Synthesis of novel indole derivatives: Variations in the Bartoli reaction

Article abstract of DOI:10.1055/s-1999-2900

The synthesis of a range of novel substituted indoles and tricyclic indole derivatives is described using an extension of the Bartoli methodology.

Full text of DOI:10.1055/s-1999-2900

1594
LETTER
Synthesis of Novel Indole Derivatives: Variations in the Bartoli Reaction
Adrian P. Dobbs,1* Martyn Voyle, Neil Whittall
SmithKline Beecham Pharmaceuticals, Old Powder Mills, Leigh, Tonbridge, Kent TN11 9AN, UK
E-mail: adrian@apdobbs.demon.co.uk; Neil_Whittall-1@sbphrd.com
Received 29 June 1999
Table 1 Results of the Bartoli Reaction employing various substra-
tes and Grignard Reagents
Abstract: The synthesis of a range of novel substituted indoles and
tricyclic indole derivatives is described using an extension of the
Bartoli methodology.
Key words: Grignard reactions, indoles, Bartoli reaction, regiose-
lectivity
In the course of our work, it was necessary to prepare a
range of novel di– and tri–substituted and fused indoles of
varied substitution pattern. A number of methods already
exist for the synthesis of such indoles from benzenoid or
aniline precursors and these initially appeared attractive
e.g. via palladium–mediated coupling and/or cyclisation
reactions.2 However due to the disappointing yields ob-
tained with certain substrates via these methods, attention
turned instead to the Bartoli method for indole synthesis.3
This method involves treating an aromatic nitro–com-
pound with three equivalents of a Grignard reagent (usu-
ally vinylmagnesium bromide) to form the indole but has
so far been of limited use, generally being restricted to ar-
omatic nitro–compounds bearing an ortho–substituent,
thus giving selectivity in the cyclisation and an efficient
route to 7–substituted indoles. (Indeed, the original work
A) Consistent with literature yield3a; B) Prepared by treatment of the
of Bartoli has shown that yields drop dramatically in the
corresponding cycloalkene with bromine to give the di–bromocyclo-
alkane followed by base–promoted hydrobromide elimination to the
absence of an ortho–substituent3a).
vinylbromide and finally treatment with magnesium turnings to gene-
However some interesting results have been encountered,
which show that this methodology may be synthetically
rate the Grignard reagent.4
useful for other substrates (Scheme 1 and Table 1).
Grignard reagent failed to push the reaction to completion
and indeed a lowering of yield was observed (1i) while the
levels of reduced products increased. It has been reported
that using compounds with meta– or para–methyl groups
gives drastically lower yields of indoles.3a It was found
that employing either meta– or para–trifluoromethyl (1d
and 1e) groups failed to give any indole product but rather
a complex mixture of compounds, including the corre-
sponding reduced aniline and nitroso derivatives as well
as azo–coupled products, but not the oxindole (a com-
monly reported side product of these reactions). In fact in
all these reactions (even the high indole–yielding ones)
the reduced aniline and nitroso compounds were always
detected in varying levels and ratios by GCMS analysis, a
result which is consistent with the proposed mechanism
for this reaction.3c Yields were upheld using disubstituted
nitrobenzenes (1f) where one substituent was still ortho to
the nitro group, giving good yields of the indole as the ma-
jor product.
Scheme 1
A range of simple substituted nitrobenzenes were first
used in the reaction. As expected, both ortho–methyl (1a)
and ortho–trifluoromethyl (1c) groups gave good yields
of the corresponding 7–substituted indoles when treated
with vinylmagnesium bromide in THF at ca. –40 °C for
between 30 minutes and one hour. (Interestingly, the yield
was lower with vinylmagnesium chloride by ca. 15% un-
der identical conditions). Employing a further excess of
Synlett 1999, No. 10, 1594–1596 ISSN 0936-5214 © Thieme Stuttgart · New York

