Synthesis of indoles by intermolecular cyclization of unfunctionalized nitroarenes and alkynes, catalyzed by palladium-phenanthroline complexes

Article abstract of DOI:10.1021/jo060073m

Palladium-phenanthroline complexes efficiently catalyze the reaction of nitroarenes with arylalkynes and CO to give 3-arylindoles by an ortho-C-H functionalization of the nitroarene ring. Both electron-withdrawing and electron-donating substituents are tolerated on the nitroarene, except for bromide and activated chloride. Nitroarenes bearing electron-withdrawing substituents react faster, but the selectivity of the reaction depends on both polar and radical stabilization effects. Among those tested, only arylalkynes afforded indoles under the investigated conditions. The reaction mechanism was partly investigated. The kinetics is first order in nitroarene concentration and the rate-determing step of the cycle is the initial nitroarene reduction. No primary isotope effect is observed on either rate or selectivity, implying that the cyclization step is fast.

Full text of DOI:10.1021/jo060073m

Synthesis of Indoles by Intermolecular Cyclization of
Unfunctionalized Nitroarenes and Alkynes, Catalyzed by
Palladium-Phenanthroline Complexes
Fabio Ragaini,* Andrea Rapetti, Elena Visentin, Michela Monzani, Alessandro Caselli, and
Sergio Cenini
Dipartimento di Chimica Inorganica, Metallorganica e Analitica, UniVersita` di Milano, and
ISTM-CNR, Via Venezian 21, 20133 Milano, Italy
fabio.ragaini@unimi.it
ReceiVed January 12, 2006
Palladium-phenanthroline complexes efficiently catalyze the reaction of nitroarenes with arylalkynes
and CO to give 3-arylindoles by an ortho-C-H functionalization of the nitroarene ring. Both electron-
withdrawing and electron-donating substituents are tolerated on the nitroarene, except for bromide and
activated chloride. Nitroarenes bearing electron-withdrawing substituents react faster, but the selectivity
of the reaction depends on both polar and radical stabilization effects. Among those tested, only arylalkynes
afforded indoles under the investigated conditions. The reaction mechanism was partly investigated. The
kinetics is first order in nitroarene concentration and the rate-determing step of the cycle is the initial
nitroarene reduction. No primary isotope effect is observed on either rate or selectivity, implying that the
cyclization step is fast.
Introduction
application is severely limited by the low activity of the catalyst.
Despite the fact that the substrate/Ru ratio is only 10, the reaction
takes 48 h at 170 °C to be complete. Some experimental
evidence was provided that the reaction proceeds via the
nitrosoarene, which reacts with the alkyne in an off-metal
reaction to afford the N-hydroxyindole.7,8 Palladium-phenan-
throline complexes are at present the most active catalysts for
reductive carbonylation of nitroarenes,9-11 and we have recently
shown that [Pd(phen)2][BF4]2 is an excellent catalyst for the
synthesis of oxazines and N-arylpyrroles from nitroarenes and
conjugated dienes.12 This reaction also apparently proceeds by
The indole skeleton is part of many pharmaceutically active
molecules, and its synthesis continues to attract much interest,1
but the overwhelming majority of the reported synthetic
strategies require the availability of a nitrogen-functionalized
arene (e.g., an arylamine or a nitroarene2-5) also having a
suitable functional group in the ortho position with respect to
the nitrogen substituent. Penoni and Nicholas have recently
reported that [RuCp*(CO)2]2 catalyzes the reaction of nitroarenes
with alkynes to give indoles.6 The presence of a substituent in
the ortho position of the nitroarene is not required. Selectivities
are only moderate, but the procedure saves several synthetic
steps. The reaction is new and interesting, but, in our view, its
(7) Penoni, A.; Volkmann, J.; Nicholas, K. M. Org. Lett. 2002, 4, 699-
701.
(8) Just before submission of this paper, the same authors have reported
the extension of the reaction based on the use of the nitrosoarenes to the
synthesis of N-methoxyindoles: Penoni, A.; Palmisano, G.; Broggini, G.;
Kadowaki, A.; Nicholas, K. M. J. Org. Chem. 2006, 71, 823-825.
(9) Ragaini, F.; Gasperini, M.; Cenini, S. AdV. Synth. Catal. 2004, 346,
63-71.
* Corresponding Author. Fax: (+39)0250314405.
