Form III of 2,2′,4,4′,6,6′-hexanitro-azobenzene (HNAB-III)

Article abstract of DOI:10.1107/S0108270105000569

The crystal structure of form III of the title compound, HNAB [systematic name: bis(2,4,6-trinitrophenyl)diazene], C₁₂H₄N ₈O₁₂, has finally been solved as a pseudo-merohedral twin (monoclinic space group P2₁, rather than the orthorhombic space group C222₁ suggested by diffraction symmetry) using a dual space recycling method. The significant differences in the room-temperature densities of the three crystalline forms allow examination of molecular differences due to packing arrangements. An interesting relationship with the stilbene analog, HNS, is discussed. Interatomic separations are compared with other explosives and/or nitro-containing compounds.

Full text of DOI:10.1107/S0108270105000569

organic compounds
Acta Crystallographica Section C
Crystal Structure
Communications
satellites. These methods appeared to show very slight split-
ting at high diffraction angles for re¯ections down the short
axis. There was a remarkable similarity with the precession
photographs also taken along the short axis of a crystal of
HNS (2,20,4,40,6,60-hexanitro-trans-1,2-diphenylethene), with
the exception of the h0l re¯ections with odd l, as required by
the c-glide0 in the latter compound. HNS is also known as
ISSN 0108-2701
Form III of 2,20,4,40,6,60-hexanitro-
azobenzene (HNAB-III)
0
0
Ê
2,2 ,4,4 ,6,6 -hexanitrostilbene [P21/c, a = 22.351 A, b =
ꢀ
Ê
Ê
5.572 A, c = 14.668 A and ꢀ = 110.05 (Duke, 1977); Duke
commented on the remarkable structure similarities of his
HNS results to those of the monoclinic form of TNT].
Systematic absences and the symmetry of the reciprocal lattice
indicated the space group to be consistent with orthorhombic
Mark A. Rodriguez,a* Charles F. Campana,b A. David
Rae,c Edward Graebera and Bruno Morosina
aPO Box 5800, MS 1411, Sandia National Laboratories, Albuquerque, New
Mexico 87185-1411, USA, bBruker AXS Inc., 5465 East Cheryl Parkway, Madison,
Wisconsin 53711-5373, USA, and cResearch School of Chemistry, The Australian
National University, Canberra, ACT 0200, Australia
Ê
Ê
Ê
C2221 [a = 15.4015 (8) A, b = 41.471 (18) A, c = 5.5240 (3) A,
3
V = 3528.3 (3) A and Z = 8], but possibly of monoclinic P21 or
Ê
P21/m symmetry. Attempts to solve the structure using this
orthorhombic cell with disordered molecules using various
molecular packing models were unsuccessful.
Correspondence e-mail: marodri@sandia.gov
Received 4 November 2003
Accepted 6 January 2005
Online 12 February 2005
The crystal structure of form III of the title compound,
HNAB [systematic name: bis(2,4,6-trinitrophenyl)diazene],
C12H4N8O12, has ®nally been solved as a pseudo-merohedral
twin (monoclinic space group P21, rather than the ortho-
rhombic space group C2221 suggested by diffraction
symmetry) using a dual space recycling method. The
signi®cant differences in the room-temperature densities of
the three crystalline forms allow examination of molecular
differences due to packing arrangements. An interesting
relationship with the stilbene analog, HNS, is discussed.
Interatomic separations are compared with other explosives
and/or nitro-containing compounds.