LETTER
Synthesis of Novel Indole Derivatives: Variations in the Bartoli Reaction
1595
The surprising results were obtained when combining
some of these groups (particularly the methyl, trifluoro-
methyl and bromo groups) in the starting material. These
groups on their own, in either the meta– or para positions,
failed to yield any significant levels of indoles, vide supra.
However, it was found that combinations of them in the 3
and 4 positions gave modest yields of the desired indoles,
for example, 3–trifluoromethyl–4–methylnitrobenzene
produced 5–methyl–6–trifluoromethylindole (1g) when
treated with vinyl Grignard reagent (with only traces of
the other possible regioisomer detected). Comparable lev-
els of 5–methyl–6–bromoindole (1h) were similarly ob-
tained from 3–bromo–4–methylnitrobenzene under the
same conditions. The explanation for this is unclear at
present, particularly since the simple meta– and para–tri-
fluoromethylnitrobenzenes gave no indole products, al-
though it could be due to a subtle interplay of steric and
electronic effects favouring the cyclisation to the desired
position on the aromatic nucleus.
Acknowledgement
The authors wish to thank the staff of the Mass Spectrometry and
NMR units at SmithKline Beecham, Tonbridge for their help with
analyses and structure determination.
References and Notes
(1) Present Address: Department of Chemistry, The Open
University, Walton Hall, Milton Keynes, MK7 6AA, England.
(2) For example see: a) Palladium catalysed aryl halide/acetylene
coupling and base promoted cyclisation: Kondo, Y., Kojima,
S., Sakamoto, T. Heterocycles 1996, 43, 2741–2746; Kondo,
Y., Kojima, S., Sakamoto, T. J. Org. Chem. 1997, 62, 6507–
6511; b) Palladium mediated coupling and cyclisation
reactions of aryl halides with acetylenes: Ezquerra, J.,
Pedregal, C., Lamas, C., Barluenga, J., Pérez, M., García–
Martín, M. A., González, J. M. J. Org. Chem. 1996, 61, 5804–
5812; Rudisill, D. E., Stille, J. K. J. Org. Chem. 1989, 54,
5856–5866; c) Palladium catalysed couplings with boranes:
Satoh, M., Miyaura, N., Suzuki, A. Synthesis 1987, 373–377;
d) Via palladium catalysed annulations between iodoaniline
and ketones: Chen, C.–Y., Lieberman, D.R., Larsen, R.D.,
Verhoeven, T.R., Reider, P.J. J. Org. Chem. 1997, 62, 2676–
2677; e) Via Heck–type reactions: Kasahara, A., Izumi, T.,
Kikuchi, T., Xiao–ping, L. J. Heterocyclic Chem. 1987, 24,
1555–1556.
(3) a) Bartoli, G., Palmieri, G., Bosco, M., Dalpozzo, R.
Tetrahedron Lett. 1989, 30, 2129–2132; b) Bartoli, G., Bosco,
M., Dalpozzo, R., Palmieri, G., Marcantoni, E. J. Chem. Soc.
Perkin Trans. 1 1991, 2757–2761; c) Bosco, M., Dalpozzo,
R., Bartoli, G., Palmieri, G., Petrini, M. J. Chem. Soc. Perkin
Trans.II 1991, 657–663; d) Dobson, D., Todd, A., Gilmore, J.
Synth. Commun. 1991, 21, 611–617; e) Dobson, D., Todd, A.,
Gilmore, J. Synlett 1992, 79–80
(4) Giacomoni, J.C., Cambon, A., Rouvier, E. Bull. Soc. Chim.
France 1970, 3097–3098 (bromination) and Bandodakar,
B.S., Nagendrappa, G. Synthesis 1990, 843–844 (elimination).
(5) IR spectra were recorded neat or in solution on a Nicolet
Avatar 360 FT–IR machine. 1H and 13C NMR spectra were
recorded using either a Jeol GSX–400 or a Bruker DRX–400
using CDCl3 as solvent and TMS or solvent peaks as internal
references and are quoted in the form dH (integration,
multiplicity, coupling constant(s), assignment) or dC
(assignment). Coupling constants (J values) are quoted to one
decimal place with values in Hz. Mass spectra were recorded
on a HP–5890GC–VG–Masslab Trio–2 gas chromatography–
mass spectrometer.
(6) Typical procedure for the Bartoli reactions: The nitro–
compound (5 mmol) was placed in a two–necked round–
bottomed flask fitted with a gas inlet (argon) and rubber
septum. The flask was purged several times with argon before
adding THF (35–40 ml) and cooling to between –40 and –
45 °C. The Grignard reagent (3 eq.) was then added rapidly in
one portion to the THF solution and stirring continued for a
further 30 mins to 1 hour (exact length of time had little effect
on yield). Saturated ammonium chloride solution was added
to the reaction mixture (at ca. –40 °C) before allowing the
mixture to warm to room temperature. The mixture was
thoroughly extracted with diethyl ether (2 x 200 ml), the ether
extracts combined and thoroughly washed with further
ammonium chloride (300 ml), water (300 ml) and brine (300
ml) before drying (MgSO4) and concentrating in vacuo to give
a dark brown gum, which was purified by flash column
chromatography (hexane:ethyl acetate 9:1) to give the title
compound. Compounds were in good agreement with
literature data; new compounds gave satisfactory
In an attempt to improve the yields in these reactions the
effects of solvent and temperature were investigated. It
was found that warming the reaction (up to reflux in THF)
gave lower yields of indole (15% at 0 °C and 0% at re-
flux). Similarly lowering the temperature to –78 °C also
caused a decrease in indole yield (albeit less dramatic, to
16%) and –40 to –45 °C was found to be the optimum for
these reactions. As far as the reaction solvent was con-
cerned, only THF gave any significant levels of indole
products, even when employing an ortho–substituent to
assist in the reaction, as in the earlier examples. (Other
solvents tested included dry diethyl ether, hexane and tol-
uene at temperatures ranging from reflux down to –78 °C
but none gave any appreciable levels of indoles). Longer
reaction times had no effect on the yields of indole ob-
tained and it is believed that these reactions are very rapid
indeed, possibly proceeding via a radical/single electron
transfer mechanism and being complete almost immedi-
ately.3c
Finally the effects of the Grignard reagent were investi-
gated using more complex Grignards, (prepared by estab-
lished literature procedures4). It was found that these
carbocyclic Grignard reagents reacted with 2–nitrotolu-
ene, thus affording very rapid synthesis of a range of fused
tricyclic indole derivatives (1 l,m,n). (N.B. Although cy-
clic vinylbromide derivatives gave good yields, heteroar-
omatic Grignards, for example, thien–2–yl and N–
methylpyrrolidin–2–yl reagents, failed to give any fused
indoles, presumably due to the aromatic nature of these
compounds).
In summary we have investigated the scope and extended
the range of the Bartoli reaction to give a number of syn-
thetically useful substituted indoles which are being fur-
ther utilised in our studies.
Synlett 1999, No. 10, 1594–1596 ISSN 0936-5214 © Thieme Stuttgart · New York

