Organometallics
Article
19.3 + 0.5), and 20.8 (TSCC10′; 20.3 + 0.5) kcal/mol. Given
the considerably lower activation energy for TSCC9′, it should
be the dominant reaction channel and indeed leads to (R)-3 in
agreement with experiment.
Additional comparisons are provided by the Newman
projections in Figure 11. For example, the C7−C5 substituents
in the highest energy species TSCC10′ are nearly eclipsed (X−
C7−C5−Y torsion angles for synperiplanar X/Y: 27, 26, 15°),
whereas those in the other transition states are staggered. In
terms of factors favoring TSCC9′ over TSCC9, both the enolate
and NO2 oxygen atoms are more extensively hydrogen bonded
in the former. Although comparisons of TSCC9′ with TSCC10
and TSCC10′ are more challenging due to the lower structural
homology, transition states in which the NO2 oxygen atoms no
Figure 13. Differences in electronic energies for GBI ligands excised
from the transition states TSCC9′ and TSCC12-I.
+
(RRuRCRC)-2+: Carbon−Carbon Bond Formation. Given
the comparable enantioselectivities and identical product
configurations obtained with the diastereomeric catalysts
longer bind to the HNMe2 moiety always exhibit higher
activation energies.
Eight additional transition states derived from (SRuRCRC)-2+
in the Supporting Information. For the two shown in Figure
12, TSCC12-I and TSCC12-II, the ΔG(CH2Cl2) values (17.5
and 19.6 kcal/mol) also represent activation energies, as there
are no intermediates like 8-III with energies lower than those
of the educts. These constitute the only other transition states
with activation energies lower than the highest energy in
Figures 9−11 (TSCC10′, 20.3 + 0.5 kcal/mol). The second,
TSCC12-II, leads to (R)-3 and shares several features with
TSCC9 but involves the opposite CsiCrePh enantioface of
trans-β-nitrostyrene and addition anti to the cyclopentadienyl
ligand. Since it has a much higher activation energy in
comparison to TSCC9′ (14.4 + 0.5 kcal/mol), the dominant
channel to (R)-3, it is not further analyzed.
(SRuRCRC)- and (RRuRCRC)-2+PF6 (Schemes 1 and 2), it
−
would not be surprising to compute similar sets of transition
states. The former complex can be converted to the latter by
simply exchanging the positions of the cyclopentadienyl and
carbonyl ligands. Both ligands are remote from the sites where
the catalyst and educts interact, as well as the 1,2-
diaminocyclohexane moiety. There are no van der Waals
contacts in any of the structures computed above or in the
crystal structures.5b
Accordingly, through similar procedures, the transition states
and ΔG(CH2Cl2) values depicted in Figure 14 were computed.
In this series, the latter are equivalent to the free energies of
activation. There are four transition states, TSCC13-I through
TSCC13-IV, corresponding to those in Figures 8−10, and two
more, TSCC13-V and TSCC13-VI, corresponding to those in
Figure 12. The complete reaction coordinates are illustrated in
HNMe2+ units of the 1,2-diaminocyclohexane moieties are anti
as opposed to syn to the cyclopentadienyl ligands for TSCC13-I
through TSCC13-IV, and syn as opposed to anti for TSCC13-V
and TSCC13-VI. This is a logical consequence of the
“cyclopentadienyl/carbonyl flip” noted in the preceding
paragraph.
The activation energy for TSCC12-I (17.5 kcal/mol) renders
it the most favorable channel to (S)-3. Here, the CreCsiPh
face of trans-β-nitrostyrene adds from a direction anti to the
cyclopentadienyl ligand, as opposed to the CsiCrePh face
adding syn as in TSCC9′. The lower activation energy for
TSCC9′ versus TSCC12-I appears to be due to a number of
factors. The hydrogen-bonding motifs are similar, but the
+
distances involving the NO2 oxygen atoms and HNMe2
moiety in TSCC9′ are generally shorter (1.73 vs 1.77 Å; 2.39
vs 2.57 Å; 2.63 vs 2.50 Å). The shortest O···HNMe2 linkage in
TSCC9′ exhibits a higher degree of linearity (164.9° vs 156.4°).
Also, the H−C7−C5−H torsion angle of TSCC9′ indicates a
more staggered arrangement of substituents (55.4° vs 47.4°).
Furthermore, the 1,2-diaminocyclohexane rings adopt much
different conformations in TSCC9′ and TSCC12-I. To help
gauge this effect, the substrates and the metal fragment were
removed to give the free substituted GBI ligands and their
:NMe2-protonated forms. As illustrated in Figure 13, the
conformations differ by a ca. 180° rotation about the
Here, the lowest energy transition state, TSCC13-II, leads to
the major product enantiomer, (R)-3 (addition of the Csi
CrePh face anti to the cyclopentadienyl ligand). It features
exactly the same grouping of hydrogen bonds as was found for
the lowest energy transition state derived from the
diastereomer (SRuRCRC)-2+ in Figures 8−10 (TSCC9′). Their
relative energies (15.1 vs 14.9 kcal/mol) represent the cost of
the “cyclopentadienyl/carbonyl flip”. The two lowest energy
transition states that lead to the minor product enantiomer,
(S)-3, are analogous to those in Figures 8−10 (TSCC13-V,
comparable to TSCC12-I (17.2 vs 17.5 kcal/mol) and involving
addition of the CreCsiPh face syn to the cyclopentadienyl
ligand; TSCC13-I, comparable to TSCC9 (18.4 vs 20.3 kcal/
mol) and involving addition of the CreCsiPh face anti to the
cyclopentadienyl ligand). In any case, the high enantioselectiv-
2
cyclohexyl−NH(Csp ) bond. That in TSCC12-I (B) is 1.2−
1.6 kcal/mol less stable than that in TSCC9′ (A). This can be
ascribed to interactions involving two axial C−H groups of the
cyclohexane ring and the CNH substituent of the NH
group. The stability difference increases to 2.1 kcal/mol when
the substrates are added.29
−
ities obtained with (RRuRCRC)-2+PF6 in Schemes 1 and 2
largely derive from two competing transition states with a
computed 2.1 kcal/mol difference in activation energies
(TSCC13-II and TSCC13-V).
In any case, from a computational standpoint, the high
enantioselectivities for the additions of 1,3-dicarbonyl com-
pounds to trans-β-nitrostyrene catalyzed by (SRuRCRC)-2+PF6
−
in Schemes 1 and 2 largely derive from two competing
transition states with a 2.6 kcal/mol difference in free energies.
This agrees well with the >99:<1 to 92:8 product enantiomer
ratios.
DISCUSSION
■
The data in Figures 5 and 6 show that enantiopure 1+PF6
−
would be expected to catalyze additions of 1,3-dicarbonyl
J
Organometallics XXXX, XXX, XXX−XXX