THERMAL STABILITY AND SENSITIVITY OF ENERGETIC FORMULATIONS

Explosive mixtures have found widespread use in both military applications and as components of improvised explosive devices (IEDs). Knowledge of how the components of these formulations interact with each other will benefit both military and anti-terrorism organizations. Since there are significant differences in the explosive properties desired by the military verses those involved in illicit activities, it is important to study both military and improvised formulations. The use of improvised explosive devices by terrorist organizations is a significant problem that has resulted in destruction of property and loss of both military and civilian lives in countries throughout the world. There are many different materials that can be used to make homemade explosives (HMEs), but they are often combinations of a fuel and an oxidizer. These materials are popular because they are generally readily available due to their use in various industrial and household processes. Knowledge of which fuel-oxidizer combinations are potentially dangerous can help anti-terrorist organizations focus resources on detecting potential threats and preventing the use of potential HME components. In the first manuscript, titled “Fueloxidizer Mixtures: Their Stabilities and Burn Characteristics”, various fuel/oxidizer combinations were examined by differential scanning calorimetry (DSC) and simultaneous differential scanning calorimetry/thermogravimetric analysis (SDT). It was found that the reaction between the fuel and the oxidizer was generally triggered by a thermal event such as a melt, phase change, or decomposition. When the fuel used was a polyalcohol or sulfur, the triggering event was often the melt of the fuel, which usually occurred at a lower temperature than that of the oxidizer. However, three of the oxidizers, potassium nitrate, potassium perchlorate, and ammonium perchlorate, generally did not react until they underwent a phase change or began to decompose, and as a result, reactions with these oxidizers tended to occur at much higher temperatures. Reactions with hydrocarbon fuels containing fewer or no alcohol groups also tended to occur at higher temperatures. Regardless of the fuel used, the mixtures containing potassium chlorate, ammonium perchlorate and ammonium nitrate generally released the greatest amount of heat, around 2000 J/g, while mixtures containing potassium dichromate were the least energetic, generally releasing less than 200 J/g. For some formulations, reactions did not occur until temperatures higher than 500°C. In order to reach higher temperatures, it was necessary to use unsealed samples in the SDT rather than the sealed capillaries used in the DSC. It was noted that when samples were not in sealed capillaries, other processes such as sublimation effectively competed with the exothermic reactions experienced by the formulations. As a result, the heat release values obtained by SDT for some formulations were artificially low. The second manuscript “Thermal Stability Studies on IMX-101 (Dinitroanisole/Nitroguanidine/NTO)” examines the interactions among the components of an insensitive munitions formulation, IMX-101, which has been developed and qualified for use as a replacement for TNT (2,4,6-trinitrotoluene). IMX-101 contains the energetic materials 2,4-dinitroanisole (DNAN), nitroguanidine (NQ), and 3-nitro-1,2,4-triazol-5-one (NTO). DNAN is a nitroarene that is very similar in structure to TNT, but with only two nitro groups, and with an anisole functional group in place of the methyl group in TNT. 2,4-Dinitrotoluene (DNT), which also contains only two nitro groups, is even more similar to TNT, because it is a toluene rather than an anisole. DSC and isothermal analyses were used to compare DNAN and DNT, to see if increased thermal stability made the use of DNAN more appealing than DNT in IMX-101 and other insensitive munitions formulations. The isothermal studies showed that neat DNAN was more stable than neat DNT. However, when mixed with either or both of the other components of the IMX-101 formulation, the thermal stability of both DNAN and DNT was decreased, with a greater impact on DNAN. The thermal decomposition of both DNAN and DNT was significantly accelerated by the presence of NQ. NTO also enhanced the decomposition of both nitroarenes, but this compound had a significantly greater impact on DNAN than on DNT. An examination of the decomposition products from the various mixtures showed that 2,4-dinitroaniline (DNA) was produced from the decomposition of both DNAN and DNT with either of the two additives; DNA was not observed during the neat decomposition of either nitroarene. It was thought that ammonia, which has been detected in either gaseous form or as ammonium ions during decomposition studies on both NQ and NTO, might be one cause of the decreased stability imparted to the nitroarenes by the two additives. Heating DNAN and DNT in the presence of ammonia generated from ammonium carbonate produced dinitroaniline and had an accelerating effect on the decomposition of the two nitroarenes, with the greater impact, both in the acceleration level and the amount of dinitroaniline produced, on DNAN.


