J . Phys. Chem. 1985,89, 2309-2315
In Figure 5 we show the results calculated for H6Si207plotted in the same way together with experimental data for the coesite?* a - q ~ a r t zand , ~ ~a-cristoba1ite"O forms of SiOz. The agreement is very nearly as good as for the phosphates. These results show that we can use calculations at this level on molecular fragments to predict structural features of solids (such as bond length-bond angle correlations) with high precision. Conclusions We have presented optimum or near-optimum geometries for 13 phosphate or silicate species at the 6-31G* level. These should provide valuable guides to expected structures in related molecules and crystals. Comparison with available experimental data shows a generally excellent agreement, although, in order to reproduce observed T-0-T angles, polarization functions must be added to bridging oxygen atoms. Calculated and observed (in crystals) bond length-bond angle correlations are also in good agreement and point clearly to the existence of significant off-diagonal stretch-bend force constants. (38) G.V. Gibbs, C. T. Prewitt, and K. J. Baldwin, Z . Kristallogr., 145, 108 (1977). (39) Y. LePage and G. Donnay, Acta Crystallogr., Sect. B, 32, 2456 (1976). (40) D. R. Peacor, Z . Kristallogr., 138, 274, (1973). (41) W. H. Baur, Trans. Am. Crystallogr. Assoc., 6, 129 (1970). (42) J. A. Tossell and G. V. Gibbs, Phys. Chem. Mineral, 2, 21 (1977). (43) M. S.Gordon, J. A. Boatz, and M. W. Schmidt, J . Phys. Chem., 88, 2998 (1984); M. W. Schmidt, S. Yabushita, and M. S.Gordon, J . Phys. Chem., 88, 382 (1984).
2309
Although these are often omitted in structure simulations and in lattice dynamics, this practice is clearly not advisable. The existence of stretch-bend force constants cannot be unequivocably ascribed to any one cause (nor should they). For example, in the T-0-T force field, the linear dependence of d ( T - 0 ) on f, sugg e s t ~ ~an' interpretation in terms of sp hybridization arguments. However, we also note the large (and relatively independent of basis set) value of n(T.-T) in Table I1 that suggests that direct T-T interacitons may also be a contributing factor. Similar considerations apply to the O - T a force field: the intratetrahedral 0-0 populations are of similar magnitude and significantly larger in phosphates (in which the 0-0 distances are shorter) than in silicates. Heats of reactions in which charge separation is not involved in solution are closely paralleled by energy differences calculated for isolated molecules, suggesting an approximate conservation of hydration energy for reactants and products, and suggesting also that, where experimental data are not available (as for hydrolysis of the monomeric metaphosphae ion), SCF calculations will provide useful estimates. Acknowledgment. This research was supported by Grants DMR 81 19061 (MOK) and EAR 82 18743 (GVG) from the National Science Foundation. Registry NO. PO:-, 14265-44-2; H4P207, 2466-09-3; H6Si207, 20638-18-0; H4Si04,10193-36-9; H3P04,7664-38-2; HP03, 10343-62-1; H4P04+, 26902-99-8; PO3-, 15389-19-2; H,PO,, 14939-10-7; PF,, 7647-19-0; H6P2072+,95864-34-9; H2P04-, 14066-20-7; H,SiO,-, 18102-72-2.
