J. Phys. Chem. 1993,97, 8152-8758
8752
Thermal Decomposition of Energetic Materials. 61. Perfidy in the Amino-2,4,6-trinitrobenzene Series of Explosives T. B. Brill' and K. J. James Department of Chemistry, University of Delaware, Newark, Delaware 19716 Received: February 16, 1993; In Final Form: May 27, 1993
The numerous electronic, molecular, crystal, and explosive variables in the series 1,3,54rinitrobenzene (TNB), 2,4,6-trinitroaniline (MATB), diamino-2,4,6-trinitrobenzene (DATB), and triamino-2,4,6-trinitrobenzene (TATB) give 153 nearly linear correlations. While some of these positive correlations have been cited previously as important, the molecular mechanism of the trend in the shock and impact sensitivity remains unclear because these correlations disguise the fact that significant differences exist in the critical thermal reactions. Arrhenius data suggest that C-NO2 homolysis is the initial decomposition reaction during impact and shock initiation in all cases. The lower activation energy cyclization process of MATB, DATB, and TATB to furazan/furoxan products can be overstepped at shock and impact initiation temperatures. However, C-NO2 homolysis does not account for the trend in sensitivity because the activation energy is relatively insensitive to the ring substituents. The trend in energy released by the intermediate reactions as reflected in the gaseous product distribution of TNB, MATB, DATB, and TATB at 520 OC appears to be important. The contribution to the total heat from the exothermic products (CO, C02, HNCO) decreases and that from the endothermic products (NO, NzO, HCN) increases as NHI groups are added.
I. Introduction A strong motivation for basic research in the field of explosives is the need to understand the controlling mechanisms of impact and shock sensitivity. The compounds 1,3,5-trinitrobenzene (TNB), 2,U-trinitroaniline (MATB), dia~n0-2946-tfini~obenzene (DATB), and triamino-2,4,5-trinitrobenzene(TATB) are
02Ng-"2 NO2
02N+Noz NO2
TNB
MATB
NO2 DATB
NO2 TATB
a category of explosives whose molecular properties, sensitivity to various stimuli, and explosive performance are frequently compared for this purpose. Upon closer examination, an astonishing number of molecular and crystal parameters are found to correlate with the impact and shock sensitivity in this series. In some cases the correlations are chemically absurd but nevertheless have high statistical coefficients of correlation. In other cases high coefficients of correlation have been previously cited as evidence that a particular molecular or crystal variable is importantto the impactand shocksensitivityor the-performance. The existence of a large number of correlations tends to cloud rather than clarify understanding about this series, because it overemphasizes many coincidental similarities and underemphasizes the fact that there are also major differences in the chemistry. Determination of the chemistry that begins theexplosion process in thecondensedphaseis complicated by the fact that the pressure and temperature increase rapidly and dramatically during the event and that bimolecular reactions occur. Chemical reactions begin at the voids and defects in the crystals or melt rather than
in the homogeneousphase.' The vast majority of literature dealing with initiation chemistry of explosives employs temperature as the variable. Studies in which the effect of pressure is probed usually involve a static pressure. For example, the thermal decomposition rate of TNT is relatively independent of static pressure up to 50 kbar.2 The temperature dimensionis considered in the discussion that follows. An examination of kinetic data indicates that C-NO2 homolysis will be a common, quasi-unimolecularreaction of TNB, MATB, DATB, and TATB during shock and impact initiation. Isomerization of the C-NO2 bond to C-ONO is also a feasible reaction but is impossible to assess because little concrete information is a~ailable.~ Therefore, we fall back on the C-NO2 homolysis reaction as a central, plausible reaction in the initiation process. Internal cyclization to furazans and furoxans that are found in subcritically shocked and impacted samples of the amino derivatives dominates because of the subthreshold temperature. However heuristically attractive they are, furazans and furoxans fail to account for the shock and impact sensitivity trend of fresh TNB, MATB, DATB, and TATB because TNB cannot undergo this reaction. Instead, a common initial reaction of this series immediately leads to a different intermediate gaseous product distribution whose subsequent reactions influence both the explosive performance as well as the sensitivity of the TNB, MATB, DATB, TATB series. The pattern of these reactions is revealed by T-jump/FTIR spectroscopy, where the IR active gaseous products are detected from samples heated at 2000 "C/s to a constant temperature of 500 O C or greater. In this article, reconciliation of the differences of the impact and shock sensitivity of TNB, MATB, DATB, and TATB is proposedin terms of chemical processes as opposed to the previous approachesthat makeuse of thesmooth trends among their many ground-state properties.
