Factors Influence the Safety of Unconventional Explosives

Factors Influencing Safety of Unconventional Explosives This dissertation details three studies regarding safety aspects for the formation, decomposition and use of unconventional explosives. The first study is a mechanistic study for the formation and decomposition of triacetone triperoxide (TATP). Using GC-MS, LC-MS, and NMR, the mechanism for the formation of TATP was elucidated detailing how experimental conditions affect the product composition. The presence of water had a significant impact on the distribution of the products TATP and diacetone diperoxide (DADP), a common contaminant in the synthesis. Water also had an impact on the decomposition of TATP resulting in a slower and more complete decomposition. The second study pertains to copper acetylide, a primary explosive sensitive towards initiation by impact, friction or spark. Copper acetylide is regarded as a safety risk in the petrochemical industry due to the presence of copper catalysts in refinery gas streams contaminated with acetylene. Analysis of the products formed after catalyst samples were exposed to acetylene gas suggested that acetylene readily reacts with many copper catalysts, likely via a copper acetylide intermediate, to form an amorphous phase of carbon. It was found that proprietary catalyst compositions inhibit the reaction significantly, reducing the potential risk of acetylide formation and subsequent accidental explosion. The third study is aimed at mitigating potential failure of direct borohydride fuel cells. The incompatibility of sodium borohydride and hydrogen peroxide poses a serious safety risk for electrochemical cells using these substances. The hydrolysis kinetics of sodium borohydride in dilute hydrogen peroxide solutions was studied to obtain fundamental information regarding the reaction. Hydrolysis by water is slow and yields hydrogen gas while hydrolysis by hydrogen peroxide is fast and yields a potentially combustible mixture of hydrogen and oxygen gases.


Introduction
Organic peroxides are often used as polymerization catalysts or bleaching agents [1][2][3]. However, a few with high ratios of peroxide functionality to ketone have found use as illicit explosives [4]. We have previously reported attempts to prevent synthesis of TATP in improvised settings [5]. That work pointed out a need for a Syntheses of TATP and DADP were previously reported [5]. For synthesis of deuterium-labeled TATP d 6 acetone was used. Precipitates were filtered, rinsed with water, dried under aspiration 30 minutes and re-crystallized in methanol. Final products were analyzed by GC/MS, GC/uECD, 1 H NMR and 13 C NMR. Anhydrous hydrogen peroxide was prepared by dissolving 20 g of L-serine in 20 mL 65 wt% hydrogen peroxide. 6 To confirm its concentration, the hydrogen peroxide was dissolved in acetonitrile (ACN) and titrated with 0.25N potassium permanaganate.

Effect of water (GC/MS):
Acetone and HP were mixed and chilled to 0°C.
Water was mixed with sulfuric or hydrochloric acid, chilled to 0°C, and added dropwise to the acetone/HP mixture keeping the temperature below 5°C. The ratio of HP:acetone:acid was maintained at 1:1:1 (8.6 mmol). Once all acid was added the mixture was removed from the ice water bath and allowed to stir at room temperature 24 hours; resulting products were analyzed by GC/MS. Resolution was set to high (50,000 at 2 Hz), and the maximum injection time was 250 ms.

TATP Formation (LC/MS):
Acetone and HP (67 wt%) mixtures were prepared as molar ratios of 1:1, 2:1 and 1:2 and held at room temperature (r.t.) without stirring. For LC/MS analysis 100 uL of each mixture was diluted to 1 mL with methanol.

Nuclear Magnetic Resonance (NMR) Method
A Bruker Avance III nuclear magnetic resonance (NMR) spectrometer with 7.

Formation of TATP with acid
Previous studies have shown that the best yield of TATP is obtained from a 1:1 mole ratio of acetone and hydrogen peroxide [5,7]. When using an acid catalyst such as hydrochloric or sulfuric acid a white precipitate is quickly formed that can be washed and re-crystallized yielding high purity TATP, DADP or a mixture of the two [5,8].
To fully understand the mechanism of TATP and DADP formation it was necessary to conduct experiments using a co-solvent that would prevent precipitation from solution and not interfere with the analysis of the products and intermediates. Using GC/MS and NMR, TATP, DADP and intermediate species were observed and monitored over time.

Formation of TATP with no acid
GC/MS analysis of the products when 70 wt% hydrogen peroxide (HP) and acetone were mixed highlighted the importance of the ratios. When HP was in excess 5:1 over acetone more TATP was produced than DADP. When the ratio of HP to acetone was adjusted from 5:1 to 1:1 and then to 1:5, the total amount of solid product decreased and the amount of DADP increased relative to TATP. The reaction between HP (70 wt%) and acetone without acid was monitored for up to 14 days. A number of peaks appeared in 1 H NMR and 13 C spectra as well as in GC/MS chromatogram/spectra.
Assignments of intermediates by NMR and GC/MS are given in  and DADP (108.7 ppm CO) did not become discernible until day 5, although their presence was detected on day 1 using GC/MS (  favored, and increasing amount of HP enhances their formation. Only TATP and DADP precipitated under reaction conditions where acid was present and when no cosolvent, employed. In the absence of acid, solid TATP precipitated when the samples were aged at room temperature for up to two months. The asymmetric peroxides and longer chain oligomers were not observed by GC/MS, but using LC/MS they were observed in trace amounts.   Although 2-hydroxy-2-hydroperoxypropane (I) was most abundant when there was no acid catalyst, formation of 2,2-dihydroperoxypropane (II) was favored under acidic conditions [7,11]. TATP formation was greatly accelerated by addition of acid; yet acid also caused TATP decomposition, as evident from following 1

Effect of Water
Previously reported results indicate that a change in the concentration of acid and hydrogen peroxide can dramatically affect the outcome of TATP syntheses [5]. Dilute reagents result in poor yield of solid products, and concentrated reagents in the presence of higher acid loadings result in increased DADP formation. Water appears to play an important role in the synthesis. In an attempt to understand how water affects DADP versus TATP formation, several syntheses were attempted using 30 wt% and 50 wt% HP, acetone and concentrated sulfuric or hydrochloric acid. The ratios of the three reagents were maintained at 1:1:1 (8.6 mmol scale), but excess water, over that contributed by the reagents, was added (Table 3, Fig. 6). At the lowest levels of added water, the white solid formed was 100% DADP. At the highest levels of added water, the white solid precipitating was 100% TATP [12]. When the acid added was HCl, only a small amount of DADP was observed; this is attributed to the large amount of water (63wt%) in HCl. This phenomenon can be explained by the tendency of 2-hydroxy-2-hydroperoxypropane to disproportionate to 2,2- dihydroperoxypropane (II) and acetone in aqueous media [9]. The formation of the dihydroperoxy species appears to be a key step in TATP formation.

