Palladium-catalyzed asymmetric allylic amination using ferrocenyl pyrazole ligands: Steric control of η3-allyl configuration and site-selective nucleophilic attack

Article abstract of DOI:10.1021/ja953031g

The reaction of 1,3-diphenylallylethyl carbonate (1) with benzylamine to afford the secondary amine 2 is effectively catalyzed by Pd-complexes containing chiral ferrocenyl pyrazole ligands. The highest enantioselectivity (99% ee) was obtained using ligand

Full text of DOI:10.1021/ja953031g

J. Am. Chem. Soc. 1996, 118, 1031-1037
1031
Palladium-Catalyzed Asymmetric Allylic Amination Using
3
Ferrocenyl Pyrazole Ligands: Steric Control of η -Allyl
Configuration and Site-Selective Nucleophilic Attack
,†
†
‡
†
Antonio Togni,* Urs Burckhardt, Volker Gramlich, Paul. S. Pregosin, and
Renzo Salzmann†
Contribution from the Laboratory of Inorganic Chemistry and Institute of Crystallography and
Petrography, Swiss Federal Institute of Technology, ETH-Zentrum, CH-8092 Z u¨ rich, Switzerland
X
ReceiVed September 5, 1995
Abstract: The reaction of 1,3-diphenylallylethyl carbonate (1) with benzylamine to afford the secondary amine 2 is
effectively catalyzed by Pd-complexes containing chiral ferrocenyl pyrazole ligands. The highest enantioselectivity
(99% ee) was obtained using ligand 3a, 1-{(S)-1-[(R)-2-(diphenylphosphino)ferrocenyl]ethyl}-3-(1-adamantyl)-1H-
pyrazole. Four different cationic Pd-allyl intermediates, 4a, 4c, 4j, and 4k (containing ligands 3a, 3c, 3j, and 3k,
respectively), formed during the catalytic reaction were studied in solution by 2D-NMR spectroscopy, with the aim
of clarifying configurational aspects. Depending on the size and shape of the substituent in position 3 of the pyrazole
ring, it was found that the major diastereoisomeric form of these complexes either adopts an exo-syn-syn (ligands 3a
and 3c, 1-{(S)-1-[(R)-2-(diphenylphosphino)ferrocenyl]-ethyl}-3-phenyl-5-methyl-1H-pyrazole) or an exo-syn-anti
configuration (ligands 3j, 1-{(S)-1-[(R)-2-(diphenylphosphino)ferrocenyl]ethyl}-3-(9-anthryl)-5-methyl-1H-pyrazole
and 3k, 1-{(S)-1-[(R)-2-(diphenylphosphino)ferrocenyl]ethyl}-3-(9-triptycyl)-1H-pyrazole). This analysis allows the
site of nucleophilic attack on the allyl ligand to be established unequivocally, i.e., the carbon atom trans to phosphorus.
3
The reasons responsible for this pronounced site-selectivity are discussed. The Pd-allyl complexes [Pd(η -
3
PhCHCHCHPh)(3c)]PF6, 4c, and [Pd(η -PhCHCHCHPh)(3k)]PF6, 4k, were characterized by X-ray diffraction. 4c
crystallizes with 1 equiv of Et2O in the orthorhombic space group P212121: a ) 13.026(2) Å, b ) 14.784(2) Å, c
)
25.124(5) Å, Z ) 4. Crystals of 4k contain 3 equiv of hexane and 1 equiv of acetone and H2O in the unit cell
belonging to the trigonal system, space group P3221: a ) 24.07(3) Å, c ) 22.48(3) Å, Z ) 6.
Introduction
Chart 1
Enantioselective palladium-catalyzed substitution reactions of
1
allylic substrates have received a new impetus in recent years
thanks to the development of novel chiral chelating ligands.
2
Thus, nitrogen-containing auxiliaries, in particular, have been
shown to belong to the most effective chiral inducing agents
for reactions of allylic acetates or carbonates with soft carbon
3
nucleophiles. Assuming that the chiral ligands lack any
element of symmetry, plus symmetrically substituted allylic
substrates, the enantioselectivity observed will depend upon
three main factors: (1) the relative concentrations of the different
generally agreed that the nucleophilic attack takes place
preferentially at the allyl terminus trans to phosphorus.4
However, uncertainty still exists concerning the profound
3
possible configurational isomers of the Pd(η -allyl) complexes
5
reasons leading to such a site-selectivity. Despite its inherent
present as intermediates (see Chart 1), (2) their relative rates of
reaction with the nucleophile, and (3) the site of nucleophilic
attack at the inequivalent allylic positions 1 and 3.
(3) For recent pertinent examples, see: (a) Leutenegger, U.; Umbricht,
G.; Fahrni, C.; von Matt, P.; Pfaltz, A. Tetrahedron 1992, 48, 2143-2156.
(
(
b) von Matt, P.; Pfaltz, A. Angew. Chem., Int. Ed. Engl. 1993, 32, 566
Angew. Chem. 1993, 105, 614). (c) von Matt, P.; Loiseleur, O.; Koch, G.;
Concomitant with the successful development of new ligands,
affording enantioselectivities as high as 99% ee for the reaction
of 1,3-diphenylallyl acetate with dimethyl malonate, studies
addressing the subtle aspects concerning the origin of enan-
tioselectivity have been reported recently. For the P,N-ligand
Pfaltz, A.; Lefeber, C.; Feucht, T.; Helmchen, G. Tetrahedron: Asymmetry
1994, 5, 573. (d) Sprinz, J.; Helmchen, G. Tetrahedron Lett. 1993, 34,
1769-1772. (e) Sprinz, J.; Kiefer, M.; Helmchen, G.; Reggelin, M.; Huttner,
G.; Walter, O.; Zsolnai, L. Tetrahedron Lett. 1994, 35, 1523-1526. (f)
Frost, C. G.; Williams, J. M. J. Tetrahedron Lett. 1993, 34, 2015-2018.
