A simple iridium catalyst with a single resolved stereocenter for enantioselective allylic amination. Catalyst selection from mechanistic analysis

Article abstract of DOI:10.1021/ja054331t

A study of the relationship between the stereochemical elements of a phosphoramidite ligand and the stereoselectivity of iridium-catalyzed amination of allylic carbonates is reported. During catalyst activation, a complex of a phosphoramidite ligand possessing one axial chiral binaphtholate group and two resolved phenethyl substituents converts to a more reactive cyclometalated complex containing one distal chiral substituent at nitrogen, one substituent that becomes part of the metalacycle, and one unperturbed binaphtholate group. Systematic changes were made to the different stereochemical elements. Replacement of the distal chiral phenethyl substituent with a large achiral cycloalkyl group led to a catalyst that reacts with rates and enantioselectivities that are similar to those of the original catalyst with the phenethyl group. Studies of the reactions of diastereomeric ligands containing (R) or (S) binaphtholate groups on phosphorus, along with one (R)-phenethyl and one achiral cyclododecyl group on nitrogen, show that the complexes of the two diastereomeric ligands undergo cyclometalation at much different rates. To access both diastereomeric catalysts and to determine if the reaction can occur selectively with an even simpler ligand containing a phenethyl substituent at nitrogen as the only resolved stereochemical element, the catalyst derived from a phosphoramidite containing a biphenolate group was studied. Catalysts generated from this ligand were shown to react in all cases examined with nearly the same rates, regioselectivities, and enantioselectivities as catalysts derived from the original more elaborate ligand. The absolute stereochemistry of the product implies that the major enantiomer is formed from the (R _a,R_c)-atropisomer of the catalyst containing the biphenolate group.

Full text of DOI:10.1021/ja054331t

Published on Web 10/12/2005
A Simple Iridium Catalyst with a Single Resolved Stereocenter
for Enantioselective Allylic Amination. Catalyst Selection from
Mechanistic Analysis
Andreas Leitner, Shashank Shekhar, Mark J. Pouy, and John F. Hartwig*
Contribution from the Department of Chemistry, Yale UniVersity, P.O. Box 208107,
New HaVen, Connecticut 06520-8107
Received June 30, 2005; E-mail: john.hartwig@yale.edu
Abstract: A study of the relationship between the stereochemical elements of a phosphoramidite ligand
and the stereoselectivity of iridium-catalyzed amination of allylic carbonates is reported. During catalyst
activation, a complex of a phosphoramidite ligand possessing one axial chiral binaphtholate group and
two resolved phenethyl substituents converts to a more reactive cyclometalated complex containing one
distal chiral substituent at nitrogen, one substituent that becomes part of the metalacycle, and one
unperturbed binaphtholate group. Systematic changes were made to the different stereochemical elements.
Replacement of the distal chiral phenethyl substituent with a large achiral cycloalkyl group led to a catalyst
that reacts with rates and enantioselectivities that are similar to those of the original catalyst with the
phenethyl group. Studies of the reactions of diastereomeric ligands containing (R) or (S) binaphtholate
groups on phosphorus, along with one (R)-phenethyl and one achiral cyclododecyl group on nitrogen, show
that the complexes of the two diastereomeric ligands undergo cyclometalation at much different rates. To
access both diastereomeric catalysts and to determine if the reaction can occur selectively with an even
simpler ligand containing a phenethyl substituent at nitrogen as the only resolved stereochemical element,
the catalyst derived from a phosphoramidite containing a biphenolate group was studied. Catalysts generated
from this ligand were shown to react in all cases examined with nearly the same rates, regioselectivities,
and enantioselectivities as catalysts derived from the original more elaborate ligand. The absolute
stereochemistry of the product implies that the major enantiomer is formed from the (Ra,Rc)-atropisomer of
the catalyst containing the biphenolate group.
Introduction
selectivity. Along with providing information on the interplay
between stereochemical elements, these studies have led to a
Information on the structures and reactions of intermediates
in catalytic reactions can guide the selection of improved
catalysts. At the same time, the translation of this structural
information into improved catalysts for enantioselective pro-
cesses can be challenging. A recent study from our laboratories
provides one example of how distinct an active catalyst can be
from its catalyst precursors.1 An iridium-phosphoramidite
complex, which catalyzes the amination and etherification of
allyliccarbonateswithhighregioselectivityandenantioselectivity,2-7
catalyzes these reactions after cyclometalation drastically alters
its structure.1 To help determine the relationship between the
structure and reactivity of the resulting complex, we have
conducted studies to determine the influence of the different
portions of the cyclometalated structure on reactivity and
catalyst that reacts with yields, rates, and enantioselectivities
that are comparable to those of the original more elaborate
catalyst, but with a ligand that contains one phenethyl group as
the sole resolved stereochemical component.
The aminations and etherifications of terminal allylic carbon-
ates catalyzed by iridium complexes of certain phosphoramidites
form branched chiral amines and ethers in high yields, with
high branched-to-linear ratios and with high enantiomeric
excesses.1-14 The regioselectivity of this process complements
the typical regioselectivity of palladium-catalyzed allylic ami-
nation. Further, the simplicity of the synthesis of phosphor-
amidite ligands contrasts with the more lengthy synthesis of
some ferrocenyl ligands that have provided branched allylic
amines with high branched-to-linear ratios and high enantiose-
(1) Kiener, C. A.; Shu, C.; Incarvito, C.; Hartwig, J. F. J. Am. Chem. Soc.
2003, 125, 14272.
(8) For early iridium-catalyzed allylic aminations with tosylamides in low ee’s,
see the next two references.
(9) Bartels, B.; Helmchen, G. Chem. Commun. 1999, 741.
(10) Bartels, B.; Garcia-Yebra, C.; Rominger, F.; Helmchen, G. Eur. J. Inorg.
Chem. 2002, 2569.
(11) Lipowsky, G.; Helmchen, G. Chem. Commun. 2004, 116.
