Gas-phase pyrolysis of 1-(2-azidophenyl)imidazole

Article abstract of DOI:10.1039/a700457e

Flash vacuum pyrolysis (FVP) of the azide 4 leads to imidazo[1,2-a]benzimidazole 8 exclusively, via highly regioselective insertion of the triplet nitrene intermediate 5 into the 2-CH bond of the imidazole ring. The X-ray crystal structure and NMR spectro

Full text of DOI:10.1039/a700457e

View Article Online / Journal Homepage / Table of Contents for this issue
Gas-phase pyrolysis of 1-(2-azidophenyl)imidazole
Alexander J. Blake,†,a Bernard A. J. Clark,b Hamish McNab*,a and
Craig C. Sommervillea
a
Department of Chemistry, The University of Edinburgh, West Mains Road, Edinburgh,
UK EH9 3JJ
b Kodak European Research and Development, Headstone Drive, Harrow, Middlesex,
UK HA1 4TY
Flash vacuum pyrolysis (F VP) of the azide 4 leads to imidazo[1,2-a]benzimidazole 8 exclusively, via highly
regioselective insertion of the triplet nitrene intermediate 5 into the 2-CH bond of the imidazole ring. The
X-ray crystal structure and N M R spectroscopic properties of 8 are discussed in detail.
In a recent communication,1 we have shown that pyrolysis of
N
N
F
1-(2-azidophenyl)pyrazole 1 in the gas-phase under flash
NO2
N
N
N
vacuum pyrolysis (FVP) conditions leads to much more effi-
cient cyclisation processes via the corresponding nitrene than
take place when the reactions are conducted in solution.2 In
particular, the triplet component of the nitrene pathway under-
goes C–H insertion to give pyrazolobenzimidazole 2 whereas
only hydrogen abstraction and dimerisation products were
found under solution conditions. The singlet component of the
nitrene pathway is trapped by the lone pair on the pyrazole 2-
nitrogen atom to give the pyrazolobenzotriazole 3 (Scheme 1).
i
ii
NO2
NH2
N
H
6
7
iii
N
N
N
N
iv
N
N
N3
N
N+
H
N
N
N
N–
N
FVP
N3
5
4
Scheme 2 Reagents and conditions: i, K2CO3–pyridine, 16 h; ii, H2–Pd/
C, 3 atm, 4 h; iii, HNO2, NaN3, 30 min; iv, FVP
1
2
3
Scheme 1
ingly involatile and showed evidence of extensive decom-
position at temperatures at or below its sublimation point.
(CAUTION: azides are potentially explosive. Although we had
no problems with this pyrolysis, all precautions were taken and
particularly a metal inlet heater was used to contain any explo-
sion which might take place.) The situation was improved
somewhat by the use of a diffusion pump (10Ϫ4 Torr), but even
then only small quantities of the azide could be pyrolysed in a
single run, and the yield of recovered products was poor (ca.
12%). The pyrolysate was obtained as white crystals whose
mass spectrum showed a molecular ion at m/z 157, as expected
following loss of dinitrogen from the azide 4. However it was
clear from the 13C NMR spectrum of the crude pyrolysate that
just one heterocyclic product was formed, even though two
insertion products, imidazo[1,2-a]benzimidazole 8 and imid-
azo[1,5-a]benzimidazole 9, could in principle be obtained.
In view of these results, we have extended our studies to the
parent 1-(2-azidophenyl)imidazole 4 case. Although there is
no corresponding trapping mechanism for the singlet nitrene,
we anticipated that two possible products might be formed by
triplet insertion into the 2- or 5-CH bonds of the imidazole.
