Effects of catalyst activation and ligand steric properties on the enantioselective allylation of amines and phenoxides

Article abstract of DOI:10.1021/ol050029d

(Chemical Equation Presented) The yields, enantioselectivities, and regioselectivities of the reactions of amines and phenoxides with allylic carbonates in the presence of a metallacyclic iridium catalyst were compared. These data show that both preactiva

Full text of DOI:10.1021/ol050029d

ORGANIC
LETTERS
2005
Vol. 7, No. 6
1093-1096
Effects of Catalyst Activation and
Ligand Steric Properties on the
Enantioselective Allylation of Amines
and Phenoxides
Andreas Leitner, Chutian Shu, and John F. Hartwig*
Department of Chemistry, Yale UniVersity, P.O. Box 208107,
New HaVen, Connecticut 6520-8107
john.hartwig@yale.edu
Received January 7, 2005
ABSTRACT
The yields, enantioselectivities, and regioselectivities of the reactions of amines and phenoxides with allylic carbonates in the presence of a
metallacyclic iridium catalyst were compared. These data show that both preactivation of the catalyst and the size of the ligand affect the
yield, enantioselectivity, and regioselectivity. With the activated catalyst containing a bis-naphthethylamino group, the allylic amination and
etherification of a broad range of allylic carbonates occurred in high yields and with high regioselectivities and enantioselectivities.
Allylic amines and ethers with stereocenters R to a hetero-
atom are valuable synthetic building blocks. A large number
of single enantiomer drug candidates and biologically active
molecules can be derived from these materials because of
the combination of heteroatom and olefinic functional groups.
A prominent method to prepare these compounds is transition
metal-catalyzed allylic substitution.1
Allylic substitution of acyclic, monosubstituted allylic
electrophiles catalyzed by complexes of Ru,2 Rh,3 and Ir4-12
often generate the more hindered, chiral, branched allylic
amine or ether products from either linear or branched allylic
carbonates or both. Recently, we reported the first highly
enantioselective, iridium-catalyzed allylation of amines and
alkoxides.7-12 Achiral linear allylic carbonates reacted with
high enantioselectivity when the iridium catalyst was gener-
(2) (a) Renaud, J.-L.; Bruneau, C.; Demerseman, B. Synlett 2003, 408.
(b) Mbaye, M. D.; Demerseman, B.; Renaud, J.-L.; Toupet, L. C.; Bruneau,
C. Angew. Chem., Int. Ed. 2003, 42, 5066. (c) Trost, B. M.; Fraisse, P. L.;
Ball, Z. T. Angew. Chem., Int. Ed. 2002, 41, 1059. (d) Morisaki, Y.; Kondo,
T.; Mitsudo, T. A. Organometallics 1999, 18, 4742. (e) Zhang, S.-W.;
Mitsudo, T.-a.; Kondo, T.; Watanabe, Y. J. Organomet. Chem. 1993, 450,
197.
(3) (a) Evans, P. A.; Nelson, J. D. Tetrahedron Lett. 1998, 39, 1725. (b)
Evans, P. A.; Robinson, J. E.; Nelson, J. D. J. Am. Chem. Soc. 1999, 121,
6761. (c) Evans, P. A.; Leahy, D. K. J. Am. Chem. Soc. 2000, 122, 5012.
(d) Evans, P. A.; Leahy, D. K.; Andrews, W. J.; Uraguchi, D. Angew. Chem.,
Int. Ed. 2004, 43, 4788. (e) Evans, P. A.; Lawler, M. J. J. Am. Chem. Soc.
2004, 126, 8642.
(1) (a) Trost, B. M. Chem. Pharm. Bull. 2002, 50, 1. (b) Negishi, E.-I.;
Liou, S.-Y. In Handbook of Organopalladium Chemistry for Organic
Synthesis; John Wiley & Sons, Inc.: New York, 2002; Vol. 2, pp 1769-
1793. (c) Acemoglu, L.; Williams, J. M. J. In Handbook of Organopalladium
Chemistry for Organic Synthesis; Wiley Interscience: New York, 2002;
Vol. 2, pp 1945-1979. (d) Pfaltz, A.; Lautens, M. In ComprehensiVe
Asymmetric Catalysis I-III; Springer: New York, 1999; Vol. 2, pp 833-
884.
(4) (a) Takeuchi, R.; Kashio, M. Angew. Chem., Int. Ed. Engl. 1997,
36, 263. (b) Takeuchi, R.; Kashio, M. J. Am. Chem. Soc. 1998, 120, 8647.
(c) Takeuchi, R. Polyhedron 2000, 19, 557. (d) Janssen, J. P.; Helmchen,
G. Tetrahedron Lett. 1997, 38, 8025. (e) Bartels, B.; Helmchen, G. Chem.
Commun. 1999, 741. (f) Bartels, B.; Garcia-Yebra, C.; Helmchen, G. Eur.
J. Org. Chem. 2003, 1097. (g) Fuji, K.; Kinoshita, N.; Tanaka, K.; Kawabata,
T. Chem. Commun. 1999, 2289. (h) Kanayama, T.; Yoshida, K.; Miyabe,
H.; Kimachi, T.; Takemoto, Y. J. Org. Chem. 2003, 68, 6197.
10.1021/ol050029d CCC: $30.25
© 2005 American Chemical Society
Published on Web 02/15/2005