(g) Dawson, G. J.; Frost, C. G.; Williams, J. M. J.; Coote, S. J. Tetrahedron
Lett. 1993, 34, 7793-7796. (h) Frost, C. G.; Williams, J. M. J. Tetrahe-
dron: Asymmetry 1993, 4, 1785-1788. (i) Bovens, M.; Togni, A.; Venanzi,
L. M. J. Organomet. Chem. 1993, 451, C28-C31. (j) Kubota, H.; Nakajima,
M.; Koga, K. Tetrahedron Lett. 1993, 34, 8135-8138. (k) Brown, J. M.;
Hulmes, D. I.; Guiry, P. J. Tetrahedron 1994, 50, 4493-4506. (l) Brenchley,
G.; Merifield, E.; Wills, M.; Fedouloff, M. Tetrahedron Lett. 1994, 35,
2791-2794. (m) Tanner, D.; Andersson, P. G.; Harden, A.; Somfai, P.
Tetrahedron Lett. 1994, 35, 4631-4634. (n) Gamez, P.; Dunjic, B.; Fache,
M.; Lemaire, M. J. Chem. Soc. Chem. Commun. 1994, 1417-1418. (o)
Kang, J.; Cho, W. O.; Cho, H. G. Tetrahedron: Asymmetry 1994, 5, 1347-
1352. (p) von Matt, P.; Lloyd-Jones, G. C.; Minidis, A. B. E.; Pfaltz, A.;
Macko, L.; Neuburger, M.; Zehnder, M.; R u¨ egger, H.; Pregosin, P. S. HelV.
Chim. Acta 1995, 78, 265-284.
systems developed by Pfaltz and Helmchen3 and Brown it is
e
3k
†
Laboratory of Inorganic Chemistry.
Institute of Crystallography and Petrography.
Abstract published in AdVance ACS Abstracts, January 1, 1996.
‡
X
(
1) For reviews, see: (a) Consiglio, G.; Waymouth, R. M. Chem. ReV.
1
989, 89, 257-276. (b) Hayashi, T. In Catalytic Asymmetric Synthesis;
Ojima, I., Ed.; VCH: New York, 1993; pp 325-365. (c) Godleski, S. A.
In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming, I., Eds.;
Pergamon: Oxford, 1991; Vol. 4, Chapter 3.3, pp 585-661. (d) Frost, C.
G.; Howarth, J.; Williams, J. M. J. Tetrahedron: Asymmetry 1992, 3, 1089-
1
122.
2) For a recent review, see: Togni, A.; Venanzi, L. M. Angew. Chem.,
Int. Ed. Engl. 1994, 33, 497-526 (Angew. Chem. 1994, 106, 517-548).
(
0
002-7863/96/1518-1031$12.00/0 © 1996 American Chemical Society

1
032 J. Am. Chem. Soc., Vol. 118, No. 5, 1996
Togni et al.
Scheme 1
Table 1. Enantioselectivities Observed for the Reaction of Scheme
1
as a Function of the Ligand Used
liganda
R1
R2
% ee (abs. conf.)
b
(S)-(R)-3a
(S)-(R)-3b
(S)-(R)-3c
(S)-(R)-3d
(S)-(R)-3e
(S)-(R)-3f
(S)-(R)-3g
(S)-(R)-3h
(S)-(R)-3i
(S)-(R)-3j
(S)-(R)-3k
(S)-(R)-3l
1-adamantyl
cyclohexyl
Ph
H
99 (R)
95 (R)
96 (R)
95 (R)
94 (R)
94 (R)
96 (R)
97 (R)
95 (R)
40 (S)
Me
Me
H
Me
Me
Me
H
Me
Me
H
Ph
1-naphthyl
2-naphthyl
4-pyridinyl
2
3-NO -Ph
2,4(MeO)
9-anthryl
2
Ph
9-triptycyl
Me
no reaction
90 (R)
Me
interest, we note that less attention has been devoted to the
analogous transformations involving amines as the nucleophilic
component.6
a
Control reactions were also performed with the (R)-(S)-enantiomers
of the ligands. b Determined by HPLC (see Experimental Section).
9
We recently reported the successful application of pyrazole-
12 h, the amination product (R)-2 was isolated after chromato-
graphic purification in 90-95% yield. A brief screening of the
7
containing ferrocenyl phosphines (P,N-ligands) of type 3 in
7b
the Rh-catalyzed hydroboration of styrene with catecholborane
affording enantioselectivities up to 98% ee. These encouraging
results prompted us to extend the scope of these ligands to the
model reaction of racemic 1,3-diphenylallylethyl carbonate (1)
with benzylamine, reported here and illustrated in Scheme 1.
The focus of our study is thereby more mechanistic than
synthetic in nature and on these lines we present evidence that
the nucleophilic attack occurs, not unexpectedly, preferentially
at the allylic terminus trans to the phosphine moiety. Further-
more, and more importantly, we show that the configuration of
available pyrazole ligands 3 indicated that a sterically relatively
large substituent is needed at position 3 of the heterocycle, in
order to obtain enantioselectivities in the range of 94-99 %
ee. A collection of selectivity data as a function of the ligand
used is given in Table 1. Thus, several aryl substituents are
well tolerated as well as the relatively bulky cyclohexyl group
(ligands 3b-i). The very rigid, compact, and large substituent
1-adamantyl (ligand 3a) was found to afford the unprecedented
value of 99% ee for this reaction. Ligand 3l containing the
less bulky methyl group at position 3 gives a significantly lower
stereoselectivity (90% ee).
3
the Pd(η -allyl) intermediate can effectively be controlled by
the steric nature of the ligand, this aspect providing an
unprecedented mechanistic tool.
Much to our surprise, the reaction involving the analogous
ligand (S)-(R)-3j containing a 9-anthryl substituent gave the (S)-2
product in only 40% ee under the same conditions. This drastic
change (inversion) of enantioselectivity can only be due to a
major steric influence of the new substituent on the configuration
Results and Discussion
Reactions of carbonate 1 with benzylamine were carried out
in THF at 40 °C in the presence of 1.5 mol % of [Pd2(dba)3]‚
CHCl3 as a catalyst precursor and 4.5 mol % of the correspond-
3
of the Pd(η -allyl) intermediate and/or on the site of nucleophilic
attack, since the absolute configuration of the ligand is constant,
10
and no significantly different electronic effects are anticipated.