11. Chemistry in Relation to Shock and Impact Sensitivity The shock and impact sensitivity of a solid is influenced by numerous physical variables,' such as the particle size, crystal hardness and orientation, defect size andvolume,and atmospheric conditions. Key observations that connect thermochemical kinetics to impact sensitivity are those of Robertson and Yoffe4
0022-3654/93/2097-8152$04.00/0 0 1993 American Chemical Society
Amino-2,4,6-trinitrobenzeneExplosives
The Journal of Physical Chemistry, Vol. 97, No. 34, 1993 8753
TABLE I: Sensitivity and Performance Parameters compd Hs,cm Pw,kbarC Pa,GPad A€fe~,kJr~ TNB loo0 15 22.9 4.734 MATB DATB TATB e
177" 3200 4906
28 46 70
24.7 26.7 29.3
4.263 4.138 3.990
Reference 13. Estimated, but frequently used, see refs 8 and 19. Reference 8. Reference 11. e Reference 12.
and Wenograd.5 Bulusu and Autera6 connected the shock sensitivity of TNT to the low-temperaturethermal decomposition mechanism; however, the connection is a loose one? The initial chemical reaction that is responsible for the slow thermal decomposition and the shock and impact sensitivity need not be the same. For the TNB, MATB, DATB, TATB series, Storm et aL8bolstered the possible connection between shock and impact sensitivityby noting theexcellent correlation (R2> 0.99) between the shock and impact sensitivities. Although statistically less convincing, performance characteristics, such as the detonation velocity,"l the Chapman-Jouguet (C-J) pressure,I1and the heat of explosion,12have been proposed to be influenced by the kinetics of the initial thermal reaction steps. Taken together, there is broad acceptance that the thermal decomposition reactions of the material play a significant role in its sensitivity to mechanical stimuli and the explosive properties. The specific mechanisms are less clearly established.
III. Molecular, Crystal,and Explosive Properties of TNB, MATB, DATB, and TATB Extensive data are available for this series which are grouped below according to whether the basis is predominantly (1) the explosive characteristics, (2) the composition and crystal properties, or (3) the molecular and electronic structure. Most entries are self explanatory,but a few of the more specialized parameters require brief description. A. Explosive Characteristics. Table I lists the trends of the sensitivity and explosive behavior: the impact sensitivity!J3 HSO; the shock initiation threshold pressure,14 Pw; the heat of explosion,'2 AHex;and the C-J pressure! PCJ. The impact sensitivity is usually measured by the drop-weight impact test. The sensitivity index is the height, HSO,that a weight of a given mass must be dropped onto the sample to produce an explosion 50%of the time. The duration of heating time averages to about 250 ~ s , ~and J 3 the instantaneous pressure is 7-1 5 kbare8 The shock initiation threshold pressure, Pw, is frequentlyobtained from the gap test, in which a booster charge of a standard configuration is used to transmit a high-pressure shock wave through a buffer plate to the explosive of interest that is packed to 90% of its theoretical maximum density. The resulting shock sensitivity, Pw, is the pressure required to produce a detonation 50%of the time. The C-J pressure, PCJ,is the pressuredeveloped at the rear of the reaction zone (the sonic transition point) in the detonation wave. PCJis proportional to the sample density, p, and the detonation velocity, D, according to hydrodynamics, by eq 1. The heat of explosion is frequently obtained experimentally
PcJ 3~ 0.25pD2 from the cylinder test, but values in Table I were taken from Zeman's compilation.12 B. Composition and Crystal Properties. Table I1 lists the molecular weight, M,the crystal density,lsl* p; the density of NO2 groups in the crystal lattice,19 pN; the standard heat of formation, A?ffo;themelting point, mp; and the oxidant balance,l3 OBioo. The NO2 group density was defined by Odiotl9 according to eq 2, where 2 is the number of molecules per unit cell and Vis the volume of the unit cell. The oxidant balance,l3 OBlm, is
TABLE LI: Compositional and Crystal Parameters M, P, PN', compd gmol-I gcm3 gem-' calgl mp,K
OBlm
TNB MATB DATB TATB
1.688" 0.653 213 -41.8 396 -1.46 1.7606 0.646 228 -78 461 -1.75 1.837' 0.628 -122.6 243 559 -2.06 1.93gd 0.623 -143 254 623, -2.33 Reference 15. *Reference16. Reference 17. Reference 18.e Reference 19.fNot well characterized.