The structure of HNAB-III was ®nally solved and re®ned as
a pseudo-merohedral twin (monoclinic space group P21
emulating orthorhombic space group C2221) using a dual
space recycling method (Sheldrick & Gould, 1995). Since Z =
4, space group P21 requires two structurally independent
molecules, labeled A and B below, in the cell. The structure
was also re®ned independently by one of us (ADR) using
RAELS2000 (Rae, 2000), with essentially identical results. The
SHELXTL (Bruker, 1998) results are principally given in the
present report, with the RAELS2000 results also indicated
where appropriate. The twin ratio for re¯ections h, k, l and h,
Comment
2,20,4,40,6,60-Hexanitroazobenzene (HNAB) is employed as an
explosive and is extremely complex in its crystallographic
behavior, as evidenced by the relationships among the various
phases and their transformations. There are at least ®ve
crystallographic polymorphs of HNAB. Two of these struc-
tures, stable at ambient conditions, were solved over 30 years
ago (HNAB-I and -II; Graeber & Morosin, 1974), while the
third, the subject of the present study, eluded sporadic struc-
ture solution attempts over that time frame. The ®ve crystal-
line polymorphs were initially identi®ed by hot-stage optical
microscopy (McCrone, 1967). Forms I and III are stable from
room temperature to 458 K, and form II from room
temperature to 478 K; above these temperatures, solid±solid
transformations begin to occur. The very unstable forms IV
and V are formed only from the melt on supercooling during
recrystallization. Forms II and III also transform rapidly
through the solution phase into form I. Bulk samples of
HNAB, regardless of the polymorphic forms present, melt at
496 K.
h
k, l related by a twofold axis parallel to the c axis, was
re®ned to 0.551 (1)/0.449 (1) using the program RAELS2000
and can be compared with 0.552/0.448 determined by
SHELXTL. Re¯ections with even h + k values were
approximately three times more intense than the others.
Historically, precession and Weissenberg photography were
used to determine preliminary unit-cell dimensions and space-
group extinctions, and to inspect the diffraction images for
Figure 1
The atom-labeling scheme used for molecule A of HNAB-III. Displace-
ment ellipsoids are shown at the 50% probability level.
Acta Cryst. (2005). C61, o127±o130
DOI: 10.1107/S0108270105000569
# 2005 International Union of Crystallography o127

organic compounds
Re¯ections were monitored according to the parity of h and k,
and showed a consistently good ®t over all such classes.
Fig. 1 shows our labeling scheme for molecule A; that for
molecule B is identical. In this regard, even the displacement
ellipsoids of the nitro groups of the two molecules appear very
similar; in fact, the absolute structures of the two independent
molecules are inverted with respect to one another. The
shorter intermolecular contacts include, but are not limited to,
possible hydrogen-bond interactions, with HÁ Á ÁO distances of
Ê
2.462, 2.491 and 2.516 A, CÁ Á ÁO contacts of 2.954, 2.969 and
Ê
Ê
2.979 A, and NÁ Á ÁO contacts of 2.919, 2.935 and 2.969 A.
Fig. 2(a) shows the packing of the molecules as viewed
along the b axis. The a and c axes are interchanged relative to
those given above, and the cell contents are shifted from the
normal placement of the 21 axes by a/4, so that the remarkable
similarity (noted earlier) with the corresponding projection
Â
for HNS can be shown (Fig. 2b; Gerard & Hardy, 1988). Note
that HNS belongs to P21/c and each of its molecules resides on
an inversion center. Examination of the dihedral angles for the
HNAB molecules shows signi®cant differences between not
only those molecules in HNS but also that for HNAB-I. In
HNAB-III, the phenyl planes form smaller angles with respect
to the linkage (CÐNÐNÐC) planes ['l = 27.7 (3), 32.0 (3),
50.8 (3) and 57.4 (3)ꢀ] and the phenyl planes are twisted with
respect to one another ['p = 82.7 (3) and 84.9 (3)ꢀ for A and
B], while in HNS, the phenyl planes form larger angles with
respect to the linkage (CÐCÐCÐC) planes ['l = 67.1 (4) and
72.0 (4)ꢀ] and the phenyl planes are in parallel alignment as
required by the inversion center ('p = 0ꢀ). Interestingly,
HNAB-I also shows this parallel disposition for the phenyl
planes ('p = 0ꢀ, with a phenyl±linkage angle of 'l = 43.2 (4)ꢀ],
while in HNAB-II, the phenyl planes are twisted, but to a
smaller degree than in HNAB-III ['p = 81.1 (5)ꢀ, with phenyl±
linkage angles of 'l = 48.0 (5) and 51.3 (5)ꢀ]. Table 1
summarizes the angles between the appropriate planes formed
by various entities of the molecules in these crystal structures.