8
ing pyrazole ligand (S)-(R)-3. After a typical reaction time of
In order to understand this unexpected observation, we decided
3
to study the structure of the corresponding cationic Pd(η -1,3-
(4) The reported studies base their conclusion upon the analysis of the
diphenylallyl) complexes (1) in the solid state by X-ray
structures both in solution and in the solid state of Pd-allyl intermediates.
3
11
They correlate the configuration of the major isomer of the η -allyl
diffraction and (2) in solution by 2D NMR spectroscopy. This
intermediate with the absolute configuration of the product. This is a typical
ground-state argument, as it has been previously put forward by Bosnich;
see: (a) Auburn, P. R.; Mackenzie, P. B.; Bosnich, B. J. Am. Chem. Soc.
kind of intermediate in the catalytic reaction is easily accessible
3
in crystalline form from the dinuclear compound [Pd2(η -1,3-
6
a
1
985, 107, 2033-2046. (b) Mackenzie, P. B.; Whelan, J.; Bosnich, B. J.
diphenylallyl)2Cl2], the chelating ligand, and a chloride
scavenging agent, such as TlPF6.
Am. Chem. Soc. 1985, 107, 2046-2054. An alternative way of directing
the incoming nucleophile by means of secondary interactions with a
functional group contained in the ligands has been developed by Hayashi
and Ito. For a review, see: Sawamura, M.; Ito, Y. Chem. ReV. 1992, 92,
Thus, the crystal and molecular structure of the compound
3
[Pd(η -1,3-diphenylallyl)((S)-(R)-3c)]PF6 (4c) clearly shows the
8
57-871, and references cited therein.
allyl ligand in an exo-syn-syn configuration (exo refers to the
relative orientation of the central allylic C-H vector pointing
away from the ferrocene core), as it is apparent from the ORTEP
view of Figure 1. Table 2 collects a selection of bond lengths
and angles. The structure turns out to be rather routine, with a
pseudo square planar geometry around palladium, and all
3
e
(
5) Helmchen pointed out the importance of an early transition state,
3
k
whereas Brown indicated the significance of repulsive steric interactions
in a late transition state.
(6) For successful approaches utilizing 1,3-disubstituted acyclic allylic
substrates, see: (a) Hayashi, T.; Yamamoto, A.; Ito, Y.; Nishioka, E.; Miura,
H.; Yanagi, K. J. Am. Chem. Soc. 1989, 111, 6301-6311. (b) Hayashi, T.;
Kishi, K.; Yamamoto, A.; Ito, Y. Tetrahedron Lett. 1990, 31, 1743-1746.
12
(
1
c) Hayashi, T.; Yamamoto, A.; Ito, Y. Tetrahedron Lett. 1988, 29, 99-
bonding parameters falling in the expected ranges. However,
02. (d) von Matt, P.; Loiseleur, O.; Koch, G.; Pfaltz, A.; Lefeber, C.;
Feucht, T.; Helmchen, G. Tetrahedron: Asymmetry 1994, 5, 573-584.
Furthermore, Trost and co-workers developed the amination of cyclic allylic
substrates, based on the enantiotopic group differentiation. See: (e) Trost,
B. M.; Van Vranken, D. L.; Bingel, C. Angew. Chem., Int. Ed. Engl. 1992,
(9) Assignment of the absolute configuration has been made by compar-
ing the sign of the optical rotation with that reported in the literature (ref
6a).
(10) For the observation of important electronic effects on stereoselec-
tivity using this kind of ligands, see ref 7a.
(11) A detailed account on the characterization and comparison of the
structural features in solution, as obtained by 2D-NMR methods, is in
preparation.
3
1, 228-229 (Angew. Chem. 1992, 104, 194-196). (f) Trost, B. M.; Van
Vranken, D. L.; Bingel, C. J. Am. Chem. Soc. 1992, 114, 9327-9343. (g)
Trost, B. M.; Van Vranken, D. L. J. Am. Chem. Soc. 1993, 115, 444-458.
(7) (a) Schnyder, A.; Hintermann, L.; Togni, A. Angew. Chem., Int. Ed.
Engl. 1995, 34, 931-933 (Angew. Chem. 1995, 107, 996-998). (b)
(12) (a) Orpen, A. G.; Brammer, L.; Allen, F. H.; Kennard, O.; Watson,
D. G.; Taylor, R. J. Chem. Soc., Dalton Trans. 1989, Supplement S1-
S83. (b) Allen, F. H.; Kennard, O.; Watson, D. G.; Brammer, L.; Orpen,
A. G.; Taylor, R. J. Chem. Soc., Perkin Trans II 1987, Supplement S1-
S19.
Burckhardt, U.; Hintermann, L.; Schnyder, A.; Togni, A. Organometallics
1
995, 14, 5415-5425.
8) These reaction conditions correspond to those reported by Hayashi
and co-workers, see ref 6a.
(

Palladium-Catalyzed Asymmetric Allylic Amination
J. Am. Chem. Soc., Vol. 118, No. 5, 1996 1033
Figure 1. ORTEP view (30% ellipsoids) and atom numbering scheme
for the cation (S)-(R)-4c.
Figure 2. Projections of the allyl ligands of compounds (S)-(R)-4c
(a)) and (R)-(S)-4k (b)) along the Pd-P-N(2) plane.
a
a
Table 2. Selected Bond Distances (Å) and Angles (deg) for
S)-(R)-4c and (R)-(S)-4k
(
the angles formed by the Pd-P-N(2) plane and the planes
defined by Pd, C(36), and C(37) and Pd, C(37), and C(38) are
(
S)-(R)-4c
(R)-(S)-4k
11.8° and 46.5°, respectively. This means that the unit C(37)-
Bond Distances
2.321(4) Pd-P
2.131(11) Pd-N(2)
2.138(16) Pd-C(43)
2.173(16) Pd-C(44)
2.268(13) Pd-C(45)
C(36) (trans to nitrogen) is already almost coplanar with the
atoms Pd, P, and N(2), as schematically illustrated in Figure
Pd-P
2.344(7)
Pd-N(2)
Pd-C(36)
Pd-C(37)
Pd-C(38)
2.225(15)
2.161(18)
2.064(24)
2.213(19)
1.391(35)
1.369(31)
1.342(34)
2
a.
