Synthetic Methodology for the Construction of Structurally Diverse Cyclopropanes

This Dissertation will cover cyclopropane methodology for the synthesis of 1,2,3-trisubstituted cyclopropanes. We envisioned building a cyclopropane precursor with two main features. The first feature was to use the known equilibrium between the homoallylic cation, the cyclobutane cation, and the cyclopropyl carbinyl cation. We incorporated this feature by having a homoallylic group (X) displaced with allylsilane forming the cyclopropyl carbinyl cation. This cation would then begin to equilibrate. However, the second feature was to include a silyl group ? to the olefin and in a position to stabilize the forming cyclopropyl carbinyl cation with the ?-silicon effect and therefore drive the equilibrium toward the cyclopropane. The silicon was necessary to make sure that ring fragmentation and subsequent 1,2-elimination did not occur. Also, the facile elimination of silicon allowed us to trap the cyclopropane, which resulted in a terminal olefin which can be elaborated further.

The 'state of the art' at the time of my arrival entailed building the cyclopropane precursor with a boron acetylide opening of mono-substituted epoxides. This methodology allowed us to prove our concept, but was limited toward the synthesis of trisubstituted cyclopropanes. Therefore we proposed a method highlighted by the formation of the cyclopropane precursor using RCM chemistry.

The goal was to investigate the closure of substituted cyclopropane precursors and determine the stereochemical course of the reaction. The new method allowed us to start from a substituted homoallylic alcohol and in three transformations yield the desired precursor. There are many methods known for the preparation of suitable homoallylic alcohols. However, for this summary I will only highlight the results obtained from homoallylic alcohols derived from an indium mediated crotylation of hydrocinnamaldehyde.

The indium mediated crotylation resulted in a 3:1 mixture of anti:syn diastereomers. The predominance of anti is due to a closed six membered transition state where the crotyl indium species coordinates the aldehyde and delivers the crotyl group when the ethyl ester and the R group on the aldehyde are suedo equatorial in relationship. The diastereomers were separated by flash column chromatography. Then, each was taken on independently to yield a single 1,2,3-tribubstitued cyclopropane. The homoallylic alcohol was protected with allylchlorodimethylsilane followed by closure with Grubbs' ruthenium catalyst to afford the silyloxycycloheptene. The seven-membered ring was opened with HF'¢pyridine to give the desired cyclopropane precursor bearing a fluorosilane ? to the olefin and a homoallylic alcohol ready for activation. Activation was accomplished with triflic anhydride and 2,6-lutidine in methylene chloride and afforded a single 1,2,3-trisubstituted cyclopropane from the displacement of triflate and subsequent elimination of the fluorosilane.

After both diastereomers were taken through the sequence we postulated the transition state structure that would result in the formation of the 1,2,3-cyclopropane that resulted. We determined the relative cyclopropane stereochemistry after several 1H-NMR decoupling experiments. We concluded that the reaction was an SN2 displacement and that the steric environment controls the resulting stereochemistry. Once we limit the discussion to an SN2 reaction we find there are only two possible transition states that allow for the anti-bonding orbital of the carbon-triflate bond to overlap with the ? bond of the olefin. The allylsilane can be either on the same side as or oppose the ethyl ester. We found that a classic A-strain argument can be used to determine what transition state is preferred. Therefore the allylsilane and the ethylester will oppose each other, or the hydrogen ? to the carbonyl prefers to eclipse the olefin (A-strain). The transition states shown lead to the cyclopropane formed.

The ethyl ester substituted case has also been further established using methyl substituted homoallylic alcohols showing that the argument was universal. Starting from the homoallylic alcohol we have established a four-step sequence to give a single diastereomeric cyclopropane. The method is highlighted by the RCM strategy and has shown to be quite robust.

After finishing the trisubstituted cyclopropane work we looked back into the sequence at our silyloxycycloheptene intermediates and realized that within the molecule laid a latent homoallylic alcohol with a ?-silyl group. This is shown when we took the intermediate and reduced the ethyl ester to the corresponding primary alcohol. This reaction was carried out at low temperature to ensure that silicon-oxygen bond cleavage did not occur. With the silyloxy ring intact we activated the primary hydroxy and the cyclopropanation took place to afford a disubstituted cyclopropane with cis orientation of the two substituents. We then took the same intermediate and treated it with excess LAH which provided the corresponding diol from reduction of the ester followed by silicon-oxygen bond cleavage. This molecule was activated using standard conditions and cleanly yielded a disubstituted cyclopropane where the orientation of the two groups were trans to each other. This second case once again proved the A-strain argument.

We have proposed that the reason for the diol to afford a cyclopropane of trans stereochemistry was because of the steric environment during the carbon-carbon forming step. We have postulated that the olefin and the ?-hydrogen are eclipsed or in other words, that the group containing the secondary hydroxyl and the olefin oppose each other. This theory is in concert with the transition states we have seen before in the trisubstituted case where the 1,3 interaction is the dominant force.

All this work was published in Organic Letters, Vol. 2, No. 5, 601-603. The paper summarizes the work that used the RCM strategy for the synthesis of trisubstituted cyclopropanes and disubstituted cyclopropanes of both cis and trans geometry.

Our methodology had now developed from a limited three-step sequence from epoxides to a four-step sequence starting from a homoallylic alcohol that exploited the RCM reaction with Grubbs' ruthenium catalyst. At this time Grubbs et. al. began publishing on a second generation catalyst that was quite promising for cross-metathesis reactions and exhibited a large range of functional groupcompatibility. We chose to exploit the new catalyst for our third generation synthesis of cyclopropanes. We envisioned using allyltrimethylsilane and a homoallylic alcohol as coupling partners in a cross metathesis reaction.

We accomplished crossing both substituted and simple homoallylic alcohols with allyltrimethylsilane using the new Grubbs' catalyst in yields ranging from 70% to 95%. This new procedure was then used to make all the cyclopropanes previously published. However, during the development of this two step method we found that when the cyclopropane precursor of substituted alcohols was activated we got a 50:50 mixture of desired cyclopropane and trimethylsilane protected starting material. Previously we did not encounter this problem when the fluorosilane was the stabilizer. Since the result ruined the efficacy of our route we explored other activation conditions (Table #). After attempting multiple conditions we found that thionyl chloride at 0o C in methylene chloride was a viable substitute. With these new activating conditions we then were able to synthesis a variety of cyclopropanes, both disubstituted and trisubstituted.

We have now shown that cyclopropanes can be synthesized in two steps from a variety of homoallylic alcohols. We have demonstrated the viability of the Grubbs' catalyst for the cross-metathesis of allyltrimethylsilane and varied homoallylic alcohols . We have also shown that thionyl chloride is an excellent reagent for the activation of the cyclopropane precursor. All this data was published as a full paper in The Journal of the American Chemical Society.