Synthesis and application of chiral cyclopropane-based ligands in palladium-catalyzed allylic alkylation

Article abstract of DOI:10.1021/jo048782s

A series of chiral, cyclopropane-based phosphorus/sulfur ligands have been synthesized and evaluated in the palladium-catalyzed allylic alkylation of 1,3-diphenylpropenyl acetate with dimethyl malonate. Variation of the ligand substituents at phosphorus,

Full text of DOI:10.1021/jo048782s

Synthesis and Application of Chiral Cyclopropane-Based Ligands
in Palladium-Catalyzed Allylic Alkylation
Gary A. Molander,* Jason P. Burke, and Patrick J. Carroll†
Roy and Diana Vagelos Laboratories, Department of Chemistry, University of Pennsylvania,
Philadelphia, Pennsylvania 19104-6323
gmolandr@sas.upenn.edu
Received July 16, 2004
A series of chiral, cyclopropane-based phosphorus/sulfur ligands have been synthesized and
evaluated in the palladium-catalyzed allylic alkylation of 1,3-diphenylpropenyl acetate with dimethyl
malonate. Variation of the ligand substituents at phosphorus, sulfur, and the carbon backbone
revealed 24d to have the optimal configuration for this reaction, giving the product in high yield
and with good enantioselectivity (93%). A model for the observed enantioselectivity is discussed
within the context of existing models, using X-ray crystallographic data, solution-phase NMR studies,
and the absolute stereochemistry of the products. Selected ligands were also evaluated in the
palladium-catalyzed intermolecular Heck reaction and the rhodium-catalyzed hydrogenation of a
dehydroamino acid.
Introduction
The asymmetric catalysis of organic reactions to pro-
vide enantiomerically enriched compounds has became
one of the most powerful tools in modern synthetic
organic chemistry.1 The majority of recent progress in
this field has come from metal complexes containing
chiral organic ligands. A quick survey of the most efficient
ligands, however, reveals an interesting fact: most belong
to just a few structural classes. This is true whether
referring to the C2-symmetric diphosphine ligands that
originally dominated the field or to the more recently
introduced bidentate ligands with mixed coordinating
atoms. Examples of the former include binaphthyl and
biaryl derivatives, bisoxazolines, salens, and tartrate-
based ligands, all of which are C2-symmetric. Examples
of the latter include those derived from ferrocenylphos-
phines, phosphinooxazolines, and cinchona alkaloids.
Until a method exists for the accurate and reliable
rational design of ligands, an empirical exploration of
various ligand frameworks and chelating atoms remains
one of the best ways to discover new catalyst systems.
Although a large variety of structurally diverse ligands
have already been tested in asymmetric catalysis, cyclo-
propane skeletons have received relatively little attention
to date. This is surprising because the cyclopropane ring
offers an advantageous combination of structural rigidity,
low molecular weight on a well-defined and highly
variable platform, and unusual bond angles. Moreover,
recent improvements in the stereoselective synthesis of
this ring make exploring the use of cyclopropanes even
more attractive.2,3
FIGURE 1. Cyclopropane-based chiral ligands.
The first appearance of a cyclopropane-based ligand
in the literature was in 1979, by Colleuille and co-workers
(1, Figure 1).4 In a survey of several other ring structures,
a cyclopropyl-based diphosphine ligand was tested in the
rhodium-catalyzed hydrogenation of dehydroamino acids,
providing a 23% ee of the reduction product.
In 1992, Minami and co-workers evaluated diphos-
phine 2 in an asymmetric allylic alkylation catalyzed by
palladium, and in this case a 61% ee was achieved.5 More
recently, Sibi and co-workers have used bisoxazoline 3
in a variety of different transformations with excellent
enantioselectivity. In these latter studies, however, ad-
vantage was not taken of the unique structural features
of the cyclopropyl moiety in orienting the heteroatoms
bound to the metal.6,7
(2) Lebel, H.; Marcoux, J. F.; Molinaro, C.; Charette, A. B. Chem.