HNS shows pseudosymmetry (Fig. 2b), and imposing a
1
4
mirror at y = creates a 1:1 disordered structure of approx-
imate Pnma symmetry in the cell a0 = a + c/2, b0 = b, c0 = c/2, but
with ꢀ0 = 90.87ꢀ. Upon ordering this structure to create the
B-centered cell 2a0, b0, 2c0, equivalent to the real structure, the
symmetry elements associated with the a0 and c0 directions
must be destroyed. However, they suggest a mechanism for
stacking faults and twinning if ꢀ0 had been 90ꢀ. The structure
of HNAB-III has ꢀ0 = 90.00ꢀ within the estimated error for the
cell a0 = a/2 + c, b0 = b, c0 = a/2, but although the structure
looks similar to that of HNS in projection, the centers of non-
1
4
equivalent molecules now differ in y by about . However, a
twofold screw operation parallel to the a axis is the likely twin
operation, as it leaves NO2 groups adjacent to the twin plane
1
at z = unchanged to a ®rst approximation.
4
The most striking differences between the three HNAB
forms are in the CringÐNazo bond lengths, which increase with
the corresponding decrease in the crystal densities (1.795,
1.744 and 1.703 Mg m 3 forms I, II and III). These differences
are not considered signi®cant by the usual statistical criteria
(Hamilton, 1964); however, the increasing values of
Ê
Ê
1.426 (5) A for HNAB-I, 1.430 (5) and 1.438 (5) A for HNAB-
Ê
II, and 1.441 (7), 1.442 (6), 1.447 (6) and 1.457 (7) A for
HNAB-III, together with the increasing phenyl ring twisting
from the parallel con®guration, must contribute toward the
packing contributions responsible for the density differences.
The examination of various speci®c bond lengths led to a
further comparison with other nitro-containing and other
aromatic energetic materials, considering how the electron-
Figure 2
(a) A packing diagram, viewed down the b axis, for HNAB-III. The a and
c axes are interchanged, and the cell contents are shifted from the normal
placement of the 21 axes by a/4 (see Comment). (b) A packing diagram,
viewed down the b axis, for HNS.
ꢁ
o128 Rodriguez et al. C12H4N8O12
Acta Cryst. (2005). C61, o127±o130

organic compounds
Ê
TNA; Holden et al., 1972), 1.32 A (1,3-diamino-2,4,6-trinitro-
Ê
benzene or DATB; Holden, 1967), 1.320 A (1,3,5-triamino-
Ê
2,4,6-trinitrobenzene or TATB; Cady & Larson, 1965), 1.312 A
(2,3,4,6-tetranitroanaline; Dickinson et al., 1966), and 1.311
Ê
and 1.310 A (TATB). The interesting result of this plot is the
relative position of the values for the nitro-containing HNAB
molecules. The CÐN lengths are longer than those for
aromatic amines containing no nitro groups; a possible reason
for the shortened values is the electron-withdrawing proper-
ties of the nitro group. It had previously been noted from the
position of 2-amino-3-methylbenzoic acid that the acid group
also functions in such a capacity. This suggests that the azo
linkage must be an even greater electron-withdrawing group
in order to neutralize the affect of the nitro groups.
withdrawing characteristics of various substituting groups
might impose on the ring bond lengths. The variation of the
CÐCring bond lengths in other related nitro-containing and
aromatic explosive materials, particularly the long values,
were initially attributed to steric effects of the crowded
substituents; other reports have since related this variation to
electronic effects associated with resonance structures, as well
as the quinonoid contribution involving the ring. This led
Holden et al. (1972) to demonstrate an inverse correlation
between the lengths of the ring bonds and the length of the
CÐN bond in various aromatic amine compounds. Fig. 3
shows this relationship and includes the values for the
different HNAB forms, as well as some other recent structure
Ê
determinations. The CÐN values range from 1.310 to 1.466 A.