Effect of Solvent and Temperature
To probe whether order of reactant addition had an effect on formation of TATP vs.
DADP, it was varied (Table 1.4). The final precipitates, as well as in-situ products, were monitored by extracting aliquots of the reaction at intervals during reagent addition and analyzing by GC/MS. There were no notable differences in the final products obtained regardless of whether the acid was added first to the acetone, to the HP, or to both together (cf. exp. 1 to 3, Table 1 Table 1.4). The effect of temperature has been discussed by others without agreement [8,11,13]. Generally, TATP is favored at lower temperatures, but the effect of temperature can be manipulated by other factors such as solvent. While DADP was the major solid product in ACN at 0 o C, at lower temperatures even in ACN, TATP was favored (cf. 1 to 1' and 3 to 3', Table 4). When TFA was substituted for sulfuric acid (exp 1") no precipitate formed within the same time interval as previous reactions (~30 minutes).

Rate of TATP Decomposition
The decomposition of TATP by acid in CD 3 CN or CDCl 3 was monitored by 1 H NMR.
In ACN and CDCl 3 decomposition of TATP was pseudo-first order and formed DADP and acetone ( isopropanol or t-butanol. Methanol showed the highest rate of TATP destruction followed by ethanol, n-propanol and isopropanol. When t-butanol was used, an anomalous effect was observed. The reaction proceeded very quickly to a mixture of TATP and DADP and ceased. The decomposition of TATP by 37 wt% HCl generated some DADP, but the amount was very small compared with other acids (TFA SA) and chlorinated acetone species were detected.

Mechanism
A proposed mechanism for the production of TATP and DADP is given in Figure 1.7.
In a reaction between HP and acetone without acid catalyst, 2-hydroxy-2hydroperoxypropane (I) was observed in high quantities soon after mixing. When acid was added 2,2-dihydroperoxy-propane (II) was the primary species observed shortly after mixing [11]. Symmetric species, where two acetones are linked by a peroxide linkage, 2,2'-dihydroxy-2,2'-diisopropylperoxide (III) and 2,2'-dihydroperoxy-2,2'diisopropylperoxide (V) were also observed, but at later reaction times. The asymmetric species similar to 2-hydroxy-2'-hydroperoxy-2,2'-diisopropylperoxide (IV) were not directly observed by NMR or GC/MS and only trace amounts were observed using LC/MS. We speculate that when these are formed the hydroxyl group exchanges with a hydroperoxy group, or they rapidly convert to DADP and TATP, respectively [7]. The effect of water is apparent at this point. When water content is low, 2-hydroxy-2-hydroperoxypropane (I) can be protonated facilitating formation of (IV) and a pathway to cyclization forming DADP. When water content is high, disproportionation of (I) becomes favored resulting in the formation of the dihydroperoxy species (II) and, ultimately, the formation of TATP [9]. Under high water conditions, water, itself, is protonated; and the overall reaction proceeds more slowly. Alcohol solvents also become involved in this competition for protonation.
Formation reactions performed in methanol or ethanol under highly acidic conditions produced almost 100% TATP versus the same reactions performed in acetonitrile which produced 90 to 98% DADP (Table 1.

Conclusions
We have reported that the synthesis of TATP is achieved in best yield by use of a 1 to 1 molar ratio of HP to acetone with modest amounts of acid (10-50 mole %).
However, acid catalyzes TATP synthesis and decomposition, especially at high acid levels. Herein we examine the intermediates of the acetone/HP reaction and the rates of formation and destruction of TATP and postulate a mechanistic pathway. The oxidation of acetone by HP occurs whether or not acid is added; it is dramatically slower without acid, taking weeks and months to precipitate TATP and DADP [15]q.
Other linear and even cyclic peroxides have been observed in the reaction solution, but only TATP and DADP precipitate out due to their low solubility in aqueous media. If an organic solvent is present, TATP does not precipitate out of solution and, in the presence of acid, converts to DADP. Whether or not acid is present as soon as HP and acetone are mixed 2-hydroxy-2-hydroperoxypropane (I) forms. This then proceeds to form 2-hydroxy-2'-hydroperoxy-2,2'-diisopropylperoxide (IV) (the precursor to DADP) or 2,2'-dihydroperoxy-2,2'-diisopropylperoxide (V) (the precursor to TATP) depending on whether water content in the reaction mix is low or high, respectively.
Water, which comes into the reaction via the acid and the HP, slows the formation of TATP and DADP, DADP most dramatically. Indeed, the same effect is observed when alcohol, which like water is susceptible to protonation, is added to the reaction mixture. acetone is removed leaving hydroperoxy species [14]. The presence of completely substituted TATP also implies one of two things: the TATP molecule opens and equilibrium with the intermediates is reestablished or the ring continues to open and reclose allowing all acetone molecules to be replaced with their deuterated counterparts. Both these possibilities seem reasonable; which dominates may depend upon the solvent. Regarding formation, the insertion of acetone seems to indicate that (V) reacts with acetone to form the asymmetric hydroxy hydroperoxy trimeric species, prior to cyclizing to TATP. This supports the claim that the asymmetric species are short-lived intermediates leading to the cyclic peroxide products.

