We have analyzed the results of thermal decomposition experimentswith HMX, performed with a simultaneous thermogravimetricmodulated beam mass spectrometry (STMBMS) apparatus, andhave created a reaction cycle that captures the physical processesand chemical reactions that control its decomposition. The reactioncycle is based on 9 different physical processes and 7 generalchemical reactions that play a significant role in controlling the de-composition of HMX. The reaction cycle captures both the mainreaction pathways and the interactions between the various path-ways. In addition, we find that these physicochemical processeslead to the emergence of several different types of morphologicalstructures within the HMX particles. These structures emerge dur-ing the early stages of the decomposition process and eventuallyevolve as active components of the rate controlling processes. Thedevelopment of a mathematical model to characterize the reactioncycle and the emergent phenomena is described.

THERMAL DECOMPOSITION OF HMX: MORPHOLOGICAL AND CHEMICAL CHANGES INDUCED AT SLOW DECOMPOSITION RATES

Richard Behrens Sandia National Laboratories Combustion Research Facility

Livermore, California 94551-0969

INTRODUCTION

The desire to predict the response of ex-plosives when exposed to fire has created an interest in understanding the thermal decom-position processes in HMX at slow heating rates. Previous work[1] on the thermal de-composition of HMX below its melting point (~270°C) indicated that the nucleation and growth of bubbles containing gaseous decom-

position products was important in the solid phase. When this behavior is coupled with current understanding of how increased sensi-tivity and violent combustion is associated with the dramatic effect of microscopic de-fects, such as cracks, holes, and voids on the combustion of the energetic materials through localized phenomena such as jetting and shear heating,[2-7] the importance of understanding and characterizing the thermal decomposition processes becomes apparent. The initial goal

of our research on HMX has been to identify the chemical reaction mechanisms and the physical processes that control its thermal de-composition and to understand how these processes lead to the evolution of the morpho-logical structure observed in our experiments. Our current goal is to develop mathematical models of our current understanding of the physicochemical processes, parameterize these models from our data collected with the STMBMS apparatus, and provide predictive models that can be used to assess the response of HMX containing explosives in fires.

Over the years, thermal analysis studies us-ing differential thermal analysis (DTA),[8] differential scanning calorimetry (DSC)[9-11] and thermogravimetric analysis (TGA)[12] have been conducted with the aim of deter-mining reaction kinetics of HMX. Reaction kinetics for HMX derived from these types of experiments assume simple reaction mecha-nisms in homogeneous materials. Since the overall information content provided by these methods is limited, it is not possible to decon-volve the effects of physical and morphologi-cal changes that occur in these materials as they decompose or to the effect these changes may have on the chemical reaction mecha-nisms and the overall rates of reaction from the data. The results from our STMBMS ex-periments on HMX[1, 13-16] and RDX[17-19] provide more information on the decom-position processes than standard thermal analysis methods and allow us to gain more insight into the underlying processes. Our work on HMX has shown that several differ-ent reaction pathways control the decomposi-tion process. Nucleation and growth proc-esses, transport processes and emergent phe-nomena all play a role in the thermal decom-position of HMX in the solid phase. HMX also undergoes significant morphological

changes that play a critical role in the overall thermal decomposition process. The ramifi-cations of these morphological changes are twofold. First, the morphological changes create microscopic defects that are associated with increased sensitivity due to ‘hot spot’ effects. Second, the overall rates of reaction are controlled by reactions in these localized microscopic regions. Thus, the number den-sity, size and interfacial processes in these microscopic reaction regions all play a role in the overall reaction process. Models of HMX thermal decomposition based on homogene-ous processes on a molecular level do not provide a sufficient framework to describe the slow thermal decomposition of HMX in the solid phase.

In this paper we present an overview of the information obtained from the STMBMS ex-periments, a qualitative model of the chemical and physical processes that constitute the HMX reaction cycle, and several aspects of the emergence and evolution of the morpho-logical features that are being incorporated into a mathematical model of the decomposi-tion process.