(12) For seminal work on iridium-catalyzed allylic substitutions with achiral
catalysts, see the next two references.
(13) Takeuchi, R.; Kashio, M. Angew. Chem., Int. Ed. Engl. 1997, 36, 263.
(14) Takeuchi, R. Polyhedron 2000, 19, 557.
(2) Ohmura, T.; Hartwig, J. F. J. Am. Chem. Soc. 2002, 124, 15164.
(3) Lopez, F.; Ohmura, T.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125,
3426.
(4) Leitner, A.; Shu, C. T.; Hartwig, J. F. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 5830.
(5) Shu, C. T.; Hartwig, J. F. Angew. Chem., Int. Ed. 2004, 43, 4794.
(6) Shu, C. T.; Leitner, A.; Hartwig, J. F. Angew. Chem., Int. Ed. 2004, 43,
4797.
(7) Leitner, A.; Shu, C. T.; Hartwig, J. F. Org. Lett. 2005, 7, 1093.
9
15506
J. AM. CHEM. SOC. 2005, 127, 15506-15514
10.1021/ja054331t CCC: $30.25 © 2005 American Chemical Society

Iridium Catalyst with a Single Resolved Stereocenter
A R T I C L E S
elements and the basis for the difference in activity between
catalysts generated from different diastereomeric ligands was
not clear from the initial studies with the catalyst generated in
situ or even after identifying the cyclometalated structure of
the activated catalyst.
We reported in communication form that a ligand (L2) with
a resolved binaphthyl group, one phenethyl group, and one
achiral N-benzyl group distal to the metal generates a complex
that catalyzes the allylation of cinnamyl carbonate with enan-
tioselectivities higher than 90%.4,32 Reactions of this catalyst
were much slower than those of the original catalyst, but these
results demonstrated that the more distal stereocenter in the
cyclometalated complex could be omitted while maintaining
high selectivity.33-35 We have now conducted detailed studies
on the effect of the distal substituent, the origins of the difference
in reactivity of diastereomeric catalysts, and studies that evaluate
the effect of eliminating each of the resolved stereochemical
elements. These studies have (1) led to a family of new C1-
symmetric phosphoramidite ligands that reveal the relative
importance of different structural features of the ligands on
reaction rate and enantioselectivity, (2) shown an unexpected
origin of the difference in reactivity of diastereomeric catalysts,
and (3) led us to design a catalyst that reacts with nearly equal
enantioselectivity as the original catalyst, but which contains a
single phenethylamino substituent as the sole resolved stereo-
chemical element. Because phenethylamine is an inexpensive
optically active building block, these studies uncover a practical
catalyst for allylic substitution in concert with revealing concepts
that should further advance efforts to design catalysts for
enantioselective transformations.
Figure 1. Original phosphoramidite ligand L1 and activated cyclometalated
catalyst 1.
Figure 2. Stereochemical elements of the cyclometalated Ir(I) complex
generated with ligand L1.
lectivity from amination reactions with palladium catalysts.15
The selectivity of the iridium-catalyzed reactions is more akin
to that of molybdenum-catalyzed allylic substitutions,16-21 but
the scope of the allylation reactions catalyzed by the iridium
complexes encompasses heteroatom nucleophiles that are not
included, as least currently, in the scope of reactions catalyzed
by molybdenum complexes. These reactions also contrast with
rhodium-catalyzed allylic aminations and etherifications that
have been conducted so far with achiral catalysts.22-31
The first highly enantioselective, iridium-catalyzed aminations
and etherifications of allylic carbonates were conducted with
phosphoramidite ligand L1 containing a binaphtholate unit and
a bis-phenethylamino group.2,3,6 The active catalyst in these
reactions is generated by cyclometalation at one methyl group
of the phenethylamino substituent.1 This cyclometalation breaks
the C2 symmetry of the ligand and generates a product with C1
symmetry. The structure of the presumed active catalyst
stabilized by a fifth dative ligand, complex 1, is shown in Figure
1.
In principle, this information that the active catalyst contains
the cyclometalated structure should allow one to prepare new
ligands for this catalytic process by choosing structures in a
logical manner. Moreover, the cyclometalated structure provides
a platform to study the origin of the effects of changes to the
different stereochemical elements of the ligand (see Figure 2)
and the interconnections between these elements on enantiose-
lectivity. A hierarchy among the different stereochemical
Results and Discussion
1. Initial Ligand Design. Our first studies were conducted
to determine if the distal phenethyl group in the cyclometalated
complex of Figure 2 could be replaced by an achiral substituent
that would impart steric and electronic properties that are
analogous to those of a phenethyl group. To do so, we prepared
a series of phosphoramidites L3-L16 in Table 1 that contain
(R)-1,1′-bi-2-naphthol ((R)-BINOL) and an amino group con-
taining one arylethyl substituent and one achiral cyclic or acyclic
aliphatic or benzylic substituent. After cyclometalation at the
arylethyl group, the achiral substituent would lie distal to the
metal. We focused on the synthesis of ligands with achiral
substituents that lacked a methyl group because the absence of
a methyl group in the achiral substituent would prevent
competitive cyclometalation at the arylethyl substituent and the
achiral substituent. Further, we chose structures that could be
prepared from amines that are either available commercially or
could be prepared by simple reductive amination of the
appropriate cyclic or acyclic ketones with an optically active
arylethylamine. As described in the final section of this paper,
these studies led to further development of catalysts that lacked
an optically active biaryl group as the second stereochemical
element.
(15) You, S. L.; Zhu, X. Z.; Luo, Y. M.; Hou, X. L.; Dai, L. X. J. Am. Chem.
Soc. 2001, 123, 7471.
(16) Trost, B. M.; Hachiya, I. J. Am. Chem. Soc. 1998, 120, 1104.
(17) Kocovsky, P.; Malkov, A. V.; Vyskocil, S.; Lloyd-Jones, G. C. Pure Appl.
Chem. 1999, 71, 1425.
