Structural features and reactivity of (sparteine)PdCl2: A model for selectivity in the oxidative kinetic resolution of secondary alcohols

Article abstract of DOI:10.1021/ja804955e

The chiral ligand (-)-sparteine and PdCl₂ catalyze the enantioselective oxidation of secondary alcohols to ketones and thus effect a kinetic resolution. The structural features of sparteine that led to the selectivity observed in the reaction were not clear. Substitution experiments with pyridine derivatives and structural studies of the complexes generated were carried out on (sparteine)PdCl₂ and indicated that the C₁ symmetry of (-)-sparteine is essential to the location of substitution at the metal center. Palladium alkoxides were synthesized from secondary alcohols that are relevant steric models for the kinetic resolution. The solid-state structures of the alkoxides also confirmed the results from the pyridine derivative substitution studies. A model for enantioinduction was developed with C₁ symmetry and Cl^- as key features. Further studies of the diastereomers of (-)-sparteine, (-)-α-iso- and (+)-β- isosparteine, in the kinetic resolution showed that these C₂- symmetric counterparts are inferior ligands in this stereoablative reaction.

Full text of DOI:10.1021/ja804955e

Published on Web 10/31/2008
Structural Features and Reactivity of (Sparteine)PdCl : A
2
Model for Selectivity in the Oxidative Kinetic Resolution of
Secondary Alcohols
Raissa M. Trend and Brian M. Stoltz*
The Arnold and Mabel Beckman Laboratories of Chemical Synthesis, DiVision of Chemistry and
Chemical Engineering, California Institute of Technology, 1200 East California BouleVard,
MC 164-30, Pasadena, California 91125
Received June 27, 2008; E-mail: stoltz@caltech.edu
2
Abstract: The chiral ligand (-)-sparteine and PdCl catalyze the enantioselective oxidation of secondary
alcohols to ketones and thus effect a kinetic resolution. The structural features of sparteine that led to the
selectivity observed in the reaction were not clear. Substitution experiments with pyridine derivatives and
structural studies of the complexes generated were carried out on (sparteine)PdCl
2
and indicated that the
C
1
symmetry of (-)-sparteine is essential to the location of substitution at the metal center. Palladium
alkoxides were synthesized from secondary alcohols that are relevant steric models for the kinetic resolution.
The solid-state structures of the alkoxides also confirmed the results from the pyridine derivative substitution
-
studies. A model for enantioinduction was developed with C
1
symmetry and Cl as key features. Further
studies of the diastereomers of (-)-sparteine, (-)-R-iso- and (+)-ꢀ-isosparteine, in the kinetic resolution
showed that these C -symmetric counterparts are inferior ligands in this stereoablative reaction [Mohr,
2
J. T.; Ebner, D. C.; Stoltz, B. M. Org. Biomol. Chem. 2007, 5, 3571-3576].
a
Introduction
Scheme 1
Palladium-catalyzed aerobic oxidation has been known for
nearly a century, since the discovery by Wieland that this metal
1
could convert alcohols to aldehydes in aqueous media. Modern
developments of this reaction were initiated in 1977 by Schwartz
and Blackburn, who reported that palladium(II) acetate and
sodium acetate could aerobically oxidize alcohols to carbonyl
2
,3
compounds. More recently, our laboratory has developed the
enantioselective aerobic oxidation of secondary alcohols by
4
,5
palladium(II) and (-)-sparteine (1). An example of this
transformation is shown in Scheme 1. While numerous im-
provements to oxidative kinetic resolution have been made and
a
nbd ) norbornadiene. 0.1 M in toluene. Five mol % Pd(nbd)Cl2.
5a,e,6
the mechanism has been analyzed kinetically,
there had been
no reports detailing a structural model for asymmetric induction
(
(
1) Wieland, H. Ber. Dtsch. Chem. Ges. 1912, 45, 484–493.
by 1 in these reactions. We recently communicated such a
2) Schwartz, J.; Blackburn, T. F. J. Chem. Soc., Chem. Commun. 1977,
7,8
model, which was subsequently verified by calculations.
1
57–158.
Herein, we describe in full the experiments that support our
model for the absolute stereochemical outcome. This model is
based on the reactivity and solid-state structures of a series of
palladium(II) complexes, including the first reported structures
of chiral palladium alkoxides relevant to this system. New
studies of the diastereomers of 1 and of another catalyst
precursor lend further support to our model and highlight the
alkaloid 1 as an atypical c chiral ligand.
(
3) For recent Pd(II) aerobic alcohol oxidations, see: (a) Peterson, K. P.;
Larock, R. C. J. Org. Chem. 1998, 63, 3185–3189. (b) Nishimura, T.;
Onoue, T.; Ohe, K.; Uemura, S. J. Org. Chem. 1999, 64, 6750–6755.
(
2
5
c) Brink, G.-J.; Arends, I. W. C. E.; Sheldon, R. A. Science 2000,
87, 1636–1639. (d) For a review, see: Muzart, J. Tetrahedron 2003,
9, 5789–5816.
(
(
4) (a) Ferreira, E. M.; Stoltz, B. M. J. Am. Chem. Soc. 2001, 123, 7725–
7
726. (b) Bagdanoff, J. T.; Ferreira, E. M.; Stoltz, B. M. Org. Lett.
2
003, 5, 835–837. (c) For conditions employing chloroform and air,
see: Bagdanoff, J. T.; Stoltz, B. M. Angew. Chem., Int. Ed. 2004, 43,
3
53–357.
Since the development of the oxidative kinetic resolution
of secondary alcohols in our laboratories, several improve-
5) For a similar system, see: (a) Jensen, D. R.; Pugsley, J. S.; Sigman,
M. S. J. Am. Chem. Soc. 2001, 123, 7475–7476. (b) Mueller, J. A.;
Jensen, D. R.; Sigman, M. S. J. Am. Chem. Soc. 2002, 124, 8202–
8
203. (c) Jensen, D. R.; Sigman, M. S. Org. Lett. 2003, 5, 63–65. (d)
(6) Mueller, J. A.; Cowell, A.; Chandler, B. D.; Sigman, M. S. J. Am.
Chem. Soc. 2005, 127, 14817–14824.