STMBMS RESULTS

We have used STMBMS, optical micros-copy and scanning electron microscopy (SEM) to probe the thermal decomposition processes in HMX and summarize our find-ings in this section. We have examined the thermal decomposition of HMX over a tem-perature range from 175°C to 265°C. The details of the STMBMS experimental meth-ods,[20-22] the chemical reactions of HMX and RDX,[1, 14, 17, 18, 23] and the morpho-logical changes that occur during the thermal decomposition of HMX[16] have been de-scribed in detail elsewhere.

A typical STMBMS experiment consists of placing either a single HMX particle or a sample of HMX powder in a reaction cell (Fig. 1), leak-checking the seals and placing the cell in the STMBMS apparatus. As the sample is heated, a vapor is formed in the re-action cell, which may consist of both HMX and its gaseous decomposition products. The gaseous species flow through an orifice in the top of the reaction cell, pass through a vac-uum and are detected with a modulated beam mass spectrometer. This data provides the identity of the decomposition products. Si-multaneously, the rate of force change (due to mass loss and thrust produced at the exit ori-fice) is measured. Analysis of the mass spec-trometry data coupled with the rate of force change data provides the gas formation rate and the partial pressure of each vapor species in the reaction cell as a function of time.

The gaseous species that evolve from the reaction cell in the STMBMS experiments

can originate from reactions that occur at sev-eral different locations. These include: 1) sublimation and unimolecular decomposition of HMX in the gas phase, 2) bimolecular re-actions between gaseous decomposition prod-ucts and HMX in the gas phase, 3) decompo-sition in the solid particles, followed by re-lease of gaseous products from the particles, 4) reactions at interfacial boundaries, such as on the surface of the particles or at interfacial regions formed by nucleation and growth processes within the HMX particles, 5) de-composition of nonvolatile secondary prod-ucts, and 6) on the walls of the reaction cell.

Data from a typical STMBMS experiment with HMX, shown in Figure 2, illustrates how insight into the various processes that control the decomposition of HMX may be derived from the results. Figure 2 shows the identities and gas formation rates (GFR) of products formed during the thermal decomposition of HMX at 235°C. The main decomposition products are: CH2O, NO, N2O, H2O and CO. Note the unusual variation in the temporal behaviors of the GFRs for the products in this isothermal experiment. After the sample reaches 235°C, there is a period of time, re-ferred to as the induction period, in which the rate of formation of the gaseous products is relatively slow. This is followed by another period of time, referred to as the acceleratory period, in which the rate of evolution of the products from the reaction cell increases. This is followed by a third period, the late stage, in which the rate of gas evolution from the reaction cell decreases with time. This type of temporal behavior has been observed in over one hundred similar experiments used to probe the decomposition process.[16] From these results we have drawn the follow-ing general conclusions: 1) most of the prod-ucts formed during the induction period arise from sublimation of HMX and its decomposi-tion in the gas phase, 2) the evolution of gaseous products during the acceleratory stage originates from the reaction and growth

Figure 1. Cross section of reaction cell used in STMBMS experiments.

of bubbles within the HMX particles, 3) the evolution of products during the late stage originates from complex reactions involving HMX and its gaseous and condensed phase decomposition products.

To understand how the evolution of gase-ous decomposition products is related to the physical process occurring within the HMX particles, STMBMS experiments were con-ducted in which the particles were partially decomposed, removed from the reaction cell and then examined with optical and SEM pic-tures. An SEM picture of the cross section of a particle that underwent 30% decomposition is shown in Figure 3. The central region of the HMX particle, where HMX remains, shows the granular structure formed when HMX is heated. The outer region, where HMX has been sublimed from the particle,

reveals the shell-like structure of the non-volatile residue (NVR) formed during the de-composition process. Note the size and uni-formity of the shell-like structure in the region where the HMX has been removed by subli-mation.