(18) Palucki, M.; Um, J. M.; Conlon, D. A.; Yasuda, N.; Hughes, D. L.; Mao,
B.; Wang, J.; Reider, P. J. AdV. Synth. Catal. 2001, 343, 46.
(19) Krska, S. W.; Hughes, D. L.; Reamer, R. A.; Mathre, D. J.; Sun, Y.; Trost,
B. M. J. Am. Chem. Soc. 2002, 124, 12656.
(20) Trost, B. M.; Dogra, K.; Hachiya, I.; Emura, T.; Hughes, D. L.; Krska, S.;
Reamer, R. A.; Palucki, M.; Yasuda, N.; Reider, P. J. Angew. Chem., Int.
Ed. 2002, 41, 1929.
(21) Trost, B. M.; Dogra, K. J. Am. Chem. Soc. 2002, 124, 7256.
(22) Evans, P. A.; Nelson, J. D. Tetrahedron Lett. 1998, 39, 1725.
(23) Evans, P. A.; Nelson, J. D. J. Am. Chem. Soc. 1998, 120, 5581.
(24) Evans, P. A.; Robinson, J. E.; Nelson, J. D. J. Am. Chem. Soc. 1999, 121,
6761.
(32) Leitner, A.; Hartwig, J. F. Abstracts of Papers; 228th ACS National
Meeting, Philadelphia, PA, August 22-26, 2004, ORGN.
(33) For initial disclosure of unsymmetrical ligands containing binaphtholate
groups and groups at nitrogen other than benzyl that generate more reactive
catalysts, see ref 32. For subsequent related studies by Helmchen, see the
next two references.
(34) Welter, C.; Dahnz, A.; Brunner, B.; Streiff, S.; Dubon, P.; Helmchen, G.
Org. Lett. 2005, 7, 1239.
(35) Streiff, S.; Welter, C.; Schelwies, M.; Lipowsky, G.; Miller, N.; Helmchen,
G. Chem. Commun. 2005, 2957.
(25) Evans, P. A.; Leahy, D. K. J. Am. Chem. Soc. 2000, 122, 5012.
(26) Evans, P. A.; Kennedy, L. J. J. Am. Chem. Soc. 2001, 123, 1234.
(27) Evans, P. A.; Robinson, J. E.; Moffett, K. K. Org. Lett. 2001, 3, 3269.
(28) Evans, P. A.; Leahy, D. K. J. Am. Chem. Soc. 2002, 124, 7882.
(29) Evans, P. A.; Uraguchi, D. J. Am. Chem. Soc. 2003, 125, 7158.
(30) Evans, P. A.; Leahy, D. K.; Andrews, W. J.; Uraguchi, D. Angew. Chem.,
Int. Ed. 2004, 43, 4788.
(31) Evans, P. A.; Lawler, M. J. J. Am. Chem. Soc. 2004, 126, 8642.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15507

A R T I C L E S
Leitner et al.
Table 1. Effect of Substituents in C1-Symmetric Phosphoramidite
Ligands (Ra,Rc)-L3-16 on the Allylic Amination of Eq 3 with
Catalysts Generated by the Procedure in Eq 1a
Equation 1 shows the protocol followed to generate the
activated catalyst from the various phosphoramidite ligands, and
Table 1 includes the substituents on the phosphoramidites L3-
L16 that were studied as part of this work. The cyclometalated
complexes were generated by addition of 1 equiv of the ligand
to 0.5 equiv of [Ir(COD)Cl]2 to form [Ir(COD)Cl(LX)] (X )
3-16), followed by heating this iridium chloride complex with
30 equiv of propylamine at 50 °C for 20 min and evaporation
of the volatile materials to generate an air-stable and moisture-
stable solid. We have shown with ligand L6 that this procedure
forms 1 equiv of the trigonal bipyramidal complex [(COD)-
Ir(κ2-L6)(κ1-L6)] (2), which is analogous to 1, in 91% yield
(determined by 31P NMR spectroscopy with an internal standard
in a capillary tube). Because complex 2 contains one cyclo-
metalated and one monodentate phosphoramidite ligand, this
procedure regenerates 1 equiv of [Ir(COD)Cl]2 during the
cyclometalation. Thus, the final solid that is added as catalyst
is a 1:1 mixture of 2 and 1/2 [Ir(COD)Cl]2. We have shown
that the presence of [Ir(COD)Cl]2 accelerates the catalytic
process by accepting the κ1-phosphoramidite and promoting the
equilibrium for dissociation of this ligand from 1.38
31P NMR spectra of the crude solutions generated upon
heating [Ir(COD)Cl]2 with phosphoramidites L3-L16 and
propylamine at 50 °C indicated that the complexes generated
from ligands that lack methyl groups on the amino substituents
undergo cyclometalation to form complexes that are analogous
to 1 and 2. The formation of these cyclometalated complexes
was revealed by the appearance of two doublets in the 31P NMR
spectrum at chemical shifts similar to those of 1 after heating
with propylamine. Only the square planar Ir(I) complex from
[Ir(COD)Cl]2 and isopropyl-substituted ligand L10 generated
mixtures of species upon heating with propylamine. Different
species could be generated from L10 by cyclometalation at the
phenethyl group or at one of the diastereotopic methyl groups
of the isopropyl substituent. Several products formed from
heating with propylamine, and we were unable to isolate a pure
species. Nevertheless, the formation of mixtures of products was
consistent with our hypothesis that competitive cyclometalation
could occur at the achiral substituent if it contained methyl
groups â to nitrogen.
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:amine:catalyst of
100:120:2. b Enantioselectivities were determined by chiral HPLC; the R
enantiomer was the major enantiomer in all reactions. c Branched-to-linear
ratios were determined by crude 1H NMR. d Reactions were monitored by
gas chromatography. e Not determined because of the low yield of substitu-
tion product.