Mandal, S. K.; Jensen, D. R.; Pugsley, J. S.; Sigman, M. S. J. Org.
Chem. 2003, 68, 4600–4603. (e) Mueller, J. A.; Sigman, M. S. J. Am.
Chem. Soc. 2003, 125, 7005–7013. (f) Mandal, S. K.; Sigman, M. S.
J. Org. Chem. 2003, 68, 7535–7537.
(7) Trend, R. M.; Stoltz, B. M. J. Am. Chem. Soc. 2004, 126, 4482–4483.
(8) Nielsen, R. J.; Keith, J. M.; Stoltz, B. M.; Goddard, W. A. J. Am.
Chem. Soc. 2004, 126, 7967–7974.
10.1021/ja804955e CCC: $40.75

2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 15957–15966 9 15957

A R T I C L E S
Trend and Stoltz
Table 1. Comparison of Conditions for the Kinetic Resolution of
Secondary Alcohols
Figure 1. General mechanism of alcohol oxidation by palladium(II).
a
nbd ) norbornadiene, s ) selectivity factor. 500 mg of MS3Å/
b
mmol substrate was used for all four sets of conditions. 5 mol %
palladium catalyst, 0.20 equiv of 1, 0.1 M in toluene. 5 mol %
c
palladium catalyst, 0.20 equiv of 1, 0.5 equiv of Cs2CO3, 1.5 equiv of
d
t-BuOH, 0.25 M in toluene. 5 mol % palladium catalyst, 0.12 equiv of
e
1
, 0.40 equiv of Cs2CO3, 0.25 M in CHCl3. (sp)PdCl2 was used as
catalyst (5 mol %), 0.07 equiv of 1, 0.4 equiv of Cs2CO3, 0.25 M in
CHCl3. Reactions were run open to air through a short drying tube of
Drierite.
4
b,c,5
ments have been made to the catalytic system.
These
results are summarized in Table 1 for three different
9
substrates. The original conditions, which employ toluene,
can be accelerated by the addition of Cs2CO3 and tert-butanol.
The use of chloroform as a solvent permits reaction at a lower
temperature and facilitates the use of air as a reoxidant.
Extensive studies of the substrate scope under these reaction
conditions have been carried out, and the reaction has been
employed in the synthesis of certain pharmaceutical inter-
1
0
mediates. Despite these advancements, we were unable to
find any ligand other than (-)-sparteine (1) capable of
inducing high levels of selectivity in the reaction, and we
were thus limited to one enantiomeric series, as the other
Figure 2. ((-)-Sparteine)PdCl2 (14, (sp)PdCl2). The molecular structures
are shown with 50% probability ellipsoids. The hydrogen atoms in the (-)-
sparteine (1) backbone in the side view have been omitted for clarity.
1
1,12
enantiomer of 1 is not readily available.
In general, the oxidation of alcohols to carbonyl compounds
by palladium(II) most likely involves associative alcohol
substitution at the palladium center (Figure 1, 7), deprotonation
of the resulting palladium alcohol complex (8) by base to form
a palladium alkoxide (9), and ꢀ-hydrogen elimination to the
carbonyl compound (13, Figure 1, catalyst regeneration steps
not shown). Sigman and co-workers have carried out extensive
kinetic studies of the oxidative kinetic resolution using ((-)-
sparteine)PdCl2 (14, (sp)PdCl2, Figure 2), (-)-sparteine (1), and
O2. In their original publication, the authors reported that
13
(
9) The selectivity factor, s, or krel was determined by the following
equation: s ) krel ) ln[(1- C)(1-ee)]/ln[(1- C)(1 + ee)], where C
is conversion and ee is enantiomeric excess.
ꢀ
-hydrogen elimination is rate-determining at high base con-
centration and proposed that the observed enantioselectivity
results from a combination of the respective energy barriers for
the elimination and the thermodynamic stabilities of the alkoxide
(
10) Caspi, D. D.; Ebner, D. C.; Bagdanoff, J. T.; Stoltz, B. M. AdV. Synth.
Catal. 2004, 346, 185–189.
(
11) O’Brien and co-workers have synthesized analogues of (+)-sparteine
that are somewhat effective as sparteine surrogates, see: (a) Dearden,
M. J.; Firkin, C. R.; Hermet, J.-P. R.; O’Brien, P. J. Am. Chem. Soc.
5e
complexes. A more recent publication from the same group
has revised this original proposal but maintains that ꢀ-hydrogen
2
002, 124, 11870–11871. (b) Hermet, J.-P. R.; Porter, D. W.; Dearden,
elimination from and reprotonation of the intermediate palladium
M. J.; Harrison, J. R.; Koplin, T.; O’Brien, P.; Parmene, J.; Tyurin,
6
alkoxide are the key kinetic influences on the selectivity. The
V.; Whitwood, A. C.; Gilday, J.; Smith, N. M. Org. Biomol. Chem.
2
003, 1, 3977–3988. (c) Dearden, M. J.; McGrath, M. J.; O’Brien, P.
J. Org. Chem. 2004, 69, 5789–5792. (d) Ebner, D. C.; Trend, R. M.;
Genet, C.; McGrath, M. J.; O’Brien, P.; Stoltz, B. M. Angew. Chem.,
Int. Ed. 2008, 47, 6367–6370.
(13) (a) Sheldon, R. A.; Arends, I. W. C. E. In AdVances in Catalytic
ActiVation of Dioxygen by Metal Complexes; Sim a´ ndi, L. I., Ed.;
Kluwer: Dordrecht; 2003; Vol. 26,Chap. 3, p 123. (b) Steinhoff, B. A.;
Stahl, S. S. Org. Lett. 2002, 4, 4179–4181. (c) Stahl, S. S.; Thorman,
J. L.; Nelson, R. C.; Kozee, M. A. J. Am. Chem. Soc. 2001, 123,
7188–7189, and references therein.
(
12) The use of a binaphthyl-derived N-heterocylic carbene ligand with
PdI2 hasrecentlybeenreported,withmoderatelevelsofselectivity:Chen,
T.; Jiang, J.-J.; Xu, Q.; Shi, M. Org. Lett. 2007, 9, 865–868.