INTRODUCTION
Explosive mixtures have found widespread use both in military applications [1][2][3][4] and as components of improvised explosive devices (IEDs) [5][6][7]. Knowledge of how the components of these formulations interact with each other is beneficial to both military and anti-terrorism organizations. Since there are significant differences in the explosive properties desired by the military [3] and those engaged in illicit activities, it is important to study both military and improvised formulations.
For many years, 2,4,6-trinitrotoluene (TNT) has been one of the most commonly used military explosives. However, there is concern about the sensitivity of munitions based on conventional explosives, which can react violently if exposed to fire or impact from an armor piercing bullet or shape charge [3,8,[9][10][11]. In addition, TNT and some of its decomposition products have significant levels of toxicity and, as a result, unexploded ordnance containing TNT has become an environmental and health concern [8]. These issues have led to increased interest in the development of insensitive munitions formulations that can serve as replacements for TNT and other conventional explosives [3][4]. One of the main components of many of these formulations has been 2,4-dinitroanisole (DNAN) [3-4, 8, 12].
DNAN is similar to TNT in that it contains a phenyl ring substituted with nitro groups; however, it has one less nitro group than TNT, and it has a methoxy (-OCH 3 ) substituent in place of the methyl (-CH 3 ) group that is present in TNT [8]. DNT is even more similar to TNT; it lacks only the nitro group attached to carbon six in the phenyl ring. Both DNAN and DNT are energetic but less sensitive than TNT [12], so in theory either could serve as a potential component in insensitive replacements for  [13]. Both components are generally materials that are readily available because they are used for some peaceful purpose in industry and elsewhere [6,[13][14]. For example, common fuels include sucrose (table sugar) and coal, while frequently used oxidizers include ammonium nitrate, a major component of many fertilizers, and potassium chlorate, which is used in safety matches, printing, dying, and pyrotechnics [15][16]. The increased level of concern about terrorist attacks has led to a greater focus on the materials that can be used to make explosives; however, most published research has dealt with methods to detect HME components, either before or after an attack has occurred [6-7, 13, 17-18]. While there has been some research into the interactions between specific fuel-oxidizer pairs [19][20][21][22][23]      To go to temperatures above 500°C, unsealed crucibles were necessary, and with these containers, the endothermic volatilization of reactants and products effectively competed against the exothermic decomposition so that heat release values were artificially low.

Introduction
Fuel-Oxidizer (FOX) mixtures are commonly used in the pyrotechnic and mining industries, with applications ranging from oxygen sources to sources of energy and propulsion. Examples of such uses include ammonium perchlorate with hydroxyterminated polybutadiene for rocket fuel and ammonium nitrate with fuel oil for commercial mining. The wide availability of many fuels and oxidizers has also resulted in their illicit use as components of improvised explosive devices (IEDs) [1,2].
In this study, a number of solid oxidzers, with varying oxidizing power, were tested on lab-scale in mixtures with a variety of fuels. The purpose of these tests was to assess the hazard and threat potential of the different mixtures, and to allow assessment of the usefulness of small-scale tests. Many of the oxidizers were oxyhalide salts. The potassium salts were used because they tend to be less hygroscopic than those of sodium. Since ammonium salts have different chemical behavior than the potassium salts of the same anion, because they carry and use, if required, their own fuel, the ammonium salts of nitrate and perchlorate were also included in the study. The choice of fuels was limited to solids, including polyalcohols, hydrocarbons, benzoic acid, sulfur, charcoal, and aluminum.

Materials and Methods
Eleven oxidizers and twelve fuels were used in various different combinations.
