3118
J . Phys. Chem. 1985,89, 3118-3126
Deuterium Isotope Effects in Condensed-Phase Thermochemical Decomposition Reactions of Octahydro-l,3,5,7-tetranitro-l ,3,5,7-tetrazoclnet S. A. Shackelford,*'* M. B. Coolidge, B. B. Goshgarian, B. A. Loving, Air Force Rocket Propulsion Laboratory (AFSC), Edwards AFB, California 93523
R. N. Rogers,*lb J. L. Janney, and M. H. Ebinger Los Alamos National Laboratory, Los Alamos. New Mexico 87545 (Received: September 1 7 , 1984)
The deuterium isotope effect was applied to condensed-phase thermochemical reactions of H M X and HMX-d8 by using isothermal techniques. Dissimilar deuterium isotope effects revealed a mechanistic dependence of H M X upon different physical states which may singularly predominate in a specific type of thermal event. Solid-state H M X thermochemical decomposition produces a primary deuterium isotope effect (DIE), indicating that covalent C-H bond rupture is the rate-controlling step in this phase. An apparent inverse DIE is displayed by the mixed melt phase and can be attributed to C-H bond contraction during a weakening of molecular lattice forces as the solid H M X liquefies. The liquid-state decomposition rate appears to be controlled by ring C-N bond cleavage as evidenced by a secondary DIE and higher molecular weight products. These results reveal a dependenceof the H M X decomposition process on physical state and lead to a broader mechanistic interpretation which explains the seemingly contradictory data found in current literature reviews.
Introduction The scientific literature is replete with decomposition studies of the cyclic nitramines, octahydro-l,3,5,7-tetranitro-l,3,5,7tetrazocine ( H M X ) and hexahydro-1,3,5-trinitro-s-triazine ( R D X ) , but their critical reaction steps and overall chemical mechanisms are still subject to spirited scientific debate.2-s Depending on how one counts them, as many as 8-13 chemical mechanisms or partial mechanisms have been p r ~ p o s e d , ~and -~ collectively, they eventually identify each type of covalent bond in the HMX molecule (N-N, N-0, C-N, C-H) as being an important participant in the decomposition process. In addition to mechanistic proposals outlining irreversible covalent bond rupture or reversible solid-state chemical reactions: it was recently suggested that the thermophysical weakening of strong intermolecular forces between HMX molecules in the crystal lattice may play an important role in the thermal decomposition rate.' From a chemical standpoint, the most important step in a thermal decomposition process is the rate-controlling step. This ratecontrolling step is the slowest chemical reaction taking place during an entire complex decomposition process, and it can be identified by observing kinetic isotope effects.*-" The deuterium kinetic isotope effect is a proven and powerful research technique for elucidating chemical reaction mechanisms. It can be used as a direct chemical probe to identify the ratecontrolling step and associated covalent bond rupture which occurs in a complex chemical reaction sequence. Although mainly used in gas-phase and solution-phase chemical reactions, it recently was applied successfully to the complicated reaction sequences found in high-temperature isothermal decomposition processes of neat, condensed-phase energetic compounds. This application was first demonstrated with liquid 2,4,6-trinitrotoluene (TNT) and its a,a,a-trideuteriomethyl analogue (TNT-d3). Here a primary deuterium isotope effect demonstrated that carbon-hydrogen bond rupture in the methyl moiety constitutes the ratecontrolling step.8 This study further illustrated the importance of identifying the rate-controlling reaction during the early portion of a complex decomposition process and verified the use of induction times in lieu of unavailable kinetic rate constants to study a previously undetected chemical reaction which catalyzes TNT's rapid exothermic decomposition. The same technique was subsequently employed to study the solid-phase decomposition of 'Portions of this paper were presented at the following scientific meetings: 12th North American Thermal Analyses Society Conference, Williamsburg, VA, 25-29 Sep 83; 20th JANNAF Combustion Meeting, Monterey, CA, 17-20 Oct 83; 1983 Pacific Conference on Chemistry and Spectroscopy, Pasadena, CA 26-28 Oct 83.
