Beruflich Dokumente
Kultur Dokumente
PHYSICAL CHEMISTRY
Registered in U.S. Patent Oflice @ Copyright, 1967, by the American Chemical Society
RDX, HMX, PETN, TNT, and composition B were burned under helium in a closed bomb
a t initial pressures of from 1 to 120 atm and the combustion products were analyzed. At
low initial pressures (below 10 atm), burning was erratic and it appears that the products
were formed in part, a t least, a t a relatively low temperature. The products formed a t
high pressures consisted primarily of Nz and the species involved in the water-gas reaction,
except in the case of TNT, where significant amounts of carbon were formed. The high-
pressure data are interpreted in terms of a mechanism which involves the freezing out
of the water-gas equilibrium in the rapidly cooling combustion products. This freeze-out
process is discussed analytically and certain numerical results obtained with a high-speed
computer also are presented. Simple procedures for estimating freeze-out temperatures
and final product compositions, valid over a wide range of conditions, are given and are
used to estimate cooling rates in these and certain related experiments.
2381
2382 RICHARD A N D LOUISC. SMITH
R. BERNECKER
a heavy, stainless steel sample cup and sample support gas other than water was computed as though no water
cage.2 An electrically heated Pt wire was used to were present.
ignite the samples. The explosives were of standard For those explosives expected to produce only vola-
ordnance manufacture and quality. The powdered tile products, a second set of data mas obtained by
samples were obtained from stock used as standards adding sufficient water to the water-free data to give
for the drop-weight impact test at Los Alamos. The the proper O/C or, in a few cases, O/N ratio. Justi-
high-density samples were prepared by pressing the fication for this procedure is that in most such cases
explosive (HE) in a mold to the desired density and it does produce an excellent material balance, in-
then machining the pressed pellets to the desired cy- cluding the correct value for the independent H/C
lindrical shape (0.200-in. diameter X 0.900-in. length). (or H/N) ratio.
The platinum ignition wire was 18 mils in diameter
Results
and was wound in the shape of a helix. For the
powders, the helix was l/s-in. in diameter and was Cyclotrimethylenetrinitramine ( R D X ) and P-Cyclo-
partially buried in the powdered HE; for the pressed tetramethylenetetranitramine ( P - H M X ) . The gaseous
charges, the diameter of the helix was just large enough products from the burning of R D X were studied at
to allow the cylindrical charge to be slipped inside. various pressures and HE densities, while the products
The powdered samples were placed directly in the from 0-HMX were obtained only on the powdered ma-
sample cup, while the pressed cylinders were suspended terial at several pressures. Table I contains the ex-
just above the bottom of the cup by the Pt helix. perimentally determined (water-free) compositions
After the bomb was assembled, it was evacuated, and the corresponding material balances and also
filled with helium, and connected to a low-voltage, the compositions and material balances obtained after
high-amperage ignition system powered by several correcting the data for water, as already described.
6-v batteries. Since the ignition circuit was not a con- For samples in which >99% decomposition occurred,
stant-resistance network, the start of burning could be the calculated material balances match the molecu-
observed by a sudden drop in current. Current was lar formula quite well. Infrared analyses of the few
applied to the sample only long enough to initiate milligrams of material left in the sample holder in
burning. the high-pressure runs indicated that it was pure
The gases were sampled by allowing the bomb prod- starting material for the R D X experiments and a mix-
ucts to expand into an evacuated 10-ml high-pressure ture of the HMX polymorphs for the HMX experi-
bomb. The contents of the 10-ml bomb were ex- ments.
panded, in turn, into several 1-1. glass bulbs for analy- The effect of explosive density on gas composition
sis. The gas samples were analyzed on a Consoli- was briefly investigated for RDX. Powdered samples
dated Electrodynamics Corp. RIodel 21-103 mass burned a t 359, 640, and 1740 psi and a pressed sample
spectrometer. Analysis by mass spectrometry is (1.62 g/cc) burned at 646 psi show nearly identical
expected to give routinely good results for CO, C02, gas compositions. Two pressed charges (1.755 g/cc,
Nz, H2, N20, S O , and CH4, but less accurate results 96.5% of crystal density) burned at approximately
for some of the other products. NO2 does not produce 1780 psi show slightly different compositions.