All materials were reagent grade with the exception of the charcoal, which was purchased locally, and aluminum, which was pyrotechnic grade. The oxidizer/fuel mixtures were examined fuel-rich at 50/50 mass% and closer to stoichiometric at 80/20 mass%. Benzoic acid, which is often used as a burn rate modifier, and aluminum were added only at the 20 mass% level. Individual components with larger particle sizes (i.e. sugars and most oxidizers) were ground prior to mixing. Those materials that were already fine powders, such as sulfur, were used as received.
Materials used to make the 80/20 sucrose mixtures were sieved to 50-100 mesh.
Mixtures were generated by gently stirring the fuel and oxidizer together with a wooden stick or by mixing in a Resodyn LabRAM acoustic mixer (two minutes, 50% intensity, auto frequency). Batch sizes ranged from 100 mg to 1 g depending on the analyses to be performed. Samples were usually run in triplicate, but where marked variations in the thermograms were apparent, up to seven samples were run. Variations in the detailed appearance of the DSC thermograms were likely a result of inhomogeneities in the oxidizer and fuel mix, especially considering that samples were usually less than a milligram. Because multiple endotherms and exotherms were often observed in the DSC and SDT traces, and because many of the exotherms covered a wide temperature range, the major exotherm of a trace is usually reported with either the onset temperature or the temperature at which a deviation from baseline was initially detected, followed by the temperature(s) at which "peak maxima" were observed, with the highest in bold, and the heat of reaction in parentheses (Jg -1 ) calculated from peak area using baselines established by the operator.
For burn tests the oxidizers and sucrose were dried overnight in a vacuum oven at 50°C, then ground and sieved to 50-100 mesh. Pyrotechnic-grade (median particle size 23 μm) aluminum powder (Obron) was used. Samples were mixed with a as zero), is one approach to quantifying oxidizing power. Standard reduction potentials are listed below starting from the left with species having most positive potential [4,5]: Actual potentials depend on the pH of the solution and the final products: An alternative approach to rating oxidizing power is a burn test. The U.N.
Manual of Tests and Criteria rates an oxidizer by comparing its burn rate in admixture with cellulose (2:3 and 3:7 ratios) to mixtures of potassium bromate/cellulose [6]. Our burn tests used 250 mg instead of 30 g of material, and sucrose or aluminum powder instead of cellulose. Burn rates are shown in Table 1. Thermal stability was assessed via the temperature at peak maximum of the DSC exotherm. The higher the exotherm temperature, the more thermally stable the species. Some salts decomposed with an exclusively endothermic response (Table 2).
Among salts releasing heat (exothermic response), the amount of heat varied from more than 1000 Jg -1 for ammonium salts, which can undergo self-oxidation, to a few hundred joules per gram for other oxidizers. Thermal traces of the oxidizers alone were not simple; they included phase change(s), decompositions, and heats of fusion of the decomposition product. In systems where oxygen was not allowed to escape, the pairs perchlorate/ chlorate [7][8][9] and nitrate/nitrite [10,11] can establish a psuedoequilibrium (eq 1-2) [10]. At high temperatures the melts of KCl, K 2 O, and KI were observed, and the DSC traces showed the decomposition of periodate to iodate around 330°C (eq 3); thereafter, their thermograms were identical [12][13][14][15][16][17]. (~1300 Jg -1 ). The SDT results appeared quite different. Immediately after the 245°C phase change, a small exotherm (~360 Jg -1 ) at ~318°C was observed followed by a second endotherm centered around 435°C (Fig. 1). This apparent difference in AP behavior has been explained by the sublimation of AP above 350°C competing with its decomposition [19,20]. Sublimation can be dramatically reduced by pressure; thus, when possible sealed DSC pans were used [20]. As heating of the open pan in SDT was continued, a small endotherm at 757°C was observed for the melt of KCl.  Table 2.
