Angewandte
Chemie
enantioselectivity (97% to > 99% ee); the Ru/BArF system,
(R,R)-5e, provided the corresponding enantiomers
(R,R)-2a–l in good diastereoselectivity (3:1 to 14:1 d.r.) and
excellent enantioselectivity (88% to > 99% ee). With both
catalytic systems, hydrogenation of substrates bearing Cl (1h)
and Br (1i) substituents in the meta position of the phenyl ring
gave relatively lower diastereoselectivity, albeit with high
enantioselectivity (Table 2, entries 15–18). Notably, when
using para-substituted substrates in reactions catalyzed by
(R,R)-5e, the enantioselectivity of the products was found to
increase gradually as follows: F (1c, 88% ee), Cl (1d,
93% ee), and Br (1e, 96% ee). The reactions of a substrate
bearing a more electron-withdrawing group (1 f) were less
diastereoselective for both catalytic systems, albeit the
product was obtained in higher enantioselectivity when
using (R,R)-5e as the catalyst (Table 2, entries 11 and 12
versus entries 3 and 4). In addition, the presence of the
electron-donating methoxy group on the fused phenyl ring
resulted in slightly lower enantioselectivity (Table 2,
entries 21 and 22). Remarkably, a complete reversal in the
sense of asymmetric induction ((S,S)-2j, > 99% ee to (R,R)-
2j, > 99% ee) was observed when using substrate 1j (Table 2,
entries 19 and 20).
forms an electrostatic ion pair with the activated substrate
(see the Supporting Information, Scheme S4). To explain the
reversal in the sense of asymmetric induction by switching to
a more weakly coordinating anion, we propose the presence
of an alternative set of CH–p interactions in the transition
state, in this case, interactions between the h6-arene ligand in
the ruthenium complex and the substituted phenyl rings of the
substrate (see the Supporting Information, Scheme S4). This
hypothesis is supported by the observation of changes in the
stereoselectivity of the reaction upon modification of the
electronics of the phenyl substituents at the 2- and 4-positions
of the substrate when the Ru/BArF catalyst system was used
(Table 2); a similar effect was previously observed in the
asymmetric transfer hydrogenation of aryl ketones using
similar half-sandwich RuII complexes as the catalyst.[18] Thus,
unlike the Ru/(PhO)2PO2 system, in which both hydrogen
bonding and CH–p interactions guide the hydride transfer,
the selectivity of the Ru/BArF system is probably only
governed by CH–p interactions alone. To support this
hypothesis, we carried out the hydrogenation of 1a with
(R,R)-5g in a mixture of MeOH/CH2Cl2. The diastereo- and
enantioselectivities decreased gradually in the presence of
increasing amounts of MeOH (see the Supporting Informa-
tion, Table S5) and a reversal in the sense of asymmetric
induction was observed when the ratio of MeOH/CH2Cl2
reached 1:2 (v/v) or higher. A similar decrease in diastereo-
and enantioselectivity was observed when the same inves-
tigation was carried out with the ruthenium/BArF catalyst,
(R,R)-5e, although the reversal in the sense of enantioselec-
tivity did not occur. This result indicates the weakening or
even the breaking of hydrogen bonds between the
Although a detailed mechanistic investigation will form
part of future studies, we herein propose that the hydro-
=
genation of the C N bond can proceed via either of two
different transition states depending on the ability of the
catalyst counteranion to participate in hydrogen bonding. In
the case of the Ru/(PhO)2PO2 catalyst system, (R,R)-5g, we
propose a pericyclic transition state that is similar to the one
we proposed previously for the AH of quinoline (Fig-
ure 1);[14b,17] this transition state involves a CH–p interaction
À
(PhO)2PO2 ion, activated substrate, and ruthenium species,
in the presence of protic solvent. Notably, almost identical
diastereo- and enantioselectivities were observed for both
catalytic systems when only MeOH was used as the solvent
(see the Supporting Information, Table S5, entries 9 and 10).
To further support this hypothesis and extend the
substrate scope, we investigated the AH of 2,4-dialkyl
substituted substrates, in which only one phenyl ring on the
substrate is available for the formation of CH–p interactions.
Initially, 2,4-dimethyl-1,5-benzodiazepine (3a) was used as
the substrate for catalyst screening (Table 3, entries 1—4, and
the Supporting Information, Table S3). Unsurprisingly, the
counteranion-induced reversal in enantioselectivity was not
observed despite an obvious effect on diastereoselectivity.[19]
On the other hand, upon catalyst screening, we found that the
hydrogenation of 3a with 1.0 mol% of (R,R,S)-6h (see
Scheme 1) gave fairly good diastereoselectivity and excellent
enantioselectivity (Table 3, entry 5; 10:1 d.r. and > 99% ee).
After investigating the effects of solvent, temperature and
pressure on the reaction outcome (see the Supporting
Information, Table S5), several 2,4-dialkyl-1,5-benzodiaze-
pines were hydrogenated under optimized conditions
(Table 3, entry 5). It was found that all substrates were
efficiently reduced to afford the corresponding products in
full conversions with good diastereoselectivity and excellent
enantioselectivity (Table 3, entries 6–11; 5:1 to 9:1 d.r., and
> 99% ee).
Figure 1. Proposed transition state involving (R,R)-5g as the catalyst.
between the h6-arene ligand in the ruthenium complex and
the fused phenyl ring of the substrate. Based on this model,
we propose that a hydride is transferred from the ruthenium
=
center to the Re face of the C N moiety to give the S-
configured product, consistent with the experimental results.
In addition, this model also explains the pronounced differ-
ence in enantioselectivity between the reaction catalyzed by
the complex bearing a OTfÀ and that by the complex bearing
a OMsÀ (39% ee for OTfÀ and 98% ee for OMsÀ), as these
counterions have very similar structures but different hydro-
gen bond forming ability.
On the other hand, the ruthenium hydride intermediate
bearing weakly coordinating anions, such as BArFÀ, probably
Angew. Chem. Int. Ed. 2012, 51, 1 – 6
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!