pN = Z M / V
(2)
defined according to eq 3 as the number of equivalents of oxygen
OB,, = 100(2n, - nH- 2nc - 2nc,)/M
(3)
in 100 g of a CHNO explosive required to convert all of the H to H20 and C to CO. n is the number of atoms of each element in the formula. The melting point is partly a reflection of the intermolecular cohesive force in the crystal lattice. Hydrogen bonding is an important cohesive force in the crystals of these c ~ m p o u n d s , although ~ ~ - ~ ~ interplanar *-electron interaction and possibly dipole-dipole interaction among the NO2 groups contribute as well. C. Molecular and Electronic Structure Properties. Table I11 lists largely intramolecularparameters: the electrostatic potential calculated at the midpoint of the C-NO2 bond, Vmid,determined by a Mulliken orbital population analysis at the level of C I N D W and ab initio SCF-MO with STO-3Gz1or STO-5G22 basis sets; the number of ?r electrons in the benzene ring computed at the STO-3G level;l4 the shake-up promotion energy separation of N(1s) of the NO2 groups obtained by X-ray photoelectron spectroscopy, ASPE23 and the average bond distances in the crystal structure, d.15-"3
IV. Similarity of Properties of TNB, MATB, DATB, and TATB Perusal of the columns of Tables 1-111 and of the tables relative to one another reveals that a trend exists in all columns. To emphasize the similarity, the correlation coefficient, R2, and the confidence level of statistical significance accordingto the t test>* eq 4, are given in Table IV for all columns plotted vs the impact t = I?(-> n - 2 112 (4) 1-R2 sensitivity, HSO. To exceed 95% confidence, the value of t for four data points, n, must exceed 3.18. In all cases except one, this confidence level is achieved. In fact, 12 data sets exceed 99% confidence (t > 5.84). The only case that fails 95% exceeds 90% confidence (t > 2.35). Furthermore, two sets of data that are linearly related to a third are linearly related to each other. Therefore, among the 18 sets of data in Tables 1-111, there are 153 positive correlations. A few of course are dependent or, at least, interconnected. The density and C-J pressure are related through the detonation velocity by eq 1 The molecular weight enters into OBlw by eq 3. The increase in the average C-C and N-O distances parallels the decrease in the average C-NO2 distance as NH2 groupsare added. In some cases the correlations are pure folly even though R2 is high. For example, melting point vs shake-up promotion energy separation, heat of formation vs density, and molecular weight vs C-NO2 midpoint electrostatic potential are among the more chemically ridiculous correlations. On the other hand, several of the correlations of the columns in Tables 1-111 have been proposed to identify crystal and molecular factors that influence the impact and shock sensitivity of TNB, MATB, DATB, and TATB. Sharma and Beard23J5 correlated H ~ in o Table 1 with the shake-up promotion energy separation (Table 111) and proposed it as evidence of the I
8754
Brill and James
The Journal of Physical Chemistry, Vol. 97, No. 34, 1993
TABLE III: Molecular and Electronic Parameters v d d , au compd
STO-5G"
STO-3Gb
CINDOe
TNB MATB DATB TATB
0.208 0.174 0.146 0.103
0.197 0.168 0.130 0.102
0.012 -0.018 -0.048 -0.063
a
no. re5.96d 6.16 6.36 6.54
ASPE: ev
C-NO2
0 0.8 2.1 2.7
1.482/ 0.469g 1.454h 1.419'
d, A N-O 1.19v 1.219g 1.233h 1.243'
C-C 1.380/ 1.393g 1.411h 1.442'
Reference 22. b Reference 21. Reference 20. Reference 14. * Reference 23. /Reference 15. Reference 16. Reference 17. Reference 18.
TABLE I V Coefficients of Correlation and the Confidence Level Based on the "tTest" for Plots of All Entries in Tables 1-111 vs the Impact Sensitivity Index, HSO parameter correl coeff, R2 t (eq 4)" 0.998 0.992 0.773 0.952 0.991 0.927 0.931 0.972 0.971 0.982 0.972 0.919 0.950 0.968 0.981 0.955 0.994
31.6 15.75 2.61 6.30 14.83 5.04 5.20 8.33 8.18 10.45 8.49 4.76 6.17 7.78 10.17 4.55 18.2
Significant at confidence level 90%2 2.35;95%13.18; 99%2 5.84. importanceofexcited states in determiningthe impact sensitivity.25 Correlation of H50 with vfid in Table I11 has been suggested to show that C-NO2 bond strength and homolysis play a role21,22 or are rate controlling20 in the impact and shock sensitivity of the TNB, MATB, DATB, TATB series. Table IV reveals that the C-NO2 bond distance correlates as well or better with H50 as does Vmid. In fact, the trend in all columns of Table I11principally results from progressive ground-state structural changes along this series. That is, the C-C, C-NO2, and N-0 bond distances are no more or less "controlling" of the sensitivity than any other factor at this level of analysis. Systematic differences in the bond distances figure importantly into the computed electrostatic potentials and measured relative orbital energies. Other correlations of data in Table I1 with H50 have been made. Odiotlgnoted that the density of N02groups in the crystal correlated with H50. Kamlet and Adolphl3 correlated OBlw with the log H50 for TNB, MATB, DATB, and TATB and other compounds. The trends implied that the oxidizing potential of the molecule may be important. Table I11 reveals that for this series of compounds,H50 correlates as well with OBlw as log H50. Storm et a1.8 noted that within Table 111, H50 and Pw correlate well. Rogers et al.14 proposed that the increasing amount of intermolecular hydrogen bonding as NH2 groups are added is responsible for the decrease in shocksensitivity. As intermolecular hydrogen bonds increase in number, they are proposed to absorb energy from the shock front and prevent it from being localized in the molecule.*4 Although difficult to quantify with the data available, qualitative indications of the trend in intermolecular association in the series are the melting point and, possibly, the density. Both increase as NH2 groups are added to the ring, and the shock sensitivity, Pw, decreases. However, this proposal must be accommodated into the fact that the chemistry begins in the voids and not in the bulk homogeneous material.' An inescapable conclusion of Tables I-IV is that single column correlations of the molecular, lattice, and explosive properties are not especially helpful for identifying the primary chemical mechanisms of shock and impact sensitivity of TNB, MATB, DATB, and TATB. It is especially risky to pick a column of one
TABLE V compd TNB MATB
Arrhenius Data for T b e r d Decomposition kcal mol-' log A, s-l phase T,OC 67.3 17.2 gas 380-470 38.5 8.8 gas 301-343 41.8 11 gas 240-280 liquid 240-280 32 7.2 liquid
DATB
liquid liquid solid
gas/solid solid solid solid a This low value may
TATB
249-259 320-323 220-270 250-300 297-382 331-332 345-375
48 46.3 50 47 43' 59.1 59.9 62A4
15.07 15 13.2 18.57 19.5 19
ref 27 41 42 42 43 44 45 45 46 38 44 47
result from the domination of sublimation.