31P, 13
), and specifically a 2D H NOESY spectrum of complex 4c
1
C, and H spectra (selected data are presented in Table
1
14
3
C(36)-C(37)
C(37)-C(38)
N(1)-N(2)
1.436(22) C(43)-C(44)
1.380(23) C(44)-C(45)
1.359(17) N(1)-N(2)
(see Figure 3a) measured in THF-d8 at 40 °C, to reflect the
reaction medium used for the catalytic experiments, confirmed
that the major isomer (85%) observed in solution had the same
structure shown in the solid state. However, of the three
remaining isomers two could be identified, i.e., the endo-syn-
syn form (6%) and the endo-syn-anti form (6%). The latter is
Bond Angles
95.0(3)
94.5(5)
P-Pd-N(2)
N(2)-Pd-C(38)
P-Pd-C(36)
P-Pd-N(2)
101.3(5)
97.0(7)
97.3(5)
N(2)-Pd-C(45)
105.3(4) P-Pd-C(43)
1
5
exchanging with the major isomer, and has the anti phenyl
on the allyl carbon atom in pseudo trans position with respect
to the pyrazole moiety. The structure of the fourth isomer (3%)
could not be assigned unequivocally. The assignment of the
C(36)-C(37)-C(38) 121.0(16) C(43)-C(44)-C(45) 115.8(27)
C(30)-C(36)-C(37) 122.4(15) C(44)-C(43)-C(46) 124.3(23)
C(37)-C(38)-C(39) 126.0(19) C(44)-C(45)-C(52) 135.9(23)
a
Number in parentheses are esd’s in the least significant digits.
1
allyl configuration is based on (1) H spin-spin coupling
constants and classical intra-allyl NOEs (it is worth remembering
that a strong NOE between the central allylic proton and just
one of the two terminal protons indicates a syn-anti arrange-
ment), (2) an NOE between the central allyl proton and ortho
protons on the phenyl group attached to the pyrazole determines
the exo configuration, and (3) NOEs between the anti terminal
allyl proton and the ortho protons of the phenyl group on the
pyrazole. Note that the absence of such an NOE involving the
central allyl proton, indicates the endo orientation.
several characteristics are worth noting. Thus, the conformation
of the ligand qualitatively corresponds to that found for free
7b
ligands of type 3, with the pyrazole fragment in a pseudo axial
position. The phenyl group on the allyl carbon trans to nitrogen
intrudes into the region between the two phenyl rings on
phosphorus. This arrangement could rationalize the preference
for the exo orientation of the allyl ligand. The Pd-C bond
lengths are significantly different from one another, the carbon
atom trans to phosphorus displaying the expected longer distance
Under the assumption that the observed major enantiomer
of the product (absolute configuration R) of the catalytic reaction
reflects preferential nucleophilic attack on the major exo-syn-
syn isomer, then a correlation between the observed absolute
configuration and the site of nucleophilic attack is possible. Such
a correlation indicates preferential attack at the allylic carbon
atom trans to phosphorus. However, it is important to realize
(
Pd-C(38) ) 2.268(13) Å), as compared to its partner trans to
nitrogen (Pd-C(36) ) 2.138(16) Å). This is an expression of
the higher trans influence13 of the phosphine ligand. The plane
of the allyl ligand is tilted 32.6° from the perpendicular to the
Pd-N-P plane, and the latter forms an angle of 36.0° with the
“
upper” Cp ring. In view of the overall mechanistic discussion
vide infra), an important distorsion of the Pd-allyl fragment is
(
observed. The allyl ligand, with respect to its idealized
orientation, is rotated in an anti-clockwise manner around the
Pd-allyl axis, as seen from the π-ligand toward the Pd atom
(
14) For details concerning the application of NOESY spectroscopy to
the structural characterization of Pd-allyl complexes in solution, see, e.g.:
a) Pregosin, P. S.; Salzmann, R.; Togni, A. Organometallics 1995, 14,
(
8
42-847. (b) Abbenhuis, H. C. L.; Burckhardt, U.; Gramlich, V.; K o¨ llner,
C.; Pregosin, P. S.; Salzmann, R.; Togni, A. Organometallics 1995, 14,
759-766, and pertinent references cited therein.
(15) A section of the phase-sensitive NOESY spectrum showing this
particular selective exchange occuring on the NMR time scale is provided
as supporting information. A simple π-σ-π equilibrium could account
for this interconversion.
(for the (S)-(R) absolute configuration of the ferrocenyl moiety).
This is reflected, e.g., by two very different interplanar angles
that should be identical in the case of an ideal geometry. Indeed,
(13) For a review see, e.g.: Appleton, T. G.; Clark, H. C.; Manzer, L.
E. Coord. Chem. ReV. 1973, 10, 335-422.

1
034 J. Am. Chem. Soc., Vol. 118, No. 5, 1996
Togni et al.
Table 3. Selected 500 MHz H, 125.8 MHz 13C, and 202.6 MHz 31P NMR data (d
1
8
-THF) for compounds 4a, 4c, 4j, and 4k
compound
exo-syn-syn-4a
exo-syn-syn-4c
endo-syn-anti-4c
endo-syn-syn-4c
exo-syn-anti-4j
exo-syn-syn-4j
exo-syn-anti-4k
b
b
b
13.4b
12.0b
b
b
1
1.8
14.2
12.0
8.2
11.4
1
13
1
13
1
13
1
13
1
13
1
13
1
13
δ H (δ C)
7.26 (107.9)
6.49 (113.8)
5.34 (65.9)
2.37 (18.0)
7.08 (59.1)
4.07 (73.7)
4.46 (71.3)
4.72 (67.0)
6.00 (106.4)
7.85 (130.0)
3.95 (71.8)
δ H (δ C)
6.43 (104.4)
5.94 (111.5)
5.15 (67.9)
2.35 (18.5)
6.85 (60.4)
4.05 (73.7)
4.56 (72.0)
4.80 (67.1)
5.93 (109.3)
2.46 (13.0)
3.95 (71.4)
δ H (δ C)
5.57 (95.0)
6.54 (106.7)
6.67 (75.9)
1.84 (17.3)
6.32 (60.1)
δ H (δ C)
4.50 (93.0)
6.92 (108.7)
5.10 (79.5)
2.03 (17.6)
6.53 (60.3)
4.02a
δ H (δ C)
5.18 (102.4)
3.63 (106.5)
5.43 (71.3)
2.37 (17.4)
6.12 (59.8)
3.93 (74.6)
4.68 (71.6)
5.36 (70.4)
6.64 (113.5)
2.79 (12.0)
3.80 (71.5)
δ H (δ C)
7.40 (115.1)
5.18 (107.0)
4.64 (67.8)
2.52 (19.2)
7.49 (60.1)
4.28 (74.1)
4.68(70.8)
5.18 (67.4)
6.29 (115.6)
2.77 (12.3)
4.05 (72.1)
δ H (δ C)
5.41 (98.7)
3.98 (105.5)
4.81 (75.3)
2.45 (20.3)
6.54 (58.5)
4.34(73.6)
4.53 (70.6)
5.02 (67.6)
7.37 (113.9)
8.65 (131.2)
3.47 (71.0)
CH(1)c
CH(2)
CH(3)
CH(4)
CH(5)
CH(6)
CH(7)
CH(8)
CH(9)
a
3.99
a
a
4.59
4.53
a
4.77a
4.82
a
a
CH(10)
2.23 (12.6)
2.29 (12.9)
a
5
η -C
5
H
5
a
a
Resonance could not be unequivocally assigned. b δ31P. c Arbitrary atom numbering scheme:
complex, indicating that the different isomers as well as the
two different allyl termini should display different rates for
nucleophilic attack.