Thermal Decomposition of Energetlc Materials. 2.+ The Thermolysis of NO3- and C104Salts of the Pentaerythrityltetrammonium Ion, C(CH,NH,).,"+, by Rapid-Scan FTIR Spectroscopy. The Crystal Structure of [C(CH2NH3)4](N03)4 Y. Oyurni, T. B. Brill,* A. L. Rheingold, Department of Chemistry, University of Delaware, Newark, Delaware 19716
and C. Lowe-Ma Chemistry Division, Research Department, Naval Weapons Center, China Lake, California 93555 (Received: November 22, 1984)
Details of the pyrolysis mechanism of pentaerythrityltetrammonium nitrate, [C(CH2NH3),](N03)4,PTTN, were determined by rapid-scan FTIR spectroscopy (-ten interferograms s-l). The initial gas product at about 490 K is desorbed HN03 resulting, most likely, from the transfer of a proton from the cation to the anion. C-N bond fission also takes place liberating NH3 which reacts with HN03(g) to produce ",NO3@). HN03(g) is rapidly consumed while oxidized fragments of the hydrocarbon portion appear. Application of gas pressure (3200 psig, 1.4 MPa) suppresses the formation of NH, that results from C-N bond cleavage but enhances the concentration of products, such as HCN and HNCO, that retain the C-N bond. The sequence of appearance of products suggests that decomposition of the cation begins at the exterior and progresses inward, a fact that is consistent with chemical attack by an external agent, such as HN03. Determination of the thermal decomposition mechanism of [C(CH2NH3),](C1O4), is complicated by the fact that deflagration or detonation takes place. The crystal structure of PTTN at 296 K [tetragonal, P41 (or P43),Z = 4, a = 10.156 (2), c = 14.629 (4) A] aids in interpreting the IR spectrum of the solid. Strong N-H- s.0 hydrogen bonds are responsible for the lattice cohesion. The point symmetry of the cation, S4, is the same as pentaerythritol, C(CH20H),.
Introduction The properties and thermal decomposition of NH4C104(AP) and ",NO, (AN) have been thoroughly investigated because of their deflagration and detonation characteristics. The initial thermolysis reaction in these salts is proton transfer from NH4+ to the anion librating adsorbed and gaseous NH3 and HC104 or 'Paper 1 of this series is ref 10.
0022-3654/85/2089-2309$01 SO10
HN03.3,4 These species then react by radical reactions resulting in products that include N2, H 2 0 , C12, nitrogen oxides, and so forth. (1) Jacobs, P. W. M.; Whitehead, H. M. Chem. Reu. 1969, 69, 551. ( 2 ) Keenan, A. G.;Siegmund, R. F. Q. Reu. Chem. SOC.1969, 23, 430. (3) Bircumshaw, L. L.; Newman, B. H. Proc. R. SOC.London, Ser. A 1955, 227, 228. (4) Rosser, W. A,; Inami, S.H.; Wise, H. J. Phys. Chem. 1963, 67, 1753.
0 1985 American Chemical Society
2310 The Journal of Physical Chemistry, Vol. 89, No. 11, 1985
Much less extensively studied has been the thermal decomposition of alkylammonium derivatives of AP and Of particular interest are alkylammonium salts that mimic the desirable properties of the simple ammonium salt while eliminating undesirable ones. A cation that is, prima facie, similar to NH4+ in terms of its electrostatic exterior is the pentaerythrityltetrammonium ion, C(CH2NH3);’. The resulting nitrate salt, [C(CH2NH3)4](N03)4,pTTN,8 is intriguing because its melting-decomposition point is higher than AN (-490 K vs. 440 K) and it does not undergo the solidsolid phase transitions that AN experiences between room temperature and its melting point? The perchlorate analogue, [C(CH2NH3)4](C104)4,PTTP,8 although excessively shock sensitive, is intrinsically interesting for comparison with the nitrate salt and with AP. In the present project, rapid scanning FTIR spectroscopy was used to follow the sequence of appearance of gas products from thermolysis of PTTN in order to establish the decomposition mechanism under conditions of variable heating rates (5 K min-’ to 170 K s-l) and pressure (0-1000 psig, 0.1-6.9 MPa). Rapid-scan FTIR spectroscopy is highly desirable for thermal decomposition research1*’* when the lifetime of decomposition products is less than 1 s. The thermal decomposition of PTTN is unusual because several key gas products are produced successively on the time scale of the experiment, thereby permitting a qualitative degradation mechanism to be advanced. A similar study of the thermal decomposition of PTTP was unsuccessful because of its very rapid decomposition. The crystal structure of F I T N reported here aids in developing the structurereactivity relationships.
