Trisannulated benzene derivatives by acid catalyzed aldol cyclotrimerizations of cyclic ketones. methodology development and mechanistic insight

Article abstract of DOI:10.1021/jo070080q

(Chemical Equation Presented) Several factors that contribute to the success of aldol cyclotrimerizations have been clarified as part of an effort to shed light on the inner workings of this century old reaction. The use of 4,7-di-tert-butylacenaphthenone (11) as a mechanistic probe molecule has led to intriguing discoveries about temperature, solvent, and solubility effects. Solvents that are both polarizable and somewhat polar, e.g., o-dichlorobenzene (ODCB), work best for the aromatic ketones examined. Certain Bronsted acids were found to work better than Lewis acids as catalysts for the archetypal aldol cyclotrimerization of indanone (2) in aprotic solvents, and a strong dependence on the pK_a of the acid was observed. A standardized protocol, using p-toluenesulfonic acid monohydrate, is shown to work well in a number of test cases.

Full text of DOI:10.1021/jo070080q

Trisannulated Benzene Derivatives by Acid Catalyzed Aldol
Cyclotrimerizations of Cyclic Ketones. Methodology Development
and Mechanistic Insight
Aaron W. Amick and Lawrence T. Scott*
Merkert Chemistry Center, Boston College, Chestnut Hill, Massachusetts 02467-3860
lawrence.scott@bc.edu
ReceiVed January 13, 2007
Several factors that contribute to the success of aldol cyclotrimerizations have been clarified as part of
an effort to shed light on the inner workings of this century old reaction. The use of 4,7-di-tert-
butylacenaphthenone (11) as a mechanistic probe molecule has led to intriguing discoveries about
temperature, solvent, and solubility effects. Solvents that are both polarizable and somewhat polar, e.g.,
o-dichlorobenzene (ODCB), work best for the aromatic ketones examined. Certain Brønsted acids were
found to work better than Lewis acids as catalysts for the archetypal aldol cyclotrimerization of indanone
(2) in aprotic solvents, and a strong dependence on the pKa of the acid was observed. A standardized
protocol, using p-toluenesulfonic acid monohydrate, is shown to work well in a number of test cases.
Introduction
concern, because rotational barriers in the “trienol” are not
expected to be large, at least in the case illustrated.
For the construction of triply annulated benzene rings that
have C3 or C3h symmetry, nothing rivals the one-pot, acid-
catalyzed, head-to-tail cyclotrimerization of cyclic ketoness
when it works.1,2 The heptacyclic C27H18 hydrocarbon truxene
(1), for example, can be prepared in 85% isolated yield simply
by refluxing indanone (2) in a 1:2 mixture of concentrated HCl
and acetic acid.1b,3 Scheme 1 outlines the presumed overall
mechanistic pathway for this remarkable transformation. The
stereochemistries of the “dimer” and “trimer” are of little
Though first discovered in the 1800s,4 this reaction has
enjoyed scant popularity as a synthetic tool, largely on account
of its unpredictability. Many cases give low yields or fail
completely.1 Under conditions that work well with indanone,
for example, the closely related aromatic ketone R-tetralone (3)
yields no more than traces of the expected cyclotrimerization
product (4), and acenaphthenone (5) gives less than 5% yield
of decacyclene (6).5
In this paper, we provide insights into some of the factors
that can scuttle such cyclotrimerizations and offer recommenda-
tions on how to choose solvents, temperatures, acid catalysts,
etc. that will circumvent some of the pitfalls. Certain ketones,
(1) (a) For a review of early work on the aldol cyclotrimerization reaction,
see: Boorum, M. M.; Scott, L. T. In Modern Arene Chemistry; Astruc, D.,
Ed.; Wiley: New York, 2002; Chapter 1. (b) De Frutos, O.; Gomez-Lor,
B.; Granier, T.; Monge, M. A.; Gutierrez-Puebla, E.; Echavarren, A. M.
Angew. Chem., Int. Ed. 1999, 38, 204-207. (c) Boorum, M. M. Ph.D.
dissertation, Boston College, 2001. (d) Zhang, W.; Cao, X.-Y.; Hong, Z.;
Pei, J. Org. Lett. 2005, 7, 959. (e) Aly, A. A. Tetrahedron Lett. 2005, 46,
443. (f) Terai, H.; Takaya, H.; Naota, T. Tetrahedron Lett. 2006, 47, 1705.
(g) Yuan, M.-S.; Fang, Q.; Liu, Z.-Q.; Guo, J.-P.; Chen, H.-Y.; Yu, W.-T.;
Xue, G.; Liu, D.-S. J. Org. Chem. 2006, 71, 7858.
(3) Dehmlow, E. V.; Kelle, T. Synth. Commun. 1997, 27, 2021-2026.