From the analysis of many similar experi-ments with HMX, we have found that the fol-lowing physical processes play an important role in controlling its rate of decomposition and creating new morphological features: 1) Sublimation of HMX from the surface of the particle and its subsequent decomposition in the gas phase. 2) Formation of a microscopic granular structure and larger scale cracks within the HMX particles as the sample un-dergoes the β δ phase transition. 3) De-composition of HMX at the inter-granular boundaries during the earliest stage of de-

NNO

H3C CH3

C2H5N

H2N CO

H

NH3C

HCHO

N C

O

HC

CH3

H3

C4N2H8

H N

N

TIME (sec)0 2000 4000 6000 8000

0

0.4

0.8

1.2

1.6

0

0.04

0.08

0.12

0.16N2 O

CH2 O

NO20°C

235°C

HCN

CO

H2 O

TIME (sec)0 2000 4000 6000 8000

0

4

8

12

061218243036

Gas

For

mat

ion

Rat

e (x

10-1

0 m

oles

/sec

)

Figure 2. Gas formation rates of the decomposition products formed during the decomposi-tion of a 1mg single particle of HMX in a reaction cell with a 25µm diameter orifice. The temperature of the sample is shown in the upper left panel. The maximum pressure of the gas products in the reaction cell is ~ 4 Torr. The three shaded regions denote the induction, acceleratory and late stages of the decomposition process.

composition (0 to 5%). 4) Nucleation and growth of reaction regions within the HMX grains, creating bubbles containing gaseous decomposition products within the grains. This process is important during the first 30% of decomposition. 5) Formation of a non-volatile residue by reaction between several

of the gaseous decomposition products (e.g., HCN and CH2O). This occurs in both inter-granular regions during the earliest stage of decomposition and in the bubbles that grow within the grains. 6) Release of gaseous de-composition products from the bubbles within the grains as they intersect the grain bounda-

Holes

HMX + Residue

Residue A

B

B

CD

Shells

Strands

Figure 2. Scanning electron micrographs of the cross section through a partially decom-posed HMX particle. To prepare the particle, it is heated to 235°C and held at this tempera-ture until ~30% of the HMX has decomposed. Next, approximately half of the remaining HMX is sublimed and then the particle is removed from the STMBMS and cut in half for examination. The black line in A demarcates the boundary between the outer region, where the residue formed in the decomposition processes has been uncovered through the sublimation of part of the remaining HMX, and the inner region, where HMX remains. Panels B and C show this boundary region more closely. Panel D shows the morphological features of the residue that is formed during the thermal decomposition process.

ries. 7) Flow of gaseous decomposition prod-ucts and gaseous HMX through the inter-granular region, which is more dominant dur-ing the later stages of an experiment. 8) Re-action between HMX, the NVR and a subset of the gaseous decomposition products. This occurs on the surface of the NVR, leading to growth of the NVR. 9) Decomposition of the NVR to form gaseous decomposition prod-ucts.

The main chemical reactions that occur in conjunction with these physical processes and lead to the emergence of new morphological structures have been elucidated from our re-cent experiments with HMX as well as from previous experiments with isotopically la-beled HMX[14] and RDX[18]. The main re-actions may be summarized as follows: 1) Elimination of HONO from HMX resulting in the formation of HCN and the subsequent re-action products of HONO: H2O, NO, and NO2. In our experiments this reaction is pri-marily associated with decomposition of HMX in the gas phase. 2) The formation of the mononitroso analogue of HMX by the re-placement of an NO2 group with NO. This pathway was elucidated from decomposition experiments using mixtures of 15N-labeled HMX and 14N-labeled HMX in which scram-bling of the N-NO bond in the mononitroso product was observed.[14] 3) The formation of the mononitroso analogue of HMX by the abstraction of oxygen from the NO2 group in a reaction between HMX and the NVR. This was also determined from experiments with mixtures of 15N-labeled HMX and 14N-labeled HMX. In this case, as the amount of NVR present in the sample increased, the amount of mononitroso HMX product without N-NO bond scrambling also increased, indicating interaction with the NVR. 4) The decomposi-tion of mononitroso HMX primarily to N2O, CH2O several other minor products (this de-composition pathway is by analogy to a study of the decomposition of the mononitroso ana-logue of RDX[24]). 5) The formation of a

NVR from the decomposition of the mononi-troso analogue of HMX. 6) The decomposi-tion of HMX on the surface of the NVR yield-ing N2O and CH2O as the primary products. We find that some of the HMX is incorpo-rated into the NVR leading to its growth. We also find that this reaction is dependent on the pressure of the confined gaseous decomposi-tion products, indicating that this reaction channel probably involves the NVR, HMX and a subset of the gaseous decomposition products. Further insight into this reaction channel comes from our work with RDX in which this reaction can be studied without the complication of the physical processes that occur during the decomposition of HMX. 7) The decomposition of the NVR forms various amides as decomposition products.

