Angewandte
Research Articles
Chemie
as Zn have no effect on the asymmetric hydrogenation (for
favorable transition state 2ATS1S leading to the desired
details, see the Supporting Information). These results
revealed that the carboxyl group was essential in this
[Co]/BPE system. It is possibly involved in both the activation
of the pre-catalyst and the control of activity and enantio-
selectivity.
(S)-form final product PS was computed to be lower in free
energy than 2ATS1R forming the (R)-form product PR by
about 2.1 kcalmolÀ1 in solution.[19,20a] This computational
result (equivalent to 94.4% ee) is qualitatively consistent
with the experimental result (90% ee, Scheme 2). Metalla-
2
2
In the presence of D2, 1b was hydrogenated smoothly to
give deuterated product 2b-d2 in > 98% conversion (Sch-
eme 4b). Furthermore, no deuterated product was detected
when the solvent was replaced by tBuOD (Scheme 4c). The
results demonstrated that protonation of the Co-alkyl inter-
mediate was probably not involved in this transformation and
the hydrogen source was H2. The hydrogenation of 1b with
H2/D2 mixture was also conducted, and some mono-deuter-
ated product 2b-d1 was observed, indicating that the hydro-
gen atoms in the product may origin from two molecules of
hydrogen gas (for details, see the Supporting Information).
Besides, a series of CoII-catalyzed asymmetric hydrogenation
of 1b were conducted with (S,S)-Ph-BPE with different
enantiopurities as the ligand under the standard reaction
conditions. A linear effect was observed in this asymmetric
reduction, indicating a 1: 1 binding pattern between (S,S)-Ph-
BPE and cobalt(II) stearate (Supporting Information, Fig-
ure S6).[16] EPR experiments were also conducted to monitor
the process of asymmetric hydrogenation using 1b as model
substrate. Our EPR results suggested that a paramagnetic
CoII species could be involved in the catalytic cycle (Figur-
es S7–S9).
cycle intermediates A2S and A2R can be formed after this
2
insertion. Coordination of one hydrogen molecule to A2S
2
2
2
and A2R then takes place to form A3S and A3R before s-
2
2
bond metathesis via ATS2S and ATS2R to afford the final
products and CoII-hydride species. Alternatively, more stable
intermediates 4A2S and 4A3S can be formed after spin
4
transition prior to ATS2S. Such spin crossing is supported
by our PCM B3LYP-D3//B3LYP-D3, PCM PBE0-D3//PBE0-
D3 and PCM M06-L//M06-L methods (Table S7). Finally,
ligand exchange of the product PS or PR by another substrate
molecule completes the catalytic cycle and regenerates the
4
active species A1S or 4A1R.[20b]
(Relative) distortion/interaction analysis[5b, 21] further sug-
2
gests that, compared to favorable ATS1S, a larger distortion
2
energy in ATS1R plays the dominant role in the enantiose-
lectivity (Figure 3b). In this connection, the phenyl group of
the substrate is oriented into a close (top left) quadrant of the
2
catalyst in ATS1S, in which one phenyl group of the chiral
ligand adapts to stack with the phenyl group of the substrate
(Figure 3a). Whereas, owing to the CoII-carboxylate chela-
tion, the phenyl group of the substrate is forced to be
2
positioned into the two right quadrants in ATS1R, in which
DFT study. On the basis of the previous computational
studies on asymmetric hydrogenation employing 3d metal
catalysts[12f,17] and our control experiments (Scheme 4), sys-
tematic density functional theory (B3LYP-D3 (mainly))
calculations were conducted to understand the origin of the
enantioselectivity.[18] Our computational results suggest that
the active CoII-carboxylate hydride intermediates (A1R or
A1S) preferentially undergo hydrogen insertion into C1 atom
of the substrate (Figure 2) via 2ATS1S and 2ATS1R, which
were computed to be the rate-, regio- and stereo-determining
step with the barriers of about 15.8–17.9 kcalmolÀ1 in solution
by the PCM B3LYP-D3//B3LYP-D3 method. The most
the two phenyl groups have a poor geometric complemen-
tarity. The steric map can also show the steric hindrance of the
chiral ligand in 4A1R, 2ATS1S and 2ATS1R (Figure 3c).[22]
Indeed, the steric repulsion can be partly attributed to
a shorter H-H distance between the substrate and ligand in
2
2ATS1R (1.96 vs. 2.12 in ATS1S; Figure 3a). Moreover,
their free energy difference was slightly increased by 0.2 kcal
molÀ1 in the absence of dispersion correction (Table S18).
These computational results and our additional calculations
(replacement of the phenyl groups by a smaller methyl
group(s) and its substitution effect evaluated by isodesmic-
reaction approach;[23] Figure S13) support that the steric
effect between phenyl groups of the ligand and the substrate
should play a more critical role in the observed enantio-
selectivity.
Based on our combined mechanistic and DFT (Figure 2)
investigations, a plausible mechanism of cobalt(II)-catalyzed
asymmetric hydrogenation of tetrasubstituted a,b-unsaturat-
ed carboxylic acids was proposed (Scheme 5). After ligand
exchange and heterolytic cleavage of hydrogen, the cobalt(II)
monohydride species 4A1S is formed and then enters the
catalytic cycle.[12b] Metallacycle intermediates 2A2S can be
formed after the migratory insertion of cobalt(II) monohy-
dride 4A1S via transition state 2ATS1S. The migratory insertion
is suggested to be the rate-, regio- and stereo-determining
2
step. Coordination of one hydrogen molecule to A2S then
4
takes place to form A3S. Subsequently, CoII-hydride species
4A4S is produced via s-bond metathesis (4ATS2S). The ligand
exchange of 4A4S with another unsaturated carboxylate
substrate 1a releases the less acidic hydrogenation product
2a and regenerates the cobalt(II) hydride species 4A1S.
Scheme 4. Mechanistic studies for cobalt(II)-catalyzed asymmetric
hydrogenation.
Angew. Chem. Int. Ed. 2021, 60, 11384 –11390
ꢀ 2021 Wiley-VCH GmbH