An advantage of SDT thermal analysis was that it allowed scanning to higher temperatures. However, since the crucibles were not sealed, the SDT thermal traces

Fig. 5 KIO 4 + 20 mass% Sucrose
For three oxidizers this general trend with sugars was not observed. These oxidizers may have exhibited a small exotherm immediately after the sugar melt, but the majority of the exothermic reaction only occurred after the oxidizer underwent a melt, phase change, or decomposition, and we labeled them "oxidizer-controlled." The two resistant anions were perchlorate and nitrate, but for the latter, nitrate, only the potassium salt failed to react immediately after the sugar melt. This counter-trend was true regardless of the type of sugar ( Fig. 6 and Fig. 7). Generally, the thermograms did not change drastically in appearance when 20 mass% rather than 50 mass% sucrose was used (compare Fig. 2 and 5 or see Fig. 8).
The exception was ammonium perchlorate (AP), one of the three oxidizers resistant to sugar melt. With 50 mass% sucrose a wide exotherm was observed immediately after the melt of sucrose and a second exotherm started about 270°C. With only 20 mass% sucrose, no exotherm was observed until ~ 470°C, in dramatic contrast to the thermogram with 50 mass% sucrose (Fig 9).  The heat released from the oxidizers with 20 mass% sucrose was comparable (~1400 Jg -1 ) to the heat released with 50 mass% sucrose (Table 4). There was a large deviation in observed heat released (+25%) run to run which we have attributed to the slow response of the DSC thermocouples. K 2 Cr 2 O 7 fuel mixtures were notably low in energy release, averaging less than a tenth of the other fuel/oxidizer mixtures (Table   4).  [42]. Only two oxidizers with erythritol (KMnO 4 and KIO 4 ) showed immediate decomposition after the melt of erythritol, although all the "sugar-controlled" oxidizers that were examined with this fuel decomposed at lower temperatures than their own phase changes or decomposition point (Fig. 11). Five of the oxidizers were heated with pentaerythritol. KClO 3 and KBrO 3 , which had been labeled "sugar-controlled", remained triggered by the fuel, while KNO 3 remained oxidizer controlled. AN, which with the four sugars exhibited an exotherm around 170°C, did not react with the melt of pentaerythritol (PE) at 190°C. Instead it began to release heat around 260°C, a phase change for PE. In some thermograms the exotherm at 260°C was the only peak; in others a second peak was observed at the normal decomposition temperature of AN ( Fig. 12). Potassium dichromate, which was one of the "sugar controlled" oxidizers, did not react near the melting point of pentaerythritol, but showed a small exotherm following its own melting point around 400°C. To examine samples that did not exhibit heat releases in the temperature range of the DSC, SDT was used. Because the SDT was designed to allow monitoring of mass loss as well as heat flow, samples were scanned unsealed. This immediately proved to be a problem. In some cases exothermic events appeared as endothermic events; a classic example is a scan of an unsealed sample of AN. When not contained in a sealed ampoule, AN will show an endotherm around its 300°C decomposition rather than the actual exotherm (Fig. 13). Occassionally the exothermic event was only partially countered by the endothermic evaporation of the reactant or products; in such cases the exotherm was observed, but heat release was significantly lower than it would have been in a sealed sample. Therefore, whenever possible, sealed samples were examined by DSC. To date we have found no satisfactory method for sealing DSC samples that remains gas tight over 550°C.   [43], and has long been used in energetic formulations [44,45], exhibited behavior much more similar to that of the sugars. We observed two, and sometimes three, endotherms between 107 and 120°C, assigned to phase change and melting, and there was also a small endotherm around 180°C. The oxidizers that were initiated by the sugar melt also showed exothermic decomposition with sulfur beginning around 180°C. A common characteristic of this exothermic decomposition was slow heat release rising to a recognizable exotherm (Fig. 14). The same three oxidizers classified as oxidizercontrolled do not show an exotherm until higher temperatures (Fig. 15).   when KBrO 3 is mixed with naphthalene ( Fig. 16), the difference between the onset and first deviation from baseline is not large (~30°C), but for KIO 3 and sulfur (Fig. 17) the difference between the calculated onset and the deviation from baseline is huge (~160°C). (Note that this trace of oxidizer and sulfur is typical for sulfur mixtures.)  The temperature at which an oxidizer/fuel mixture begins to react depends on both the susceptibility of the fuel to oxidation and the oxidizer's tendency to be reduced. In comparing the carbonaous fuels, cyclododecanol, hexatriacontane, naphthalene, benzoic acid, and charcoal, we had hoped to see a reactivity trend across all oxidizers, and, indeed, the following trend in the initiation temperature of the decomposition exotherm was observed with over half the oxidizers:

Fig 16 KBrO 3 + 50 mass% Naphthalene
benzoic acid < cyclododecanol ~ hexatriacontane < charcoal < napththalene Interestingly, hexatriacontane and cyclododecanol, which had boiling points only one degree apart, produced DSC traces almost identical to each other, suggesting the reaction of cyclododecanol was that of a hydrocarbon rather than an alcohol. It was also observed that with fuels other than aluminum, ammonium perchlorate mixtures decomposed at lower temperatures than those with potasssium perchlorate.