1
5958 J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008

Oxidative Kinetic Resolution of Secondary Alcohols
A R T I C L E S
a
Scheme 2
Figure 3. Transition state for ꢀ-hydrogen elimination superimposed on a
quadrant diagram for a C2-symmetric ligand.
alkoxide of the enantiomer that is oxidized undergoes ꢀ-hy-
drogen elimination more rapidly than the other and is reproto-
nated more slowly.
In the analysis described above, 1 was treated as a C2-
14
symmetric ligand. Only two diastereomeric palladium alkox-
ides are considered, while in fact four diastereomers are possible
because (-)-sparteine is C1-symmetric. Since the introduction
1
5
of the diisooctyl phthalate (DIOP) ligand by Kagan in 1972,
C2 symmetry has been a dominant guiding force in the design
of improved ligands for enantioselective transition-metal-
catalyzed reactions. In this respect, a C2-symmetric scaffold
offers distinct advantages: the presence of a symmetry axis
reduces the number of competing diastereomeric reaction
pathways, enables a more straightforward analysis of substrate-
catalyst interactions, and simplifies mechanistic and structural
a
The molecular structures are shown with 50% probability ellipsoids-.
The hydrogen atoms in the ligand backbone in the side view and the SbF6
ion in both views have been omitted for clarity.
1
6
studies. The chiral scaffold of a C2-symmetric ligand can be
modeled by a quadrant diagram of the ligand-metal complex,
in which quadrants I and III (or II and IV) are equivalently
hindered (Figure 2A). (sp)PdCl2 (14) can be mapped onto this
quadrant representation, but the hemispheres of this complex
ligands for all enantioselective reactions. We show also the
significance of the halide ligand in enantioinduction, which is
also borne out by calculations.
17
are clearly nonequivalent. As can be seen from the solid-state
Results and Discussion
1
2
structure, Cl and Cl are located in different steric environ-
ments, due to the different influence of each of the flanking
piperidine rings of (-)-sparteine (1). Thus, it is perhaps more
accurate to depict 14 with quadrant I fully blocked but quadrant
III only partially blocked (Figure 2B).
2
Reactivity of (sp)PdCl (14) with Pyridine Derivatives and
Their Solid-State Structures. To experimentally test the degree
to which the C1-symmetric ligand (-)-sparteine (1) approaches
C2 symmetry, a series of substitution reactions were carried out
on (sp)PdCl2 (14). In an initial experiment, 14 was treated with
Given that ꢀ-hydrogen elimination is key to enantiodiffer-
entiation, it was also not obvious to us how a C2-symmetric
ligand could induce asymmetry in this process. Figure 3 depicts
the transition state for ꢀ-hydrogen elimination (see below)
superimposed on a quadrant diagram for a C2-symmetric ligand.
As the ꢀ-agostic interaction between the palladium center and
the C-H atom approaches a full Pd-H bond, the transition
state becomes more and more symmetrical. If the secondary
carbon atom is aligned with the origin point of the quadrant
diagram, in fact RL and RS are in identical steric environments.
To understand why (-)-sparteine (1) was the only highly
effective ligand for the oxidative kinetic resolution, we became
interested in exploring the steric environment that 1 created in
palladium complexes. We chose to address the extent to which
the C1-symmetric 1 mimics C2 symmetry when bound to PdCl2.
Reported herein are the full results of these investigations, which
culminated in an experimentally derived model for the selectivity
observed in the oxidative kinetic resolution of secondary
alcohols. Further studies involving two other members of the
1
equiv of AgSbF6 in the presence of pyridine (Scheme 2) at
2
5 °C in CH2Cl2 to provide 91% yield of a crystalline product.
We anticipated that if 1 behaved as a C2-symmetric ligand, two
cationic pyridyl complexes (16 and 17) would form in nearly
equivalent amounts. In the event, one major species was
observed by H and C NMR in acetone-d6 along with a small
amount of another compound. The analysis of a single crystal
of the product by X-ray diffraction revealed complex 16. This
was our first indication that (-)-sparteine (1) does not mimic
C2 symmetry.
Although a solution-state structural analysis was not carried
out, the major component in solution appears to be 16.
Treatment of the complex with pyridine-d5 in acetone showed
rapid deuterium incorporation and liberation of free pyridine at
1
13
19
2
5 °C (Scheme 3). It is unclear at this time what accounts for
the minor component observed in solution. Observation of 16
by H NMR at 15 degree intervals from -60 to 45 °C in
acetone-d6 showed almost no change in the product ratio.
1
18
lupin alkaloid family, (-)-R- and (+)-ꢀ-isosparteine, comple-
ment our model and belie the superiority of C2-symmetric
(
19) The same product was observed in the reaction of the chloro-bridged
dimer with 2.0 equiv of pyridine in 91% yield.
(
14) For another example that uses similar logic for (-)-sparteine, see: Li,
X.; Schenkel, L. B.; Kozlowski, M. C. Org. Lett. 2000, 2, 875–878.
15) Kagan, H. B.; Dang, T.-P. J. Am. Chem. Soc. 1972, 94, 6429–6433.
16) Whitesell, J. K. Chem. ReV. 1989, 89, 1581–1590.
(
(
(
(
17) Structure originally reported in ref 3a.
18) Leonard, N. J. In The Alkaloids, Chemistry and Physiology, Vol. III;
Manske, R. H. F., Holmes, H. L., Eds.; Academic Press: New York,
1
953; pp 119-199.
.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008 15959

A R T I C L E S
Trend and Stoltz
a
Scheme 3
Scheme 5
Scheme 4
We next investigated the reaction of 14 with a bulkier
pyridine-related compound, 2-methylisoquinoline (18). The
structure and geometry of 18 can be considered structurally
analogous to acetophenone, the product of the kinetic resolution
(Scheme 4). Under the same reaction conditions, (sp)PdCl2 (14)
reacts with AgSbF6 in the presence of 18 to produce a crystalline
product in moderate yield. If (-)-sparteine (1) were exhibiting
the properties of a C2-symmetric ligand, we anticipated that four
possible diastereomeric products could form, resulting from
substitution at either Cl or Cl and the atropisomer at each
position, but that two would dominate: 19 and 21, or 20 and
1
2
a
The molecular structures are shown with 50% probability ellipsoids-.
The hydrogen atoms in the ligand backbone in the side view and the SbF6
ion in both views have been omitted for clarity.