Rev. 2003, 103, 977-1050.
(3) Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz,
A., Yamamoto, H., Eds.; Springer-Verlag: Berlin, Germany, 1999; Vol.
2, pp 513-606.
(4) Aviron-Violet, P.; Colleuille, Y.; Varagnat, J. J. Mol. Catal. 1979,
5, 41-50.
† To whom all correspondence concerning X-ray crystal structures
should be addressed.
(1) Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz,
A., Yamamoto, H., Eds.; Springer-Verlag: Berlin, Germany, 1999.
(5) Okada, Y.; Minami, T.; Yamamoto, T.; Ichikawa, J. Chem. Lett.
1992, 547-50.
(6) Sibi, M. P.; Ji, J. J. Org. Chem. 1997, 62, 3800-3801.
10.1021/jo048782s CCC: $27.50 © 2004 American Chemical Society
Published on Web 10/19/2004
8062
J. Org. Chem. 2004, 69, 8062-8069

Synthesis of Chiral Cyclopropane-Based Ligands
FIGURE 2. X-ray crystal structure of Ni(1)Br2.
We sought to explore the three-dimensional space of
the cyclopropyl motif in greater depth to determine its
potential utility in asymmetric synthesis, and some of
our initial results are reported herein.
Results and Discussion
As an easy entry into cyclopropane-based catalysts and
a useful point of reference for further studies, ligand 1
was synthesized with use of the known alcohol 4 (eq 1).8
FIGURE 3. X-ray crystal structure of [Pd(1)Cl2]2.
fectively incorporated within the first model system.
Though many combinations of strong and weak hetero-
atom pairs are known, several recent reports demon-
strating the utility of P/S ligands in a variety of different
applications made it an appealing choice.9-17
The first generation of P/S ligands took advantage of
the known iodo alcohol 7, derived from a Charette
cyclopropanation (Scheme 1).18
Variations in the carbon framework were also explored,
although the gem-dimethyl analogue required a resolu-
tion to obtain enantiopure ligand (Scheme 2).
As might have been anticipated, when tested in the
palladium-catalyzed allylic alkylation of 1,3-diphenyl-
propenyl acetate with dimethyl malonate, ligand 1
produced disappointing results in a variety of solvents
(25-27% ee). To gain insight into why ligand 1 performed
so poorly, metal complexes were synthesized for X-ray
analysis.
The solid-state structures clearly show that although
ligand 1 is indeed able to chelate tetrahedral nickel(II)
(with a bite angle of 105.4°), the square-planar geometry
of palladium(II) results in a dimeric complex with bridg-
ing ligands (Figures 2 and 3). This suggests that the
active catalyst in the allylic alkylation reaction and the
reduction reactions of Colleuille et al. are actually dimeric
or a mixture of monomeric nonchelated catalyst com-
plexes, resulting in a highly fluxional environment that
subsequently leads to lower enantioselectivity.
For comparison, a 1,2-trans substituted ligand (17) was
also synthesized (Scheme 3).
The ligands were then tested according to the standard
palladium-catalyzed asymmetric allylic alkylation (AAA)
with 1,3-diphenylpropenyl acetate and dimethyl mal-
onate (Table 1). The ee values were reported as observed
(uncorrected) for ligands of ∼90% ee (10, 12, and 17) or
99% ee (15). Results show that having the phosphorus
directly attached to the cyclopropane ring was beneficial
(12 versus 10), but altering the carbon framework (12f
versus 15) or the cis configuration of the cyclopropane
First-Generation P/S Ligands. With these results
in mind, new frameworks were designed with more
accommodating metal-chelating centers. The systems
initially chosen (10 and 12, Scheme 1) bore a cis orienta-
tion of heteroatoms to facilitate chelation and were
accessed via a modular approach that allowed fine-tuning
of ligand performance. The smaller chelate ring required
the use of electronic effects to direct the incoming
nucleophile and thereby control asymmetric induction,
as opposed to restricting the number of diastereomeric
transition states via the C2-symmetrical paradigm inef-
(9) Albinati, A.; Eckert, J.; Pregosin, P.; Ruegger, H.; Salzmann, R.;
Stossel, C. Organometallics 1997, 16, 579-590.
