J. M. Brown, T. Hayashi et al.
Compound (R,R)-1b is known to favour the formation of
(R)-3-phenylcyclohexanone. DFT analysis was carried out as
described above. The procedure was then repeated by using
(S,S)-1b to promote formation of the same hand of product
via the diastereomeric transition state (Figure 3). In both
Figure 4. Format for analysis of distortions at the carbometallation TS.
as x. The value of x is very similar for 1a and (R,R)-1b, but
is lower on the less favourable (S,S)-1b pathway. (Table 3;
Figure 4).
Table 3. Transition-state distortions according to Figure 4 (see Support-
ing Information for detailed analysis).[a]
Ligand
Angle a[b]
Angle b
Angle (aÀb)
Oop x[c]
1a
A
161.4
165.4
172.8
172.0
166.2
164.5
168.9
161.0
156.7
149.9
146.6
156.9
159.2
155.1
0.4
8.7
22.9
25.4
9.3
0.499
0.487
0.356
0.156
0.396
0.599
0.410
A
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
G
5.3
13.8
U
[a] The preferred TS to (R)-product is emboldened; only lowest energy
pathways A are included. [b] Adjacent to bridgehead, less substituted
alkene terminus. [c] In Angstroms; Oop is out of plane.
Figure 3. Transition-state structures for the preferred pathway from
(R,R)-1bcatalyst to (R)-product precursor (i), and from (S,S)-1b catalyst
to (R)-product precursor (ii); see Supporting Information, Figure B for
space-filling models.
The same exercise was carried out for the enantiomers of
2,6-diphenylbicycloACTHNUTRGNEUNG
[3.3.1]nona-2,6-diene 2b.[20] It is known
that the (S,S)-ligand preferentially gives the (R)-product in
the Rh-catalysed phenylation of cyclohex-2-enone, and the
DFT calculations are in accord with the experiment results.
Thus the TS for the (R,R)-2b series is 2.6 kcalmolÀ1 higher
in energy than for the (S,S)-2b series. The preferred TS in
the 2b series is 5.6 kcalmolÀ1 higher in energy than its 1b
counterpart. Qualitative comparison indicates that conjugate
addition of PhB(OH)2 to cyclohex-2-enone is lower yielding
(93% vs. 97%) and slower (3 h vs. 1 h) for 2b than for
1b.[8a.11a] This is insufficient to explain the large difference
and the possibility of a change in rate-determining stage to
the carbometallation step needs to be considered. Steric
pressure is only evident at the CMA TS in the 2b series; the
cases binding of the substrate alkene is less favourable than
for the parent diene complex from 1a; this is reflected in TS
energies for carbometallation that are higher in energy, and
mirror the energies of the preceding alkene-bound ground
states. More significantly, the experimentally preferred path-
way for (R,R)-1b to (R)-product is 3.2 kcalmolÀ1 lower in
energy than the corresponding (S,S)-1b to (R)-product path-
way (CMA, Table 2). The overall results concur with the ex-
periment and provide confidence for previous mechanistic
proposals. More remarkably, the two diastereomeric transi-
tion states fail to reveal significant intracomplex steric inter-
actions.
À
How are the two diastereomeric transition states from 1b
(CMA, Table 2) differentiated by >3 kcalmolÀ1 if not just
by steric compression? A closer comparison between (R,R)-
1b and (S,S)-1b showed two significant differences. In both
carbonyl oxygen is in proximity to vinylic C H, and well
within van der Waals contact for both diastereomers.
The high ee values found with the unsubstituted bicyclo-
AHCTUNGTREG[NNNU 3.3.1]nona-2,6-diene 2a (Table 1) had provided a driving
À
cases Rh Ca bonding is well-advanced at the carbometalla-
tion TS (both 2.125 ꢁ). Firstly, in the higher energy pathway
force for these calculations. The computed energies here are
more comparable with those obtained in the [2.2.2] series,
and lower than observed in the 2b-based calculations. More
significantly, the preferred TS is favoured by 1.5 kcalmolÀ1
and the sense is reversed relative to 2b, consistent with what
is observed experimentally. For both 2a and 2b, analysis
shows accord with the 1b series (Table 3); the favoured
pathway causes less pronounced (aÀb) roll distortion; also
for the favoured pathway, the ipso-carbon atom of the mi-
grating phenyl group is further out of the RhCaCb plane for
this pathway. Hence the correct product configuration may
involving (S,S)-1b, a rolling distortion of the ligand double
bond trans- to Rh Ca is evident.[19] This distortion can be
À
simply expressed as the angle difference (aÀb) (Figure 4). It
is much more apparent here than is observed with either
ligand 1a, or (R,R)-1b. Secondly, the four-membered reac-
tion centre is non-planar. The ipso-carbon atom approaches
Cb at an angle of approximately 1158 and is out of the
RhCaCb plane exo to the cyclohex-2-enone ring; the distance
of the ipso-C atom from that plane in ꢁngstroms is defined
82
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 80 – 84