Titanium(IV) triflates in the catalysis of homoaldol reactions

Article abstract of DOI:10.1021/ol991017t

(matrix presented). The first example of the use of a titanium catalyst to effect the addition of silyloxycyclopropanes to aldehydes, the homoaldol reaction, is reported. The method features an alkoxytitanium(IV) triflate catalyst, which is conveniently prepared by treatment of Binol-Ti(OⁱPr)₂ with TMSOTf.

Full text of DOI:10.1021/ol991017t

ORGANIC
LETTERS
1999
Vol. 1, No. 10
1643-1645
Titanium(IV) Triflates in the Catalysis of
Homoaldol Reactions
Evelyn O. Martins and James L. Gleason*
Department of Chemistry, McGill UniVersity, 801 Sherbrooke St. West,
Montreal, Quebec H3A 2K6, Canada
jim@gleason.chem.mcgill.ca
Received September 7, 1999
ABSTRACT
The first example of the use of a titanium catalyst to effect the addition of silyloxycyclopropanes to aldehydes, the homoaldol reaction, is
i
reported. The method features an alkoxytitanium(IV) triflate catalyst, which is conveniently prepared by treatment of Binol−Ti(O Pr)2 with
TMSOTf.
Stereoselective aldol reactions have been the subject of
intense study in the past three decades. Recent efforts in this
area have focused on the use of chiral Lewis acids to effect
enantioselective condensations.1 Stereoselective homoaldol
reactions have, in contrast, received significantly less atten-
tion.2,3 This is despite the potential use of this reaction for
the stereoselective synthesis of 1,4-oxygenated substrates,
including peptidomimetic substructures. In this Letter, we
describe a new method for the catalysis of the homoaldol
reaction, using alkoxytitanium(IV) triflates as catalysts, which
should serve as an effective platform for the development
of a highly enantioselective process.
(trimethylsilyloxy)cyclopropanes (e.g., 1) could be ring-
opened with either titanium(IV) chloride or zinc chloride to
afford titanium or zinc homoenolates, respectively, which
add to aldehydes in good yields (Scheme 1).4 The need to
Scheme 1
The preparation of discrete homoenolates was pioneered
by Nakamura and Kuwajima, who found that 1-alkoxy-1-
form a discrete metal alkyl prior to aldehyde addition makes
this reaction fundamentally different from the Mukaiyama
aldol process, wherein the Lewis acid generally is involved
only in carbonyl activation. This difference makes catalysis
of the homoaldol reaction more difficult, reflected by the
fact that only ZnCl2 and ZnI2 are capable of acting as
substoichiometric catalysts.4 Our preliminary investigations
indicated that coordinating ligands (e.g., â-amino alcohols,
bis-oxazolines)5 inhibit both zinc halide induced ring opening
(1) (a) Seyden-Penne, J. Chiral Auxiliaries and Ligands in Asymmetric
Synthesis; Wiley: New York, 1995. (b) Ojima, I Catalytic Asymmetric
Synthesis; VCH: Weinheim, 1993.
(2) For reviews on the homoaldol reaction, see: (a) Crimmins, M. T.;
Nantermet, P. G. Org. Prep. Proc. 1993, 25, 43-81. (b) Kuwajima, I.;
Nakamura, E. In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming,
I., Eds.; Pergammon: Oxford, 1991; Vol 2, p 441.
(3) For examples of diastereoselective homoaldol reactions, see: (a)
Kano, S.; Yokomatsu, T.; Shibuya, S. Tetrahedron Lett. 1991, 32, 233. (b)
DeCamp, A.; Kawaguchi, A.; Volante, R.; Shinkai, I. Tetrahedron Lett.
1991, 32, 1867. (c) Armstrong, J.; Hartner, F.; DeCamp, A.; Volante, R.;
Shinkai, I. Tetrahedron Lett. 1992, 33, 6599. (d) Houkawa, T.; Ueda, T.;
Sakami, S.; Asaoka, M.; Takei, H. Tetrahedron Lett. 1996, 37, 1045. For
examples of stereoselective reactions using homoenolate equivalents, see
ref 2 and Ahlbrecht, H.; Beyer, U. Synthesis 1999, 365.
