Explosive Thermal Decomposition Mechanism of RDX - American

Apr 15, 1994 - Thin films of RDX (hexahydro- 1,3,5-trinitro- 1,3,5-triazine) have been subjected to transient pyrolysis using a pulsed C 0 2 laser in ...
2 downloads 0 Views 448KB Size
J . Phys. Chem. 1994,98, 5441-5444

5441

Explosive Thermal Decomposition Mechanism of RDX Tod R. Botcher and Charles A. Wight',? Department of Chemistry, University of Utah, Salt Lake City, Utah 841 12 Received: January 25, 1994; In Final Form: March 10, 1994'

Thin films of RDX (hexahydro- 1,3,5-trinitro- 1,3,5-triazine) have been subjected to transient pyrolysis using a pulsed C 0 2 laser in order to determine details of the thermal decomposition mechanism under conditions that simulate a thermal explosion. The first step, scission of an N-N bond, leads to formation of N2O4. The product is trapped in the solid film by rapid quenching to 77 K following the pyrolysis pulse and subsequently detected by transmission FTIR spectroscopy of the film. Product yield measurements show that 1.9 f 0.2 RDX molecules are destroyed for every N2O4 molecule detected in the films. Crossover experiments conducted on isotopically labeled samples containing both unlabeled and fully labeled RDX-15Na show that the N204 product consists of a statistical mixture of 14,14N204,14,*SN204, and 15+15N204 isotopomers. These results show that both halves of the dimer arise from separate RDX parent molecules and that explosive decomposition of RDX involves loss of only a single NO2 molecule.

Introduction Recent studies of thermal decomposition mechanisms in explosives and propellants have been designed to determine the detailed chemical pathways by which these large organic molecules are transformed into N2, C02, H20, and other small reaction products. This is a particularly challenging problem because the rapid release of heat during the decomposition makes it difficult to isolate and identify individual steps in the mechanism. The main driving force for these studies stems from a desire to understand the sensitivityand performancecharacteristicsof these materials in terms of their chemical mechanisms. It is hoped that the development of global chemical models will aid in the design of new energetic materials that are safer to manufacture, store, and handle. One of the most thoroughly studied systems is hexahydro1,3,5-trinitro-l,3,54riazine,which is commonly called RDX.l-15

RDX

Even so, there has been considerabledifficulty in identifying even the first step in the thermal decomposition mechanism of this molecule. Early studies suggested that reaction begins with N-N bond cleavage because this dissociation energy is only about 205 kJ/mol.'6 However, a molecular beam experiment performed by Y.T. Lee and co-workersI7provided convincing evidence that a major decomposition route of laser-heated gas-phase RDX is concerted depolymerization to form three molecules of methylenenitramine, CHzNNOz. Mass spectrometric studies have also indicated the possibility of a C-N bond scission mechanism for gas-phase RDX.I8 More recently, detailed thermal decompositionstudies of RDX have been performed by Behrens and Bulusu using simultaneous thermogravimetric modulated beam mass spectrometry (STMBMS).19-22 These authors have identified four major decomposition pathways when the solid is heated slowly, none of t Alfred. P. Sloan Research Fellow, 1990-94. e Abstract published in Advance ACS Absrracrs, April 15, 1994.

which involve formation of CH2NN02. Taken together, these studies clearly demonstrate that the decomposition mechanism depends on the physical state of the material (gas phase or condensed phase) and on the heating rate. We have recently developed a new transient IR laser pyrolysis technique for examining decomposition mechanisms under conditions that mimic condensed-phase thermal explosions. Our initial results showed that under these conditions, the first step in decomposition of RDX is unimolecular scission of an N-N bond.3*z3.z4 The principal evidence for this was detection of N204 (the dimer of N02) at the threshold laser fluence for decomposition. It struck us as curious that the initial reaction product was detected in its dimerized form, and it suggested to us that both halves of the dimer might arise from the same RDX molecule. This could happen if breaking the first N-N bond somehow lowers the barrier to breaking the second. However, preliminary measurements of the decomposition yields showed that approximately two RDX molecules are destroyed for every N2O4 product detected, suggesting that the decomposition mechanism involves loss of only a single NO2 m ~ l e c u l e . ~ We undertook the present study in order to determine definitively whether N2O4 is formed from a single RDX molecule or by dimerization of NO2 products from separate RDX parent molecules. Making this distinction is an important step toward identifying subsequent steps in the decomposition mechanism. The strategy we have adopted is to pyrolyze samples containing mixtures of normal and fully '5N-labeled RDX isotopomers. In this type of "crossovern experiment the mechanism of dimer formation can be determined by examining the relative abundances of 14,14N204,14,1sN204, and 15.15Nz04.If both NO2 molecules in a given dimer arise from the same molecule of RDX, only 14*14N204 and l5JSNz04 should be observed. However, if each RDX molecule loses only a single NOz, then a statistical distribution of 14N204:15,l4N~O4:'5N2O4 isotopomers should be observed.