table and correlate it with a column in another table. For example, any column of Table 111is as justifiable as any other to correlate with H50 in Table I. As a result, previous conclusions about molecular and electronic features that control shock and impact sensitivity in this series of compounds are contradictory: excited electronic states25vs C-NO2 electrostatic potentials22 vs intermolecularhydrogen bonding% oxidant ba1an~e.l~ With more than 100positive correlations among the data, single correlations are unlikely to elucidate the true factors. Perhaps all of these factors (or none of them) contribute to the sensitivity to some degree. V. Differences between TNB, MATB, DATB, and TATB First, if early thermal decomposition reactions play an important role in the sensitivity of the material to shock and impact, then the smooth correlations in Table IV are deceptive. The initial steps of decomposition, at least at lower temperature in this series of compounds, are different. These differences are discussed below. Second, the energy generated by the explosive is determined by the secondary, or propagation, reactions to a large extent. These reactions also appear to differ for this series of compounds. A. Initiation Chemistry. From the point of view of chemical processes, the initiation step is the first reaction or class of reactions that is stimulated by the input of energy. The initiation step is almost always endothermic because it usually involves bond rupture. In the Evans-Polanyi formalism,Mthe activationenergy, E,, is closely related to the bond dissociation energy. TNB is a thermally rather stable explosive because the dissociation energy of the weakest bond, the C-NOz b0nd,2~,2* is about 70 kcal mol-'. Arrhenius constants for thermal decomposition of TNB in the gas phase are given in Table V. E. for TNB is typical of nitrobenzene, dinitrobenzene, and many substituted nitrobenzenes, all of which have E. values in the 6575 kcal mol-' range.27-30 Several earlier studies3'J2 report values of 5 1-52 kcal mol-' and are likely in error. Note that E. resembles the C-NO2 bond dissociation energy, which is consistent with eq 5 being the rate determining step (RDS).Also consistent with eq 5 is the fact that fragmentation of the TNB cation in the gas phase produces ions whose m / e values revealed by mass spectrometry correspond to successive loss of NO2 while the aromatic ring remains intact.33
Amino-2,4,6-trinitrobenzeneExplosives
The Journal of Physical Chemistry, Vol. 97, No. 34, 1993 8755 15 I I
C-NO2 HOMOLYSIS FURAZAN/FUROXAN (ref 38)
. ‘. The initial thermal decomposition mechanism of MATB, DATB, and TATB below 500 OC is substantially morecomplicated than that of TNB because of participation by the NH2 groups. It has been known for at least 80 years that an NH2 group ortho to NOz on an aromatic ring will cyclize on heating to produce furazan and furoxan derivatives by eq 6.34.35 These reactions are
“
-.....:.
0
-25 0.0008 ~
I
,
0.001
~
0.0012
0.0014
0.0016
0.0018
1/T K Figure 1. Arrhenius plots of C-NOz homolysis of TNB (eq 5) and the two sets of gas-phase kinetics in Table V for MATB. The MATB data reflect cyclization to furazan/furoxan products (eq 6).