2
2. In the event, two major compounds are indeed observed by
1
H NMR in a 1.3:1 ratio in acetone-d6 at 25 °C. However,
analysis in the solid state by X-ray crystallography revealed
1
only substitution of Cl , with both orientations of the isoquino-
line ligand related by a rotation of 180° around the Pd-N bond
occupying the site, i.e., 19 and 20 (for crystallographic
2
0
Figure 4. (-)-Sparteine (1) bound to PdCl2, which is able to differentiate
structures, see Supporting Information). This outcome pro-
vided further evidence against the possibility that 1 behaved
with pseudo-C2 symmetry in this complex. In addition, the
quadrants I and III.
°C in acetone-d6 (see Supporting Information). The addition of
an equivalent of 2-mesitylpyridine (23) resulted in an increase
in the minor peaks in the aryl region of the spectrum and a
disappearance of the minor peaks in the region from 1 to 5 ppm
corresponding to palladium-bound (-)-sparteine (1). Thus, the
minor product observed by NMR is likely dissociated ligand
and an acetone·(sp)PdCl cation, and not any of the other
possible positional or rotational isomers.
These ligand substitution experiments present convincing
evidence that (-)-sparteine (1) does not mimic a C2-symmetric
ligand when bound to PdCl2, and they provide information about
the steric environment this ligand creates. The unique archi-
tecture favors substitution at site A on the palladium center, as
depicted in Figure 4, and can sterically distinguish quadrants
III and IV for sufficiently bulky unsymmetrical ligands. Most
importantly, quadrants I and III are differentiated, unlike in a
truly C2-symmetric ligand.
2
mixture of products indicated that the (sp)PdCl fragment was
not able to completely discriminate the prochiral faces of a
planar molecule, albeit one with a steric environment similar
to that of acetophenone.
2
In order to probe the steric environment of the (sp)PdCl
fragment further, we turned to an even bulkier ligand, 2-mesi-
tylpyridine (23). In this instance, under conditions identical to
those used for the syntheses of 16 and 19/20, we observed a
1
good yield of one major product out of four possible by H
NMR in acetone-d6 with approximately 5% of a minor
component (Scheme 5). The presence of a single major product
indicated not only that substitution occurred at a single site of
the palladium center but also that a single atropisomer about
the Pd-N bond was favored for a suitably bulky ligand. X-ray
1
crystallographic analysis showed that Cl had been substituted
as in 16 and 19/20 to give 24 and that the mesityl group resided
exclusively in quadrant IV.
The ratio of major and minor products observed in the H
NMR spectrum remained nearly unchanged from -60 to 45
Synthesis and Reactivity of Palladium Alkoxides Relevant
to the Oxidative Kinetic Resolution. To better understand the
enantioselective oxidation itself, we set out to synthesize an
1
1
5960 J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008

Oxidative Kinetic Resolution of Secondary Alcohols
A R T I C L E S
a
Scheme 6
Figure 5. R-(Trifluoromethyl)benzyl alcohol (29), a meaningful steric
model for the prototypical oxidative kinetic resolution substrate 1-phenyl-
ethanol (28).
alkoxide complex relevant to our system. While numerous
palladium alkoxides have been characterized, few bear ꢀ-hy-
2
1
drogen atoms due to the ease of ꢀ-hydrogen elimination.
A
meaningful steric model for the prototypical oxidative kinetic
resolution substrate 1-phenylethanol (28) is R-(trifluorometh-
2
2
yl)benzyl alcohol (29, Figure 5). Under our kinetic resolution
conditions, 29 does not react, presumably because the electron-
withdrawing trifluoromethyl group disfavors the ꢀ-agostic
interaction between the palladium atom and the benzylic C-H
bond that precedes elimination. Thus, we hoped that a
palladium-alkoxide complex of this alcohol would be relatively
stable and serve as a model of a relevant intermediate in the
oxidation reaction. In this way, such an alcohol would act as a
“suicide inhibitor”, or as a sort of trap for a reactive intermediate.
Alcohol (+)-29, which corresponds to the more reactive
enantiomer of 1-phenylethanol ((+)-28) in the resolution
chemistry, was chosen as an initial target for palladium-alkoxide
formation.
a
The molecular structures are shown with 50% probability ellipsoids.
The hydrogen atoms in the ligand backbone in the side view have been
omitted for clarity.
Treatment of complex 14 with the sodium alkoxide of (+)-
2
9 produced a 64% yield of recrystallized material as a single
1
major product as observed by H NMR in CD2Cl2 (Scheme 6).
The product was shown to be alkoxide complex 31 by X-ray
analysis. A single atropisomer is observed, and again substitution
Scheme 7
1
of Cl occurs exclusively. The phenyl moiety is located in
quadrant IV in an orientation similar to that of the mesityl group
of complex 24. As in the cationic pyridyl complexes described
above (i.e., 24), the square plane of 31 is distorted such that
2
Cl is displaced toward open quadrant II. In addition, the
benzylic C-H bond is oriented toward the palladium center
and parallel to the distorted Pd-Cl bond.
yield was observed along with unidentified palladium decom-
2
3
Thermolysis of 31 in toluene-d8 led to little, if any, ketone.
However, when 31 was treated with 1.5 equiv of AgSbF6 in
CD2Cl2 at 25 °C, immediate production of ketone 32 in 92%
position products (Scheme 7). Silver cation appears to act as
a trigger that releases the “trapped” intermediate to undergo
ꢀ-hydrogen elimination.
The synthesis of the other diastereomer of palladium alkoxide,
arising from the opposite enantiomer of the alcohol ((-)-29),
was also pursued. When (-)-29 was treated with sodium hydride
and added to 14, however, two major products were observed
by H NMR in an approximately 1:1 ratio (Scheme 8). This
was our initial indication that the interaction between the
(
(
20) Two molecules were found in the asymmetric unit. See Supporting
Information for further details.
21) For examples that possess ꢀ-hydrogen atoms, see: (a) Bouquillon, S.;
du Moullinet d’Hardemare, A.; Averbuch-Pouchot, M.-T.; Henin, F.;
Muzart, J.; Durif, M. Acta Crystallogr., Sect. C: Cryst. Struct.
