and Vincenzo Barone. 1999. “Toward Reliable Density Functional Methods Without Adjustable Parameters: The PBE0 Model.” The Journal of Chemical Physics 110 (13): 6158. doi:10.1063/1.478522. http://link.aip.org/link/JCPSA6/v110/i13/p6158/s1&Agg=doi.

Asay, Blaine W., Steven F. Son, James R. Busse, and David M. Oschwald. 2004. “Ignition Characteristics of Metastable Intermolecular Composites.” Propellants, Explosives, Pyrotechnics 29 (4) (August): 216–219. doi:10.1002/prep.200400049. http://onlinelibrary.wiley.com/doi/10.1002/prep.200400049/pdf.

ASTM E698. 2011. “Standard Test Method for Arrhenius Kinetic Constants for Thermally Unstable Materials Using Differential Scanning Calorimetry and the Flynn / Wall / Ozawa Method 1.” Annual Book of ASTM Standards. doi:10.1520/E0698-11.

Babaitsev, I. V., S. Yu. Arzhevitov, and O. a. Presnakova. 2007. “Evaluating the Blast Pressure of Iron-aluminum Thermites.” Metallurgist 51 (7-8) (July): 348–350. doi:10.1007/s11015-007-0065-6. http://www.springerlink.com/index/10.1007/s11015-007-0065-6.

Becke, Axel D. 1993. “Density-functional Thermochemistry. III. The Role of Exact Exchange.” The Journal of Chemical Physics 98 (7): 5648. doi:10.1063/1.464913. http://link.aip.org/link/JCPSA6/v98/i7/p5648/s1&Agg=doi.

Beckstead, M. W. 2005. “Correlating Aluminum Burning Times.” Combustion, Explosion, and Shock Waves 41 (5) (September): 533–546. doi:10.1007/s10573- 005-0067-2. http://www.springerlink.com/index/10.1007/s10573-005-0067-2.

Beckstead, M. W., Y. Liang, and K. V. Pudduppakkam. 2005. “Numerical Simulation of Single Aluminum Particle Combustion (Review).” Combustion, Explosion, and Shock Waves 41 (6) (November): 622–638. doi:10.1007/s10573-005-0077-0. http://www.springerlink.com/index/10.1007/s10573-005-0077-0.

Blatchley, Ernest R, A M Asce, Anne Meeusen, Arthur I Aronson, and Lindsay Brewster. 2005. “Inactivation of Bacillus Spores by Ultraviolet or Gamma Radiation” (September): 1245–1252. doi:10.1061/ASCE0733-93722005131:91245.

Bockmon, B. S., M. L. Pantoya, S. F. Son, B. W. Asay, and J. T. Mang. 2005. “Combustion Velocities and Propagation Mechanisms of Metastable Interstitial Composites.” Journal of Applied Physics 98 (6): 064903. doi:10.1063/1.2058175. http://link.aip.org/link/JAPIAU/v98/i6/p064903/s1&Agg=doi.

Bowler, Peter J., and Iwan R. Morus. 2005. Making Modern Science: A Historical Survey. Chicago and London: The University of Chicago Press.

iv Texas Tech University, Cory Farley, May 2013

Bradley, Derek, Z. Chen, and J.R. Swithenbank. 1989. “Burning Rates in Turbulent Fine Dust-air Explosions.” Symposium (International) on Combustion 22 (1) (January): 1767–1775. doi:10.1016/S0082-0784(89)80190-0. http://linkinghub.elsevier.com/retrieve/pii/S0082078489801900.

Brown, Lemay, and Bursten. 2000. Chemistry: The Central Science. Ed. John Challice. 8th ed. Prentice-Hall. Inc.

Clark, Billy R, and Michelle L Pantoya. 2010. “The Aluminium and Iodine Pentoxide Reaction for the Destruction of Spore Forming Bacteria.” Physical Chemistry Chemical Physics : PCCP 12 (39) (October 21): 12653–7. doi:10.1039/c0cp00473a. http://www.ncbi.nlm.nih.gov/pubmed/20730185.