3
The complex [Pd(η -1,3-diphenylallyl)((S)-(R)-3a)]PF6 (4a),
containing the 1-adamantyl ligand 3a which affords the highest
enantioselectivity, exists in solution as a single diastereomeric
form, the exo-syn-syn. Assignment of this structure is based
on the observation of NMR characteristics similar to those
described above for exo-syn-syn-3c.
3
For the analogous complex [Pd(η -1,3-diphenylallyl)((S)-(R)-
j)]PF6 (4j), containing the 9-anthryl substituent, a mixture of
3
two isomers was observed in solution (see Chart 2). The
assignment of these two structures is again based on the
observation of several diagnostic NOEs, as detailed above for
derivative 4c (see Figure 3c). Thus, one of the two isomers
(29%) shows the exo-syn-syn configuration (syn-syn as shown
by NOEs between the terminal anti allyl protons and exo via
the NOE between the central proton and protons on the anthryl
group). However, the major component is now an exo-syn-
anti isomer (71%), the anti configurated allyl carbon being the
one cis to the N-moiety, based on intraligand NOEs between
two allyl protons and the anthryl group.
Nucleophilic attack at the carbon atom trans to phosphorus
should lead preferentially to the formation of the S-configurated
product during catalysis. This is indeed what we find, and this
observation thus corroborates the hypothesis that the allylic
terminus trans to the phosphine behaves as the more electrophilic
center. However, this observation on 4j, taken alone, cannot
conclusively exclude attack at the carbon trans to nitrogen which
would also lead to the S-product. Nevertheless, it is worth
noting that, in this case, the ratio of the two configurational
isomers reflects the observed ratio of enantiomers for the product
of the catalytic reaction within 1-2%. If the nucleophilic attack
indeed takes place in both cases only at the carbon atom trans
to phosphorus, this would imply a very similar reaction rate of
the two isomers. Finally, it is important to note that on the
NMR time scale, up to 60 °C, we did not observe any
1
Figure 3. Sections of the phase-sensitive H 2D NOESY spectra of
compounds 4c (a), 4a (b), 4j (c) (A, exo-syn-anti isomer; B, exo-syn-
syn isomer), and 4k (d). 1, 2, and 3 indicate the three allylic protons,
H being adjacent to nitrogen. The oblique arrows show the NOEs
between the allylic protons and the square boxes mark the absence of
cross-signals. These characteristic “absences” are also important for
distinguishing between syn and anti configurations. The horizontal
1
3
arrows show the NOE between anti proton H and assigned phosphine
ortho phenyl protons, indicating the exo form (see also text).
1
6
that the enantiomeric composition of the product (98:2) does
not reflect the isomeric distribution of the intermediate allyl
equilibrium between the two isomers.
With these results in hand, we decided to construct a ligand

Palladium-Catalyzed Asymmetric Allylic Amination
J. Am. Chem. Soc., Vol. 118, No. 5, 1996 1035
Chart 2
which would induce the formation of an exo-syn-anti configu-
rated isomer exclusively. Since the formation of this isomer
seemed to be favored by the presence of a very bulky substituent
at position 3 of the pyrazole, we reasoned that this could be
achieved by a group larger than a 9-anthryl. Selective stabiliza-
Figure 4. ORTEP view (20% ellipsoids) and atom numbering scheme
for the cation (R)-(S)-4k.
Chart 3
3
tion of the anti configuration in (η -allyl)palladium complexes
based on steric repulsion has been reported previously by
A° kermark, Vitagliano, and co-workers, using 2,9-disubstituted-
1
,10-phenanthroline ligands.17 In our case, and for the sake of
simplicity, the group exerting the necessary steric bulk should
also be as symmetrical as possible. This was achieved with
the 9-triptycyl substituted ligand 3k (see Chart 3). In fact in
the major component (90%) of the corresponding complex [Pd-
3
(
η -1,3-diphenylallyl)((R)-(S)-3k)]PF6 (4k), only one form, the
exo-syn-anti configuration, could be identified in solution by
NOESY spectroscopy (see Figure 3d). This is confirmed by
(1) the usual absence of NOEs between the terminal allyl protons
(
syn-anti structure, anti cis to N), (2) an NOE between the
central allyl proton and the terminal one trans to the phosphine,
and (3) NOEs between these two protons and protons on the
triptycyl fragment (exo form). The proximity between the allyl
ligand and the triptycyl group also seems to constitute a
sufficiently high barrier such that at room temperature no free
rotation of the latter is observed on the NMR time scale (all
protons on the triptycyl fragment are inequivalent).18 A minor
component (10%) observed in solution shows very broad signals
and could not be assigned a defined structure.19
of the allyl ligand is nicely imbedded into one of the three
cavities created by the propeller-shaped triptycyl moiety (see
Figure 4; for a selection of important bond distances and angles,
see Table 2). The general geometric features of the cation 4k
are very similar to those found for cation 4j (vide supra). Thus,
with respect to the rest of the molecule, the orientation of the
palladium coordination plane, as defined by the atoms Pd, P,
and N(2), forms angles of 108.6° and 34.5° with the allyl ligand
and with the “upper” Cp ring, respectively. The anti arrange-
ment of the phenyl group on C(45) is reflected by the observed
torsion angle C(43)-C(44)-C(45)-C(52) of 32°, as opposed
to a value of 170° for the partner angle C(45)-C(44)-C(43)-
C(46) (corresponding similar absolute values of 168° and 172°
were observed in complex 4c). The only crystallographically
characterized Pd-allyl complexes known in the literature con-
taining an anti oriented aryl substituent involve 1,1,3-triarylallyl
The X-ray crystal structure analysis of 4k confirms the exo-
syn-anti configuration and shows that the anti-configurated half
(16) Such an interconversion ought to display a high activation barrier
since, if operating, it could not take place via simple π-σ-π-equilibria
only but probably would also require an apparent rotation of the allyl ligand.