Experimental Section The experimental procedure has been described previously,3.23.24 and only a brief outline is presented here. A thin film (- 13 r m ) of explosive is formed on an infrared window by vapor deposition from a Knudsen oven. Following deposition, the sample is annealed to about 35 OC to form a polycrystalline film. The vapor deposition technique ensures that the isotopomers are randomly distributed within the sample. One side of the film is left exposed to a vacuum so that any portion of the sample that

0022-3654/94/2098-5441~04.50/00 1994 American Chemical Society

5442

The Journal of Physical Chemistry, Vol. 98, No. 21, 1994

is vaporized by the pyrolysis laser will be lost. In this way, the experiment issensitive only to reactions that occur in thecondensed phase. The crystalline film is cooled to 77 K and then pyrolyzed from the vacuum side using a single shot from a C02 laser (Pulse Systems Model LP140-G) operating at 944 cm-1 and 1.8 J/cm2/ pulse. This heats the sample to approximately 1200 K in 35 ps (the nominal pulse width of the laser); the sample then cools below 400 K in 3 ms by conduction of the heat into the 77 K substrate. These temperatures are estimated using a computer simulation in which the one-dimensional heat diffusion equation is solved during the laser heating and conductive cooling cycles, as reported previously.3 The strategy is to heat the RDX to extremely high temperatures for brief periods, followed by trapping of the initial decomposition products within the sample at 77 K for subsequent spectroscopic detection. Some samples were pyrolyzed from the substrate side by passing the laser beam through the CsI window. A somewhat higher laser fluence of 2.2 J/cm2/pulse is required to reach a peak temperature of 1200 K in this geometry because the hottest portion of the sample is closest to the substrate, where cooling is most efficient, but the results are otherwise unaffected. A separate series of experiments was conducted to determine the yield of N2O4 product in the laser pyrolysis. These experiments were carried out by depositing a thin film of RDX on the window at 77 K, warming to room temperature, and covering the film with a second CsI window. This “sandwich” arrangement was remounted in the vacuum dewar vessel and recooled to 77 K before laser pyrolysis. The purpose of covering the sample is to ensure that no reactants or products are lost to vaporization by the intense laser pulse. Transmission FTIR spectra were obtained with a Mattson Model Polaris FTIR spectrometer before and after laser pyrolysis to monitor product formation. Spectra were obtained by averaging 64 scans at 0.5-cm-1 resolution. A sample of fully labeled RDX-ISN6 was supplied by Dr. S. Bulusu (U.S.Army Research Laboratory, Picatinny Arsenal) and had an isotopic purity of no less than 99% as determined from its 70-eV mass spectrum. Mixed samples of RDX-I4N6/ RDX-ISN6were prepared by loading the Knudsen oven with 6.71 mg of RDX-I4N6 and 6.70 mg of RDX-ISN6. Thin film samples of N204 were also prepared by vapor deposition onto a 77 K infrared window for comparison with spectra of RDX pyrolysis reaction products. Nitrogen dioxide was introduced into a known volume of a glass manifold and allowed to reach equilibrium between the monomer and dimer (about 15min). Theq9.3 X 1Wmolofthemixturewasdeposited onto a CaF2 substrate at 77 K at a rate of 3.72 X IO-* mol/s, and an FTIR spectrum was obtained. Isotopically labeled ISNO2 was obtained from Cambridge Isotope Laboratories and had an isotopic purity of 99%. Unlabeled NO2 was obtained from Matheson and was purified by vacuum distillation prior to use.