Table 1. Effect of Catalyst Activation and Ligand Steric Properties on the Yield, Enantioselectivity, and Regioselectivity of the
Amination and Etherification of Allylic Carbonatesa
a Reactions were conducted at room temperature on a 1.0 mmol scale in THF (0.5 mL) with relative mole ratios of carbonate:amine:catalyst of 100:120:2.
b Isolated yields out of two independent runs. c Reactions were conducted with relative mole ratios of carbonate:amine:catalyst of 100:120:4. d Ratio of
linear:branched:diallylation product. e Reactions conducted with 2 equiv of aryloxide.
ated in situ from commercially available [Ir(COD)Cl]2 and
a phosphoramidite ligand L1.8,10-12 This ligand was originally
prepared and applied to copper-catalyzed processes by
Feringa and de Vries.13
We showed that the square planar [Ir(COD)(Cl)(L1)],
which was generated from L1 and [Ir(COD)(Cl)]2, reacts with
aliphatic amines or other basic nucleophiles to generate a
metallacyclic complex 1a in which the monodentate ligand
has become bidentate.5,10,14,15 Complex 1a appears to generate
the active catalyst by dissociation of the second monodentate
phosphoramidite.
We have published our use of this activated catalyst and
its analogues to improve the scope of the reactions. We
showed that the reactions of arylamines, which did not occur
with the combination of [Ir(COD)Cl]2 and phosphoramidite
ligand L1, occurred with broad scope and in high yield and
enantioselectivity if an aliphatic amine was added to the
system to induce the cyclometalation.8 We also showed that
a catalyst generated from the bis-naphthethyl analogue of
L1, phosphoramidite L2,16 formed an iridium complex that
catalyzed the first highly enantioselective allylation of
alkoxides.7
Considering that activation of the catalyst with amine prior
to addition of the nucleophile improved one set of reactions
(those of aromatic amines) and that use of the naphthethyl
ligand L2 instead of L1 improved another set of reactions
(those of alkoxides), we have conducted a study to reevaluate
the effect of catalyst activation with ligand L2 on the
reactions of amine and phenoxide nucleophiles we published
initially. We conducted these further studies both to deter-
mine if these two changes to the catalyst would improve
some of the less selective reactions of these two types of
nucleophiles and to determine whether one of the two
changes was more important than the other. We report that
these reactions occur with a combination of faster rates,
higher yields, increased regioselectivities, or increased enan-
tioselectivities as a result of these changes to the catalyst
and that both catalyst activation and changes to the ligand
structure contribute to these improvements.
We initially studied four reactions to determine if the
combination of the catalyst activated by cyclometalation prior
to the addition of the reagents and the use of ligand L2 to
generate the activated catalyst would improve the yields,
regioselectivities, and enantioselectivities of the allylation
processes (eq 1 and Table 1). These four reactions were (1)
the addition of amines to aliphatic allylic methyl carbonates,
which occurred with high enantioselectivities with the
original catalyst but lower yields and regioselectivities than
reactions of cinnamyl carbonates;12 (2) reaction of benzyl-
amine with methyl p-nitrocinnamyl carbonate, which oc-
curred with modest enantioselectivity and regioselectivity
under the original conditions;12 and (3) two reactions of aryl-
oxides with an aliphatic methyl carbonate, which occurred
with modest to good regioselectivities and enantioselectivities
with the original catalyst.11 Table 1 summarizes the reactions
we published originally, the reactions conducted by addition
of the reagents after activation of the catalyst generated from
L1, and the reactions conducted by adding the reagents after
activation of the catalyst containing ligand L2. The catalyst
(5) Bartels, B.; Garcia-Yebra, C.; Rominger, F.; Helmchen, G. Eur. J.
Inorg. Chem. 2002, 2569.
(6) Lipowsky, G.; Helmchen, G. Chem. Commun. 2004, 116. The authors
report for the reaction of the phenyl-substituted dienyl carbonate with benzyl
amine a b/l ratio of 99:1. The corresponding reaction with the TBDMSOCH2-
CH2- and PMBOCH2CH2-substituted dienyl carbonate gave b/l ratios of
94:6 and 96:4, respectively.
(7) Shu, C.; Hartwig, J. F. Angew. Chem., Int. Ed. 2004, 43, 4794.
(8) Shu, C.; Leitner, A.; Hartwig, J. F. Angew. Chem., Int. Ed. 2004, 43,
4797.
(9) Leitner, A.; Shu, C.; Hartwig, J. F. Proc. Natl Acad. Sci. U.S.A. 2004,
101, 5830.
(10) Kiener, C. A.; Shu, C.; Incarvito, C.; Hartwig, J. F. J. Am. Chem.
Soc. 2003, 125, 14272.
(11) Lopez, F.; Ohmura, T.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125,
3426.
(12) Ohmura, T.; Hartwig, J. F. J. Am. Chem. Soc. 2002, 124, 15164.
(13) (a) de Vries, A. H. M.; Meetsma, A.; Feringa, B. L. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 2374. (b) Feringa, B. L. Acc. Chem. Res. 2000, 33,
346.
(14) Helmchen and co-workers mentioned in ref 5 the potential that
cyclometalation of triphenyl phosphite leads to the active catalyst upon
addition of nucleophiles to the related [Ir(COD)(L)Cl] complex with L )
triphenyl phosphite. See also: Lipowsky, G.; Miller, N.; Helmchen, G.
Angew. Chem., Int. Ed. 2004, 43, 4595.
(15) Tissot-Croset, K.; Polet, D.; Alexakis, A. Angew. Chem., Int. Ed.
2004, 43, 2426. The authors report reactions with a phosphoramidite ligand
with an anisyl group, which could be hemilabile, and which generates an
Ir catalyst for allylic amination with cinnamyl methyl carbonate and
alkylamines with rates and regio- and enantioselectivities that are similar
to those of the catalyst in ref 10.
(16) Naasz, R.; Arnold, L. A.; Minnaard, A. J.; Feringa, B. L. Angew.
Chem., Int. Ed. 2001, 40, 927.
1094
Org. Lett., Vol. 7, No. 6, 2005