(1) Sundberg, R. J. Indoles; Academic Press: London, 1996.
(2) Cenini, S.; Ragaini, F. Catalytic ReductiVe Carbonylation of Organic
Nitro Compounds; Kluwer Academic Publishers: Dordrecht, The Nether-
lands, 1996.
(3) Soderberg, B. C. G. Curr. Org. Chem. 2000, 4, 727-764.
(4) Davies, I. W.; Smitrovich, J. H.; Sidler, R.; Qu, C.; Gresham, V.;
Bazaral, C. Tetrahedron 2005, 61, 6425-6437.
(5) Ragaini, F.; Cenini, S.; Gallo, E.; Caselli, A.; Fantauzzi, S. Curr.
Org. Chem., in press.
(10) Ragaini, F.; Cognolato, C.; Gasperini, M.; Cenini, S. Angew. Chem.,
Int. Ed. 2003, 42, 2886-2889.
(11) Gasperini, M.; Ragaini, F.; Cazzaniga, C.; Cenini, S. AdV. Synth.
Catal. 2005, 347, 105-120.
(12) Ragaini, F.; Cenini, S.; Brignoli, D.; Gasperini, M.; Gallo, E. J.
Org. Chem. 2003, 68, 460-466.
(6) Penoni, A.; Nicholas, K. M. Chem. Commun. 2002, 484-485.
10.1021/jo060073m CCC: $33.50 © 2006 American Chemical Society
Published on Web 04/11/2006
3748
J. Org. Chem. 2006, 71, 3748-3753

Synthesis of Indoles by Intermolecular Cyclization
TABLE 1. Synthesis of Indoles from 1-Phenylpropyne and
Different Nitroarenesa
SCHEME 1
ArNO2 conv.b,c indole sel.d
ArNO2
R1
(%)
(%)
σHamm
σJJ•
1a
1b
1c
1d
1e
1f
H
Cl
68.4
25.8
74.1
90.4
52.3
1.6
29.7 (3ab)
21.5 (3bb)
36.7 (3cb)
28.7 (3db)
44.7 (3eb)
0
0
0.23 0.22
-0.17 0.15
0.45 0.33
CH3
C(O)OCH3
OCH3
-0.27 0.23
1g
1h
1i
OC(O)CH3
NO2
CN
100
100
100
31.4 (3gb)
24.3 (3hb)
14.6 (3ib)
8.2 (3jb)
0.31 0.35
0.78 0.36
0.66 0.42
1j
44.7
a Experimental conditions: [Pd(phen)2][BF4]2 ) 4.7 × 10-3 mmol, mol
ratios Pd/phen/nitroarene/PhCtCCH3 ) 1:20:300:1800, T ) 170 °C, PCO
) 60 bar, in DME (10 mL), for 10 h. b Calculated with respect to the starting
nitroarene. c Measured by GC with naphthalene as the internal standard.
1
d With respect to the reacted nitroarene, measured by H NMR, with 2,4-
dinitrotoluene as the internal standard.
phenylacetylene-coupling products strongly decreases with the
CO pressure, and 60 bar was chosen as the operating pressure
for this reason. Under these conditions (170 °C, 60 bar), the
alkyne dimers are barely detectable by GC-MS. (iv) Medium
polarity, aprotic solvents, THF, and 1,2-dimethoxyethane (DME)
are the best solvents. More polar solvents such as DMF give a
faster but less-selective reaction, while lower polarity solvents
(diethyl ether, dioxane) give a slower reaction without any
increase in selectivity with respect to THF and DME. (v) The
addition of bases or acids strongly inhibits the reaction. (vi)
The amount of phenanthroline and phenylacetylene were also
optimized, and the values employed for the reactions in Table
1 are the best ones. Under these conditions, a complete
conversion of nitrobenzene was achieved in 3 h with a 0.33
mol % catalyst and affording 3-phenylindole in a 53% yield. A
comparison with the published data shows that [Pd(phen)2]-
[BF4]2 is almost 500 times more active than [RuCp*(CO)2]2
(based on the number of metal atoms) at the same temperature,6
although the selectivity in indole is only marginally improved
(39% with ruthenium). The palladium catalyst is also cheaper
and easier to prepare, making the present synthetic strategy now
competitive.
a metal-catalyzed reduction of the nitroarene to nitrosoarene,
followed by an off-metal reaction with the diene. Thus, we
considered that the same catalyst would also perform well in
the synthesis of indoles. This expectation was confirmed to be
true, although the experimental conditions originally optimized
for the oxazines were not suitable for the synthesis of indoles.