2. Ligand and Catalyst Preparation and Structure. All of
the phosphoramidite ligands were prepared following protocols
reported by Alexakis36 and closely related to those reported by
Feringa37 and their co-workers for the preparation of C2-
symmetric phosphoramidites. This method simply involves
reaction of PCl3 with a secondary amine and NEt3, followed
by addition of (R)-BINOL. All ligands in this study were isolated
in pure form and were characterized by 1H, 13C, and 31P NMR
spectroscopy, as well as microanalysis.
To obtain a more crystalline product, we prepared a complex
with PPh3 as the κ1 ligand by cyclometalation of the Ir(I)
complex of phosphoramidite L5 and replacement of the mono-
dentate phosphoramidite with PPh3 (eq 2). This complex was
(36) Alexakis, A.; Rosset, S.; Allamand, J.; March, S.; Guillen, F.; Benhaim,
C. Synlett 2001, 1375.
(37) Feringa, B. L. Acc. Chem. Res. 2000, 33, 346.
(38) Other more practical methods, involving addition of copper, to promote
this dissociation have also been developed and will be reported in due
course.
9
15508 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Iridium Catalyst with a Single Resolved Stereocenter
A R T I C L E S
formed in the crude reaction as a 1:0.4 mixture of isomers that
appears to result from varied stereochemistry at the metal. The
complex crystallized as a single isomer, and this form of 3 was
characterized by conventional spectroscopic methods and X-ray
diffraction. An ORTEP diagram of 3 is shown in Figure 3. This
complex is nearly superimposable on the structure of the PPh3
adduct of the catalyst generated from L1,1 except that the
cycloalkyl group resides at the position of the phenethyl group.
Figure 3. ORTEP diagram of the trigonal bipyramidal Ir(I)-complex 3
with PPh3 as the monodentate phosphorus ligand.
3. Reactions Catalyzed by Cyclometalated Complexes of
L3-L6. The ability of ligands L3-L16 to generate catalysts
for enantioselective amination of terminal allylic carbonates was
probed by conducting the reaction of methyl cinnamyl carbonate
with p-methoxy benzylamine at room temperature (eq 3)
catalyzed by the complex generated according to eq 1. The
branched-to-linear ratios, enantioselectivity of the branched
product, and the approximate times required to reach >95%
conversion are shown in Table 1 for the reactions catalyzed by
cyclometalated complexes of the various ligands.
Third, one sees from these data that the reactions with the
catalyst containing an isopropyl substituent on nitrogen react
with reasonably high enantioselectivity and fast rates. Although
this catalyst is not the most reactive or selective, and conclusions
from studies of this ligand can be made only tentatively, the
results with this ligand do warrant a brief comment. We showed
previously that the reaction of cinnamyl carbonate with ben-
zylamine catalyzed by the complex generated from the phos-
phoramidite derived from diisopropylamine occurs with a
considerable 61% enantioselectivity.1 The major enantiomer
from this reaction was the same as that from reactions catalyzed
by the complex of the phosphoramidite with a bis-phenethyl-
amino group. We assume that the active catalyst components
that result from cyclometalation of the isopropyl group of the
ligand L10 would also form the same enantiomer as the catalysts
generated from the bis-phenethyl ligand L1 and the previously
studied2 diisopropylamino ligand (binaphtholate)P(NPri2). If so,
then the composite selectivity of the species generated from
cyclometalation at the phenethyl and isopropyl groups of L10
would lie between the 61% and 95% ee of the two catalysts
generated from these two ligands. The product from the reaction
catalyzed by complexes generated from L10 was formed in 89%
ee. The greater similarity of this value to the enantiomeric excess
from the reactions catalyzed by the complexes generated from
L1 suggests that the species resulting from cyclometalation at
the phenethyl group is formed in larger quantities or is more
reactive than the species resulting from cyclometalation at the
isopropyl group.
From these data, one can extract several trends. First, a
comparison of the results in entries 1-7 shows that increased
size of the achiral distal substituent attached to the nitrogen leads
to improved enantioselectivities. As the size of the cycloalkyl
group was increased from six to seven to eight and then to a
12-membered carbocycle, the enantioselectivity increased. The
reaction also required the shortest times when catalyzed by the
complex derived from the ligand with the largest 12-membered
carbocycle. A comparison of these results to those of catalysts
derived from ligands with benzylic or fluorenyl substituents
showed that the rates were much faster when this substituent
was more three-dimensional than planar.
Second, one can conclude that variation of the aryl group on
the arylethyl substituent does not dramatically affect the
enantioselectivity or regioselectivity but does allow for fine-
tuning of enantioselectivity and presumably matching of sub-
strate with catalyst. The highest enantioselectivity was observed
with the catalyst derived from the ligand with a 2,4-dichloro-
phenyl group. The fastest rates were observed with the catalyst
derived from the ligand with a 2-methoxy group. This fast rate
parallels the results reported by Alexakis with a bis-arylethyl
substituent on the phosphoramidite39 and a recent result of
Helmchen.34 The highest branched-to-linear selectivity was
measured with the catalyst derived from ligands containing a
2-naphthyl or a 4-fluorophenyl group, although these small
differences in an already high selectivity are difficult to clearly
1
assess by H NMR spectroscopy.
Allylic aminations catalyzed by the mixture of [Ir(COD)Cl]2
and cyclometalated complex 2 generated from L6, which
(39) Tissot-Croset, K.; Polet, D.; Alexakis, A. Angew. Chem., Int. Ed. 2004,
43, 2426.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15509

A R T I C L E S
Leitner et al.
Table 2. Amination of Allylic Carbonates Catalyzed by Complexes
Generated from Ligand L6a
Figure 4. Comparison of relative rates for the reactions of aniline with
methyl cinnamyl carbonate catalyzed by (1) (Ra,Rc)-2 formed from (Ra,Rc)-
L6; (2) (Sa,Rc)-2 formed from (Sa,Rc)-L6; and (3) (raca,Rc)-2, which refers
to the stereoisomeric forms of 2 generated from a 1:1 mixture of (Ra,Rc)-
L6 and (Sa,Rc)-L6.
stereoisomers of L6 at 50 °C for 20 min, followed by removal
of all volatile materials.