Commun. 1999, 55, 2028–2030. (b) Achternbosch, M.; Klufers, P.
Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 1994, 50, 175–
1
(
sp)PdCl fragment and the alcohol corresponding to the slow-
1
4
78. (c) Hubel, R.; Polborn, K.; Beck, W. Eur. J. Inorg. Chem. 1999,
reacting enantiomer of 1-phenylethanol ((-)-29) was more
complicated than that for (+)-29. The products did not crystal-
lize readily, and only after standing for a period of several
months at -18 °C was a mixture of orange and yellow crystals
obtained. X-ray analysis revealed the yellow crystals to be
71–482. (d) Kapteijn, G. M.; Grove, D. M.; van Koten, G.; Smeets,
W. J. J.; Spek, A. L. Inorg. Chim. Acta 1993, 207, 131–134. (e)
Klufers, P.; Kunte, T. Angew. Chem., Int. Ed. 2001, 40, 4210–4212.
(
6
f) Kastele, X.; Klufers, P.; Kunte, T. Z. Anorg. Allg. Chem. 2001,
27, 2042–2044. (g) Kapteijn, G. M.; Baesjou, P.; Alsters, P. L.; Grove,
D. M.; Smeets, W. J. J.; Kooijman, H.; Spek, A. L.; van Koten, G.
Chem. Ber. 1997, 130, 35–44. (h) Kapteijn, G. M.; Dervisi, A.; Grove,
D. M.; Kooijman, H.; Lakin, M. T.; Spek, A. L.; van Koten, G. J. Am.
Chem. Soc. 1995, 117, 10939–10949. (i) Ahlrichs, R.; Ballauff, M.;
Eichkorn, K.; Hanemann, O.; Kettenbach, G.; Klufers, P. Chem.-Eur.
J. 1998, 4, 835–844. (j) Klufers, P.; Kunte, T. Eur. J. Inorg. Chem.
1
bisalkoxide 34, the H NMR spectrum of which was not present
in the spectrum of the original product mixture. The orange
crystals were dichloride complex 14, also not observed in the
1
original H NMR spectrum. We propose that the initially formed
2
002, 1285–1289.
22) A racemic Pd(II) alkoxide complex of (()-29 has been reported:
(PMe3)2Pd(Me)(OR)], where R ) PhCH(CF3). See: Kim, Y.-J.;
Osakada, K.; Takenaka, A.; Yamamoto, A. J. Am. Chem. Soc. 1990,
12, 1096–1104.
products are a mixture of alkoxide atropisomers that cannot
(
[
1
(23) Yield was determined by H NMR using an internal standard; see
1
Supporting Information.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008 15961

A R T I C L E S
Trend and Stoltz
a
Scheme 8
Figure 7. ꢀ-Hydrogen elimination from a square planar d8 complex.
an interaction could also lead to rotation around the Pd-O bond,
giving rise to the mixture of products initially observed. In the
1
2
bisalkoxide complex, the N -Pd-O angle is less acute than
1
the corresponding N -Pd-Cl angle in monoalkoxide complex
3
1 (173.14° vs 164.59°), a geometry which perhaps can better
accommodate a bulkier group in quadrant III.
Experimentally Derived Model for the Stereoselectivity
Observed in the Oxidative Kinetic Resolution of Secondary
Alcohols. On the basis of the reactivity of (sp)PdCl2 (14) and
the structures of 16, 19/20, 24, 31, and 34, we have proposed
a general model for asymmetric induction in the palladium-
a
25
The molecular structures are shown with 50% probability ellipsoids.
catalyzed oxidative kinetic resolution of secondary alcohols.
The hydrogen atoms in the ligand backbone in the side view have been
omitted for clarity.
Upon reaction of complex 14 with a racemic mixture of alcohol
1
2
(
35), Cl is substituted preferentially over Cl to form two
diastereomeric palladium alkoxides (Figure 6, 36 and 37). Either
could be reprotonated, leading to alcohol dissociation, or could
undergo ꢀ-hydrogen elimination to afford ketone. Given the
geometry of the solid-state structure of palladium-alkoxide
complex 31, we propose that for the reactive diastereomer (37),
the unsaturated moiety (RL) is located in open quadrant IV. In
this geometry, the secondary C-H bond is positioned opposite
the oblique Pd-Cl bond. The same orientation of RL in quadrant
IV of the less reactive diastereomer 36 requires that the C-H
bond point away from Pd (not shown). Alternatively, for
diastereomer 36, the C-H bond could approach from the
opposite face of the square plane, as depicted by 36a. This
geometry, however, would be expected to induce destabilizing
interactions between the alkoxide moiety and the piperidine ring
in quadrant III, and between the chloride ligand and the
piperidine ring in quadrant I. On the basis of the solid-state
structure of bisalkoxide 34, we expect that unreactive dia-
stereomer 36 likely has the geometry shown in 36b.
Figure 6. Model for the enantioselection in the oxidative kinetic resolution
by (sp)PdCl2.
undergo ꢀ-hydrogen elimination, and which eventually dispro-
portionate to 14 and 34.
The transition state for ꢀ-hydrogen elimination (i.e., 38 or
It is unclear why alcohol (-)-29 forms a more complex initial
product mixture than the other enantiomer, or why dispropor-
tionation occurs. On the basis of the solid-state structure of the
bisalkoxide, one possibility is that the aryl group, now oriented
3
9) is expected to involve a developing cationic palladium
26
species with four-coordinate square planar geometry, although
-
calculations for this system have shown that Cl remains closely
associated below the square plane (see below). A schematic
representation of ꢀ-hydrogen elimination for our ligand set is
1
toward quadrant III, experiences a steric clash with Cl , which
24
destabilizes the complex and leads to disproportionation. Such
(
25) While the structural studies we have undertaken have been almost
exclusively in the solid state, the consistency of the results and the
support of theoretical calculations give us confidence in our model.
(
24) Another possibility is that the thermodynamics of solubility play a
role.
1
5962 J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008

Oxidative Kinetic Resolution of Secondary Alcohols
A R T I C L E S
a
Table 2
Scheme 9
shown in Figure 7. The C-H bond likely moves into the square
plane by an associative substitution mechanism and displace-
ment of the chloride ligand (41). ꢀ-Hydrogen elimination results
in an intermediate possessing a hydride in a basal position (43).