Note: De Frutos et al. report an 85% yield for purified truxene,1b whereas
the 98% yield reported by Dehmlow et al. is based on the weight of
unpurified precipitate from the reaction.
(4) (a) Wislicenus, W. Ber. Dtsch. Chem. Ges. 1887, 20, 589-595. (b)
Hausmann, J. Ber. Dtsch. Chem. Ges. 1889, 22, 2019-2026. (c) Kipping,
F. S. J. Chem. Soc. 1894, 65, 269-289. (d) Kipping, F. S. J. Chem. Soc.
1894, 65, 480-503.
(2) For more lengthy stepwise alternatives, see: (a) Hagen, S.; Scott, L.
T. J. Org. Chem. 1996, 61, 7198-7199. (b) Mehta, G.; Srirama Sarma, P.
V. V. Tetrahedron Lett. 2002, 43, 9343-9346 and references cited therein.
(5) Not previously published; see the Supporting Information.
10.1021/jo070080q CCC: $37.00 © 2007 American Chemical Society
Published on Web 03/24/2007
3412
J. Org. Chem. 2007, 72, 3412-3418

Aldol Cyclotrimerizations
SCHEME 1
such as R-tetralone (3), are doomed from the start, for reasons
discussed below, whereas useful conditions can be found for
others, such as acenaphthenone (5). These and related case
studies are presented herein. Hopefully, our results will prove
helpful to others who seek to exploit this powerful reaction.
We note at the outset, however, that many questions remain
unanswered.
In the R-tetralone example (Scheme 2), dimer 7 builds up in
concentration until all of the starting material has dimerized.
At that point, the dimer finds no monomer left with which to
condense, and the reaction stops.6 One is forced to conclude
that, in this case, the second aldol step (dimer + monomer f
trimer) must be significantly slower than the first one (monomer
+ monomer f dimer). By contrast, in the indanone example
(Scheme 1), the second aldol step must be significantly faster
than the first, because virtually no dimer is left stranded when
the last molecules of monomer disappear.
Why should the position of the double bond affect the rate
of the second aldol condensation step? The answer lies in the
details of the second reaction (dimer + monomer f trimer).
The C-C bond-forming step for this aldol condensation involves
nucleophilic attack by the enol of one partner on the protonated
carbonyl group of the other. In principle, that coupling could
be accomplished in either of two ways, as illustrated in eqs 1
and 2 for indanone. Protons are used in these equations as
generic representations for the acid catalyst (discussed further
below).
Results and Discussion
First Requirement: Dimer Must be R,â-Unsaturated.
Dehydration of the initial aldol product to produce a â,γ-
unsaturated dimer as the thermodynamically preferred enone
spells death for the aldol cyclotrimerization.1 The failure of
R-tetralone (3) to give a cyclic trimer (4) can be traced to
precisely this problem. Under all conditions we have examined,
the acid-catalyzed aldol condensation of R-tetralone (3) produces
the â,γ-unsaturated dimer (7) and stops there (e.g., Scheme 2).2a
Density functional theory confirms that both geometric isomers
of the R,â-unsaturated dimer lie higher in energy than 7 in this
case.1
The first of these pathways, in which the enol of the dimer
(8) functions as the nucleophile (eq 1), can be dismissed as
improbable for electronic reasons. The γ-carbon atom of the
dienol (8) will be relatively nucleophilic only if the diene
SCHEME 2
(6) Continued heating of this reaction mixture after the tetralone has been
consumed leads only to the slow destruction of dimer 7 and no significant
increase in the small amount of cyclic trimer 4 already formed: Amick, A.
W.; Dombrowski, J. M.; Scott, L. T. Unpublished observation.
J. Org. Chem, Vol. 72, No. 9, 2007 3413

Amick and Scott
SCHEME 3
SCHEME 4
remains conjugated. Steric repulsions, however, will greatly
diminish the population of the planar conformation required to
achieve good orbital overlap. In such twisted systems, the
R-carbon atom of a dienol generally exhibits greater nucleo-
philicity than the γ-carbon atom7 (see the Supporting Informa-
tion).
The pathway depicted in eq 2, on the other hand, looks quite
reasonable. No serious steric impediments protrude above or
below the plane of the protonated carbonyl group. Furthermore,
the enhanced basicity of the ketone imparted by the additional
R,â-unsaturation8 ensures that the concentration of electrophile
9 will be high enough to compete for the enol of the monomer
as the reaction proceeds.
With a â,γ-double bond, the dimer of R-tetralone (7) not only
loses the kinetic advantage resulting from a high concentration
of protonated dimer, but the protonated dimer is also a more
crowded electrophile than the protonated monomer. Conse-
quently, the second aldol step becomes slow, and dimer 7 builds
up in concentration much faster than it proceeds on to the next
step.