(10) Enders, D.; Peters, R.; Runsink, J.; Bats, J. W. Org. Lett. 1999,
1, 1863-1866.
(11) Evans, D. A.; Campos, K. R.; Tedrow, J. S.; Michael, F. E.;
Gagne, M. R. J. Am. Chem. Soc. 2000, 122, 7905-7920.
(12) Yan, Y. Y.; RajanBabu, T. V. Org. Lett. 2000, 2, 199-202.
(13) Nakano, H.; Okuyama, Y.; Yanagida, M.; Hongo, H. J. Org.
Chem. 2001, 66, 620-625.
(14) Sugama, H.; Saito, H.; Danjo, H.; Imamoto, T. Synthesis 2001,
2348-2353.
(15) Mancheno, O. G.; Priego, J.; Cabrera, S.; Arrayas, R. G.; Llamas,
T.; Carretero, J. C. J. Org. Chem. 2003, 68, 3679-3686.
(16) Faller, J. W.; Wilt, J. C.; Parr, J. Org. Lett. 2004, 6, 1301-
1304.
(17) Nakamura, S.; Fukuzumi, T.; Toru, T. Chirality 2004, 16, 10-
12.
(18) Charette, A. B.; Juteau, H.; Lebel, H.; Molinaro, C. J. Am.
Chem. Soc. 1998, 120, 11943-11952.
(7) Sibi, M. P.; Ma, Z.; Jasperse, C. P. J. Am. Chem. Soc. 2004, 126,
718-719.
(8) Barrett, A. G. M.; Hamprecht, D.; White, A. J. P.; Williams, D.
J. J. Am. Chem. Soc. 1997, 119, 8608-8615.
J. Org. Chem, Vol. 69, No. 23, 2004 8063

Molander et al.
SCHEME 1
SCHEME 2a
TABLE 1. Best Results of P/S Ligands Tested in
Pd-Catalyzed AAA
L*
R
R′
ee, %
10aa
10ba
12a
12b
12c
Ph
CH3
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Cy
i-Pr
Ph
Ph
Ph
22
23
70
69
76
48
59
78
33
48
65
74
72
76
64
24
16
Ph
t-Bu
1-Ad
Et
a Reagents and conditions: (a) (+)-3-bromocamphor-8-sulfonic
acid chloride, Et3N; (b) resolve; (c) 2,6-(Me)2PhSH, K2CO3, DMF,
86%; (d) Zn, AcOH, 6:1 dr, 38%; (e) t-BuLi; Ph2PCl, 60%.
12db
12e
Ph
3,5-(Me)2Ph
2,6-(Me)2Ph
2,6-(Me)2Ph
4-(CF3)Ph
12f
trans-12f
12g
12h
12i
SCHEME 3a
4-(MeO)Ph
2,4,6-(Me)3Ph
2,4,6-(i-Pr)3Ph
2,6-(Me)2Ph
2,6-(Me)2Ph
2,6-(Me)2Ph
2,6-(Me)2Ph
12jb
12kc
12la
15
17
a THF used as solvent. b Toluene used as solvent. c CH2Cl2 used
as solvent.
a
Reagents and conditions: (a) PPh3, Br2, 63%; (b) 2,6-
(Me)2PhSH (1 equiv), K2CO3, DMF, 43%; (c) LiPPh2, 61%.
(e.g., to trans-12f and 17) was not. Additionally, bulky
aromatic substituents on the sulfur gave the best results
(12f), with electron-donating groups giving improved
yield and enantioselectivity (12g versus 12h). However,
increasing the steric bulk on the sulfide aromatic ring
did not always increase the enantioselectivity (12j).
Varying substituents on the phosphine did not show any
improvement either (12k,l).