0022-3654/85/2089-3 118$01.50/0
1,3,5-triamin0-2,4,6-trinitrobenzene (TATB) and several other high energy compound^.^ This investigation verified a solid-state primary deuterium isotope effect in TATB indicating nitrogenhydrogen bond rupture to be the rate-limiting step. Additionally, an apparent relationship between isothermal DSC deuterium isotope effects and critical-temperature data was found. Both of these studies provide precedental scientific approaches and mechanistic information directly relevant to our HMX investigation. Examination of the HMX molecule's structure (Figure 1) illustrates the types of deuterium isotope effects possible from its decomposition. The 1.41 value is the minimum theoretical high-temperature limit for a primary deuterium isotope effect,'* although a value of 1.35 has been reported as a valid minimum experimental limit.13 The decomposition mechanism of H M X appears to be an even more complicated process than that followed by TNT, TATB, or other nitroaromatic compounds. Therefore, our use of deuterium isotope effects proved especially important in characterizing individual rate-controlling steps in three distinctly ( I ) (a) Current address: European Office of Aerospace Research and Development, 223/231 Old Marylebone Road, London NW1 5TH, England. (b) Full address: Los Alamos National Laboratory (MS C920), P.O. Box 1663, Los Alamos, NM 87545. (2) Bogs, T. L. 'Fundamentals of Combustion of Solid Propellants"; Kuo, K. K., Ed.; AIAA, Inc.: New York, 1984; Prog. Astronaut. Aeronaut. Ser., Vol. 90. (3) (a) Schroeder, M. A. CPIA Publication 366, 19th JANNAF Combustion Meeting, Greenbelt, MD, Oct 1982; Vol. I, 321. (b) CPIA Publication 347, 18th JANNAF Combustion Meeting, Pasadena, CA, Oct 1981; Vol. 11, 395. (c) CPIA Publication 329, 17th JANNAF Combustion Meeting, Hampton, VA, Sep 1980; Vol. 11, 498. (4) Shaw, R.; Walker, F. E. J . Phys. Chem. 1977, 81, 2572. (5) McCarty, K. P. AFRPL-TR-76-59, 1976. ( 6 ) (a) Fifer, R. A. CPIA Publication 366, 19th JANNAF Combustion Meeting, Greenbelt, MD, Oct 1982; Vol. I, 31 1. (b) 'Fundamental Directions for Energetic Material Decomposition Research"; Joint ONR, AFOSR, ARO Workshop Report, Berkeley, CA, 198 1. (7) Brill, T. B.; Karpowicz, R. J. J . Phys. Chem. 1982, 86, 4260. (8) Shackelford, S. A,; Beckmann, J. W.; Wilkes, J. S . J . Org. Chem. 1977, 42, 420 1. (9) Rogers, R. N.;Janney, J. L.; Ebinger, M. H. Thermochim. Acta 1982, 59, 287. (10) Shackelford, S. A,; Beckmann, J. W.; Wilkes, J. S.; Gunziger, M. L. 7th Nitroaromatic Seminar, Dover, NJ, Oct 1977. (11) Matveev, V. G.; Dubikhin, V. V.; Nazin, G. M. Izu. Akad. Nuuk SSR,Ser. Khim. 1978, 474. (1 2) Bigeleisen, J.; Wolfsberg, M. "Theoretical and Experimental Aspects of Isotope Effects in Chemical Kinetics"; Prigogine, I., Ed.; Interscience: New York. 1958: Adv. Chem. Phvs. Vol. I. DD 15-31 and reference 8. (13) Streitwieser, A,; Jagow, R. H.;Fghey, R. C.; Suzuki, S . J . Am. Chem. SOC.1958, 80, 2326.
0 1985 American Chemical Society
Reactions of H M X
The Journal of Physical Chemistry, Vol. 89, No. 14, 1985 3119 TABLE I: Induction Period Solid-state Decomposition
compd
temp, K
HMX-da HMX HMX-d, HMX HMX-d, HMX
553 553 552 552 551 551
11, s 105 f 9 53 f 6 198 f 27 78 f 3 215 f 28 101 f 6
av
DIE
(~ID/~IH)
1.98 f 0.27 2.53 f 0.36 2.13 f 0.31 2.21 f 0.18
TABLE II: Acceleratory Mixed Melt Thermochemical Decomposition
1
> 1.41 2" EFFECT: 1 .OO- 1.34 lNVERS€ EFFECT: < 1.00
L 1" EFFECT:
Figure 1. Possible deuterium isotope effects in the H M X molecule.