The nine physical processes and the seven different chemical reactions have been identi-fied and characterized from a wide range of different STMBMS experiments on both HMX and RDX. These results form the ex-perimental basis for creation of a qualitative model to describe these processes and a mathematical model to capture the parameters for the model from the STMBMS data and provide a basis for predicting the behavior of HMX over a range of heating conditions.

QUALITATIVE MODEL

We have created a reaction cycle that cap-tures the physical and chemical processes that control the thermal decomposition of HMX and leads to the emergence of new morpho-logical features. The reaction cycle is shown in Fig. 4.

The reactions start with the reactants in the solid phase (upper left) and proceed to the main exit products shown on the right. The main reactive processes that control the rates of reaction are captured in the reactions and reaction cycles shown in the center of the dia-gram. The bold circles show the main

reaction cycles. The smaller arrows indicate how the reactive intermediates move between the different reaction cycles and how the exit products evolve from the reactions and reac-tion cycles. The exit products, listed in Fig. 4, are classified as either non-reactive or reac-tive. The non-reactive products undergo no further reactions under the conditions of our thermal decomposition experiments. The reactive products decompose in the reaction cell, as well as evolve through the reaction cell orifice. The relative rates of decomposi-tion and evolution of the reactive products are determined by the experimental conditions, with the diameter of the reaction cell orifice having the most significant effect. The minor products that evolve during the decomposition process are shown in the box at the lower left. Although the minor products represent only a

small fraction of the total products formed during the decomposition process, their iden-tities and the temporal behaviors of their rates of evolution provide insight to the different cycles in the overall reaction mechanism.

The solid-phase reaction cycle is encoun-tered first by the reactants. The importance of this cycle is determined by the heating rate of the sample. If the heating rate is sufficiently rapid, then the length of time spent in this cy-cle is short and the sample will liquefy before extensive reaction has occurred in the solid-phase. However, if the heating rate is slow and the sample is maintained at a temperature between 160°C and its melting point, then reactions are observed in the solid phase. These reactions are complex and nonlinear with respect to the reactant. Processes involv-

Figure 4. Reaction scheme showing the physical and chemical processes that control the thermal decomposition of RDX and HMX.

ing grain formation, nucleation and growth of reactive sites, bubble formation, and nonlinear reactions between the reactant and its non-volatile secondary products are observed in the decomposition of HMX.

Thermal decomposition of liquefied RDX or HMX is also a complex process. The main reaction pathways of this process have been described pre-viously for the case of RDX.[17, 18] One pathway in-volves the elimina-tion of HONO and HON from RDX, leading to the formation of oxy-s-triazine (OST -- V). HONO and HON undergo further reaction, forming H2O, NO, and NO2. This reaction was found to be first order in RDX, with all atoms constituting OST originating from a single RDX molecule. This is the only reac-tion pathway, observed in our experiments with RDX and HMX, that is first order in the reactant, with the rate of formation of the gaseous products being proportional to the amount of RDX remaining in the sample. OST is not observed in the decomposition of HMX or in the decomposition of RDX under higher gas confinement conditions. Under these conditions HCN appears in place of OST.