Little can be done to rectify this problem when it spoils a
planned aldol cyclotrimerization reaction, unfortunately, but with
the help of modern computational methods, the potential for
encountering such a pitfall can at least be assessed in advance
with some confidence.1
the hypothesis is correct that this factor is playing a critical
role in determining the outcome of the aldol cyclotrimerization,
one would predict that the incorporation of solubilizing groups
might salvage the reaction and prevent it from stalling at the
dimer stage. To test our hypothesis, we decided to study the
aldol cyclotrimerization of a soluble mechanistic probe molecule,
4,7-di-tert-butylacenaphthenone (11). A short synthesis of 11
has recently been reported by Amick et al.10
As anticipated, the addition of tert-butyl groups to acenaph-
thenone provides soluble intermediates that can be identified
1
easily by H NMR spectroscopy (see the next section), and
cyclic trimer 12 can be obtained in better than 50% isolated
yield (Scheme 4).10. This reaction yields no more than a trace
of cyclic tetramer, a common byproduct of many aldol cyclo-
trimerizations.
We conclude that solubility of the aldol dimer is a second
prerequisite for success in the aldol cyclotrimerization reaction.
Temperature Requirements. Many aldol cyclotrimerizations
seem to work best only at relatively high temperatures, e.g.,
refluxing acetic acid (bp 117 °C) and even refluxing o-
dichlorobenzene (bp 180 °C).1 We wished to determine whether
elevated temperatures are required to overcome high-energy
barriers at one or more steps along the reaction pathway (Scheme
1) or if they simply serve to keep the aldol dimer dissolved.
From the following studies on ketone 11, we conclude that there
are no intrinsically high-energy barriers along the reaction
pathway; hexa-tert-butyldecacyclene (12) begins to appear in
the cyclotrimerization of 11 even at 0 °C!
After screening several solvents and acids, we settled on
boron tribromide in o-dichlorobenzene (ODCB) for the tem-
perature studies described here. To the best of our knowledge,
boron tribromide has not previously been used for the aldol
cyclotrimerization reaction; however, we found that it gives
cleaner reaction mixtures for NMR monitoring in the present
case than TiCl4 and the other Lewis acids examined. We will
have more to say later about the choice of solvent.
Second Requirement: Dimer Must Stay in Solution.
Unlike R-tetralone (3), acenaphthenone (5) gives an R,â-
unsaturated dimer (10), rather than a â,γ-unsaturated isomer.
Nevertheless, the cyclotrimerization still fails under conditions
that work well for indanone (1). In this case, it appears that the
reaction fails because 10 precipitates out of the reaction mixture
as soon as it is formed (Scheme 3).5
For the second aldol condensation (dimer + monomer f
trimer) to compete successfully with the first one (monomer +
monomer f dimer), the concentration of protonated dimer in
solution must be high enough to compensate for the fact that it
is a somewhat more hindered electrophile than the protonated
monomer. Apparently, the protonated 18-carbon dimer of
indanone (9) remains sufficiently soluble in hot acetic acid/
HCl to complete the trimerization, whereas protonation does
not render the 24-carbon acenaphthenone dimer (10) sufficiently
soluble to be kinetically viable.
The first aldol condensation between two molecules of 11
gives rise to two diastereomeric R,â-unsaturated dimers (Z- and
E-13). Both are soluble in ordinary organic solvents, and both
give characteristic singlets that are well separated from other
1
peaks in the H NMR spectra taken of reaction mixtures prior
The dimer of acenaphthenone (10), being large and planar,9
exhibits relatively poor solubility in most organic solvents. If
to completion of the cyclotrimerization of 11 (see the Supporting
Information). Dimer Z-13 has been isolated in pure form as a
high melting, bright-yellow solid.10
Aldol cyclotrimerizations of 11 catalyzed by BBr3 in ODCB
were run in triplicate, for 5 h each, at ten different temperatures
from 0 to 180 °C. After quenching, each reaction mixture was
(7) See, for example: Noyori, R.; Inoue, H.; Kato, M. Bull. Chem. Soc.
Jpn. 1976, 49, 3673-3678.
(8) See ref 1, p 28.
(9) No X-ray crystal structure of 10 is available, but DFT calculations
at the B3LYP/6-31G* level of theory predict a planar geometry as the energy
minimum. X-ray crystal structures of substituted derivatives of 10 are
planar.20
(10) Amick, A. W.; Griswold, K. S.; Scott, L. T. Can. J. Chem. 2006,
84, 1268.
3414 J. Org. Chem., Vol. 72, No. 9, 2007

Aldol Cyclotrimerizations
1
worked up by a uniform protocol and analyzed by H NMR
spectroscopy. The relative intensities of the characteristic signals
for dimers Z-13 and E-13, cyclic trimer 12, and residual starting
material 11 present at the end of each 5 h reaction are plotted
in Figure 1A. No NMR signals could be detected for any of
the reaction intermediates between the dimers (Z- and E-13)
and the final cyclic trimer (12).