a parent peak in our mass spectrometer, but is readily At lower pressures, combustion of the samples is less
detected by its color. Xone was seen in any of the complete and the composition of the product gases
sample bulbs. Water, a major product, is very dif- changes noticeably. This is consistent with the ob-
ficult to determine quantitatively in a routine mass servations of Andreev and P l y a ~ u n o v ,who
~ reported
spectral analysis. Water was found in the analyses, that at pressures of a few atmospheres intermediate
but the amount detected depended upon the condi- products of combustion (such as the nitrogen oxides)
tions of the experiment and the method of sampling tend to be formed, whereas at high pressures only the
(liquid water was frequently present in the bomb when products of complete conversion are found. Their
it was opened). Similar difficulties are encountered data for the combustion of RDX under 1 atm of Nz
with S H a , HCK, and HzCO; however, in our experi- also are given in Table I. A comparison of the data
ments these were present in minor amounts and repre- from our low-pressure runs with their data shows reason-
sent a relatively small source of error. For these able agreement, the main discrepancy being in the
reasons we have presented the analytical data for many amount of Nz found. This is reflected in their material
of the samples in two ways. In the first method, used
(2) Described in Spec. No. 1100, Parr Instrument Go., Moline, Ill.
for all the samples, the analyses are reported on a
(3) K. K. Andreev and hl. S. Plyasunov, Z h . Vses. Khim. Obsh-
water-free basis. That is, the mole per cent of each chestza im. D . I . Mendeleeta, 8, 586 (1963).
balance for Nz and may represent an analytical dif- some recovered samples, it appeared that substantial
ficulty causcd by the large amount of Nz initially pres- melting occurred before the sample ignited. In one
ent in their experiments. unsuccessful run in which ignition did not occur, the
The HMX data confirm the trend observed with upper part of the sample had melted and was fused onto
RDX, but here we have, in addition, two cases of very the platinum wire, while the lower part of the sample
incomplete combustion, accompanied by drastic changes was in its initial state.
in composition. Evidently either ignition or burning The data in Table I show the gas compositions from
becomes erratic at sufficiently low pressure under the the burning of T N T over the pressure range of 21-
conditions of our experiments and it is quite possible 1811 psi. The high-pressure combustion experi-
that these composition changes are the accidental ments show essentially constant compositions and poor
result of a mixed burning-thermal decomposition proc- material balances, especially for carbon. The latter
ess rather than :L direct effect of pressure per se. is to be expected, as a fluffy deposit of carbon was
Evidence supporting this interpretation is contained in found in the bomb. We have, therefore, reported
the data of Robertson4 (Table I), which show the these results on the basis 0 = 6 rather than C = 7.
thermal decomposition gases obtained from molten A low-pressure (21-psi) combustion run was made to
H M X during a 2-min reaction at 280. Large amounts compare our results with those reported by Andreev and
of NO and NzO are common in both experiments; how- Plyasunov13 whose data are shown for comparison.
ever, the presence of equally large amounts of HCN in The agreement is good, although the Russians have
our sample, but not in Robertsons, shows that the two a slightly poorer nitrogen balance. This 21-psi run
results are not strictly comparable. gave only a partial burn (<50% decomposition of
Pentaerythritol Tetranitrate ( P E T N ) . Data on the TNT) and thus the material balances are not expected
combustion products of P E T N have not been re- to be good. It is evident from the material balances
ported in the literature and hence a more thorough that there are serious shortages in carbon and hy-
study of the effect of pressure was undertaken on drogen-more serious at the lower pressures than a t the
powdered samples. The gas compositions and ma- higher pressures. It seems clear that the decomposi-
terial balances are shown in Table I for a pressure tion produces nonvolatile products other than carbon
range of 19-724 psi. Higher pressures were not studied, under these conditions.