Results and Discussion
Synthetic Results. A list of the nitroarenes and alkynes
investigated in this study is shown in Scheme 1.
Nitrobenzene (1a) and phenylacetylene (2a) were chosen for
the optimization of the experimental conditions. 3-Phenylindole
(3aa; in the numbering of indoles, the first letter refers to the
nitroarene and the second to the alkyne employed) was the only
detected product containing both the nitrobenzene- and the
phenylacetylene-derived moieties. Not a trace of the isomeric
2-phenylindole was detected. This is noteworthy, because most
reported syntheses of arylindoles selectively afford the 2-sub-
stituted product, and few methods exist for the 3-substituted
isomer.13,14 The major nitrobenzene-derived byproduct was
azoxybenzene, but minor amounts of azobenzene and aniline
were also formed. Diphenylurea was detected, but could not be
quantified by gas chromatographic analysis, because it decom-
posed during the analysis. Small amounts of a phenylacetylene
dimer and of the coupling compound PhCtCsCtCPh were
the only alkyne-derived byproducts.
The main results of the optimization study are the following.
(i) The reaction is slow at temperatures below 130 °C, and its
rate increases with an increase in the temperature up to 190
°C. (ii) The selectivity is almost constant in the range 130-
170 °C but decreases at higher temperatures. Thus, we chose
170 °C as the standard temperature. (iii) The selectivity in indole
is almost independent of the CO pressure between 20 and 60
bar, but decreases outside this range. However, the amount of
Some experiments were also conducted employing nitroben-
zene as substrate, but in the presence of 4-toluidine. In no case
were indoles, azo-, or azoxy-arenes containing the methylated
arene detected, showing that anilines are not intermediates in
the formation of any of these products.
The results of a series of experiments that were run with
1-phenylpropyne (2b), but different nitroarenes, are reported
in Table 1. This alkyne was chosen because the presence of a
methyl group allowed the quantification of the indole produced
1
by H NMR of the crude reaction mixture without workup,
although all indoles were later isolated and fully characterized.
The experimental conditions optimized with 1a and 2a were
employed, except for the reaction time that was increased to
10 h as a result of the lower reaction rate observed with this
disubstituted alkyne. In all cases, only the regioisomer with the
aryl group in the 3-position was obtained. In the table, the
Hammett σ values for the substituents on the nitroarene15 are
also reported, together with a parameter, σJJ•, which describes
radical stabilization effects.16
(13) Lane, B. S.; Brown, M. A.; Sames, D. J. Am. Chem. Soc. 2005,
127, 8050-8057.
(14) Taber, D. F.; Tian, W. W. J. Am. Chem. Soc. 2006, 128, 1058-
(15) Hansch, C.; Leo, A.; Taft, R. W. Chem. ReV. 1991, 91, 165-195.
(16) Jiang, X. K. Acc. Chem. Res. 1997, 30, 283-289.
1059.
J. Org. Chem, Vol. 71, No. 10, 2006 3749

Ragaini et al.
TABLE 2. Synthesis of Indoles from 4-Nitrotoluene and Different
Alkynesa
was attempted. Because the C-Br bond is activated more easily
than the C-Cl bond, the deactivation effect should be even more
evident. In accord with our hypothesis, no indole was formed
at all, and the amount of 4-bromonitrobenzene at the end of
the reaction was indistinguishable (by GC) from the initially
added one.21 Second, the reaction of [Pd(phen)2][BF4]2 with
respect to 4-chloronitrobenzene was tested under the catalytic
reaction conditions but in the absence of any alkyne. Because
the catalyst amount under usual conditions is too small (less
than 1 mg of palladium) to isolate any organometallic product,
the catalyst amount was increased 10-fold and the reaction time
was also prolonged to 10 h. At the end of the reaction, an insol-
uble precipitate was present, which was dissolved in DMSO-d6
group in the
4-position on 1c convb,c indole sel.d
alkyne
ArCtCH
(%)
(%)
σHamm
σJJ•
2a
2e
2h
2i
2j
2k
2l
H
100
100
100
100
100
100
100
29.9 (3ca)
38.1 (3ce)
23.2 (3ch)
33.7 (3ci)
16.1 (3cj)
11.9 (3ck)
9.3 (3cl)
0
0
NH2
OCH3
Cl
C6H5
CF3
-0.66
-0.27
0.23
-0.01
0.54 -0.01
1.00e
0.23
0.22
0.47
a Experimental conditions: [Pd(phen)2][BF4]2 ) 4.7 × 10-4 mmol, in
DME (1 mL); other conditions as in Table 1. b-d See corresponding
footnotes in Table 1. e Value for -NMe2.