Manuscript 2 Factors Influencing Triacetone Triperoxide (TATP) and
Diacetone  There are few publications that have addressed safe, effective, field-usable methods for destroying TATP; two have suggested copper and tin salts to effect destruction at elevated temperature; 2,3 one used mineral acids and elevated temperature. 4 These articles offered guidance in the search for a room-temperature answer for gentle chemical destruction of peroxides. Ideal protocols would involve a homogeneous liquid chemical solution to spray over solid peroxide stashes or a method involving immersion of peroxide saturated materials into a solution that would quiescently destroy the explosive in hours without further handling. Our first approach was to seek a general solution applicable to peroxide explosives with no prior characterization. Concentrated sulfuric acid was found to effectively destroy milligram amounts of TATP; however, when scaled-up to even 1 gram, the excessive heat release caused violent rapid release of energy, perhaps detonation. 5 TATP is formed by reaction of acetone and hydrogen peroxide. Under suitable conditions the two reagents can slowly form TATP at room temperature. [6][7][8] However; the usual methods for synthesis of TATP involve controlled addition of acid. Excess acid and/or elevated temperature can favor the formation of DADP. If the heat of the reaction is substantial, it can initiate the peroxide mixture, causing detonation. Herein, we explore the region where acid can be used to affect quiescent decomposition of TATP. This work mainly focused on the destruction of 0.5 g or 3g quantities of solid TATP, but it was helpful to obtain kinetics for the destruction of TATP in solution.
Field tests were performed on 50, 100 and 460g quantities of TATP.

Synthesis of TATP and DADP
TATP and DADP were synthesized in our laboratory. 7-9 TATP was prepared by stirring hydrogen peroxide (7 g, 50 wt% in water) and acetone (5.8 grams) below a temperature of -5°C with slow addition of 0.5 mL of HCl (18% m/m). The mixture was kept at -14°C overnight (14-18 hours). Water was added to the mixture; and the precipitate filtered out and rinsed with copious amounts of water. Crude yields were typically 5 g (67.6%), melting point 88-92°C; recrystallization from hot methanol yielded a white, finely divided crystalline product, melting point 94-95°C. DADP was prepared by adding concentrated sulfuric acid (10.7 g, 96%, 105 mmol) with stirring to a cold (< 3°C) acetonitrile solution of hydrogen peroxide (3.00 g, 70%, 62 mmol). Acetone (2.9 g, 50 mmol) and acetonitrile (10 mL) were combined and chilled (~ 0 o C). The acetone mixture was added drop-wise to the hydrogen peroxide mixture while the temperature was maintained between -4 and 4°C, and the mixture was allowed to stir for 90 minutes, during which time a white precipitate formed. The precipitate was collected by vacuum filtration and rinsed with copious amounts of cold water. The crude solid (2.9 g, 76% yield) had a melting point of 131-132°C and was recrystallized from ethyl acetate.
This was followed by addition of 0.5, 1, 2, 3, 4, 5, 9 mL of acids in varying concentrations. More than 600 individual experiments were performed. All mixtures were allowed to react at room temperature uncovered for 2-24 h before extraction with 10 mL dichloromethane (TATP solubility at room temperature >1g/4mL) and rinsing with 3 mL distilled water followed by 3 mL of 1% Na 2 CO 3 . The organic layer was dried over anhydrous magnesium sulfate and analyzed via gas chromatography with mass selective detector (GC/MS). Each analytical run began with a series of five or more authentic TATP samples ranging in concentration from 10-10,000 µg/mL. These samples were used to monitor instrument responses and plot calibration curves.
An Agilent 6890 GC with Agilent 5973i MSD detector was used (i.e electron impact). The inlet was operated with a 5:1 split at 150°C. The column was an HP-5MS (30m x 0.25mm x 250µm), operated in constant flow mode with a flow rate of 1.5 mL/min and average velocity of 45 cm/sec. The transfer line for GC to MS was held at 250°C. The oven was programmed 50°C for two minutes before ramping to 200°C at 10°C/min. The MS had a solvent delay of 2 minutes and scanned from 14-500 m/z.

Kinetics for Destruction of TATP
Solutions of TATP (5 mL) were measured into 40 mL screw-top vials. Two vials were prepared; one with 5 mL of an acidic alcohol solution and the other with 5 mL of a TATP solution. These solutions were equilibrated at specified temperatures in a water bath or GC oven. After equilibration, the 5 mL acid solution was poured into the 5mL TATP solution, and the mixture was held at constant temperature for the duration of the experiment. At recorded time intervals, an aliquot of the reaction mixture was removed by syringe, placed in a separate 15 mL vial containing dichloromethane (DCM), rinsed with 2 to 3 mL of 3% NaHCO 3, followed by a rinse with distilled water, removing the aqueous layer each time. The organic layer was dried over a small amount of MgSO 4 (anhydrous) and transferred to a GC vial for quantification of remaining TATP. A parallel experiment with 5 mL of solvent (i.e. no acid) was used as a control.
For destruction of solid TATP with aqueous acid 5 mg TATP was placed into a 16 mL screw cap vial and 1 mL of acid was added. At recorded intervals the reaction was quenched by addition of ~3 mL 3 wt% sodium bicarbonate followed by 5 mL DCM. The aqueous layer was discarded; a second rinse with bicarbonate was performed; and a third with distilled water. The organic layer was dried over anhydrous magnesium sulfate and analyzed by GC/MS.
To quantify TATP, an Agilent 6890 gas chromatograph with 5973i mass selective detector (GC/MS) was used. The inlet temperature was 110°C and total flow was 24.1 mL/min (helium carrier gas). The inlet was operated in splitless mode, with a purge flow of 20 mL/min at 0.5 minutes. The column was a Varian VF-200MS (15m x 0.25mm x 250µm), operated in constant flow mode with a flow rate of of 1.5 mL/min. The oven program was initial temperature of 40°C for 2 minutes followed by a 10°C/min ramp to 70°C, a 20°C/min ramp to 220°C and a post-run at 310°C for 3 minutes. The transfer line temperature was 250°C and the mass selective detector source and quadrupole temperatures were 230°C and 150°C, respectively. Electron impact ionization at 70 eV was used.