Since HMX decomposes at temperatures well below its melting point, it reacts via a set of more complicated pathways. This is de-picted in Fig. 4 by the NO, nitroso and non-volatile residue (NVR) reaction cycles. Iso-topic scrambling experiments have shown that reaction between NO and RDX[18] or HMX[14] results in the formation of the mononitroso analogue of the reactant. Thus, one of the reaction cycles involves the reac-tion of NO, which may be generated in sev-eral of the different reaction cycles, with the

reactant to form the mononitroso analogue of HMX. The next reaction cycle involves de-composition of the mononitroso analogue. To understand the reaction mechanism of the mononitroso analogue, mononitroso RDX (MNRDX) was independently synthesized and its decomposition process studied by STMBMS methods.[24, 25] The references provide details on the complex decomposition behavior of MNRDX. It was found that ONDNTA decomposes in two different tem-perature ranges. One is between 110-145 °C and the other is between 155-210 °C. Iso-thermal experiments in the lower temperature range show that the fraction of the sample that decomposes in this channel is limited, and the fraction depends on the pressure of the gase-ous decomposition products contained with MNRDX in the reaction cell. Higher pres-sures of the gaseous decomposition products result in larger fractions of the MNRDX de-composing in the lower temperature channel. Above pressures of ~ 1atm all of the MNRDX decomposes in the lower-temperature chan-nel. The temporal behaviors of the rates of formation of the products in the higher tem-perature channel show that a nonlinear “auto-catalytic-like” process plays a significant role in the reaction mechanism for this channel. A nonvolatile residue is formed during the de-composition of MNRDX in both temperature channels. The products formed in the decom-position of ONDNTA are NO2, CO/N2, CH2O, N2O, NO and the minor products shown in Fig. 4. The NO2, CH2O, NO and NVR are involved in other reaction cycles in the decomposition process. Observation of the mononitroso analogue of HMX (MNHMX), plus the presence of the same decomposition products associated with the decomposition of MNRDX, provide firm evi-dence that the nitroso reaction cycle is impor-tant in the decomposition of HMX.

N

NN

N

NN

OHO

V

The nonvolatile residue reaction cycle is most dominant under experimental conditions characterized by the presence of higher pres-sures of the gaseous decomposition products. This reaction cycle initially involves the crea-tion of a NVR via reactions of secondary products from the HMX decomposition. As the amount of NVR present in the sample in-creases, this reaction cycle becomes the main rate-controlling process. The reaction be-tween the reactant and the NVR results in part of each HMX molecule being released as gaseous products and the remainder being in-corporated into the NVR. Thus, the rate of HMX decomposition that is controlled by the NVR reaction cycle increases as the amount of NVR present in the sample increases. The NVR is not completely stable at the tempera-tures of the experiments. During an experi-ment, the NVR grows through reaction with RDX or HMX and decomposes thermally, releasing the minor products shown.

The reaction cycle shown in Fig. 4 pro-vides the basis for creating mathematical models to describe the decomposition of HMX.

DEVELOPMENT OF MATHE-MATICAL MODELS

Our initial attempt to develop a mathe-matical model describing the decomposition of HMX was reported previously.[26] While

this model captured the features of the chemi-cal reactions, more recent experimental re-sults have shown that nucleation and growth processes and emerging morphological fea-tures play a major role in controlling the de-composition process. Mathematical represen-tations of these processes are being developed for use in new mathematical models of the HMX decomposition cycle.

Models of the morphological features as-sociated with rate controlling processes are being developed. The different types and relative scales of these features are illustrated in Fig. 5. The illustration represents a particle of HMX. Heating the particle through the β δ phase transition creates a grain structure composed of flat “platelet-like” grains. The newly formed grain boundaries provide new interfaces where decomposition occurs during the initial stage of an experiment. Within the grains, reaction centers nucleate and grow forming bubbles within the grains. Both gaseous and nonvolatile reaction products form within the bubbles. The NVR contained

Figure 5. Illustration of morphological fea-tures that play a role in the decomposition of HMX.

Figure 6. Illustration of growth of bubbles within grains.

within the bubbles opens a new reaction pathway at the NVR interface where reactions with HMX and the gaseous decomposition products may occur. When the surface of a growing bubble intersects a grain boundary, the gaseous products flow into the inter-granular region and may exit from the particle and pass into the free volume of the reaction cell. The NVR from the bubble is no longer exposed to the high-pressure gas conditions associated with the bubble, but is now in the reaction environment associated with the in-ter-granular region and the free volume of the reaction cell. This leads to the final type of morphological structure observed in our ex-periments, flakes of NVR. The flakes are formed during the later stages of decomposi-tion and probably originate from the reaction of HMX and one of the gaseous decomposi-tion on the surface of the NVR. We suspect that the “flake-like” structure arises from the

flow conditions of the HMX and the decom-position products through the inter-granular region of the particle where it reacts with the NVR, resulting in the growth of the NVR in the flow paths.