Oyumi et al. TABLE I: Cnstd and Refmment h t a
formula crystal system space group a, A c, A
v. A3
2’
formula weight p(calcd), g cm-, temp, O C crystal dimension, mm radiation
C5H20N801
tetragonal
P4, (or P4,) 10 .156 (2) 14.629 (4) 1508.9 (9) 4 328.3 1.70
23 0.28 X 0.30 X 0.31 graphite-monochromated
Mo Ka (A = 0.71073 A)
diffractometer absorption coefficient, cm-l scan speed, deg/min 26 scan range, deg scan technique data collected weighting factor g“ unique data unique data with F, > 2.50(F0) standard reflections
Nicolet R3
R(I) RF,bRwF;GOFd
0.0226
highest peak, final diff map, e A-’
2
1.34
variable, 4-20 4 I28 I 4 8 w , full profile ih,+k,+l
0.0029 1249 reflections (2646 collected) 1179 reflections 3/97 (no decay observed) 0.0548, 0.0610, 1.360 0.38
The structure was solved in P4, by the direct methods routine which, after difference Fourier recycling, revealed the positions of all non-hydrogen atoms. A difference Fourier map produced after anisotropic refinement revealed the locations of all 20 hydrogen atoms, but these were not refined because of insufficient data. Instead, the hydrogen atoms were added to the refined model as idealized, fixed, and updated contributions based on sp3 geometry and E-H distances of 0.96 A. A staggered rotational relationship of NH3+to the adjacent CH2 group was maintained. The found and calculated H atom positions were identical within the esd’s of the fractional coordinates. Neutral atom scattering factors and anomalous dispersion corrections were taken from ref 13. The true enantiomorph (P4, or P4,) was indeterminate. A noncrystallographically imposed twofold rotation axis passes through C( 1) and is parallel to the z axis. The fractional atomic coordinates for C ( l ) [-0, - l f 2 , z] suggested that symmetry higher than that required for P4, (or P4,) potentially existed. However, the addition of this symmetry produces the Laue symmetry, 4/mmm, a result inconsistent with the photographic and diffraction intensity data. FTIR Studies. The IR spectra were recorded on a Nicolet 60SX FTIR spectrometer employing an HgCdTe detector. The variable-temperature, atmospheric pressure studies were conducted on 2-4 mg of sample with cells and procedures that are described elsewhere,12except that N 2 instead of Ar was used as the atmosphere. The heating rates given are the maximum rates during the run. Toward the end of the heating time increment the heating rate levels off to the final filament temperature. The study of gas products from thermal decomposition under Nz gas pressure was accomplished with a home-built gas-tight cell containing a nichrome alloy filament and 0.5-in. thick ZnSe windows. The pressure was directly measured with a McDaniel gauge. A significant difference in the rate of heat dissipation from the filament occurs as a function of the applied pressure. Thus it was necessary to calibrate the filament temperature at each pressure. This was accomplished by attaching a type K thermocouple to the filament with a small silver clip (1.5 X 1 X 0.2 mm) and recording the thermocouple output on a Tektronix 100-MHz storage oscilloscope. The temperature calibration was confirmed by measuring the melting points of known compounds SOLV
Experimental Section Compounds. [C(CH2NH3)4](S04)2was provided by W. S. Anderson, United Technologies, Chemical Systems Division. Following his recommendation, NO< was exchanged for by soaking (3X) 1 g of the sample on a medium-frit glass filter with 9 M HN03and then washing the residue with distilled HzO. The colorless sample was recrystallized from H20 and was, according to its I R spectrum, sulfate free. The C104- salt was prepared on a 50-mg scale from the sulfate salt by adding an H20 solution of Ba(C104)2. After filtration, the solution was evaporated and the colorless residue recrystallized from H20. PTTN AND PTTP ARE EXPLOSION HAZARDS! The detonation velocity of PTTN (>300 m s-l) is 2.7 times greater than ANG9To our knowledge values for PTTP have not been measured. Structure Determination. Simultaneously and without awareness of the other’s work, the crystal structure of P’M” was determined by C.L.M. and A.L.R. Both efforts produced similar crystal data and structure solution. Preliminary photographic evidence, including a Laue photograph with the X-ray beam parallel to c showing only 4 symmetry, combined with unit cell dimensions obtained from the angular settings of 25 reflections (22’ 6 28 6 26O) indicated the Laue group 4/m. The observed systematic absences in the reflection data (001,l = 4n 1, 2, 3) limited the possible space groups to P4, or P4,. The reflection data (two octants) were processed with a learned profile routine to improve the accuracy of the measurements of weak data. Corrections for Lorentz and polarization effects, but not absorption, were applied to the data. Table I provides the crystal and refinement data for PTTN. The computer programs used are those distributed by the Nicolet Corporation (P3 and SHELXTL).