Clark, Stewart J., Matthew D. Segall, Chris J. Pickard, Phil J. Hasnip, Matt I. J. Probert, Keith Refson, and Mike C. Payne. 2005. “First Principles Methods Using CASTEP.” Zeitschrift Für Kristallographie 220 (5-6-2005) (May): 567–570. doi:10.1524/zkri.220.5.567.65075. http://www.oldenbourg- link.com/doi/abs/10.1524/zkri.220.5.567.65075.

“Combustion”. “Combustion.” Merriam-Webster.com. http://www.merriam- webster.com/.

Cox, J. D., D. D. Wagman, and V. A. Medvedev. 1987. CODATA Key Values for Thermodynamics. New York, New York: Hemisphere Publishing Corporation.

Crane, Charles A., Michelle L. Pantoya, and Jerry Dunn. 2010. “Infrared Measurements of Energy Transfer from Energetic Materials to Steel Substrates.” International Journal of Thermal Sciences 49 (10) (October): 1877–1885. doi:10.1016/j.ijthermalsci.2010.06.005. http://linkinghub.elsevier.com/retrieve/pii/S129007291000164X.

Curtiss, Larry A., Krishnan Raghavachari, Paul C. Redfern, and John A. Pople. 1997. “Assessment of Gaussian-2 and Density Functional Theories for the Computation of Enthalpies of Formation.” The Journal of Chemical Physics 106 (3): 1063. doi:10.1063/1.473182. http://link.aip.org/link/JCPSA6/v106/i3/p1063/s1&Agg=doi.

Dikici, Birce, Steven W. Dean, Michelle L. Pantoya, Valery I. Levitas, and R. Jason Jouet. 2009. “Influence of Aluminum Passivation on the Reaction Mechanism: Flame Propagation Studies.” Energy & Fuels 23 (9) (September 17): 4231–4235. doi:10.1021/ef801116x. http://pubs.acs.org/doi/abs/10.1021/ef801116x.

Dreizin, Edward L., Carlo Badiola, Shasha Zhang, and Yasmine Aly. 2011. “PARTICLE COMBUSTION DYNAMICS OF METAL-BASED REACTIVE MATERIALS.” International Journal of Energetic Materials and Chemical Propulsion 10 (4): 297– 319. doi:10.1615/IntJEnergeticMaterialsChemProp.2012005014. http://www.dl.begellhouse.com/journals/17bbb47e377ce023,forthcoming,5014.html.

iv Texas Tech University, Cory Farley, May 2013

Dunning, Thom H. 1989. “Gaussian Basis Sets for Use in Correlated Molecular Calculations. I. The Atoms Boron Through Neon and Hydrogen.” The Journal of Chemical Physics 90 (2): 1007. doi:10.1063/1.456153. http://link.aip.org/link/JCPSA6/v90/i2/p1007/s1&Agg=doi.

Fan, Run-Hua, Hong-Liang Lü, Kang-Ning Sun, Wan-Xia Wang, and Xin-Bing Yi. 2006. “Kinetics of Thermite Reaction in Al-Fe2O3 System.” Thermochimica Acta 440 (2) (January 15): 129–131. doi:10.1016/j.tca.2005.10.020. http://linkinghub.elsevier.com/retrieve/pii/S0040603105005344.

Farley, Cory, and Michelle Pantoya. 2010. “Reaction Kinetics of Nanometric Aluminum and Iodine Pentoxide.” Journal of Thermal Analysis and Calorimetry 102 (2) (June): 609–613. doi:10.1007/s10973-010-0915-5. http://www.springerlink.com/index/10.1007/s10973-010-0915-5.