We have created mathematical representa-tions of these processes and illustrate the ap-proach with a model for the nucleation and growth of the bubbles within the grains. The geometrical aspects of the model are illus-trated in Fig. 6. The model assumes that the nucleation centers are all present at time zero and uniformly distributed throughout the grain. This is consistent with the experimen-tal results and suggests that the nucleation centers may be formed during the β δ phase transition. For one reaction scenario, we may assume the reaction occurs at the surface of the bubble (e.g., at the interface between the NVR and the solid HMX at the boundary of

Figure 7. Illustration of the growth of bubbles within grains of HMX and the resulting rates of HMX decomposition (B), formation of NVR (C), rate of gas release from the grains (D).

the bubble). This implies that the radius of the bubbles grow as a linear function of time, r(t) = k t, and the maximum radius is limited to h/2, half the thickness of the grain.

The number of bubbles that have released their gaseous products as they intersect the grain boundaries is given by

nr (t) = 2 ρn l w dr =0

r( t )

∫ 2 l w ρn r (t) (1)

and the number remaining confined within the grains is given by ng (t) = ρn l w (h −2 r(t) ) (2)

where ρn is the density of nucleation sites and the other parameters associated with the di-mensions of the grains are defined in Fig. 6. Corresponding expressions can be derived to describe amounts of HMX that has been con-verted to products and has been either re-leased (Vv) or remains (Vg) in the bubbles

Vr( t) = 43

0

r( t )

∫ π r3 2 ρn l w dr = 23 π ρn l w r (t)4

Vg(t) = 43π r (t )3 ρn l w ( h − 2 r ( t ) )

Furthermore, the corresponding terms repre-senting the surface area of the NVR that forms at the bubble interface which either re-main confined in the bubble environment (Ag) or is transferred to the inter-granular envi-ronment (Av) is given by Ag (t) = 4π r (t )2 ρn l w (h −2r ( t ) )

Ar (t) = 4 π r 2

0

r( t )

∫ ρn l w dr = 83 π ρn l w r (t )3

respectively.

A model of this behavior for decomposi-tion of particle is shown in Figure 7. The di-mensions of the grains are set at 10µm x 10µm x 1µm. A Gaussian distribution of grain heights is assumed with an average height of 1.0µm and a width of 0.5µm. Fig. 7A shows the linear growth of the radius of the bubble with time. Fig. 7B shows, the vol-ume of HMX that has decomposed and whose

gaseous products have been released from the bubbles, the volume of HMX that has decom-posed but whose gaseous products still remain within the bubbles in the grains, and the vol-ume of the remaining HMX. Fig. 7C shows the surface area of the NVR in bubbles that are still growing within the grains and the NVR from the bubble that have released their gas, resulting in an increase of NVR in the inter-granular region. This NVR is now available to react with the HMX and gaseous products flowing out from the particle and may provide the initial NVR that grows into flake-like structures. Finally, Fig. 7D shows the rate of gas release from the grains of HMX. Comparison of the temporal behavior rate of gas release with the data shown in Fig-ure 2 shows how this relatively simple nuclea-tion and growth model captures the general behavior the decomposition of HMX through the acceleratory stage.

Future work will incorporate the other morphological features, as well as the chemi-cal reactions, into a mathematical model that can be used to characterize the entire HMX reaction cycle.

CONCLUSIONS

The decomposition of HMX in the solid phase at slow heating rates is controlled by a com-plex set of coupled physical and chemical processes. These physicochemical processes lead to the emergence of distinct morphologi-cal and chemical features in a sample that are incorporated into the rate controlling reaction processes as the reaction progresses. Analysis of the data from STMBMS experiments over a wide range of conditions has provided the basis for creation of a reaction cycle that de-scribes the processes that control the thermal decomposition of HMX. The reaction cycle summarizes 9 physical processes and 7 gen-eral chemical reactions that play a major role in controlling the thermal decomposition of HMX. Mathematical representations of these

processes are under development with the ul-timate goal of creating predictive models that can be used to characterize the type and ex-tent of damage created in HMX based explo-sives that have been exposed to fires.