a small subset of ortho-related substituent reactions that have wide synthetic ~tility.3633~Rogers et al.3* found a primary deuterium isotope effect ( k ~ / =k 1.5) ~ when TATB thermally decomposed at 297-382 OC. This finding is consistent with N-H rupture occurring in the transition state of the RDS, which is possible in eq 6. Sharma et a1.25,39,40found furazans and furoxans in samples of TATB that had been heated or subcritically shocked or impacted. They proposed a role for these furazansand furoxans in shock and impact initiation. Table V lists Arrhenius data for MATB, DATB, and TATB in the gas, liquid, and solid phases. The tendency to react by eq 6 is the most logical explanation for the much smaller values of Ea and the prefactor A compared to those for TNB. Unfortunately, further analysis is necessarily qualitative because of the scatter of the data, even from the same researchers, and the fact that measurements were made under different conditions (DTA, DSC, manometry, electron beam heating traces, and effusion cell mass spectroscopy). In the liquid and solid phase, matrix effects and catalysis by impuritiesand the decomposition products influence the Arrhenius values.3 If the data are accepted, then there is a rough pattern of increasing E, and A from MATB to TATB. These trends in the Arrhenius parameters are consistent with the decreasein impact sensitivityalong this series. However, the Arrhenius parameters for TNB are clearly out of line because the initial reaction (C-NO2 homolysis) is altogether different. A major conclusion is that the impact and shock sensitivity cannot be determined by the low-temperature decomposition mechanism of MATB, DATB, and TATB if TNB is retained in the comparison. Otherwise, TNB, which requires C-NO;! homolysis, should be the least sensitive to impact while the aminotrinitrobenzenecompounds, which easily undergo furazan and furoxan formation, should be more impact sensitive. The observation of furazan and furoxan formation in samples recovered from shock and impa~t~s~3~.40 treatment is understandable from the fact that the temperature achieved did not induce an explosion or detonationand thereby favored this lower temperature reaction channel during rapid heat-up and cool-down of the sample. These products could sensitize a “rough handled” explosive.25 If the lower activation energy decomposition channel of the aminotrinitrobenzeneseries fails to account for the impact and shock sensitivity trend in Table I, then how is it explained? Two alternate factors stand out. The first is the probability that the C-NO2 bond homolysis really is the universal initiating reaction. The second is the role of the intermediate chemistry in the impact and shock sensitivity and the performance. These two factors are now analyzed. Owens et a1.20.21 and Politzer et a1.22calculatedtheelectrostatic potential at the midpoint of the C-NO2 bond, and concluded
from the trend (Table 111) that the C-NO2 bond characteristics are the rate-controlling factor in the shock and impact sensitivity. More simply, it might be said that the trend in the C-NO;! bond strength as reflected in the bond distance (Table 111) is the important variable. For this to be the case, the instantaneous temperature rise during shock or impact initiation would have to be sufficient to overstep the lower E, furazan-furoxan routes of eq 6 directly into the higher Ea C-NO2 homolysis reaction of eq 5. Only a rough estimate of the temperature required is justified because of the spread in the aminonitrobenzeneArrhenius data in Table V. The gas-phase data for TNB and MATB are probably the least encumbered, although they are also the least relevant. Figure 1 is the Arrhenius plot of the C-NO;! homolysis kinetics of TNB and the two gas-phase measurements on MATB. The Arrhenius values for C-NO2 homolysis are essentially independant of temperature over this range.’ The isokinetic temperature of eqs 5 and 6 is 475 or 625 OC depending on the MATB data used. The fact that these are isokinetic temperatures in the gas phase makes their suitability uncertain for use in the condensed phase. If trusted, these temperatures are easily achieved during impact and shock, where Wenograds estimated that TNB reached 1060 OC in 250 ps during the drop-weight impact test. Estimates of isokinetic temperatures for eqs 5 and 6 in the condensed phase can be made, but with much less confidence. If the C-NO2 bond dissociation activation energy is fixed at the gas-phase value2 of 65-70 kcal mol-’ and the liquid-phase DATB valueu of E, = 46.3 kcal mol-’ is used, then the isokinetic temperature of eqs 5 and 6 is about 1880 OC. The other liquid and solid data in Table V also produce very high isokinetic temperatures, or in the case of TATB, impossible negative temperatures. Comparablekinetic data on this series of compounds is definitely needed. Even though C-NO2 homolysis stands out as a dominating initial reaction of these compounds under shock and impact conditions, it is doubtful that C-NO2 homolysis controls the smooth trend of impact and shock sensitivity in TNB, MATB, DATB, and TATB. The C-NO2 bond dissociation activation energy simply is not strongly sensitive to ring sub~tituents,2~-30 being in the range of 61-70 kcal mol-’ for a wide variety of substituent^.^ Therefore, the intermediate chemistry following C-NO2 homolysis becomes crucially important. B. Propagation Chemistry. After the initiation step, an immensely complex network of rapid, largely exothermic propagation reactions occur that determine how fast and how exothermicallythe explosion or detonation develops. Some success at experimentally identifying the first energy-producingclass of reactions has been achieved for the energetic materials HMX,4* RDX,48?49azide and NH4+ saltssO~sl of NO3-, ClO4-, and N(N02)z- by using T-jump/FTIR spectroscopy.52 Although the same degree of success was not achieved with TNB, MATB,
Brill and James
0756 The Journal of Physical Chemistry, Vol. 97, No. 34, I993
1 HNCO
--- HCN
/
*,\.-.. .......................
;
0
4
2
0
6
0 1
0
6
TIME (sec) Figure 2. T-Jump/FTIR data for TNB heated at 2000 s-l to 520 OC under 40 atm of Ar. The control voltage trace shows a sharp exotherm
at about 2.5 s; no other gas products were detected except a small amount of H20 which was not quantified. 1.7 ,
I;\\
2 0
F: C
I
p: 1.2 -
I
Tf = 5 2 0 ' ~
.
V W
\
:; . J
E-
2
1
I
z
,. '.
I
I .
'\
'.,
.:
.......
'.------
....
.................
1.5
I 1 E-
I/:.: f
i T f =
I! I:
..........................................
......................
I:: 1:
2 0.5
0
2
I
520°C
4
6
6
TIME (sec) Figure 4. T-Jump/FHR data for DATB heated at 2000 0C s-* to 520 'C under 10 atm of Ar. The exotherm is broad and ragged for this compound. A small amount of H 2 0 was detected but not quantified.