Fischer, SH, and MC Grubelich. 1998. “Theoretical Energy Release of Thermites, Intermetallics, and Combustible Metals.” In 24th International Pyrotechnics Seminar. Monterey, CA. doi:10.2172/658208. http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=658208.

Flockhart, B.D., K Y Liew, and R C Pink. 1974. “Electron-Transfer at Alumina Surfaces” 24: 20–24.

Friedman, R, and A. Maček. 1963. “Combustion Studies of Single Aluminum Particles.” Symposium (International) on Combustion 9 (1) (January): 703–712. doi:10.1016/S0082-0784(63)80078-8. http://www.sciencedirect.com/science/article/pii/S0082078463800788.

Gang, Xiao, Yimin Zhu, Henning Birch, Hans Aage Hjuler, and Niels J. Bjerrum. 2004. “Iodine as Catalyst for the Direct Oxidation of Methane to Methyl Sulfates in Oleum.” Applied Catalysis A: General 261 (1) (April): 91–98. doi:10.1016/j.apcata.2003.10.039. http://linkinghub.elsevier.com/retrieve/pii/S0926860X03009682.

Gill, Robert J., Carlo Badiola, and Edward L. Dreizin. 2010. “Combustion Times and Emission Profiles of Micron-sized Aluminum Particles Burning in Different Environments.” Combustion and Flame 157 (11) (November): 2015–2023. doi:10.1016/j.combustflame.2010.02.023. http://linkinghub.elsevier.com/retrieve/pii/S0010218010000787.

Glassman, Irvin, and Richard A Yetter. 2008. Combustion. 4th ed. San Diego CA: Elsevier Inc.

Goyne, C P, R J Stalker, A Paull, C P Brescianini, H BRADLEY, G L DEHORITY, M IBIRICU, and E W PRICE. 1966. “Theory of Ignition of Solid Propellants.” AIAA

iv Texas Tech University, Cory Farley, May 2013

Journal 4 (7) (July 1): 1153–1181. doi:10.2514/3.3645. http://dx.doi.org/10.2514/3.3645.

Granier, John J., and Michelle L. Pantoya. 2004. “Laser Ignition of Nanocomposite Thermites.” Combustion and Flame 138 (4) (September): 373–383. doi:10.1016/j.combustflame.2004.05.006. http://linkinghub.elsevier.com/retrieve/pii/S0010218004001373.

Granier, John, and Michelle Pantoya. 2004. “The Effect of Size Distribution on Burn Rate in Nanocomposite Thermites: a Probability Density Function Study.” Combustion Theory and Modelling 8 (3) (September): 555–565. doi:10.1088/1364- 7830/8/3/007. http://www.informaworld.com/openurl?genre=article&doi=10.1088/1364- 7830/8/3/007&magic=crossref||D404A21C5BB053405B1A640AFFD44AE3.

Grant, Daniel J, Edward B Garner, Myrna H Matus, Minh Tho Nguyen, Kirk A Peterson, Joseph S Francisco, and David A Dixon. 2010. “Thermodynamic Properties of the XO(2), X(2)O, XYO, X(2)O(2), and XYO(2) (X, Y = Cl, Br, and I) Isomers.” The Journal of Physical Chemistry. A 114 (12) (April 1): 4254–65. doi:10.1021/jp911320p. http://www.ncbi.nlm.nih.gov/pubmed/20187618.

Harigel, GG. 2000. “The Concept of Weapons of Mass Destruction: Chemical and Biological Weapons, Use in Warfare, Impact on Society and Environment.” Proceeding of Seventh ISODARCO-Beijing Seminar on …: 1–18. http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:The+concept+of+ weapons+of+mass+destruction:+Chemical+and+biological+weapons,+use+in+warf are,+impact+on+society+and+environment#0.

Hettich, Robert L. 1989. “Structural Investigations of Aluminum Cluster Ions, Aln- (n = 3-50).” Journal of the American Chemical Society 111 (23) (November): 8582– 8588. doi:10.1021/ja00205a004. http://pubs.acs.org/doi/abs/10.1021/ja00205a004.