different H M X physical states (crystalline solid, mixed melt, and homogeneous liquid) wherein each state contributes differently to the overall decomposition process. Many researchers believe that a detailed mechanistic description is needed for each physical state to understand adequately the decomposition of HMX.14 This paper describes our application of deuterium isotope effects to three types of thgrmally initiated processes, slow decomposition, rapid pyrolysis, andthermal explosion. Our purpose was to identify in situ the rate-coqtro!ling chemical reactions that determine the overall decomposition rate of the HMX molecule. The data gained by isothermal qifferential scanning calorimetry (DSC), rapid pyroprobe pyrolysis studies, and critical-temperature measurements made with a time-to-explosion test reveal that HMX displays three distinctly different deuterium isotope effects depending on its predominant physical state (solid, mixed mel;, liquid). Our deuterium isotope effect results and their resultant mechanistic implications permitted us to elucidate a more general and unified mechanistic model for the thermochemical decomposition of HMX. Experimental Section
CAUTION HMX is a high explosive material that is 1.6 times more powerful than TNT. Care should be taken to use proper laboratory shielding and safety procedures in all handling operations. Isothermal DSC Analysis. All isothermal DSC measurements were made with Perkin-Elmer Model DSC-IB instruments. Most runs were conducted with unsealed flat aluminum sample pans, Perkin-Elmer Part No. 21 9-0041, because their inherent design kept the cell's free volume to a minimum and permitted the rapid escape of gaseous products. This significantly reduced contributions from interfering heterogeneous vapor-phase reactions. All data presented were obtained with the flat aluminum pans; however, several experimental runs were made with low-freevolume perforated sealed cells, Perkin-Elmer Part No. 219-0062, which contained an aluminum cylindrical plug again to reduce heterogeneous vapor-phase r e h i o n contributions. The deuterium isotope effect trends were similar with both types of sample pans, although the several runs made with the aluminum cylindrical plugs gave a higher albeit less precise deuterium isotope effect (1.5) in the liquid state. Prior to use, the acetone recrystallized H M X was vacuum-dried overnight at 30-50 "C. Instrumental temperatures were calibrated daily under identical conditions with a 5 K/min scanning rate. This rate was the slowest which still produced a meping endotherm without the endotherm being masked by HMX's rapid exothermic response which actually begins during the melting process. The onset of the H M X endotherm was assigned a melting point value of 555 K (282 "C). Since the melting point of HMX is dependent upon its extent of (14) Karpowicz,
R. J.; Brill, T.B. Combust. Flume 1984, 56, 317.
compd
temp, K
HMX HMX-d, HMX HMX-d, HMX HMX-d,
553 553 552 552 551 551
k , s-I 1.06 1.11 0.36 0.45 0.30 0.38
f 0.18 f 0.24
f 0.04 f 0.16 f 0.05 f 0.19
av
DIE (kH/kD) 0.95 f 0.23 0.80 f 0.28 0.79 f 0.22 0.85 i 0.22
TABLE 111: Decay Period Liquid-State Decomposition compd temp, K k X lo2, S-' DIE (kH/kD) HMX HMX-dg HMX HMX-d8 HMX HMX-d8 av
553 553 552 552 55 1 55 1
5.81 5.07 4.85 4.35 4.43 3.87
f 0.60
f 0.30 f 0.50 f 0.40 f 0.40 f 0.20
1.15 f 0.14 1.12 f 0.15 1.11 f 0.11 1.13 f 0.08
TABLE IV: Identically Synthesized HMX Sample Thermochemical Decomposition DIE
compd HMX DMX" HDMXb HMX DMX HDMX HMX DMX HDMX
State solid solid solid
tl,
S
k x lo2, S-I
(tlD/tIH or kH/kD) 1.74 f 0.22
30.8 f 3.1 29.3 f 4.4 27.6 f 8.3 5.67 f 0.51 4.43 f 0.61 5.64 f 0.79
1.05 f 0.18
98 f 8 1170 f 16 62 f 6
m meltc m melt mmelt liquid liquid liquid
" D M X = HMX-d,.
1.28 f 0.21
b H D M X = 98% H M X / 2 % HMX-d, copre-
cipitated. CMixedmelt. decomposition, great care was taken to conduct the daily calibrations under identically controlled conditions in order to obtain the same reproducible temperature setting. Calibrations taken both before data acquisition and after all daily isothermal DSC runs showed no temperature drift occurred. An empty cell and cover (identical with that used to hold the HMX sample) always remained upon the reference support. A sample size of 1.50 f 0.04 mg was weighed into the flat aluminum pans. A lid was placed into the sample pan and was then tamped with an appropriately sized metal rod to distribute the HMX powder evenly over a flat surface area. Analyses were conducted at one degree temperature intervals from 548 (275 "C) to 553 K (280 "C) on a temperature equilibrated DSC under a nitrogen atmosphere. Samples of H M X and HMX-d8 were usually run alternately in the overlapping temperature range suitable for HMX-ds; this provided a continuous experimental check that daily temperature drift did not occur after initial calibration. Best HMX-ds results were obtained between 551 and 553 K. Data produced during individual experimental runs were recorded directly with a Series 2090 Nicolet digital oscilloscope, and the stored data were transferred to a Hewlett Packard 85A desk top computer. The data were evaluated with the HP-85 computer and graphed with an HP 7225B plotter to produce the curves illustrated in Figure 2. Analysis of autocatalytic rate curves such as shown in Figure 2 (bottom) provided the induction time ratios and rate constant
The Journal of Physical Chemistry, Vol. 89, No. 14, 1985
3120
INDUCTION PERIOD -+cGELER I
I
AlORY
-
PHASE D
ti
OECAl PERIOO-
1
!
