Thermal Ring Contraction of Dibenz[b,f]azepin-5-yl Radicals: New Routes to Pyrrolo[3,2,1-jk]carbazoles

Article abstract of DOI:10.1021/jo800637u

(Chemical Equation Presented) Flash vacuum pyrolysis (FVP) of N-allyl- or N-benzyldibenz[b,f]azepine at temperatures from 750 to 950°C gives pyrrolo[3,2,1-jk]carbazole as the major product. The mechanism of the ring contraction involves dibenzazepin-1-yl radical formation, followed by transannular attack and formation of a 2-(indol-1-yl)phenyl radical which cyclizes. The mechanism is supported by independent generation of 2-(indol-1-yl)phenyl radicals by two different methods, and the use of 1-(2-nitrophenyl)indole as a radical generator gives an optimized synthetic route to pyrrolo[3,2,1-jk]carbazole (54% overall yield in two steps from indole). The first substituted pyrrolo[3,2,1-jk]carbazoles have been synthesized by FVP methods and also by reactions of the parent compound with electrophiles, leading to a range of 4-substituted pyrrolocarbazoles.

Full text of DOI:10.1021/jo800637u

Thermal Ring Contraction of Dibenz[b,f]azepin-5-yl Radicals: New
Routes to Pyrrolo[3,2,1-jk]carbazoles
Lynne A. Crawford, Hamish McNab,* Andrew R. Mount, and Stuart I. Wharton
School of Chemistry, The UniVersity of Edinburgh, West Mains Road, Edinburgh EH9 3JJ, U.K.
H.McNab@ed.ac.uk
ReceiVed March 25, 2008
Flash vacuum pyrolysis (FVP) of N-allyl- or N-benzyldibenz[b,f]azepine at temperatures from 750 to
950 °C gives pyrrolo[3,2,1-jk]carbazole as the major product. The mechanism of the ring contraction
involves dibenzazepin-1-yl radical formation, followed by transannular attack and formation of a 2-(indol-
1-yl)phenyl radical which cyclizes. The mechanism is supported by independent generation of 2-(indol-
1-yl)phenyl radicals by two different methods, and the use of 1-(2-nitrophenyl)indole as a radical generator
gives an optimized synthetic route to pyrrolo[3,2,1-jk]carbazole (54% overall yield in two steps from
indole). The first substituted pyrrolo[3,2,1-jk]carbazoles have been synthesized by FVP methods and
also by reactions of the parent compound with electrophiles, leading to a range of 4-substituted
pyrrolocarbazoles.
SCHEME 1
Introduction
Very little is known of the chemical properties of 5- or
7-membered heterocyclic species in which the heteroatom bears
a radical center. In previous work, we have shown that 1,2,4-
triazol-4-yl species, generated by flash vacuum pyrolysis (FVP)
of a 4-amino compound, can cyclize with high regioselectivity
onto a 3-S-allyl group to provide thiazolo[3,2-b]triazoles.1 Here,
we extend this work to the 7-membered ring series and
demonstrate that, under FVP conditions, dibenzazepin-5-yl
radicals 1 undergo ring contraction leading to pyrrolo[3,2,1-
jk]carbazole 2 (Scheme 1) via cyclization of an intermediate
2-(indol-1-yl)phenyl radical. By further exploiting this phenyl
radical cyclization, we have optimized the synthetic route to
pyrrolo[3,2,1-jk]carbazole 2 and present the first studies of its
chemical properties.
SCHEME 2
moderate yield by alkylation of the parent compound 3
(“iminostilbene”) under conditions used for the N-alkylation of
pyrroles and indoles3 (Scheme 2).
When the N-benzyl compound 4 was subjected to FVP at
750 °C, four products were identified from the reaction mixture
(Scheme 3).