Large-Scale Decomposition
For all large-scale experiments addition of reagents was done remotely. A pumping apparatus was erected and an electronic means of actuating the pumps via remote control was assembled. TATP (460g) was placed in a 4 L beaker with thermocouples and tubes from the output of the pumps already in place. A secondary means of adding acid was included in case of pump failure. This was accomplished by securing a Nalgene bottle with a spigot above the beaker containing the TATP.
Tygon tubing attached to the spigot was placed in the beaker. The valve could be operated remotely by mechanical means ensuring that if some acid were added and the pump failed that more acid could be added without approaching the acidified TATP.
Two thermocouples were used in this experiment. One was attached to the outside of the beaker and one submerged in the TATP. Alcohol solution (950 mL 50wt% isopropanol/water) was pumped onto the TATP first at approximately 100 mL/min using an aquarium pump. The TATP did not appear wet. The acid was then metered (120 mL/min) in the mixture using a peristaltic pump with acid resistant tubing; when the temperature rose to 70 o C the pump was stopped. A total addition of 425 mL of acid was added. Once the experiment was completed the products of the reaction were neutralized with sodium bicarbonate and put into a 4 L glass waste container. A sample of the waste was extracted in DCM followed by GC/MS analysis.

Heat Release
Heat released during the reaction of acid with dissolved TATP was measured using a Thermal Hazards Technologies micro-calorimeter. To calibrate the instrument two amber GC vials containing 1 mL reagent alcohol were placed in the sample and reference positions of the instrument. In calibration mode, the number of pulses was set to 3; the pulse size to 300 mJ; the pulse interval to 300 seconds; and the lead time to 30 seconds; samples were stirred at 200 rpm. To determine the heat of mixing between sulfuric acid and reagent alcohol, the instrument was set to collection mode with an experimental duration of 1000 seconds. A modified acid injection method was designed to accommodate the corrosive nature of strong acids. A glass capillary syringe needle was attached to a 1 mL plastic syringe. The syringe was primed to remove excess air and reduce dead volume, and the desired mass of acid was pulled into the syringe. Once a stable baseline was achieved, data collection began followed by manual injection of acid into alcohol. To determine heat released during the reaction between acid and TATP, the steps described above were followed using 1 mL of a 40 mg/mL TATP/alcohol solution in the sample position and an experimental duration of 50,000 seconds.

Decomposition Product Identification
The type and concentration of acid used to destroy TATP determined reaction progress and products formed. Experiments, in duplicate, were conducted to examine the effect of acid type. TATP (500 mg) and 1 mL of 50% water/alcohol (either ethanol or isopropanol) were combined. To this mixture was added 2 mL of one of the following: sulfuric acid (65%), hydrochloric acid (36%), nitric acid (70%), phosphoric acid (85%), methanesulfonic acid (99%), boron trifluoride (48% in diethyl ether), trifluoroacetic acid (99%), or perchloric acid (99%). WARNING: The addition of nitric acid resulted in violent fuming. Mixtures reacted for 3 hours before extraction as described above. Products were identified by comparison of mass spectra to authentic samples of TATP, DADP, and various chlorinated acetones or by spectral matching to the NIST database. Relative amounts of each material in solution are expressed as percentage of the total chromatographic signal.

Relative rates of TATP Decomposition with Acid
It was proposed that the application of mineral acid, an inexpensive and widely available liquid, applied as a spray or mist, could be a field approach to destruction of TATP. Addition of concentrated sulfuric acid (3 mL of 80% or 90%) to solid TATP (3 g) resulted in detonation. In an attempt to slow the reaction, solvents were added to the TATP (3 mL of diesel fuel, various alcohols). Addition of 98% sulfuric to the TATP moistened with a solvent resulted in violent decomposition, but not detonation.
To avoid violent reactions, experiments were designed to screen different solvent and acid combinations. TATP destruction did not occur with bases, but many acids, even  (Table 2.1) and neat, solid TATP (Table 2.2). The results in Table 2.1 are expressed as percent TATP remaining after a specified time interval. The solid TATP was first moistened with the solvent followed by addition of the acid. Note that in Table 2.1 decomposition of TATP is more complete in the same time interval, when using 36 wt% HCl than when using 65wt% H 2 SO 4 , though the molar concentrations of these acids were roughly the same. This may be explained by the higher pK a value for HCl. In Table 2.2, aqueous acid was added directly to the TATP and the amount remaining vs time was determined by quenching the reaction at specified time intervals. The first-order rate constants from this data also indicated that HCl decomposed TATP more quickly than sulfuric acid at highest concentration (i.e. 12M) Table 2.3 shows first-order rate constants for the decomposition of TATP dissolved in the solvent system indicated. Decomposition is much faster in solution than in solid so that lower concentrations of acid can decompose TATP relatively quickly (Table 2.3); again the effectiveness of HCl is noticeable.  We previously reported that in synthesis water content affected the ratio of TATP/DADP; high water favoring TATP. 8 Water also affects the rate of decomposition as well as the decomposition products. Water, entering the reaction with the acid, and in some cases with the solvent, slows the rate of TATP decomposition (Table 2.3). Solubility is part of the effect. TATP is soluble in the alcohols and acetonitrile but practically insoluble in water, yet the acid can more freely dissociate in water. The highest observed decomposition rate constant was for TATP in acetonitrile with no water, and in that solvent TATP converted to DADP.
This conversion was not observed in alcohol solutions of TATP, nor when 10% or more water was added to the acetonitrile solutions of TATP. Furthermore, use of an alcohol solvent or addition of water slowed the decomposition of TATP. Similar observations were noted when using alcohols as co-solvents in TATP formation reactions. 8 Rates of TATP decomposition were dependent on the type of alcohol.
The rate constants for TATP decomposition in alcohol are at a maximum in pure alcohol but pass through a minimum as the amount of water increases. The formation of alcohol/water complexes were shown to have a significant impact upon protonation of organic acids and bases and is attributed to preferential solubility by water or the organic solvent depending on the nature of the substance. [10][11][12]  Increased amounts of water, and reduced amounts of acid slowed decomposition to an extent that decomposition was incomplete (i.e. more than 50% TATP remaining).
When acetone, ethylene glycol, and ethyl acetate were used as the wetting agents, the acid decomposition of TATP proceeded but somewhat slower than it did with alcohol wetting agents. Interestingly, with 50wt% ethylene glycol/water wetting agent, 65% sulfuric acid did not destroy TATP in 24 hr while 36% HCl did. Although TATP was soluble in iso-octane, toluene and diesel (Table 2.4), using these as wetting agents rendered acid treatment rather ineffective (65% sulfuric destroying 20-25% and 36% HCl destroying 40% of the TATP). This is likely due to the immiscibility of the aqueous acids with these solvents.