Ê
[The CÐN value of 1.497 A for an average ring bond of
Ê
1.391 A found in tetryl (Cady, 1967) is an exception to this
linear relationship and is probably due to steric intramolecular
interactions.] In decreasing order the values are 1.457, 1.447,
The benzene ring is also distorted by the increase in the CÐ
C lengths given in the above inverse correlation. This is
achieved in part by the CÐCÐC ring angles increasing at the
location of the corresponding nitro groups. Other uncorre-
lated and minor contributions to the distortion of the ring are
the quinonoid and resonance shortening of the other CÐC
lengths (details available from authors).
Ê
1.442 and 1.441 A (HNAB-III), 1.438 A (HNAB-II), 1.433 A
Ê
Ê
Ê
(azotoluene; Brown, 1966a), 1.434 A (azobenzene; Brown,
Ê
Ê
Ê
1966b), 1.430 A (HNAB-II), 1.426 A (HNAB-I), 1.426 A
(2,4,6-tribromoaniline; Christensen Stromme, 1969),
&
1.412 A (1,3,5-triaminobenzene 1,3,5-trinitrobenzene or TAB±
Ê
Experimental
Ê
TNB; Iwasaki & Saito, 1970), 1.407 A (2,5-dichloroaniline;
Ê
Sakurai et al., 1963), 1.40 A (4-chloroaniline; Palm, 1966),
A sample of hexanitroazobenzene was obtained from the Explosive
Component Division of (our) Sandia National Laboratories; the
compound was prepared by reacting puri®ed picryl chloride with
hydrazine and potassium acetate in ethanol followed by acidi®cation
with hydrochloric acid. The yield of crude hexanitrohydrazobenzene
was subsequently oxidized with nitric acid to give orange crystals of
HNAB. About 70% of the crystals from this preparation were form I
crystals, the balance being form II. Larger crystals of forms I and II
may be obtained by recrystallization from nitromethane at room
temperature, and those of form III from hot nitromethane.
Morphological characteristics for forms I and II were described
previously. Equant crystals of form I consist of rhombohedra with
monoclinic symmetry, and well developed dipyramid {111} faces and
{100} basal pinacoids; form II crystals consist of long obliquely
truncated rods with {110} prism and {001} basal pinacoids; form III
mostly consists of ¯at bladed crystals of apparent orthorhombic
symmetry bound by {100} and {110} prism and {101} basal pinacoids
along the long direction.
Ê
1.392 and 1.391 A for TAB±TNB, 1.386 A (2-chloro-4-nitro-
Ê
Ê
aniline; McPhail & Sim, 1965), 1.371 A (4-nitroanaline; True-
Ê
blood et al., 1961), 1.367 A (2-amino-3-methylbenzoic acid;
Ê
Brown & Marsh, 1963), 1.358 A (2,6-dichloro-4-nitroaniline;
Ê
Hughes & Trotter, 1971), 1.340 A (2,4,6-trinitroanaline or
Crystal data
3
C12H4N8O12
Mr = 452.23
Monoclinic, P21
a = 15.4015 (8) A
Ê
b = 5.5240 (3) A
c = 22.1182 (11) A
ꢀ = 110.367 (2)ꢀ
Ê
V = 1764.13 (16) A
Z = 4
Dx = 1.703 Mg m
Mo Kꢁ radiation
Cell parameters from 770
re¯ections
Ê
ꢂ = 2.8±22.2ꢀ
ꢃ = 0.16 mm
T = 298 (2) K
1
Ê
3
Acicular, orange
0.40 Â 0.10 Â 0.08 mm
Data collection
Bruker SMART CCD area-detector
diffractometer
' and ! scans
Absorption correction: empirical,
using intensity measurements
(SADABS; Sheldrick, 1999)
Tmin = 0.94, Tmax = 0.99
10 081 measured re¯ections
6076 independent re¯ections
4219 re¯ections with I > 2ꢄ(I)
Rint = 0.028
ꢂmax = 26.0ꢀ
h = 19 ! 10
Figure 3
CÐNamine bond lengths versus average adjacent CÐCring bond lengths
(see Comment).