The formation of bibenzyl 6 (ca. 20%) provided confirmation
that radical generation had been successful. Recovered 5H-
dibenz[b,f]azepine 3 (20%) suggested that hydrogen capture (a
well-known reaction of aminyl and phenoxyl radicals under FVP
Results and Discussion
Cleavage of N-benzyl or N-allyl groups at ca. 700 °C under
FVP conditions is a good route to phenylaminyl and related
radical species.2 In the dibenz[b,f]azepine series, the corre-
sponding FVP precursors 4 and 5 were easily obtained in
(1) Cartwright, D. D. J.; Clark, B. A. J.; McNab, H. J. Chem. Soc., Perkin
Trans. 1 2001, 424–428.
(2) Cadogan, J. I. G.; Hickson, C. L.; McNab, H. Tetrahedron 1986, 42,
2135–2165.
(3) Heaney, H.; Ley, S. V. J. Chem. Soc., Perkin Trans. 1 1973, 499–500.
6642 J. Org. Chem. 2008, 73, 6642–6646
10.1021/jo800637u CCC: $40.75 2008 American Chemical Society
Published on Web 08/12/2008

New Routes to Pyrrolo[3,2,1-jk]carbazoles
SCHEME 3
SCHEME 4
SCHEME 5
conditions2) is able to compete with other processes at this
temperature. In addition, two ring contraction products were
obtained. The first, a trace of 9-methylacridine 7 (2%), is likely
to be formed by ring contraction of 5H-dibenz[b,f]azepine 3; a
control pyrolysis of 3 under more vigorous conditions (950 °C)
gave 9-methylacridine 7 as the sole product in 60% isolated
yield. Ring contractions of this type are known to provide
acridine byproduct in the commercial synthesis of 3 by catalytic
gas-phase dehydrogenation of its 10,11-dihydro derivative,
particularly at higher temperatures.4,5 A 1,5-shift, electrocy-
clization, ring cleavage, hydrogen shift sequence is the most
likely mechanism (Scheme 4).
However, the major product formed from FVP of the
N-benzyl compound 4 was pyrrolo[3,2,1-jk]carbazole 2 (ca.
35%), identified by comparison of its melting point and 1H NMR
spectrum with those previously reported.6,7 On a preparative
scale, the use of the N-benzyl precursor 4 proved to be
inconvenient because it was difficult to separate 2 from the
bibenzyl coproduct 6. Pyrolysis of the N-allyl compound 5 was
therefore studied; volatile coproducts (allene and/or biallyl) were
expected to be generated, thus eliminating any separation issues.
In this case, only the pyrrolocarbazole 2, the dibenzazepine 3,
and the acridine 7 were obtained in 56:43:1 relative yields at
750 °C. The temperature of the pyrolysis proved to have a
dramatic effect on the product distribution, with the amount of
pyrrolocarbazole 2 increasing with temperature (Figure 1). The
optimum temperature for formation of 2 proved to be 950 °C
and on a preparative scale (>1.0 g) a 63% yield of pure 2 was
obtained after workup and chromatography.
3 (and hence the low levels of the acridine 7). Alternatively, a
trans-annular interaction of the aminyl center of 1 with its 10,11-
double bond leads via the tetracyclic transition state (or
intermediate) 8 to the phenyl radical 9 which can cyclize to
provide the strained pyrrolo[3,2,1-jk]carbazole 2. Many ex-
amples of phenyl radical cyclizations to provide strained ring
systems are known,8 and it is not surprising that this process is
relatively favored at the higher temperatures.
It is clear from the proposed mechanism that the phenyl
radical 9 is the key intermediate in pyrrolocarbazole formation,
and we sought to obtain supporting evidence by generating such
radicals by an independent route. In previous work,9 we have
used FVP of allyl esters as a source of substituted phenyl
radicals, and the required precursor 11 of the phenyl radical 9
was made by the method shown in Scheme 6. The carboxylic
acid 10 was not isolated but was esterified in situ to provide 11
in 47% yield for the two steps.
FVP of the allyl ester 11 at 950 °C provided pyrrolo[3,2,1-
jk]carbazole 2 in 42% isolated yield as essentially the only
product (Scheme 6). This result is consistent with the mechanism
proposed in Scheme 5 and with the involvement of the phenyl
radical 9 in the process.