since the differences between the 375 and 724 psi data Composition B-3 (60% RDX :40% T N T ) . Compo-
indicated constant compositions, as already noted sition B-3 is prepared by cooling a slurry of R D X in
for R D X and HMX. The data of Table I include molten T N T and it thus represents an intimate mix-
both the water-free and corrected compositions. These ture of these two materials. The data given in Table
corrected data show that at high pressures Nz,CO, I were obtained on the loose powder and on pressed
COZ,Hz,and HzOare the final products. pellets prepared from the powder. Excellent material
At low pressures we again find incomplete combus- balances are obtained from the corrected gas analysis
tion of the sample, accompanied by a marked change data, which suggests that equilibrium between the
in gas composition. Data on the thermal decom- products of the two components has been reached in
position products of P E T N at 210, obtained by the flame. There is no variation in composition over
Robertson5 and given in the table, again suggest the the 350-1800-psi pressure region or in going from the
possibility of a mixed burning-thermal decomposi- bulk explosive to charges pressed to 98% of the theo-
tion process for our low-pressure experiments, al- retical maximum density. Decomposition of the
though we do not find the large amounts of N 2 0 and samples was nearly complete, although trace amounts
NO2 reported by him. of carbon black and other colored material were found
2,4,6-Trinitrotoluene ( T N T ) . The burning char- in the sample cup after burning.
acteristics of T N T differ from those of the three ex-
plosives already discussed, in part because of its lower Discussion
melting point (81). Attempts to burn pressed T N T
In the higher pressure runs with RDX, HMX,
met with failure. Even though the Pt helix in which
PETN, and composition B, the combustion products
the cylindrical charge was placed covered 90% of the
consist primarily of Nz and the species involved in the
length of the charge, the T N T would merely melt
water-gas reaction given as eq 1.
where the wire contacted it and the sample would
slide through the wire without igniting.
Powdered T N T samples could be burned; however, (4) A. J. B. Robertson, Trans. Faraday Soc., 45, 85 (1949).
from the ignition characteristics and examination of (5) A. J. B. Robertson, J. SOC.Chem. Ind. (London), 67,221 (1948).
5 aa'
m N ro 0
'9
d
?
d
N
'9
d
?
10
h
4
19 a
d
'3
g
a
8
d
V
0
V
i
,
Approx initial
bomb press..
psi Pellets
RDX
Powder
- HMX
powder
PETN
powder
-Corhposition
Pellets
B
Powder
four values.
If K is equated to the equilibrium constant K , the (6) K as a function of T was obtained from F. D. Rossini, et al.,
Selected Values of Physical and Thermodynamic Properties of
temperatures corresponding to the values shown in Hydrocarbons and Related Compounds, A. P. I. Research Project
Table I1 range from about 1750 to 2100K.6 These 44, Carnegie Press, Pittsburgh, Pa., 1953, p 748.
temperatures are much lower than the adiabatic flame (7) H. Behrens, Z. Phyaik. Chem. (Leipzig), 195, 24 (1950). See
also 9.S.Penner, Chemistry Problems in Jet Propulsion, Pergamon
temperatures and presumably represent the tempera- Press, New York, N . Y., 1957, Chapter 23.
where p (uv/xy) (x y) u+ + +v will be assumed Here we have introduced the relaxation time r = - dt/d
constant during the "freeze-out" process. The total In [(uvlxy) - K ] from eq 2 and an Arrhenius expres-
differential with respect to time is d/dt = (a/&) + sion for the dependence of k-1 on T , L-1 = Ae- E / R T
(dT/dt) (d/dT) and, therefore, the time dependence It is clear that eq 5 may fail (and undoubtedly does)
of (uv/zy) - K is given by the integral of the equation long before (d/dt) [(uvlzy) - K ] attains its maximum
value. The point we are interested in here, however,
d(y - K) = -kw1fi(g dtfi
- K ) - dTdK is that eq 5 cannot be true when the inequality expressed
(3)
dt xy in eq 6 is satisfied. That is, eq 6 is a sufficient (but not
This equation cannot be integrated in a form useful necessary) condition for (uv/xy) - K t o increase
for our purposes. We can, however, use it to obtain rapidly, or, in other words, for freeze-out to occur.
