Organic Letters
Letter
a
Table 1. Reaction Condition Optimization
Figure 1. Dynamic kinetic resolution in asymmetric hydrogenation of
ketones.
interactions other than steric hindrance would be more
practical.
The NH effect in transition metal-catalyzed homogeneous
hydrogenation10 has been demonstrated to be important for
carbonyl substrates. One plausible explanation involved
intermolecular hydrogen bonding between the NH proton
and the carbonyl oxygen atom. However, DFT computational
studies by Dub et al.11 and our group12 revealed an alternative
participation of alkali metal cation to NH proton in ruthenium-
and iridium-catalyzed hydrogenation of ketones. We envi-
sioned that either the hydrogen bonding between the covalent
fluorine atom and the protic NH or the through-space
interaction between the fluorine atom and alkali metal cation
could be utilized to discriminate fluorine from the hydrogen
atom in the DKR13 (Figure 1c).
Our study was initiated with the discovery of the
advantageous catalytic system. 2-Fluoro-1-phenylpropan-1-
one (1a) was selected as the model substrate, and both
ruthenium and iridium catalysts were tested at the beginning.
In order to ensure a well-established equilibrium between the
two enantiomeric starting materials, the hydrogenation rate
should be limited to a low level. We therefore carried out the
hydrogenation reaction in isopropanol under mild conditions
(room temperature, 10 bar of hydrogen pressure). In addition,
a relatively high base concentration was applied to promote the
ketone−enolate interconversion. With potassium tert-butoxide
as the base, Noyori and co-workers’s Ru(II) bisphosphine/
diamine catalyst14 yielded α-fluoro alcohol with full conversion
and high enantioselectivity (93% ee for the major diaster-
eomer), but the diastereoselectivity was poor (Table 1, entry
1). The iridium tridentate catalysts developed by our group,
Ir/f-amphox,15 Ir/f-amphol,16 and Ir/f-ampha,17 gave a desired
reactivity (full conversion) and enantioselectivity (>95% ee) as
expected. However, the diastereomeric induction was not
satisfactory. The highest dr value was obtained with Ir/f-ampha
L7 (entry 8, 76:24). After systematically screening different
bases, we found that 10 mol % sodium hydroxide gave much
improved diastereoselectivity under 1 bar of hydrogen pressure
(entry 10).
entry
catalyst
base (mol %)
dr
ee (%)
1
2
3
4
5
6
7
8
9
[Ru]
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
KOtBu (5)
NaOH (10)
54:46
47:53
56:44
71:29
67:33
76:24
69:31
76:24
71:29
88:12
93
98
97
99
98
98
98
99
>99
99
L1/[Ir(COD)Cl]2
L2/[Ir(COD)Cl]2
L3/[Ir(COD)Cl]2
L4/[Ir(COD)Cl]2
L5/[Ir(COD)Cl]2
L6/[Ir(COD)Cl]2
L7/[Ir(COD)Cl]2
L8/[Ir(COD)Cl]2
L7/[Ir(COD)Cl]2
b
10
a
Reaction conditions: substrate (0.2 mmol), iPrOH (0.5 mL), H2 (10
bar), 25 °C. H2 (1 bar).
b
dr and ee values. The absolute configurations of products 2e
and 2p were determined by XRD analysis of a single crystal
structure, and an anti-selectivity was presented. This reaction
could be carried out at a larger scale without erosion of either
the dr or ee value (2d).
DFT computational studies (for details, see the Supporting
Information) were carried out to get insights into how both
enantioselectivity and diastereoselectivity were achieved in this
chemical transformation. Since the hydrogen activation was
already understood in our previous studies,11,12,16 we only
focused on the catalyst−substrate complexation in the
transition states of the hydride transfer step. The structures
of the transition states that lead to two pairs of diastereomers
were optimized in isopropanol (Figure 2). Due to the steric
hindrance of the bulky aryl group on the phosphorus atom, the
phenyl group from the substrate should only be placed at the
coordinating oxygen side. The calculated diastereomeric and
enantiomeric ratios18 were roughly in agreement with the
observed values (dr calculated 14:1 vs observed 9:1; ee
calculated >99.9% vs observed 99%). The stereocenter of
benzylic carbon was dictated by the facial approach of the
ketone substrate to the catalyst (top or bottom). Interestingly,
the Na−F and Na−O distances indicated a pronounced O,F-
bidentate chelation with sodium cation was present. This
charge−diploe interaction19 was believed to lower the energy
barriers in the hydride transfer step and therefore enhance the
With this optimized condition, we examined the reaction
scope of this chemical transformation (Table 2). When a
phenyl is attached to the carbonyl group, both the
enantioselectivities and diastereoselectivities were satisfactory
with 0.1% catalyst loading. The substituents on the benzene
ring, regardless of the position, did not have a significant
influence on the conversion and the stereoselectivities. The
substituent adjacent to the targeting C−F bond could vary
from alkyl group to aryl group (2p), with the retention of high
B
Org. Lett. XXXX, XXX, XXX−XXX