As Figure 1A shows, the aldol cyclotrimerization of 11
catalyzed by BBr3 in o-dichlorobenzene reaches completion in
5 h at 55 °C. At higher temperatures, the amount of cyclic trimer
(12) in this case actually decreases during the 5 h reaction time.
1
In the H NMR spectra, new signals begin to appear that may
result, in part, from the onset of acid-catalyzed “dealkylation”
of 12. This unwanted side reaction limits the use of 11 as a
probe molecule for reactions run under forcing conditions.
At temperatures below 55 °C, the cyclotrimerizations proceed
only part way in the 5 h before they are quenched. Signals for
both dimers (Z- and E-13) can be seen, with the latter
predominating at lower temperatures, but the concentration of
dimers, taken together, always remains lower than that of starting
material (11). Even at 0 °C, the first aldol step converts 32%
of the starting material to dimers Z-13 and E-13 within the first
5 h, and a 1-2% yield of trimer already begins to appear.
Clearly, none of the steps in this aldol cyclotrimerization require
high temperatures.
Closer examination of the results in Figure 1A allows us to
draw a second, tentative conclusion. The trends suggest that
the relatiVe rates of the first and second aldol steps may respond
differently to temperature, at least in this test case. We note
that when the reaction is run at low temperatures, the concentra-
tions of the two dimers build up quite high, and only small
amounts of trimer are formed during the first 5 h of the reaction.
At progressively higher temperatures, the ratio of dimers to
trimer at the end of 5 h is less skewed. In light of this trend,
one would be wise, when optimizing the conditions for a new
aldol cyclotrimerization, to examine the reaction not only at
the lowest practical temperature but also at higher temperatures
for shorter times.
Solvent Requirements. Figure 1B shows the results obtained
by repeating the aldol cyclotrimerization of di-tert-butylacenaph-
thenone (11) with BBr3 in 1,1,2,2-tetrachloroethane. This solvent
has also been used in the past for aldol cyclotrimerizations,1 in
part because it boils at a relatively high temperature (bp 147
°C) and is stable to acid.
Qualitatively, the temperature profile in Figure 1B looks very
similar to that in Figure 1A, but it is shifted. Not much happens
at 0 °C; the buildup of dimers and the onset of trimerization is
not seen until 25 °C, and the reaction is only 90% complete in
5 h at 55 °C. The composition of the reaction mixture after 5 h
at 25 °C in 1,1,2,2-tetrachloroethane appears virtually identical
with that of the reaction mixture in ODCB after 5 h at 0 °C
FIGURE 1. Compositions of the reaction mixtures (1H NMR) after 5
h at different temperatures for the aldol cyclotrimerizations of 11 (50
mg each) with BBr3 (11.2 equiv) in o-dichlorobenzene (A) and in
1,1,2,2-tetrachloroethane (B).
(compare parts A and B of Figure 1). From this comparison,
then, o-dichlorobenzene seems to be a somewhat better choice
as the solvent than 1,1,2,2-tetrachloroethane. The two have very
similar dielectric constants (9.93 and 8.20, respectively)11 and
even more similar solvent ET(30) values (38.0 and 39.2,
respectively);12 however, the delocalized π-system of o-di-
chlorobenzene undoubtedly makes it more polarizable than
1,1,2,2-tetrachloroethane, and this may well contribute to its
superiority. The difference in refractive index between these
two solvents (1.552 and 1.494, respectively, at 20 °C)13 can be
taken as an indication of the difference in their polarizabilities.14
To probe the importance of solvent polarity, we searched for
a solvent that matches one of these two with respect to
polarizability but differs significantly in polarity. Toluene
emerged as a reasonable choice: the refractive index of toluene
is virtually identical with that of 1,1,2,2-tetrachloroethane (1.496
vs 1.494),13 but toluene has a much lower dielectric constant
(2.38 vs 8.20)11 and solvent ET(30) value (33.9 vs 39.2)12 than
1,1,2,2-tetrachloroethane.
A single point comparison between reactions run in toluene
and 1,1,2,2-tetrachloroethane is revealing. At the end of 5 h at
55 °C, the reaction mixture from aldol cyclotrimerization of 11
with BBr3 in 1,1,2,2-tetrachloroethane contains 90% trimer (12),
7% starting material, and 3% dimer Z-13 (Figure 1B). By
contrast, the corresponding mixture from the same reaction run
(11) Riddick, J. A.; Bunger, W. B. Organic SolVents: Physical Properties
and Methods of Purification, 3rd ed.; Wiley-Interscience: New York, 1971.
(12) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry,
3rd ed.; Wiley-VCH: Weinheim, Germany, 2003.