Aluminum, which has been used as a fuel in mixtures with ammonium nitrate and perchlorate, was used as the highest melting fuel. With two exceptions, the oxidizers decomposed long before the aluminum reacted, and aluminum did not react until over 800°C (Fig. 18, 19). In two cases (KClO 3 , KNO 2 ) the exotherm appeared at a significantly lower temperature indicating that these oxidizers react readily with the aluminum (Fig. 20, 21). All the oxidizer/Al samples were examined by SDT, and a few were also examined by DSC. The low temperature exotherm observed for KIO 4 was its conversion into KIO 3 . The low temperature (i.e. under 800°C) exotherms recorded for other oxidzers reflect the decomposition of the oxidizer.  When fuels were added to the oxidizer, the phase changes of the individual oxidizers and fuels were often still observed. Most of the fuels were added at the 50 mass% level, but thermograms of 20 mass% sucrose were examined and shown to be very similar to 50 mass% sucrose in terms of appearance and heat release. Variations in appearance and heat release (+25%) were attributed to inhomogeneity in the samples and variations in particle size [48][49][50][51], although even neat ammonium nitrate exhibited 15% variation in heat release. We suspect that with energetic materials it is difficult for the DSC thermocouples to accurately track the fast release of heat.
Differences in DSC and SDT traces appeared to be related to the ability of reactants/products to vaporize in the open or lightly capped SDT containers.
We found that a phase change in the fuel or oxidizer or decomposition of the oxidizer typically was the trigger causing their reaction; therefore, we classified the  (Table 4). Oxidizers consistently releasing the most heat were KClO 3 , AP, and AN. (The heat release values for potassium perchlorate may have been artifically low due to the fact they were only observable by SDT, which could allow material to escape prior to the exothermic event). The amount of heat released appeared dependent on the oxidizer rather than the fuel. No fuel stood out as clearly the 'best'; they averaged 1500 Jg -1 by DSC analysis.
Response to hot-wire ignition was assessed by the length of the burn and the light output. Table 1 orders the oxidizers left to right as highest oxidizing power to lowest in terms of electromotive potential; this roughly followed their thermal stability. Light output, when the fuel was aluminium, also roughly followed this trend.

Thermal Stability Studies on IMX-101 (Dinitroanisole/Nitroguanidine/NTO) Abstract
The also have initiatives to develop and use less sensitive munitions [2][3][4]. The mandate that explosive materials be safer is not easily met. An insensitive munitions formulation should have good explosive properties but should also be thermally stable and must not react violently when subjected to unplanned events [5]. In general, materials that have good explosive properties are not particularly thermally stable and tend to be sensitive to accidental ignition [4]. Finding materials that have acceptable explosive performance and low sensitivity has proven to be quite difficult, even after many years of research. In general, the quest for insensitive munitions has followed one of two approaches; either explosive materials are encased in less sensitive polymeric materials to form polymer bonded explosives (PBXs), or entirely new explosive formulations are developed [5,6].