ACKNOWLEDGEMENTS

The author would like to thank Surya Bu-lusu (deceased), Sabrina Mack, Jeanette Wood for their efforts in collecting and ana-lyzing the data over the years. We would also like to acknowledge the support provided in part by the U.S. Army Research Office, under contract No. 38302-CH, The DoD Office of Munitions and the U.S. Department of Energy under Contract No. DE-AC0494AL85000.

REFERENCES 1. Behrens, R., Jr., Thermal Decomposition of Ener-

getic Materials: Temporal Behaviors of the Rates of Formation of Gaseous Pyrolysis Products from Condensed-Phase Decomposition of Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine. Journal of Physical Chemistry, 1990. 94: p. 6706-6718.

2. Field, J.E., et al., Hot-Spot Ignition Mechanisms for Explosives and Propellants. Philosophical Transactions of the Royal Society of London Se-ries a-Physical Sciences and Engineering, 1992. 339(1654): p. 269-283.

3. Belmas, R. and J.P. Plotard, Physical Origin of Hot-Spots in Pressed Explosive Compositions. Journal De Physique Iv, 1995. 5(C4): p. 61-87.

4. Mellor, A.M., D.A. Wiegand, and K.B. Isom, Hot-Spot Histories In Energetic Materials. Combustion and Flame, 1995. 101(#1-2): p. 26-35.

5. Luebcke, P.E., P.M. Dickson, and J.E. Field, De-flagration-to-detonation transition in granular pentaerythritol tetranitrate. Journal of Applied Physics, 1996. 79(7): p. 3499-3503.

6. Tokmakoff, A., M.D. Fayer, and D.D. Dlott, Chemical-Reaction Initiation and Hot-Spot For-mation In Shocked Energetic Molecular Materials. Journal Of Physical Chemistry, 1993. 97(#9): p. 1901-1913.

7. Tarver, C.M., S.K. Chidester, and A.L. Nichols, Critical Conditions For Impact-Induced and Shock-Induced Hot-Spots In Solid Explosives. Journal Of Physical Chemistry, 1996. 100(#14): p. 5794-5799.

8. Huang, C.C. and T.S. Wu, Obtaining Kinetic Data Of Energetic Material Decomposition By Differen-tial Thermal-Analysis. Thermochimica Acta, 1992. 204(#2): p. 239-250.

9. Rogers, R.N., Differential Scanning Calorimetric Determination of Kinetics Constants of Systems that Melt with Decomposition. Thermchimica Acta, 1972. 3: p. 437-447.

10. Vyazovkin, S. and C.A. Wight, Model-free and model-fitting approaches to kinetic analysis of iso-thermal and nonisothermal data. Thermochimica Acta, 1999. 341: p. 53-68.

11. Hussain, G. and G.J. Rees, Thermal-Decomposition Of Hmx and Mixtures. Propellants Explosives Pyrotechnics, 1995. 20(#2): p. 74-78.

12. Rocchio, J.J. and A.A. Juhasz, eds. HMX Thermal Decomposition Chemistry and Its Relation to HMX-Composite Propellant Chemistry. 11th JANNAF Combustion Meeting. Vol. 261. 1974, Chemical Propulsion Information Agency. 247-265.

13. Behrens, R., Jr., Simultaneous Thermogravimetric Modulated Beam Mass Spectrometry and Time-of-flight Velocity-Spectra Measurements: Thermal Decomposition of RDX and HMX. 24th JANNAF Combustion Meeting. Vol. 476. 1987, Chemical Propulsion Information Agency: Monterey, Cali-fornia. 333-342.

14. Behrens, R. and S. Bulusu, Thermal-Decomposition Of Energetic Materials .2. Deute-rium-Isotope Effects and Isotopic Scrambling In Condensed-Phase Decomposition Of Octahydro-1,3,5,7-Tetranitro-1,3,5,7-Tetrazocine. Journal Of Physical Chemistry, 1991. 95(#15): p. 5838-5845.