X-ray Structures. Solid-state structures confirmed
that the cis ligands did indeed chelate without forming
dimers (Figures 4-6). The trans influence could also be
seen, with the Pd-Cl bond trans to phosphorus measur-
ing 0.06 Å longer than the corresponding bond trans to
sulfur for complex 18 (Figure 4). The triisopropyl complex
19 was relatively similar in both bite angle and magni-
tude of the trans influence (Figure 5), while 20 showed
a smaller difference in the trans effect (0.04 Å). Addition-
ally, while complexes 18 and 19 showed a twist boat
conformation, the gem-dimethyl complex 20 was es-
FIGURE 4. X-ray crystal structure of Pd(10a)Cl2.
sentially planar (Figure 6). On this basis, there is no
obvious, direct correlation between either the effective
trans effect or the conformation of the chelate with the
ee values observed.
Second-Generation P/S Ligands. Although these
P/S ligands represented a drastic improvement upon the
8064 J. Org. Chem., Vol. 69, No. 23, 2004

Synthesis of Chiral Cyclopropane-Based Ligands
SCHEME 4a
a Reagents and conditions: (a) n-BuLi; I2, 98%; (b) KO2CNdN-
CO2K, AcOH, 87%; (c) Et2Zn, CF3CO2H, CH2I2, 74%; (d) PPh3, Br2,
5:1 dr, 91%; (e) RSH, Cs2CO3, 35-87%; (f) t-BuLi, Ph2PCl, 45-
84%.
FIGURE 5. X-ray crystal structure of Pd(12j)Cl2.
TABLE 2. Best Results of Second Generation P/S
Ligands in Pd-Catalyzed AAA
L*
24a
24a‚BH3
24b
24b‚BH3
24c
24d
Pd
yield, %
ee, %
[Pd(allyl)Cl]2
Pd(OAc)2
[Pd(allyl)Cl]2
Pd(OAc)2
[Pd(allyl)Cl]2
[Pd(allyl)Cl]2
78
>95
>95
>95
>95
>95
75
44
74
64
91
93
FIGURE 6. X-ray crystal structure of Pd(15)Cl2.
tively cyclopropanated to afford a single isomer by NMR
analysis. The selectivity of this reaction is rationalized
by steric strain in the transition structure as a result of
A1,3-interactions. The relative stereochemistry of 22 was
determined by X-ray crystallographic analysis of the
p-bromobenzoyl ester derivative.
The standard palladium-catalyzed AAA utilized to
benchmark this second generation of ligands provided
much more promising results (Table 2). Surprisingly, the
smaller sulfide substituents clearly gave the best results,
with the S-methyl ligand 24d affording the best enanti-
oselectivity of all ligands tested in this study. This was
counterintuitive based upon the known inversion of
sulfides chelating to metals and the resulting preponder-
ant use of bulky alkyl groups employed in most successful
P/S ligands to prevent this phenomenon.11,12,15,17,20 Sub-
jecting the borane-protected ligands (24a‚BH3, 24b‚BH3)
to the same reaction conditions with Pd(OAc)2, forming
palladium(0) in situ, reduced both the rate and the
enantioselectivity of the reaction.21
FIGURE 7. Isomerization of the S-stereocenter upon metal
coordination.11
initially tested ligand 1, a second series of P/S ligands
was designed to overcome some of the weaknesses
inherent in the first-generation design (ligands 10, 12,
15, and 17). By placing an additional stereocenter
between the sulfide and the cyclopropane ring, we
perceived that two key problems of the first-generation
ligands could be solved: (1) iodo alcohol 7 was generated
in e90% ee, which, neglecting any nonlinear effects,
thereby limited the potential enantiopurity of the cata-
lyzed reaction products, and (2) inversion of stereochem-
istry at the newly formed stereogenic site at sulfur upon
coordination with the metal created the possibility of
isomerization and the generation of competing diaster-
eomeric complexes (Figure 7).19
To our chagrin, the S-methyl ligand 24d was the most
difficult to prepare, because only by using a nucleophile
prepared in situ from methanethiol (a gas at room
temperature) could success be achieved in incorporation
of the thioalkoxy group. All other attempts at its syn-
Synthesis of the second-generation ligands began with
the commercially available (R)-3-butyn-2-ol (>98% ee)
that was easily iodinated and reduced with diimide
(Scheme 4). The vinyl iodide was then diastereoselec-
(20) Albinati, A.; Pregosin, P. S.; Wick, K. Organometallics 1996,
15, 2419-2421.