1
and analyzed by H NMR. Comparison with the signals of an
authentic species showed that about half of the residue is indeed
Pd(phen)Cl2, although a second major species and at least a
third one present in a much lower amount are present also. These
last two species could not be identified but are not the starting
complex.22 Thus, it can be conclusively stated that the lower or
null reactivity observed with halonitroarenes is due to the
carbon-halogen bond activation by the palladium catalyst.
As far as the alkyne series is concerned, only arylalkynes
afforded the indole. Nitroarene conversion was fast (100%
conversion) in the presence of alkylalkynes (2c, 2g), but not
even a trace of the cyclized product was formed. An alkeny-
lalkyne (2f) gave a mixture of products, apparently mostly
derived from a “diene-type” reactivity.12 Dimethylacetylenedi-
carboxylate (2d) completely inhibited the reaction. Because this
alkyne is known to dimerize in the presence of palladium
complexes with chelating nitrogen ligands,23 it is likely that a
related reactivity is involved here. Bistrimethylsilylacetylene
(2n) gave a low conversion (30%) and no indole. Several
substituents on the aryl ring of the arylalkynes were tolerated,
including a free amino group. This is noteworthy, because many
synthetic procedures for the synthesis of indoles would not
tolerate such a functional group. Surprisingly, the reaction did
not proceed with tolylacetylene (2m). However, the com-
mercially available compound was found (by GC-MS) to be
contaminated by brominated impurities (2-bromo-4-methylsty-
rene, â-bromo-4-methylstyrene, and bromomethyl-4-tolyl ke-
tone) that would cause catalyst deactivation (see above). After
we purified 2m to the point that the impurities were no longer
observable by GC-MS, the catalytic reaction partially proceeded
and the formation of the expected indole (3am) was observed
by GC-MS. However, the reaction was still incomplete. We note
that at the alkyne/palladium ratio employed in the catalytic
reactions, a 0.05 mol % amount of brominated impurities in
the alkyne would be enough to deactivate all of the catalyst.24
FIGURE 1. Nitroarene conversion against the Hammett σ constant
for the reactions in Table 1.
Another series of experiments were run keeping the nitroarene
(4-nitrotoluene) constant and changing the alkyne. Positive
results are reported in Table 2. The reactions were performed
under the same experimental conditions of Table 1, but on a
smaller scale.
From the results reported in Table 1, it is immediately evident
that both electron-withdrawing and electron-donating substitu-
ents in the para position of the nitroarene are tolerated (Figure
1), but a methyl group in the ortho position slows down the
reaction and lowers the selectivity.17 In the case of 2-chloro-
3-nitropyridine (1f) only, the reaction failed completely, and
the substrate was recovered essentially unaltered.
Because the reactivity of 4-chloronitrobenzene was also lower
than expected based on the electron-withdrawing power of
chlorine (Figure 1), we suspected that activation of the C-Cl
bond in the nitroarene was occurring. We recall that compounds
of the kind Pd(phen)X2 (X ) halogen) are notoriously in-
active in nitroarene reductive carbonylation reactions2,18 and
that activation of the C-Cl bond by palladium is known to be
easier when electron-withdrawing groups are present on the
aryl ring.19,20 The C-Cl bond in 1f is, thus, doubly activated
by the contemporary presence of a pyridine ring and a nitro
group. To test this assumption, two experiments were under-
taken. First, a catalytic reaction employing 4-bromonitrobenzene
A methyl group on the other terminal position of phenyl-
acetylene (2b) slows down the reaction (compare Table 1),
but the second phenyl group in diphenylacetylene (2e) acceler-
ates it.
(21) Since only a 0.33 mol % catalyst was employed, a complete
conversion to Pd(phen)Br2 would consume only 0.66% of the starting
nitroarene, making it impossible to detect its consumption.