15. Behrens, R. and S. Bulusu, Thermal Decomposi-tion of HMX: Low Temperature Reaction Kinetics and their Use for Assessing Response in Abnormal Thermal Environments and Implications for Long-Term Aging. Decomposition, Combustion and Detonation Chemistry of Energetic Materials, ed. T.B. Brill, et al. Vol. 418. 1996, Material Research Society: Boston, MA. 119 - 126.

16. Behrens, R., S. Mack, and J. Wood, Thermal De-composition Mechanisms of HMX: The Interrela-tionship of Chemical and Physical Processes. JANNAF 17th Propulsion Systems Hazards Sub-committee Meeting. Vol. CPIA Publication #681, Vol. 1. 1998, CPIA: Chemical Propulsion Information Agency. 21-44.

17. Behrens, R. and S. Bulusu, Thermal-Decomposition Of Energetic Materials .3. Tempo-ral Behaviors Of the Rates Of Formation Of the Gaseous Pyrolysis Products From Condensed-Phase Decomposition Of 1,3,5-Trinitrohexahydro-

S-Triazine. Journal Of Physical Chemistry, 1992. 96(#22): p. 8877-8891.

18. Behrens, R. and S. Bulusu, Thermal-Decomposition Of Energetic Materials .4. Deute-rium-Isotope Effects and Isotopic Scrambling (H/D; C-13/O-18; N-14/N-15) In Condensed-Phase Decomposition Of 1,3,5-Trinitrohexahydro-S-Triazine. Journal Of Physical Chemistry, 1992. 96(#22): p. 8891-8897.

19. Maharrey, S., R. Behrens, and L. Johnston, Physi-cal and Chemical Processes Controlling the Solid-Phase Thermal Decomposition of hexahydro-1,3,5-trinitro-s-triazine (RDX). CPIA, presented at the 37th JANNAF Combustion Meeting, 2000: p. to appear in 2001.

20. Behrens, R., Jr., New simultaneous thermogravim-etry and modulated molecular beam mass spec-trometry apparatus for quantitative thermal de-composition studies. Review of Scientific Instru-ments, 1987. 58(3): p. 451-461.

21. Behrens, R., Jr., Identification of Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) Pyro-lysis Products by Simultaneous Thermogravimetric Modulated Beam Mass Spectrometry and Time-of-Flight Velocity-Spectra Measurements. Interna-tional Journal of Chemical Kinetics, 1990. 22: p. 135-157.

22. Behrens, R., Jr., Determination of the Rates of Formation of Gaseous Products from the Pyrolysis of Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) by Simultaneous Thermogravimetric Modulated Beam Mass Spectrometry. International Journal of Chemical Kinetics, 1990. 22: p. 159-173.

23. Behrens, R. and S. Maharrey, Chemical and Physical Processes that Control the Thermal De-composition of RDX and HMX, in Combustion of Energetic Materials, K.Kuo and.L.T. DeLuca, Editor. 2002, Begell House: New York. p. 3 - 21.

24. Behrens, R., Jr., T.A. Land, and S. Bulusu, Ther-mal Decomposition Mechanisms of 1-Nitroso-3,5-dinitro-s-triazine (ONDNTA). 30th JANNAF Combustion Meeting. Vol. CPIA-PUB-606-VOL-II. 1993, Chemical Propulsion Information Agency: Monterey, California. 47-59.

25. Behrens, R. and S. Bulusu, The Importance of Mononitroso Analogues of Cyclic Nitramines to the Assessment and the Safety of HMX-Based Pro-pellants and Explosives, in Challenges in Propel-lants and Combustion 100 Years after Nobel, K.K. Kuo, Editor. 1997, Begell House, Inc.: New York. p. 275 - 289.

26. Behrens, R., S.B. Margolis, and M.L. Hobbs, A Zero-Dimensional Model of Experimental thermal

Decomposition of HMX, in 11th International Detonation Symposium. 1998, Office of Naval Re-search: Snowmass, CO. p. 533-543.