Results When an unlabeled sample of RDX is pyrolyzed with a single C02 laser pulse at 1.8 J/cm2, the initial reaction product is Nz04. The infrared absorption spectrum of this product is shown in Figure l a as a triplet of bands centered at 1736 cm-I. A similar spectrum is also shown in Figure 1b for the 15J5N204reaction product formed from pyrolysis of a thin film of fully labeled RDX-15N6; this band is centered at 1705 cm-I. It should be noted that under these experimental conditions, approximately 35% of the sample is vaporized by the laser pulse. A mixture of labeled and unlabeled RDX was prepared by codeposition of a mixture of isotopomers. The relative concentrations of isotopomers in the sample was not 1:1, probably because the two crystalline samples have somewhat different grain sizes and the opening in the Knudsen oven is too large for the full

Botcher and Wight

unlabeled

Ir‘ c Alabeled

I

J

k I

~

I i

1750 1700 Wavenumbers (cm-’) Figure 1. Infrared bands of N2O4 produced following transient infrared laser pyrolysis of unlabeled RDX (trace a) and RDX-IS& (trace b).

C I

1750 1700 Wavenumbers (cm-’) Figure 2. Infrared spectrum showing N204 absorption bands arising from laser pyrolysis of a 2:l mixture of unlabeled and labeled RDX isotopomers (trace a). For comparison, spectra are also shown for neat Nz04 films containing a 2:l ratio of I4N and I5N isotopes. In trace b, the isotopes are scrambled by premixing a 2:l mixture of unlabeled and labeled NO;! prior to vapor deposition. Trace c was generated by first obtaining a spectrum of a layered sample containing equal amounts of The intensity of the labeled layer was reduced 14*14N;!04 and 1S*1SN204. by half using spectral subtraction in order to produce a spectrum of N2O4 having a 2:l ratio of unscrambled isotopomers.

equilibrium vapor pressures to be established. The isotopic composition of this sample was determined to be 2.0: 1 (unlabeled: labeled) from an FTIR spectrum obtained prior to pyrolysis by comparing the integrated intensities of two bands at 1132 cm-1 (RDX) and 1118 cm-1 (RDx-lSN6). This calculation includes a correction factor of 1.26 to account for the larger characteristic integrated cross section of the 1 132-cm-1 band compared with its unlabeled counterpart at 1118 cm-l. The comparison of the integrated cross sections was made using spectra of the isotopically pure samples and assumes that the integrated intensity of the C-H stretching vibrations is not dependent on the mass of the nitrogen isotopes. Absorbance at the pyrolysis laser frequency (944 cm-l) for the unlabeled and fully labeled isotopomers is 1.3 X 10-19 and 8.2 X 10-20 cm2, respectively. Pyrolysis of the isotopically mixed sample yields the FTIR spectrum shown in Figure 2a. Note that the mixture of NzO4 isotopomers produces a complicated spectrum spanning the region from about 1660 to 1800 cm-1. The triplet structure of the N204 absorption band for each isotopomer makes it difficult todetermine the distribution of isotopomers formed as RDX reaction products simply by examining the peak intensities. Therefore, two control experiments were performed to generate comparison spectra. In

Explosive Thermal Decomposition Mechanism of RDX

The Journal of Physical Chemistry, Vol. 98, No. 21, 1994 5443

the first, a 2: 1 mixture of 14NOz/15N02was prepared, equilibrated, and deposited as a thin film at 77 K. A spectrum of this sample, which contains a statistical (4:4:1) distribution of 14,14N204: l4J~N2O4:15J5N2O4 isotopomers, is shown in Figure 2b. In the second control experiment a layered sample was prepared by first depositing a sample of pure 'SNOz, followed by an equal amount of 14NOz. Because the isotopomers are never mixed, the spectrum of this sample contains no contribution from the mixed 14JsN204 isotopomer. The spectrum shown in Figure 2c was generated by subtracting the pure ISNO2 spectrum (scaled by a factor 0.5) from the spectrum of the layered sample. In this way, it represents the spectrum of a 2:l mixture of 14J4N~04/1sJsN204. The decomposition yield was determined in a set of experiments in which the RDX film was covered with a second CsI window to prevent laser vaporization of the reactants and products. The fraction of RDX molecules destroyed was determined by comparing the integrated IR band intensities before and after photolysis in the regions 700-1150 cm-l and 1200-1300 cm-I. This was converted to the number of RDX molecules decomposed per unit area in the film using the known density of RDX and the film thickness (as determined by laser interferometry during deposition). The corresponding number of N2O4 product molecules per unit area was determined by comparing the integrated intensity of the triplet IR band centered a t 1736 cm-l with spectra of thin films containing measured amounts of NzO4. This comparison shows that an average of 1.9 f 0.2 RDX molecules are destroyed for every N2O4 molecule detected in the pyrolyzed film.