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/amine/catalyst of
100:120:2. b Isolated yields from two independent runs. c Branched-to-linear
1
ratio-determined from crude H NMR spectra.
contains a cyclododecyl group as the distal substituent at the
nitrogen in the metalacycle, occur in yields and enantioselec-
tivities that are as high as those with the catalyst derived from
the original ligand L1. Results from a series of reactions of
aromatic and aliphatic amines with both aliphatic and cinnamyl
carbonates catalyzed by the complex generated from L6 are
summarized in Table 2. The reactions of methyl cinnamyl
carbonate with benzylamine and two heteroarylmethylamines
occurred in good to excellent yields with high branched-to-linear
ratios and high enantiomeric excesses with 1 mol % of
[Ir(COD)Cl]2 and 2 mol % of L6. Further, reactions of acyclic
secondary amines with this allylic carbonate occurred in
excellent yield with high selectivities, and reactions of aromatic
amines occurred with similarly high chemoselectivity and
enantioselectivity. Finally, the reactions of a linear aliphatic
carbonates occurred in equally high yield and selectivity with
the simpler ligand L6 as with the original more elaborate ligand
L1.
4. Origins of the Difference in Activity of Diastereomeric
Catalysts. a. Relative Rates of Reaction of Catalysts Derived
from Diastereomeric Ligands. With a catalyst that is highly
active and selective and that contains only two stereochemical
elements, the phenethyl group and the axial chirality of the
binaphtholate unit, we sought to determine the relative contribu-
tions of these two elements to the enantioselectivity. To begin
these studies, we determined the relative rates and enantiose-
lectivity of reactions with each of the two diastereomeric ligands
and with a mixture of the two diastereomers. These studies of
the reactivity of the diastereomers revealed an unexpected origin
of the difference in reactivity.
The reaction is clearly faster when catalyzed by the complex
derived from the (Ra,Rc)-diastereomer of ligand L6 than it is
from that derived from the (Sa,Rc)-diastereomer. These data are
consistent with our previous observation that the reactions
catalyzed by complexes derived from (Ra,Rc,Rc)-L1 occurred
much faster than those derived from the diasteromeric (Sa,Rc,Rc)-
L1. The allylic amine products formed from the allylic substitu-
tion catalyzed by complexes generated from (Ra,Rc)-L6 and
(Sa,Rc)-L6 are antipodes of each other. We had previously shown
that the major enantiomers of the allylic amines formed from
catalysts generated from (Ra,Rc,Rc)-L1 and from (Sa,Rc,Rc)-L1
are also antipodes of each other.2
We also evaluated the rate of the same reaction catalyzed by
complexes of the mixture of diastereomeric (raca,Rc)-L6 ligands.
Reactions catalyzed by the complexes derived from the mixture
of diastereromeric (raca,Rc)-L6 ligands occurred at rates that
are closer to those of reactions catalyzed by the complex
generated from pure (Ra,Rc)-L6 than from pure (Sa,Rc)-L6.
However, this allylic amination is somewhat slower when
catalyzed by complexes derived from (raca,Rc)-L6 than from
pure (Ra,Rc)-L6. The enantioselectivity of the reaction initiated
with the mixture of diastereomeric (raca,Rc)-L6 ligands was
Figure 4 shows the appearance of product from the reaction
of aniline with methyl cinnamyl carbonate (eq 4) catalyzed by
the complexes (Ra,Rc)-2 and (Sa,Rc)-2 (generated from the (Ra,Rc)
and (Sa,Rc) diastereomers of ligand L6) and the catalyst
generated from (raca, Rc)-2 in THF-d8 at 25 °C. The appearance
1
of product was determined by monitoring the reactions by H
NMR spectroscopy. The catalyst was generated by reaction of
propylamine with a mixture of [Ir(COD)Cl]2 and the different
9
15510 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Iridium Catalyst with a Single Resolved Stereocenter
A R T I C L E S
These data show that the difference in reactivity of complexes
generated from the diasteromeric catalysts is due, at least in
part, to differences in the propensity of the two diastereomeric
square planar complexes {Ir(COD)Cl[(Ra,Rc)-L6)]} and {Ir-
(COD)Cl[(Sa,Rc)-L6)]} to undergo cyclometalation. The small
difference in enantioselectivity between the reaction of p-
methoxybenzylamine and methyl cinnamyl carbonate catalyzed
by the complex generated from the pure (Ra,Rc)-L6 ligand and
the mixture of diastereomers (raca,Rc)-L6 (90%) implies that
only a few percent of the minor enantiomer of the product forms
from the reaction catalyzed by the intermediates containing the
(Sa,Rc) diastereomer of L6.
Figure 5. Relative rates of cyclometalation of [Ir(COD)Cl((Ra,Rc)-L6)]
and [Ir(COD)Cl((Sa,Rc)-L6)] in the presence of propylamine at room
temperature monitored by 31P NMR spectroscopy.
These data also suggest that a single metalacycle core would
be formed by cyclometalation of {Ir(COD)Cl[(raca,Rc)-(L6)]}.
The faster rate of cyclometalation of the (Ra,Rc) ligand makes
it seem likely that the (Ra,Rc) ligand of the (raca,Rc)-L6 mixture
would preferentially undergo cyclometalation. Indeed, heating
the mixture of (raca,Rc)-L6 and [Ir(COD)Cl]2 in a 4:1 ratio of
phosphoramidite to iridium dimer with 30 equiv of propylamine
at 50 °C for 30 min led to a 1.9:1 ratio of two diastereomers
[(COD)Ir{κ2-(Ra,Rc)-L6}(κ1-(Ra,Rc)-L6)] and [(COD)Ir{κ2-
(Ra,Rc)-L6}(κ1-(Sa,Rc)-L6)] in which the core of the metalacycle
generated from the (Ra,Rc)-L6 ligand and the κ1 ligands is
(Ra,Rc)-L6 and (Sa,Rc)-L6, respectively.