a method of estimating freeze-out temperatures via Shock tube studies of the macroscopic kinetics of the
the following argument. reaction 1 have been made by RiIatthews*in a tempera-
For the case we are interested in here, dT/dt < 0, ture and pressure range close to that of interest here.
and, over a limited temperature range, we can assume Neither A nor E was determined with great precision,
-dK/dt E constant > 0. If initially (uv/zy) - K =O, but from a least-squares fit of his data Matthews
then (dldt) [(uvlxy) - K ] 2 0 for all t , and (uv/xy) - summarized his results for kl in the form
K is a positive, nondecreasing, continuous function of kl = 3.6 X 1011e --59,40O/RT 1. sec-~
t for all t > 0. It follows that (uvlxy) - K can re-
main small only if Matthews' measurements were made in the tempera-
ture range 2180-2580"K1 whereas we are interested
(4) primarily in the rate of the reverse reaction a t tempera-
tures around 1800K. From IC-, = Ll/K we have
(except., possibly, for momentary fluctuations) or, from = En, - AH
(3) and our assumptions on K(t), if
and
- K) A ki
k-d3(: g dt constant (5) log - = log K
Ak-1
+-AH
2.3RT
Now, k-1 decreases as t increases, and eq 5 thus defines
where AH is the heat of reaction. At 1800"K, AH =
the rate at which (uv/xy) - K must increase if (4)is
6.5 kcal/mole and K = 3.8@and hence
to hold. In particular, by differentiating eq 5 we ob-
tain
Using this result in eq 6 with T = 1800"K, we obtain
-dt/dT = RT"ldt! (6) (8) J. C. Matthews, "A Shock Tube Study of the Reaction Between
d l n k-lp/dT E dT H2 and COO," Dissertation, Washington University, 1965.
-E 62,400----
Tr
(eq 8) K ( Tr) (UV/ZY) f
1 x 103 1880 4.21 4.03 2230 5.44 5.22 2560 6.28 6.11
3 x 103 1750 3.69 3.50 2060 4.88 4.70 2380 5.86 5.67
1 x 104 1620 3.14 2.96 1910 4.33 4.15 2200 5.35 5.17
3 x 104 1510 2.67 2.54 1790 3.85 3.68 2060 4.88 4.72
1 x 105 1410 2.25 2.13 1670 3.35 3.20 1930 4.41 4.23
3 x 105 1330 1.92 1.81 1570 2.93 2.81 1820 3.97 3.81
1 x 106 1250 1.61 1.52 1480 2.54 2.43 1710 3.52 3.38
3 x 106 1180 1.34 1.28 1400 2.21 2.12 1620 3.14 3.02
1 x 10' 1120 1.13 1.07 1330 1.92 1.82 1540 2.80 2.66
3 x 10' 1070 0.970 0.905 1270 1.68 1.59 1460 2.46 2.37
1 x 108 1010 0.786 0.751 1200 1.42 1.36 1390 2.16 2.08
3 x 108 970 0.673 0.633 1150 1.24 1.18 1330 1.92 1.85
1 x 109 920 0.543 0.523 1100 1.07 1.01 1270 1.68 1.62
3 x 108 890 0.471 0.439 1060 0.938 0.878 1220 1.49 1.43
1 x 10'0 850 0.383 0.360 1010 0.786 0.750 1170 1.31 1.25
peratures, they give no information concerning the and the error can be reduced considerably (to less than
final value of uv/xy, which is, of course, the quantity 3%) if 0.955 K(Tf) is used instead of K(Tf). This is
ultimately determined in the experiments. There- remarkable in view of the wide range of parameters
fore, eq 3 in the equivalent form (eq 9) was integrated represented in Table 111.
numerically on a computer for a range of values of It remains to discuss the data of Table I1 in terms of
the process believed to be occurring in the bomb, which,
for our high-pressure runs, can be described as follows.