It is possible that ꢀ-hydrogen elimination could occur directly
from an apical position (Figure 5B). In this case, elimination
would result in an apical hydride ligand (45). However, because
the weakly antibonding dz2 orbital of a square planar d8 complex
a
The molecular structures are shown with 50% probability ellipsoids.
The hydrogen atoms in the ligand backbone in the side view have been
omitted for clarity.
Table 3. Kinetic Resolution with (sp)PdBr2 (45)
is already filled, the introduction of a σ-bonding apical ligand
creates a fully antibonding interaction. Because hydride is a
stronger σ-donor than chloride, apical ꢀ-hydrogen elimination
would result in a higher-energy intermediate (45) than if chloride
were forced into the apical position (42). In examples of
ꢀ
-hydrogen elimination from alkyl complexes, the hydride is
entry
X
conditions
time (h)
conversion (%)
ee (%)
s
delivered to a vacant orbital, not a filled one. Furthermore,
calculations by Hoffmann have shown that for five-coordinate,
square pyramidal species, hydride ligands prefer to be basal.
In structure 31, and in 37 by analogy, the reactive C-H bond
is poised to achieve the conformation for ꢀ-hydrogen elimination
via 39 after displacement of Cl into quadrant II (Figure 6).
This sequence of events minimizes potential steric interactions
en route to ketone. In contrast, achievement of a similar
conformation by diastereomer 36 would entail a destabilizing
1
2
3
4
Br
Br
Cl
Cl
A
B
A
B
28
6.5
96
48
62
56
60
60
99
94
98
99
28
22
23
31
2
7
2
intermediate. ꢀ-Hydrogen elimination proceeds by displacement
of the chloride anion by the benzylic C-H bond, although the
chloride ion, while not fully bound, remains closely associated
to the metal center in the transition state. Significantly, when
the chloride ligand was left out of the calculation of the barriers
to ꢀ-hydrogen elimination from each diastereomer, little dif-
ference was found, which indicates that the rates would be
similar. Thus, the chloride anion appears to be essential to the
selectivity.
Effect of the Halide Ligand on Selectivity and Reactivity. On
the basis of the experiments and calculations described above,
chloride anion appears to play an essential role in communicat-
ing the chirality of (-)-sparteine (1) to the bound substrate.
Table 2 compares the angles around the square plane determined
from the solid-state structures for the (sp)Pd-derived complexes
reported herein. The largest disparity among the complexes is
2
interaction between RL and quadrant III or between Cl and
quadrant I. Furthermore, and perhaps more importantly, a closely
associated chloride ion in the apical position below the square
plane could further disfavor the diastereomeric transition state
(38) by steric crowding with RL. Thus, 37 reacts to form ketone
4
0, while 36 protolytically dissociates to the observed enanti-
omer of resolved alcohol 35. This model predicts the absolute
stereochemical outcome of every (sp)PdCl2 (14)-catalyzed
oxidative kinetic resolution performed to date.
Extensive calculations by Goddard and co-workers on the
oxidative kinetic resolution catalyzed by (sp)PdCl2 (14) provide
8
theoretical support to the model described herein. The high-
level calculations determine the geometry of the alkoxide
intermediate for sec-phenethyl alcohol to be similar to that found
in the solid-state structure of 31, our model compound for this
1
1
in the N -Pd-B, A-Pd-B, and N -Pd-A angles. The
-mesitylpyridine cationic complex (24) and the monoalkoxide
31) are the most distorted, the sum of the six angles between
2
(
ligands being 699.45° and 701.58°, respectively, compared to
720° in a perfect square plane and 657° for a tetrahedral
geometry. Because, as proposed, Cl remains close to the
palladium atom throughout the reaction, it may effectively block
quadrant II, leaving only one quadrant open for substrate
(
26) (a) Bryndza, H. E.; Tam, W. Chem. ReV. 1988, 88, 1163–1188. (b)
Zhao, J.; Hesslink, H.; Hartwig, J. F. J. Am. Chem. Soc. 2001, 123,
2
7
220–7227. (c) For an alternative with platinum, see: Bryndza, H. E.;
Calabrese, J. C.; Marsi, M.; Roe, D. C.; Tam, W.; Bercaw, J. E. J. Am.
Chem. Soc. 1986, 108, 4805–4813.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008 15963

A R T I C L E S
Trend and Stoltz
a
Scheme 10
Figure 8. Sparteine and isosparteine diastereomers and their complexes
with PdCl2.
binding. In this way, the transition state for ꢀ-hydrogen
elimination is further desymmetrized (see Figure 3).
If this proposed role for chloride is correct, then a larger but
still coordinating ligand at position B should improve the kinetic
resolution. To investigate this possibility, (sp)PdBr2 (45) was
pursued. Ligation of 22 to palladium precursor 46 occurred in
moderate yield to provide 45 (Scheme 9). The molecular
structure of 45, obtained by X-ray diffraction of a single crystal,
a
The molecular structures are shown with 50% probability ellipsoids.
The hydrogen atoms in the ligand backbone in the side view have been
omitted for clarity.
a
Scheme 11
2
2
shows that the Pd-Br bond is further distorted than the Pd-Cl
bond of 14. The sum of the angles between the six ligands is
99.22°, which is comparable to the most tetrahedrally distorted
complex containing a chloride at position B (24, Table 2).
5 was tested in the kinetic resolution reaction of 1-phenyl-
6
4
ethanol (28). The results for two sets of conditions are shown
9
in Table 3 (entries 1 and 2). Interestingly, high selectivites are
obtained at rates 3-7 times faster than with (sp)PdCl2 (14,
entries 3 and 4). The effect of the bromide ligand appears to be
rate acceleration with comparable selectivity.
Properties and Reactivity of r-Isosparteine and ꢀ-Iso-
sparteine in the Oxidative Kinetic Resolution. Having developed
a model for selectivity, we again turned our attention to the
issue of symmetry and whether the C2-symmetric diastereomers
of (-)-sparteine (1) would be more or less selective in the kinetic
resolution. The PdCl2 complexes of the three diastereomers are
a
The molecular structures are shown with 50% probability ellipsoids.