Holleman, A.F., and E Wiberg. 2001. Inorganic Chemistry. 34th ed. San Diego CA: Academic Press, Inc.

Huang, Ying, Grant A Risha, Vigor Yang, and Richard A Yetter. 2009. “Effect of Particle Size on Combustion of Aluminum Particle Dust in Air.” Combustion and Flame 156 (1) (January): 5–13. doi:10.1016/j.combustflame.2008.07.018. http://linkinghub.elsevier.com/retrieve/pii/S0010218008002861.

Inglesby, T. V. 1999. “Anthrax as a Biological Weapon: Medical and Public Health Management.” JAMA: The Journal of the American Medical Association 281 (18) (May): 1735–1745. doi:10.1001/jama.281.18.1735. http://jama.ama- assn.org/cgi/doi/10.1001/jama.281.18.1735.

iv Texas Tech University, Cory Farley, May 2013

Jian, Guoqiang, Snehaunshu Chowdhury, Kyle Sullivan, and Michael R. Zachariah. 2013. “Nanothermite Reactions: Is Gas Phase Oxygen Generation from the Oxygen Carrier an Essential Prerequisite to Ignition?” Combustion and Flame 160 (2) (February): 432–437. doi:10.1016/j.combustflame.2012.09.009. http://linkinghub.elsevier.com/retrieve/pii/S0010218012002702.

Kendall, RA, TH Dunning Jr, and RJ Harrison. 1992. “Electron Affinities of the First- row Atoms Revisited. Systematic Basis Sets and Wave Functions.” The Journal of Chemical … 96 (May). http://link.aip.org/link/jcpsa6/v96/i9/p6796/s1.

Kim, Joonghan, Hyotcherl Ihee, and Yoon Sup Lee. 2010. “Spin-orbit Density Functional and Ab Initio Study of HgX(n) (X=F, Cl, Br, and I; N=1, 2, and 4).” The Journal of Chemical Physics 133 (14) (October 14): 144309. doi:10.1063/1.3497189. http://www.ncbi.nlm.nih.gov/pubmed/20950002.

Kuo, Kenneth K. 1986. Principles of Combustion. 1st ed. Hoboken, NJ: John Wiley and Sons.

Lee, Chengteh, Weitao Yang, and Robert G. Parr. 1988. “Development of the Colle- Salvetti Correlation-energy Formula into a Functional of the Electron Density.” Physical Review B 37 (2) (January): 785–789. doi:10.1103/PhysRevB.37.785. http://link.aps.org/doi/10.1103/PhysRevB.37.785.

Lee, Sang Yeon. 2004. “Computational Study of Enthalpies of Formation of OXO (X = Cl, Br, and I) and Their Anions.” The Journal of Physical Chemistry A 108 (48) (December): 10754–10761. doi:10.1021/jp0467550. http://pubs.acs.org/doi/abs/10.1021/jp0467550.

Levitas, Valery I., Blaine W. Asay, Steven F. Son, and Michelle Pantoya. 2006. “Melt Dispersion Mechanism for Fast Reaction of Nanothermites.” Applied Physics Letters 89 (7): 071909. doi:10.1063/1.2335362. http://link.aip.org/link/APPLAB/v89/i7/p071909/s1&Agg=doi.

Losada, Martin, and Santanu Chaudhuri. 2010. “Finite Size Effects on aluminum/Teflon Reaction Channels Under Combustive Environment: a Rice-Ramsperger-Kassel- Marcus and Transition State Theory Study of Fluorination.” The Journal of Chemical Physics 133 (13) (October 7): 134305. doi:10.1063/1.3480020. http://www.ncbi.nlm.nih.gov/pubmed/20942537.