(21) Riegel, N.; Darcel, C.; Stephan, O.; Juge, S. J. Organomet.
Chem. 1998, 567, 219-233.
(19) Murray, S. G.; Hartley, F. R. Chem. Rev. 1981, 81, 365-414.
J. Org. Chem, Vol. 69, No. 23, 2004 8065

Molander et al.
TABLE 3. Results of Ligand 24d with Various Nucleophiles and Electrophiles in Pd-Catalyzed AAA
FIGURE 9. X-ray structure of Pd(24c)Cl2.
FIGURE 8. X-ray structure of Pd(24b)Cl2.
thesis failed, whether using previously deprotonated
methanethiol or forming the thiolate in situ through
reduction of the disulfide. Nevertheless, 24d could be
obtained in quantities suitable for further studies.
Scope. Ligand 24d was evaluated in other AAA
reactions with various electrophiles and nucleophiles to
determine the scope of its performance (Table 3). Dis-
appointingly, ligand performance was not as successful
in these trials. Thus, both the bulkier diethyl methyl-
malonate and benzylamine nucleophiles provided inferior
results, and less reactive allylic acetates also performed
poorly.
X-ray Structures. Not surprisingly, several of the
second-generation ligands were shown to chelate on
square-planar palladium(II) (Figures 8-10). The six-
FIGURE 10. X-ray structure of [Pd(II)(1,3-diphenylallyl)-
(24c)]SbF6 (counterion omitted for clarity).
membered chelating rings again adopted a twist boat
conformation. Whether sterically bulky or small, the
sulfur substituent maintained a pseudoaxial orientation
(Figures 8 and 9). Bite angles for the palladium com-
plexes also remained fairly constant between 95 and 97°.
A crystal suitable for X-ray analysis was also obtained
for the allyl complex by allowing ligand 24c to react with
1,3-diphenypropenylpalladium(II) chloride followed by an
anion exchange with silver hexafluoroantiminate (Figure
10).
Solution-Phase Conformation. To determine if the
solid-state configuration was the same as that in solution,
complexes 26 and 27 were inspected by 1H and 31P NMR
spectroscopy. Interestingly, while 26 showed only one 31
P
8066 J. Org. Chem., Vol. 69, No. 23, 2004

Synthesis of Chiral Cyclopropane-Based Ligands
SCHEME 5
FIGURE 11. Possible allyl-palladium isomers.
be quicker than either nucleophilic reaction, because the
observed enantiomeric ratio of the product (91% ee, 21:1
er in CH3CN) is greater than the solution-phase ratio of
complex isomers (4.7:1 dr in CD3CN).
The proposed mechanism predicts that an (S,R,R)-
ligand (the enantiomer depicted in Figure 12) generates
the (S)-enantiomer as the major product. This is in fact
what is observed for ligand 24c (based upon optical
rotation measurements).
X-ray structures of complex 27 and the Evans ligand
complex 28 allow structural comparisons that might
reveal why the two catalyst systems, though sharing
many similar features, still differ in the levels of observed
enantioselectivity (Figures 13 and 14).11
A front-on comparison shows that in both complexes,
the 1,3-diphenylallyl ligand is rotated slightly counter-
clockwise relative to the perpendicular ligand/metal
plane. Quantitatively, in 27 the C1-Pd-C3 plane is 18.5°
rotated from the P-Pd-S plane, while in 28 the differ-
ence is only 9.4°. Both catalyst systems have similar face-
on/edge-on arrangements of the phenyl rings. The Pd-
C1 and Pd-C3 bond lengths also are similar, though 28
shows a slightly stronger trans influence (∆ 0.11 Å vs ∆
0.09 Å).