The pyrolysis results can be rationalized by the proposed
mechanisms shown in Scheme 5. The dibenzazepin-5-yl radical
1 may undergo hydrogen capture (a comparatively low energy
process, favored at low temperatures) to give the dibenzazepine
Because of the availability of the starting materials, the phenyl
radical cyclization route to pyrrolo[3,2,1-jk]carbazoles is more
general in principle than the dibenzazepine ring contraction. As
(4) Kricka, L. J.; Ledwith, A. Chem. ReV. 1974, 74, 101–123.
(5) Inoue, M.; Itoi, Y.; Enomoto, S.; Ishizuka, N.; Amemiya, H.; Takeuchi,
T. Chem. Pharm. Bull. 1982, 30, 24–27.
(6) Brown, R. F. C.; Choi, N.; Coulston, K. J.; Eastwood, F. W.; Ercole,
J. M.; Horvath, J. M.; Mattinson, M.; Mulder, R. J.; Ooi, H. C. Liebigs Ann.
Recl. 1997, 1931–1940.
(7) Klingstedt, T.; Hallberg, A.; Dunbar, P.; Martin, A. J. Heterocycl. Chem.
1985, 22, 1547–1550.
(8) For example: Peng, L.; Scott, L. T. J. Am. Chem. Soc. 2005, 127, 16518–
16521.
(9) Cadogan, J. I. G.; Hutchison, H. S.; McNab, H. Tetrahedron 1992, 48,
7747–7762.
J. Org. Chem. Vol. 73, No. 17, 2008 6643

Crawford et al.
SCHEME 8
TABLE 1. 1H and 13C NMR Data for Pyrrolo[3,2,1-jk]carbazole 2
(Numbering Scheme Shown in Scheme 3)
position
δC
δH
JHH/Hz
pattern
4
1
2
3
4
5
7
8
9
117.84
123.91
121.54
109.86
122.93
111.91
127.01
122.63
123.62
7.90
7.51
7.79
6.86
7.73
7.68
7.47
7.32
8.08
3J 7.4, J 0.5
dd
t
dd
d
d
ddd
td
FIGURE 1. Temperature profile for the pyrolysis of 5.
3J 7.4
4
3J 7.4, J 0.5
SCHEME 6
3J 3.1
3J 3.1
4
3J 8.0, J 1.0,5J 0.7
4
3J 8.0, J 1.2
4
3J 7.8, J 1.0
td
ddd
4
10
3J 7.8, J 1.2,5J 0.7
isolation13 and appears to be the best explanation of certain
cyclization reactions observed by Rees and co-workers.14 In our
case, 1-(2-nitrophenyl)indole16 was readily made in 90% yield
by reaction of indole 15 with 2-fluoronitrobenzene, and FVP
of 16 provided pyrrolo[3,2,1-jk]carbazole 2 (60%) after chro-
matography (Scheme 8). The effect of the coformation of
reactive NOx species as pyrolysis byproducts from the nitro-
phenylindole method could be minimized by using a coldfinger,
cooled by a dry ice-acetone mixture, in place of our usual
U-tube trap.
SCHEME 7
As a second example of this strategy, 4-cyanopyrrolo[3,2,1-
jk]carbazole 19 was made in two steps (64% overall yield) from
indole-3-carbonitrile 17 via the nitro compound 18 (Scheme 8).
We believe that this method is generally the most efficient
of our new synthetic routes to the pyrrolo[3,2,1-jk]carbazole
ring system so that the parent compound 2 is now readily made
in two steps from indole, on a gram scale, in 54% overall yield.
Other published routes suffered from low overall yields,6
mixtures of products,6 and/or inconvenient starting materials,6,7
such that only milligram quantities were previously available.
On the other hand, the reactive NOx coproducts are the main
drawback of the route involving the nitro precursors, especially
for the preparation of electron rich species such as 14, for which
the allyl ester method remains preferable (see the Supporting
Information).