activation of the catalyst before addition of the reagents and
by increasing the size of the amino group.
With the information that both the activation of the catalyst
and increased size of the amino group improve rates, yields,
and selectivity, we assessed the scope of the reactions of
amines and phenoxides with this catalyst system. A com-
parison of the yield, enantiomeric excess, and branched-to-
linear regioselectivity of reactions with the activated catalysts
containing ligands L1 and L2 is shown in Table 2. These
data show that yield, enantioselectivity, or branched-to-linear
regioselectivity is improved with the catalyst containing L2
over the values obtained with the catalyst containing L1.
For example, entries 1-3 of Table 2 show that the
reactions of a simple, unhindered aliphatic allylic carbonate
occurred in good to excellent yields, enantioselectivities, and
regioselectivities with benzylic and heteroarylmethylamines
in the presence of the activated catalyst generated from ligand
L2. Although none of the reactions of this allylic carbonate
occurred in poor yield or with selectivities, with ligand L1
either yield or selectivity was improved substantially by the
generation of the catalyst from L2. The reaction of benzyl-
amine with the linear aliphatic carbonate occurred in higher
yield and with measurably higher enantioselectivity (entry
1), while reactions of the heteroarylmethylamines occurred
with higher yields, enantioselectivities, and regioselectivities
(entries 2 and 3) than did reactions with the catalyst generated
from L1. The change from ligand L1 to L2 also improved
reactions of benzylamine with more hindered aliphatic allylic
carbonates. For example, the yield of the reaction of the
allylic carbonate in entry 4, which contains a branch point
R to the allyl unit, occurred in only 66% yield, 89% ee, and
roughly 10:1 branched-to-linear regioselectivity with the
catalyst derived from L1, but it occurred in 90% yield, 94%
Figure 1. Phosphoramidites used to generate catalysts 1a and 1b.
was “activated” by heating [Ir(COD)Cl]2 and L1 or [Ir-
(COD)Cl]2 and L2 with propylamine at 50 °C for 20 min to
generate the metallacyclic core structure, followed by
evaporation of the volatile materials before the addition of
the reaction solvent and the two reagents.8
An evaluation of the results in Table 1 shows that both
activation of the catalyst and a change in the steric properties
of the ligand led to improvements in yields, regioselectivities,
and enantioselectivities. For example, the branched-to-linear
regioselectivity of the reaction of benzylamine with the linear
aliphatic allylic carbonate in entry 1 was improved by
generating the activated catalyst prior to the addition of the
substrates, and the enantioselectivity was improved by
increasing the size of the arylethyl group. In contrast, the
enantioselectivity of the reaction of benzylamine with the
electron-poor cinnamyl substrate in entry 2 was improved
by activation of the catalyst prior to addition of the reagents,
and the regioselectivity was improved by increasing the size
of the arylethyl group. In contrast to each of these results,
both the enantioselectivity and the regioselectivity of the
reactions of the phenoxides were improved incrementally by
Table 2. Comparison of the Yields and Selectivities of the Activated Catalysts Derived from Ligands L1 and L2a
a Reactions were conducted at room temperature on a 1.0 mmol scale in THF (0.5 mL) with a relative mole ratio of carbonate/nucleophile/catalyst of
100:120:2. b Isolated yields are an average from two independent runs. c Room temperature; reaction times were not optimized. d Catalyst was activated in
situ with DABCO; relative mole ratio of carbonate:amine:DABCO/[Ir(COD)Cl]2/L was 100:120:10:1:2. e Performed with 4% catalyst. f Performed with 2
equiv of nucleophile.
Org. Lett., Vol. 7, No. 6, 2005
1095