DATB, and TATB, the gas products observed qualitatively describe the trend. Figures 2-5 are T-jump/FTIR data for TNB, MATB, DATB, and TATB. The growth rate of the gas products through the exotherm and the final gas products formed under elevated pressure and the same temperature are shown. The exotherm is the negative excusion of the Pt filament voltage from horizontal. For all four compounds the gas products rise in concentration after the exotherm because the elevated pressure in the cell suppresses the diffusion rate into the IR beam. The products
2
4
6
8
-I 1
TIME (sec) F i p e 5. T-Jump/FTIR data for TATB heated at 2000 OC s-l to 520 OC under 10 atm of Ar. The exotherm is broad and ragged. A small amount of HzO was detected but not quantified.
TABLE VI: QUnnWiedCas Products from Heating at 2OOO O C s-l to 520 O C under 10-40 atm of Ar "xntage of auantified products compd CO2 CO NO HCN N20 HNCO TNB 6 94 MATB 17 38 28 13 3 DATB 20 31 21 18 2 8 TATB 27 20 26 13 3 11 spikein concentrationbefore dispersingto fill the cell completely. Note that the relative concentrations change very little during this process. Figure 2 shows the results for TNB heated at 2000 OC/s to 520OCunder4OatmofArandthen heldat520OC. Thispressure of Ar was needed to suppress evaporation. A sharp exotherm occurs at about 2.5 s. The only IR active gaseous products are CO and C02 along with a small amount of H20 (not quantified). N2 and H2 undoubtedly form under these conditions because no IR active nitrogen-containing products and a little quantity of hydrogen-containingproducts were detected. Figures 3-5 show T-jump/FTIR data for MATB, DATB, and TATB at 520 OC and 10 atm of Ar, respectively. Less pressure is required to suppress evaporation/sublimation of these compounds than was required for TNB. MATB and DATB were visually observed to melt on the Pt filament before the exotherm. TATB remained solid and rapidly discolored before the exotherm. The control voltage traces reflecting the characteristics of heat release are markedly different for the four compounds. The traces for DATB and TATB are ragged and broad. Visual observation reveals that these compounds ignite at random sites and bum outward from the ignition spots. This behavior produces spikes and breadth in the release of energy. MATB and TNB ignite uniformly giving rise to a sharp release of energy. The relative percentages of the gas products after the exotherm are compiled in Table VI for the four compounds. Products having endothermic heats of formation (NO, HCN, and N20) form when NH2 groups are present on the ring. HNCO also increases as the NH2 content increases. Even though the C/O ratio of the parent compounds remains fixed, the COdCO ratio increasesas the NH2 content increases. The energy distribution among the gaseous products reflects the fact that the heat of explosion of TNB is higher than those of the aminonitrobenzeneexplosives. The high heat of explosion of TNB results from the dominance of products with negative AHfo (CO, COz. and HNCO) and the absence of products with positive AHro (NO, HCN, and NzO). For MATB, DATB, and TATB the amount of energy from the exothermic products (CO and CO2) is lower than that for TNB and is further aggravated by the formation of endothermic products (NO, HCN,and N20). In sum, the high energyreleased
Amino-2,4,6-trinitrobenzeneExplosives
The Journal of Physical Chemistry, Vol. 97, No. 34, 1993 8757
by TNB results from a different set of propagation reactions than occurs for MATB, DATB, and TATB because of the presence of NH2 groups in the latter. An alternate view of the source of the trend in initiation sensitivity in TNB, MATB, DATB, and TATB is based on the role of intermolecular hydrogen bonding. Hydrogen bonding was proposed to absorb shock energy and prevent it from localizing in the internal bonds of the molecule.14 The effect of hydrogen bonding differences is difficult to judge kinetically but could possibly be accommodated in theories of how mechanical energy couples to chemical processes.53-57 The cohesive forces definitely increase as NH2 groups are added as evidenced by the trend in melting points in Table 11. Raman spectroscopy during dynamic shock loading also suggests that the N-H.-O bond is compre~sed.~~ While an increase in the amount of hydrogen bonding in this series matches the decrease in shock and impact sensitivity, it should be noted that the presence of hydrogen bonding does not guarantee low impact and shock sensitivity.59 This factor can be overridden by the addition of reactive energetic sites.59 Crucial to any argument of how hydrogen bonding affects sensitivity is the need to account for its selective role at defects in the crystal as opposed to its shock absorbing potential in the homogeneous bulk crystal. This might be analyzed by molecular dynamics calculations or the Dlott-Fayer
VI. Conclusions It is widely accepted that thermal decomposition pathways are induced by shock and impact stimulus and that this chemistry plays a role in the trends of sensitivity. The frequently cited trends in the relationships among the properties of TNB, MATB, DATB, and TATB obscures the fact that major differences exist in the thermal decomposition chemistry along this series. Reconciliation of these differences helps identify the molecular sources of the trend in impact and shock sensitivity. The unifying unimolecular initial chemistry appears to involve C-NOz homolysis. Cyclization reactions of MATB, DATB, and TATB that produce furazans and furoxans are lower activation energy processes that can be overstepped by the rapid, large temperature rise during shock and impact stimulation. Unfortunately, the scatter and lack of appropriate Arrhenius data cloud the precise assessment of the isokinetic temperature. The influence of the intermediate propagation reactions is felt immediately. The relative percentages of the gaseous products from TNB, MATB, DATB, and TATB suggest that the decrease in the heat of explosion along this series is largely tied to the decrease in the amount of exothermic products and the increase in the amount of endothermic products in these steps.