This mixture of diastereomers was observed in the 31P NMR
spectrum as two pairs of doublets, and the identity of these
diastereomers was confirmed by independent synthesis. [(COD)-
Ir{κ2-(Ra,Rc)-L6}(κ1-(Ra,Rc)-L6)] was prepared from the reaction
of (Ra,Rc)-L6 with [Ir(COD)Cl]2 (vide supra). [(COD)Ir{κ2-
(Ra,Rc)-L6}(κ1-(Sa,Rc)-L6)] was generated independently as a
mixture with [(COD)Ir{κ2-(Ra,Rc)-L6}(κ1-(Ra,Rc)-L6)] by ad-
dition of 1 equiv of (Sa,Rc)-L6 to pure [(COD)Ir{κ2-(Ra,Rc)-
L6}(κ1-(Ra,Rc)-L6)]. The assignments of the cyclometalated
products were confirmed by reaction of PPh3 with the mixture
of cyclometalated complexes generated from (raca,Rc)-L6
(Scheme 1). This addition led to the formation of PPh3-ligated
(Ra,Rc)-2, as determined by 31P NMR spectroscopy.
The mixture of diastereomers was formed in about 40% yield.
This yield is about one-half of the yield of [(COD)Ir{κ2-(Ra,Rc)-
L6}(κ1-(Ra,Rc)-L6)] after heating [Ir(COD)Cl]2 with (Ra,Rc)-
L6 and 30 equiv of propylamine for 30 min. This lower yield
for formation of the cyclometalated species from (raca,Rc)-L6
is consistent with the slightly slower rate of reaction when the
catalyst is generated from (raca,Rc)-L6 instead of (Ra,Rc)-L6.40
5. A Highly Active and Selective Catalyst from a Single
Optically Active Component. These studies of diastereomeric
ligands led to information on the mode of catalyst activation,
but they did not reveal information about the relative reactivity
of the two diastereomers within the catalytic cycle or information
on whether the phenethyl group or the binaphthyl group was
the dominant component that controlled enantioselectivity. To
allow both diastereomers to be generated in the catalytic cycle
90%. This value is close to, but slightly lower than, the 96% ee
obtained from the reaction initiated with the pure (Ra,Rc)-L6
ligand.
b. Relative Rates of Cyclometalation of the Two Diaster-
eomeric Ligands. Studies of the rate and yield of cyclometa-
lation of the square planar Ir(I)-complexes [Ir(COD)(Cl)(L6)]
derived from (Ra,Rc)-L6, (Sa,Rc)-L6, and (raca,Rc)-L6 were
conducted. These studies revealed the origin of (1) the difference
in rates between the reactions catalyzed by complexes containing
diastereomeric ligands and (2) the difference in rates between
reactions catalyzed by complexes generated from the diaste-
reomerically pure (Ra,Rc)-L6 and from the mixture of diaster-
eomers. The cyclometalation of the two diastereomers of
[Ir(COD)(Cl)(L6)] in the presence of 30 equiv of propylamine
(eq 5) was monitored by 31P NMR spectroscopy. As illustrated
in Figure 5, {Ir(COD)Cl[(Ra,Rc)-L6]} converted at room tem-
perature in less than 2 h to the cyclometalated, trigonal
biypyramidal Ir(I) compound 2. In contrast, addition of an excess
of propylamine to {Ir(COD)Cl[(Sa,Rc)-L6)]} led to little conver-
sion of the starting Ir(I) chloride. Heating of {Ir(COD)Cl[(Sa,Rc)-
L6)]} at 50 °C for 1 h led to an increased conversion of this
starting complex, but several products were formed instead of
the single product from heating of {Ir(COD)Cl[(Ra,Rc)-L6]}.
This information led us to test the relative rates for cyclo-
metalation of [Ir(COD)(L1)Cl] containing the (Ra,Rc,Rc) and
(Sa,Rc,Rc) forms of the original ligand L1. [Ir(COD)Cl]2 and
(Ra,Rc,Rc)-L1 (2:1 ratio of ligand to iridium) and 30 equiv of
propylamine reacted over the course of 1.25 h at room
temperature to form the cyclometalated complex 1 (Figure 1)
reported previously.1 After this short time at room temperature,
only 25% of the square planar complex [Ir(COD)((Ra,Rc,Rc)-
L1)Cl] remained. In contrast, [Ir(COD)Cl]2 and the diastereo-
meric ligand (Sa,Rc,Rc)-L1 did not form the corresponding
cyclometalated complex at room temperature; only the square
planar complex {Ir(COD)Cl[(Sa,Ra,Rc)-L1]} and free ligand was
observed after standing with 30 equiv of propylamine at room
temperature for 1.25 h.
(40) Because the catalysts with (Ra,Rc)-L6 and (Sa,Rc)-L6 as the κ1 ligand are
diastereomeric, the rates of reactions catalyzed by the two species will be
different. Although we have not isolated pure complex, the complex with
(Sa,Rc)-L6 as the κ1 ligand, a 2:1 ratio of complexes in favor of the species
with (Ra,Rc)-L6 is formed, and this greater stability of the compound with
(Ra,Rc)-L6 as the κ1 ligand implies that this diastereomer will react slightly
more slowly. This effect would make the catalyst from (raca,Rc)-L6 more
reactive instead of less reactive, as we observe experimentally. Thus, we
attribute the slightly slower reaction of the catalyst generated from (raca,Rc)-
L6 to the lower yield for formation of the cyclometalated complex from
(raca,Rc)-L6.
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15511

A R T I C L E S
Scheme 1
Leitner et al.
with a complex that is related in structure to that derived from
ligand L6, we studied reactions with catalysts generated from
L17 in Figure 6, which is the biphenolate analogue of binaph-
tholate ligand L6. Because ligand L17 can undergo atropisom-
erism, the catalysts with these ligands would be able to access
both diastereomers during the catalytic cycle. In addition to
providing information on the reactivity of diastereomeric
complexes within the catalytic cycle, an enantioselective process
with this ligand would be practical because a single, inexpensive
phenethylamine would be the only optically active reagent used
to prepare the entire catalyst.