This could occur, e.g., via a dissociation of the pyrazole fragment leading
to a T-shaped intermediate, followed by a rotation around the Pd-P bond
and reformation of the chelate ring, as recently reported by Gogoll, B a¨ ckvall,
and co-workers for analogous systems. See: Gogoll, A.; O¨ rnebro, J.;
Grennberg, H.; B a¨ ckvall, J.-E. J. Am. Chem. Soc. 1994, 116, 3631-3632.
20
ligands. Similarly to compound 4c, one observes a distorsion
of the allyl ligand in 4k, such that the angles between the Pd-
(
17) (a) A° kermark, B.; Hansson, S.; Vitagliano, A. J. Am. Chem. Soc.
990, 112, 4587-4588. (b) Sj o¨ gren, M. P. T.; Hansson, S.; Norrby, P.-O.;
A° kermark, B.; Cucciolito, M. E.; Vitagliano, A. Organometallics 1992, 11,
1
(18) A bipyridine system containing a triptycyl substituent has recently
been described as “molecular brake”, by virtue of a conformational change
induced by the complexation to a metal ion that blocks the free rotation of
the triptycyl fragment. See: Ross Kelly, T.; Bowyer, M. C.; Vijaya Bhaskar,
K.; Bebbington, D.; Garcia, A.; Lang, F.; Kim, M. H.; Jette, M. P. J. Am.
Chem. Soc. 1994, 116, 3657-3658.
(19) We presume that the minor isomer adopts a different conformation
of the chelate ring, thus allowing for a slow rotation of the triptycyl moiety
on the NMR time-scale.
3
954-3964. For related studies, see also: (c) Hansson, S.; Norrby, P.-O.;
Sj o¨ gren, M. P. T.; A° kermark, B.; Cucciolito, M. E.; Giordano, F.; Vitagliano,
A. Organometallics 1993, 12, 4940-4948. (d) A° kermark, B.; Zetterberg,
K.; Hansson, S.; Krakenberger, B.; Vitagliano, A. J. Organomet. Chem.
1
987, 335, 133-142. (e) Brown, J. M.; MacIntyre, J. E. J. Chem. Soc.,
Perkin Trans. II 1985, 961-970. (f) Gogoll, A.; O¨ rnebro, J.; Grennberg,
H.; B a¨ ckvall, J.-E. J. Am. Chem. Soc. 1994, 116, 3631-3632.

1
036 J. Am. Chem. Soc., Vol. 118, No. 5, 1996
Togni et al.
P-N(2) plane and the two planes Pd-C(43)-C(44) and Pd-
C(44)-C(45) are significantly different from one another (20.7°
and 43.2°, respectively, see Figure 2b). A point worth noting
about the structure of complex 4k is the angle C(44)-C(45)-
C(46) of 135.9(23)°, representing a large deviation from the
expected 120°.
Chart 4
To our knowledge, compound 4k is the first example of a
complex with an open 1,3-disubstituted allyl ligand selectively
adopting a syn,anti configuration that has been characterized
by X-ray diffraction.
Catalytic experiments using ligand 4k did not afford any
product, even at reflux temperature in THF for days. This at
first sight disappointing observation is in fact very instructive
and can be rationalized as follows. The nucleophilic attack
presumed to take place at the carbon atom adjacent to the
triptycyl fragment is now hampered for steric reasons, presum-
ably. Due to the anti arrangement of the phenyl group at that
center and its position out of the allyl plane away from the metal,
the trajectory of the incoming amine nucleophile is likely to be
different from that for the corresponding syn orientation. In
the former case this trajectory can be described as “from above”,
i.e., from a region of space where the bulky triptycyl is located.
Another possible explanation relates to severe steric interactions
within the Pd(0) olefin complex to be formed upon nucleophilic
attack. In order to form the coordinated olefin, the allyl
fragment has to slide toward the bulky triptycyl fragment,
intruding even more into its cavity. This movement seems
unlikely, if one considers the relatively short distance between
C(45) (the electrophilic center) and the plane of the phenylene
ring C(12)-C(17) of 3.51 Å observed in the crystal structure
of 4k (shortest nonbonding interatomic distance: C(45)-C(13)
preference for nucleophilic attack and strongly indicate that an
early (reagent-like) transition state is involved: (1) the carbon
atom trans to phosphorus is electronically set up to be more
electrophilic, and (2) the other two carbon atoms are already
“
prepared” to lie in the main coordination plane of the Pd(0)
olefin complex forming upon nucleophilic attack. From these
considerations and from the fact that complex 4a exists as single
exo-syn-syn isomer, one can draw the following conclusion. The
observed enantiomer ratio of the amination product 2 (>ca. 200:
1
) in the reaction involving intermediate 4a corresponds to the
ratio of the rates for nucleophilic attacks at the carbon trans to
phosphorus and trans to nitrogen, respectively. In other words,
the activation barrier for C-N bond formation trans to
)
3.55 Å). Furthermore, the lack of reactivity of complex 4k
-1
phosphorus must be at least ca. 3 kcal‚mol lower than for the
toward benzylamine also indicates that the electrophilicity of
the second allylic terminus is not high enough to afford the
aminated reaction product, i.e., the alternative site of attack is
being “ignored” by the nucleophile. This very fact constitutes
a further indirect piece of evidence that in the present Pd(allyl)
system containing P,N-ligands, the nucleophile benzylamine
attacks preferentially or probably even exclusively at the carbon
atom in trans position with respect to the phosphine.
analogous process trans to nitrogen.