only monomers are present in the vapor phase.27 Therefore, we can state with confidence that the two halves of the NO2 dimer arise from separate RDX molecules, but we cannot make any conclusion about whether the RDX parent molecules belong to the same dimer pair in the crystal. The organic radical produced in reaction 1 may decompose by subsequent reactions that break up the ring, e.g.,

A possible complication arises from the fact that crystalline RDX is composed of interlocked dimers.2s.26 The influence of thisdimer structureon thedecomposition mechanismisnot known. In our experiment, preparation of the samples by vapor deposition ensures that the dimers would be isotopically scrambled because

Acknowledgment. We thank Drs. Robert Fifer and Suryanarayana Bulusu of the US. Army Research Laboratory for generously supplying the unlabeled and labeled samples of RDX, respectively. This research was supported by the U S . Army Research Office under Contract No. DAAL03-90-G-0043. The

H2 6

-

N,O c H 2 0

+

HCN

+

NO + CHzO

(3)

Of course, we have not yet obtained direct evidence to elucidate the ring fragmentation reactions. However, this scheme is similar to the one originally proposed by Melius, which is based on the idea that N-N bond scission promotes unraveling of the ring by significantly reducing the amount of energy required to cleave three C-N bonds.16 The mechanism is consistent with our earlier observation of N20, NO, and HCN reaction products in highfluence laser pyrolysis experiment^.^ Also, Hz0 and CHzO are also known reaction products from other ~ t u d i e s .Such ~ a scheme provides a plausible explanation for why the second and third NO2 groups can be converted to N O and NzO, rather than being released as NO2. It should be stressed that there is no conclusive evidence for a depolymerization mechanism leading to formation of CHz"02 in the decomposition of condensed-phaseRDX under any Discussion experimental conditions. Comparison of the spectra in Figure 2 shows that the pyrolysis In the scheme outlined above, most of the reactions are product formed from the isotopically mixed RDX sample most unimolecular, and the products retain much of the bond likely consists of a statistical 4:4: 1 mixture of 14~14Nz04:14~1sN204:connectivity that is present in the parent RDX molecules. The lSJSN2O4isotopomers. Note that in the region near 1700 cm-I, exceptions are bimolecular reactions that form HzO and CH20. spectra a and b exhibit maxima at 1705 and 1690 cm-I, with It makes intuitive sense that unimolecular reactions should matching minima a t 1720 and 1695 cm-l. There are no dominate the mechanism of thermal decomposition under rapid corresponding features in the spectrum of layered sample c. Also, heating (Le., under conditions found in thermal explosions). If spectra a and b have significant band intensity extending to the we take thesimplisticview that therateofeachstepin thereaction red edge near 1670 cm-1, whereas this feature is considerably less can be characterized by an activation energy and a preexponential intense in layered sample c. The match between a and b is not factor, then we expect that both quantities will tend to be lower perfect, particularly in the region near 1735 cm-I, where there for bimolecular steps than for unimolecular. The lower activation are some differences in band shape. These differences could be energies for bimolecular steps can be thought of as arising from caused by matrix shifts, i.e., the fact that spectrum a is N2O4 in a combination of bond-making and bond-breaking processes RDX whereas b is neat solid N204. A slight blue shift of the (compared with unimolecular bond scission). The lower preexcenter peak of the NzO4 triplet was observed previously for ponential factors of bimolecular reactions arise from steric unlabeled NzO4 in RDX compared with the neat sample,3 and requirements of bringing two reactants together in a specific this could account for the differences between spectra a and b. relative orientation that is favorable for reaction to occur. Note that such a matrix shift cannot be used to improve the Although this comparison of bimolecular and unimolecular agreement between spectra a and c because the change in reaction kinetics relies on broad generalizations, it provides a frequency is in the opposite direction. qualitative basis for understanding the differences between our The observation of isotopically scrambled N2O4 products clearly results (where samples are heated at -3 X lo7 K/s) and those shows that these dimers are formed by joining NO2 radicals from of Behrens and Bulusu (where heating is at -8 X 10-3 K/s).21,2* separate RDX parent molecules. In addition, the decomposition When samples are subjected to rapid heating, reactions will tend yield measurements show that, on average, 1.9 f 0.2 RDX to occur at higher temperatures, and unimolecular reactions will molecules disappear for every Nz04 molecule produced. Both of tend to dominate the mechanism. Slow heating experiments, on these observations provide firm support for the conclusion that the other hand, will tend to have larger contributions from each RDX molecule reacts by loss of a single NO2 radical, Le. bimolecular steps in the overall mechanism (for which direct evidence has been obtained).21-22 The former are useful for investigating mechanisms apropos to combustion and explosion, whereas the latter have more application for hazard assessment, degradation of material by unintentional heating, and disposal by incineration.

