The new method was applied to a variety of aldol reactions, using experimental data
from Brown (Figure 5) [13]. For each system,
a conformation search was carried out with the constraint in place.
Figure 5: RCOMe reacts with R'CHO
A force constant of 25 kJ/mol
was chosen, which is a rather weak bond, and an optimal bond length of 2.4 Å was
used, which is quite a long bond. When the search was complete, the output files were filtered
using the Acca program, in
order to remove all structures for which the constrained distance was greater tha 2.9 Å.
In some cases, unconstrained conformation searches were also performed, in order
to compare the results. We tried to save time by reminimising all of the local minima found using the unconstrained searches, with the constraint in place. This turned out not to give
good coverage of the constrained conformational space.
The results are given in Table 2:
Table 2: Comparison with experimental results
RCOMe | R'CHO | Experimental | Monte Carlo | Constrained Monte Carlo | Filtered |
|
Me | Et | 62 % S | 33 % R | 6 % S | 15 % S |
Me | i-Pr | 61 % R | 45 % S | 0.4 % R | 10 % R |
Me | t-Bu | 83 % R | 7 % R | 15 % R | 19 % R |
Me | Ph | 57 % R | 35 % S | 13 % S | 29 % S |
| | | | | |
Et | Et | 61 % S | 16 % R | 10 % S | 16 % S |
Et | i-Pr | 48 % R | | 13 % R | 21 % R |
Et | t-Bu | 81 % R | | 18 % S | 16 % S |
Et | Ph | 39 % R | 33 % R | 65 % S | 13 % S |
| | | | | |
Ph | Et | 65 % S | | 37 % R | 13 % S |
Ph | i-Pr | 74 % R | | 60 % S | 16 % R |
Ph | t-Bu | 89 % R | 19 % S | 21 % R | 63 % R |
Ph | Ph | 58 % R | 45 % S | 65 % S | 41 % S |
| | | | | |
t-Bu | Et | 17 % S | | 3 % R | 4 % R |
t-Bu | i-Pr | 13 & R | | 15 % S | 16 % S |
t-Bu | t-Bu | 26 % R | | 54 % S | 53 % S |
t-Bu | Ph | 4 % R | 12 % S | 28 % S | 28 % S |
It is clear from Table 2 that the improved force field does not completely account for the selectivity of this reaction. In particular, it appears to be confused by phenyl groups and tertiary-butyl groups. However, It represents a considerable improvement on the original force field.
Chiral Ketones
The most severe test of the new procedure is to apply it to chiral ketones. Two examples were chosen, one with an alpha chiral centre [14], and the other with a beta chiral centre [15]. These are illustrated in Figure 6.
Figure 6: Aldol reactions of chiral ketones
The procedure correctly calculates a value of 71 % de for the ketone with an
-chiral centre. For the
centre, however, the incorrect sense of induction is calculated. The effects of a
-chiral
centre must depend on long range interactions to a greater extent that the effect of an
-chiral centre, and so the force field may be expected to be less accurate. Further investigation is required to optimise the improvements to the force field.