Like the phosphoramidites with binaphtholate groups, phos-
phoramidite L17 with a biphenolate group was prepared by
treatment of N-cyclododecylphenethylamine and triethylamine
with PCl3, followed by 2,2′-biphenol. The enantiopure ligand
was obtained in 60% yield by this trivial two-step synthesis
starting with phenethylamine.
a. Reactions Catalyzed by Complexes of Biphenolate
Ligand L17. The reactions of primary and secondary amines
with terminal allylic carbonates catalyzed by the complex
generated from this truncated ligand occurred with rates, yields,
regioselectivities, and enantioselectivities that are nearly identical
to those of the ligand L6, which contains the resolved binaph-
tholate substituent. They were even comparable to those obtained
with the ligand L1 that contains a resolved binaphtholate
substituent and two phenethyl substituents on the nitrogen. A
series of reactions of primary and secondary amines with
aliphatic and cinnamyl carbonates catalyzed by the complex
generated from biphenolate ligand L17 are summarized in Table
3. Like the reactions with the complexes generated from the
other phosphoramidites, we conducted these reactions after
inducing cyclometalation by heating [Ir(COD)Cl]2 with L17 and
propylamine at 50 °C for 30 min.
with benzylic heteroarylmethylamines, acyclic dialkylamines,
and cyclic aliphatic amines all occurred in good yields and with
high regio- and enantioselectivity. Reactions of aromatic amines
occurred in similar yields and selectivities. Although we have
not explored a wide range of substrate combinations at this time,
the reactions of aromatic and benzylic amines with the simple
aliphatic methyl carbonate derived from trans-2-hexene-1-ol
occur in a fashion similar to those of methyl cinnamyl carbonate.
Like reactions of acyclic secondary aliphatic amines with this
aliphatic carbonate catalyzed by complexes of other phosphor-
amidites, the reaction of diethylamine with this allylic carbonate
did not occur in high yield or with high regioselectivities when
conducted with ligand L17.
The observation of high branched-to-linear ratios and high
enantiomeric excess with such a simple ligand is remarkable.
However, the biphenolate unit presumably provides a stereo-
chemical environment at the metal that is similar to that provided
by the binaphtholate group in the original ligand, and the large
cycloalkyl group presumably plays a role in diverting attack
from a trajectory that intersects the space occupied by the distal
group. Studies to reveal the interplay between the stereochem-
istry of the biphenolate group and the stereochemistry of the
phenethyl group are described in sections 5b and 6.
b. Dynamics of the Complexes Generated by Coordination
and by Cyclometalation of Biphenolate Ligand L17. Variable-
temperature 31P NMR spectra of free L17, [Ir(COD)Cl(L17)],
the cyclometalated complex analogous to 1 that results from
heating this square planar complex with propylamine, and
phosphine derivatives of the cyclometalated complex were
obtained. These data are provided in the Supporting Information
as Figures S1-S3. These data allowed us to relate the dynamics
of stereoisomerism to the rate of the catalytic reaction.
A single resonance in the 31P NMR spectrum of free
phosphoramidite L17 was observed at room temperature, but
two broad resonances, which presumably correspond to the two
atropisomers of L17, were observed in a 1:0.7 ratio at -80 °C.
A single resonance in the 31P NMR spectrum of [Ir(COD)Cl-
(L17)] was also observed at room temperature. This signal
broadened at 0 °C, and two resolved singlets were observed at
-40 °C in a 1:0.2 ratio. These data suggest that atropisomerism
of the free and coordinated ligand is fast on the NMR time scale
at room temperature and that the difference in energy between
the two atropisomers is small.
In general, most reactions occurred to completion at room
temperature between 3 and 12 h and with enantioselectivities
between 91% and 97%. The branched-to-linear ratios were
typically over 20:1. Reactions of methyl cinnamyl carbonate
The 31P NMR spectrum of the cyclometalated complex
resulting from reaction of [Ir(COD)Cl]2 with L17 and propyl-
amine (Scheme 2) consisted of broad resonances at room
Figure 6. Ligand L17 with a biphenol backbone, an achiral N-alkyl group,
and one phenethyl group from resolved phenethylamine.
9
15512 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005

Iridium Catalyst with a Single Resolved Stereocenter
A R T I C L E S
Table 3. Amination of Allylic Carbonates Catalyzed by Complexes Generated from Biphenolate Ligand L17a
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/amine/catalyst of
100:120:2. b Isolated yields from two independent runs.
149.7 and -57.0 ppm in the 31P NMR spectrum at room
Scheme 2
temperature. At -40 °C, the two doublets broadened and were
unresolved. At -70 °C, the signals split into two sets of broad
resonances, and at -90 °C, two sets of broad resolved doublet
resonances were observed in a 1:0.2 ratio. These spectral
changes are again consistent with an atropisomerism that occurs
on the NMR time scale at low temperature and that is rapid at
room temperature. These data are also consistent with a
relatively small difference in population of the two atropisomeric
conformations of the cyclometalated ligand.
This body of spectroscopic data implies that the atropisom-
erism is fast enough that the cyclometalated catalyst formed
from biphenolate ligand L17 samples both atropisomeric
conformations within the catalytic cycle.41 Moreover, the lack
of large difference in the ratio of atropisomers of the free ligand,
of the square planar complex, and of the cyclometalated complex
implies that the ratio of atropisomers within the catalytic cycle
temperature. At -40 °C, these resonances sharpened, and two
predominant sets of doublets were observed, along with at least
is much smaller than the ratio of enantiomers formed by the
catalytic process. Therefore, the high enantioselectivity of the
two additional sets of doublets of lesser intensity. The multiple
catalyst with the biphenolate group on phosphorus does not
sets of resonances of the cyclometalated complex could cor-
appear to result from a predominant conformation of the
respond to those of diastereomeric complexes resulting from
biphenolate group that is created by the benzylic stereocenter
atropisomeric conformations of the biphenolate groups in the
in the metalacycle. Instead, the catalyst generated from L17
κ1 and κ2 phosphoramidites or from two different geometries
reacts under classic Curtin-Hammett conditions: the difference
at the metal center (see Scheme 2). This mixture of isomers
in energy between the conformers is smaller than the difference
was too complicated for further characterization.