When the nitrene 5 was generated in solution by phosphite-
mediated deoxygenation of the nitro compound 6, the only
reported product was 2-nitroaniline (29%).2 Other reactions
of imidazoles with electron-deficient intermediates are sur-
prisingly scarce.3 Busby et al. have studied the gas-phase
intermolecular reaction of a number of imidazoles with
dichlorocarbene and obtained ring expansion products which
were rationalised by insertion mechanisms into C᎐C or C᎐N
bonds.4
For the gas-phase generation of the nitrene, we employed the
azide precursor 4 which was efficiently prepared by the method
shown in Scheme 2. The nitro compound 6 was made in 92%
yield by reaction of imidazole with 2-fluoronitrobenzene using
a minor modification of the literature method,2 and reduction
to the amine 7 was achieved by catalytic hydrogenation in 99%
yield. Diazotisation and reaction with sodium azide gave the
azide 4 (98%).
N
N
H
H
N
N
N
N
Upon attempted FVP at 500 ЊC under our standard vacuum
8
9
conditions (10Ϫ2–10Ϫ3 Torr), the azide 4 proved to be exceed-
Only the imidazo[1,2-a]benzimidazole 8 is a known com-
pound,5 but its spectra were not reported in sufficient detail to
allow unambiguous identification in this case.
† Present address: Department of Chemistry, University of Notting-
ham, University Park, Nottingham, UK NG7 2RD.
J. Chem. Soc., Perkin Trans. 1, 1997
1605

View Article Online
Table 1 Selected bond lengths (Å) and angles (Њ) for 8
7.58
H
N
N(1)–C(1a)
N(1)–C(2)
C(2)–C(3)
C(3)–N(4)
N(4)–C(1a)
N(4)–C(4a)
C(4a)–C(5)
C(4a)–C(8a)
C(5)–C(6)
C(6)–C(7)
C(7)–C(8)
C(8)–C(8a)
C(8a)–N(9)
N(9)–C(1a)
1.319(2)
1.401(2)
1.350(2)
1.395(2)
1.364(2)
1.399(2)
1.376(2)
1.407(2)
1.386(2)
1.392(2)
1.390(2)
1.389(2)
1.398(2)
1.372(2)
7.10 H
H
N
N
1%
7.72
7.48
H
H
1%
1%
7.14
1%
H
H
7.26
1%
Fig. 2 Proton chemical shift ([2H6]acetone) and NOE data for 8
of N(9)–C(8a) [1.398(2) Å], indicating that electron delocalis-
ation for the N(9) lone pair takes place predominantly into the
imidazole ring rather than the benzenoid ring. The expected
angular distortion at each of the ring junction positions is
present, with all the exocyclic angles being >130Њ and that at
N(4) [C(3)–N(4)–C(4a)] being >140Њ. In addition, the ring
fusion leads to some distortion of both the bond lengths and
angles in the benzenoid ring.
C(1a)–N(1)–C(2)
C(3)–C(2)–N(1)
C(2)–C(3)–N(4)
C(1a)–N(4)–C(3)
C(1a)–N(4)–C(4a)
C(3)–N(4)–C(4a)
C(5)–C(4a)–N(4)
C(5)–C(4a)–C(8a)
N(4)–C(4a)–C(8a)
C(4a)–C(5)–C(6)
C(5)–C(6)–C(7)
C(8)–C(7)–C(6)
C(7)–C(8)–C(8a)
C(8)–C(8a)–N(9)
C(8)–C(8a)–C(4a)
N(9)–C(8a)–C(4a)
C(1a)–N(9)–C(8a)
N(1)–C(1a)–N(4)
N(1)–C(1a)–N(9)
N(4)–C(1a)–N(9)
102.22(13)
113.17(14)
104.34(15)
106.54(12)
109.38(12)
143.99(14)
132.10(15)
122.54(15)
105.36(13)
117.1(2)
121.2(2)
121.7(2)
117.6(2)
131.22(15)
119.84(14)
108.94(13)
106.90(13)
113.73(14)
136.96(15)
109.31(13)
Each individual ring of 8 is almost planar [RMS deviation
of the imidazole-derived ring N(1)–C(2)–C(3)–N(4)–C(1a) is
0.0011 Å, that of the central ring C(1a)–N(4)–C(4a)–C(8a)–
N(9) is 0.012 Å and that of the benzenoid ring is 0.005 Å], but
the molecule as a whole is very slightly bowl-shaped (angle
between normals to the two 5-membered rings is 2.4Њ and
between the central ring and the benzenoid ring is 1.6Њ). As
shown in Fig. 1, the molecules are linked into dimeric pairs by
hydrogen bonding between H(9)N and N(1Ј), where N(1Ј) is in
a molecule related by inversion. The H ؒ ؒ ؒ NЈ distance is 1.98 Å,
and the N–H ؒ ؒ ؒ NЈ angle is 166Њ. The two molecules in the
dimeric unit are parallel but not coplanar, the centres of the
best planes being offset by 0.466 Å. The dimers pack in parallel
stacks related by inversion (1 Ϫ x, 1 Ϫ y, 1 Ϫ z); the closest
approach of molecules within individual stacks is 3.23 Å.