k = 6 ! 6
l = 26 ! 27
ꢁ
Acta Cryst. (2005). C61, o127±o130
Rodriguez et al. C12H4N8O12 o129

organic compounds
Re®nement
Atom N1B was ®xed at y = 1.0. H atoms were constrained (CÐH =
Ê
0.93 A), with their Uiso(H) values set at 1.2Ueq(C). An independent
re®nement (conducted by ADR) used similarly constrained H atoms
Ê
at 0.93 A; however, in this case, these H atoms were given the same
Re®nement on F2
R[F2 > 2ꢄ(F2)] = 0.038
wR(F2) = 0.086
S = 1.02
6076 re¯ections
578 parameters
H-atom parameters constrained
w = 1/[ꢄ2(F2o) + (0.004P)2]
where P = (F2o + 2Fc2)/3
(Á/ꢄ)max = 0.004
3
Ê
Áꢅmax = 0.16 e A
atomic displacement parameters as the atom to which they were
attached. In the re®nement of ADR, the origin along y was ®xed by
setting the mean y coordinate of the two pairs of central N atoms to
y = 0.625.
3
Ê
0.17 e A
Áꢅmin
Extinction correction: SHELXTL
=
Extinction coef®cient: 0.0028 (6)
Table 1
Data collection: SMART (Bruker, 1997); cell re®nement: SMART;
data reduction: SAINT (Bruker, 1997) and SADABS (Sheldrick,
1999); program(s) used to solve structure: XM in SHELXTL
(Bruker, 1998); program(s) used to re®ne structure: XL in
SHELXTL; molecular graphics: XP and XSHELL in SHELXTL;
software used to prepare material for publication: XCIF in
SHELXTL.
Angles (ꢀ) between various molecular entities for HNAB and HNS
molecules.
HNAB-I
C1ÐN1ÐN1iÐC1i
43.2
C1ÐC6
C1ÐC6
O1ÐN2ÐO2
O3ÐN3ÐO4
O5ÐN9ÐO6
129.6
126.1
148.3
Sandia is a multiprogram laboratory operated by Sandia
Corporation, a Lockheed Martin Company, for the United
States Department of Energy under contract No. DE-AC04-94
A L85000. The authors acknowledge the technical assistance
of Ralph G. Tissot Jr and Dale Huber.
HNAB-II
C1ÐN1ÐN11ÐC11
C1ÐC6
C11ÐC61
C1ÐC6
C11ÐC61
48.0
51.3
81.1
87.5
1.4
O1ÐN2ÐO2
O3ÐN4ÐO4
O5ÐN6ÐO6
O11ÐN21ÐO21
O31ÐN41ÐO41
O51ÐN61ÐO61
151.2
Supplementary data for this paper are available from the IUCr electronic
archives (Reference: GA1037). Services for accessing these data are
described at the back of the journal.
52.4
21.3
143.4
HNAB-III (two molecules)
C1AÐN1AÐN5AÐC7A C1AÐC6A C7AÐC12A
References
Brown, C. J. (1966a). Acta Cryst. 21, 153±158.
Brown, C. J. (1966b). Acta Cryst. 21, 146±153.