Thus, we replaced the κ1-phosphoramidite in the cyclometa-
lated complex with the achiral phosphine PMe3 (Scheme 2) to
simplify the stereochemistry of the cyclometalated complex. The
in energy between the two transition states, and the conformer
that reacts with the larger rate constant leads to the major
enantiomer.
(41) Because the catalytic process appears to occur after dissociation of the κ1
ligand to generate a four-coordinate iridium complex, any stereochemistry
at the metal in the five-coordinate complex would be lost upon dissociation
and should not affect the enantioselectivity of the catalytic process.
PMe3 complex was isolated in 60% yield by adding the
phosphine after the cyclometalation step. The product from this
ligand substitution displayed a single set of sharp doublets at
9
J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005 15513

A R T I C L E S
Leitner et al.
6. The Relationship between the Structures of Catalysts
Containing the Simple Ligand L17 and Those Containing
the More Elaborate Phosphoramidites. The information on
the various diastereomeric catalysts allows one to draw several
conclusions about the structures of the catalysts derived from
the simple ligand L17 and to deduce which is the more reactive
atropisomer within the catalytic cycle. As mentioned previously
in this paper, we have published that the catalysts derived from
the (Ra,Rc,Rc) and (Sa,Rc,Rc) diastereomers of ligand L1 generate
enantiomeric products. Consistent with this assertion, reactions
catalyzed by complexes of ligands (Ra,Rc)-L6 and (Sa,Rc)-L6
generate enantiomeric products, although the yield of product
from reactions catalyzed by the (Sa,Rc)-L6 was low. Thus, the
absolute configuration, not just the ee, appears to be affected
by the axial chirality of the binaphtholate group. Because the
rates and selectivities of biphenolate ligand L17 are similar to
those of binaphtholate ligands L1 and L6, and because the major
enantiomer of the product from reaction of the catalyst contain-
ing (R)-L17 is the same as that from the catalyst derived from
either (Ra,Rc,Rc)-L1 or (Ra,Rc)-L6, we surmise that the diaster-
eomeric conformation of the catalyst containing ligand L17 that
gives rise to the major enantiomer also has an Ra configuration
of the biphenolate group. Our data do not show whether the Ra
diastereomer is the more stable isomer in solution, but the
assertion that the product is derived from this atropisomer, in
combination with the conclusion that the ratio of populations
of the two atropisomers in the observed complexes is smaller
than the ratio of enantiomeric products of the catalytic reaction,
makes it likely that the diastereomer with an Ra configuration
of the biphenolate group is more reactive than that with an Sa
configuration within the catalytic cycle.
reactions catalyzed by complexes containing a ligand L17
possessing a biphenolate group results from reaction of the
(Ra,Rc) atropisomer. The final ligand L17 presented in this work
is prepared by a short synthesis using trivial reactions with
inexpensive bulk chemicals as reagents, and this catalyst behaves
in a fashion similar to the more complex catalyst with three
resolved stereochemical elements studied originally.
Several principles emerge from this work. (1) Most generally,
we show that the identification of the structure of an active
catalyst can be integral to the rational selection or design of
ligands, even for enantioselective catalysis. (2) We show that
bidentate ligands can be generated from a simple monodentate
ligand by cyclometalation and that this sequence can provide a
simple synthesis of a catalyst for enantioselective reactions. (3)
We show that cyclometalation can be a pathway to catalyst
activation, rather than catalyst deactivation. (4) We show a
means by which a highly selective catalyst can be generated
from a ligand containing a single resolved stereocenter. It is
interesting to consider that the manner in which this single
phenethyl stereocenter affects the population and reactivities
of an atropisomeric biphenolate group is reciprocal to the manner
in which axial chirality in commonly used ligands, such as
BINAP, influences the gearing of aryl substituents on a
phosphine.42-47
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. Dr.
Klaus Ditrich (BASF AG, Ludwigshafen) is acknowledged for
a generous gift of chiral amines.
Supporting Information Available: Full experimental section,
including figures of variable-temperature NMR data. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Summary
We have developed a catalyst that possesses stereochemistry
controlled by a single phenethyl group for enantioselective
allylic amination of linear allylic carbonates to form branched
amines with high regioselectivity and enantioselectivity. This
ligand resulted from a design process that used as its foundation
a cyclometalated structure of the active catalyst. In addition,
studies of diastereomeric catalysts showed that the origin of
the difference in reactivity of diastereomeric catalysts is due,
in part, to different rates of catalyst activation. Finally, these
studies show that the major enantiomer of the product from
JA054331T
(42) For representative previous studies on amplification of a chiral environment,
see the work and reviews of Reetz and Walsh in the following references.
(43) Reetz, M. T.; Neugebauer, T. Angew. Chem., Int. Ed. 1999, 38, 179.
(44) Reetz, M. T.; Becker, M. H.; Klein, H.-W.; Stockigt, D. Angew. Chem.,
Int. Ed. 1999, 38, 1758.
(45) Costa, A. M.; Jimeno, C.; Gavenonis, J.; Carroll, P. J.; Walsh, P. J. J. Am.
Chem. Soc. 2002, 124, 6929.
(46) Walsh, P. J.; Lurain, A. E.; Balsells, J. Chem. ReV. 2003, 103, 3297.
(47) Walsh, P. J. Acc. Chem. Res. 2003, 36, 739.
9
15514 J. AM. CHEM. SOC. VOL. 127, NO. 44, 2005