1596
A. P. Dobbs et al.
LETTER
compounds and Grignards may be explosive in nature,
spectroscopic and analytical data, as illustrated by the
following representative compounds:4,7–Dibromoindole
(2f): Pale yellow oil, slowly crystallising to pale yellow/
brown solid; Found (M 79Br, 79Br)+272.8795. C8H5N79Br79Br
requires 272.8789; dH (400 MHz) 6.63 (1H, C(3)H), 7.15 (2H,
m, C(5)H & C(6)H), 7.21 (1H, d, J C(2)H), 8.36 (1H, s (br),
N–H); dC (100 MHz) 104.1 (C(7)Br), 104.8 (C(3)H), 114.3
(C(4)Br), 124.2 & 125.4 (C(5)H & C(6)H), 129.9 (C(3a)),
135.0 (C(7a));5–Methyl–6–trifluoromethylindole (2g):Dark
brown oil, crystallising to a dark brown/black solid; Found
M+199.0619. C10H8NF3 requires 199.0609; dH (400 MHz)
2.55 (3H, –CH3), 6.51 (1H, d, J 5.6, C(3)H), 7.28 (1H, d, J 5.6,
C(2)H), 7.51(1H, s, C(4)H), 7.67(1H, C(7)H), 8.22 (1H, s
(br), N–H); m/z:199.0 [(M)+, 100%], 130.2 {(M–CF3)+, 68%].
(N.B. Care should be taken in these reactions as it has been
reported7 that the reaction between aromatic trifluoromethyl
although no such problems were encountered during the
course of this work); 2,3,7–Trimethylindole (2k): Yellow oil,
slowly crystallising to pale yellow solid; Found M+159.1046.
C11H13N requires 159.1048; dH (400 MHz) 2.21 (3H, s, –CH3),
2.36 (3H, s, –CH3), 2.43 (3H, s, –CH3), 6.90 (1H, d, J 7.1,
C(5)H), 7.00 (1H, dd, J 7.3, C(6)H), 7.32 (1H, d, J 7.5,
C(4)H).
(7) Laird, T. Org. Proc. Res. Dev. 1998, 2, 210–213; Beck, C.M.,
Park, Y.–J., Crabtree, R.H. J. C. S. Chem. Commun. 1998,
693–694 and reference therein.
Article Identifier:
1437-2096,E;1999,0,10,1594,1596,ftx,en;L11199ST.pdf
Synlett 1999, No. 10, 1594–1596 ISSN 0936-5214 © Thieme Stuttgart · New York