(13) Refractive index values at 20 °C were taken from the following:
CRC Handbook of Chemistry and Physics, 75th ed.; Lide, D. R., Ed.; CRC
Press: Boca Raton, FL, 1994; Section 3.
(14) The polarizability of a liquid is directly proportional to its index of
refraction: CRC Handbook of Chemistry and Physics, 75th ed.; Lide, D.
R., Ed.; CRC Press: Boca Raton, FL, 1994; Section 10.
J. Org. Chem, Vol. 72, No. 9, 2007 3415

Amick and Scott
in toluene contains only 4% trimer (12), 96% dimer Z-13, and
no remaining starting material. The dramatic reversal in the aldol
product ratios clearly indicates that some degree of solvent
polarity, in addition to polarizability, is advantageous for the
cyclotrimerization. Toluene also proved to be less impervious
to BBr3 than 1,1,2,2-tetrachloroethane, giving rise to a number
of solvent-derived hydrocarbon byproducts.
To probe next the importance of solvent polarizability, we
searched for a solvent that has exceptionally high polarizability.
Diiodomethane emerged as an interesting choice: the refractive
index of diiodomethane at 20 °C (1.743)13 is substantially higher
than that of either ODCB or 1,1,2,2-tetrachloroethane (1.552
and 1.494, respectively). The dielectric constant (5.32)11 and
solvent ET(30) value (36.5)12 for diiodomethane are both
somewhat lower than those for ODCB and 1,1,2,2-tetrachloro-
ethane, but its polarity is not as low as that of toluene. For the
aldol cyclotrimerization of 11 with BBr3 in diiodomethane, after
5 h at 55 °C, the reaction mixture contains 56% trimer (12),
40% starting material, and 4% dimer Z-13. Thus, the reaction
proceeds more slowly than in ODCB and 1,1,2,2-tetrachloro-
ethane, but the dimer continues rapidly on to trimer; its
concentration does not build up. Polarizability, therefore, seems
to be an important solvent attribute for this reaction.
FIGURE 2. Yield of the cyclotrimerization of indanone (2) versus
pKa of the acid catalyst used. Each reaction was run with 1.0 mmol of
2 and 3.5 equiv of acid in 1,1,2,2-tetrachloroethane at 105 °C for 16 h.
neutral enol never rises high enough for the aldol condensation
to occur. This is entirely reasonable, since the pKa of protonated
indanone (-3.65)15 is much higher than that of triflic acid (ca
-14).16
When the above experiment was repeated, using 3.5 equiv
of triflic acid but with 3.5 equiv of acetic acid also added,
truxene was formed in 4% yield. Under these conditions,
protonated acetic acid (pKa ) -6.1)17,18 serves as the Brønsted
acid catalyst. Using 3.5 equiv of methanesulfonic acid as the
catalyst (pKa ) -2.0)18. raises the yield of truxene to 20%, but
moving slightly further up the pKa scale to p-toluenesulfonic
acid monohydrate (i.e., hydronium tosylate, pKa ) -1.7)16a
gives truxene in 54% yield (Figure 2).
These results indicate that the Brønsted acid catalyzed aldol
cyclotrimerization of indanone in 1,1,2,2-tetrachloroethane
works best when the pKa of the acid catalyst is at least 2 pKa
units higher than that of the protonated ketone. Empirically,
this suggests that high levels of protonated ketone (more than
a few percent) are detrimental to the aldol cyclotrimerization.
One possible explanation for this correlation is that high levels
of protonated ketone could promote polymerization by intercep-
tion of the trienol (see Scheme 1) before it can cyclize. High
molecular weight materials invariably constitute a portion of
the product mixture in these reactions.
Overall, in our experience, o-dichlorobenzene has the best
combination of polarity and polarizability of any solvent we
have found for the aldol cyclotrimerization. Furthermore, its
high boiling point makes it convenient to use at temperatures
well above 100 °C when such measures are required to prevent
the dimers from precipitating out of solution.
Brønsted Acids as Catalysts. The preparation of truxene
from indanone (2 f 1, Scheme 1) in 85% isolated yield stands
out as one of the best aldol cyclotrimerizations ever reported.3
The highest yield has been achieved by using a classical
Brønsted acid system, concentrated HCl in hot acetic acid.
Unfortunately, the inhospitable nature of aqueous acetic acid
for larger ketones and their aldol dimers (e.g., acenaphthenone,
5, see above) severely limits the generality of these conditions.
Fortunately, o-dichlorobenzene and other chlorinated hydro-
carbons dissolve a much wider range of organic compounds
and have enough polarity and acid stability to serve as
convenient media in which to run aldol cyclotrimerizations. BBr3
and TiCl4 both work as Lewis acid catalysts in these solvents,
but the isolated yield of truxene from aldol cyclotrimerization
of indanone in 1,2-dichloroethane is only 36% when BBr3 is
used as the catalyst, and the yield is even lower with TiCl4.