Decomposition Products
Minor amounts of peroxo-acetone species have been previously identified in the acid destruction of TATP. 8 Depending on the reaction conditions DADP could be a significant decomposition product.  transitioned from blue to yellow as a fireball grew. A similar experiment was conducted with HMTD. The first observation was a burst of smoke or fine particulate.
In the same fashion as TATP the pile disappeared in linear progression, but before the entire pile of HMTD was gone a yellow flame was already forming above the HMTD.

Decomposition Mechanism
All the strong acids decomposed TATP, but rates (Tables 2.1, 2.2) and final products (Table 2.5) differed. If a protic solvent, e.g. water or alcohol, is used the carbocation formed when the TATP ring opens is stabilized and the intermediate will react more slowly and decompose into smaller molecules. If no water was present, the carbocation is not stabilized and the intermediate quickly cyclizes to DADP. In both cases acetone was formed. Figure 2.3 illustrates these alternate routes.

Calorimetry
During calorimetry experiments when 65 wt% H 2 SO 4 was added to 1 mL (40 mg/mL) alcohol solutions of TATP, no reaction was observed until the solution was raised to 50°C. At 50°C temperature the reaction started within minutes. The experiment was repeated using 80 wt% sulfuric acid. Within minutes heat release was visible and after about 11 hours it appeared to be complete. Duplicate experiments were run using sulfuric, hydrochloric and nitric acids taking care to deliver similar quantities of water while delivering the same number of moles of acid because previous work had shown that water affects the formation and destruction of TATP. 8 Under the same conditions hydrochloric acid resulted in a faster reaction rate but with less heat released overall than tests using nitric or sulfuric acid. Figure 2

Scaleup
When the 500 mg tests were scaled to 3 g TATP wetted with 3 mL alcohol, the acid (3mL) was added remotely. With concentrated HCl (36%), the decomposition went quiescently; when it was 90% sulfuric acid, the reaction was violent (see Figure 2.6).
Outdoor field tests were conducted on 100g and 460 g quantities of TATP. Quantities of acid and solvent are detailed in Table 2.7. Hydrochloric acid was chosen due to its ability to decompose TATP quickly without the formation of DADP and its reduced heat of reaction with TATP. Aqueous ethanol and isopropanol (50/50 with water) were tested on the 100 g scale with similar results (Table 2.7). Aqueous isopropanol was chosen for the 460g experiment over the lower molecular weight alcohols due to its higher boiling point (82.5 o C); butanol was not considered due to its reactivity with acid. Data points for these experiments are also labeled on the ternary diagrams in

Introduction
Mercury removal from heavy naptha feedstocks can be accomplished in a number of ways. Using a non-regenerable metal sulfide adsorbent has several benefits but one potential catastrophic drawback. Copper sulfide effectively scavenges mercury; but since acetylene is a possible contaminant under these circumstances, the potential formation of explosive copper acetylide is a cause for concern.
Copper acetylide is a general term used somewhat ambiguously for compounds containing copper and acetylene moieties. If the alkyne is acetylene (C 2 H 2 ), both hydrogen atoms can be substituted with a cuprous ion (Cu + ), yielding a molecule with the formula Cu 2 C 2 . If acetylene reacts with the cupric ion (Cu 2+ ), CuC 2 may be formed. To eliminate ambiguity Cu 2 C 2 will be referred to as cuprous acetylide and CuC 2 will be referred to as cupric acetylide for the purposes of this research. acetylide, a primary explosive, has sometimes been cited as the cause of accidental explosions. 2 Although it has been assigned a CAS number, only qualitative data is available. Cupric acetylide, or copper carbide, is brown to black in color, sometimes forming lustrous plates and is extremely sensitive towards initiation. Its structure is envisioned as the copper ion coordinated to the pi system of acetylene. The lack of physical data makes this assignment speculative, and most reports are only qualitative. 3 A fundamental understanding of "copper acetylides" is necessary for evaluation of the hazards of using copper materials for mercury removal. Herein we determine whether copper adsorbents on an alumina substrate--copper oxide, copper carbonate, and copper sulfide-form hazardous acetylides when exposed to acetylene.
Various analytical techniques were employed in an attempt to make that determination.

Phenylacetylene experiments:
A 5:1 molar ratio of phenylacetylene to copper material mixture was stirred at room temperature for 16 hours followed by extraction into methylene chloride and analysis by GC/MS.

Elemental Analysis/Isotopic Ratio Mass Spectrometry: A Thermo Flash 2000
organic elemental analyzer coupled to a Thermo Delta V Advantage isotopic ratio mass spectrometer was used to determine the carbon content of samples before and after exposure to acetylene. Urea was used as an external standard for carbon quantification.