(4) (a) Nakamura, E.; Oshino, H.; Kuwajima, I. J. Am. Chem. Soc. 1986,
108, 3745-3755. (b) Nakamura, E.; Aoki, S.; Sekiya, K.; Oshino, H.;
Kuwajima, I. J. Am. Chem. Soc., 1987, 109, 8056-8066.
10.1021/ol991017t CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/15/1999

of 1 and the subsequent addition of 2 (M ) Zn) to aldehydes,
thus making an enantioselective zinc-catalyzed process
unlikely. For this reason, we became interested in the
possibility of developing titanium(IV) complexes as catalysts
for the homoaldol reaction. Although there have been no
reports of titanium catalysis of this reaction, the potential
advantages of titanium catalysts, the most important of which
is the relative ease with which chiral ligands may be
incorporated,6 encouraged us to explore their development.
We envisioned a catalytic enantioselective homoaldol
process in which a chiral titanium complex, presumably
containing a bidentate alkoxide ligand, would react with
cyclopropane 1 to form an alkoxytitanium homoenolate.
Subsequent addition of this species to an aldehyde, followed
by in situ silylation of the resulting homoaldolate, would
complete a catalytic cycle. Modification of the known TiCl4-
mediated reaction by incorporation of a bidentate alkoxide
ligand presented two main concerns. The first was that the
alkoxide ligand would reduce the Lewis acidity of the metal
complex, retarding the rate of ring opening of 1. Second,
the large difference in bond strengths between Ti-Cl and
Ti-O bonds,7 when compared to the slight difference in bond
strengths for Si-Cl and Si-O bonds,8 indicates that silylation
of titanium homoaldolates with trimethylsilyl chloride is
thermodynamically unfavorable. Thus, catalytic turnover with
alkoxytitanium chlorides was not expected.9 To address these
concerns, we chose to explore titanium triflates as catalysts.
It was anticipated that a triflate counterion would offset the
decrease in Lewis acidity, thus allowing the ring opening of
1 to proceed at a reasonable rate. Additionally, although the
bond dissociation energy for titanium triflates has not been
reported, it seemed reasonable to assume that the difference
in bond strength between a titanium triflate and a titanium
alkoxide would be much smaller, thus making catalytic
turnover more likely.
the subsequent reaction by 1H NMR revealed the formation
of TMS ether 3b and lactone 5. Benzaldehyde and most of
the cyclopropane were consumed within 40 h. After workup,
analysis of the mixture by capillary GC indicated a 63% yield
of homoaldol adducts (16% 3b, 47% 5), indicating that some
catalyst turnover had taken place. To assess the enantiose-
lectivity of the reaction, all homoaldol products were
converted to lactone 5 by treatment with HF in acetonitrile.
Analysis by capillary GC (Chirasil-dex column) revealed that
the reaction had occurred with a modest level of enantiose-
lectivity (17% ee). The reaction catalyzed by monotriflate
complex 4b12 (0.25 equiv) was slower, requiring 143 h for
complete consumption of 1. However, the yield of products
(61%) and enantioselectivity (15% ee) were comparable to
those obtained with the ditriflate 4a.
A significant enhancement in the reaction rate was
achieved using an alternative catalyst preparation. Addition
of TMSOTf (0.25 equiv) to a solution of (R)-Binol-Ti(Oi-
13
Pr)2 (0.25 equiv) in CDCl3 at 0 °C rapidly generates
1
TMSOiPr, as observed by H NMR.14 Presumably, catalyst
4b is also formed, for subsequent addition of 1 (1 equiv)
and benzaldehyde (1 equiv) proceeds to completion within
20 h to produce 3b in 74% isolated yield (10% ee).15 This
modified catalyst preparation is operationally more facile than
the silver triflate procedure and, more importantly, results
in a significantly faster reaction rate.16 Further optimization
of the reaction was achieved by examining a variety of
solvents. Although the reaction rate was retarded in strongly
coordinating solvents such as Et2O or THF, it was signifi-
cantly enhanced in the presence of a weakly coordinating
solvent, CD3CN. The optimum conditions utilize a mixture
of CD3CN and CDCl3 (3:1), which results in a decrease in
the reaction time to less than 2 h and an increase in the yield
(82%). The mixture of solvents is necessary as the titanium
(R)-Binol-Ti(OTf)2 (4a) was prepared by treatment of (R)-
10
(8) Bond strengths for Si-X are estimated at 113 and 116 kcal/mol for
TMS-Cl and TMS-OEt, respectively. See: Walsh, R. In The Chemistry
of Organic Silicon Compounds; Patai, S., Rappoport, Z., Eds.; Wiley: New
York, 1989; p 371.