The hydrogen atoms in the ligand backbone in the side view have been
omitted for clarity.
2
8,29
shown in Figure 8 for comparison.
For 1, the reactions of
ligands of 49 would be in an environment identical to that of
(
sp)PdCl2 (14) we have thus far observed have occurred
2
Cl in 14. On the other hand, for the (+)-ꢀ-isosparteine (48)
primarily at site A, which appears to be the relevant site of
substitution in the kinetic resolution. For the (-)-R-isosparteine
complex ((ꢀ-isosp)PdCl2, 50), in which both flanking piperidine
rings are directed away from the metal center, only A sites are
available. It was unclear to us what effect this would have on
the rate of oxidation. If the C1 symmetry of 1 is essential to its
unique selectivity in the kinetic resolution, a comparison of the
three diastereomers would reveal this point.
(
47) complex ((R-isosp)PdCl2, 49), reaction, if any, would be
forced to occur at a B site, by analogy to 1. That is, both anionic
(
(
27) Thorn, D. L.; Hoffmann, R. J. Am. Chem. Soc. 1978, 100, 2079–
090.
2
(
-)-R-Isosparteine (47) was prepared by the method of
28) For references pertaining to R-isosparteine, see: (a) Winterfeld, K.
Arch. Pharm. 1928, 266, 299–325. (b) Marion, L.; Turcotte, F.; Ouellet,
J. Can. J. Chem. 1951, 29, 22–29. (c) Galinovsky, F.; Knoth, P.;
Fischer, W. Monatsh. Chem. 1955, 86, 1014–1023. (d) Leonard, N. J.;
Thomas, P. D.; Gash, V. W. J. Am. Chem. Soc. 1955, 77, 1552–1558.
30
Leonard (Scheme 10). Dehydrogenation with mercuric acetate
and diastereoselective reduction of the resulting bisenamine 51
provides 47. The ligand was reacted with palladium dichloride
to give (R-isosp)PdCl2 (49), and a solid-state structure was
obtained by X-ray crystallography. In this structure, both Pd-Cl
bonds are deflected from planarity by the flanking piperidine
rings of the sparteine ligand, and the sum of the six angles
around the square plane is 693.72°, the smallest for our sparteine
complexes thus far.
(e) Oinuma, H.; Dan, S.; Kakisawa, H. J. Chem. Soc., Chem. Commun.
1
983, 654–655. (f) Blakemore, P. R.; Kilner, C.; Norcross, N. R.;
Astles, P. C. Org. Lett. 2005, 7, 4721–4724. (g) Przybylska, M.;
Barnes, W. H. Acta Crystallogr. 1953, 6, 377–384.
(
29) For references pertaining to ꢀ-isosparteine, see: (a) Greenhalgh, R.;
Marion, L. Can. J. Chem. 1956, 34, 456–458. (b) Carmack, M.;
Douglas, B.; Martin, E. W.; Suss, H. J. Am. Chem. Soc. 1955, 77,
4
1
1
435. (c) Carmack, M.; Goldberg, S. I.; Martin, E. W. J. Org. Chem.
967, 32, 3045–3049. (d) Wanner, M. J.; Koomen, G.-J. J. Org. Chem.
996, 61, 5581–5586. (e) Childers, L. S.; Folting, K., Jr.; Streib, W. E.
(30) Leonard, N. J.; Beyler, R. E. J. Am. Chem. Soc. 1950, 72, 1316–
Acta Crystallogr. 1975, B31, 924–925.
1323.
1
5964 J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008

Oxidative Kinetic Resolution of Secondary Alcohols
A R T I C L E S
Table 4. Oxidative Kinetic Resolution with (-)-R-Iso- and
indication that site A of the (sp)Pd fragment (Figure 6) is the
relevant reactive site, and that a Curtin-Hammett situation
seems unlikely to be operative in the reaction.
(
+)-ꢀ-Isosparteine
(
R-isosp)PdCl2 (49) is likely a less-reactive catalyst because
approach of the substrate and formation of the postulated
palladium-alkoxide intermediate are hindered by the additional
steric bulk; reaction at site B of (sp)PdCl2 (14) was never
favored in the formation of cationic pyridine complexes. On
the other hand, because (ꢀ-isosp)PdCl2 (50) features two type
A sites, it likely forms an alkoxide complex, but this is slower
to undergo ꢀ-hydrogen elimination. This may be the case if a
entry
catalyst
base
time (h) conversion (%) ee (%)
s
a
b
(
R-isosp)PdCl
2
2
(49) 47
72
24
12
12
17
17
34
58
17
36
31
45
29
93
5
35
4
4.7
17.3
2
6.0
1
1
c
a
a
a
b
2
3
4
5
6
(sp)PdCl
2
(14)
1
d
d
(R-isosp)PdCl
(49) Cs
CO
CO
2
2
3
3
(sp)PdCl
2
(14)
Cs
b
(ꢀ-isosp)PdCl
(ꢀ-isosp)PdCl
2
(50) 48
2
d
distorted Pd-Cl bond is necessary for good reactivity; such
2
(50) Cs CO
2
3
6
1
2
distortion may facilitate displacement of Cl to an axial position
a
b
c
Average of two runs. 0.15 equiv of ligand. Average of three
below the square plane by a C-H agostic interaction (Figure
d
runs. 1.0 equiv of Cs2CO3.
5). Another possibility is that the more exposed palladium center
can easily form a stable bisalkoxide that is resistant to
(
+)-ꢀ-Isosparteine (48) was prepared by the method of
31
ꢀ
-hydrogen elimination.
These results provide strong evidence that the truly C2-
Winterfeld. Thermolysis of (-)-sparteine (1) in the presence
of 1.17 equiv of AlCl3 at 180-200 °C in a sealed tube provided
a mixture of all three sparteine diastereomers (Scheme 11).
symmetric diastereomers of (-)-sparteine (1) are inferior ligands
for the oxidative kinetic resolution in terms of both selectivity
and reactivity. 1 is perhaps especially effective for this reaction
because it provides a highly specific steric environment at the
palladium center. One coordination site is accessible to the
substrate (site A, or quadrant IV), whereas the other sites contain
the steric bulk necessary to desymmetrize the transition state
enough to effect selectivity (Cl at site B, quadrants I and II).