Focusing on the p-toluenesulfonic acid monohydrate condi-
tions, we examined the effects of various additives and
discovered that the combined use of p-toluenesulfonic acid
monohydrate and acetic acid (3.5 equiv of each) in 1,1,2,2-
tetrachloroethane at 105 °C raises the yield for conversion of
indanone to truxene from 54% to 64%. We speculate that the
added acetic acid may play a role in solvation of the tosylate
The effectiveness of Brønsted acids in chlorinated hydro-
carbon solvents in promoting aldol cyclotrimerizations has never
been thoroughly explored. We decided to screen some Brønsted
acids, using indanone (2), rather than di-tert-butylacenaph-
thenone (11), which suffers from acid-catalyzed dealkylation
under forcing conditions. Our objective was to find conditions
for converting indanone into truxene (2 f 1, Scheme 1) in good
yield, using a high-boiling organic solvent that would be suitable
for a wide variety of cyclic ketones. The 85% yield of truxene
from indanone in acetic acid/HCl was the benchmark that we
strove to replicate.
(15) Noto, R.; D’Anna, F.; Gruttadauria, M.; Lo Meo, P.; Spinelli, D. J.
Chem. Soc., Perkin 2 2001, 2043-2046.
(16) (a) Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456-63. See also:
(b) Engelbrecht, A.; Rode, B. M. Monatsh. Chem. 1972, 103, 1315-19.
(c) Rode, B. M.; Engelbrecht, A.; Schantl, J. Z. Phys. Chem. 1973, 253,
17-24. (d) Benoit, R. L.; Buisson, C. Electrochim. Acta 1973, 18, 105-
10. (e) Fujinaga, T.; Sakamoto, I. J. Electroanal. Chem. Interfacial
Electrochem. 1977, 85, 185-201. (f) Guthrie, J. P. Can. J. Chem. 1978,
56, 2342-54.
Adding 3.5 equiv of triflic acid (CF3SO3H) to a solution of
indanone in 1,1,2,2,-tetrachloroethane produces an immediate
color change to red. Heating the reaction mixture overnight at
105 °C fails to yield any truxene (1); only starting material (2)
is recovered. Apparently, triflic acid simply protonates all of
the ketone, essentially irreversibly, and the concentration of
(17) Arnett, E. M. In Progress in Physical Organic Chemistry; Cohen,
S. G., Streitwieser, A., Jr., Taft, R. W., Eds.; Interscience: New York, 1963;
Vol. 1, p 253.
(18) Ionization Constants of Organic Acids in Solution Serjeant, E. P.,
Dempsey, B., Eds.; IUPAC Chemical Data Series No. 23; Pergamon:
Oxford, UK, 1979.
3416 J. Org. Chem., Vol. 72, No. 9, 2007

Aldol Cyclotrimerizations
TABLE 1. Aldol Cyclotrimerization Reactions Run with 3.5 equiv
Each of p-TsOH‚H2O and Propionic Acid as the Catalyst System in
o-Dichlorobenzene at 105 °C for 16 h
always given poor yields of complex product mixtures that
contain not only the trimer but also the cyclic tetramer.20 With
the TsOH/propionic acid system, trimer 21 was obtained in a
usable 34% yield, uncontaminated by cyclic tetramer.
Conclusions
From the results of the limited screening summarized in Table
1, it is already clear that progress has been made toward the
delineation of a widely applicable recipe for conducting aldol
cyclotrimerizations of aromatic ketones. The use of p-toluene-
sulfonic acid monohydrate and propionic acid (3.5 equiv of each)
in o-dichlorobenzene at 100 °C is recommended as a good
starting point for new systems. The reaction is generally
complete in less than 16 h and should be monitored to produce
the optimum yield. If all of the starting ketone is consumed
before the dimer is converted to cyclic trimer, it may be
beneficial to repeat the reaction at a higher temperature.
There is still much to learn about the aldol cyclotrimerization
reaction, but some key factors have been clarified. Dehydration
of the initial aldol dimer to a â,γ-unsaturated ketone (e.g., 7)
will generally eliminate any chances of producing the cyclic
trimer. Fortunately, DFT calculations can be used to determine
in advance, with some confidence, whether the most stable
enone will be â,γ-unsaturated (undesirable) or R,â-unsaturated
(required for a successful aldol cyclotrimerization).
Precipitation of the dimer from the reaction medium will
likewise diminish the yield of cyclic trimer, sometimes to 0%.
Several strategies can be invoked, either individually or in
combination, to combat this problem: (1) incorporate solubi-
lizing groups on the ketone, (2) run the reaction at a higher
temperature, (3) switch to a better solvent. Polarizability and
some degree of polarity both seem to be important prerequisites
for a good solvent in these reactions; o-dichlorobenzene (ODCB)
has about the best balance between these two qualities of any
solvent we have examined.
