Published on Web 10/01/2008
Strategy for Employing Unstabilized Nucleophiles in Palladium-Catalyzed
Asymmetric Allylic Alkylations
Barry M. Trost* and David A. Thaisrivongs
Department of Chemistry, Stanford UniVersity, Stanford, California 94305-5080
Received August 27, 2008; E-mail: bmtrost@stanford.edu
Palladium-catalyzed asymmetric allylic alkylations (AAA) have
found extensive application in the enantioselective construction of
stereogenic centers, enabling the synthesis of a broad diversity of
complex molecules.1 This family of reactions includes a vast array
of stabilized or “soft” carbon and heteroatom nucleophiles (those
from conjugate acids with a pKa less than 25). While nonenanti-
oselective palladium-catalyzed allylic substitution reactions with
unstabilized or “hard” nucleophiles (those from conjugate acids with
a pKa greater than 25) are known,2 there are no reports of such
reactions proceeding with high enantioselectivity.3 One significant
difference between these two classes of nucleophiles is the distinct
mechanistic pathways through which each undergoes transition
metal-catalyzed allylic substitution reactions: “soft” nucleophiles
Figure 1. Chiral ligands used in optimization experiments.
externally attack the allylic ligand outside the metal’s coordination
sphere to directly afford the product,4 while “hard” nucleophiles
internally attack the metal to form a neutral complex that liberates
the product by reductive elimination.5
Table 1. Selected Optimization Experimentsa
The importance of pyridine and related heterocycles, both as
common structural motifs in natural products and pharmaceutical
agents and as synthetic building blocks, led us to consider using
metalated 2-methylpyridine, a “hard” nucleophile, in palladium-
catalyzed AAA reactions. Not unexpectedly, when the 2-methyl-
pyridyl anion generated with n-BuLi was employed in an AAA
reaction, no desired product was observed. In considering ways to
“soften” this nucleophile, we were attracted to the BF3 complex of
2-methylpyridine, which has not heretofore been metalated stoi-
chiometrically to serve as a nucleophile in any type of alkylation
reaction.6
entry
ligand
solvent
temp (°C)
yield (%)b
ee (%)c
1
2
3
4
5
6
7
8
9
L1
L1
L1
L1
L2
L3
L4
L4
L4
L4
THF
THF
THF
THF
THF
THF
THF
DME
toluene
dioxane
25
40
4
-25
25
25
25
25
25
13
11
68
18
55
15
70
50
31
86
-30d
1.0
-8.4
-4.0
-20
-43
86
94
59
95
Initial experiments studied the reaction catalyzed by [(η3-
C3H5)PdCl]2 and ligand L1 (Figure 1) of tert-butyl cyclopent-2-
enyl carbonate with the complex generated in situ from 2-meth-
ylpyridine and BF3•OEt2 (Table 1, entry 1). From the bases
examined, LiHMDS emerged as the optimal reagent, although
several equivalents were required for the reaction to proceed to
full conversion. Similar observations have been made for lithium
amide bases employed in palladium-catalyzed AAA reactions with
ketone enolates,7 a requirement that suggests lithium aggregates
may be forming.8 Conducting the reaction above or below ambient
temperature proved deleterious to both the yield and the observed
enantioselectivity (Table 1, entries 2-4). Of the chiral ligands tested
(Figure 1), L4 provided the desired product in the highest yield
and enantiomeric excess (Table 1, entries 1, 5-7). Using a
noncoordinating solvent substantially decreased the efficiency and
selectivity of the reaction (Table 1, entry 9). Of the solvents
surveyed, dioxane was differential, delivering the desired product
in 86% yield and 95% ee (Table 1, entries 7-10).9
10
25
a Reactions run on a 0.16 mmol scale at 0.08 M using 1.0 equiv
tert-butyl cyclopent-2-enyl carbonate, 1.5 equiv 2-methylpyridine, 1.3
equiv BF3•OEt2, 3.5 equiv LiHMDS, 2.5 mol % [(η3-C3H5)PdCl]2, and
6.0 mol % ligand. b Isolated yield. c Determined by chiral HPLC.
d Negative ee value signifies that the opposite enantiomer of the product
as drawn was formed.
trace amounts of the desired adduct. When 2,4-lutidine was
employed in a competition experiment between the 2- and 4-position
of pyridine, substitution was observed exclusively at the 2-position
(Table 2, entry 4). In contrast, when 2,6-lutidine was employed
under the optimized conditions, no reaction was observed; presum-
ably 2,6-disubstitution inhibits efficient coordination of the nitrogen
atom with boron. Substitution at the 3-, 4-, and 5-positions was
tolerated, even for sterically bulky alkyl and aromatic groups (Table
2, entries 5-10). Notably, aryl chlorides are not susceptible to
oxidative addition by the palladium(0) species presumably generated
during the reaction and potentially competitive coordination of
dibenzofuranyl ethers to BF3 does not inhibit the reaction (Table
2, entries 9 and 10, respectively). Finally, 1-methylisoquinoline also
underwent the desired transformation, although longer reaction times
were needed for full conversion (Table 2, entry 11).
Subsequent experiments began to define the substrate scope of
this reaction (Table 2). Five-, six-, and seven-membered allylic
carbonates were all competent electrophilic partners for the reaction
with 2-methylpyridine (Table 2, entries 1-3). Attempts to conduct
an analogous AAA reaction with 4-methylpyridine provided only
9
14092 J. AM. CHEM. SOC. 2008, 130, 14092–14093
10.1021/ja806781u CCC: $40.75
2008 American Chemical Society
Trost, Barry M.