Binol-TiCl2 with silver triflate in toluene (Scheme 2).11
(9) Consistent with these concerns, we observed that while TiCl3(OiPr)
and TiCl2(OiPr)2 were both capable of mediating the homoaldol reaction,
the reactions were very slow, afforded low yields of homoaldol adducts,
and, as with TiCl4, required stoichiometric amounts of the metal salt.
(10) (R)-Binol-TiCl2 was prepared by treatment of (R)-2,2′-binaphthol
with n-BuLi (2 equiv) in ether followed by addition of TiCl4. See: Mikami,
K.; Terada, M.; Nakai, T. J. Am. Chem. Soc. 1990, 112, 3949.
(11) For synthesis of alkoxytitanium(IV) triflates, see: (a) Mikami, K.;
Sawa, E.; Terada, M. Tetrahedron: Asymmetry 1991, 2, 1403. (b)
Joergensen, K. A.; Gothelf, K. V. J. Chem. Soc., Perkin Trans. 2 1997,
111.
Scheme 2
(12) Catalyst 4b was prepared by treatment of (R)-2,2′-binaphthol with
n-BuLi (2 equiv) in ether followed by addition of Ti(OiPr)Cl3. The resulting
Binol-Ti(OiPr)Cl was then added to a suspension of silver triflate in toluene
to generate triflate 4b.
(13) Prepared by ligand exchange. See: Seebach, D.; Plattner, D. A.;
Beck, A. K.; Wang, Y. M.; Hunziker, D.; Petter, W. HelV. Chim. Acta 1992,
75, 2171.
(14) Chen, H.; White, P. S.; Gagne, M. R. Organometallics 1998, 17,
5853-5366.
Addition of 1 (1 equiv) and benzaldehyde (1 equiv) to a
solution of 4a (0.25 equiv) in CDCl3 at 0 °C and monitoring
(15) The exact structure of the active catalyst in these reactions has not
1
yet been determined. H NMR spectra of the putative catalysts 4a and 4b
(5) (a) Noyori, R.; Kitamura, M. Angew. Chem., Int. Ed. Engl. 1991,
30, 49-69. (b) Soai, K.; Niwa, S. Chem. ReV. 1992, 92, 833.
(6) Duthaler, R. O.; Hafner, A. Chem. ReV. 1992, 92, 807
(7) Bond strengths for Ti-X are estimated at 103 and 115 kcal/mol for
TiCl4 and Ti(OiPr)4, respectively. See: Reetz, M. T. Organotitanium
Reagents in Organic Synthesis, Springer-Verlag: 1986.
indicate the presence of a mixture of interconverting species at room
temperature.
(16) Ditriflate complex 4a apparently cannot be prepared by this method,
as the addition of 2 equiv of TMSOTf to Binol-Ti(OiPr)2 followed by
addition of 1 and benzaldehyde results in complete decomposition of the
starting materials.
1644
Org. Lett., Vol. 1, No. 10, 1999

complexes are only very slightly soluble in acetonitrile
alone.
Scheme 3
With these modifications in hand, we developed an
optimized protocol for the reaction which utilizes 10 mol %
of catalyst17 and 1.5 equiv of cyclopropane 1 in 3:1 CD3-
CN/CDCl3. These conditions were designed to maximize
conversion of the carbonyl substrate and afford excellent
yields for many aldehydes (Table 1). Highlighting the thermal
Table 1. Titanium-Catalyzed Homoaldol Reactions
substrate
PhCHO
Ph-CtCCHO
Me3Si-CtCCHO
p-ClPhCHO
t-C4H9CHO
conditionsa
0 °C, 24 h
0 °C, 36 h
0 °C, 36 h
yield, %
99
76
82
84
52
78
formation of a significant amount of TMSOiPr. In contrast,
during the homoaldol reaction, no formation of TMSOiPr is
observed after the initial catalyst preparation, nor at all when
the catalysts are prepared using AgOTf. These observations
are inconsistent with a direct addition of 6 to an aldehyde,
followed by silylation, and suggest addition of 6 to a silyl-
activated aldehyde. The absence of coordination of the
aldehyde to the titanium complex might also explain the low
levels of enantioselection that are observed in this process,
as it would result in an open transition state.