High temperatures are not absolutely necessary for the aldol
cyclotrimerization, but they can be advantageous when the
intermediate dimers exhibit poor solubility. High temperatures
also appear to suppress the build up of the aldol dimers. Lewis
acids can function as catalysts in aprotic media but are often
inferior to p-toluenesulfonic acid monohydrate.
anion, although how this improves the yield remains unclear.
In anticipation that cyclotrimerizations of other ketones might
require temperatures well above the boiling point of acetic acid
(117 °C), we investigated also the influence of added propionic
acid (bp 141 °C) and found that it works just as well as acetic
acid, or even slightly better (67% yield).
In light of these results, we set out to see how broadly
applicable the combined use of p-toluenesulfonic acid mono-
hydrate and propionic acid (1:1) might be as a catalyst system
for other aldol cyclotrimerizations in ODCB.
Applications of the TsOH‚H2O/CH3CH2COOH/ODCB
Catalyst System. A particularly problematical case in our
experience has been the aldol cyclotrimerization of unsubstituted
acenaphthenone (5) to decacyclene (6). As reported in the
introduction, heating acenaphthenone in acetic acid/concentrated
HCl produces decacyclene in less than 5% yield. Using
p-toluenesulfonic acid monohydrate and propionic acid (3.5
equiv of each) in o-dichlorobenzene at 105 °C for 16 h, however,
converts acenaphthenone to decacyclene in 66% isolated yield.
This is by far the best yield we have ever been able to achieve
for the cyclotrimerization of acenaphthenone. Our previous best
was 29%, using SiCl4 in THF.19
Table 1 summarizes the results obtained by using the
p-toluenesulfonic acid monohydrate/propionic acid catalyst
system with indanone, acenaphthenone, and several derivatives
of these aromatic ketones. In all cases but one, the cyclic trimers
could be isolated in 59-89% yield, without any individual
optimization of the reaction conditions.
We hope these observations will prove useful to others who
wish to harness the power of the aldol cyclotrimerization
reaction for the one-pot synthesis of triply annulated benzene
rings.
Experimental Section
General Procedure for Aldol Cyclotrimerizations with Use
of p-Toluenesulfonic Acid Monohydrate and Propionic Acid.
A 25 mL round-bottom flask was charged with the ketone (0.810
mmol), p-toluenesulfonic acid monohydrate (0.538 g, 2.83 mmol),
propionic acid (0.21 mL, 2.8 mmol), and 0.69 mL of o-dichloro-
benzene. The reaction mixture was heated to 105 °C for 16 h. After
16 h the crude reaction mixture was poured into methanol, which
was slowly neutralized with 5 M NaOH. The precipitate was
collected by filtration and washed with methanol and then ethanol
to leave behind the desired cyclic trimer.
Truxene (1). From indanone (2) (1.36 g, 1.03 mmol), the general
procedure described above gave 0.758 g (67%) of 1 as a pale yellow
1
Previous attempts to effect the cyclotrimerization of 5-chloro-
acenaphthenone (20) with TiCl4 in o-dichlorobenzene have
solid: mp >300 °C; H NMR (400 MHz, CDCl3) δ 7.97 (d, J )
(20) Ansems, R. B. M. Ph.D. Dissertation, Boston College, Chestnut Hill,
2004.
(19) McComas, C. C. B.S. thesis, Boston College, Chestnut Hill, 1996.
J. Org. Chem, Vol. 72, No. 9, 2007 3417

Amick and Scott
7.2 Hz, 3 H), 7.70 (d, J ) 7.6 Hz, 3 H), 7.51 (t, J ) 7.6 Hz, 3 H),
7.40 (t, J ) 7.6 Hz, 3 H), 4.29 (s, 6 H).3a,21
described above gave 0.051 g (34%) of 21 as a light brown solid:
mp >300 °C; 1H NMR22(400 MHz, CDCl3) δ 8.59 (d, J ) 7.6 Hz,
1H), 8.58 (dd, J ) 8.0 Hz, J ) 2.0 Hz, 1H), 7.96 (d, J ) 8.0 Hz,
1H), 7.91 (d, J ) 7.6 Hz, 1H), 7.91 (d, J ) 8.4 Hz, 1H), 7.78 (d,
J ) 7.6 Hz, 1H), 7.57-7.54 (m, 1H), 7.51 (d, J ) 7.2 Hz, 1H),
6.99 (dd, J ) 8.4 Hz, J ) 7.2 Hz, 1H), 5.05 (d, J ) 7.2 Hz, 1H);
13C NMR22 (125 MHz, CDCl3) δ 163.0, 136.0, 135.4, 135.3, 135.1,
135.0, 134.3, 133.3, 133.0, 132.1, 131.9, 131.5, 130.3, 129.4, 128.7,
128.5, 127.7, 127.3, 126.4, 124.8, 124.3, 122.99, 53.6; HRMS ESI
(m/z) [M]+ calcd for C36H15Cl3 (M+) 552.0239, found 552.0248.