(22) Wehman, P.; Kaasjager, V. E.; Delange, W. G. J.; Hartl, F.; Kamer,
P. C. J.; van Leeuwen, P. W. N. M.; Fraanje, J.; Goubitz, K. Organometallics
1995, 14, 3751-3761.
(23) van Belzen, R.; Elsevier, C. J.; Dedieu, A.; Veldman, N.; Spek, A.
L. Organometallics 2003, 22, 722-736.
(24) Purification of 3cm proved difficult because this indole and the
starting 1c shows very close Rf values in a chromatographic separation. An
analytically pure sample of 3cm could not be obtained, and accordingly,
this compound is not mentioned in the tables or in the Experimental Section.
(17) The lower selectivity may be due to the availability of only one
ortho position on the nitroarene, but the lower conversion is likely due to
a more difficult reduction of the nitroarene, which appears to be the slow
step of the reaction (see also later). Due to the low yield of this indole, it
could not be isolated in a pure form, and the identification of 3jb is to be
considered only as a proposal.
(18) Gasperini, M.; Ragaini, F.; Remondini, C.; Caselli, A.; Cenini, S.
J. Organomet. Chem. 2005, 690, 4517-4529.
(19) Stille, J. K.; Lau, K. S. Y. Acc. Chem. Res. 1977, 10, 434-442.
(20) Portnoy, M.; Milstein, D. Organometallics 1993, 12, 1665-1673.
3750 J. Org. Chem., Vol. 71, No. 10, 2006

Synthesis of Indoles by Intermolecular Cyclization
FIGURE 3. Plot of log(k/kH) vs Hammett σ, where k is the second-
order rate constant for the carbonylation of all nitroarenes for which a
total conversion was not observed and kH is the same constant for
nitrobenzene. The straight line (F ) +0.61, R2 ) 0.85) is the best linear
fit to the data, with the exclusion of the point corresponding to
4-ClC6H4NO2.
FIGURE 2. Plot of log([PhNO2]i/[PhNO2]f) vs reaction time for a series
of reactions between 1a and 2a. [Pd(phen)2][BF4]2 ) 1.67 × 10-3 mmol;
mol ratios 1a/2a/phen/Pd ) 900:5400:20:1, at 170 °C, PCO ) 60 bar,
and in 1,2-dimethoxyethane (10 mL). The straight line is the best linear
fit to the data (R2 ) 0.96).
Reactivity Trends and Mechanistic Aspects. Before analyz-
ing in more detail the results, it is useful to recall some
mechanistic information. Reduction of nitroarenes by CO
initially produces nitrosoarenes, and the reaction is accelerated
by electron-withdrawing substituents.2 No mechanistic informa-
tion is available on the reaction between nitrosoarenes and
alkynes, but the reaction with alkenes (nitroso-ene reaction)25
has been investigated even from a theoretical point of view.26
The nitrosoarene and alkene interact to initially generate a
polarized adduct with partial diradical character. This in turn
rearranges by abstracting a hydrogen atom on the allylic position
of the olefin to give the finally observed allylhydroxylamine.
The nitroso-ene reaction is also favored by electron-withdrawing
substituents on the nitrosoarene.25 In the case of the reaction
between nitrosoarenes and alkynes, propargylic hydrogens are
either not present or their abstraction appears not to be possible,
probably for geometric reasons, so that the reaction takes another
pathway.
To get more information on the reaction mechanism, the
kinetics of the reaction between 1a and 2a was investigated.
The previously optimized conditions were employed, but the
catalyst amount had to be decreased to 0.11 mol % to avoid a
complete consumption of the nitroarene. Plotting log([PhNO2]i/
[PhNO2]f) versus reaction time ([PhNO2]i and [PhNO2]f are the
nitrobenzene concentration, respectively, at the beginning and
at the end of the reaction) for a series of reactions between 1a
and 2a, a straight line is obtained, indicating a first-order
dependence of the reaction rate on the nitroarene concentration
(Figure 2).
We can now examine the synthetic results in more detail. As
previously mentioned, a quick examination of the data in Table
1 immediately shows that nitroarenes having electron-withdraw-
ing substituents generally showed higher reactivity, and for most
of them, a complete conversion was observed (Figure 1).