Lynch, Patrick, Herman Krier, and Nick Glumac. 2009. “A Correlation for Burn Time of Aluminum Particles in the Transition Regime.” Proceedings of the Combustion Institute 32 (2): 1887–1893. doi:10.1016/j.proci.2008.06.205. http://linkinghub.elsevier.com/retrieve/pii/S1540748908002514.

Martirosyan, K.S., L. Wang, and D. Luss. 2009. “Novel Nanoenergetic System Based on Iodine Pentoxide.” Chemical Physics Letters 483 (1-3) (November 24): 107–110.

iv Texas Tech University, Cory Farley, May 2013

doi:10.1016/j.cplett.2009.10.038. http://linkinghub.elsevier.com/retrieve/pii/S0009261409012974.

Moore, David S., Steven F. Son, and Blaine W. Asay. 2004. “Time-Resolved Spectral Emission of Deflagrating Nano-Al and Nano-MoO3 Metastable Interstitial Composites.” Propellants, Explosives, Pyrotechnics 29 (2) (April): 106–111. doi:10.1002/prep.200400038. http://doi.wiley.com/10.1002/prep.200400038.

Osborne, D.T. 2006. “The Effects of Fuel Particle Size on the Reaction of Al/Teflon Mixtures.” Master’s Thesis, Texas Tech University. Available Electronically from Http://hdl. Handle. Net/2346/1009. http://repositories.tdl.org/handle/2346/1009.

Osborne, Dustin T., and Michelle L. Pantoya. 2007. “Effect of Al Particle Size on the Thermal Degradation of Al/Teflon Mixtures.” Combustion Science and Technology. doi:10.1080/00102200601182333. http://www.informaworld.com/openurl?genre=article&doi=10.1080/0010220060118 2333&magic=crossref||D404A21C5BB053405B1A640AFFD44AE3.

Pagni, RM, GW Kabalka, and Richard Boothe. 1988. “Reactions of Unsaturated Compounds with Iodine and Bromine on. Gamma. Alumina.” The Journal of Organic Chemistry (4): 4477–4482. http://pubs.acs.org/doi/abs/10.1021/jo00254a011.

Park, K, D Lee, a Rai, D Mukherjee, and M R Zachariah. 2005. “Size-resolved Kinetic Measurements of Aluminum Nanoparticle Oxidation with Single Particle Mass Spectrometry.” The Journal of Physical Chemistry. B 109 (15) (April 21): 7290–9. doi:10.1021/jp048041v. http://www.ncbi.nlm.nih.gov/pubmed/16851834.

Perry, W. Lee, Bettina L. Smith, Christopher J. Bulian, James R. Busse, Clay S. Macomber, Rob C. Dye, and Steven F. Son. 2004. “Nano-Scale Tungsten Oxides for Metastable Intermolecular Composites.” Propellants, Explosives, Pyrotechnics 29 (2) (April): 99–105. doi:10.1002/prep.200400037. http://doi.wiley.com/10.1002/prep.200400037.

Peterson, Kirk A, Benjamin C Shepler, Detlev Figgen, and Hermann Stoll. 2006. “On the Spectroscopic and Thermochemical Properties of ClO, BrO, IO, and Their Anions.” The Journal of Physical Chemistry. A 110 (51) (December 28): 13877–83. doi:10.1021/jp065887l. http://www.ncbi.nlm.nih.gov/pubmed/17181347.

Prakash, a., a. V. McCormick, and M. R. Zachariah. 2005. “Synthesis and Reactivity of a Super-Reactive Metastable Intermolecular Composite Formulation of Al/KMnO4.” Advanced Materials 17 (7) (April): 900–903. doi:10.1002/adma.200400853. http://doi.wiley.com/10.1002/adma.200400853.

Prentice, Daniel, Michelle L Pantoya, and Brady J Clapsaddle. 2005. “Effect of Nanocomposite Synthesis on the Combustion Performance of a Ternary Thermite.”

iv Texas Tech University, Cory Farley, May 2013

The Journal of Physical Chemistry. B 109 (43) (November): 20180–5. doi:10.1021/jp0534481. http://www.ncbi.nlm.nih.gov/pubmed/16853608.