One of the early candidates for a TNT replacement was 2,4-dinitroanisole (DNAN), which, like TNT, is melt-castable but also toxic [7][8][9][10]. The use of DNAN is not new; it was used as part of the explosive formulation Amatol 40 during WWII, but this use was most likely due to a shortage of TNT rather than because of concerns about the sensitivity of munitions [7,11]. More recently, the improved sensitivity of DNAN based munitions has led to the development of numerous DNAN formulations, some of which have been qualified for use by the U.S. National Service Authority [11][12][13][14][15][16][17][18]. The TNT replacement IMX-101, which contains 43.5% DNAN, 19.7% 3-nitro-1,2,4-triazol-5-one (NTO), and 36.8% nitroguanidine (NQ), was certified for use in 2010 [16]. IMX-101 is listed as having a theoretical maximum density of 1.67 g/cc and a detonation velocity of 6900 m/s; it has passed the STANAG fast and slow heating tests 4240 and 4382, respectively [16,19]. However, while IMX-101 did pass the various ageing and stability tests to which is has been subjected, it has not always shown better results than TNT and RDX. In the vacuum thermal stability test at 100°C, IMX-101 evolved slightly more gas than TNT and RDX, and events occurred at lower temperatures than TNT in the Woods Metal bath and Henkin time to explosion tests, and at lower temperatures than both TNT and RDX in the 1-liter spherical cook-off test. IMX-101 did outperform both TNT and RDX in the smallscale ESD, ERL/Bruceton impact, and BAM friction tests [16]. This study examined  Note, had a flow rate of 0.75 mL/min [22]. The initial eluent was 26% aqueous acetonitrile; the organic component was increased to 40% then 55% over two ten minute increments, and then raised to 100% over the next 14 minutes. The system was held at 100% acetonitrile for one minute before returning to the initial composition.
Each set of samples was accompanied by combined DNAN and DNT standards covering the range from 0.5 to 100 μg/mL, and yielding R 2 values that were generally 0.999 or better. New standards were made when the R 2 values fell below this value.
Chromatographic results were analyzed using the Agilent Chem Station software.  were introduced into an initial mobile phase of 98% solvent A (water) and 2% solvent B (acetonitrile). Following injection, this was held for 1 minute and then ramped linearly to 98% solvent B and 2% solvent A over 7 minutes. This was held for 1 minute before returning to initial conditions over 30 seconds and re-equilibrated for 2.5 minutes prior to the next injection (total run time of 12 minutes). Because the polarities of these 4 components vary significantly, some compromise was required in this analysis. Normal phase methods retained NTO and NQ, but DNT and DNAN eluted close to the void. Other reverse phase columns retain DNT and DNAN very well, but peak shape, retention or resolution of NTO and NQ were unacceptable. The chosen method provided good peak shape for all compounds with reasonable resolution of all components; however, NTO did not retain acceptably. Since the intention of this analysis was to determine the decomposition products of these compound mixtures, this compromise was made because NTO decomposition is well documented. Compounds smaller than NTO that may elute earlier are not likely to be detected with this system. Data collection and analysis was performed with Thermo Xcalibur software version 2.2, SP 1.48.

Results
Differential Scanning Calorimetry (DSC): DSC scans obtained at a scan rate of 20°C /minute confirmed that DNAN was somewhat more stable than DNT. Both these nitroarenes are more thermally stable than NQ and NTO, and slightly more stable than TNT (see Figure 2).  To emulate IMX-101 a three-part mix of DNAN (43%), NTO (20%), and NQ (37%) and a similar one using DNT instead of DNAN were scanned by DSC (Figure 4 a & b). Both three-part mixtures showed a large broad exotherm immediately after an endotherm around 215°C, which was assumed to be the melt of nitroguanidine, though slightly depressed. The average total heat released by the DNAN three-part mixture was slightly more than that produced by the DNT three-part mixture (2900+250 J/g vs. 2100+230 J/g, respectively). It has been claimed that the DSC of the three-part mixture IMX 101 appears to be the superposition of the individual components; therefore, the decomposition of one does not affect the other [23]. This was not found to be the case. There was sufficient heat being generated at temperatures below the decomposition exotherm of neat DNAN or neat DNT that decomposition of the three-part mixture was nearly complete by that temperature. The results from this study do agree with Cuddy's findings that decomposition of the IMX-101 mixture begins below 200°C. To examine these observations in more detail, isothermal analyses were performed.   Because ammonia is a likely decomposition product of both NQ and NTO, the rate of decomposition of DNAN and DNT under ammonia was examined at 200°C. Figure 6 shows that while ammonia has an accelerating effect on the decomposition of both nitroarenes, the impact on DNAN is much larger.