Unlike many catalytic enantioselective processes in which the
chiral ligand blocks one face of a prochiral substrate from
reaction, the oxidative kinetic resolution reaction creates a
prochiral molecule. This unusual scenario seems to require a
ligand environment more exotic than that provided by standard
3
2
Separation by column chromatography provided 48. Ligation
to palladium dichloride was achieved in CH2Cl2, and a solid-
state structure of (ꢀ-isosp)PdCl2 (50) was obtained by X-ray
crystallography. Without the steric intrusion of the trans-
quinolizidine ring system at the metal center, the complex is
more planar. The sum of the angles around the palladium atom
is 710.68°, comparable to bisalkoxide 34 and the closest to
square planar geometry of any sparteine-derived complexes
containing a chloride ligand.
The oxidative kinetic resolution of 1-phenylethanol (28),
our prototypical substrate, was attempted under our standard
conditions with 5 mol % of the catalyst (49 or 50) and 0.15
equiv of excess (-)-R- or (+)-ꢀ-isosparteine (47 or 48) in
toluene at 80 °C in the presence of O2 and MS3Å (Table 4).
After 72 h with 47, 34% conversion with 29% ee was
obtained, for a selectivity factor (s) of 4.7, compared to s )
3
5
2
C -symmetric ligands.
Summary and Conclusion
We have developed a model for the stereoselectivity in the
Pd-catalyzed aerobic oxidative kinetic resolution of secondary
alcohols. The model is based on the solid-state structures of
coordination complexes and general reactivity trends of (sp)-
PdCl2 (14). The first solid-state structure of a non-racemic chiral
palladium alkoxide is presented and further demonstrates the
subtle steric influences of the ligand (-)-sparteine (1). High-
level calculations support our model and emphasize the essential
role that the halide ligand plays in selectivity. The C2-symmetric
diasteromers of 1, (-)-R-isosparteine (47) and (+)-ꢀ-iso-
sparteine (48), were synthesized and investigated from a
structural standpoint as well as in the oxidation reaction itself.
Both were less selective and less reactive than 1, which supports
the unusual conclusion that a C1-symmetric ligand is particulary
effective for this oxidative kinetic resolution.
1
7.3 with (sp)PdCl2 (14, entries 1 and 2). 47 appears to induce
a much lower reactivity and selectivity in the kinetic
resolution. To account for any effect that free ligand may
have on the selectivity, the kinetic resolution was carried
out in the absence of excess ligand, but with cesium carbonate
as a stoichiometric base (entries 3 and 4). The reaction with
9 as catalyst showed little conversion and almost no
3
3
4
3
4
selectivity. For 48, whether with excess ligand (entry 5)
or inorganic base (entry 6), there is essentially no selectivity.
The poor selectivity and low reactivity of (R-isosp)PdCl2 (49)
in the kinetic resolution support our hypothesis that (-)-sparteine
(
1) is a particularly effective ligand due to its C1 symmetry.
While the increased steric bulk that (-)-R-isosparteine (47)
provides close to the metal center could be expected to increase
the steric interactions that control selectivity, it instead hampers
even reactivity. In addition, no benefit is gained from the
increased symmetry of the ligand, which stands in contrast to
many asymmetric reactions for which a C2-symmetric ligand
Acknowledgment. This work is dedicated to the memory of
Prof. Nelson J. Leonard. The authors thank the NIH-NIGMS (R01
GM65961-01), Bristol-Myers Squibb Co. and the American Chemi-
1
6
provides better selectivity. These results also provide further
(35) References describing C1-symmetric ligand in asymmetric processes: (a)
Resconi, L.; Cavallo, L.; Fait, A.; Piemontesi, F. Chem. ReV. 2000,
1
00, 1253. (b) Helmchen, G.; Pfaltz, A. Acc. Chem. Res. 2000, 33,
(
(
(
(
31) Winterfeld, K.; Bange, H.; Lalvani, K. S. Justus Liebigs Ann. Chem.
336. (c) Williams, J. M. J. Synlett 1996, 705. (d) Hatano, M.;
Tamanaka, M.; Mikami, K. Eur. J. Org. Chem. 2003, 2552. (e) Nozaki,
K.; Komaki, H.; Kawashima, Y.; Hiyama, T.; Matsubara, T. J. Am.
Chem. Soc. 2001, 123, 534. (f) Kocovsky, P.; Vyskocil, S.; Smrcina,
M. Chem. ReV. 2003, 103, 3213. (g) Bolm, C.; Verruci, M.; Simic,
O.; Cozzi, P. G.; Raabe, G.; Okamura, H. Chem. Commun. 2003, 2826.
(h) Ohashi, A.; Kikuchi, S.-i.; Yasutake, M.; Imamoto, T. Eur. J. Org.
Chem. 2002, 2535. (i) Hoge, G.; Wu, H.-P.; Kissel, W. S.; Pflum,
D. A.; Greene, D. J.; Bao, J. J. Am. Chem. Soc. 2004, 126, 5966.
1
966, 698, 230–234.
32) Although the sign of rotation is different than for (-)-sparteine (1)
and (-)-R-isosparteine (47), it is in the same enantiomeric series.
33) Cesium carbonate (along with excess (-)-sparteine) is a component
of the “rate-accelerated” conditions for oxidative kinetic resolution.
34) While the selectivity for the reaction using (sp)PdCl2 (14) without
added ligand was also diminished, we believe this is a result of
(
-)-sparteine (1) decomplexation in the absence of excess ligand.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008 15965

A R T I C L E S
Trend and Stoltz
cal Society (graduate fellowship to R.M.T.), Merck Research
Laboratories, Abbott Laboratories, Pfizer, Amgen, Boehringer
Ingelheim, GlaxoSmithKline, Lilly, and Johnson and Johnson for
generous financial support. Mr. Lawrence Henling and Dr. Michael
Day are gratefully acknowledged for X-ray crystallographic struc-
tural determination.
Supporting Information Available: Experimental details,
X-ray data, enlarged schemes from the text, and NMR spectra
PDF, CIF). This material is available free of charge via the
Internet at http://pubs.acs.org.
(
JA804955E
1
5966 J. AM. CHEM. SOC. 9 VOL. 130, NO. 47, 2008