Proust, C., and B. Veyssiere. 1988. “Fundamental Properties of Flames Propagating in Starch Dust-Air Mixtures.” Combustion Science and Technology 62 (4-6) (December): 149–172. doi:10.1080/00102208808924007. http://www.tandfonline.com/doi/abs/10.1080/00102208808924007.

Pu, Y.K., F. Jia, S.F. Wang, and T. Skjold. 2007. “Determination of the Maximum Effective Burning Velocity of Dust–air Mixtures in Constant Volume Combustion.” Journal of Loss Prevention in the Process Industries 20 (4-6) (July): 462–469. doi:10.1016/j.jlp.2007.04.036. http://linkinghub.elsevier.com/retrieve/pii/S0950423007000526.

Rao, B. K., and P. Jena. 1999. “Evolution of the Electronic Structure and Properties of Neutral and Charged Aluminum Clusters: A Comprehensive Analysis.” The Journal of Chemical Physics 111 (5): 1890. doi:10.1063/1.479458. http://link.aip.org/link/JCPSA6/v111/i5/p1890/s1&Agg=doi.

Saikia, Monmi, Dwipen Kakati, Maria S. Joseph, and Jadab Chandra Sarma. 2009. “Iodine-Alumina Catalyzed Aza- Michael Addition Under Solvent Free Conditions.” Letters in Organic Chemistry 6 (8) (December 1): 654–658. doi:10.2174/157017809790442961. http://openurl.ingenta.com/content/xref?genre=article&issn=1570- 1786&volume=6&issue=8&spage=654.

Sanders, V Eric, Blaine W Asay, Timothy J Foley, Bryce C Tappan, Adam N Pacheco, and Steven F Son. 2007. “Reaction Propagation of Four Nanoscale Energetic Composites (Al/MoO3, Al/WO3, Al/CuO, and B12O3).” Journal of Propulsion and Power 23 (4) (July): 707–714. doi:10.2514/1.26089. http://arc.aiaa.org/doi/abs/10.2514/1.26089.

Sarbak, Z. 1997. “Effect of Fluoride and Sodium Ions on Structural and Thermal Properties of $\gamma$-Al2O3.” Crystal Research and Technology 32 (4): 491– 497. http://onlinelibrary.wiley.com/doi/10.1002/crat.2170320402/abstract.

Segall, M D, Philip J D Lindan, M J Probert, C J Pickard, P J Hasnip, S J Clark, and M C Payne. 2002. “First-principles Simulation: Ideas, Illustrations and the CASTEP Code.” Journal of Physics: Condensed Matter 14 (11) (March 25): 2717–2744. doi:10.1088/0953-8984/14/11/301. http://stacks.iop.org/0953- 8984/14/i=11/a=301?key=crossref.242efa60e2e8ff32bb67f73813bd334f.

Sottery, T. W. 1985. “Chemical Principles, Sixth Edition (Masterton, William L.; Slowinski, Emil J.; Stanitski, Conrad L.).” Journal of Chemical Education 62 (12) (December): A325. doi:10.1021/ed062pA325. http://pubs.acs.org/doi/abs/10.1021/ed062pA325.

iv Texas Tech University, Cory Farley, May 2013

Starink, M.J. 2004. “Analysis of Aluminium Based Alloys by Calorimetry: Quantitative Analysis of Reactions and Reaction Kinetics.” International Materials Reviews 49 (3) (June 1): 191–226. doi:10.1179/095066004225010532. http://www.ingentaselect.com/rpsv/cgi- bin/cgi?ini=xref&body=linker&reqdoi=10.1179/095066004225010532.