In view of the paucity of literature data on the imidazo[1,5-a]-
benzimidazole system, the NMR spectroscopic properties of
8 were investigated in some detail. The 1H NMR spectrum was
highly solvent dependent, but solutions in [2H6]acetone were
fully resolved at 360 MHz and could be fully assigned on the
basis of NOE studies (Fig. 2), which showed the order of
increasing chemical shift in the benzenoid portion of the
molecule to be H(6) < H(7) < H(8) < H(5). In the fully coupled
13C NMR spectrum, all the methine signals appeared as doub-
lets of doublets. The signals at δC 126.68 and 105.74, which had
1
large values of JCH (189–196 Hz), were assigned as the C(2)
and C(3) resonances, respectively, with the signal of lower
chemical shift corresponding to the more electron-rich position.
The absence of a 1JCH value >200 Hz provides good supporting
evidence for the imidazo[1,2-a]benzimidazole rather than the
isomeric imidazo[1,5-a]benzimidazole structure, since it is
Fig. 1 A view of a dimer of molecule 8 showing crystallographic
numbering scheme
It proved surprisingly difficult to distinguish the two systems
unambiguously by 1H NMR spectroscopy, in particular
because 3- and 4-bond proton coupling constants in imidazole-
like systems are of comparable magnitude,6 and because NOE
experiments involving the two imidazole-derived protons at δH
7.58 and 7.10 (J 1.8 Hz) were inconclusive (see below). The
problem was finally solved unambiguously by single crystal
X-ray analysis which confirmed that the unknown compound
was imidazo[1,2-a]benzimidazole 8. Structural parameters are
given in Table 1 and a view of the molecule with the crystallo-
graphic numbering system is shown in Fig. 1. The bond lengths
N(1)–C(1a) [1.319(2) Å] and C(2)–C(3) [1.350(2) Å] in particu-
lar serve to exclude the alternative structure 9, as well as the
location of each hydrogen atom as a difference electron density
feature in the range 0.64–0.78 electrons ÅϪ3 with the remaining
1
known that the JCH values of C(2) in imidazole systems are
especially large in magnitude.8 No attempt was made to corre-
late the benzenoid carbon signals of 8, which occurred in the
range δC 123.36–110.89. One-bond and long-range coupling
constants nJCH are quoted in the Experimental section. The fine
couplings were due to 2-bond interactions in the case of the
imidazole-derived ring, and to 3-bond interactions in the case
of the benzenoid ring.
In conclusion, these results show that thermolysis of the
azide 4 leads to a nitrene intermediate which inserts exclusively
into the 2-CH bond of the imidazole ring with no correspond-
ing reaction at the 5-CH bond. We believe that this reaction is
the first example of such high regioselectivity in the imidazole
system and an unusual example of selectivity in reactive inter-
mediate chemistry. It seems unlikely that a concerted insertion
reaction of a singlet nitrene would be able to distinguish H(2)
and H(5) in this way. However, hydrogen abstraction by a triplet
nitrene would generate an intermediate diradical 10 in which
one electron is located in an imidazole σ-orbital (Scheme 3), a
features less than 0.20 electrons ÅϪ3
.