Finally, the virtual lack of reactivity of the allylic carbon trans
to nitrogen could be explained by the fact that the necessary
distorsion of the Pd-allyl moiety should take place in the
opposite sense, thus requiring a more drastic rearrangement in
the transition state, and hence not obeying the principle of least
22
motion. Furthermore, if this were to be the case, probable
unfavorable steric interactions would develop during the transi-
tion state. For instance, the evolving pyramidalization of the
carbon atom being attacked by benzylamine would “push back”
the allylic phenyl group toward the phosphine phenyls, a region
of the molecule where not sufficient space appears to be
available for this movement.
Conclusions
An explanation of the pronounced site-selectivity observed
can be based on a simple ground-state argument. The crucial
carbon atom is subjected to a much higher trans influence than
the one trans to the pyrazole nitrogen, as reflected by both the
longer bond distance to palladium (see Table 2) and by the larger
deshielding observed in 13C NMR spectra (see Table 3). These
characteristics render this carbon atom more electrophilic than
its companion trans to nitrogen. The simple ground state
argument will of course have validity only for an early transition
state of the nucleophilic attack (Hammond postulate, see Chart
In order to shed more and conclusive light on these subtle
mechanistic aspects ab initio molecular dynamics calculations
(projector augmented wave method) on this system are currently
2
3
being carried out.
Experimental Section
General Considerations. All reactions with air- or moisture-
sensitive materials were carried out under an argon or nitrogen
atmosphere using standard Schlenk techniques. Freshly distilled, dry,
and oxygen-free solvents were used throughout. 1D- and 2D-NMR
experiments were carried out using a Bruker AMX 500 spectrometer
4
). However, in the course of C-N bond formation an
important distorsion of the allyl ligand should take place.
Specifically, the latter should rotate around the Pd-allyl axis in
order for the developing double bond to move into coplanarity
with the Pd-P-N(2) plane, a situation rather consistent with a
3 8
on 0.02 mmol samples in CDCl or THF-d . Standard pulse sequences
1
24 13
1
25
31
1
were employed for H-2D-NOESY, C- H-, and P- H-correlation
late (product-like) transition state. Precisely this distorsion is
studies.26 The latter two types of experiments were used for resonance
obserVed in the crystal structures described above.21 Thus, in
the ground-state of the intermediates examined, two important
structural features indicate and explain the very high site-
(
21) This important aspect has been brought to our attention by Dr. J.
M. Brown, Oxford University, who observed similar distorsions for the
system reported in ref 3k.
(
20) Farrar and Payne reported the complex [Pd(S,S-chiraphos)(η3-
xylyl)2CCHCHPh)]BPh4 , see: Farrar, D. H.; Payne, N. C. J. Am. Chem.
Soc. 1985, 107, 2054-2058, and we described more recently the compound
(22) For a review of this principle, see: Hine, J. AdV. Phys. Org. Chem.
(
1977, 15, 1-61.
(23) Bl o¨ chl, P. E.; Togni, A., manuscript in preparation.
(24) Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. J. Chem. Phys.
1979, 71, 4546-4553.
3
[
Pd(sparteine)(η -(Ph)2CCHCHPh)][CF3SO3], see: Togni, A.; Rihs, G.;
Pregosin, P. S.; Ammann, C. HelV. Chim. Acta 1990, 73, 723-732.

Palladium-Catalyzed Asymmetric Allylic Amination
J. Am. Chem. Soc., Vol. 118, No. 5, 1996 1037
3
assignment purposes. The phase-sensitive NOESY experiments used
mixing times of 0.8 s. IR spectra of KBr pellets were recorded on a
Perkin-Elmer Paragon 1000 FT spectrometer. Ligands 3 were prepared
[Pd(η -PhCHCHCHPh)(µ-Cl)]
2
(42.5 mg, 0.063 mmol). After the
(44 mg, 0.127 mmol) was
solution had turned clear again, solid TlPF
6
added and stirring continued for 1 h. Workup as detailed above,
followed by recrystallization from acetone/hexane yielded 118 mg
(80%) of compound 4k as dark orange crystals, containing 1 equiv of
acetone: IR (cm ) 3057, 1623, 1512, 1490, 1457, 1242, 1163, 1097,
1002, 842 (PF
(M - PhCHCHCHPh, 100%), 716 (822-Pd), 651 (716-cp), 502
(822-Hpz), 397 (716-Hpz), 331 (397-cp), 289, 212 (397-PPh
Anal. Calcd for C62
Found: C, 63.82; H, 4.93; N, 2.37.
X-Ray Crystallographic Study of (S)-(R)-4c and (R)-(S)-4k. Crys-
tallographic and data collection parameters are provided as supporting
information. Data were measured with variable scan speed to ensure
constant statistical precision on the collected intensities. One standard
reflection was measured every 120 reflections; no significant variation
was detected. However, because of their fast decay when taken out of
the mother liquor (loss of enclosed solvent molecules), crystals of 4k
were embedded in silicon grease and a selected one mounted in a glass
capillary. Scan speed was raised to prevent crystal decay during data
collection. Because of the silicon mantle, outer reflections were
generally weak, and no reflections beyond 2θ ) 40° could be detected.
For both compounds no correction for absorption was applied. The
structures were solved either by Patterson (4c) or direct (4k) methods
and refined by full-matrix least squares using anisotropic displacement
parameters for all non-hydrogen atoms (4c) and for both non-hydrogen
and non-carbon atoms (4k), respectively. Hydrogen atoms were refined
in their idealized position (riding model with fixed isotropic U ) 0.080
7
b
as reported from our laboratory.
Catalytic experiments and the
assignment of the absolute configuration of the product 2 were carried
6
a
-1
out following a reported procedure. Enantiomeric excesses were
determined by HPLC using a Daicel Chiralcel OJ column and eluting
with an hexane/i-PrOH mixture (87:13 v/v; 0.5 mL/min; T ) 25 °C;
retention times: (S)-2 16.8 min, (R)-2 19.5 min)). Elemental analyses
as well as mass spectra were performed by the “Mikroelementar-
+
+
6
), 751, 697, 633, 558; MS (FAB ) m/z 1015 (M ), 822
+
2
), 154.