Decomposition Products
Both DNT and DNAN can undergo oxidation of the methyl or methoxy group, reduction of the nitro groups, Meisenheimer complex formation and various oligomerization reactions . DNT has been shown to experience elimination of a nitro group to form p-and o-nitrotoluene and, under aerobic conditions, to eventually yield nitrite and catechols [35,37]. Anaerobic reduction and biotransformation of DNT produces nitroso-, amino-, aminonitro-, and diaminotoluenes, as well as azoxy compounds [29,34,35]. For DNAN, loss of the methoxy group to yield 2,4 dinitrophenol has been reported under various different reaction conditions, such as mammalian metabolism and reactions with piperidines and sodium hydroxide [11,[38][39][40]. The methoxy group has also been shown to undergo aromatic and aliphatic nucleophilic substitution reactions resulting in the replacement of either the methyl or the entire methoxy group by amines or other nucleophiles [11,39]. As with DNT, the nitro groups of DNAN can be reduced microbially to form amino-and aminonitroanisole [12,[29][30][31][32]; arylnitroso and arylhydroxylamino intermediates, azoxy-and azo-dimers, demethylated and acetylated products, and ring cleavage have also been reported [33,41].
Under our experimental conditions, in which DNT and DNAN were heated in glass capillaries at 200°C for four and five days, respectively, to achieve approximately 50% decomposition, numerous products were observed (Figures 7 and   8). Assignment of chemical formulas was based on the high-resolution mass spectrometry results where compositions could be determined within 5 ppm of their calculated mass. Masses associated with proposed structures here, and throughout the paper, are those obtained from the LC/MS for the M-1 adducts detected in negative ion mode. (Table 4):   Nitroguanidine is reported to exist in two tautomeric forms (Figure 9) with A being predominant under all but extremely basic conditions [42,43].
In addition to the production of ammonium ions through electrochemical reduction, Fan also noted the creation of ammonia, in the form of ammonium ions, via thermal decomposition [54]. When NTO was thermally decomposed with TNT, the products observed, which included TO, triazole, 2,4-DNT, 2,6-DNT, trinitrobenzene (TNB), and aminodinitrobenzoic acid, were similar to the products observed when TNT and NTO were decomposed alone [55]. A number of studies have reported the decomposition kinetics of NTO [52,62], and labeling studies have been used to elucidate the decomposition mechanisms [54]. A number of routes have been proposed, including bond homolysis with or without hydrogen transfer, mono-or bi-molecular nitro-nitrite rearrangement, or a combination of both [51,54,[63][64][65]. These may be manifast by the evolution first of NO 2 , HONO, or CO 2 , but in all cases a polymeric residue results. Various proposed decompostion mechanisms, as summarized by Smith, are shown in Figure 11 [62]. all other decomposition products that were identified in the mixtures contained some form of the phenyl ring from the nitroarene.
When DNAN or DNT was heated at 200°C with NQ, most of the decomposition products appeared to be related to NQ addition to the nitroarene; however, formation of dinitrophenol from DNAN and oxidation processes (e.g. conversion of the methyl group in DNT to a carboxyl group) had also occurred ( Table   4, Figures 12 & 13).