Bond lengths in the imidazole-derived ring show some evi-
dence of increased localisation with respect to imidazole itself;7
in particular, all the formal single bonds are much longer than
in the parent molecule. In the central ring of 8, the length of the
N(9)–C(1a) bond [1.372(2) Å] is significantly shorter than that
1606
J. Chem. Soc., Perkin Trans. 1, 1997

View Article Online
1-(2-Azidophenyl)imidazole 4
N
N
N
To stirred solution of 1-(2-aminophenyl)imidazole 7 (1.03 g, 6.5
mmol) in concentrated hydrochloric acid solution (35%, 3 cm3)
and water (10 cm3) at 0 ЊC was added a solution of sodium
nitrite (0.50 g, 7.2 mmol) in water (3 cm3). The resulting solu-
tion was added carefully, with stirring, to a solution of sodium
azide (0.47 g, 7.2 mmol) and sodium acetate (2.5 g) in water (15
cm3) at 0 ЊC. On addition a light brown precipitate appeared,
however the mixture was stirred at room temperature for a
further 30 min. The mixture was extracted with diethyl ether
(3 × 50 cm3), the organic extracts were washed with dilute
aqueous sodium hydroxide (2 , 10 cm3) and dried (MgSO4).
The solvents were removed under reduced pressure at a tem-
perature of less than 40 ЊC to yield the title compound 4 as a
brown oil (1.17 g, 98%) (HRMS: found, 185.0704. C9H7N5
requires M, 185.0701); νmax/cmϪ1 2130 and 2097; δH 7.64 (1 H, t,
H
N
H
N
N
N
N
N
H
10
Scheme 3
conclusion supported by independent evidence on the multi-
plicity of the nitrene responsible for insertion reactions in the
analogous pyrazolyl case.1 It is well known that the imidazole
2-position conveys special stability to reactive intermediates
located at that site. Examples include the remarkable isolable
imidazol-2-ylidenes9 11 and the well known reactivity of imid-
3
4
4J 1.1), 7.38 (1 H, ddd, J 8.1, 7.0, J 1.9), 7.25–7.13 (3 H, m)
and 7.11–7.09 (2 H, m); δC 137.25, 134.05 (q), 129.02, 128.11
(q), 126.34, 125.18, 120.02 and 119.27 (one quaternary overlaps
with another signal); m/z 185 (Mϩ, 74%), 157 (100), 130 (65),
104 (26), 103 (94), 90 (33), 77 (35) and 76 (45).
R
N
N
R
11
Flash vacuum pyrolysis of 1-(2-azidophenyl)imidazole 4
1-(2-Azidophenyl)imidazole 4 (0.09 g, 0.4 mmol) was allowed
to sublime at room temperature and 10Ϫ4 Torr (mercury dif-
fusion pump) over a period of 7 h through the silica furnace
tube (35 × 2.5 cm) which was maintained at 500 ЊC by an elec-
trical heater. The inlet tube was covered with a metal cylinder
to aid volatilisation and act as a safety screen in the event of
an azide explosion. Substantial decomposition of the sub-
strate in the inlet was observed. The pyrolysis was repeated
three times in order to obtain enough material for NMR
analysis. Upon completion of each pyrolysis the trap was
allowed to warm to room temperature under an atmosphere
of nitrogen. Acetone was then distilled into the trap, the
resulting solution was removed and the combined products
purified by dry-flash chromatography eluting with a 1:20
mixture of methanol and ethyl acetate. The pyrolyses
afforded only 9H-imidazo[1,2-a]benzimidazole 8 (24 mg,
total from 3 runs, 12%), mp 188 ЊC (decomp.) [lit.,5 190 ЊC
azole 2-hydrogen atoms towards strong bases which generate an
imidazolyl 2-carbanion. In the present case, stabilisation may
be conferred by lateral overlap of the orbital containing the
single electron with the lone pair orbital on C(3). Finally, our
results reinforce our previous comments1 on the utility of gas-
phase FVP methods in such fundamental studies of nitrene
reactions.