The l-g sample of explosive, contained in or suspended
01 and E and using the least-squares polynomial fit just above the sample cup, burns in a time of the order
of Rossini's data to express the dependence of K on T of a few tenths of a s e ~ o n d . The
~ volume of gas
evolved, at the adiabatic flame temperature and a
K = 7.9395 x - 4.8006 x I O - 6 ~ 2 +
10-427 total pressure of 100 atm, is about 100 cc. The hot
8.1576 X 10-9T3 - 4.2506 X 10-12T4 + combustion products are propelled through the support
cage into the cold helium. In the initial stages of their
9.3176 X 10-"T6 - 7.4736 X 10-20T6 motion, they do not mix with the helium and hence
remain hot. At some point, a rapid, turbulent mixing
The results of the calculations are given in Table 111,
occurs, accompanied by high cooling rates and freeze-
together with the freeze-out temperatures obtained
out of the water-gas equilibrium.
from eq 8 and the corresponding values of K.
To estimate the cooling rate, we must first assign a
Two methods of estimating the final value of uv/xy
value to 8. For a run in which the initial pressure of
for a given 01 and E can be obtained from these results.
helium is 100 atm, the final pressure, if no heat is lost
First, the computer results can be summarized in the
to the walls or metal parts of the bomb, will be about
convenient form
400 atm. If we take 200 atm as the average pressure
and 1900K as the approximate freeze-out tempera-
ture and assume that the products have mixed with an
equal volume of helium at the time freeze-out occurs,
This reproduces the computer results to within 15% then g 1.4. For E = 52,900 cal/mole, a value of 4
for the entire range of 01's and E's given in the table and for (uu/xy)f corresponds, from eq 10, to an CY of 8.9 X
is good to better than 10% in all but a few cases. lo3. If A = 1.5 X 1010, then B = / ~ A / c= Y 2.4 X
Alternatively, one can calculate the freeze-out tem- lo6deg/sec. Tf (from eq 8) is 1900K.
perature Tf from eq 8 and use the corresponding value This result is not very sensitive to errors in A and E
of K as the estimate of (uv/zy)r. The error in the esti-
mate obtained in this way will be less than 7%. All the (9) Estimated from burning rate data given by A. P. Glaskova and
deviations are positive, however [K(Tt) > (uv/zy)fl, I. A. Tereshkin, Russ. J. Phys. Chem., 3 5 , 795 (1961).
within the range permitted by Matthews data. From Thus, the cooling rates for these three experiments
his data we calculate a 95% confidence limit for E of can be summarized (Table V). We view those results
*21,100 cal/mole.1 However, the 95% confidence as being qualitatively consistent.
limit of log kl a t the midpoint (in 1/T) of his measure-
ments is only It0.08. Thus, unless there is a large
systematic error in his rate measurements, errors Table V
in E must be accompanied by compensating errors in
Estimated
A . In Table IV are shown the values of A / a and Tf cooling
computed from eq 10 and 8 for (uv/zy)f = 4 and with Heat Estimated Tr, rate,
transfer turbulence O K deg/sec
Thus, while the mixing-cooling process in our experi- of decomposition in those cases in which the products
ments cannot be sufficiently well defined to provide a are formed above the freeze-out temperature. At
quantitative test of the details of the analysis, the sufficiently low pressures, the products of burning may
procedure described does lead to a qualitatively con- be formed below t,hat temperature and under these
sistent interpretation of these and other related experi- conditions they may be characteristic of that mode of
ments. The technique would appear to be applicable decomposition.
to other reactions and other systems as well and should Acknowledgments. The authors gratefully acknowl-
lead to quantitatively useful results under more favor- edge the assistance of Dr. Howard H. Cady, who per-
able circumstances. formed the numerical integrations on the Los Alamos
Finally, it is obvious that the product composition computers. This work was performed under the aus-
cannot be used to distinguish between different modes pices of the U. S. Atomic Energy Commission.