Because we have determined that the reaction rate is first order
in nitroarene concentration, the kinetic constants (assuming also
a first-order rate dependence on palladium concentration, as is
always found for this kind of catalytic systems) for all reactions
that did not reach complete conversion could be determined,
and a plot of log(k/kH) versus the Hammett σ constant of the
substituents, where k is the second-order rate constant for a
generic nitroarene and kH is the same constant for nitrobenzene,
could be drawn (Figure 3). Although the number of useful data
points is limited, it is evident that a linear correlation exists,
with the remarkable exception of the point due to 4-chloroni-
trobenzene (1b), and a positive slope (F ) +0.61) is observed.
The reason for the anomalous behavior of halogen-substituted
nitroarenes has been already discussed previously. On the other
hand, all the other data, first-order dependence of the rate on
nitroarene concentration, positive slope in the Hammett plot,
and lower reactivity of 2-nitrotoluene with respect to 4-nitro-
toluene,27,28 indicate that the rate-determining step of the cycle
is the initial activation of the nitroarene and are consistent with
an electron transfer from the metal to the nitroarene in this step.
The latter is a common feature of these reactions and has been
observed for all other systems for which this step has been
investigated.2,29,30
The selectivity in indole is correlated with the ratio between
the rate of formation of indole and the rate of formation of
possible byproducts. Other factors being constant, this ratio is
the function of the substituents on the nitroarene. No evident
clear correlation exists between the log(selectivity in indole)
and the Hammett σ constant or the σJJ• alone for the reactions
in Table 1. However, a multivariate correlation including both
individuates a straight line passing close to all data except for
the points for the 4-chloro and the 4-cyano groups (Figure 4).
The reason for the deviation of the chlorinated derivative has
already been discussed. The reason for which also the cyano
substituent gives a lower selectivity than expected is not obvious,
but it must be recalled that nitriles can coordinate to palladium.
It may be also noticed that the cyano group itself contains a
triple bond that may enter in competition with the CtC one,
(27) It is well-known that ortho-substituted nitroarenes are more difficult
to reduce than the corresponding meta- and para-substituted isomers, because
steric hindrance causes a tilting of the nitro group out the arene plane,
making charge delocalization in the reduced species more difficult.28
(28) Wheeler, O. H. Can. J. Chem. 1963, 41, 192-194.
(29) The F value determined in this study is much lower than that
measured by Gladfelter in a study on the reduction of nitroarenes by
Ru(CO)3(Ph2P-CH2CH2PPh2) (+3.45).30 However, this difference is to be
expected because Gladefter’s study was conducted at 25 °C, whereas our
reactions were performed at 170 °C. Because the nitroarene reduction
becomes easier at higher temperatures, both ∆G and ∆∆G values for
different nitroarenes will decrease. This will accordingly affect the
differences in rate constants and result in a lower slope in the Hammett
plot.
(25) Adam, W.; Krebs, O. Chem. ReV. 2003, 103, 4131-4146.
(26) Leach, A. G.; Houk, K. N. Org. Biomol. Chem. 2003, 1, 1389-
1403.
(30) Skoog, S. J.; Gladfelter, W. L. J. Am. Chem. Soc. 1997, 119, 11049-
11060.
J. Org. Chem, Vol. 71, No. 10, 2006 3751

Ragaini et al.
TABLE 3. Use of Deuterated Nitrobenzenea
PhN(O)dNPh
PhNO2-d5: PhNO2 conv.b,c (%) indole sel.c,d (%)
PhNO2
sel.c,d (%)
(H/D mol ratio)e
(H/D mol ratio)e (H/D mol ratio)e
0:1
49.5
53.5
20.4
1:0
0.5:0.5
50.5
49.9 (0.85)
50.3
48.4 (1.08)
21.2
18.8 (1:2.3:1.3)f
a Experimental conditions: [Pd(Phen)2][BF4]2 ) 4.7 × 10-4 mmol; mol
ratios Pd/phen/PhNO2/PhCtCH ) 1:20:300:1800, T ) 170 °C, PCO ) 60
bar, in DME (1 mL) for 1 h. b Calculated with respect to the starting
nitroarene. c Measured by GC with naphthalene as the internal standard.
d With respect to the reacted nitrobenzene. e Mol ratio deuterated/undeu-
terated compound in the final solution. For details on the calculation of
this ratio and the response factor of deuterated compounds, see the
Supporting Information. f The ratio d0/d5/d10. Note that the statistical ratio
in the absence of any isotope effect is 1:2:1.