Experimental
1H and 13C NMR Spectra were recorded at 200 and 50 MHz
respectively for solutions in [2H]chloroform unless otherwise
stated. Coupling constants (J) are quoted in Hz.
1-(2-Nitrophenyl)imidazole 6
A
stirred mixture of imidazole (1.09 g, 16 mmol),
3
2-fluoronitrobenzene (2.26 g, 16 mmol), anhydrous potassium
carbonate (2.24 g, 16 mmol) and copper() oxide (0.08 g) in
pyridine (4.0 cm3) was heated under reflux for 16 h under nitro-
gen. Dichloromethane (100 cm3) and activated charcoal (2.0 g)
were added to the resulting dark brown residue and the mixture
was heated under reflux for 30 min. The mixture was filtered
through a Celite pad then washed thoroughly with dichloro-
methane. The solvents were removed under reduced pressure to
give the title compound 6 as orange crystals (2.80 g, 92%), mp
96–98 ЊC (from cyclohexane–chloroform) (lit.,2 mp 96–98 ЊC);
(decomp.)]; δH ([2H6]acetone) 7.72 (1 H, d, J 6.7), 7.58 (1 H,
3
3
3
4
d, J 1.8), 7.48 (1 H, d, J 6.0), 7.26 (1 H, td, J 7.7, J 1.4),
7.14 (1 H, td, 3J 7.7, 4J 1.3) and 7.10 (1 H, d, 3J 1.8); δC
149.56 (q), 137.89 (q), 126.88 (1J 189.7, J 10.3), 124.73 (q),
2
123.36 (1J 157.3, J 5.7), 119.56 (1J 161.1, J 7.7), 113.55 (1J
3
3
163.9, J 8.8), 110.89 (1J 161.4, J 8.8) and 105.74 (1J 196.1,
2J 14.8); m/z 157 (Mϩ, 33%), 88 (12), 86 (62), 84 (100), 49
(13), 47 (20), 40 (12) and 32 (100).
3
3
X-Ray crystallography
3
4
3
4
δH 7.92 (1 H, dd, J 8.0, J 1.6), 7.67 (1 H, td, J 7.7, J 1.6),
7.58–7.52 (2 H, m), 7.40 (1 H, dd, 3J 7.8, 4J 1.5), 7.11 (1 H, br s)
and 7.00 (1 H, t, 3J 1.2).
A red–brown crystal of dimensions 0.50 × 0.35 × 0.25 mm was
mounted on a glass fibre and transferred to a Stoe Stadi-4
four-circle diffractometer equipped with an Oxford Cryo-
systems low-temperature device.10
1-(2-Aminophenyl)imidazole 7
Crystal data. C9H7N3, M = 157.18, monoclinic, space group
P21/n with a = 9.6989(11), b = 5.4692(7), c = 13.863(2) Å,
β = 100.315(14)Њ, U = 723.5 Å3 {from 2θ values of 31 reflections
measured at ±ω [30 р 2θ р 33Њ, λ(Mo-Kα) = 0.71 073 Å,
Palladium-on-charcoal (10%, 0.27 g) was carefully added to a
solution of 1-(2-nitrophenyl)imidazole 6 (2.00 g, 10.6 mmol) in
ethanol (150 cm3) and the mixture was hydrogenated at medium
pressure (3 atm) and ambient temperature for 4 h. The mixture
was filtered through a Celite pad and washed thoroughly with
ethanol. The solvents were removed under reduced pressure
to give the amine 7 as a brown solid (1.67 g, 99%) which did
not require further purification; mp 103–106 ЊC (from cyclo-
hexane–toluene) (Found: C, 67.9; H, 5.85; N, 26.2. C9H9N3
requires C, 67.9; H, 5.65; N, 26.4%); δH 7.59 (1 H, s), 7.25–
7.16 (2 H, m), 7.09–7.05 (2 H, m), 6.82–6.73 (2 H, m) and
4.24 (2 H, br s); δC 141.81 (q), 137.44, 129.68, 129.59, 126.91,
123.01 (q), 119.92, 118.25 and 116.17; m/z 159 (Mϩ, 51%),
132 (100), 131 (98), 119 (34), 104 (21), 77 (15), 65 (28), 52
(19) and 39 (16).