FIGURE 12. Interpretation of NOESY data in determining
M- and W-isomers.
resonance, 27 existed as two isomers in a 5.3:1 ratio at
room temperature in CD2Cl2. This suggests that sulfur
inversion is negligible at room temperature, but that
there is some equilibration between allyl conformations
in the η3-complex. Again, the lack of S-inversion is
surprising because previous reports required large alkyl
groups (e.g., tert-butyl) to prevent inversion (Figure 8).11
Of the several possible allyl isomers, COSY and
NOESY data suggested that both isomers of 27 in
solution have the syn-syn configuration, rather than the
sterically unfavorable syn-anti (or anti-anti) structure
(Figure 11).
In the observed NOESY spectrum, no NOE interac-
tions between the allyl protons and the S-ethyl group
were seen in the major isomer, suggesting that it is the
M-isomer; the same diastereomer was observed in the
solid state (Figure 12). Weak interactions were seen in
the minor isomer, however, between a terminal allylic
proton and the CH3 protons on the S-ethyl group,
suggesting that it is the opposite W-isomer. These
findings are also consistent with reports in the litera-
ture.11
A side-on comparison reveals similar tilts of the C1-
C2-C3 ligand plane relative to the P-Pd-S plane, with
complex 27 tilted 61.2° toward the face-on phenyl ring
Mechanistic Considerations. These conclusions al-
low mechanistic predictions and comparison to observed
products. Nucleophilic attack should occur at the carbon
trans to the stronger σ-donating and π-accepting phos-
phorus atom according to the trans influence. Attack at
this carbon on the M-isomer would give one enantiomer,
while attack on the analogous carbon in the W-isomer
would result in the opposite enantiomer (Scheme 5).
Attack on the M-isomer should be fast in that it relieves
steric strain between the allyl phenyl group and the
S-ethyl substituent, while attack on the minor isomer
would be slow owing to the steric strain generated in the
transition state between the S-ethyl group and the allyl
phenyl ring in the palladium(0)-olefin product. Curtin-
Hammett kinetics appear to be operating in this system.
Thus, interchange between the two allyl isomers must
FIGURE 13. Front-on comparison of 27 and the Evans
complex 28.
J. Org. Chem, Vol. 69, No. 23, 2004 8067

Molander et al.
TABLE 4. Results of Selected Ligands in Pd-Catalyzed
Intermolecular Heck Reaction
ee, %
L*
solvent temp, °C time, h conv, % A/B
A
B
1a
1
10a
10b
12ab
12ab
12c
THF
C6H6
THF
THF
C6H6
C6H6
THF
70
70
80
70
65
65
70
70
60
65
70
40
40
40
65
65
65
20
20
48
48
60
60
48
60
48
60
60
18
18
18
60
60
60
99
48
<5
99
99
93
81
<5
99
99
99
51
17
28
57
43
71
0.05
8
7
0.14 <5
∼1
8.0
3.9
16 <5
36 45
54 55
10.6 63 43
2.6
17 12
trans-12f THF
24da
THF
THF
C6H6
C6H6
C6H6
C6H6
THF
THF
THF
2.6
8.1
9.7
3.4
55 36
55 31
50 19
61 20
24da
24da
FIGURE 14. Side-on comparison of 27 and Evans complex
28.
24da,b
24da,c
24da
10.7 55
11.2 52
and the Evans system 61.9° toward the same direction
(Figure 14). The complexes differ, however, in the orien-
tation of the sulfur substituent: the ethyl group of
complex 27 points toward the allyl ligand while the tert-
butyl group of complex 28 points away from the allyl
ligand. This is also revealed in the nonbonded distance
between the first carbon off the sulfur and the first
phenyl ring carbon cis to the sulfur atom: 3.52 Å in 27
versus 4.35 Å in 28. The much shorter distance in
complex 27 might explain why smaller R groups on the
sulfide performed better; larger groups (e.g., tert-butyl)
would cause severe steric congestion. Although we can
only speculate, it may also explain the difference in the
levels of enantioselectivity observed between the Evans
ligand and the cyclopropyl series of ligands. Curiously,
as the allylic carbon is rehybridizing in the transition
structures leading to product, the S-Et group in 27 may
provide more steric congestion than the S-t-Bu group of
28 simply due to its orientation. Thus, although in both
systems the electronic effects favor alkylation at the site
cis to the S-ligand on palladium, steric effects in the
Evans systems at that same position are minimized
relative to those of the cyclopropyl ligand motif, and thus
higher enantioselectivities are observed.
