Angewandte
Communications
Chemie
Table 1: Optimization of the model reaction.[a]
[12]
À
a stereogenic C F unit (Figure 1b).
To the best of our
knowledge, the chemistry provides the first example of an
intermolecular-aldol-based desymmetrization of achiral cyclic
1,3-diketones. Since the pioneering work of Hajos and
Parrish,[13] symmetric 1,3-diketones of type 1 have been
used extensively to design new desymmetric aldol processes.
However, these have proceeded through intramolecular
manifolds exclusively.[7a,13]
Our exploratory studies focused on the intermolecular
aldol reaction between the commercially available 2-methyl-
benzophenone (2a) and 2-benzyl-2-fluorocyclopentane-1,3-
dione (1a; see Table 1). The aldol acceptor is easily prepared
from cyclopentane-1,3-dione following a Knoevenagel con-
densation/reduction/electrophilic fluorination (using Select-
fluor) sequence. The catalytic experiments were conducted in
toluene over 16 hours and under irradiation by a single black-
light-emitting diode (black LED, lmax = 365 nm). We initially
evaluated a large set of different chiral organic catalysts with
a known propensity to activate substrates by hydrogen-
bonding or ion-pairing interactions.[14] Specifically, we used
a screening platform to automate the parallel execution of
small-scale reactions to determine the efficiency of a library
of 60 novel and known organocatalysts in this desymmetric
aldol reaction (see Figure S1 in the Supporting Information
for selected results). This evaluation quickly identified the
commercially available amido-thiourea 4a[15] as the most
promising catalyst.
The optimization studies were then continued on a mean-
ingful experimental scale (0.1 mmol). We first confirmed that
the catalyst 4a (20 mol%) afforded the product 3a with good
chemical yield and stereoselectivity (Table 1, entry 1; single
diastereoisomer). Modification of the benzyl amide moiety in
4 revealed that incorporating an additional stereogenic center
imparted a slightly higher stereoinduction, but with a minimal
matched/mismatched effect (entries 2 and 3). These results
suggested that the steric profile of the tertiary benzyl amide
component could play a more prominent role than the
presence of a second stereocontrol element in dictating the
reactionꢀs stereoselectivity. In consonance with this reasoning,
the best results were achieved with the catalysts 4d and 4e,
both containing a more encumbered amide moiety but
a single stereogenic center (entries 4 and 5).
The N-benzhydryl-substituted amido-thiourea catalyst
4e[15b] was selected for further optimization studies. A solvent
screening identified o-dichlorobenzene as the best reaction
medium (Table 1, entry 6). Since our illumination system
consisted of a black LED connected to an external power
supply, we could finely tune and control the intensity of light
emission (see Figure S2 for details of the illumination set-up).
Increasing the irradiance from 10 Æ 1 (used in the initial
experiments) to 15 Æ 1 mWcmÀ2, while extending the reaction
time to 30 hours provided 3a in 89% yield, as a single
diastereoisomer, and with 90% ee (entry 7). Then, control
experiments were conducted to glean insights into the
mechanism of the photochemical organocatalytic desymmet-
rization aldol process. The absence of light irradiation
completely suppressed the reaction (entry 8), while the
racemic adduct 3a was isolated in 15% yield in the absence
of 4 (entry 9). The last result, along with the high stereocon-
Entry
Catalyst
Solvent
Yield [%][b]
ee [%][c]
1
2
3
4
5
6
4a
4b
4c
4d
4e
4e
4e
4e
none
toluene
toluene
toluene
toluene
50
31
41
45
40
69
89
0
76
83
87
89
90
91
90
–
0
toluene
o-Cl2C6H4
o-Cl2C6H4
o-Cl2C6H4
o-Cl2C6H4
7[d]
8[e]
9
15
[a] Reactions performed over 16 h on a 0.1 mmol scale using 3 equiv of
2a under illumination by black LED (lmax =365 nm) with an irradiance of
10Æ1 mWcmÀ2. [b] Yield of the isolated 3a. [c] Enantiomeric excess of
3a determined by UPC2 analysis on a chiral stationary phase. In all cases
a single diastereoisomer was observed by 19F NMR analysis of the crude
reaction mixture. [d] Irradiance of 15Æ1 mWcmÀ2 and 30 hours of
reaction time. [e] In the dark.
trol achieved with the chiral amido-thiourea 4e, implies that
the rate acceleration facilitated by 4e is large enough to
overcome a racemic background process.
We then evaluated the generality of the light-driven
organocatalytic aldol desymmetrization process using the
optimized reaction conditions described in Table 1, entry 7.
We studied the reactivity of a variety of achiral 2-substituted-
2-fluorocyclopentane-1,3-diketones (1). As displayed in
Figure 2, different substituents at the aromatic ring of the
benzyl moiety in 1 could be used, regardless of either their
electronic and steric properties or position on the phenyl ring
(3a–h). To probe the synthetic utility of the method, we
demonstrated that good efficiency was maintained when
running the reaction on a 1 mmol scale (3a). A product
bearing a heteroaryl framework could be synthesized, as
shown for the furyl-substituted adduct 3i. The desymmetric
aldol reaction is also effective for 2-alkyl-2-fluoro cyclo-
pentane-1,3-diones, thus affording the corresponding adducts
3j,k with high enantioselectivities. For this substrate class, we
observed a correlation between the steric profile of the 2-alkyl
substituent and the relative stereocontrol of the desymmet-
rization process, since the diastereoselectivity increases with
the length of the alkyl chain (d.r. from 5.5:1 to 19:1 when
moving from a methyl to a butyl substituent; 3j and 3k). X-
ray crystallographic analysis[16] performed on crystals from
the adduct 3h allowed the assignment of the absolute
configuration for the newly formed fluorine-containing qua-
ꢀ 2017 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 11875 –11879