ee, and nearly 20:1 branched-to-linear regioselectivity with
the catalyst from L2.
with or without activation prior to addition of the reagents,
the rates are not. To avoid a long induction period, the cyclo-
metalation process must occur on the time scale of minutes,
and cyclometalation at room temperature requires hours.
Thus, reactions without activation of the catalyst prior to
addition of the reagents occur with an induction period and
without the full concentration of active catalyst. Consistent
with this analysis, the reaction of benzylamine with the
carbonate derived from (E)-2-hexen-1-ol (Table 2, entry 1)
without activation of the catalyst required 10 h to proceed
to completion, while the reaction with initial activation with
propylamine occurred within 2 h.
In summary, we have shown that a combination of
generation of the metallacyclic catalyst prior to addition of
the reagents and a change in the bis-arylethylamino group
on the phosphoramidite from the bis-phenethylamino group
in L1 to the bis-naphthethylamino group in L2 improves the
rates, yields, regioselectivity, and enantioselectivity of many
reactions of allylic carbonates with amines and phenoxides.
Both preactivation of the catalyst and a change in the steric
properties of the ligand improve these reactions. Structural
information on the allyl intermediates would help to explain
the changes in selectivity, and the synthesis of such a
complex is one goal of our future work.
The reactions of dienyl carbonates to give optically active
amines with two different olefinic units have been reported
by us with anilines and by Helmchen et al.6 with benzyl-
amine. Although reactions catalyzed by complexes generated
from ligand L1 occurred with good enantioselectivity, the
yields were modest and the branched-to-linear regioselec-
tivities of reactions of the aliphatic dienyl carbonate were
lower than that of the reaction of the analogous phenyl-
substituted dienyl carbonate. In contrast, the reactions with
the catalyst generated from ligand L2 occurred with uni-
formly high regioselectivities, enantioselectivities, and yields
(entries 5-8). For example, the regioselectivity of the
reactions of furylmethylamine and thienylmethylamine were
improved from 4:1 to greater than 10:1, and the enantiose-
lectivity of the reaction of thienylmethylamine improved
from 92 to 98%. Although the reactions of anilines occurred
with high enantioselectivities with the catalyst generated from
ligand L1, the regioselectivity was improved from roughly
4:1 to over 30:1 by generating the catalyst from L2.
Most reactions of cinnamyl carbonates with aliphatic and
benzylic amines occurred with high regio- and enantiose-
lectivity with ligand L1, but reaction of the strongly electron-
poor cinnamyl carbonate with a p-nitro group occurred with
low regioselectivity, even with the activated catalyst. In
contrast, the reaction of benzylamine catalyzed by the
complex derived from ligand L2 occurred in high yield, ee,
and regioselectivity (entry 9).
Acknowledgment. We thank the NSF for support of this
work. A.L. is a recipient of the FWF Austria Schro¨dinger
Fellowship. We also thank Boehringer Ingelheim for an
unrestricted gift and Johnson-Matthey for iridium chloride.
One might wonder whether reactions of basic nucleophiles
conducted with a mixture of [Ir(COD)Cl]2 and ligand L2
without activation prior to addition of the reagents are as
effective as reactions of the same nucleophiles conducted
with the preformed metallacyclic catalyst derived from L2.
Although the selectivities and yields may be equally high
Supporting Information Available: Experimental sec-
tion containing reaction procedures and characterization of
reaction products. This material is available free of charge
via the Internet at http://pubs.acs.org.
OL050029D
1096
Org. Lett., Vol. 7, No. 6, 2005