24da
10.8 52 28
24da,d
24da,e
all B
6.8
0
49 16
a Et3N used as base. b Pd(OAc)2 used as catalyst. c Pd2(dba)3‚
CHCl3 used as catalyst. d 3 Å mol sieves added; Bu4NCl (1 equiv)
added. e K2CO3 used as base.
TABLE 5. Results of Various Ligands in Rh-Catalyzed
Hydrogenation
pressure,
psi
time,
h
conv,
%
ee,
%
L+
10b
12a
24d
24d
24d
24da
24db
100
100
14.7
120
200
100
100
42
42
48
19
18
18
18
27
93
99
41
19
99
7
5
46
41
47
47
45
a
b
MeOH used as solvent. Catalyst formed in situ with
Rh(COD)Cl and 24d.
Heck Reaction. To examine the reliability of the
allylic alkylation reaction as a ligand benchmark, selected
ligands were tested in the palladium-catalyzed inter-
molecular Heck reaction between phenyl triflate and 2,3-
dihydrofuran (Table 4). Overall, most ligands gave ee
values 20-30% lower in this asymmetric Heck reaction
than when evaluated in the palladium-catalyzed AAA.
Surprisingly, the best-performing ligand 12a (91% optical
purity) out-performed the previously successful ligand
24d. The highest A/B ratios were obtained with the
benzene/i-Pr2NEt combination; other bases, such as Et3N
and K2CO3, always gave less selective results. Moreover,
palladium sources other than Pd2(dba)3 generally gave a
larger percentage of isomerization product B.
Rh-Catalyzed Hydrogenation. A few ligands were
also tested in the rhodium-catalyzed hydrogenation of
dehydroamino acids (Table 5). Higher pressures of hy-
drogen increased enantioselectivity while at the same
time decreasing the rate of the reaction. Also, as the last
entry clearly demonstrates, the pre-catalyst had to be
synthesized beforehand and could not be formed in situ
without greatly lowering the reaction rate.
Conclusions
A new series of P/S ligands have been synthesized and
evaluated in the palladium-catalyzed allylic alkylation
reactions with moderate to good enantioselectivity. An
X-ray crystal structure of the active catalyst was ob-
tained. This, combined with NMR spectroscopic studies,
was used to assign the major diastereomer in solution,
which in turn was utilized to propose a mechanism for
the enantioselectivity of the allylic alkylation. The ligand
system was also compared to the Evans P/S catalyst
system, a “gold standard” among this ligand class. Subtle
structural differences were noted that may explain the
rather dramatic differences in the performance of the two
systems.
The scope of the best-performing ligand in this reaction
was also evaluated in the palladium-catalyzed intermo-
8068 J. Org. Chem., Vol. 69, No. 23, 2004

Synthesis of Chiral Cyclopropane-Based Ligands
lecular Heck reaction and the rhodium-catalyzed hydro-
genation of dehydroamino acids, both with somewhat less
successful results.
support and Johnson Matthey for a loan of the pal-
ladium catalysts used in this study. J.P.B. thanks Eli
Lilly for a Graduate Fellowship.
Experimental Section
Supporting Information Available: Experimental data,
spectral characterization, and general procedures for asym-
metric reactions; crystal data in CIF format. This material is
available free of charge via the Internet at http://pubs.acs.org.
See Supporting Information.
Acknowledgment. We thank the National Insti-
tutes of Health (NIGMS 48580) for their generous
JO048782S
J. Org. Chem, Vol. 69, No. 23, 2004 8069