VII. Experimental Section Samples of TNB, MATB, DATB, and TATB were obtained from the Wright Laboratory, Armament Directorate, Eglin AFB, FL, and were used without further treatment. T-Jump/FTIR spectroscopy has been described in detail elsewhere.52 Approximately 200 r g of sample was thinly spread on a Pt ribbon filament which was housed in a gas-tight cell so that the beam of a Nicolet 20SX rapid-scan FTIR spectrometer passed 2-3 mm above the surface. The atmosphere in the cell was Argas adjusted to the desired pressure. The Pt filament was activated by a high-gain, fast-response control circuit, the control voltage of which was recorded throughout the experiment. A heating rate of 2000 OC/s to the chosen constant temperature was sought. Endothermic and exothermic events of the sample were detected as positive and negative excusions, respectively, of the control voltage difference trace. The difference trace was produced by subtracting thecontrolvoltage with nosample present from the control voltage with sample present. Correlated in time with the control voltage trace were the IR spectraofthegasproductsrecordedat 10scans/s. Theabsorbance
values of the individual products were converted to relative concentrations through the absolute absorbance by using a procedure described e1sewhere.a The relative rate of growth of the products through the exotherm and the final concentrations after all of the sample decomposed are shown in Figures 2-5. The absolute absorbance of the NCO stretch of HNCO is unknown, and so it was assumed to be similar to the structually and electronically related COZand N20 molecules. A small amount of HzO was observed in all cases but was not quantified.
Acknowledgment. We are grateful to Drs. R. L. McKenney, Jr., and P. R. Bolduc of the Wright Laboratory, Armament Directorate, Eglin AFB, FL, for samples of compounds and financial support of this work. Discussions with C. B. Storm and J. P. Ritchie were most helpful. References and Notes (1) Field, J. E.; Bourne, N. K.; Palmer, S. J. P.; Walley, S.M. Proc. R. SOC.London A 1992, 339, 269. Field, J. E. Acc. Chem. Res. 1992, 25,489. (2) Lee, E. L.; Sanborn, R. H.; Stromberg, H. D. Fifth Symp. (In?.) Detonation; Office of Naval Research: Washington, DC, 1970; p 331. (3) Brill, T. B.; James, K. J. Chem. Rev., in press. (4) Robertson, A. J. B.; Yoffe, A. D. Nature 1948, 161, 806. (5) Wenograd, J. Trans. Faraday SOC.1961, 57, 1612. (6) Bulusu, S.; Autera, J. R. J . Energ. Mater. 1983, 1. 133. (7) Brill, T. B.; James, K. J. J . Phys. Chem., following article in this issue. (8) Storm, C. B.; Stine, J. R.; Kramer, J. F. In Chemistry and Physics of Energetic Materials;Bulusu, S.,Ed.; Kluwer Academic Publ.: Dordrecht, The Netherlands, 1990, p 605. (9) Cook, M. A.; Horsley, G. S.; Partridge, W. S.; Ursenbach, W. 0.J . Chem. Phys. 1956, 24,60. (10) Cook, M. A,; Mayfield, E. B.; Partridge, W. S. J . Chem. Phys. 1975, 59, 675. (11) Zeman, S. Thermochim. Acta 1980, 41, 199. (12) Zeman, S.; Dimun, M.; Truchlik, S. Thermochim. Acta 1984, 78, 181. (13) Kamlet, M. J.; Adolph, H. G. Propellants Explos. 1979, 4, 30. (14) Rogers, J. W., Jr.;Peebles,H. C.;Rye,R. R.;Houston, J. E.;Binkley, J. S. J. Chem. Phys. 1984,80, 4513. (15) Choi, C. S.; Abel, J. S. Acta Crystallogr. 1972, 823, 193. (16) Holden, J. R.; Dickenson, C.; Bock, C. M. J . Phys. Chem. 1972,24, 3597. (17) Holden, J. R. Acta Crystallogr. 1967, 22, 545. (18) Cady, H. H.; Larson, A. C. Acta Crystallogr. 