Thermal analysis:
A TA Instruments Q100 differential scanning calorimeter was used to collect DSC data. Approximately 2.00 mg of material was placed into a hermetically sealed aluminum DSC pan. The sample and empty reference sample were heated at a rate of 20 o C/minute from 30 o C to 450 o C. Flowing nitrogen (50 mL/min) purged the DSC cell during the experiments.
A TA Instruments Q5000 thermogravimetric analyzer was used for the TGA analysis. In a high temperature platinum pan the sample was heated under an inert nitrogen atmosphere (25 mL/min) from room temperature to 900 o C at a rate of 20 o C/min.

Infrared Spectroscopy:
A Thermo Nicolet 6700 fourier transform infrared spectrometer was use to collect 64 scans at a resolution of 4 cm -1 from 375-4000 cm -1 .
Approximately 2 wt% of sample was mixed for 2 minutes with KBr in a wiggle bug followed by pressing under 15000 psi for approximately 10 minutes into a pellet.

Raman Spectroscopy:
A Bruker Senterra Raman microscope was used to collect Raman data. The laser source was 785 nm set at 100 mW. With a resolution of 9-15 cm -1 spectra were collected from 148-3818 cm -1 using an integration time of 3 seconds and 2 coadditions.

Gas Chromatography/Mass Spectrometry:
An Agilent GC6980 gas chromatograph coupled to an Agilent 5973 mass selective detector was used for analysis of the phenylacetylene experiments. The inlet was operated at 225 o C in split mode with a 2:1 split. Helium was used as the carrier gas in constant flow mode using a Varian VF200MS column. The oven program started at 50 o C followed by a 20 o C/min ramp to 300 o C.

X-Ray Photoelectron Spectroscopy: A Physical Electronics 5500 Multi-
Technique Surface Analyzer was used to collect ESCA (electron spectroscopy for chemical analysis) surveys as well as multiplex analysis of specific elements. An aluminum source was used with a resolution of 0.125 eV and a spot size of 600 microns. Peaks were calibrated to a C1S peak of 284.5 eV.

Elemental Analysis:
Three adsorbents were exposed to acetylene: copper oxide (GB238); copper carbonate (GB220); and copper sulfide (GB562S from two different batches; batch B was derived from copper carbonate). Both moist and dry acetylene were used. Table 3.1 shows the elemental analyses of these samples. All samples contained more carbon after exposure than the initial samples; the increase in carbon content depended on the copper compound used. Copper carbonate showed the highest carbon content post-exposure, followed by copper oxide, and then copper sulfide batch B. The response of copper carbonate to acetylene exposure was unique in that it changed color from teal green to brown upon exposure to dry or moist acetylene for 1 week at 40 o C and a flame test of this brown materials resulted in ignition. In contrast, the acetylene-exposed copper oxide and copper sulfide did not change color nor ignite in a flame.          (Figs. 3.14-3.19). The TGA traces of the cooper carbonate before exposure to acetylene showed a mass loss of 29% (peak at 282 o C) which is a little low for complete loss of carbonate (36%). (Fig. 3.16). 4 Cooper carbonate exposed to either moist or dry acetylene showed a TGA loss of 17-20% (peak 205 o C) (Figs. 3.18 (Table 3.1). Likewise copper oxide exposed to moist acetylene exhibited a weight loss of 19% (Fig. 3.15), an increase of 7% by weight, consistent with elemental analysis.
The low weight loss from the copper oxide exposed to dry acetylene was unique ( Fig.   3.16); its Raman and IR spectra were unique, as well.

Raman & Infrared Spectroscopies:
The copper oxide and copper carbonate exposed to acetylene showed a new peak in the Raman spectra around 1370 cm -1 (Figs. 3.23-3.28). There was little to distinguish these except the intensity of the peak near 1370 cm -1 and the appearance of multiple peaks in this region for the copper oxide sample exposed to dry acetylene (Fig. 3.25). Reported literature values suggest the 1370 cm -1 peak is due to the presence of a highly disordered phase of carbon. [5][6][7] Acetylide or carbyne carbons which are sp hybridized should be have a peak around 2100cm -1 similar to that seen in the Raman spectra of phenylacetylene (figure 3.35); 8-10 however, no such peak was observed in any of the exposed samples. The acetyleneexposed copper oxide and carbonate samples also exhibited a small peak around 1860 cm -1 (Figs. 3.24, 3.27 & 3.28). This may be assigned to cumulenic carbon associated with a highly oxidized form of copper at the terminal ends of the carbon chain. 8 Cupric oxide had sharp peaks at 285 and 600 cm -1 which broaden out to reduced intensity as the cupric oxide was reduced to the cuprous oxide. 11,12 Peaks in these regions were observed in the as-received copper oxide and copper carbonate samples   Raman:**GB238*exposed*to*moist*acetylene*for*1* week*at*40C* Fig. 3.25 Raman GB238 exposed to a flow of dry acetylene; 40 o C for 1 week.
Raman GB220 (copper carbonate) Fig. 3.26 Raman GB220 as received Fig. 3.27 Raman GB220 exposed to a flow of acetylene bubbled through water; 40 o C for 1 week.  Fig. 3.34 Raman GB562S (from original starting material) exposed to a flow of dry acetylene; 40 o C for 1 week. Infrared (IR) spectroscopy indicates that all samples, except for the copper oxide exposed to dry acetylene, remained relatively unchanged after exposure to moist and dry acetylene (Figs. 3.36-3.47). As in Raman, the IR spectrum of the cooper oxide exposed to dry acetylene produced a unique spectrum with multiple peaks in the 1700 to 1100 cm -1 region (Fig. 3.38). In no case did the acetylene exposed samples exhibited peaks in the 2200 to 2100 cm -

X-ray Photoelectron Spectroscopy (XPS):
After an initial survey of 0 to 1000 eV binding energies, the Cu 2p (970-920 eV) and the C1s regions (280-300eV) were examined in detail. Copper peaks were observed at 935 and 955 eV with CuO satellite peaks at 943 and 963 eV in the as-received cooper oxide and carbonate and in the acetylene exposed samples (Figs. 3.48-3.60). Readily observed in the acetyleneexposed copper carbonate sample is a decrease in the intensity of the CuO satellites with a concomitant increase the Cu peaks, i.e. reduction is observed by XPS as well as Raman. (Fig. 3.59). This change is not as dramatic in the copper oxide sample although the reduced intensity of the satellite peaks suggests that reduction may have occurred (Fig. 3.53). Decomposition of the carbonate in the exposed copper carbonate samples is also observed by XPS (Fig. 3.60). In the as received copper carbonate sample a peak is observed near 290 eV that is absent in the exposed samples.
In the carbon region, the C1s peaks of the acetylene-exposed samples were asymmetric, centered at 284.