2,7,12-Tribromotruxene (15). From 5-bromo-1-indanone (14)
(0.171 g, 0.810 mmol), the general procedure described above gave
0.100 g (64%) of 15 as a pale yellow solid: mp >300 °C; 1H NMR
(300 MHz, [D2]1,1,2,2-tetrachloroethane, 100 °C) δ 7.86 (s, 3H),
7.80 (d, J ) 7.8 Hz, 3H), 7.66 (d, J ) 8.1 Hz, 3H), 4.27 (s, 6H);
HRMS ESI (m/z) [M]+ calcd for C27H15Br3 (M+) 575.8724, found
575.8735. Anal. Calcd for C27H15Br3: C, 56.00; H, 2.61. Found:
C, 55.90; H, 2.34.
1,6,11-Trimethyltruxene (17). From 4-methyl-1-indanone (16)
(0.118 g, 0.810 mmol), the general procedure described above gave
0.92 g (89%) of 17 as a pale yellow solid: mp >300 °C; 1H NMR
(400 MHz, [D2]1,1,2,2-tetrachloroethane, 100 °C) δ 7.90 (d, J )
8.0 Hz, 3H), 7.50 (t, J ) 8.0 Hz, 3H), 7.28 (d, J ) 8.0 Hz, 3H),
4.18 (s, 6H), 2.59 (s, 9H); 13C NMR (125 MHz, [D2]1,1,2,2-
tetrachloroethane, 83 °C) δ 142.6, 141.5, 137. 6, 135.2, 127.6, 127.4,
119.6, 35.5, 19.0; HRMS ESI (m/z) [M]+ calcd for C30H24 (M+)
384.1878, found 384.1876.
3,8,13-Trimethyltruxene (19). From 6-methyl-1-indanone (18)
(0.118 g, 0.810 mmol), the general procedure described above gave
0.061 g (59%) of 19 as a pale yellow solid: mp >300 °C; 1H NMR
(300 MHz, [D2]1,1,2,2-tetrachloroethane, 100 °C) δ 7.78 (s, 3H),
7.64 (d, J ) 7.2 Hz, 3H), 7.26 (d, J ) 7.2 Hz, 3H), 4.27 (s, 6H),
2.57 (s, 9H); HRMS ESI (m/z) [M]+ calcd for C30H24 (M+)
384.1878, found 384.1875. Anal. Calcd for C30H24: C, 93.71; H,
6.29. Found: C, 93.46; H, 6.01.
Acknowledgment. Financial support from the National
Science Foundation is gratefully acknowledged. We thank Dr.
Kenneth R. Metz for technical assistance, Dr. Ronald B. M.
Ansems for a sample of 5-chloroacenaphthenone (20), and
Restek Corp. for generously supplying GC columns.
Supporting Information Available: NMR spectra for new
cyclic trimers 15, 17, 19 and 21; experimental details for the
attempted aldol cyclotrimerization of acenaphthenone (5) in con-
centrated HCl/HOAc (Scheme 3); representative NMR spectra,
integration data, and experimental details for the solvent/temperature
studies summarized in Figure 1; DFT calculations on dienol 8.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JO070080Q
Decacyclene (6). From acenaphthenone (5) (0.136 g, 0.810
mmol), the general procedure described above gave 0.080 g (66%)
of 6 as a brown solid: mp >300 °C; 1H NMR (300 MHz,
[D2]1,1,2,2-tetrachloroethane) δ 8.65 (d, J ) 7.2, 6H), 7.93 (d, J
) 7.8 Hz, 6H), 7.75 (t, J ) 7.2 Hz, 6H).21
(22) Two sets of peaks are seen in the 1H and 13C NMR spectra of
decacyclene derivative 21. Strong aggregation/face-to-face dimerization is
expected in solution for large polycyclic compounds such as this,23 and
from the symmetry of the monomer, one would expect a meso-dimer and
a d,l-dimer to be formed with nearly equal probability.
(23) De Frutos, O.; Granier, T.; Gomez-Lor, B.; Jimenez-Barbero, J.;
Monge, A.; Gutierrez-Puebla, E.; Echavarren, A. M. Chem. Eur. J. 2002,
8, 2879.
4,10,16-Trichlorodecacyclene (21). From 5-chloroacenaph-
thenone (20)20 (0.164 g, 0.810 mmol), the general procedure
(21) Drake, J. A. G.; Jones, D. W. Org. Magn. Reson. 1980, 14, 272.
3418 J. Org. Chem., Vol. 72, No. 9, 2007