Pyrrolo[3,2,1-jk]carbazole 2 is an unusual heterocyclic skel-
eton which is significantly strained, owing to the presence of
the bis-fused benzene ring spanning two bonds of a pyrrolizine
system. With quantities of 2 now available, a brief examination
was made of its spectroscopic and chemical properties. The
previous assignment of its 1H NMR spectrum7 was fully
confirmed by NOESY methods and extension to the 13C
dimension was made by HSQC (see the Supporting Informa-
tion). The full assignments are shown in Table 1.
a second example, 4-methylpyrrolo[3,2,1-jk]carbazole 14 was
obtained via the esters 12 and 13 by the route shown in Scheme
7. This appears to be the first substituted pyrrolo[3,2,1-
jk]carbazole recorded in the literature.
In practice, the preparation of allyl esters such as 11 proved
less robust than we had hoped, and we therefore sought an
alternative generator of the phenyl radical 9 which could be
readily made from indole by SNAr methodology. The nitro
compound 16 was an attractive possibility synthetically, though
the literature reveals conflicting evidence on the FVP reactions
of aromatic nitro compounds. Under flow conditions,10 (later
extended to FVP11) nitro compounds have been shown to give
phenols by rearrangement and cleavage of NO. However, Marty
and De Mayo presented good evidence that aryl nitro groups
could undergo C-N cleavage to give phenyl radical intermedi-
ates;12 this route has been used to generate diradicals for matrix
Despite the inherent ring strain, pyrrolo[3,2,1-jk]carbazole 2
was found to behave as a 1-substituted indole in its reactions
with electrophiles. Thus, treatment with oxalyl chloride provided
(13) (a) Sander, W.; Exner, M.; Winkler, M.; Balster, A.; Hjerpe, A.; Kraka,
E.; Cremer, D. J. Am. Chem. Soc. 2002, 124, 13072–13079. (b) Winkler, M.;
Cakir, B.; Sander, W. J. Am. Chem. Soc. 2004, 126, 6135–6149.
(14) Houghton, P. G.; Pipe, D. F.; Rees, C. W. J. Chem. Soc., Perkin Trans.
1 1985, 1471–1479.
(10) Fields, E. K.; Meyerson, S. Acc. Chem. Res. 1969, 2, 273–278.
(11) Wiersum, U. E. Polycyclic Aromat. Compd. 1996, 11, 291–300.
(12) Marty, R. A.; De Mayo, P. J. Chem. Soc., Chem. Commun. 1971, 127–
128.
6644 J. Org. Chem. Vol. 73, No. 17, 2008

New Routes to Pyrrolo[3,2,1-jk]carbazoles
SCHEME 9
4 (35%): mp 81-83 °C (from hexane) (lit.16 mp 80 °C); NMR δH
7.03-7.63 (11H, m), 6.72-6.77 (2H, m), 6.57 (2H, s) and 4.81
(2H, s); m/z 283 (M+, 6), 217 (100), 194 (14), 149 (2), and 91
(77). Anal. Calcd for C21H17N: C, 89.05; H, 6.0; N, 4.95. Found:
C, 89.2; H, 6.2; N, 5.15.
5-Allyl-5H-dibenz[b,f]azepine (5). Application of the general
method using allyl bromide gave 5-allyl-5H-dibenz[b,f]azepine 5
(46%): mp 55-56 °C (lit.17 mp 55-57 °C); MS M+, 233.1200,
C17H15N requires 233.1205; NMR δH 7.22-7.29 (2H, m), 6.96-7.10
(6H, m), 6.77 (2H, m), 5.80 (1H, m), 5.32 (1H, m), 5.12 (1H, m)
and 4.40 - 4.43 (2H, m); NMR δC 150.5 (2 × quat), 135.1 (CH),
133.6 (2 × quat), 132.1 (2 × CH), 129.0 (2 × CH), 128.5 (2 ×
CH), 123.2 (2 × CH), 120.4 (2 × CH), 117.5 (CH2), and 53.4
(CH2); m/z 233 (M+, 19), 192 (100), 165 (15), 139 (7), 89 (6), and
77 (6).