Discussion
The typical method for synthesizing cuprous acetylide is to dissolve a cuprous salt such as cuprous iodide in aqueous ammonia in the presence of a reducing agent such as hydroxylamine. 19 Upon bubbling acetylene into the solution a bright red precipitate forms. This precipitate is sensitive to oxygen and begins to undergo Glaser oxidative polymerization forming copper polyynides. 2   The absence of a peak near 2100 cm -1 in any of these samples, including the red powder produced by the "standard" preparation method, suggests an acetylide was not formed or, if formed, and rapidly underwent further reaction. There is precedence for substituted acetylene molecules coupling in the presence of copper in various oxidation states without solvent. [17][18][19][20] To probe the reactivity of the copper compounds with other acetylene bearing molecules, each was stirred with phenylacetylene for ~16 hours, followed by GC/MS analysis ( Fig. 3.71). Identification of 1,4-diphenylbutadiyne indicates the ability of the copper compounds to couple acetylene moieties without any further assistance from solvents or added base. With copper sulfide much less diphenylbutadiyne was observed than when copper oxide or copper carbonate were used. This result is in keeping with the lack of reaction between the sulfide material and acetylene.

Conclusions
Under the experimental conditions used, acetylene exposure of copper oxide or copper carbonate formed energetic materials, but these resulting materials do not appear to be cuprous acetylide (Fig. 3.1a). Indeed previous reports of cuprous acetylide synthesis do not isolate it. 3,19 If formed, cuprous acetylide likely underwent further polymerization and crosslinking resulting in a very disordered form of carbon, as observed by Raman spectroscopy. The absence of copper acetylide as a product in the exposed samples reduces the risks associated with exposing these compounds to acetylene. Nevertheless, with copper oxide and carbonate there is evidence that the end products are energetic as evidenced by the exotherm in their DSC traces and the energetic behavior of the post-exposure copper carbonate sample when exposed to flame. This is likely due to the exothermic reaction between copper oxide and carbon at elevated temperatures. 20

Introduction
Fuel cells are attractive because they provide an efficient, safe and renewable alternative to traditional petroleum fuels. A number of technologies, such as polymer electrolyte fuel cells (PEFCs) and direct methanol fuel cells (DMFCs) are promising, but their commercialization has been restricted due to various problems such as electrode poisoning. 1 Direct borohydride fuel cells (DBFCs) have been developed and studied in order to overcome some of these obstacles, but they suffer from the potential of catastrophic failure of the fuel cell due to chemical incompatibilities between hydrogen peroxide and sodium borohydride. To examine the potentially hazardous reactivity between hydrogen peroxide and sodium borohydride, we studied the kinetics and heat release of the reaction by gas evolution and reaction calorimetry.

Reagents and Chemicals
Sodium hydroxide pellets of 99% purity were obtained from Sigma Aldrich.
Sodium borohydride (98% purity) was obtained from Alfa Aesar. Hydrogen peroxide (30 wt% and 50 wt%) were obtained from Univar. Concentrated solutions of hydrogen peroxide were obtained by distillation. Concentrations were confirmed by refractive index measurements.

Micro reaction calorimetry
Isothermal calorimetry experiments were conducted using a Thermal Hazards Technology (THT) micro-reaction calorimeter. The instrument was calibrated at 25 o C by supplying 300 mJ pulses after an initial period of 30 seconds. Data was collected for a total of 500 seconds, and the experiment was repeated 3 times. For the isothermal experiments, data was collected at a rate of 1 data point/second in titration mode, using 1mL solvent as reference. One mL of the borohydride solution was equilibrated in the sample cell. Hydrogen peroxide (HP) was measured into a 100 µL THT syringe, its tip was placed into the equilibrated borohydride solution, the syringe secured, to the instrument, and the entire setup allowed to equilibrate. The experimental duration, duration of baseline data collection, dosage, dosage rate, and other necessary parameters were set, before the experiment, initiated.

Gas Evolution
A gas burette was made by inverting a 50 mL burette and securing a ¼" ID

Product Identification
Gaseous products were analyzed using an Agilent 7890 gas chromatograph coupled to a thermal conductivity detector (TCD). Argon was used as a carrier gas and kept at a constant flow of 5 mL/min. A 5 µL injection split 5:1 was injected into a 50 o C inlet. A Varian Molsieve 5A column was used, and the oven was kept isothermal for 3 minutes. Gases were identified by retention times compared to injections of known gas samples.