FIGURE 4. Plot of log(indole selectivity %/100) vs the best multi-
variate combination of Hammett σ and σJJ• constants for the data in
Table 1 (R2 ) 0.985). The data for 1b and 1d are shown but were not
included in the calculation of the best fit.
although, at this stage, we could not detect any product clearly
assignable to such a competitive reaction. The equation describ-
ing the best fit to all other points (R2 ) 0.985, Figure 3) has
the form:
SCHEME 2
log(indole selectivity %/100) )
-0.292σ + 0.368σJJ• - 0.529
While the absolute values of the coefficient are somewhat
arbitrary, it is important that the coefficients associated with
the two constants are comparable in magnitude, indicating that
both polar and radical stabilization effects are relevant in
determining the selectivity.
An analogous data analysis was attempted with the results
in Table 2, but no clear correlation could be obtained. However,
it should be considered that alkynes can coordinate to palladium,
and the role of such a complex (intermediate or dead branch of
the catalytic cycle) is not known at this stage. That the alkyne
plays a role before the rate-determining activation of the
nitroarene is indicated by the fact that the conversion of the
latter depends on both the alkyne amount and its identity (see
also later).
To investigate whether the cyclization step was important in
determining the selectivity of the reaction, we investigated the
isotopic effect running three parallel reactions employing either
pure C6H5NO2 and C6D5NO2 or a 1:1 mixture of the two as
substrates. Results are reported in Table 3. Close to the
experimental error, both the nitroarene conversion and the
selectivity in indole were the same in all cases. Even in the
competition experiment, an isotope effect of around 1.08-1.15
can explain the data. This is consistent with a secondary
deuterium effect but is inconsistent with a primary isotopic
effect, indicating that the selectivity-determining step is not the
cyclization step.
We can now draw a tentative reaction scheme on the basis
of the supposition, following the work by Nicholas et al.,6,7 that
the reaction between the nitrosoarene and the alkyne occurs
outside the coordination sphere of the metal (Scheme 2).
By an analogy with the reactions with olefins, the nitrosoarene
should interact reversibly with the alkyne to give an intermediate
having a polarized diradical character.26 Only two of several
possible resonance structures are depicted in the scheme. Note
that this intermediate (A) has been drawn with a linearly
arranged benzylic carbon atom because this is the geometry that
has been determined to be most stable both for R-styryl cations
(by theoretical calculations31) and R-styryl radicals (by EPR32
and fluorescence spectroscopy33). Cyclization would give the
hydroxyindole, which, in the presence of CO and the catalyst,
is reduced to indole. Alternatively, the nitrosoarene can react
with itself to give, in the presence of CO and the catalyst, the
azoxyarene, a well-precedented reaction.2
The selectivity should be higher if steps (a) and (b) are fast
with respect to (c). Step (b) must be fast because no isotope
effect is observed. The radical character of the intermediate
adducts nicely explains the need for an aryl group on the alkyne,
the regioselectivity (the aryl ring stabilizes charges or a radical
in the R position), and also the beneficial role of radical-
stabilizing groups on the nitroarene. On the other hand, the
polarization that can be inferred from the sign of the Hammett
σ coefficient of the plot in Figure 4 is opposite to that depicted
in the scheme and would place a negative charge on the benzylic
carbon. Such a polarization is unlikely unless the intermediate
is bound to palladium through that carbon atom.
To gain more information on the cyclization step, we
performed some stoichiometric experiments. Since palladium-
(0) carbonyl complexes, such as those that are likely formed
by reduction of [Pd(phen)2][BF4]2, under the reaction conditions
are not stable, we employed Pd(phen)(dba) as a starting material.
(31) Galli, C.; Gentili, P.; Guarnieri, A.; Kobayashi, S.; Rappoport, Z.
J. Org. Chem. 1998, 63, 9292-9299.
(32) Bennett, J. E.; Howard, J. A. Chem. Phys. Lett. 1971, 9, 460-462.
(33) Brocklehurst, B.; Robinson, J. S.; Tawn, D. N. Chem. Phys. Lett.
1972, 12, 610-611.