Brown, G. M. & Marsh, R. E. (1963). Acta Cryst. 16, 191±202.
Bruker (1997). SMART and SAINT. Bruker AXS Inc., Madison, Wisconsin,
USA.
Bruker (1998). SHELXTL. Bruker AXS Inc., Madison, Wisconsin, USA.
Cady, H. H. (1967). Acta Cryst. 23, 601±609.
Cady, H. H. & Larson, A. C. (1965). Acta Cryst. 18, 485±496.
Christensen, A. T. & Stromme, K. O. (1969). Acta Cryst. B25, 657±664.
Dickinson, C., Stewart, J. M. & Holden, J. R. (1966). Acta Cryst. 21, 663±670.
Duke, J. R. C. (1977). Private communication.
C1AÐC6A
C7AÐC12A
50.8
32.0
97.3
130.0
160.0
148.1
O1AÐN2AÐO2A
O3AÐN3AÐO4A
O5AÐN4AÐO6A
O7AÐN6AÐO8A
O9AÐN7AÐO10A
O11AÐN8AÐO12A
67.7
31.6
47.1
C1BÐN1BÐN5BÐC7B C1BÐC6B C7BÐC12B
C1BÐC6B
C7BÐC12B
57.4
27.7
95.1
41.1
16.7
155.1
Â
Gerard, P. F. & Hardy, A. (1988). Acta Cryst. C44, 1283±1287.
O1BÐN2BÐO2B
O3BÐN3BÐO4B
O5BÐN4BÐO6B
O7BÐN6BÐO8B
O9BÐN7BÐO10B
O11BÐN8BÐO12B
Graeber, E. J. & Morosin, B. (1974). Acta Cryst. B30, 310±317.
Hamilton, W. (1964). In Statistics in Physical Science. New York: The Ronald
Press Co.
Holden, J. R. (1967). Acta Cryst. 22, 545±550.
Holden, J. R., Dickerson, C. & Bock, C. M. (1972). J. Phys. Chem. 76, 3597±
3602.
70.8
26.7
123.3
HNS (two molecules)
Hughes, D. L. & Trotter, J. (1971). J. Chem. Soc. A, pp. 2181±2184.
Iwasaki, F. & Saito, Y. (1970). Acta Cryst. B26, 251±260.
McCrone, W. C. (1967). Crystallographic Study of Hexanitroazobenzene
(HNAB), Report SAND, pp. 75±7087. Sandia Laboratories, Albuquerque,
NM, USA.
C11ÐC17ÐC17iÐC11i
C11ÐC16
C11ÐC16
67.1
O12ÐN12ÐO13
O14ÐN14ÐO15
O16ÐN16ÐO17
5.5
18.4
48.7
McPhail, A. T. & Sim, G. A. (1965). J. Chem. Soc. pp. 227±236.
Palm, J. H. (1966). Acta Cryst. 21, 473±476.
C21ÐC27ÐC27iÐC21i
72.0
C21ÐC26
Rae, A. D. (2000). RAELS2000. The Australian National University,
Canberra, Australia.
Sakurai, T., Sundaralingam, M. & Jefferey, G. A. (1963). Acta Cryst. 16, 354±
363.
È
Sheldrick, G. M. (1999). SADABS. Version 2.03. University of Gottingen,
Germany.
C21ÐC26
O22ÐN22ÐO23
O24ÐN24ÐO25
O26ÐN26ÐO27
43.8
21.7
13.5
Sheldrick, G. M. & Gould, R. O. (1995). Acta Cryst. B51, 423±431.
Trueblood, K. N., Goldish, E. & Donohue, J. (1961). Acta Cryst. 14, 1009±1017.
Note: The labeling schemes are those of the original publications. Labels including the
superscript i signify inversion-symmetry-related atoms.
ꢁ
o130 Rodriguez et al. C12H4N8O12
Acta Cryst. (2005). C61, o127±o130