.
50 6 2 2 3 6
H F FeN P Pd C H O: C, 64.03; H, 4.63; N, 2.30.
analytisches Laboratorium der ETH”.
3
[
Pd(η -PhCHCHCHPh)((R)-(S)-3a)][PF
6
] (4a). To a solution of
ligand (R)-(S)-3a (100 mg, 0.167 mmol) in acetone (10 mL) was added
3
[
Pd(η -PhCHCHCHPh)(µ-Cl)]
2
(50.9 mg, 0.076 mmol). After the
(53.1 mg, 0.152 mmol)
solution had turned clear again, solid TlPF
6
was added, and stirring continued for 1 h. The suspension was filtered
over Celite, the solvent was evaporated in vacuo except for a few
milliliters, and the product precipitated with hexane. Repeated filtration
over Celite, followed by recrystallization from dichloromethane/hexane
yielded compound 4a as an amorphous solid (155 mg, 98%). An
analytical sample was dried under vacuo: IR (cm- ) 3056, 2903, 2849,
1
1
8
-
546, 1511, 1490, 1448, 1434, 1356, 1304, 1241, 1204, 1162, 1002,
+
+
42 (PF
6
), 755, 696, 558; MS (FAB ) m/z 1941 (M
2
PF
6
), 1042 ((MPF
6
+
+
+
H) ), 897 (M , 100%), 704 (M - PhCHCHCHPh), 307, 154. Anal.
Calcd for C52
52 6 2 2
H F FeN P Pd: C, 59.87; H, 5.02; N, 2.69. Found: C,
6
0.12; H, 5.53; N, 2.57.
3
[Pd(η -PhCHCHCHPh)((S)-(R)-3c)][PF
6
] (4c). To a solution of
ligand (S)-(R)-3c (85 mg, 0.15 mmol) in acetone (15 mL) was added
2
3
Å ). Because of the relatively poor quality of the data of 4k, all carbon
rings as well as the PF
[
Pd(η -PhCHCHCHPh)(µ-Cl)]
2
(45 mg, 0.067 mmol). After the
(47 mg, 0.135 mmol) was
-
6
counteranion were refined as rigid groups.
solution had turned clear again, solid TlPF
6
Solvent molecules found in the crystal packing of both 4c and 4k were
highly disordered and could not be unequivocally assigned. For 4k in
particular, best results were obtained by assuming the presence of three
molecules of hexane, one molecule of acetone, and one of water, per
one equiv of 4k, and by refining them without constraints. All
calculations were carried out by using the Siemens SHELXTL PLUS
system.
added, and stirring continued for 1 h. Workup as detailed above,
followed by recrystallization from hexane/diethyl ether yielded 110 mg
(
82%) of compound 4c as orange needles suitable for X-ray, containing
equiv of Et O. An analytical sample was dried in vacuo, thus
releasing the co-crystallized Et
1
2
-
1
2
O; IR (cm ) 1548, 1482, 1437, 1372,
+
1
321, 1245, 1163, 1101, 1045, 842 (PF
6
), 756, 696, 558; MS (FAB )
+
+
+
m/z 853 (M , 100%), 732 (M - FeCp), 660 (M - PhCHCHCHPh),
5
44 6 2 2
02 (660-Hpz), 369, 291, 212, 154. Anal. Calcd for C49H F FeN P -
Acknowledgment. This research was supported by LONZA
Ltd. (Ph.D. grant to U.B.).
Pd: C, 58.91; H, 4.44; N, 2.80. Found: C, 58.79; H, 4.42; N, 2.87.
3
[
Pd(η -PhCHCHCHPh)((S)-(R)-3j)][BF
4
] (4j). To a solution of
ligand (S)-(R)-3j (50 mg, 0.076 mmol) in dichloromethane (10 mL)
3
Supporting Information Available: Tables of crystal data
and refinement details, atomic coordinates, complete listing of
bond distances and angles, tables of anisotropic displacement
coefficients for non-carbon atoms, and coordinates of hydrogen
atoms for compounds (S)-(R)-4c and (R)-(S)-4k, a section of
was added [Pd(η -PhCHCHCHPh)(µ-Cl)]
the solution had turned clear again, AgBF
2
(24 mg, 0.036 mmol). After
(14 mg, 0.072 mmol), solved
4
in 5 mL of methanol, was added and stirring continued for 1 h. Workup
as detailed above, followed by recrystallization from dichloromethane/
hexane yielded 57 mg (76%) of compound 4j as a bright yellow
-
1
1
powder: IR (cm ) 1654, 1636, 1560, 1541, 1490, 1458, 1437, 1386,
the phase sensitive H-2D-NOESY spectrum of compound 4c,
+
1
9
312, 1161, 1084, 1061 (BF
4
), 831, 744, 697, 534; MS (FAB ) m/z
showing the exchange occurring between the exo-syn-syn and
+
+
53 (M ), 760 (M -PhCHCHCHPh), 327, 281, 207, 147, 73 (100%).
FeN Cl : C, 63.74; H, 4.56;
1
the endo-syn-anti isomers, and 500 MHz H NMR spectra of
Anal. Calcd for C57
N, 2.58. Found: C, 63.67; H, 4.74; N, 2.84.
H48BF
4
2
PPd‚0.5 CH
2
2
compounds 4a, 4c, 4j, and 4k (28 pages). This material is
contained in many libraries on microfiche, immediately follows
this article in the microfilm version of the journal, can be ordered
from the ACS, and can be downloaded from the Internet; see
any current masthead page for ordering information and Internet
access instructions.
3
[
Pd(η -PhCHCHCHPh)((R)-(S)-3k)][PF
6
] (4k). To a solution of
ligand (R)-(S)-3k (100 mg, 0.14 mmol) in acetone (15 mL) was added
(
25) Summers, M. F.; Marzilli, L. G.; Bax, A. J. Am. Chem. Soc. 1986,
08, 4285-4294.
26) Sklen a´ r, V.; Miyashiro, H.; Zon, G.; Miles, H. T.; Bax. A. FEBS
Lett. 1986, 208, 94-98.
1
(
JA953031G