LIQUID
ENERGY
nUTOCATnLvTIC RATE PLOT
I
XI1 - X J
Figure 2. (Top) Isothermal DSC decomposition curve of H M X at 551 K. Time of entire scan ranged between 350 and 400 s. (Bottom) Autocatalytic rate plot of the H M X isothermal DSC decomposition curve.
TABLE V: Rapid Pyroprobe Pyrolysis of HMX and HMX-d8 run comDd temp, K tl, s DIE (tmlfw) 1
2
HMX-dB HMX HMX-d, HMX
534 534 534 534
216 71 210 68
f f f f
11 13 20 09
3.02 f 0.57 3.06 f 0.51
ratios presented in Tables I-IV. These results were used to determine the deuterium isotope effect data at 551, 552, and 553 K in the solid, mixed-melt, and liquid states. Pyroprobe Pyrolysis. All pyroprobe pyrolysis reactions were obtained with a Chemical Data Systems, Inc., 120 Pyroprobe apparatus with a CSD 382 Extended Pyroprobe module. The pyroprobe pyrolysis curves were obtained on a Varian Aerograph Model A-25 millivolt strip chart recorder. A 1.50 f 0.04 mg sample of H M X was tamped into an open ended 2.0 mm X 15 mm quartz tube atop a small flat glass wool plug to form a disk-shaped sample. Meanwhile, the high heating rate pyroprobe apparatus was temperature cycled in 15-30-min durations at the desired temperature until the same isothermal temperature was reproduced from one blank run to another. The quartz tube containing the disk-shaped H M X sample was placed into the platinum coil of the pyroprobe gun such that the probe’s thermocouple rested directly against the thin glass wool plug next to the bottom of the disk-packed HMX sample. The entire pyroprobe containing the sample was then placed into a large tubular sleeve, and a low flow of nitrogen was passed through the outer tubular sleeve. The pyroprobe was activated, and a curve like that illustrated in Figure 6 was produced. The pyrolysis temperature was taken as that present a t the end of the induction period immediately prior to the exothermic response. The same point defined the end of the solid-state induction period. Deuterium isotope effects were obtained by comparing the HMX-d8 and normal HMX industion times (Table V). Time-to-Explosion E~periments.’~Critical temperatures for H M X and its deuterium analogue (HMX-d8) were determined with a LANL-designed time-to-explosion apparatus which measures the time to a catastrophic self-heating event as a function of temperature. The lowest temperature a t which a sample of given size and shape will self-heat to destruction represents the critical temperature. Critical temperatures are dependent upon sample size and shape. Thinner samples produce higher critical temperatures. Lacking accurate kinetics constants for these systems, we cannot normalize the samples to the same thickness. This, however, was not a factor in correlating critical temperatures (15) Rogers, R. N. Thermochim. Acta 1975, 1 1 , 131.
Shackelford et al. of HMX and HMX-d8 to an inverse deuterium isotope effect since the thinner HMX-d8 samples provided a lower critical temperature than normal HMX. The H M X and HMX-d8 were recrystallized in an identical procedure to ensure rate process differences were not caused by sample purity, perfection, or crystal habit dissimilarity. A 40-mg H M X or HMX-d8 sample was pressed into an empty DuPont E-38 blasting-cap shell. The shell was sealed such that gas overpressure alone could not be interpreted as a thermal event. Next, each sample thickness was measured accurately. The sample shell was placed in the sample holder assembly and lowered into a Woods metal bath at a preset constant temperature. The time-teexplosion was measured for each sample; new samples were introduced at succdssively lower temperatures until no explosion mugred. The average HMX sample thickness was 0.072 cm while the HMX-d8 samples averaged 0.056 cm; the thinner HMX-d8 samples reflect its greater density. Considerable data scatter results in these tests when the compound’s critical temperature is below its nominal melting point; however, critical temperatures can still be determined with good precision. Although no-go explosion points are not displayed in Figure 7, H M X did not explode below 257 OC and HMX-d8 not below 242 O C (mp HMX 282 “C). HMX, HMX-d8, and 98% HMX/2% HMX-d8 Sample Preparation. The fine particle-sized 0-HMX (