5444

The Journal of Physical Chemistry, Vol. 98, No. 21, 1994

Chemistry Department Mass Spectrometry Facility is funded in part by grants from the National Science Foundation (No. CHE9002690) and from the University of Utah Institutional Funds Committee.

References and Notes (1) Adams, G. F.; Shaw, R. W., Jr. Annu. Rev. Phys. Chem. 1992,43, 311. (2) Alexander, M. H.;Dagdigian, P. J.; Jacox, M. E.; Kolb,C. E.; Melius, C. F. Prog. Energy Combust. Sci. 1991, 54, 203. (3) Botcher, T. R.; Wight, C. A. J . Phys. Chem. 1993, 97, 9149. (4) Brill, T. B.; Brush, P. J. Philos. Trans. R. SOC.London A 1992,339, 377. (5) Brill, T. B. In Chemisfryand Physics of Energetic Materials; Bulusu, S.N., Ed.; Kluwer: London, 1990, p 255. (6) Brill, T. B.; Karpowicz, R. J. J . Phys. Chem. 1982, 86, 4260. (7) Karpowicz, R. J.; Brill, T. B. J . Phys. Chem. 1984, 88, 348. (8) Oyumi, Y.; Brill, T. B. Combust. Flame 1977, 62, 233. (9) Mesaros, D. V.; Oyumi, Y.; Brill, T. B.; Dybowski, C. J . Phys. Chem. 1986, 90, 1970. (IO) Oxley, J. C.; Hiskey, M.; Naud, D.; Szekeres, R. J. Phys. Chem. 1992, 96, 2505.

Botcher and Wight (1 1) Bulusu, S.;Weinstein, D. I.; Autera, J. R.; Velicky, R. W. J. Phys. Chem. 1986, 90, 4121. (12) Suryanarayana, B.; Graybush, R. J.; Autera, J. R. Chem. Znd. 1967, 2177. (1 3) Suryanarayanan, K.; Bulusu, S.J . Phys. Chem. 1972, 76, 496. (14) Meyer, R. Explosives; Verlag Chemie: Weinkeim, 1977. (15) Federoff, B. T.; Sheffield, 0. E. Encyclopedia of Explosives and Relafed Items; Picatinny Arsenal: Dover, NJ, 1966; Rpt. No. PATR-2700, Vol. 111. (16) Melius, C. F. In Chemistry and Physics of Energetic Materials; Bulusu, S.N., Ed.; Kluwer: London, 1990; p 21. (17) Zhao, X.;Hintsa, E. J.; Lee, Y. T. J . Chem. Phys. 1988,88, 801. (18) Farber, M. Mass Spectrom. Rev. 1992, 11, 137. (19) Behrens, R., Jr. J. Phys. Chem. 1990, 94, 6706. (20) Behrens, R., Jr.; Bulusu, S.J . Phys. Chem. 1991, 95, 5838. (21) Behrens, R., Jr.; Bulusu, S.J. Phys. Chem. 1992, 96, 8877. (22) Behrens, R., Jr.; Bulusu, S.J . Phys. Chem. 1992, 96, 8891. (23) Botcher, T. R.; Wight, C. A. InStructureandPropertiesofEnergetic Materials; Liebenberg, D. H., Armstrong, R. W., Gilman, J. J., Eds.; Materials Research Society: Pittsburgh, PA, 1993; p 47. (24) Wight, C. A.; Botcher, T. R. J . Am. Chem. Soc. 1992, 114, 8303. (25) Karpowicz, R. J.; Brill, T. B. J . Phys. Chem. 1983, 87, 2109. (26) Choi, C. S.; Prince, E. Acta Crystallogr., Sect. B 1972, 28, 2857. (27) Doyle, R. J., Jr.; Campana, J. E. J . Chem. Phys. 1985, 89, 4251.