Results & Discussion
Spot tests were conducted to probe the reactivity between hydrogen peroxide and sodium borohydride as a solid and in solution. Low concentrations of reactants (<1 wt% sodium borohydride and < 30 wt% hydrogen peroxide) were deemed safe so that larger scale studies could be conducted. Observations are recorded in Table 4.1. hydroxide the addition of HP resulted in an even more highly exothermic reaction. To remove the potential additive effects to the overall heat release (HP decomposition in NaOH and HP reaction with borohydride), sodium hydroxide was not used; rather sodium borohydride was dissolved in isopropanol With 30wt% and 70 wt% hydrogen peroxide, the heat released per mole of sodium borohydride in isopropanol was similar to that obtained by adding 30wt% hydrogen peroxide to an aqueous 3wt% sodium hydroxide solution containing sodium borohydride. However, when 70wt% HP was added to a sodium borohydride solution in aqueous 3wt% sodium hydroxide the heat released was much greater than when 30wt% HP was used. This is most likely due to the decomposition of hydrogen peroxide by strong base being dependent upon the ratio of hydrogen peroxide to hydroxide ion. 3 In comparison to hydrolysis of borohydride by water, hydrolysis by hydrogen peroxide is considerably more exothermic, 250 kJ/mole 4 versus more than 900 kJ/mole (Table 4.2).
To analyze the progress of the reaction, we attempted both 1 H and 11 B NMR experiments; but the rate of the reaction and the formation of gas bubbles made this very difficult. NMR was useful in identifying the end product as boric acid.
Monitoring the reaction by gas evolution was successful. The difficulty with that approach was the fact more than one reaction was taking place. Sodium borohydride reacts with water (eq. 4.1); but, the presence of base greatly retards this reaction. 5 Unfortunately, while alkaline conditions stabilize borohydride in water, they catalyze the decomposition of hydrogen peroxide (eq. 4. Hydrolysis of sodium borohydride by water is reasonably slow at room temperature.
Hydrolysis of borohydride by hydrogen peroxide was much faster than by water (Tables 4.3

and 4.4). It is interesting to note that hydrolysis of borohydride in water
resulted in 4 molar equivalents of hydrogen gas; 5 whereas hydrolysis by hydrogen peroxide only resulted in 1 molar equivalent of hydrogen gas (Table 4.3). This was confirmed using gas chromatography coupled to a thermal conductivity detector (GC-TCD) (Fig. 4.1).

in 3 wt% NaOH by hydrogen peroxide
The use of strong base to stabilize sodium borohydride hydrolysis in water also effectively stabilized the sodium borohydride against hydrolysis by hydrogen peroxide (cf Table 4.3 to Table 4.4). However, the retarding of borohydride hydrolysis was not sufficient to prevent the reaction from being potentially catastrophic, and the decomposition of the HP, promoted by the base, resulted in oxygen formation (eq. 4.2). At room temperature 1 molar equivalent of gas was evolved from the reactions shown in Table 4.4, but at elevated temperatures, e.g. 40 o C (Table 4.4), more gas was evolved. Analysis of the gaseous products showed that instead of just hydrogen gas evolution there was also oxygen gas evolution, due to hydrogen peroxide decomposition in the presence of strong base (Fig. 4.1).
The difference in the amount of gas evolved when comparing hydrolysis by water or hydrogen peroxide may be rationalized by the hydroboration/oxidation reaction, which utilizes strong base and hydrogen peroxide 7 (eq. 4.3, Fig. 4.2).
Following this type of mechanism the hydride could react with the acidic proton of the hydrogen peroxide evolving hydrogen gas followed by the nucleophilic anion of

Mitigating Strategies
The reaction between sodium borohydride and HP is extremely violent, especially when both are at high concentrations. It is difficult to chemically alter the formulation in a manner that would lessen the severity of the reaction but not prevent it from being harnessed, as intended. When a drop of 70wt% HP was dripped onto a NaOH solution of 25wt% sodium borohydride sitting in a spot plate, there was a flash The reaction was sufficiently slow that a red glow could be observed at the boundary layer between the HP and borohydride. The glowing red boundary layer progressed across the drops as diffusion between them occurred. These observations suggest that efforts to increase viscosity may mitigate catastrophic mixing if engineering cannot absolutely guarantee no leakage between cells.

Conclusions
Sodium borohydride hydrolysis by water is slow compared to hydrolysis by hydrogen peroxide. The reaction is also more exothermic when hydrogen peroxide is used. Although strong base can stabilize borohydride hydrolysis in both cases, it cannot prevent the potential catastrophic failure of a DBFC if it were to be physically compromised and the components, mixed. In a DBFC the concentration of the chemicals is high, and qualitative tests have shown they react in a hypergolic fashion.
Limiting diffusion between the solutions may be the solution to mitigating the violence of this reaction. How addition of contaminants, like sodium alginate, would effect the performance of these cells is unknown, but perhaps thickening agents could be selected that would not have any adverse effects on the cell performance.

Identification of intermediates and relative rates of formation
The following GC-MS data was collected using anhydrous ammonia positive chemical ionization mass spectrometry. All peaks were analyzed in select ion monitoring mode unless otherwise specified.

TATP decomposition in acetonitrile
All of the decomposition kinetics experiments were conducted by preparing 1 vial with 5 mL 20 mg/mL TATP in acetonitrile. A second vial was prepared with the appropriate amount of acetonitrile, water and acid totaling 5mL. Both vials were equilibrated to the experimental temperature and then mixed. Aliquots of the reaction mixture were taken, extracted in methylene chloride, rinsed once with sodium bicarbonate solution, once with water and dried over anhydrous magnesium sulfate.
The extract was analyzed by GC-MS using electron impact ionization. External standards of TATP and DADP were used for quantification.

TATP decomposition with aqueous acid
All of the decomposition kinetics experiments were conducted by weighing 5mg solid TATP into a 14mL vial. To the vial 1 mL of aqueous acid was added with a volumetric pipet. At recorded intervals, the decomposition reaction was quenched by adding approximately 3mL sodium bicarbonate solution and 5mL methylene chloride.

Sodium borohydride added to hydrogen peroxide (30mg in 20mL 3wt% HP)
In a 50mL 3-neck round bottom flask 30 mg of sodium borohydride was added to 20 mL of 3wt% HP. At recorded intervals the volume of gas produced was recorded. Table D8: 20mL 3wt% HP, 30mg NaBH 4 , 0 o C Figure: ln % total gas evolved vs time for reaction of 30mg NaBH4 in 20mL 3wt% hydrogen peroxide at 0 o C