+
(5) Guillory, W. A,; King, M. J . Phys. Chem. 1969, 73, 4367. (6) Mack, J. L.;Wilmot, G. B. J . Phys. Chem. 1967, 71, 2155. ( 7 ) Miron, Y . J . Hazard. Mat. 1980, 3, 301. ( 8 ) Barnes, C. J.; Matuzko, A. J. US.Dep. Commer., Off. Tech. Sem., AD 1%3, 415,706. Chem. Abstr. 60: 10642~. (9) Anderson, W. S . , personal communication, and citations of work by M. Stinecifer. (10) Brill, T. B.; Karpowicz, R. J.; Haller, T. M.; Rheingold, A. L. J . Phys. Chem. 1984, 88, 4138. (11) Karpowicz, R. J.; Brill, T. B. Combust. Flame 1984, 56, 317. (12) Karpowicz, R. J.; Brill, T. B. J . Phys. Chem. 1983,87,2109.
(1 3) “International Tables for X-Ray Crystallography”; Kynoch Press: Birmingham, England, 1974; Vol. IV, pp 99 and 149.
Thermal Decomposition of Energetic Materials
The Journal of Physical Chemistry, Vol. 89, No. 11, 1985 2311 TABLE II: Atom Coordinates ( X lo') and Temperature Factors
x l(r) atom
07b
",A
--
\I
H2o H2c
Figure 1. An ORTEP plot of €TI" (40% probability ellipsoids) showing the closest 0-H contacts. 0(4)-H2c at 2.09 A is not shown. The translations used for NO3- with N(7a) and N(8a) are 1.000 + x, y , z; with N(7b) 1.000 + y , 1.000 - x, -0.2500 + r; with N(6c) 1.000 + x, 1.000 + y, 2.
at various pressures. The voltage needed to maintain a constant heating rate was found to increase substantially with applied pressure, but the calibration lines parallel one another. The other details of the pyrolysis experiment are the same as those at atmospheric pressure. The temperature accuracy for the high heating rate experiments is 110 K. Rapid-scan spectra (ten scans s-I) were recorded at 4-cm-' resolution. The spectra at room temperature and during slow heating were recorded at 4 cm-' with 32 accumulated scans.
Structure and IR Results Crystal Structure of PTTN. The asymmetric unit of PTTN contains one formula unit, [C(CH2NH3)4](N03)4,depicted in Figure 1. The atomic coordinates are listed in Table 11. Arranged in columns parallel to the screw axis of the unit cell are NO< ions alternating and separated by the -NH3+ groups of the cation (Figure 2). The packing of the NO3- ions suggests that a network of N-H.s.0 hydrogen bonds (vide infra) is largely responsible for the cohesion of the lattice. The conformation of the cation produces S4point group symmetry, which is the highest possible for this ion. Tables I11 and IV give the bond distances and angles, respectively. All of the -NH3+ groups ideally would be coplanar except that the C-C-C angles in the noncrystallographic mirror planes average 103.4 (4)'. This distortion could originate from the need to reduce the contacts among the hydrogen atoms of distal