SCHEME 10
Flash Vacuum Pyrolysis (FVP) Experiments. The precursor was
volatilized under rotary pump vacuum through an empty, electrically
heated silica tube (35 × 2.5 cm), and the products were collected
in a U-tube, cooled with liquid nitrogen, situated at the exit point
of the furnace. For large scale (0.5 g and greater) pyrolyses from
aromatic nitro precursors, a “coldfinger” trap, cooled by a mixture
of dry ice and acetone, was used in place of the U-tube. For all
pyrolyses, the pressure was measured by a Pirani gauge situated
between the product trap and the pump. Upon completion of the
pyrolysis, the trap was allowed to warm to room temperature under
a nitrogen atmosphere. The entire pyrolysate was dissolved in
solvent and removed from the trap. The precursor, pyrolysis
conditions [quantity of precursor (m), furnace temperature (Tf), inlet
temperature (Ti), pressure (P) and pyrolysis time (t)], and products
are quoted below for each experiment.
the 4-substituted product 20 in 48% yield; the position of
substitution was confirmed by NOESY correlation of the
1-proton singlet (δH 8.89) to a doublet of doublets of doublets
(δH 7.74) belonging to the 4-spin system (see the Supporting
Information). Similarly, Vilsmeier formylation of pyrrolo[3,2,1-
jk]carbazole 2 gave its 4-carboxaldehyde derivative 21 in 24%
(unoptimized) yield (Scheme 9).
In an attempt to extend the ring contraction work to the
previously reported diazepine 22, the one-step literature syn-
thesis of this compound was repeated (Scheme 10). In our hands,
the sole product isolated in moderate yield from this procedure,
had an identical 1H NMR spectrum and compatible melting point
with those reported for 22,15 but it was clear from its 13C NMR
spectrum that the compound contained an unchanged phenyl
group. This product proved to be 1-phenylbenzimidazole 23 (see
the Supporting Information), formed by N-formylation instead
of C-formylation, followed by cyclization.
In conclusion, the work presented in this paper has identified
a new thermal ring contraction of dibenzazepin-5-yl radicals
which has been extended to give the first convenient access to
the pyrrolo[3,2,1-jk]carbazole system 2. In reactions with
electrophiles, 2 behaves as a 1-substituted indole, with substitu-
tion reactions occurring exclusively at the 4-position. Further
examples of the phenyl radical route to new heterocyclic ring
systems related to pyrrolo[3,2,1-jk]carbazoles will be reported
in future publications.
FVP of 5-Benzyl-5H-dibenz[b,f]azepine (4). FVP of 4 (m 0.05 g,
Tf 750 °C, Ti 150 °C, P 0.04 Torr, t 20 min) produced bibenzyl 6
(ca. 20%, but could not be totally separated from 2 by chroma-
tography): NMR δH 7.16-7.87 (10H, m) and 2.92 (4H, s); NMR
δC 129.3 (2 × quat), 128.3 (4 × CH), 128.2 (4 × CH), 125.8 (2 ×
CH) and 37.8 (2 × CH2);18 pyrrolo[3,2,1-jk]carbazole 2 (ca. 35%,
but could not be separated from 6 by chromatography; see
spectroscopic data below); 9-methylacridine 7 (see below) (2%)
and 5H-dibenz[b,f]azepine 3 (20%); δH 7.03-7.10 (2H, m),
6.82-6.93 (4H, m), 6.51-6.55 (2H, m) and 6.35 (2H, s); NMR δC
148.2 (2 × quat), 132.0 (2 × CH), 130.4 (2 × CH), 129.6 (2 ×
quat), 129.3 (2 × CH), 122.9 (2 × CH) and 119.2 (2 × CH).