Stern, J. H., and a. a. Passchier. 1962. “The Heats of Formation of Triiodide and Iodate Ions.” Journal of Physical Chemistry 66 (4) (April 1): 752–753. doi:10.1021/j100810a041. http://pubs.acs.org/cgi- bin/doilookup/?10.1021/j100810a041.

Stern, KH. 1974. “High Temperature Properties and Decomposition of Inorganic Salts, Part 4. Oxy-Salts of the Halogens.” Journal of Physical and Chemical Reference Data 3 (2). doi:10.1201/9781420042344.ch9. http://jpcrd.aip.org/resource/1/jpcrbu/v3/i2/p481_s1.

Sullivan, K. T., N. W. Piekiel, S. Chowdhury, C. Wu, M. R. Zachariah, and C. E. Johnson. 2011. “Ignition and Combustion Characteristics of Nanoscale Al/AgIO3: A Potential Energetic Biocidal System.” Combustion Science and Technology 183 (3) (March): 285–302. doi:10.1080/00102202.2010.496378. http://www.informaworld.com/openurl?genre=article&doi=10.1080/00102202.2010. 496378&magic=crossref||D404A21C5BB053405B1A640AFFD44AE3.

Sullivan, K.T., N.W. Piekiel, C. Wu, S. Chowdhury, S.T. Kelly, T.C. Hufnagel, K. Fezzaa, and M.R. Zachariah. 2012. “Reactive Sintering: An Important Component in the Combustion of Nanocomposite Thermites.” Combustion and Flame 159 (1) (January): 2–15. doi:10.1016/j.combustflame.2011.07.015. http://linkinghub.elsevier.com/retrieve/pii/S0010218011002276.

Taylor, S L, L R Fina, and J L Lambert. 1970. “New Water Disinfectant: An Insoluble Quaternary Ammonium Resin-triiodide Combination That Releases Bactericide on Demand.” Applied Microbiology 20 (5) (November): 720–2. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=377032&tool=pmcentre z&rendertype=abstract.

Tillotson, T M, A E Gash, R L Simpson, L W Hrubesh, J H Satcher Jr, and J F Poco. 2001. “Nanostructured Energetic Materials Using Sol ± Gel Methodologies” 285: 338–345.

Toyohara, Masumitsu, Masaaki Kaneko, Norimasa Mitsutsuka, Hiroshi Fujihara, Noriyuki Saito, and Takuya Murase. 2002. “Contribution to Understanding Iodine Sorption Mechanism onto Mixed Solid Alumina Cement and Calcium Compounds.” Journal of Nuclear Science and Technology 39 (9) (September): 950–956. doi:10.1080/18811248.2002.9715281. http://www.tandfonline.com/doi/abs/10.1080/18811248.2002.9715281.

iv Texas Tech University, Cory Farley, May 2013

Trunov, M, M Schoenitz, X Zhu, and E Dreizin. 2005. “Effect of Polymorphic Phase Transformations in AlO Film on Oxidation Kinetics of Aluminum Powders.” Combustion and Flame 140 (4) (March): 310–318. doi:10.1016/j.combustflame.2004.10.010. http://linkinghub.elsevier.com/retrieve/pii/S0010218004002536.

Trunov, MA. 2006. “Melting and Oxidation of Nanometer Size Aluminum Powders.” MATERIALS …. http://journals.cambridge.org/production/action/cjoGetFulltext?fulltextid=8109845.

Turns, Stephen R. 2012. An Introduction to Combustion: Concepts and Applications Third Edition. 3rd ed. New York, New York: McGraw Hill.

Umbrajkar, Swati M., Mirko Schoenitz, and Edward L. Dreizin. 2006. “Exothermic Reactions in Al–CuO Nanocomposites.” Thermochimica Acta 451 (1-2) (December): 34–43. doi:10.1016/j.tca.2006.09.002. http://linkinghub.elsevier.com/retrieve/pii/S0040603106004692.