T = 150.0(2) K]}, Dc = 1.443 g cmϪ3, Z = 4, µ = 0.092 mmϪ1
.
Data collection and processing. Diffraction data were
acquired to 2θmax = 50Њ using ω–2θ scans. Of 1287 unique reflec-
tions collected, 1000 had F у 4σ(F ) and 1282 were retained in
all calculations. No crystal decay was observed and no correc-
tions were applied for absorption.
Structure solution and refinement. Automatic direct
methods11 identified the positions of all non-H atoms, which
were then refined anisotropically. Hydrogen atoms were located
from a ∆F synthesis and freely refined 12 with individual iso-
tropic thermal parameters. The weighting scheme wϪ1
=
2
2
2
2
2
1
–
[σ (Fo ) ϩ (0.035P) ϩ 0.23P], P = 3[MAX(Fo , 0) ϩ 2Fc ] led to
J. Chem. Soc., Perkin Trans. 1, 1997
1607

View Article Online
3 For example, M. R. Grimmett, in Comprehensive Heterocyclic
final convergence with R1 [F у 4σ(F )] = 0.0337, wR2 (all data) =
0.0846‡, S(F 2) = 1.062 for 138 refined parameters. An extinc-
tion correction 12 refined to 0.011(2) and the final ∆F synthesis
showed no peaks above ±0.18 e ÅϪ3. Fig. 1 was produced using
SHELXTL/PC.13
Chemistry, ed. K. T. Potts, Pergamon, Oxford, 1984, vol. 5, p. 418.
4 R. E. Busby, M. A. Khan, M. R. Khan, J. Parrick, C. J. G. Shaw and
M. Iqbal, J. Chem. Soc., Perkin Trans. 1, 1980, 1427.
5 H. Ogura, H. Takayanagi, Y. Yamazaki, S. Yonezawa, H. Takagi,
S. Kobayashi, T. Kamioka and K. Kamoshita, J. Med. Chem., 1972,
15, 923.
6 For example, G. S. Reddy, L. Mandell and J. H. Goldstein, J. Chem.
Soc., 1963, 1414,
Acknowledgements
7 S. Martinez-Carrera, Acta Crystallogr., 1966, 20, 783.
8 For example, H. McNab, J. Chem. Soc., Perkin Trans. 1, 1987, 657.
9 A. J. Arduengo III, H. V. R. Dias, R. L. Harlow and M. Kline,
J. Am. Chem. Soc., 1992, 114, 5530.
10 J. Cosier and A. M. Glazer, J. Appl. Crystallogr., 1986, 19, 105.
11 G. M. Sheldrick, SHELXS-86, Acta Crystallogr., Sect. A, 1990, 46,
467.
We are grateful to Kodak Ltd and the EPSRC for a CASE
award to C. C. S., to the SERC for the provision of a four-circle
diffractomer and to Dr S. Parsons for his help in collecting the
X-ray data.
¹
‡ wR2 = {Σ[w(Fo2 Ϫ Fc )2]/Σ[w(Fo ) ]} .
2
2 2

²
12 G. M. Sheldrick, SHELXL-93, University of Göttingen, Germany,
1993.
13 G. M. Sheldrick, SHELXTL/PC ver. 4.3, Siemens Analytical
Instrumentation Inc., Madison, WI.
References
1 B. A. J. Clark, H. McNab and C. C. Sommerville, Chem. Commun.,
1996, 1211.
2 J. M. Lindley, I. M. McRobbie, O. Meth-Cohn and H. Suschitzky,
J. Chem. Soc., Perkin Trans. 1, 1977, 2194.
Paper 7/00457E
Received 20th January 1997
Accepted 3rd March 1997
1608
J. Chem. Soc., Perkin Trans. 1, 1997