3752 J. Org. Chem., Vol. 71, No. 10, 2006

Synthesis of Indoles by Intermolecular Cyclization
dinitrogen, frozen at - 78 °C with dry ice, evacuated, and filled
with dinitrogen, after which the solvent was added. After the solvent
was also frozen, the liner was closed with a screw cap having a
glass-wool-filled open mouth, which allows for gaseous reagents
exchange, and rapidly transferred to a 200-mL stainless steel
autoclave with magnetic stirring. The autoclave was then evacuated
and filled with dinitrogen three times. CO was then charged at room
temperature at the required pressure, and the autoclave was
immersed in an oil bath preheated at the required temperature. Other
experimental conditions are reported in Table 1. At the end of the
reaction, the autoclave was cooled with an ice bath and vented,
and the products were analyzed by gas chromatography (naphtha-
lene as an internal standard) to determine the nitroarene conversion
and, in the case of the reactions involving nitrobenzene and
phenylacetylene, the amounts of 3aa, aniline, azobenzene, and
azoxybenzene formed. Then 2,4-dinitrotoluene was added as an
internal standard, and the solution was evaporated in vacuo. The
residue was dissolved in CDCl3 and analyzed by 1H NMR using a
delay of 10 s. The obtained value for the amount of indole formed
was corrected for the small amount of solution withdrawn for the
gas chromatographic analysis. The reactions in Tables 2 and 3 were
performed in a similar way, but, in this case, three 10 mm wide ×
40 mm high test tubes were employed instead of the glass liner,
each having a miniature glass-wool-filled screw cap similar to the
one of the larger liner. The three test tubes were located in the
holes of an aluminum block designed to fit the autoclave. Other
operations were analogous to the ones described above except that
stock solutions of the catalyst, the ligand, and the nitroarene were
prepared, and the reagent amounts were measured by volume to
avoid the errors in weighing very small amounts of materials.
Because the catalyst was not completely soluble at rt in some cases,
the suspension was irradiated with ultrasounds immediately before
the withdrawal to homogeneously disperse the catalyst. Stock
solutions were also employed in the experiments with deuterated
nitrobenzene. For the competitive experiment, equal volumes of
stock solutions (0.5 mL each) in PhNO2 and PhNO2-d5 were added.
We have previously reported that nitrosobenzene reacts with
this complex to give an insoluble and apparently polymeric
compound34 that cannot be employed as a model for any
intermediate in the indole synthesis. During this work, we tried
to isolate an alkyne complex of the kind Pd(phen)(alkyne)
(alkyne ) 2a, 2b, or 2e) by the reaction of Pd(phen)(dba) with
the alkyne in THF. An immediate reaction was observed in all
cases and with different reagent ratios, but insoluble and
apparently polymeric products were always observed. It should
be noted that only very few palladium-alkyne complexes have
been isolated in the solid state35,36 or even characterized in
solution and none of them having only nitrogen ligands. Thus,
although reactivity data suggest an involvement of the metal
even in the cyclization step, the exact mechanism by which this
occurs cannot be individuated at the moment.
Conclusions
In conclusion, we have shown that a palladium catalyst is
more convenient than the reported ruthenium one for the
intermolecular condensation of nitroarenes and alkynes to afford
indoles. Several functional groups are tolerated on the nitroarene,
except for bromide and activated chloride. The presence of a
C-Br bond is especially deleterious, and brominated compounds
should not be present, even in trace amounts. As far as the
alkyne is concerned, only arylalkynes gave the indole among
the substrates tested, and the aryl ring was selectively found at
the 3-position of the indole. Some mechanistic features have
been explained, and more work is in progress to clarify the
remaining ones and to improve the selectivity of the reaction.
Experimental Section
Catalytic Reactions. In a typical reaction, the reagents (see
Tables 1 and 2) were quickly weighed in a glass liner. The liner
was placed inside a Schlenk tube with a wide mouth under
Acknowledgment. We thank the MIUR (PRIN 2005035123)
for financial support.
Supporting Information Available: General Experimental
Section; details on the quantification of deuterated compounds;
characterization data for all new compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
(34) Gallo, E.; Ragaini, F.; Cenini, S.; Demartin, F. J. Organomet. Chem.
1999, 586, 190-195.
(35) Farrar, D. H.; Payne, N. C. J. Organomet. Chem. 1981, 220, 239-
250.
(36) Dervisi, A.; Edwards, P. G.; Newman, P. D.; Tooze, R. P. J. Chem.
Soc., Dalton Trans. 2000, 523-528.
JO060073M
J. Org. Chem, Vol. 71, No. 10, 2006 3753