Discussion
Dinitrophenol and dinitroaniline were observed in the decomposition of DNAN with either NQ or NTO. While dinitrophenol was also observed in the thermolysis of neat DNAN, dinitroaniline was only observed when NQ or NTO was present. Though dinitroaniline was also detected as a product when DNT was decomposed with either NQ or NTO, the formation of DNA was two orders of magnitude greater for DNAN than for DNT with either species. This is likely due to the ease with which the methoxy group is lost from DNAN as compared to methyl loss from DNT. Dinitroaniline could be formed from DNAN if, after loss of methoxy, NQ added to the ring and then was subsequently lost. However, NTO would have no such route available to it. NQ and NTO have been observed to generate ammonia or ammonium ions [42][43][44][45][46][47]50,54,59], which may replace the methoxy group with amine via a substitution reaction. Multiple studies have demonstrated DNAN's ability to undergo nucleophilic substitution reactions, and amines were the nucleophiles in some of those experiments. [11,39]. Dinitroaniline was also formed when DNT was heated in the presence of NQ or NTO, but at significantly lower levels than were observed with DNAN. Again we attribute this to the reaction of the DNT with ammonia generated from the thermolysis of NQ or NTO.
In order to examine the impact of ammonia on the decomposition of DNAN Previous studies have shown that ammonia has an acceleratory effect on the decomposition of TNT [55], and ammonia has been detected as a decomposition product of mixtures of TNT and NQ [66].
A comparison of the effects of NQ and NTO on the nitroarenes indicates that, while NQ has a similar impact on the thermal stability of both DNAN and DNT, NTO, like ammonia, has a much greater acceleratory effect on the rate of decomposition of DNAN than it does on DNT. The observation that NTO has more of a destabilizing affect on DNAN than on DNT may be related to the fact that the principle way NTO affects DNT is not via replacement of the methyl group, which is a much poorer leaving group than a methoxy, but via hydrogen transfer. We, as well as Menapace, have reported the decomposition of TNT with NTO over the temperature range 220-280°C [55,67]. In those studies, NTO accelerated the decomposition of TNT 10-fold, while TNT accelerated the decomposition of NTO 100-fold. Using deuterated analogs, Menapace found that the NH group of NTO favored reaction with the nitro groups of TNT via a process involving hydrogen abstraction, and that a similar hydrogen abstraction process also occurred between the nitro group on NTO and the methyl hydrogens on TNT. Hydrogen abstraction from TNT did not result in a loss of the methyl group, but instead produced aryl hydroxyl nitroxide radical adducts, such as that shown in Figure 18, which are similar to some products observed in our thermolysis of DNT. The decomposition of neat DNT yielded primarily molecules containing oxidized forms of the methyl group and species best described as linked DNT molecules. When DNT was thermolyzed with NTO and/or NQ, most products that were not present in the decomposition of neat DNT were again derived from NTO, NQ or combinations/fragments of these materials attached to the arene ring.  /z 178, 190, 205, 244, 246, 247).

Conclusion
The addition of NQ accelerated the decomposition of both DNAN and DNT by approximately two-orders of magnitude. A similar acceleratory effect was seen when DNAN was decomposed with NTO; however, NTO only increased the decomposition of DNT by one-order of magnitude. As would be expected from the results of the two part mixtures, both nitroarenes decomposed faster in the NTO/NQ mixture; however, DNT was not as severely accelerated as DNAN. NTO decomposed a little slower in the 50:50 DNT/NTO mixture than in the corresponding DNAN/NTO mix. In contrast, NQ decomposed a little faster in DNT/NQ than in DNAN/NQ. An IMX 101 mixture using 2,4-dinitrotoluene rather than 2,4-dinitroanisole would be more thermally stable although not as energetic. Some evidence suggests that DNAN might be more toxic than TNT [7][8]; thus, using DNT might provide a less toxic mix. Furthermore, 2,4-DNT is a widely used chemical since it is an intermediate in the production of toluene diisocyanate (TDI) used in polyurethane production. As a result it is relatively inexpensive and widely available. Despite these apparent advantages of DNT over DNAN, the 20°C higher melting point of DNAN may continue to favor it, and improvements to the formulation may come from removing NQ entirely.