Valiev, M., E.J. Bylaska, N. Govind, K. Kowalski, T.P. Straatsma, H.J.J. Van Dam, D. Wang, et al. 2010. “NWChem: A Comprehensive and Scalable Open-source Solution for Large Scale Molecular Simulations.” Computer Physics Communications 181 (9) (September): 1477–1489. doi:10.1016/j.cpc.2010.04.018. http://linkinghub.elsevier.com/retrieve/pii/S0010465510001438.

Wang, H, L Gao, Weiqun Li, and Qiang Li. 1999. “Preparation of Nanoscale α-Al2O3 Powder by the Polyacrylamide Gel Method.” Nanostructured Materials 11 (8) (November): 1263–1267. doi:10.1016/S0965-9773(99)00417-1. http://linkinghub.elsevier.com/retrieve/pii/S0965977399004171.

Watson, Kyle W, Michelle L Pantoya, and Valery I Levitas. 2008. “Fast Reactions with Nano- and Micrometer Aluminum : A Study on Oxidation Versus Fluorination.” Combustion and Flame 155 (4): 619–634. doi:10.1016/j.combustflame.2008.06.003. http://dx.doi.org/10.1016/j.combustflame.2008.06.003.

Weismiller, M.R., J.G. Lee, and R.a. Yetter. 2011. “Temperature Measurements of Al Containing Nano-thermite Reactions Using Multi-wavelength Pyrometry.” Proceedings of the Combustion Institute 33 (2) (January): 1933–1940. doi:10.1016/j.proci.2010.06.094. http://linkinghub.elsevier.com/retrieve/pii/S1540748910001616.

Weismiller, MR, JY Malchi, R.A. Yetter, and T.J. Foley. 2009. “Dependence of Flame Propagation on Pressure and Pressurizing Gas for an Al/CuO Nanoscale Thermite.” Proceedings of the Combustion Institute 32 (2): 1895–1903. doi:10.1016/j.proci.2008.06.191. http://www.sciencedirect.com/science/article/pii/S1540748908003520.

iv Texas Tech University, Cory Farley, May 2013

Zhai, Qing-Zhou, and Lu-Guo Chen. 2011. “Adsorptive Performance of Nanoscale a- Al2O3 Towards La(III).” Asian Journal of Chemistry 23 (9): 3975–3978.

Zhang, L., B. Li, C.-M. Mo, and J. Chen. 1995. “A Study of 27Al Nuclear Magnetic Resonance in Nanoscale Al2O3.” Nanostructured Materials 5 (3) (March): 299– 306. doi:10.1016/0965-9773(95)00245-A. http://linkinghub.elsevier.com/retrieve/pii/096597739500245A.

Zhang, Shasha, Carlo Badiola, Mirko Schoenitz, and Edward L. Dreizin. 2012. “Oxidation, Ignition, and Combustion of Al·I2 Composite Powders.” Combustion and Flame 159 (5) (May): 1980–1986. doi:10.1016/j.combustflame.2012.01.004. http://linkinghub.elsevier.com/retrieve/pii/S0010218012000065.

Zhao, Yan, and Donald G Truhlar. 2006. “A New Local Density Functional for Main- group Thermochemistry, Transition Metal Bonding, Thermochemical Kinetics, and Noncovalent Interactions.” The Journal of Chemical Physics 125 (19) (November 21): 194101. doi:10.1063/1.2370993. http://www.ncbi.nlm.nih.gov/pubmed/17129083.

Zhao, Yan, and Donald G. Truhlar. 2004. “Hybrid Meta Density Functional Theory Methods for Thermochemistry, Thermochemical Kinetics, and Noncovalent Interactions: The MPW1B95 and MPWB1K Models and Comparative Assessments for Hydrogen Bonding and Van Der Waals Interactions.” The Journal of Physical Chemistry A 108 (33) (August): 6908–6918. doi:10.1021/jp048147q. http://pubs.acs.org/doi/abs/10.1021/jp048147q.

iv