FVP of 5-Allyl-5H-dibenz[b,f]azepine (5). FVP of 5 [m 1.10 g
(4.7 mmol), Tf 950 °C, Ti 150 °C, P 0.04 Torr, tm 60 min] gave
pyrrolo[3,2,1-jk]carbazole 2 (0.57 g, 63%): mp 88-89 °C (lit.7 mp
89-90 °C), after dry flash chromatography on silica using hexane
3
4
5
as eluant; NMR δH (360 MHz) 8.08 (1H, ddd, J 7.8, J 1.2, J
3
4
3
4
0.7), 7.90 (1H, dd, J 7.4, J 0.5), 7.79 (1H, dd, J 7.4, J 0.5),
3
3
4
5
7.73 (1H, d, J 3.1), 7.68 (1H, ddd, J 8.0, J 1.0, J 0.7), 7.51
(1H, t, 3J 7.4), 7.47 (1H, td, 3J 8.0, 4J 1.2), 7.32 (1H, td, 3J 7.8, 4J
3
1.0) and 6.86 (1H, d, J 3.1); NMR δC (90 MHz) 141.3 (quat),
Experimental Section
139.9 (quat), 131.4 (quat), 127.0 (CH), 123.9 (CH), 123.6 (CH),
122.9 (CH), 122.6 (CH), 122.1 (quat), 121.5 (CH), 119.4 (quat),
117.8 (CH), 111.9 (CH) and 109.9 (CH); MS m/z 191 (M+, 78),
190 (84), 156 (100), 128 (31), 95 (6), 78 (14) and 51 (4).
Control pyrolysis of 5H-dibenz[b,f]azepine (3). FVP of 3 [m
0.35 g (1.8 mmol), Tf 950 °C, Ti 180-200 °C, P 0.009 Torr, t 30
min] gave 9-methylacridine 7 after dry flash chromatography on
silica using hexane/ethyl acetate as eluant (0.21 g, 60%): NMR δH
8.14-8.21 (4H, m), 7.45-7.75 (4H, m) and 3.02 (3H, s); NMR δC
Alkylation of 5H-dibenz[b,f]azepine (3). Powdered potassium
hydroxide (1.5 g, 37.5 mmol) was added to DMSO (10 cm3) and
the mixture allowed to stir for 10 min. 5H-Dibenz[b,f]azepine 3
(1.0 g, 5.2 mmol) was added, and the solution was stirred for 90
min. The appropriate alkyl halide (10.4 mmol) was added, and the
solution was stirred for a further 90 min. The solution was added
to water (100 cm3) and extracted with DCM (3 × 100 cm3). The
organic layer was washed with water (2 × 100 cm3) and dried
(MgSO4), and the solvent was removed under reduced pressure.
The residue was subjected to dry flash chromatography on silica
using hexane as eluant.
(16) Ohta, T.; Miyata, N.; Hirobe, M. Chem. Pharm. Bull. 1981, 29, 1221–
1230.
(17) Hannig, E.; Pech, R.; Dressler, C. H. R. Pharmazie 1979, 34, 670–671.
(18) Duffy, E. F.; Foot, J. S.; McNab, H.; Milligan, A. A. Org. Biomol. Chem.
2004, 2, 2677–2683.
5-Benzyl-5H-dibenz[b,f]azepine (4). Application of the general
method using benzyl bromide gave 5-benzyl-5H-dibenz[b,f]azepine
(19) Moeller, U.; Cech, D.; Schubert, F. Liebigs Ann. Chem. 1990, 1221–
1225.
(15) Narasimhan, N. S.; Chandrachood, P. S. Synthesis 1979, 589–590.
J. Org. Chem. Vol. 73, No. 17, 2008 6645

Crawford et al.
148.1 (2 × quat), 142.1 (quat), 129.9 (2 × CH), 129.6 (2 × CH),
125.5 (2 × quat), 125.2 (2 × CH), 124.3 (2 × CH) and 13.4
(CH3).19
Supporting Information Available: Full experimental details
for all new compounds; data for the FVP temperature profile
of 5; copies of 1H and 13C NMR spectra of 2 and 20 and of the
new compounds. This material is available free of charge via
the Internet at http://pubs.acs.org.
Acknowledgment. We are most grateful to the EPSRC (UK)
and The University of Edinburgh for Research Studentships (for
L.A.C. and S.I.W.).
JO800637U
6646 J. Org. Chem. Vol. 73, No. 17, 2008