1965, 18, 485. (19) Odiot, S. In Chemistry and Physics of Energetic Materials; Bulusu, S., Ed.; Kluwer Academic, Publ.: Dordrecht, The Netherlands, 1990; p 79. (20) Owens, F. J. J . Mol. Srruct. 1985, 121, 212. (21) Owens, F. J.; Jayasuriya, K.; Abrahmsen, L.; Politzer, P.Chem.Phys. Lett. 1985, 116, 434. (22) Murray, J. S.; Lane, P.; Politzer, P.; Bolduc, P. R. Chem. Phys. Lett. 1990, 168, 135. (23) Sharma, J.; Beard, B. C.; Chaykowsky, M. J . Phys. Chem. 1991,95, 1209. (24) Liteanu, C.; Rica, I. Statistical Theory and Methodology of Trace Analysis; Ellis Horwwd Publ.: Chichester, U.K., 1980; p 45. (25) Sharma, J.; Beard, B. C. In Chemistry and Physics of Energetic Materials, Bulusu, S., Ed.; Kluwer Academic Publ.: Dordrecht, The Netherlands, 1990; p 569. (26) Evans, M. G.; Polanyi, M. Trans. Faraday SOC.1938,34, 11. (27) Matveev, V. G.; Dubikhin, V. V.; Nazin, G. B. Izv. Akad. Nauk SSSR, Ser. Khim. 1978, 675. (28) Gonzalez, A. C.; Larson, C. W.; McMillen, D. S.; Golden, D. M. J . Phys. Chem. 1985, 89,4809. (29) Maksimov, Y. Y.; Sorochkin, S.B.; Titov, S. V. Tr. Inst.--Mosk. Khim.-Tekhnol. Inst. im. D. I . Mendeleeua 1980, 112, 26. (30) Tsang, W.; Robaugh, D.; Mallard, W. G. J. Phys. Chem. 1986,90, 5968. (31) Smith, R. E. Trans. Faraday SOC.1940, 36,985. (32) Maksimov, Y. Y. Symposium on Theory of Explosives; Oborongiz: Moscow, 1963; p 338. (33) Meyerson, S.; Vander Haar, R. W.; Fields, E. K. J . Org. Chem. 1972, 37, 4114. (34) Green, A. G.; Rowe, F. M. J . Chem. Soc. 1912, 101, 2443. (35) Green, A. G.; Rowe, F. M.J. Chem. SOC.1912, 101,2452. (36) Preston, P. N.; Tennant, G. Chem. Rev. 1972, 72, 627. (37) Khmelnitsky, L. I.; Novikov, S. S.; Godovikova, T. I. Chemistry of Furoxans; Nauka: Moscow, 1981. (38) Rogers, R. N.; Janney, J. L.; Ebinger, M. H. Thermochim. Acta 1982, 59, 287. (39) Sharma, J.; Forbes, J. W.; Coffey, C. S.; Liddiard, T. P. J. Phys. Chem. 1987, 91, 5139. (40) Sharma, J.; Beard, B. C.; Forbes, J.; Coffey, C. S.; Boyle, V. M. Ninth Symp. (Int.)Deron., Office of Naval Res.: Arlington, VA, 1989; Vol. 11, p 897.
8758 The Journal of Physical Chemistry, Vol. 97, No. 34, 1993 Maksimov, Y. Y . Russ. J. Phys. Chem. 1972,46,990. Maksimov, Y.Y.; Kogut, E. N. Russ.J. Phys. Chem. 1978,52,805. Zeman, S. Thermochim. Acra 1980, 39, 117. Rogers, R. N. Thermochim. Acra 1975,11, 131. Maksimov, Y. Y . Russ. J. Phys. Chem. 1%7,40,635. Farber, M.; Srivastava, R. D. Combusr. Flame 1981,42, 165. Stolovy, A.; Jones, E. C., Jr.; Aviles, J. B., Jr.; Namenson, A. I.; Fraacr, W. A. J. Chem. Phys. 1983, 78,229. (48) Brill, T. B.; Brush, P. J. Proc. R. Soc. London A, 1992, 339, 377. (49) Brill, T. B.; Brush, P. J.; Patil, D. G.;Chen, J. K. Twenty-Fourth Symp. (Inr.) Combustion;The Combustion Institute: Pittsburgh, PA, 1992; p 1907. (50) Brill, T. B.;Brush,P. J.; Patil, D. G. Combur. Flame 1993,92,178. (41) (42) (43) (44) (45) (46) (47)
Brill and James (51) Brill, T. B.; Brush, P. J.; Pati], D. G. Combusr.Flame., 1993,94,70. (52) Brill, T. B.; Brush, P. J.; James, K. J.; Shepherd, J. E.; Pfeiffer, K. J. Appl. Specrrosc. 1992,46,900. (53) Coffey, C. S.;Toton, E. T. 1. Chem. Phys. 1982, 76, 949. (54) Trevino, S. F.; Tsai, D. H. J. Chem. Phys. 1984, 81, 348. (55) ZeriUi, F. J.; Toton, E. T. Phys. Rev. B 1984, 29, 5891. (56) Walker, F. E. J. Appl. Phys. 1988,63, 5548. (57) Dlott, D. D.; Fayer, M. D. J. Chem. Phys. 1990, 92, 3798. (58) Trott, W. M.; Renlund, A. M. J. Phys. Chem. 1988, 92, 5921. (59) Oyumi,Y.;Rheingold,A. L.;Brill,T. B.PropellantsExplos.Pyrotech. 1987, 12, 46. (60) Brill, T. B. Prog. Energy Combust. Sci. 1991,18, 91.