A potential means for improving the enantioselectivity of
this process is to increase the steric bulk of the ligand. To
this end, we have found that the system is not limited to
binaphthol as a ligand, as aliphatic 1,2-, 1,3-, and 1,4-diols
may be utilized without significantly affecting either the rate
or yield of the reaction. The presence of a bidentate ligand
is a definite requirement, however, as (iPrO)3TiOTf does not
promote the reaction to any extent. This implies that the
bidentate ligands play an important role in controlling the
aggregation state, and thus the reactivity, of the metal
catalyst.
In conclusion, we have disclosed the first example of
titanium catalysis of the direct homoaldol addition of
silyloxycyclopropanes to aldehydes. The methodology fea-
tures alkoxytitanium(IV) triflates as catalysts and is ap-
plicable toward a range of substrates. Of particular advantage
for future development is the fact that the catalyst system is
tolerant of a variety of ligands and is thermally stable. The
ability to incorporate a wide range of alkoxide ligands should
allow further development toward a highly enantioselective
process, and these efforts will be reported in due course.
0 °C, 80 h
45-50 °C, 54 hb
PhCOCH3
45-50 °C, 60 hb
a All reactions were performed in CD3CN/CDCl3 (3:1) using 1.5 equiv
of 1, except where noted. b Reaction conducted using 2 equiv of 1.
stability of the catalyst, we note that even less reactive
substrates such as acetophenone and pivaldehyde undergo
addition at elevated temperatures. One limitation with the
current catalyst system is that it is not optimal for aliphatic
aldehydes containing enolizable protons. For example, reac-
tion with cyclohexanecarboxaldehyde results in significant
quenching of the homoenolate and gives low yields of the
desired product (<40%).
Although the development of this method was based upon
an envisaged three-step catalytic cycle involving (1) ho-
moenolate formation, (2) aldehyde addition, and (3) homoal-
dolate silylation, all current evidence points toward the
catalytic process shown in Scheme 3. Thus, it is believed
that ring opening of 1 with titanium triflate 4b results in the
formation of an alkoxytitanium homoenolate.18 The TMSOTf
that is liberated in this process reacts with the aldehyde to
form an activated complex which undergoes addition with
the homoenolate 6 to form 3b directly.19 Normally, titanium
homoenolates do not require silicon activation in order to
add to aldehydes.4a However, if direct addition of the
homoenolate 6 to the aldehyde occurred, a titanate containing
both homoaldolate and isopropoxide ligands would be
generated. A control experiment where a mixed titanate,
containing Binol, isopropoxide, and homoaldolate ligands,
was prepared and treated with TMSOTf resulted in the
Acknowledgment. The authors thank NSERC and Bio-
Me´ga/Boehringer Ingelheim for support of this research.
Post-graduate scholarships to E.O.M. from NSERC, FCAR,
and McGill University are gratefully acknowledged.
(17) We have found that the catalyst loading can be reduced to 2% or
5% with little change in yield for the reaction with benzaldehyde. However,
longer reaction times were required, which was not practical for less reactive
substrates.
Supporting Information Available: Experimental pro-
cedures and characterization for all reaction products. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(18) In the absence of Binol-Ti(OiPr)2, TMSOTf does not promote the
homoaldol reaction of 1 with aldehydes. For this reason, and by analogy to
ref 4, it is presumed that homoenolate 6 is an intermediate in this process.
Thus far, it has not been possible to characterize the homoenolate.
(19) Similar activation of aldehydes has been postulated in the zinc-
catalyzed process (ref 4).
OL991017T
Org. Lett., Vol. 1, No. 10, 1999
1645